DISINFECTION BY-PRODUCT FORMATION FROM BIOFILM

CHLORINATION IN DRINKING WATER PIPES

Carolina Montoya-Pachongo

Submitted in accordance with the requirements for the degree of

Doctor of Philosophy

University of Leeds

School of Civil Engineering

March 2018

1
 Intellectual property and publication statement

The candidate confirms that the work submitted is her own, except where work which has formed
part of jointly authored publications has been included. The contribution of the candidate and the
other authors to this work has been explicitly indicated below. The candidate confirms that
appropriate credit has been given within the thesis where reference has been made to the work of
others.

Chapter 3 refers to the field assessment of bacterial communities and their relationships with
engineered factors. The results reported in this chapter were used to generate a jointly authored
publication:

C. Montoya-Pachongo, I. Douterelo. C. Noakes, M.A. Carmargo-Valero, A. Sleigh, J.C. Escobar-

Rivera, P. Torres-Lozada. Field assessment of bacterial communities and total trihalomethanes:
implications for drinking water networks. Journal Science of the Total Environment. 616–617, pp.
345–354.

The candidate is the first author of the publication and her contribution is related to planning and
executing the field work, sample processing, literature review, result analysis, writing up the
manuscripts, and making corrections for further publication. Other authors’ contribution was in the
form of advice on field work planning, easing access to laboratory facilities and equipment,
laboratory techniques, academic guidance, and writing style.

This copy has been supplied on the understanding that it is copyright material and that no quotation
from the thesis maybe published without proper acknowledgment. 2018. The University of Leeds.
Carolina Montoya Pachongo.

Disinfection by-product formation from biofilm chlorination in drinking water pipes i

Carolina Montoya Pachongo. School of Civil Engineering
 Acknowledgements

Firstly, I would like to express my sincere gratitude to my supervisors Prof Catherine Noakes, Dr
Miller-Alonso Camargo-Valero, and Dr Andrew Sleigh for the continuous support of my PhD study
and related research, for their guidance and support. My sincere thanks also go to Dr Isabel
Douterelo-Soler for her guidance on molecular methods applied to drinking water research.

Secondly, special gratitude with Colciencias for funding the PhD and the Leeds experience. Thanks
to the John Fox Award and the School of Civil Engineering for funding my field work in Colombia. I
want to thank Universidad del Valle for giving me access to laboratory facilities and field-work
equipment. Thanks to staff of EMCALI EICE ESP for granting permission and support on sample
collection. In particular, I am grateful with Prof Cristian Picioreanu for introducing me to COMSOL
modelling. Special thanks to colleagues in the Faculty of Engineering at the University of Leeds for
their guidance and advice on coding and modelling.

Last but not the least, I would like to thank my family: my parents Valentina and Harold, aunts,
cousins, and friends for spiritually supporting me throughout the PhD process. Special gratitude
goes to Diana and Davidde for supporting me under especial circumstances.

Disinfection by-product formation from biofilm chlorination in drinking water pipes ii

Carolina Montoya Pachongo. School of Civil Engineering
 Abstract

Biofilms in drinking water distribution networks (DWDNs) are recognized as potential pathogen
reservoirs. Recent experiments have found that biofilm can also act as precursors for the formation
of disinfection by-products (DBPs). This project aimed to better understand the impact of the
presence of biofilms and improve prediction of DBPs in DWDNs. To study the microbial significance
of biofilms in water pipes, bacterial communities in biofilm and bulk water were identified in a DWDN
in a tropical climate country. Drinking water and biofilms were characterised by physico-chemical
parameters. Relationships between biotic, physico-chemical parameters and engineered factors
(i.e., pipe age, material and diameter; and water age) were explored by the application of statistical
tests.

Additionally, improvement of DBP prediction in DWDNs was approached by modelling the role of
biofilms as DBP precursors. Two models for predicting DBP formation potentials were developed
from chlorination of cells and extracellular polymeric substances. The first model corresponded to
stagnant conditions and a second more complex model was produced for transitional and turbulent
flow. The models were implemented in the software COMSOL Multiphysics 5.2a and sensitivity
analysis was carried out to screen the parameters influence on the response variables.

Field-work assessment allowed determining that biofilms are richer habitats than bulk water. Pipe
age, pipe material, water age, free chlorine, pH and temperature can be key to the composition of
bacterial communities. Model simulations suggested that the important DBP exposure is related to
dichloroacetronitrile, stagnant bulk, and slow flow. The microbial and chemical significance of
biofilms is important in the context of climate change and developing countries because water
managers can face multiple challenges under these conditions. Alterations of raw water properties,
increasing occurrence of extreme weather events and poor capacity to mitigate such events may
rise the chemical and microbiological risk associated to biofilms in DWDNs in tropical countries.

Disinfection by-product formation from biofilm chlorination in drinking water pipes iii
Carolina Montoya Pachongo. School of Civil Engineering
 TABLE OF CONTENT

LIST OF ABBREVIATIONS __________________________________________________________ XIX

NOMENCLATURE _________________________________________________________________ XXI

1 INTRODUCTION _________________________________________________________________ 1

1.1 MOTIVATION ____________________________________________________________________ 1

1.2 PROBLEM DEFINITION _____________________________________________________________ 3

1.3 RESEARCH QUESTIONS, AIM AND OBJECTIVES ___________________________________________ 7

1.3.1 RESEARCH QUESTIONS __________________________________________________________ 7

1.3.2 AIM _________________________________________________________________________ 8

1.3.3 OBJECTIVES __________________________________________________________________ 8

1.4 GENERAL METHODOLOGY OF THE RESEARCH PROJECT ____________________________________ 8

1.4.1 STRUCTURE OF THE THESIS _______________________________________________________ 9

2 LITERATURE REVIEW ___________________________________________________________ 11

2.1 INTRODUCTION _________________________________________________________________ 11

2.2 PHYSICAL, HYDRAULIC AND WATER QUALITY INTEGRITY OF DRINKING WATER DISTRIBUTION NETWORKS 11

2.2.1 PHYSICAL INTEGRITY OF DWDNS__________________________________________________ 11

2.2.2 HYDRAULIC INTEGRITY OF DWDNS ________________________________________________ 13

2.2.3 WATER QUALITY INTEGRITY OF DWDNS _____________________________________________ 15

2.2.4 PLUMBING SYSTEMS ___________________________________________________________ 17

2.3 BASICS OF TRANSPORT PHENOMENA _________________________________________________ 18

2.3.1 FLUXES AND CONSERVATION OF CHEMICAL SPECIES ____________________________________ 18

2.3.2 MASS TRANSFER AND BOUNDARY CONDITIONS AT INTERFACES ____________________________ 19

2.3.2.1 Sherwood number __________________________________________________________ 20

2.3.2.2 Engineering correlations ______________________________________________________ 21

2.4 BIOFILMS IN DRINKING WATER DISTRIBUTION NETWORKS __________________________________ 23

2.4.1 TAXONOMY AND DIVERSITY ______________________________________________________ 24

2.4.2 MICROBIOME IDENTIFICATION IN BULK WATER AND BIOFILMS IN DRINKING WATER NETWORKS ______ 25
Disinfection by-product formation from biofilm chlorination in drinking water pipes iv
Carolina Montoya Pachongo. School of Civil Engineering
2.4.3 BIOFILM MODELLING ___________________________________________________________ 33

2.5 WATER DISINFECTION ____________________________________________________________ 46

2.5.1 CHLORINE REACTIONS IN WATER __________________________________________________ 46

2.5.1.1 Hydrolytic reactions _________________________________________________________ 46

2.5.1.2 Redox reactions ____________________________________________________________ 47

2.5.2 DISINFECTANT DECAY MODELS____________________________________________________ 47

2.5.2.1 Wall reaction kinetics ________________________________________________________ 48

2.5.2.2 Mass transfer kinetics _______________________________________________________ 50

2.5.2.3 Overall rate________________________________________________________________ 50

2.6 DBPS IN DRINKING WATER ________________________________________________________ 50

2.6.1 DISINFECTION APPROACHES _____________________________________________________ 50

2.6.2 DBP SPECIES ________________________________________________________________ 51

2.6.3 DBP REGULATION _____________________________________________________________ 54

2.6.4 DBP PRECURSORS ____________________________________________________________ 55

2.6.5 FACTORS INFLUENCING DBP FORMATION____________________________________________ 56

2.6.6 DBPS AND PUBLIC HEALTH ______________________________________________________ 58

2.6.7 DISINFECTION BY-PRODUCTS MODELLING ____________________________________________ 59

2.7 BIOFILMS AS PRECURSORS OF DBPS ________________________________________________ 74

2.8 FURTHER RESEARCH NEEDED: IMPROVE THE UNDERSTANDING OF THE ROLE OF BIOFILMS AS DBP
PRECURSORS _______________________________________________________________________ 77

3 FIELD ASSESSMENT OF BACTERIAL COMMUNITIES AND THEIR RELATIONSHIPS WITH

ENGINEERED FACTORS ____________________________________________________________ 81

3.1 INTRODUCTION _________________________________________________________________ 81

3.2 METHODOLOGY ________________________________________________________________ 81

3.2.1 DATA COLLECTION FROM COLOMBIAN WATER COMPANIES _______________________________ 81

3.2.2 AREA OF STUDY ______________________________________________________________ 82

3.2.3 PRELIMINARY ACTIVITIES ________________________________________________________ 85

3.2.4 WATER AGE DATA _____________________________________________________________ 85

3.2.5 SAMPLE COLLECTION___________________________________________________________ 87

3.2.6 PHYSICOCHEMICAL ANALYSIS ____________________________________________________ 89

Disinfection by-product formation from biofilm chlorination in drinking water pipes v

Carolina Montoya Pachongo. School of Civil Engineering
3.2.7 MOLECULAR METHODS__________________________________________________________ 90

3.2.8 DATA ANALYSIS _______________________________________________________________ 91

3.3 RESULTS _____________________________________________________________________ 93

3.3.1 SUSPENSION OF DRINKING WATER SUPPLY IN THE STUDIED NETWORK _______________________ 93

3.3.2 WATER LEVELS IN SERVICE RESERVOIRS IN THE DWDN __________________________________ 96

3.3.3 WATER AGE_________________________________________________________________ 100

3.3.4 PHYSICAL CHARACTERISTICS OF SAMPLED PIPES _____________________________________ 100

3.3.5 WATER QUALITY AND BIOTIC PARAMETERS __________________________________________ 101

3.3.6 CHARACTERIZATION OF THE BACTERIAL COMMUNITY STRUCTURE OF BIOFILMS AND BULK WATER __ 104

3.3.7 MICROBIAL RICHNESS AND DIVERSITY ______________________________________________ 107

3.4 DISCUSSION __________________________________________________________________ 108

3.4.1 BULK WATER QUALITY AND BIOTIC PARAMETERS AND THEIR RELATIONSHIPS WITH ENGINEERED FACTORS
__________________________________________________________________________ 108

3.4.2 IDENTIFICATION OF THE BACTERIAL COMMUNITY STRUCTURE OF BIOFILMS AND BULK WATER ______ 108

3.4.3 INFLUENCE OF NETWORK CHARACTERISTICS OVER THE BACTERIAL COMMUNITIES AND RICHNESS AND
DIVERSITY INDICES IN BIOFILMS AND BULK WATER SAMPLES ____________________________________ 112

3.5 SIGNIFICANCE AND LIMITATIONS OF THE STUDY ________________________________________ 115

3.6 CONCLUSIONS ________________________________________________________________ 116

4 A SIMPLE BIOFILM CHLORINATION MODEL________________________________________ 118

4.1 INTRODUCTION ________________________________________________________________ 118

4.2 MODEL DEVELOPMENT __________________________________________________________ 119

4.2.1 MODEL DEFINITION ___________________________________________________________ 119

4.2.2 MODEL EQUATIONS ___________________________________________________________ 121

4.2.3 PARAMETERS SELECTION _______________________________________________________ 123

4.2.3.1 Biofilm thickness (BT) _______________________________________________________ 123

4.2.3.2 Initial volumetric biofilm density (Xo) ___________________________________________ 126

4.2.3.3 Initial concentration of EPS (Eo) _______________________________________________ 129

4.2.3.4 Effective diffusion coefficient of dissolved substances (C and  S) ____________________ 130

4.2.3.5 Reaction rates (k1 and k2) and yield coefficients (Y1 and Y2) _________________________ 130

4.2.3.6 Factors FX and FE __________________________________________________________ 131

Disinfection by-product formation from biofilm chlorination in drinking water pipes vi

Carolina Montoya Pachongo. School of Civil Engineering
4.2.3.7 Temperature______________________________________________________________ 131

4.2.4 BOUNDARY CONDITIONS _______________________________________________________ 132

4.2.5 GEOMETRY AND MESH _________________________________________________________ 133

4.2.6 ANALYSIS OF MESH DEPENDENCE ________________________________________________ 133

4.2.7 PROCESSING AND NUMERICAL METHODS ___________________________________________ 134

4.2.8 MODEL IMPLEMENTATION IN COMSOL MULTIPHYSICS ___________________________________ 136

4.2.9 SELECTION OF TIME STEP ______________________________________________________ 136

4.2.10 EXPERIMENTAL STUDY USED AS A REFERENCE ______________________________________ 137

4.2.11 DATA ANALYSIS _____________________________________________________________ 138

4.3 RESULTS AND DISCUSSION _______________________________________________________ 139

4.3.1 TYPICAL MODEL BEHAVIOUR ____________________________________________________ 139

4.3.2 CHLORINE PENETRATION WITHIN THE BIOFILM________________________________________ 141

4.3.3 SENSITIVITY ANALYSIS _________________________________________________________ 142

4.3.3.1 S, reaction rates k1 and k2, and initial chlorine concentration ________________________ 143

4.3.3.2 Temperature______________________________________________________________ 147

4.3.3.3 Yield coefficients Y1 and Y2 and reactor volume __________________________________ 148

4.3.3.4 Biofilm thickness, initial cell and EPS concentration, fraction of cells and EPS transformed into
DBP and biofilm area _______________________________________________________________ 149

4.3.3.5 Comparison among parameters _______________________________________________ 150

4.3.4 FACTORS AFFECTING THE DBP FORMATION _________________________________________ 155

4.3.5 COMPARISON BETWEEN MODEL RESULTS AND EXPERIMENTAL DATA _______________________ 155

4.3.6 SIMULATION OF SCENARIOS: BUILDING PLUMBING SYSTEM SUPPLIED WITH WARM WATER ________ 156

4.3.6.1 Chloroform formation compared with field assessment (Chowdhury, 2016) _____________ 157

4.3.6.2 Chloroform formation _______________________________________________________ 159

4.3.6.3 Dichloroacetonitrile formation _________________________________________________ 161

4.3.6.4 Chlorine demand __________________________________________________________ 162

4.3.6.5 Influence of S/V ratio on DBP formation_________________________________________ 163

4.3.6.6 Discussion of DBP formation potentials simulated under stagnation conditions __________ 166

4.4 MODEL APPLICATIONS AND LIMITATIONS _____________________________________________ 167

4.5 CONCLUSIONS ________________________________________________________________ 169

Disinfection by-product formation from biofilm chlorination in drinking water pipes vii
Carolina Montoya Pachongo. School of Civil Engineering
5 MODELLING DBP FORMATION FROM BIOFILM CHLORINATION UNDER HYDRODYNAMIC
CONDITIONS _____________________________________________________________________ 171

5.1 INTRODUCTION ________________________________________________________________ 171

5.2 MODEL DEVELOPMENT __________________________________________________________ 171

5.2.1 MODEL DEFINITION ___________________________________________________________ 171

5.2.2 MODEL EQUATIONS AND INITIAL AND BOUNDARY CONDITIONS – REACTIONS __________________ 173

5.2.3 MODEL EQUATIONS AND INITIAL AND BOUNDARY CONDITIONS – BULK FLOW __________________ 179

5.2.4 PRE-PROCESSING ____________________________________________________________ 184

5.2.4.1 Geometry and mesh ________________________________________________________ 184

5.2.4.2 Mesh convergence and selection of time step ____________________________________ 187

5.2.4.3 Selection of turbulence model ________________________________________________ 190

5.2.5 IMPLEMENTING THE MODEL IN COMSOL _____________________________________________ 197

5.2.6 SCENARIOS SIMULATED ________________________________________________________ 198

5.2.7 PARAMETERS SELECTION _______________________________________________________ 198

5.2.8 EFFECTIVE DIFFUSION COEFFICIENT OF DBPS _______________________________________ 199

5.2.9 MASS TRANSPORT ____________________________________________________________ 201

5.2.10 SENSITIVITY ANALYSIS ________________________________________________________ 201

5.2.10.1 Parameter screening with Morris method _______________________________________ 202

5.2.10.2 Higher-order parameter interaction effects ______________________________________ 204

5.2.10.3 Implementation of Morris method _____________________________________________ 204

5.2.11 POST-PROCESSING __________________________________________________________ 205

5.3 RESULTS AND DISCUSSION _______________________________________________________ 205

5.3.1 TYPICAL MODEL BEHAVIOUR _____________________________________________________ 205

5.3.1.1 Flow field ________________________________________________________________ 205

5.3.1.2 Concentrations of substances in biofilm and bulk water _____________________________ 208

5.3.1.3 Fluxes of dissolved substances at biofilm surface _________________________________ 212

5.3.1.4 Variation of Sh
¯ at the biofilm surface ___________________________________________ 213

5.3.1.5 Concentrations of dissolved substances at the pipe outlet ___________________________ 213

5.3.2 INFLUENCE OF FLOW REGIME ON DBP FORMATION AND TRANSPORT _______________________ 214

5.3.3 INFLUENCE OF RE ON DBP TRANSPORT _____________________________________________ 216

Disinfection by-product formation from biofilm chlorination in drinking water pipes viii
Carolina Montoya Pachongo. School of Civil Engineering
5.3.3.1 Chloroform and dichloroacetonitrile formation ____________________________________ 216

5.3.3.2 Mass transport ____________________________________________________________ 222

5.3.4 INFLUENCE OF S/V RATIO ON DBP TRANSPORT _______________________________________ 225

5.3.4.1 Different flow rates and equal Re ______________________________________________ 225

5.3.4.2 Equal flow rates and different Re ______________________________________________ 227

5.3.5 DBP FORMATION UNDER SEVERAL CONDITIONS OF DRINKING WATER QUALITY ________________ 229

5.3.5.1 Transitional flow ___________________________________________________________ 229

5.3.5.2 Turbulent flow_____________________________________________________________ 232

5.3.6 DISCUSSION ________________________________________________________________ 234

5.3.7 SENSITIVITY ANALYSIS _________________________________________________________ 235

5.3.7.1Sh
¯ - chlorine ______________________________________________________________ 235

5.3.7.2Sh
¯ - DCAN _______________________________________________________________ 237

5.3.7.3 Average DCAN____________________________________________________________ 238

5.3.7.4 Median DCAN ____________________________________________________________ 240

5.3.7.5 Summary ________________________________________________________________ 241

5.4 MODEL APPLICATIONS AND LIMITATIONS _____________________________________________ 242

5.5 CONCLUSIONS ________________________________________________________________ 244

6 GENERAL DISCUSSION, IMPLICATIONS AND APPLICATIONS ________________________ 246

6.1 GENERAL DISCUSSION __________________________________________________________ 246

6.2 BACTERIAL COMMUNITIES AND PUBLIC HEALTH ________________________________________ 250

6.3 DBPS AND PUBLIC HEALTH_______________________________________________________ 252

6.4 IMPLICATIONS_________________________________________________________________ 258

6.4.1 IMPLICATIONS FOR WATER QUALITY IN CALI’S DWDN IN THE CONTEXT OF EXTREME WEATHER EVENTS_
_________________________________________________________________________ 258

6.4.2 IMPLICATIONS FOR O&M ACTIVITIES IN DWDN _______________________________________ 260

6.5 APPLICABILITY OF THE CURRENT RESEARCH PROJECT ___________________________________ 261

7 KEY FINDINGS, CONCLUSION AND FUTURE RESEARCH ____________________________ 263

7.1 GENERAL CONCLUSION __________________________________________________________ 263

7.2 KEY FINDINGS _________________________________________________________________ 263

Disinfection by-product formation from biofilm chlorination in drinking water pipes ix

Carolina Montoya Pachongo. School of Civil Engineering
7.2.1 BACTERIAL COMMUNITIES IN A TROPICAL-CLIMATE DWDN_______________________________ 263

7.2.2 RELATIONSHIPS BETWEEN ABIOTIC AND BIOTIC PARAMETERS IN A TROPICAL-WEATHER DWDN ____ 264

7.2.3 DBP FORMATION POTENTIALS FROM BIOFILM CHLORINATION UNDER STAGNATION CONDITIONS ____ 264

7.2.4 DBP FORMATION POTENTIALS FROM BIOFILM CHLORINATION UNDER HYDRODYNAMIC CONDITIONS _ 265

7.2.5 PRACTICAL RECOMMENDATIONS ON BIOFILM AND DISINFECTION BY-PRODUCTS CONTROL IN DRINKING

WATER NETWORKS __________________________________________________________________ 265

7.3 FUTURE RESEARCH __________________________________________________________ 266

8 REFERENCES _________________________________________________________________ 268

9 APPENDICES _________________________________________________________________ 290

Disinfection by-product formation from biofilm chlorination in drinking water pipes x

Carolina Montoya Pachongo. School of Civil Engineering
 LIST OF TABLES

Table 2-1. Characteristics of plumbing systems ..............................................................................................17

Table 2-2. Summary of bacterial composition in DWDNs supplied by surface water sources .....................28

Table 2-3. Summary of studies on biofilm modelling .......................................................................................34

Table 2-4. Disinfectant decay constants reported in experimental studies ....................................................48

Table 2-5. Distinctive functional groups of chlorination DBPs ........................................................................51

Table 2-6. Regulation of concentration limits of DBPs in drinking water ........................................................55

Table 2-7. Models for DBP formation ...............................................................................................................61

Table 2-8. Main features of the interaction between disinfectant and drinking water biofilms according to
experimental studies ..........................................................................................................................................75

Table 3-1. Water sources and treatment description of the water supply system of the city of Cali ............84

Table 3-2. Description of the code used to classify the type of suspension of drinking water supply in
Colombia ............................................................................................................................................................93

Table 3-3. Hydraulic conditions of the sub-network 4 during sampling campaign .........................................99

Table 3-4. Network characteristics in sampling points ................................................................................. 101

Table 3-5. Water quality, biotic parameters and descriptive statistics ......................................................... 102

Table 3-6. Spearman correlation coefficients for bulk water parameters .................................................... 103

Table 3-7. Spearman correlation coefficients for biofilm parameters .......................................................... 103

Table 3-8. ANOSIM test for RA of species .................................................................................................... 106

Table 3-9. Medians and means of richness and diversity indices ............................................................... 107

Table 4-1. Model equations in biofilm ............................................................................................................ 122

Table 4-2. Review of biofilm thickness data reported by drinking-water-related studies............................ 124

Table 4-3. Descriptive statistics of biofilm thickness reported in Table 4-2................................................. 126

Table 4-4. Review of cell density data in drinking water biofilm reported by other researchers ................ 127

Table 4-5. Descriptive statistics of cell density in drinking water biofilm reported by other researchers ... 129

Table 4-6. Values of reaction rates (k1 and k2) and yield coefficients (Y1 and Y2) ...................................... 131

Table 4-7. Initial and boundary conditions in 1D model................................................................................ 132

Table 4-8. Number of elements of every type of mesh................................................................................. 134

Table 4-9. Main features for model implementation in COMSOL Multiphysics 5.2a .................................. 136

Disinfection by-product formation from biofilm chlorination in drinking water pipes xi

Carolina Montoya Pachongo. School of Civil Engineering
Table 4-10. Experimental conditions and results of tests developed by Wang et al. (2013a) .................... 138

Table 4-11. Base parameters used in the sensitivity analysis ...................................................................... 142

Table 4-12. Percentage of change of parameters and the respective chloroform concentrations in bulk
water ................................................................................................................................................................. 152

Table 4-13. Percentage of change of parameters and the respective chloroform concentrations in bulk
water ................................................................................................................................................................. 153

Table 4-14. Parameters used to simulate DBP formation in plumbing systems ......................................... 157

Table 4-15. TTHMs and free chlorine concentrations measured in plumbing systems (Chowdhury, 2016)
.......................................................................................................................................................................... 158

Table 4-16. Chlorine demand on plumbing system scenarios ...................................................................... 163

Table 5-1. Mass balance equations and initial conditions in 2D model – Cylindrical pipes ........................ 175

Table 5-2. Boundary conditions for transport of dissolved substances in 2D model – Cylindrical pipes ... 178

Table 5-3. Navier-Stokes equations for incompressible flow ........................................................................ 179

Table 5-4. Boundary conditions for flow field in 2D model – Cylindrical pipes ............................................ 180

Table 5-5. Navier-Stokes equations for turbulent incompressible flow ........................................................ 181

Table 5-6. Equations of the turbulent model SST.......................................................................................... 183

Table 5-7. Geometry configurations tested in the current model.................................................................. 184

Table 5-8. Characteristics of the mapped mesh for pipe diameter of 10 inches ......................................... 186

Table 5-9. Characteristics of the mapped mesh for pipe diameter of 6 inches ........................................... 187

Table 5-10. Characteristics of the mapped mesh for pipe diameter of 3 inches ......................................... 187

Table 5-11. Description of five turbulence models included in COMSOL Multiphysics 5.2a ...................... 191

Table 5-12. Percentage of difference between pairs of turbulence model – Maximal velocity ................... 193

Table 5-13. Percentage of difference between pairs of turbulence model – Median velocity .................... 193

Table 5-14. Percentage of difference between pairs of turbulence model – Average velocity ................... 193

Table 5-15. Percentage of difference between COMSOL and Fluent for the same turbulence model –
Maximum velocity ............................................................................................................................................ 195

Table 5-16. Percentage of difference between COMSOL and Fluent for the same turbulence model –
Median velocity ................................................................................................................................................ 196

Table 5-17. Percentage of difference between COMSOL and Fluent for the same turbulence model –
Average velocity .............................................................................................................................................. 196

Table 5-18. Sh
¯ for different turbulent models ................................................................................................ 196

Table 5-19. Main features for model implementation in COMSOL Multiphysics 5.2a ................................. 197

Disinfection by-product formation from biofilm chlorination in drinking water pipes xii
Carolina Montoya Pachongo. School of Civil Engineering
Table 5-20. Parameters used in the 2D model simulations ......................................................................... 199

Table 5-21. Ranges of parameters considered for sensitivity analysis ....................................................... 205

Table 5-22. Sh ¯ and 𝒌𝒇 for dissolved substances for two scenarios of drinking water quality – Transitional
flow ................................................................................................................................................................... 230

Table 5-23. Sh ¯ and 𝒌𝒇 for dissolved substances for two scenarios of drinking water quality – Turbulent
flow ................................................................................................................................................................... 233

Table 5-24. Median and standard deviation of EEs on response variable Sh

¯ - Chlorine ........................... 236

Table 5-25. Median and standard deviation of EEs on response variable Sh

¯ - DCAN .............................. 237

Table 5-26. Median and standard deviation of EEs on response variable average DCAN........................ 239

Table 5-27. Median and standard deviation of EEs on response variable median DCAN ......................... 240

Table 5-28. Ranking of parameters according to their influence on response variables............................ 242

Table 6-1. Potential pathogenic bacteria identified in Cali’s DWDN and their corresponding risk area
defined by Pérez-Vidal et al. (2012) .............................................................................................................. 251

Table 6-2. Main results of 1D model of potential of DBP formation from biofilm chlorination .................... 253

Table 6-3. Main results of 2D model of potential of DBP formation from biofilm chlorination .................... 255

Disinfection by-product formation from biofilm chlorination in drinking water pipes xiii
Carolina Montoya Pachongo. School of Civil Engineering
 LIST OF FIGURES

Figure 1-1. Chemical and biological processes occurring in water pipes ........................................................ 3

Figure 1-2. Issues arising in DWDNs ................................................................................................................. 5

Figure 1-3. Methodological approach of the current research project.............................................................. 9

Figure 2-1. Schematic representation of mass flux at a surface with a unit normal n and arbitrary
orientation .......................................................................................................................................................... 19

Figure 3-1. Location of the study area ............................................................................................................. 83

Figure 3-2. Fractions of pipe materials in the DWDN of the city of Cali ......................................................... 84

Figure 3-3. Water age ranges for the sampling sub-network.......................................................................... 86

Figure 3-4. Procedure for collection of water and biofilm samples................................................................. 87

Figure 3-5. Location of sampling points in the city of Cali............................................................................... 88

Figure 3-6. Inner walls of cast iron pipeline of sampling point 2 ..................................................................... 90

Figure 3-7. Stages in the multivariate analysis based on similarity coefficients ............................................ 92

Figure 3-8. Characterization of events of suspension of drinking water supply in the DWDN of the city of
Cali ..................................................................................................................................................................... 94

Figure 3-9. Evolution of events of suspension of WTP operation in the period 2000-2016.......................... 95

Figure 3-10. Water levels in service reservoirs during sampling activities (a) La Campiña (b) Normal (c)
Siloé (d) Nápoles (e) Ciudad Jardín ................................................................................................................. 97

Figure 3-11. Water age of sampling points .................................................................................................... 100

Figure 3-12. RA of bacterial groups to (a) phylum level and (b) genus level in water samples.................. 104

Figure 3-13. RA of bacterial groups to (a) phylum level and (b) genus level in biofilm samples ................ 105

Figure 3-14. Non-metric MDS analysis of bacterial RA. Factors (a) Habitat and (b) and pipe material -
biofilm samples ................................................................................................................................................ 106

Figure 4-1. Schematic representation of the 1D model................................................................................. 120

Figure 4-2. 1D mass balance equations for soluble and particulate components in biofilm and biofilm
surface.............................................................................................................................................................. 122

Figure 4-3. Box plot biofilm thickness data .................................................................................................... 126

Figure 4-4. Box plot cell density data ............................................................................................................. 129

Figure 4-5. Linear correlation between biofilm thickness and EPS concentration (Celmer et al., 2008) ... 130

Figure 4-6. Representation of the 1D model mesh ....................................................................................... 133

Figure 4-7. Influence of grid size on simulation results ................................................................................. 134

Disinfection by-product formation from biofilm chlorination in drinking water pipes xiv
Carolina Montoya Pachongo. School of Civil Engineering
Figure 4-8. Time step comparison among chloroform concentrations in bulk water .................................. 137

Figure 4-9. Dissolved substances within biofilm (a) Chlorine (b) Chloroform ............................................. 139

Figure 4-10. Dissolved substances in bulk water after 3 hours ................................................................... 140

Figure 4-11. Decay of particulate substances (a) Cells (b) EPS.................................................................. 140

Figure 4-12. Comparison of chlorine penetration within biofilm (a) Chen and Stewart (1996): experimental
vs model data (b) Experimental data from Chen and Stewart (1996) vs simulated data from current model
......................................................................................................................................................................... 142

Figure 4-13. Influence of S on chloroform concentration in bulk water (a) Chloroform concentrations in
bulk water at different values of S after t = 3 h (b) Temporal variation of chloroform concentrations in bulk
water for three values of S ............................................................................................................................ 144

Figure 4-14. Influence of reaction rate k1 on chloroform concentration in bulk water (a) Chloroform
concentrations in bulk water at different values of k1 after t = 3 h (b) Temporal variation of chloroform
concentrations in bulk water for several values of k1 .................................................................................... 145

Figure 4-15. Influence of reaction rate k2 on chloroform concentration in bulk water (a) Chloroform
concentrations in bulk water at different values of k2 after t = 3 h (b) Temporal variation of chloroform
concentrations in bulk water for several values of k2 .................................................................................... 146

Figure 4-16. Influence of initial chlorine concentration on chloroform concentration in bulk water (a)
Comparison between Clo and minimum cell concentration (b) Comparison between Clo and minimum EPS
concentration ................................................................................................................................................... 147

Figure 4-17. Temperature influence on chloroform concentration in bulk water (a) Chloroform
concentrations in bulk water at different temperatures after t = 3 h (b) Temporal variation of chloroform
concentrations in bulk water at the temperature range 5-25 °C .................................................................. 148

Figure 4-18. Influence of (a) Y1, (b) Y2, and (c) reactor volume on chloroform concentration in bulk water
after t = 3 h ...................................................................................................................................................... 149

Figure 4-19. Parameters with linear influence on chloroform concentration in bulk water (a) Biofilm
thickness (b) Initial cell concentration (c) Initial EPS concentration (d) FX (e) FE (f) Biofilm area .............. 150

Figure 4-20. Comparison of influence on chloroform concentration in bulk water among parameters (a) 14
parameters tested (b) 12 parameters tested excluding V and Y1 ................................................................ 151

Figure 4-21. (a) Chloroform and (b) DCAN concentrations according to results of Wang et al. (2013a) .. 156

Figure 4-22. Chloroform concentrations in bulk water for different pipe diameter ...................................... 158

Figure 4-23. Chloroform concentrations in bulk water for several scenarios of initial chlorine and cell
concentration (a) BT = 7 m (b) BT = 102 m ............................................................................................. 159

Figure 4-24. DCAN concentrations in bulk water for several scenarios of initial chlorine and cell
concentration (a) BT = 7 m (b) BT = 102 m ............................................................................................. 162

Figure 4-25. Chlorine concentrations in bulk water, under several scenarios of initial chlorine and cell
concentration (a) BT = 7 m (b) BT = 102 m ............................................................................................. 163

Figure 4-26. Influence of S/V ratio on DBP concentrations in bulk water (a) Chloroform (b) DCAN ......... 164

Disinfection by-product formation from biofilm chlorination in drinking water pipes xv

Carolina Montoya Pachongo. School of Civil Engineering
Figure 4-27. DBP concentrations in bulk water and regulation limits / guidelines (a) Chloroform, Clo = 1.66
mg/L, Xo = 4.44x10-3 mg/cm2 (b) DCAN, Clo = 1.66 mg/L, Xo = 1.28x10-4 mg/cm2 ................................... 165

Figure 4-28. Cell concentration according to pipe diameter ......................................................................... 166

Figure 5-1. Schematic representation of the 2D model................................................................................. 171

Figure 5-2. Process for building a flow-coupled model to predict DBP formation potentials from chlorination
of biofilms ......................................................................................................................................................... 172

Figure 5-3. 2D mass balance equations for soluble and particulate components in biofilm, biofilm surface,
and bulk water.................................................................................................................................................. 174

Figure 5-4. Geometry of the water pipe ......................................................................................................... 185

Figure 5-5. Mesh details of the water pipe ..................................................................................................... 185

Figure 5-6. Comparison of velocity and Sh

¯ between mesh 1 and 2 (a) Velocity monitored in the centre of
bulk water domain along r axis (b) Sh
¯ for chlorine (c) Sh
¯ for chloroform ..................................................... 189

Figure 5-7. Comparison of Sh ¯ (a) and average chloroform concentration at the pipe outlet (b) in mesh 1
.......................................................................................................................................................................... 190

Figure 5-8. Regimes of the turbulent flow near to a flat wall ......................................................................... 191

Figure 5-9. Comparison of five turbulence models included in COMSOL Multiphysics 5.2a (a) Velocity
monitored at the middle of bulk water domain (b) Average chloroform concentrations at the pipe outlet . 193

Figure 5-10. Comparison of velocity monitored at the middle of bulk water domain between software
COMSOL and Fluent (a) Laminar and SST (b) SST ..................................................................................... 195

Figure 5-11. Influence of S on (a) chloroform concentrations in bulk water at the pipe outlet and (b) Sh ¯
.......................................................................................................................................................................... 200

Figure 5-12. Influence of S on (a) DCAN concentrations in bulk water at the pipe outlet and (b) Sh
¯ ...... 201

Figure 5-13. Pressure (Pa) surface plot – Inlet: left; Outlet: right (a) Laminar flow (b) Transitional flow (c)
Turbulent flow .................................................................................................................................................. 206

Figure 5-14. Velocity profiles (a) Laminar flow (b) Transitional flow (c) Turbulent flow .............................. 207

Figure 5-15. Sketch of the non-scaled geometry........................................................................................... 208

Figure 5-16. Contour plots of substance concentrations in biofilm – BT = 102 m, Clo = 1.66 mg/L, Xo =
2.33 mg/L , Re = 2774,  = 3”......................................................................................................................... 210

Figure 5-17. Contour plots of concentrations of dissolved substances in bulk water - BT = 102 m, Clo =
1.66 mg/L, Xo = 2.33 mg/L , Re = 2774,  = 3” ............................ 211

Figure 5-18. Flux of dissolved substances at the biofilm surface (a) DCAN (b) Chlorine ........................... 212

Figure 5-19. Variation of Sh

¯ at the pipe outlet ............................................................................................... 213

Figure 5-20. Average concentration of chlorine and DCAN at the pipe outlet ............................................. 213

Figure 5-21. Average concentration of chloroform at the pipe outlet according to flow regime (a)
Chloroform (b) DCAN ...................................................................................................................................... 214

Disinfection by-product formation from biofilm chlorination in drinking water pipes xvi
Carolina Montoya Pachongo. School of Civil Engineering
Figure 5-22. Variation of local Sh according to flow regime (a) DBPs (b) Chlorine .................................... 214

Figure 5-23. Average chloroform concentration at the pipe outlet for several Re – Transitional flow (a) BT =
7 m (b) BT = 102 m.................................................................................................................................... 216

Figure 5-24. Average DCAN concentration at the pipe outlet for several Re – Transitional flow (a) BT = 7
m (b) BT = 102 m ....................................................................................................................................... 217

Figure 5-25. Average chlorine concentration at the pipe outlet for several Re – Transitional flow (a) BT = 7
m (b) BT = 102 m ....................................................................................................................................... 218

Figure 5-26. Average chloroform concentration at the pipe outlet for several Re – Turbulent flow (a) BT = 7
m (b) BT = 102 m ....................................................................................................................................... 218

Figure 5-27. Average DCAN concentration at the pipe outlet for several Re – Turbulent flow (a) BT = 7 m
(b) BT = 102 m.............................................................................................................................................. 219

Figure 5-28. Average chlorine concentration at the pipe outlet for several Re – Turbulent flow (a) BT = 7
m (b) BT = 102 m ....................................................................................................................................... 220

Figure 5-29. Sh
¯ (a) and mass transfer rate (b) of chloroform for several Re – Transitional flow ............... 222

Figure 5-30. Sh
¯ (a) and mass transfer rate (b) of chloroform for several Re – Turbulent flow .................. 222

Figure 5-31. Sh
¯ (a) and mass transfer rate (b) of DCAN for several Re – Transitional flow ...................... 223

Figure 5-32. Sh
¯ (a) and mass transfer rate (b) of DCAN for several Re – Turbulent flow ......................... 223

Figure 5-33. Sh
¯ (a) and mass transfer rate (b) of chlorine for several Re – Transitional flow ................... 223

Figure 5-34. Sh
¯ (a) and mass transfer rate (b) of chlorine for several Re – Turbulent flow ....................... 224

Figure 5-35. Average chloroform (a) and DCAN (b) concentrations at the pipe outlet for several pipe
diameters – Transitional flow.......................................................................................................................... 226

Figure 5-36. Average chlorine concentrations at the pipe outlet for several pipe diameters – Transitional
flow ................................................................................................................................................................... 226

Figure 5-37. Average Sh

¯ for DBPs (a) and Chlorine (b) for several pipe diameters – Transitional flow ... 227

Figure 5-38. Average chloroform (a) and DCAN (b) concentrations at the pipe outlet for several pipe
diameters – Turbulent flow ............................................................................................................................. 228

Figure 5-39. Average chlorine concentrations at the pipe outlet for several pipe diameters – Turbulent flow
......................................................................................................................................................................... 228

Figure 5-40. Average Sh

¯ for DBPs (a) and Chlorine (b) for several pipe diameters – Turbulent flow ....... 228

Figure 5-41. Average concentrations of (a) chloroform and (b) DCAN at the pipe outlet for two scenarios of
drinking water quality – Transitional flow ....................................................................................................... 229

Figure 5-42. Flux of DCAN at the biofilm surface for two scenarios of drinking water quality (a) Least
critical scenario (b) Most critical scenario ...................................................................................................... 231

Figure 5-43. Flux of chlorine at the biofilm surface for two scenarios of drinking water quality – Transitional
flow (a) Least critical scenario (b) Most critical scenario .............................................................................. 232

Disinfection by-product formation from biofilm chlorination in drinking water pipes xvii
Carolina Montoya Pachongo. School of Civil Engineering
Figure 5-44. Average concentrations of (a) chloroform and (b) DCAN at the pipe outlet for two scenarios of
drinking water quality – Turbulent flow ........................................................................................................... 233

Figure 5-45. EEs on response variable Sh

¯ - Chlorine (a) Median vs Standard deviation (b) Median vs
Standard deviation including upper and lower bounds ................................................................................. 237

Figure 5-46. EEs on response variable Sh

¯ - DCAN (a) Median vs Standard deviation (b) Median vs
Standard deviation including upper and lower bounds ................................................................................. 238

Figure 5-47. EEs on response variable average DCAN (a) Median vs Standard deviation (b) Median vs
Standard deviation including upper and lower bounds ................................................................................. 239

Figure 5-48. EEs on response variable median DCAN (a) Median vs Standard deviation (b) Median vs
Standard deviation including upper and lower bounds ................................................................................. 241

Figure 6-1. Dual role of biofilms studied in the current research project ...................................................... 246

Figure 6-2. Two approaches to study the microbial and chemical significance of biofilms in drinking water
.......................................................................................................................................................................... 247

Figure 9-1. Water age for subnetwork 4 at the DWDN of the city of Cali - Colombia ................................. 290

Figure 9-2. Bray Curtis similarities of the relative abundance percentage of species in biofilm samples –
Water species .................................................................................................................................................. 295

Figure 9-3. Bray Curtis similarities of the relative abundance percentage of species in biofilm samples –
Biofilm species ................................................................................................................................................. 298

Figure 9-4. Bray Curtis similarities of the relative abundance percentage of species in biofilm samples –
Water and biofilm species together ................................................................................................................ 300

LIST OF APPENDICES

Appendix 9-A. Water age map........................................................................................................................ 290

Appendix 9-B. Normality tests of correlated parameters .............................................................................. 291

Appendix 9-C. Two-dimensional plots of the Multi-Dimensional Scaling (MDS) analysis and dendrograms
for visualization of Bray Curtis similarity index............................................................................................... 292

Disinfection by-product formation from biofilm chlorination in drinking water pipes xviii
Carolina Montoya Pachongo. School of Civil Engineering
 LIST OF ABBREVIATIONS

1D One dimension

2D Two dimensions

AMD Advanced micro devices

ANOSIM Analysis of similarities

CCP Concrete cylinder pipe

Centres for disease control and

CDC
prevention

CPU Central processing unit

CV Coefficient of variation

DBP Disinfection by-product

DCAN Dichloroacetonitrile

DICL Ductile iron cement-lined

DNA Desoxyribose nucleic acid

DPD N,N diethyl-p-phenylene diamine

DWDN Drinking water distribution network

EE Elementary effects

EPS Extracellular polymeric substances

GPU Graphics processing unit

HAA Haloacetic acid

HOCl Hypochlorous acid

HPC Heterotrophic plate count

IRB Iron-reducing bacteria

MDPE Medium density polyethylene

MDS Multidimensional scaling

NOM Natural organic matter

NTU Nephelometric turbidity units

Disinfection by-product formation from biofilm chlorination in drinking water pipes xix
Carolina Montoya Pachongo. School of Civil Engineering
 O&M Operation and maintenance

OTU Operational taxonomic unit

PCR Polymerase chain reaction

PE Polyethylene

PVC Polyvinyl chloride

Q1 Quartile 1

Q3 Quartile 3

QAC Quaternary ammonium chloride

R Hydrocarbon

R2 Different chemical combinations

RA Relative abundance

RAM Random access memory

RNA Ribose nucleic acid

S/V Pipe surface area to volume ratio

Sh Local Sherwood number

SRB Sulphate-reducing bacteria

SST Shear stress transport

SUVA Specific ultraviolet absorbance

THM Trihalomethane

TOC Total organic carbon

TTHM Total trihalomethanes

UK United Kingdom

USA United States of America

UV254 Ultraviolet absorbance

WTP Water treatment plant

Disinfection by-product formation from biofilm chlorination in drinking water pipes xx

Carolina Montoya Pachongo. School of Civil Engineering
 NOMENCLATURE

A Parameter empirically determined

Ab Biofilm area (cm2) (m2)

B Parameter empirically determined

BT Biofilm thickness (m)

C Chlorine concentration (mg/L)

Co | Clo Initial concentration of chlorine (mg/L)

d Length scale

DC-B Molecular diffusion coefficient of chlorine in biofilm (m2/sec)

DS-B Diffusion coefficient of chloroform/DCAN in in biofilm (m2/sec)

E EPS concentration (mg/L)

Eo Initial EPS concentration (mg/L)

FE Fraction of EPS transformed into chloroform (dimensionless)

FX Fraction of cells transformed into chloroform (dimensionless)

k Thermal conductivity | Chlorine decay constant

k1 Disinfectant decay rate – cells disinfection (m3/g-sec)

k2 Disinfectant decay rate – EPS oxidation (m3/g-sec)

kb Reaction coefficient in bulk water

kf Mass transfer rate

̅̅̅
𝑘𝑓 Global mass transfer rate

kw Reaction coefficient at pipe walls

M Parameter empirically determined

MR Molecular mass of DBP (g/mol)

n Parameter empirically determined

Q Flow rate

Re Reynolds number

rh Hydraulic radius

Disinfection by-product formation from biofilm chlorination in drinking water pipes xxi
Carolina Montoya Pachongo. School of Civil Engineering
 S Chloroform/DCAN concentration (g/L)

Sc Schmidt number

Sh Sherwood number

Sh
¯ Global Sherwood number

T Temperature (°C)

t Time

Uo Velocity magnitude

𝑢
⃗ Velocity vector

V Reactor volume (mL) (m3)

X Cells concentration (mg/L)

Xo Initial cell concentration (mg/L)

Y1 Cell-chlorine yield coefficient (g/g)

Y2 EPS-chlorine yield coefficient (g/g)

 Pipe diameter

C Biofilm-bulk water diffusion coefficient ratio of chlorine (dimensionless)

S Biofilm-bulk water diffusion coefficient ratio of DBP (dimensionless)

Subscripts

b Biofilm
Bs Biofilm surface
C Chlorine
E EPS
L Bulk liquid
S DBP
w Wall
X Cells

Disinfection by-product formation from biofilm chlorination in drinking water pipes xxii
Carolina Montoya Pachongo. School of Civil Engineering
 1 INTRODUCTION

1.1 MOTIVATION

The supply of water and sanitation services is one of the most effective strategies for reducing
poverty around the world. The Sustainable Development Goals from the United Nations
Organization include in the goal number 6: “Ensure availability and sustainable management of
water and sanitation for all” (United Nations, 2015). However, preserving drinking water to
consumers represents a challenge for water utilities, since technical and economic resources,
appropriate knowledge and training, and adequate operation and maintenance (O&M) programmes
are required to achieve this purpose. To respond to this challenge in the drinking water distribution
networks (DWDNs) is an enormous task because several phenomena occur in the pipes such as
biofilm growth, disinfection by-products (DBPs) formation, sedimentation, corrosion, discolouration,
among others. Therefore, efforts must be addressed to minimize such phenomena and to find the
balance between acute and chronic risk in drinking water supply.

Once raw water is treated in water treatment plants (WTPs) for human consumption, DWDNs are
used to transport drinking water to consumers. DWDNs are mainly composed by pipes, fittings,
storage facilities, and back-flow prevention devices (NRCNA, 2006). In drinking water systems,
treated water at the outlet of the WTPs is not sterile and still contains cells and organic matter.
Then, microorganisms develop biofilms as a strategy to survive in DWDNs. Biofilms are habitats
fixed to wet surfaces where live bacteria, archaea, fungi, and protozoa. The ability of some
microorganisms to produce Extracellular Polymeric Substances (EPS) enables their attachment to
surfaces in hostile environments (e.g., low nutrient concentrations, bulk flow, and presence of
disinfectant). Other organisms can also adhere to the biofilm once this started forming (Wang et
al., 2014; Fish et al., 2015; Srivastava and Bhargava, 2015). DBPs are the result of the reaction
between disinfectants and natural organic matter (NOM), biofilms, anthropogenic contaminants,
bromide, and iodide during the production and/or distribution of drinking water. Such reaction is
driven by several factors such as pH, temperature, type and concentration of precursor and
disinfectant, and contact time (Chowdhury et al., 2009).

Biofilm and DBP formation has become a major concern in the operation of DWDNs since they can
cause acute and chronic effects on human health. Biofilms can lead to corrosion, discoloured
waters and odours; host pathogens which may be released to bulk water; detach and recolonize
clean surfaces; act as precursors for the formation of DBPs, and consequently, contribute to
disinfectant decay (NRCNA, 2006). In this line, biofilm modelling has shown a further development
Chapter 1. Introduction 1
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
in order to better understand phenomena which allow attachment, growing, detachment, and
survival of cells, considering hydraulic and molecular conditions (Wang and Zhang, 2010a).

On the other hand, it is widely accepted that DBPs have negative effects on human health as they
have been identified as carcinogenic, teratogenic and mutagenic substances (WHO, 2008). The
multiple factors involved in reactions between disinfectants and organic matter and the diverse
nature of precursors has resulted in numerous DBP species; to date, 600 species have been
identified (Hrudey, 2009). From those, chloroform and dichloroacetonitrile (DCAN) are approached
in this project because the first DBP is the most abundant substance in the group trihalomethanes
(THMs), which is one of the few DBP groups regulated around the world (Richardson et al., 2007).
The second DBP is the most abundant substance in the group haloacetonitriles (HANs), which has
not regulated yet and experimental tests on mice has shown that they are more toxic than regulated
DBPs such as haloacetic acids (HAAs) (Muellner et al., 2007). In this line, DBP formation modelling
has been developed to predict their concentrations by empirical models according to different
environmental, hydraulic and chemical variables such as pH, temperature, residence time,
disinfectant and bromide concentrations, and type and amount of precursors (Chowdhury et al.,
2009).

Biological and chemical processes related to biofilms in drinking water pipes are described in Figure
1-1. Recently, DBP formation in DWDNs has also been attributed to the reaction of disinfectants
with biofilm matrix, which is mainly composed by intracellular organic matter, EPS, and other
biomolecules (Wang et al., 2012c; Wang et al., 2013a; Xue et al., 2014). Therefore, recent
experimental studies have focussed on the formation of DBPs from pure cells and heterogeneous
biofilms, planktonic cells, and extracted EPS (Hong et al., 2008; Fang et al., 2010a; Fang et al.,
2010b; Wang et al., 2012c; Wang et al., 2013a; Wang et al., 2013b). Considering that biofilms act
as DBP precursors and that most of the empirical models developed to date do not include the
contribution of biomass to the bulk water concentrations of these substances, further research is
required to define the role of drinking water biofilms into the formation of DBPs in distribution
networks. The capacity of biofilms to act as both pathogen reservoir and DBP precursor was the
driver behind this research.

Chapter 1. Introduction 2
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Adapted from Prest et al. (2016)
Figure 1-1. Chemical and biological processes occurring in water pipes

1.2 PROBLEM DEFINITION

Tropical climate is characterised by average temperatures higher than 18 °C, high precipitation,
and then high humidity. In contrast to temperate weather, there are not seasons in tropical climate
regions but periods of higher or lower precipitation (The British Geographer, 2017). Tropical climate
region is located between the Tropic of Cancer and the Tropic of Capricorn (The British
Geographer, 2017); it means Latino American, African and Pacific Ocean countries are located in
this region. Furthermore, climate change has significantly impacted water availability, since
extreme events as droughts and floods are more frequent and more severe in different regions of
the world (Delpla et al., 2009). Then, it is important to highlight that less developed countries are
more vulnerable to the consequences of extreme weather events but developed regions are also
susceptible to them (Cann et al., 2012; Rataj et al., 2016). Thus, water supply in the developing
world may represent a challenge under extreme weather conditions and economical limitations.
Several factors can or may affect drinking water quality; the following describes some of them.

According to Delpla et al. (2009), surface water quality has been especially affected due to
temperature increase and the alteration of almost all physico-chemical equilibriums and biological
reactions; therefore, rise of organic and inorganic micro-pollutants, dissolved organic matter,
nutrients and pathogens is expected. Climate change could impact drinking water quality because
of the rise of DBP concentrations with increasing water temperature; and elevated levels of turbidity
and organic matter found in river waters during rainy seasons in tropical countries may deteriorate
treatment performance, demanding more sophisticated technologies and, therefore, more

Chapter 1. Introduction 3
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
economical resources (Dearmont et al., 1998; Yorkshire Water, 2013; Hutton and Varughese,
2016).

In particular, inefficient performance of water treatment processes could lead to solids accumulation
and biofilm and DBP formation in areas of DWDNs where flow velocity is low (end zones and
storage/distribution tanks). Wen et al. (2014) showed that improving coagulation can restrict
microbial regrowth in tap water by phosphate elimination in water treatment. This was concluded
by testing conventional treatment with three different coagulants and enumerating microbes by flow
cytometry. Microbial regrowth potential was identified by inoculation of bottled mineral water of
filtered de-chlorinated water and addition of nutrients such as phosphate and carbon. Their results
showed that coagulation was the crucial step for decreasing the microbial re-growth potential and
producing phosphorous-limited water. Indeed, it is known that limiting nutrient content in treated
water contributes to achieve bio-stability of the water. This procedure is used to limit carbon
concentrations in mid-northern Europe (Hammes et al., 2010) and phosphate concentrations in
northern Europe (Miettinen et al., 1997).

In addition, events which compromise hydraulic and physical integrity of networks like water supply
interruption, leakages repairs, and infrastructure replacements also promote solid accumulation
and increase of nutrient concentrations (Propato and Uber, 2004; NRCNA, 2006; Vreeburg and
Boxall, 2007). Although some authors argue that limiting particle accumulation and reducing
assimilable organic carbon are the best way to preserve drinking water quality in DWDNs (Brettar
and Höfle, 2008; Liu et al., 2013), other considerations like high variability of quality of surface water
sources, improved performance of WTPs, loss of the physical integrity of DWDNs, and the ability
of bacteria to grow in oligotrophic environments (Simões and Simões, 2013) should be recognized
in planning and O&M of drinking water supply systems.

Water treatment, including disinfection, is not completely effective to remove all biomass present
in raw water and, as a result, treated water still contains viable microorganisms. Low velocities, low
residual disinfectant concentrations, presence of nutrients, and biofilm-former microorganisms in
bulk water together promote biofilm formation in all the components of a DWDN. Biofilm represents
the large fraction of total biomass in drinking water (Abe et al., 2012). Figure 1-2 shows the existing
relationships between factors that can lead to acute and chronic risk for consumers. Such risk is
associated with bacterial re-growth and DBP formation. In addition, customer complaints to water
utilities and regulatory agencies can occur due to changes in aesthetic characteristics and
intermittence in the service (e.g., discoloured water, broken pipes). In summary, drinking water
quality can deteriorate when the physical barrier (physical integrity) of the DWDNs is broken,
Chapter 1. Introduction 4
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
hydraulic changes occurs (hydraulic integrity), and physico-chemical and biological processes
occur within the water pipes (water quality integrity).

Figure 1-2. Issues arising in DWDNs

Precisely, biofilms are a concern because they can promote corrosion of pipes, discoloured waters
and odours; host pathogens which may be released to bulk water; detach and recolonize clean
surfaces; be precursors of DBPs and, consequently, causing disinfectant decay (Lehtola et al.,
2004; Berry et al., 2006; Wang et al., 2012a; Pu et al., 2013; Wang et al., 2013a; Sun et al., 2014;
Xue et al., 2014). Biofilm control has been oriented toward application of disinfection strategies
(Berry et al., 2006; Simões and Simões, 2013) due to its role as pathogen reservoir (Wingender
and Flemming, 2011). Several studies have investigated the effects of disinfectants on the structure
of biofilms and survival of cells (Codony et al., 2005; Gagnon et al., 2005; Roeder et al., 2010).
Even though there is evidence of DBP formation from the disinfection of biofilms and that EPS exert
a protection barrier against disinfectants, recent studies still recommend to adjust operation
parameters for disinfection considering appropriate doses of disinfectant in order to inactivate
efficiently both biofilm and detached clusters (Xue et al., 2012; Xue et al., 2014; Prest et al., 2016).

Drinking water quality in distribution networks is frequently threatened by several factors such as
sudden variations of raw water quality, inefficient treatment, intrusion phenomenon, and routine
maintenance practices (Figure 1-2). In order to preserve drinking water quality in the DWDNs, water
Chapter 1. Introduction 5
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
operators carry out regular practices such as flushing and pipeline cleaning, which are known as
the simplest and most cost effective ways of reducing the risk of discolouration (Vreeburg and
Boxall, 2007). The objective of flushing pipes is to remove, in a controlled way, soft deposits from
a sector of distribution network; flushing can be effective in soft deposit removal in cast iron pipes
(Lehtola et al., 2004; Vreeburg et al., 2008) but relatively ineffective in achieving biofilms
detachment (Abe et al., 2012; Fish et al., 2012).

DBP formation is another concern in the control of water quality in DWDNs because they have a
negative effect on human health as they have been identified as potential carcinogenic, teratogenic
and mutagenic substances (WHO, 2008). DBPs are the result of the reaction between oxidant
substances, halogens and organic matter. There are more than 600 species identified to date
(Hrudey, 2009), and they are commonly classified into two groups: carbonaceous and nitrogenous
DBPs. The highest occurrence of DBPs is for chlorate, followed by chloroform, dichloroacetic acid,
dibromoacetic acid, thichloroacetic acid, bromate, and chlorite (Richardson et al., 2007).

Despite of the wide variety of DBP species, only THMs and HAAs (carbonaceous DBPs) are the
regulated groups around the world (Richardson et al., 2007). DBP formation starts, under the
presence of disinfectants, in water treatment and continues during distribution and storage,
increasing with water age. This principle applies for THMs but not for unstable DBPs such as HAAs,
since some species of this group are biodegradable and their concentrations can be lower at points
where biological activity is high within DWDNs, such as areas with high residence time and low
chlorine residuals (Tung and Xie, 2009). Similarly, DCAN is an unstable DBP species, which
degrades under certain conditions of free chorine and pH. DCAN degrades in the absence of
chlorine above pH 7and below pH 6.5. In the presence of free chlorine, DCAN degradation can be
much faster in the condition of a pH of about 6–8.5 under low to moderate chlorine residuals
(Reckhow et al., 2001).

Considering that biofilms are present in every component of DWDNs and are DBP precursors; a
particular interest has grown in the recent years in relation to DBP formation from disinfection of
pure cells and heterogeneous biofilms, planktonic cells, and extracted EPS from pure cells. For
example, studies of disinfection of microbial biomass from cyanobacteria and green microalgae
(i.e., highly relevant in tropical climate countries) were undertaken to determine the yield of THMs
and HAAs (Hong et al., 2008; Fang et al., 2010a; Fang et al., 2010b; Pu et al., 2013; Xie et al.,
2013), focusing on disinfection of surface water from reservoirs and water in pools and cooling
towers.

Chapter 1. Introduction 6
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
In addition, EPS from planktonic pure cells, biofilm pure cells, and isolated drinking water biofilms
have been tested after disinfection with chlorine and/or chloramines to quantify carbonaceous and
nitrogenous DBPs (Wang et al., 2012c; Wang et al., 2013a; Wang et al., 2013b). Another focus of
research has been developed by Yang et al. (2010) and Huang et al. (2012) towards quantification
of nitrogenous DBPs from nitrogenous organic compounds since these DBPs are considered more
toxic than carbonaceous DBPs (Muellner et al., 2007; Plewa et al., 2008). Doederer et al. (2014)
also investigated this topic on disinfection of treated wastewater.

These studies were based on laboratory tests and their results demonstrate that biofilms contribute
significantly to the DBP formation in disinfection processes. However, more information is still
required to determine parameters which allow mathematical simulation of this phenomenon in order
to apply it to experiment design and DWDN operation. It is recognized that modelling could
represent a less expensive tool when compared to experimental/field work and it allows testing
different scenarios to predict and optimize processes. Finally, it is worth noting that biofilm and DBP
modelling has been approached separately, resulting in the need for improving the development of
DBP models that take into account the role of biofilms as DBP precursors.

1.3 RESEARCH QUESTIONS, AIM AND OBJECTIVES

Following the above discussion, the current document corresponds to a research project aimed to
improve the understanding of biofilms presence in drinking water pipes by conducting a field work
in a tropical climate country and developing a flow-coupled mathematical model to predict the
contribution of biofilms to the DBP concentration in bulk water. This project addressed the gaps in
knowledge related to (i) the type of bacterial communities present in real scale DWDNs, especially
in tropical climates, and (ii) limited previous models that couples biofilms and DBPs together,
principally under transitional and turbulent flow conditions.

1.3.1 Research questions

 Is there any correlation between hydraulic factors, water quality, biofilm properties, bacterial
communities and total THMs concentrations in a tropical-climate DWDN?
 What are the bacterial communities present in bulk water and biofilms of a tropical-climate
DWDN?
 What are the main parameters influencing the formation of DBPs under stagnation conditions?

Chapter 1. Introduction 7
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Can the interaction between these parameters be quantified?
 How flow regime and pipe surface/volume ratio and biofilm characteristics affect the DBP
formation in drinking water pipes?
 Which are the practical recommendations for O&M of DWDNs in order to reduce the formation
of DBPs, related to biofilms?

1.3.2 Aim

The aim of this project is to better understand the impacts on water quality from the presence of
biofilms in DWDNs in tropical climate countries, with particular emphasis on their dual role as
pathogen reservoirs and DBPs precursors, under transitional and turbulent flow conditions.

1.3.3 Objectives

1. To understand the current knowledge on drinking water quality associated with biofilms and
DBPs, microbiological methods, and modelling.
2. To identify the bacterial communities existing in biofilm and bulk water of a tropical-climate
DWDN.
3. To identify the interrelationship between hydraulic, physical, chemical and biological factors in
a tropical-climate DWDN.
4. To build a simple biofilm model in order to understand the basis of modelling the chlorination of
biofilms, and the chloroform and DCAN formation potentials, under stagnation conditions.
5. To develop a flow-coupled model to predict chloroform and DCAN formation from chlorination
of biofilms in drinking water pipes.
6. To provide practical recommendations on biofilm and DBPs control in drinking water networks,
with a particular focus on tropical climate.

1.4 GENERAL METHODOLOGY OF THE RESEARCH PROJECT

The general methodological approach of the current research project is described in Figure 1-3.
The starting point of this study was an extensive literature review to identify the gap on the study
of biofilms and DBPs in drinking water. Then, the problem, research questions, aim, and objectives
were stablished. In order to achieve such objectives, a field work and the development of one-
dimensional (1D) and two-dimensional (2D) models were carried out. The results of such work were

Chapter 1. Introduction 8
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
analysed and discussed in order to improve the understanding of the impact of the presence of
biofilms and prediction of DBPs in DWDNs.

Figure 1-3. Methodological approach of the current research project

1.4.1 Structure of the thesis

The current thesis is structured as it follows:

 Chapter 2: Literature review. This chapter explains the basic concepts related to this study
such as physical, hydraulic, and water quality integrity in DWDNs; water disinfection, biofilms,
and DBPs. The literature review shows the evolution of research on biofilms and DBPs and
focuses on the relationship between these two elements. Finally, this chapter presents the
current gap related to such relationship and explains how this project intends to fill it.

 Chapter 3: Field assessment of bacterial communities and their relationships with

engineered factors. Here the identification of bacterial communities in bulk water and biofilms
is presented. The relationships among biotic and abiotic properties of a tropical-climate DWDN

Chapter 1. Introduction 9
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 are defined by statistical tests, and the relationship between methylotrophic bacteria and total
THMs concentrations is also explored.

 Chapter 4: A simple biofilm disinfection model. This chapter presents the basics for
developing a 1D model of biofilm chlorination, under stagnation conditions, and the consequent
chloroform and DCAN formation. The parameter influence on the bulk concentration of these
substances is also assessed, and the model is applied to plumbing building conditions.

 Chapter 5: Modelling DBP formation from biofilm chlorination under hydrodynamic

conditions. The simple biofilm model built in Chapter 4 forms the basis for the extended 2D
model, flow-coupled developed here. This chapter evaluates the influence of parameters such
as Re, flow regime, pipe surface to volume ratio (S/V), and water quality on the mass transfer
of dissolved substances through the biofilm surface.

 Chapter 6: General discussion, implications and applications. Here the results obtained in
Chapters 3, 4, and 5 are discussed together and their implications are reported in the context
of climate change and human health. Special focus is brought into the practicalities of the results
in the water industry.

 Chapter 7: Key findings, conclusion and future research. This final chapter summarize the
main findings of the current research project, presents a general conclusion linked to the aim
and mentions potential areas for further investigation.

Chapter 1. Introduction 10
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 2 LITERATURE REVIEW

2.1 INTRODUCTION
The current research project was described in the previous chapter, including the problem
identification and objectives. In order to identify the research gaps in current knowledge on biofilm
and DBP research, the current chapter outlines a conceptual framework to physico-chemical and
biological processes occurring in DWDNs, with special focus on biofilms, DBPs, and modelling.

2.2 PHYSICAL, HYDRAULIC AND WATER QUALITY INTEGRITY OF DRINKING

WATER DISTRIBUTION NETWORKS
The preservation of water quality for the consumers is a major challenge for utilities, and the origin
of substantial changes in the DWDN must be understood in order to reach a suitable performance
in O&M. Those changes can be associated with different events which affect the physical,
hydraulic, and water quality integrity such as cross-connections, leakages, metal leaching, pipe
corrosion, residual disinfectant loss, formation of DBPs, bacterial regrowth, turbidity fluctuations,
inadequate network construction, and maintenance and repairs (Kirmeyer et al., 2001).

As a result, water contamination sources in DWDNs are generally studied to identify potential health
risks and strategies for future investments and operation and maintenance activities. Because of
this, the USA National Research Council established that maintaining the physical, hydraulic, and
water quality integrity in DWDNs is necessary to preserve water quality, considering that the
networks can be less vulnerable to contamination than contamination of surface water sources
(NRCNA, 2006), but reducing risks could be more difficult if they became contaminated (Stevens
et al., 2004).

2.2.1 Physical integrity of DWDNs

Physical integrity of DWDNs applies to the primary physical barriers used to avoid the entry of
external contaminants to the distribution network resulting in poor water quality. Those barriers
comprises mains, services lines, and premise plumbing; fittings (crosses, tees, ells, hydrants,
valves, and meters); storage facilities (service reservoirs and storage tanks); and back-flow
prevention devices (NRCNA, 2006). The loss of physical integrity is related to corrosion (Kettler
and Goulter, 1985; Ahmadi et al., 2013); biodegradation (Wang et al., 2011); missing cover of

Chapter 2. Literature review 11

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
storage facilities (USEPA, 2002b); unsanitary activity during construction, replacement, or repair of
pipes and fittings (RSPH et al., 2017); among others (NRCNA, 2006). According to the NRCNA
(2006), the loss of physical integrity is mainly due to material failure as a result of its alteration, and
absence and improper installation of the barrier. Material modification can be related to chemical
interactions between the materials and the surrounding environment, eventually leading to holes,
leaks, and other breaches. Material collapse can be accelerated by high water pressure and stress
on pipes, and during natural disasters (NRCNA, 2006).

Losing the physical integrity of the system affects the water quality by the introduction of microbial
and chemical contaminants, debris, and particulate matter into the distribution system, sometimes
accompanied by changes in aesthetic characteristics (i.e., water colour, turbidity, taste, and odour).
However, aesthetic changes do not always represent a direct health impact (NRCNA, 2006).
Strategies such as monitoring, prediction and prevention of failures; disinfection of new pipes;
design and implementation of protocols for leakage repairs and accessories and pipeline
replacements, and inspections are recommended to maintain the physical integrity of DWDNs
(RSPH et al., 2017). In addition, maintenance activities such as proper material selection, corrosion
control, avoiding permeation, cleaning and inspection of storage facilities, and cross-connection
control should also be implemented (NRCNA, 2006).

Water supply interruption and environmental damage due to wasting water are the immediate
repercussions of leakages. They can impact customers’ perception on water companies and their
own water saving strategies (CCWater, 2013). The Consumer Council for Water in the United
Kingdom (UK) promotes the reduction of leaks in DWDNs in order to keep high levels of customer
satisfaction and reduce negative environmental impacts. The latest report for water companies in
England and Wales “Diving into Water 2011-12 to 2015-16” informed about the change of trend:
industry-wide leakage levels were rising since 2011-12 but more recently it was found a reduction
of 1.4% in leakage levels for 2015-16 (CCWater, 2013).

Despite of the immediate consequences of leakages, the most serious effect is the negative impact
on water quality. Therefore, strategies to reduce their number include setting leakage targets;
replacement of pipelines with cast iron and stainless steel pipes (Farley, 2015); however, others
recommend plastic materials over metal considering biological aspects of drinking water quality
(Fish et al., 2016). Additionally, changing water management by installing water meters and
controlling losses; reducing pressure transients and stress in the pipe network by setting valve
operations, pumping routines and pressure management to operate at the optimum level (pressure
districts); real time monitoring; smart asset management by identifying leaks as soon as they

Chapter 2. Literature review 12

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
propagate; integrated data management by handling “big data”; and researching to reduce the risk
of failure of plastic pipes (polyethylene) (Farley, 2015).

2.2.2 Hydraulic integrity of DWDNs

Hydraulic integrity refers to the maintenance required to keep a desirable water flow, water
pressure, and water age, taking both potable water and fire flow provision into account. Generally,
big cities have centralised drinking water supply systems and continuous service provision.
However, extreme weather events such as droughts and heavy rain are impacting the normal
operation of WTPs and, therefore, impacting drinking water quality (Khan et al., 2015). Such
impacts can lead to water interruption supply in developing countries as reported by Montoya et al.
(2011) in Colombia, but in industrialised countries like the USA and Australia main concerns are
more related to changes in the quality of water sources and subsequent formation of DBPs (Water
Research Foundation, 2014; Raseman et al., 2017). Similarly, water scarce areas have drinking
water supply systems which must be operated intermittently (Kumpel and Nelson, 2016). This
irregular operation of DWDNs can cause pressure transients, changes in flow regime, and backflow
(NRCNA, 2006; Kumpel and Nelson, 2016), which can lead to water quality deterioration.

Water demand is the driving force for the operation of DWDNs. Due to its stochastic nature,
operation of DWDNs requires an understanding of the amount of water being used, where it is
being used, and the temporal variation of water consumption (NRCNA, 2006). For instance, water
age is a factor related to the hydraulic design of the system and directly influences water quality
(USEPA, 2002a). One of the most commonly acknowledged effects of increasing water age is the
decay of disinfectant residuals such as chlorine and chloramines. DWDN operators are commonly
committed to keep residual concentrations of disinfectants in the networks, but very often none or
very low concentrations of disinfectant can be found in locations at the farthest distances from the
actual WTP, along with an increment of DBP concentrations such as THMs (USEPA, 2002a).

Low flow velocities in pipes create long travel times, resulting in pipe sections where sediments can
accumulate and microorganisms can form biofilms (USEPA, 2002a). Long detention times can also
greatly reduce corrosion control effectiveness by reversing blended metha- and poly-phosphates
to orthophosphates with time (i.e., as they tend to hydrolyse) and increasing their concentration in
bulk water; orthophosphates are less effective as corrosion inhibitors (USEPA, 2002a). The need
for pH control is another consequence of high water age, since the interaction between water and
cement linings significantly increases pH; unstable, soft, low-mineralized waters can revert to

Chapter 2. Literature review 13

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
untreated water pH conditions by the time the water reaches the point-of-use tap (USEPA, 2002a).
Thus, reducing water age is important to maintain the hydraulic integrity of the system (USEPA,
2002a; NRCNA, 2006).

Factors causing loss of hydraulic integrity include pipe deterioration, pressure transients and
changes in flow regime, hydraulic changes during maintenance and emergencies, tuberculation
and scale, inadequate operational control of DWDNs related to the lack of focus on water-quality-
related issues. Several procedures can be applied to maintain and/or recover the hydraulic integrity
of DWDNs such as providing alternative water sources, managing pressure zones, using devices
for surge protection, flushing water mains, maintaining sufficient mixing and turnover rates in
storage facilities, and cleaning pipes (NRCNA, 2006).

For instance, it is advisable to divide a new or existing DWDN into pressure zones when pressure
differentials in relation to minimal and maximal values are outside of their desirable values
(NRCNA, 2006). Pressure zones are created by installation of closed valves, and pressure-
regulating valves are placed between the zones to improve reliability and stretches of new pipe are
used to eliminate dead ends (NRCNA, 2006). Adequate pressure zones will reduce leaks by
properly controlling pressure fluctuations, breaks, and pumping costs; improving reservoir turnover
rates; and avoiding over-pressured systems (NRCNA, 2006), which can lead to reduce pipe
leakages. However, pressure districts or sectorization can also lead to deterioration of drinking
water quality since they reduce the network redundancy, increase the probabilities of sediment
resuspension, and promote the deterioration of drinking water quality due to the increase of number
of dead zones and water age (Grayman et al., 2009; Di Nardo et al., 2013; Wright et al., 2014) .

Pipe flushing is considered a maintenance activity aimed to eliminate loose deposits and poor-
quality water accumulated mainly in dead ends. This procedure can be conducted by conventional
or unidirectional flushing. Conventional flushing consists simply in opening one or more hydrant
valves to let sediments be washed out of the distribution network. In unidirectional flushing water
flow is increased in a specific “dirty” pipe by cutting off flows in other segments of the network and
increasing scouring velocities (to 1.5 – 3.0 m/s), with the resulting effect of flushing out sediments,
biofilms, corrosion products, and tuberculation (Antoun et al., 1999). Although, unidirectional
flushing requires more labour, equipment, data collection, and appropriate measures to dispose
disinfected water, it is more efficient and uses on average 40 percent less water than conventional
flushing (Barbeau et al., 2005).

Storage tanks or service reservoirs are used in the DWDNs to provide disinfectant contact time,
water reserve, and pressure regulation. Poor design and/or operation of storage facilities can
Chapter 2. Literature review 14
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
promote insufficient mixing and consequently increase water age, which promotes DBP formation,
losing disinfectant residual, microbial regrowth, organoleptic alterations and water recontamination
in the network (USEPA, 2002b). Montoya et al. (2012) and Montoya-Pachongo et al. (2016)
recommended to promote mixing times in service reservoirs to be higher than the duration of filling
times, reduce minimum water levels, avoid thermal stratification, increase turnover rates, and
change inlet configuration in order to increase jet momentum to impact positively water quality
stored in service reservoirs.

2.2.3 Water quality integrity of DWDNs

Water quality integrity refers to the maintenance of finished water quality via prevention of internally
derived contamination. It was already explained how issues with physical and hydraulic integrity
may result in deterioration of water quality; however, this can also occur even in the absence of
external contamination events due to transformations which take place within pipes, tanks, and
premises plumbing (NRCNA, 2006). The factors that affect water quality integrity include biofilm
growth (Srivastava and Bhargava, 2015), nitrification (Lipponen et al., 2002), leaching substances
from network materials (WHO, 2008), internal corrosion (WHO, 2008), scale formation and
dissolution (Peng et al., 2010), disinfectant loss (WHO, 2008), and DBP formation (WHO, 2008).
The core focus of the research work reported herein is on biofilms and DBP formation; therefore,
these subjects will be discussed in depth in the following sections.

Biological nitrification is a process in which nitrifying bacteria oxidize reduced nitrogen compounds
(e.g., ammonia) to nitrite and then nitrate (Lipponen et al., 2002). It is associated with long retention
times in distribution systems practicing chloramination (NRCNA, 2006). However, nitrification
potential has also been identified in chlorinated and non-disinfected DWDNs, but no correlations
between retention times and occurrence of nitrifying bacteria has been identified, as reported in a
study conducted in Finland covering 15 DWDNs from eight towns (Lipponen et al., 2002). One of
the most important problems exacerbated by nitrification is loss of the chloramine disinfectant
residual (NRCNA, 2006). In addition, all materials in the DWDN leach substances into the water,
which can alter the taste and odour of drinking water. For instance, leaching lead from lead-pipes
can represent a substantial health risk (NRCNA, 2006; WHO, 2008).

Internal corrosion manifests as the destruction of metal pipe interiors by both uniform and pitting
corrosion, which can lead to accumulation of scales. The latter one can also be formed by the
precipitation of some metals dissolved in bulk water as a consequence of changes in water pH and

Chapter 2. Literature review 15

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
redox potential. Corrosion and scaling in pipes offer higher surface area for biofilm growth, which
can be detached changing the organoleptic characteristics of bulk water (NRCNA, 2006). To
preserve drinking water quality, it is necessary to monitor it and carry out adequate O&M to prevent
or solve any case of water quality integrity loss (WHO, 2005). Monitoring water quality can include
the installation of on-line monitoring devices for conductivity, chlorine, pH, turbidity, and
temperature; and surveillance of biological quality indicators in both biofilms and bulk water with
emphasis on opportunistic and pathogenic organisms and other microorganisms associated with
them (e.g., amoebas, Cryptosporidium, Staphylococcus, etc.) (NRCNA, 2006).

In addition, the NRCNA (2006) stated that drinking water can serve as a transmission vehicle for a
variety of hazardous agents such as enteric microbial pathogens from human or animal faecal
contamination (e.g., noroviruses, E. coli O157:H7, Cryptosporidium, fungi like moulds and yeasts),
aquatic microorganisms that can cause harmful infections in humans (e.g., nontuberculous
mycobacteria, Legionella), and toxins from aquatic microorganisms such as cyanobacteria (Fan,
2012; Badar et al., 2015). In addition, some aquatic invertebrates have organoleptic and clinical
importance because can alter the turbidity of drinking water and serve as transport hosts for
pathogenic bacteria (e.g., free living amoeba and bacteria Legionella) (Lau and Ashbolt, 2009).
Such organisms have been found in bulk water from WTPs, main pipelines (Wolmarans et al.,
2005), and premise plumbing (Buse et al., 2013).

According to WHO (2005), monitoring can be defined as “the act of conducting a planned series of
observations or measurements of operational and/or critical limits to assess whether the
components of the water supply are operating properly.” This kind of monitoring involves design
and application of corrective actions in order to amend the deviation and maintain water quality in
the short and long term (WHO, 2005). The drinking water surveillance could be achieved by
traditional methods (water sampling by fixed stations) (Montoya et al., 2009) or computational
applications (real-time operation) (Perelman and Ostfeld, 2013); however, both methods may be
based on new technologies of both software and hardware. Although their application seems to be
sophisticated, these could be applied in large cities of developing countries if adequate knowledge,
training and funding are provided. However, developing countries still need to deal with providing
treatment facilities and increasing coverage of water supply in rural areas and small towns;
therefore, in the meantime, real-time monitoring of drinking water quality may be mainly affordable
for large DWDNs.

Procedures such as reduction of organic matter content and nutrients in WTPs; cleaning pipes by
flushing; controlling corrosion by appropriate selection of materials and maintaining suitable

Chapter 2. Literature review 16

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
chemical conditions of water; and keeping appropriate concentrations of residual disinfectant,
taking into consideration a balance between acute and chronic health risk, can be carried out to
directly control water quality (NRCNA, 2006). In summary, physical, hydraulic, and water quality
integrity are closely interrelated, which reflects the complexity of DWDNs and their O&M. The main
purpose of a water company is to deliver drinking water to the consumers with appropriate amount
and quality. This firstly requires to know in detail the distribution system itself (infrastructure and
operation), to monitor the most influencing factors on water quality, and to apply simultaneously
procedures to preserve it (NRCNA, 2006). This demands adequate technical and administrative
structure of the water company, where the different departments sustain fluent communication
channels for planning, decision making, and supporting to each other.

2.2.4 Plumbing systems

The last component of the water supply chain is the plumbing systems. Plumbing premises
comprise health and education institutions, private housing, commercial buildings, among others;
and it is in there where actual water consumption occurs (Chowdhury, 2016). However, regulation
and monitoring by agencies and water utilities take place in the distribution networks (NRCNA,
2006). The same physical, chemical, and biological processes occurring in a full scale DWDN take
place in plumbing systems, but these may be exacerbated by being located at the end of the
network, and poorer water quality can be found. Table 2-1 briefly explain the main characteristics
of plumbing buildings.

Table 2-1. Characteristics of plumbing systems

Characteristic Description
One of the most prominent characteristics of plumbing premises is the high S/V ratio
because they are lengthy sections and small-diameter tubing (NRCNA, 2006). S/V
is inversely proportional to pipe diameter (4/). According to Hallam et al. (2002),
High surface the higher S/V, the larger the surface across which mass transfer can occur and the
area to volume greater the number of reactive sites available for each unit volume of water passing
ratio (S/V) through the pipe. While the public infrastructure can have S/V of 0.26 cm2/mL,
private infrastructure can have S/V of 2.1 cm2/mL (NRCNA, 2006). This represents
that wall effects on bulk water are more important in these systems as it will be
explained later.
In plumbing systems, water flow is more dependent on household consumption
patterns. These systems are characterized for stagnation conditions since the
buildings can remain unoccupied for several hours, days or weeks in houses,
High water age
commercial buildings or schools, respectively. It should be noted that the negative
effects of water age are exacerbated if the biological stability of the water in the
DWDN is poor (NRCNA, 2006).
Presence of
Several materials can be found in plumbing premises such as copper, plastics,
different
brass, lead, galvanized iron, and stainless steel. Such variety is not usually found in
materials
Chapter 2. Literature review 17
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Characteristic Description
public DWDNs. The type of pipe material influences the interaction with
disinfectants and microorganisms (NRCNA, 2006).
Interior of buildings are naturally warmer that outdoors, especially in geographic
Extreme regions with temperate weather, which use heating systems during cold season.
temperatures This is reflected in warmer water temperatures in plumbing facilities. In addition, the
domestic hot-water distribution systems also transfer heat to adjacent pipes.
The reaction of disinfectants with pipe materials, organic and inorganic matter, and
Low or no biofilms promotes the disinfectant decay. If the disinfectant concentrations are
disinfectant already low in the service lines, it is expected total disappearance of this under
residual stagnation conditions in building facilities (Lu et al., 1999; Buamah et al., 2014; Lee
et al., 2014).
Potential Type of pipe material (Song et al., 2015), high S/V, warm water, and low
bacterial concentration of disinfectant promote the bacterial regrowth and biofilm formation or
regrowth and survival. This increases the risk by potential survival and exposure to pathogenic
biofilm formation microorganisms (NRCNA, 2006).
Plumbing premises are characterized by start-stop flow patterns. This induces a
Highly variable high variation of flow velocities, which can lead to biofilm detachment and potential
velocities consumption of pathogenic microorganisms by the consumers, if these were
released to the bulk water (NRCNA, 2006).
Exposure to Showering is an important exposure route to bioaerosols with hygienic importance
vapours and such as Mycobacterium avium and Legionella (Whiley et al., 2015) and volatile
bioaerosols DBPs such as THMs (Cantor et al., 2010; Chowdhury, 2016).
According to NRCNA (2006), service lines are the connection between the
Proximity to distribution main and the premise plumbing. The potential contamination of these by
service lines water leaks and repairs can affect individual consumers due to the proximity of
plumbing premises to service lines.
Maintenance activities of plumbing systems are developed by the household or by
Prevalence of
plumbers, who are not trained and licensed in all the cases. Thus, the risk of cross-
cross-
connections may increase leading to backflow and contamination of the drinking
connections
water within the facilities (NRCNA, 2006).

2.3 BASICS OF TRANSPORT PHENOMENA

Transport phenomena occurs when a perturbation of the equilibrium occurs in a physical system.
Such perturbation can correspond to gradients of temperature, substance concentrations, and
velocity. Energy, matter and momentum tend to move from regions of higher to lower concentration.
The central aspects of the analysis of transport phenomena is the establishment of suitable
constitutive equations to relate the fluxes of energy, matter, and momentum to local material
properties.

2.3.1 Fluxes and conservation of chemical species

The fluxes of energy, matter, and momentum are composed by convective transport, which is
related to bulk motion, and molecular or diffusive transport, which is related to small-scale
molecular displacements. For the case of heat conduction, the Fourier’s law is used as the

Chapter 2. Literature review 18

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
constitutive equation to describe the heat fluxes; Equation (2-1) represents the heat flux q (energy
flow per unit cross-sectional are). A schematic representation of the heat flux vector is presented
in Figure 2-1 for an imaginary surface, which is perpendicular to a vector n of unit magnitude.
Equation (2-2) expresses the heat flux qn normal to the surface. Here k is the thermal conductivity,
 is the gradient operator, and T is temperature.

Figure 2-1. Schematic representation of mass flux at a surface with a unit normal n and arbitrary
orientation

𝑞 = −𝑘∇𝑇 (2-1)

𝑞𝑛 = 𝑛 ∙ 𝑞 = −𝑘(𝑛 ∙ ∇𝑇) (2-2)

A similar analysis can be done for a given species in a mixture, which results from a gradient in a
concentration of that species; in this case, the Fick’s law is used as the constitutive equation to
describe the fluxes. The total flux (Ni) is given by Equation (2-3).

𝑵𝑖 = 𝐶𝑖 𝒖 + 𝐉𝑖 (2-3)

Here, Ciu is the convective flux (Ci: concentration, u: mass-average velocity) and Ji is the molar
flux of the species i. For liquid solutions, the molar flux is given by Equation (2-4), where Di is the
diffusion coefficient of the substance.

𝐉𝑖 = −𝐷𝑖 ∇𝐶𝑖 (2-4)

2.3.2 Mass transfer and boundary conditions at interfaces

The analysis of conservation equations for interfaces found in Deen (1998) defined an interfacial
balance valid at any instant in time given by the flux relative to the interface between two phases.
Such interfacial source creates a difference in the normal components of the fluxes relative to the

Chapter 2. Literature review 19

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
interface. In general, all of the quantities involved are functions of position on the interface, so that
the balance changes from point to point. This interfacial balance is a key result, because it is the
basis for many of the boundary conditions in mass transfer problems. Further simplification of such
interfacial balances led to the boundary condition flux continuity, which is commonly used in biofilm
modelling (Equation (2-5)) (IWA Task Group on Biofilm Modeling et al., 2006; Taherzadeh et al.,
2012). Here flux in the phase biofilm (B) at the biofilm surface (Bs) is equal to the flux in liquid (L)
at the same point.

𝑑𝐶 𝑑𝐶 (2-5)
𝐷𝐵 |𝐵 = 𝐷𝐿 |
𝑑𝑛 𝑑𝑛 𝐿

In the case of bulk motion, convection boundary condition at the biofilm surface is defined in the
Equation (2-6) (Deen, 1998). Here kf is the mass transfer rate specific to the substance since it
depends in part on its diffusivity, so it is different for each component of a mixture. C L-Bs is the
concentration of the substance i in liquid at the biofilm surface. C o is the concentration in the bulk
far away from the biofilm surface, concentration at the reactor inlet is usually used as Co
(Taherzadeh et al., 2012). According to Deen (1998), when mass transfer is accompanied by bulk
flow normal to the interface, the boundary condition shown in Equation (2-6) is best preserved if
the mass transfer coefficient continues to refer only to the diffusive part of the solute flux (i.e., JBs
rather than NBs). This is because this analogy has been adapted from the analysis of heat transfer;
then heat transfer coefficients have been defined to represent only the conduction heat flux and
not energy transfer by bulk flow (Deen, 1998).

𝐉𝐵𝑠 = 𝑘𝑓−𝑖 (𝐶𝑖−𝐿−𝐵𝑠 − 𝐶𝑜 ) (2-6)

2.3.2.1 Sherwood number

Despite of the simplicity of Equation (2-6), the parameter kf and variable CL-Bs are not always
available. In biofilm analysis, k f can be calculated from experimental tests of penetration of a
specific substance and monitoring its concentration in the bulk and within the biofilm by micro-
sensors. More details of this can be found in Chen and Stewart (1996) and Guimerà et al. (2016).
By combining Equations (2-4) and (2-6), kf is given by Equation (2-7); where δC represent the
thickness of the stagnant film in the fluid next to the wall; δ C is also known as boundary layer
thickness.

𝐷𝐿 (2-7)
𝑘𝑓 =
𝛿𝐶

Chapter 2. Literature review 20

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
In some cases, δC is not available; therefore, the Sherwood number (Sh) is used mainly for
modelling purposes (IWA Task Group on Biofilm Modeling et al., 2006; Tosun, 2007). Since δC
depends on the convection just outside the interface, Sh gives a measurement of the convective
and diffusive fluxes to such interface. Sh is a dimensionless form of k f and is represented by the
Equation (2-8). Here Lch is the characteristic length which can be assumed as pipe diameter or
radius, width of a channel, or radius of the bubble for the case of transport around a gas bubble in
a liquid (COMSOL, 2017). For instance, Taherzadeh et al. (2012) used the biofilm head diameter
as Lch for studying the fluid dynamics and mass transfer of biofilm streamers with a small immobile
base attached to the support and a flexible tail elongated in the flow direction, which can vibrate in
fast flows.

𝑘𝑓−𝑖 𝐿𝑐ℎ (2-8)

𝑆ℎ𝑖 =
𝐷𝑖

By combining the Equations (2-4), (2-6) and (2-8), an expression for Sh based on concentration
gradient at the biofilm surface (Equation (2-9)) is obtained, which leads to the calculation of local
Sh; then the global Sh (Sh
¯ ) can be determined by a surface integration and averaging (Equation
(2-11)). This allows comparing Sh for different system configurations and obtaining the global mass
̅̅̅𝑓 ) in biofilm modelling (Equation (2-12)). For instance, this approach was used by
transfer (𝑘
Taherzadeh et al. (2012). Here ds refers to differential of biofilm surface. In the case of biofilm
attached to the pipe wall, ds represents the differential of pipe surface.

𝑑𝐶 (2-9)
𝑘𝑓−𝑖 𝐿𝑐ℎ 𝐿𝑐ℎ 𝑑𝑛 |𝐵𝑠
𝑆ℎ𝑖 = =
𝐷𝑖 𝐶𝑖−𝐵𝑠 − 𝐶𝑖−𝑜

𝑑𝐶𝑖 𝐉 (2-10)
|𝐵𝑠 = 𝑖
𝑑𝑛 −𝐷𝑖

(2-11)
∫𝐵𝑠 𝑆ℎ𝑖 𝑑𝑠 𝑘𝑓−𝑖 𝐿𝑐ℎ
𝑆ℎ𝑖 = =
∫𝐵𝑠 𝑑𝑠 𝐷𝑖

𝑆ℎ𝑖 𝐷𝑖 (2-12)
𝑘𝑓−𝑖 =
𝐿𝑐ℎ

2.3.2.2 Engineering correlations

According to Tosun (2007), most engineering problems do not have theoretical solutions, then a
large portion of engineering analysis is concerned with experimental information, which is usually
Chapter 2. Literature review 21
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
expressed in terms of engineering correlations. In the case of mass transfer between bulk liquid
and biofilm, calculation of Sh becomes important in order to obtain the values of k f for each
substance of interest. According to Guimerà et al. (2016) and IWA Task Group on Biofilm Modeling
et al. (2006), Sh is often expressed as a function of the non-dimensional Reynolds number (Re –
Equation (2-13)) and Schmidt number (Sc – Equation (2-14)): Sh=f(Re, Sc). Sh is usually
represented by a power function Sh=A+B•RemScn. The parameters A, B, m, and n are, in most
cases, determined empirically from experimental data for a specific system.

𝑢𝐿 (2-13)
𝑅𝑒 =
𝜈

𝜈 (2-14)
𝑆𝑐 =
𝐷𝑖

Where  is the kinematic viscosity of the fluid (m2/s), u is the velocity of the fluid with respect to the
object (m/s), L is a characteristic linear dimension (m), and D is the mass diffusivity (m 2/s) of
species i. In relation to Sh, Equation (2-15) reported by Horn and Lackner (2014; cited by (Guimerà
et al., 2016)) is used for turbulent models and Equation (2-16) is applied to laminar flow (Zhang
and Bishop, 1994). Experimental data from Guimerà et al. (2016), who characterized external and
internal mass transport of dissolved oxygen in an heterotrophic biofilm, fitted very well with the
Equation (2-16).

𝑆ℎ = 0.239 ∙ 𝑅𝑒 0.8 ∙ 𝑆𝑐 0.33 (2-15) Turbulent flow

𝑆ℎ = 0.369 ∙ 𝑅𝑒 0.5 ∙ 𝑆𝑐 0.33 (2-16) Laminar flow

Due to the empirical origin of expressions to calculate Sh, their application to biofilm modelling must
be cautious since parameters depend on the geometry of the biofilm support medium, the type of
substance, and are valid only for a defined range of Re and Sc. Extrapolating from these conditions
can yield erroneous results (IWA Task Group on Biofilm Modeling et al., 2006). Additionally, most
correlations for Sh were derived for rigid particles, but the elastic and heterogeneous nature of the
biofilm can influence external mass transfer (IWA Task Group on Biofilm Modeling et al., 2006).
Despite of the mentioned cautions, engineering correlations can be useful to correlate experimental
data to simplified models; which in turn can represent an acceptable physical process. However,
with the further and accelerated development of numerical methods, computational techniques,

Chapter 2. Literature review 22

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
software, and hardware, applying theoretical concepts to physical process in drinking water industry
is more feasible every day, the more reliable results can be obtained from modelling approaches.

2.4 BIOFILMS IN DRINKING WATER DISTRIBUTION NETWORKS

Biofilms can be understood as a biological strategy of microorganisms to survive in hostile
environments. From a microscale point of view, biofilms are a group of microorganisms, mainly
bacteria but archaea, fungi and protozoa are also important (Wang et al., 2014; Fish et al., 2015),
living as a consortium, attached to a surface as a result of the secretion of EPS (Srivastava and
Bhargava, 2015). Biofilms are successful due to the protective effect of EPS to cells against oxidant
substances, grant cells attachment to surfaces under certain hydraulic conditions, and improve
availability of nutrients as a result of organic matter retention. With regards to hydraulic forces,
flushing water pipes has been proved as a proper technique to remove soft sediments and
biological material but inefficient to completely detach biofilms (Abe et al., 2012; Douterelo et al.,
2013). From a macroscale point of view, biofilms can be conceived as reservoirs of organic matter
due to the presence of lipids, proteins, carbohydrates, DNA, and organic matter particles (Wang et
al., 2012b; Fish et al., 2015).

Biofilms are naturally found in most of the solid/liquid interfaces such as showers, pools, teeth, food
industry facilities, wastewater treatment plants, water supply systems, among others. In water
systems for human consumption, biofilms grow in the walls of reactors of WTPs, pipes, valves,
tanks, pumps and all the fittings of the system. In this case, drinking water biofilms are a major
concern because they can lead to corrosion, discoloured waters and odours; host pathogens which
may be released to bulk water; detach and recolonize clean surfaces; act as precursors for the
formation of DBPs, and consequently, contribute to disinfectant decay (NRCNA, 2006).

Biofilm formation starts with the adhesion or incorporation of microorganisms depending on their
ability to produce EPS, salivary pellicle or if the cell structure includes components such as flagella
(Song et al., 2015). Material properties like surface charge, surface energy, roughness, topography,
surfaces with topographic patterns, and stiffness influence the adhesion of microorganisms. Some
bacteria like P. aeruginosa use the surface sensing ability to probably modify cell surfaces to better
attach to negatively charged surfaces (Song et al., 2015). The organisms may attach to surfaces
as primary colonizers and actively establish biofilms alone or in combination with other
microorganisms. However, they can also become integrated in pre-existing biofilms as secondary

Chapter 2. Literature review 23

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
colonizers. Bacteria such as Pseudomonas, Staphylococcus, and Methylobacterium are known as
biofilm formers (Simões et al., 2007; Simões et al., 2010; Douterelo et al., 2014c).

In a second step, incorporation of other organisms such as virus, fungi, and invertebrates occurs.
In this stage, heterotrophic bacteria, free-living protozoa and fungi can multiply and persist if they
have adapted to the oligotrophic conditions characteristic of engineered water systems (Wingender
and Flemming, 2011). Given suitable laboratory conditions, all relevant water-related pathogenic
bacterial species have actually been shown to be able to adhere to solid surfaces and/or to form
mono-species biofilms, indicating their potential as biofilm organisms (Wingender and Flemming,
2011). However, enteric viruses and parasitic protozoa are obligate parasites and dependent on
multiplication in animal or human hosts. Such organisms can only be expected to attach to and
persist in biofilms without being able to proliferate (Wingender and Flemming, 2011). Finally,
chlorine effects and shear stress can cause the detachment of the biofilm (Xue and Seo, 2013),
releasing the organisms, which can reach the customers, colonize new surfaces, or be incorporated
in other biofilms downstream.

2.4.1 Taxonomy and diversity

Population genetics and taxonomy fields are two very different aspects of biology and necessitate
very different assumptions, theories and methods, but they overlap at some degree (Wheeler,
2008). First, taxonomy refers to naming, describing and classifying organisms and includes all
plants, animals and microorganisms discovered in nature. This branch of biology is being
transformed according to the technical and technological progress made in the access of electronic
publications, molecular analysis, and bioinformatics (Wheeler, 2008). Then, the taxonomic rank
allows allocating an organism in a certain group defined by the kingdom, phylum, class, order,
family, genus, and species. In the environment, ecosystems are an assemblage of populations,
plants, animals, bacteria, fungi (Whittaker, 1975), and the recent-defined archaea, that live in an
environment and interact with one another, forming together a distinctive living system with its own
composition, structure, environment relations, development, and function (Whittaker, 1975).

Natural ecosystems are forests, lagoons, mangroves, etc. DWDNs can be considered engineered
ecosystems since all the characteristics previously mentioned are present in there, with the
difference that they have been designed, constructed, operated and maintained by humans.
Therefore, it can be inferred that most of the variables influencing the behaviour of such systems
can be modified according to the needs of the managers and drinking water consumers. In order

Chapter 2. Literature review 24

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
to study living systems in their environmental contexts, ecology arises as the area of the biology
concerning to it.

Whittaker (1975), by studying the communities in redwood forests, analysed the species diversity
by the application of indices, according to the alpha (within community); beta (between
communities; and gamma diversity (between regions) (Whittaker, 1975; Sepkoski, 1988).
Examples of alpha-diversity metrics are Margalef and Chao1 richness estimators, and Simpson
and Shannon diversity estimators (Whittaker, 1975; Gamito, 2010; Douterelo et al., 2013).
According to the scope of this project, calculation of Shannon (H) and Margalef (d) indices to
calculate alpha diversity are explained in this section (Equations (2-17), (2-18) and (2-19)).
Here S is the number of species, N is the total number of individuals in the sample, pi is the relative
importance values for these same species.

𝑆−1 (2-17)
𝑑=
𝐿𝑛 𝑁

𝑆 (2-18)
′
𝐻 = − ∑ 𝑝𝑖 𝑙𝑜𝑔𝑝𝑖
𝑖=1

𝑆 = exp(𝐻′ ) (2-19)

2.4.2 Microbiome identification in bulk water and biofilms in drinking water networks

Culture-dependent methods, routinely used by water utilities to assess the microbiological quality
of drinking water, present shortcomings such as underestimation of the amount and diversity of the
microbial community (Theron and Cloete, 2000). Heterotrophic bacterial counts only represent a
limited fraction of the whole microbial community (Douterelo et al., 2014a; Ren et al., 2015), when
used to estimate bacterial loads in water samples. Moreover only less than 5% of the biomass is
present in the water phase and 95% is living in the biofilm phase (Flemming et al., 2002). Current
drinking water quality standards rely on culture-dependent techniques to quantify specific
pathogenic organisms (e.g., E. coli, Enterococci, Coliform bacteria) (UK Parliament, 2000;
Ministerio de la Protección Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial,
2007) and heterotrophic bacteria in bulk water. Then, biofilms are only explored for research
purposes and, to date, they have not been included in routine operative and regulatory plans by
drinking water operators and agencies, respectively.

Chapter 2. Literature review 25

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Recently, water microbiology research is focused on developing and applying culture-independent
practices in order to improve the diagnostic of microbiological quality of water. Therefore, molecular
analysis is being used for studying the microbial community in bulk water and biofilms. According
to Douterelo et al. (2014a), studying microorganisms in a DWDN allows improving knowledge about
how abundant they are, which types of microorganisms are present, how their activities shape the
environment or influence other organisms, including any possible effects on human health, and
how the environment influences the structure and functions of the microorganisms present.
Douterelo et al. (2014a) also described the classification of microbiological techniques used to
study microbial communities according to their purpose and nature: culture-based methods,
microscopy, polymerase chain reaction (PCR), molecular techniques, and quantification of
metabolism by-products. In particular, molecular techniques include cloning and sequencing,
metagenomics, and next generation sequencing. The molecular techniques are applied to study
the microbial community composition and involve the extraction of nucleic acid, followed by PCR
amplification of “marker genes” to obtain taxonomic information. The most commonly used marker
gene in microbiological research is the ribosomal RNA (rRNA) gene, 16S rRNA for prokaryotes and
18S rRNA for eukaryotes (Douterelo et al., 2014a).

According to Douterelo et al. (2014a), cloning and sequencing is the conventional and more
widespread genomic approach used when detailed and accurate phylogenetic information from
environmental samples is required. After DNA extraction and amplification of the rRNA gene with
suitable primers, clone libraries using sequencing vectors must be constructed (Rondon et al.,
2000). Selected clones are then sequenced (Sangerbased) (Sanger et al., 1977) and the nucleotide
sequence of the rRNA gene retrieved, allowing estimates of the microbial diversity in the samples
by comparison with sequences available in databases (e.g., GenBank, EMBL and Silva). The
generation of DNA clone libraries followed by sequencing has being extensively applied in drinking
water microbiology. The need to analyse the extensive amount of data related to sequence reads
from environmental samples led to the development of the bioinformatics field, which involves
software and analysis tools (Douterelo et al., 2014a).

Understanding of microorganisms in DWDNs has been developed by sampling water and biofilms.
In relation to water sampling, guidelines from environmental and regulatory agencies exist for
sampling methods and interpretation of results of traditional and well-known laboratory parameters.
For instance, there are several International Organization for Standardization (ISO) standards for
detection and enumeration of faecal indicator organisms in water, which are listed in WHO (2017).
The environmental agency of the United States has designed a guide for sampling biological
contaminants in bulk water (USEPA, 2016b). In the European territory, the standard ISO 19458 is
Chapter 2. Literature review 26
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineerin
recommended for compliance samples for microbiological parameters (European Union, 2015).
Locally, the UK Environment Agency also offers guidelines for sampling drinking water to test
microbiological parameters (Environment Agency, 2010).

However, if molecular analysis and/or proteomics/metabolomics-based approaches are desired,

there is not enough guidance about the minimal representative sampling volume required to
capture the complete microbiome in a DWDN. Therefore, comparisons are difficult to make. In
relation to biofilm sampling, two approaches have been used to study biofilms in DWDNs such as
scrapping attached biofilm from cut-out pipes or directly from devices inserted into the pipes. Both
methods can be used for sampling in real-scale or laboratory DWDNs (Douterelo et al., 2014a).

Pipe cut-out sampling are labour-intensive, then it is common to take advantage of the leakage
repairs or pipe replacement activities to collect pipe samples. Due to the random nature of leakages
location, the arbitrary selection of sampling points according to the scope of the study is limited.
Therefore, the sampling campaign may last several months in order to cope with all the selected
criteria. Otherwise, the study must be based on few variables in order to find the balance time-
scope. On the other hand, excavation and cutting processes often lead to concerns with
contamination and representative sampling and preservation of the physical integrity of the sample
(Douterelo et al., 2014a). The use of devices, commonly coupons, that can be deployed repeatedly
either within a pilot-scale test facility or in an operational DWDN, allows the study of biofilm
dynamics over time in relation to changing abiotic and biotic factors in situ. According to Douterelo
et al. (2014a), commonly the main limitation of some of these devices is that they distort hydraulic
conditions in pipes and, in most cases, shear stress and turbulence regimes are different from
those expected in real pipes, artificially influencing the way biofilms develop.

Recent studies applying molecular analysis in DWDNs supplied by surface water, either simulated
or real-scale DWDNs, are summarized in Table 2-2. These studies were developed mainly in
temperate-climate areas; with water treated by different treatment processes, including chlorine
and chloramines as disinfectants. Samples were collected from several pipe materials such as
polyvinyl chloride (PVC), HDPE, ductile and cast iron, cement, copper. Pipe material was the
variable more frequently tested among these studies. Results reported in every investigation found
Proteobacteria as the predominant phylum in biofilm and bulk water samples. Particularly, research
in USA indicates that other phyla also dominating include Cyanobacteria and Actinobacteria
(Holinger et al., 2014; Kelly et al., 2014). By comparison, bacterial class, family and genera exhibit
more variety of the microbial communities in the studied samples.

Chapter 2. Literature review 27

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 2-2. Summary of bacterial composition in DWDNs supplied by surface water sources
Pipe age Composition of bacterial communities in pipelines (dominant
Reference Type
Type of Sampling (Years) / communities and relative abundance -RA-)
and study Water treatment of Pipe material
system point Biofilm
country sample Phyla Class Genera
age
Mycobacterium,
FLOC, SEDIM, pH
Acidovorax,
adj., SFILT, GAC,
Proteobacteria Alphaproteobacteria (35%) Burkholderia,
and chlorine
Pseudomonas,
Gomez- disinfection
Dechloromonas
Alvarez et
Simulated Caulobacter,
al. (2012) Tap Water PVC NS
DWDN Rhodopseudomonas,
Alphaproteobacteria (22%) Synechococcus,
USA chloramine Proteobacteria
Actinobacteria (28%) Bradyrhizobium,
disinfection Actinobacteria
Betaproteobacteria (24%) Pseudomonas
(chloraminated
system)
Alphaproteobacteria (2-30%)
Proteobacteria Chlorofexi (7-65%)
Henne et al. Pipe Biofilm NS
(9-62%) Betaproteobacteria (1-20%)
(2012) Full scale FLOC, COAG,
NS NS Actinobacteria (6-19%)
DWDN SFILT, chlorination
Alphaproteobacteria (17-28%)
Germany Proteobacteria
Tap Water Actinobacter (16-25%) NS
(27-41%)
Bacteroidetes (14-25%)
Gammaproteobacteria (6-87%) Pseudomonas
Biofilm
Douterelo Betaproteobacteria (4-60 %) (20-65%)
Flushing
et al. (2013) Simulated Methylocistis
NS points in HDPE NS Proteobacteria
DWDN (20-40%)
the network Water Alphaproteobacteria (59-88 %)
UK Methylocella
(10-20%)
Henne et al. Proteobacteria Alphaproteobacteria (7-28%)
Full scale FLOC, COAG, Cold
(2013) Tap NS NS (15-72%) Betaproteobacteria (6-28%) NS
DWDN SFILT, chlorination water
Bacteroidetes Gammaproteobacteria (2-16%)

Chapter 2. Literature review 28

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Pipe age Composition of bacterial communities in pipelines (dominant
Reference Type
Type of Sampling (Years) / communities and relative abundance -RA-)
and study Water treatment of Pipe material
system point Biofilm
country sample Phyla Class Genera
age
Germany (17-36%)
Actinobacteria
(8-30%)

NS
Disinfection with
Holinger et
chloramine: 2 Cyanobacteria Chloroplast (24-91%)
al. (2014) Full scale Taps of Mycobacterium
sampling points Water NS NS Actinobacteria Sphingomomadales (45%)
DWDNs buildings (43-52%)
Disinfection with Proteobacteria Other (29%)
USA
chlorine: 7 sampling
points
Methylomonas
(95.5%)
Mycobacterium
Kelly et al.
CLAR, OZON, BAF, (59.1%)
(2014) Full scale Proteobacteria
and chlorine Pipelines Biofilm Ductile iron 40-45 Gammaproteobacteria Unclassified
DWDN Actinobacteria
disinfection Xanthomonadaceae
USA
(50.4%)
Acinetobacter
(74.2%)
Shaw et al.
(2014) COAG, MIEX, MIEX Biofilm age:
NS Pipelines Biofilm NS Proteobacteria Alphaproteobacteria NS
+ COAG + GAC, MF 28 months
Australia
Sun et al. COAG, SEDIM, Raistoria
(2014) Full scale FILT, GACAd, Proteobacteria Bacilli (30-35%) Burkholderia
Pipelines Biofilm CI ≈ 20
DWDNs and chlorine (40-60%) Betaproteobacteria (25-30%) Delftia
China disinfection Bacillus

Chapter 2. Literature review 29

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Pipe age Composition of bacterial communities in pipelines (dominant
Reference Type
Type of Sampling (Years) / communities and relative abundance -RA-)
and study Water treatment of Pipe material
system point Biofilm
country sample Phyla Class Genera
age
Mycobacterium
(RA closest to max.
value: 42.21%)
Wang et al.
Ports
(2014) Simulated 4.0 mg/L chlorine Cement, iron Biofilm age: Proteobacteria Alphaproteobacteria (25-96%)
located in Biofilm NS
DWDN 4.8 mg/L chloramine and PVC 6 months (66-98%) Betaproteobacteria (1.2-69%)
pipelines
USA
El- COAG, FLOC, Effluent of
Chakhtoura SEDIM, OZON, treatment
et al. (2015) Full scale DMF, GACAd, and facility Cemented Proteobacteria
Water NS Gammaproteobacteria (34.0%) NS
DWDN chlorine dioxide Pipeline in steal and PVC (58.6%)
The disinfection (no water
Netherlands residual disinfectant) network
Water Brass and Proteobacteria
Lührig et al. 4-10 Alphaproteobacteria (70-85 %) NS
meters plastic (45-87 %)
(2015) Full scale
NS Biofilm Gammaproteobacteria (60-65%)
DWDN Proteobacteria
Pipelines Cast iron 45-103 Betaproteobacteria (35%) NS
Sweden (58-86 %)
Alphaproteobacteria (35 %)
Commercial
Commercial
water Acinetobacter
Mahapatra water
purifier (24.7%)
et al. (2015) Full scale purifier
NS Water PVC systems: 27 Proteobacteria Gammaproteobacteria Pseudomonas
DWDN systems
months (20.8%)
India and
Pipelines: Klebsiella (18.5%)
pipelines
NS
Conventional Pipe age:
Ji et al. Alphaproteobacteria
treatment, Building NS
(2015) Full scale CU, CPVC, Proteobacteria Betaproteobacteria
chlorine/chloramine plumbing Biofilm Staphylococcus
DWDNs CU/Lead (50-90%) Gammaproteobacteria
disinfection (5 rigs Biofilm age:
USA Actinobacteria
WTP) 1 year
Chapter 2. Literature review 30
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Pipe age Composition of bacterial communities in pipelines (dominant
Reference Type
Type of Sampling (Years) / communities and relative abundance -RA-)
and study Water treatment of Pipe material
system point Biofilm
country sample Phyla Class Genera
age
Hypromicrobium
Alphaproteobacteria (68%, (28%)
Ren et al.
FLOC, SFILT, and Pipelines: DCIP, GCIP, 83%) Desulfovibrio (17%)
(2015) Full scale Proteobacteria
chlorine ends of the Biofilm GSP, SSCP, 11-13 Betaproteobacteria (41%) Pseudomonas (43%,
DWDN (71-99%)
disinfection DWDS PVC Gammaproteobacteria (45%, 63%)
China
73%) Sphingomonas
(72%)
Biofilm age:
Coupons
Polycarbonate 3-18
incubated
months
within
Revetta et Alphaproteobacteria (34%)
biofilm Mycobacterium
al. (2016) Full scale GAC, chlorine Proteobacteria Actinobacteria (24%)
annular Biofilm (23%)
DWDN disinfection Biofilm age: Actinobacteria Betaproteobacteria (20%)
reactors Limnobacter (11%)
USA Glass 6-11 Gammaproteobacteria (12%)
and glass
months
bead
baskets
COAG, rapid gravity
Bautista-de filtration, chlorine
Comamonadaceae (a)
los Santos disinfection, and
Full scale Proteobacteria Betaproteobacteria (92.2%) (64.9%)
et al. (2016) orthophosphate Faucets Water NS NS
DWDN (> 98%) Alphaproteobacteria (76.6%) Sphingomonadaceae
dosing (another WTP
(2.22%)
UK included air flotation
after coagulation)
Alphaproteobacteria (32.58%)
Zhang et al. Sphingomonas
Full scale Bacilli (15.65%)
(2017) Biofilm Proteobacteria (25.25%)
laboratory NS Coupons PE 3 years Gammaproteobacteria (15.63%)
Actinobacteria Streptococcus
DWDN Betaproteobacteria (14.30%)
China (7.64%)
Actinobacteria (10.44%)

Chapter 2. Literature review 31

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Pipe age Composition of bacterial communities in pipelines (dominant
Reference Type
Type of Sampling (Years) / communities and relative abundance -RA-)
and study Water treatment of Pipe material
system point Biofilm
country sample Phyla Class Genera
age
Enhydrobacter
(33.84%)
Gammaproteobacteria (47.95%)
Propionibacterium
Ductile iron Actinobacteria (13.88%)
(8.08%)
Betaproteobacteria (11.95%)
Acinetobacter
(5.59%)
Betaproteobacteria (27.53%) Flavobacterium
Flavobacteria (25.15%) (24.76%)
SS
Verrumicrobiae (13.4%) Arcicella (12.03%)
Cytophagia (12.04%) Acidovorax (8.98%)
Methylophilus
Water -- -- Proteobacteria Betaproteobacteria (95.5%)
(95.41%)
(a): The authors reported these organisms as classified at the familiy/genus level | BAF: biologically active filtration | CI: cast iron | CLAR: clarification | COAG: coagulation | CPVC: polyvinyl chloride with brass fitting | CU:
copper | DCI: ductile cast iron | DMF: dual medium filtration | FILT: filtration | FLOC: flocculation | GAC: granular activated carbon | GACAd: granular activated carbon adsorption | GCI: grey cast iron | GS: galvanized steel |
HDPE: High-density polyethylene |MIEX: magnetic ion exchange contact | MF: membrane filtration |NS: not specified | OZON: ozonation | PB: polybutylene (PB) | PE: polyethylene | PVC: polyvinyl chloride | SEDIM:
sedimentation | SFILT: sand filtration | SS: stainless steel | SSC: stainless steel clad | ST: steel coated with zinc |

Chapter 2. Literature review 32

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
2.4.3 Biofilm modelling

As previously explained, biofilms in drinking water pipes are heterogeneous and includes multiple
bacteria species such as aerobic, anaerobic, heterotrophic, and autotrophic bacteria. Substances
like EPS, biomolecules and other microorganisms such as fungi, archae, and invertebrates are also
present. In addition, hydraulics affect the biofilm physical structure by the influence of shear stress,
which can cause detachment, and nutrients transport. Disinfectant in drinking water also affect the
biofilm by killing cells and oxidizing biomolecules, which can translate into genetic adaptation or
physical resistance and DBP formation. Drinking water is an oligotrophic environment and,
according to nutrient concentrations, carbon, and nitrogen or phosphorous can be the growth-
limiting nutrient.

Multiple factors affect the growth of biofilms in drinking water pipes. Table 2-3 reports 31 studies
conducted on biofilm modelling between 1994 and 2017 in order to better explain the different
approaches applied by researchers to represent biofilm dynamics. As it can be observed in Table
2-3, biofilm models have been low-dimensional continuum models (Stewart, 1994; Stewart and
Raquepas, 1995; Eberl and Demaret, 2007; D’Acunto and Frunzo, 2011; Clarelli et al., 2013;
D’Acunto et al., 2015), discrete-continuum/cellular automaton models (Picioreanu et al., 1998a;
Picioreanu et al., 1998b; Xavier et al., 2005a; Xavier et al., 2005b; Jayathilake et al., 2017),
multidimensional biofilm models with biomass and flow coupling under laminar regime (Eberl et al.,
2000; Picioreanu et al., 2000a; Xavier et al., 2005a; Eberl and Sudarsan, 2008; Zhang et al., 2008;
Duddu et al., 2009; Lindley et al., 2012; Taherzadeh et al., 2012; Zhang, 2012; Coroneo et al.,
2014; Tierra et al., 2015; Zhao et al., 2016) and other multidimensional biofilm models (Eberl et al.,
2001; Eberl and Efendiev, 2003; Duddu et al., 2008; Cogan, 2010; Cumsille et al., 2014).

Most of the models have focused on biofilm structure as a result of biofilm growth and spreading,
and substrate transport and consumption. However, other models have evolved to consider
external flow and effects of antimicrobial agents to simulate the antimicrobial penetration barrier,
persistent cells, and viscoelastic properties of biofilm. Due to the important role played by EPS in
initial formation, growth, and survival of biofilms; 15 of 31 reviewed studies included EPS production
in the biofilm models (Kommedal et al., 2001; Xavier et al., 2005a; Xavier et al., 2005b; Duddu et
al., 2008; Zhang et al., 2008; Duddu et al., 2009; Cogan, 2010; Cogan, 2011; Lindley et al., 2012;
Clarelli et al., 2013; Ghosh et al., 2013; Macías-Díaz, 2015; Tierra et al., 2015; Zhao et al., 2016;
Jayathilake et al., 2017).

Chapter 2. Literature review 33

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 2-3. Summary of studies on biofilm modelling

Model components Max. BT

System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
Simulated | Disinfection
Stewart 1D - Length: 1.4 Not Active and Dissolved Not Type of Not Not modelling was
Simulated Not specified
(1994) continuum and 87 m simulated inert cells oxygen simulated antibiotic not specified specified simulated up to 8
specified hours
Model assessed
Simulated |
Stewart and the decay of
1D - Not Active and Not Not Type of Not Not
Raquepas Not apply Not apply Not specified active cells by
continuum simulated inert cells simulated simulated antibiotic not specified specified
(1995) action of
specified
antibiotics
Length: 600 Spatial
2D - Not
m; Depth: characteristics of
Discrete simulated Common
600 m Simulated | colonies are
Picioreanu Nitrosomonas personal
Length: 120 Spatial Dissolved Not Not described
et al. and Not simulated No apply computers |
m; Depth: structure of oxygen simulated specified qualitatively.
(1998a) 3D - Nitrobacter Comp. time not
120 m; colonies Simulated Growth was
Discrete specified
Height: 120 simulated up to 20
m days
300 | 5
days
BT varied
300 | 7
Simulated | according to the
Length: days
Picioreanu Finger- Nitrosomonas biomass growth
2D - 2000 m; Not Dissolved Not Not 300 | 11
et al. and and Not simulated Not specified rate: compact
Discrete Depth: 400 simulated oxygen simulated simulated days
(1998b) mushroom- Nitrobacter biofilms
m 300 | 18
like shape
days
350 | 31 BT varied
days according to the
Chapter 2. Literature review 34
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
350 | 71 biomass growth
days rate: mushroom
350 | shape biofilm
141
days
Length: 800
m; Depth:
3D - 250 | 50
320 m; -
Discrete days
Height: 400
m
Simulated | 12.5-
Mushroom 137.5 | High perf.
Constant - Laminar |
Eberl et al. 3D - shape: Not Not Time of comp. | Comp.
1.6 mm Single species not Not simulated Re 4.1 |
(2000) continuum pores, simulated simulated growth time not
specified 32.6
cavities, not specified
channels specified
15 | 4
days 8 or on 16
20 | 18 processors
days from a Cray
50 | 5 T3E parallel
Picioreanu Height: 500 Simulated | Laminar | BT values
2D - Not Dissolved Not days machine with
et al. m; Length: Mushroom Single species Not simulated Re 0.8 | corresponded to
continuum simulated oxygen simulated 120 | 38 128
(2000a) 2000 m shape 6.7 two Re values
days processors
150 | 6 at 400 MHz | |
days Comp. time 1
200 | 80 day
days

Chapter 2. Literature review 35

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
250 | 7.5
days
300 |
130
days
BT values
150 | 88 corresponded to
and 4 two Re values
days and two substrate
concentrations
Length: 121 Simulated |
121 |
Eberl et al. 3D - Wavy and Not Not Not
m; Depth: Single species Not simulated Static 643 Not specified -
(2001) continuum mushroom simulated specified simulated
121 m hours
shapes
Kommedal 1D - Not Not
No apply P. aeruginosa Glucose Simulated Not simulated Static No apply Not specified -
et al. (2001) continuum simulated simulated
Decay or growth
of active cells
Eberl and Simulated - according to initial
3D - Not Inert and Not Dissolved Not
Efendiev Simulated Disinfectant Static - Not specified values of BT and
continuum specified active cells simulated oxygen simulated
(2003) not specified disinfectant
concentrations at t
= 1.6-90 days
Length:
Simulated | Dissolved Laminar | CFD was applied
Xavier et al. 2D - 4000 m; Heterotrophic 75 | 40
Mushroom Simulated oxygen Simulated Not simulated Re not Not specified for solving the
(2005a) Discrete Depth: 31 bacteria days
shape and specified flow
m

Chapter 2. Literature review 36

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
Length: 500 organic
3D - carbon 300 | 30
m; Depth:
Discrete days
500 m
Length: Dissolved
2D and Simulated | oxygen
Xavier et al. 1000 m; Active and Not Not 400 | 20
3D - Mushroom and Simulated Not simulated Not specified -
(2005b) Depth: 1000 inert cells simulated specified days
Discrete shape organic
m carbon
Irregular
Simulated | One/two morphology was
Eberl and Length: 500 Moving substrates observed at
1D - Not Not Not Not
Demaret m; Depth: interface: Not specified (oxygen Not simulated Not specified t=19.12 days.
continuum simulated simulated specified specified
(2007) 500 m mushroom and Channels and
shape nutrients) clusters appeared
at t=159.13 days
100 | 7.3
days
Variation of BT
300 | 31
Length: corresponded to
days
Duddu et 2D - 200/500 Not Dissolved Not different type of
Simulated Not specified Simulated Not simulated 250 | Not specified
al. (2008) continuum m; Depth: simulated oxygen specified seeds (semi-
28.6
10/500 m circular and slab
days
biofilm)
400 | 44
days

Chapter 2. Literature review 37

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
Shared
environment
Laminar | on the SGI
Re Altix 330 with
0.32x10-5 16 processors
| 1x10-5 | (4 processors
Eberl and Simulated - Colonies
2D - Not Simulated | Not Dissolved Not 0.32x10-4 for each
Sudarsan Single species Disinfectant No apply formation occurs
continuum specified Colonies simulated oxygen simulated | 1x10-4 | simultaneous
(2008) not specified in 1.1 days
0.32x10-3 run), 1.5 GHz
| 1x10-3 | and 32 GB
0.32x10-2 main memory,
| 1x10-2 4 MB cache |
Comp. time
20-60 min
BT was reported
as fractions.
Biofilm growth
was reported at
t=3.5, 4.7 and
Laminar |
Zhang et al. 2D - Not Not Not 10.5 days,
Simulated Single species Simulated Simulated Not simulated Re Not specified
(2008) continuum specified specified specified according to
9.98x10-4
different type of
seeds (single
hump, mushroom
shape, compact
biofilm)
Simulated | Laminar | 200 | This study tested
Duddu et 2D - Length: Dissolved
Mushroom Single species Simulated Simulated Not simulated Re 0.8 - 120 Not specified several previous
al. (2009) continuum 1000/2000 oxygen
shape 10 days models,
Chapter 2. Literature review 38
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
m; Height: 250 | 4.5 considering
500 m days different Re
250 | values and the
10.7 presence/absence
days of biomass
100 | 7 detachment
days
150 | 7
days
300 | 20
Static
days
100 |
Time of Fluid momentum
21-node
Simulated | Persister and Simulated | growth: is governed by the
Cogan 2D - Not Not cluster of GPU
Hump susceptible Simulated Simulated Disinfectant: Re < 1 240 Stokes equation
(2010) continuum specified specified | Comp. time 9
shape bacteria hypochlorite hours | (viscous forces
hours
300 are dominating)
hours
Simulated | Disinfectant
Length: 1 Simulated | Persister and
Cogan 2D - Not Not Disinfectant Laminar | Not effects on biofilm
cm; Depth: Hump susceptible Simulated Not specified
(2011) continuum simulated specified type not Re 0.94 specified were simulated up
0.5 cm shape bacteria
specified to 35 hours
Competition
between
D’Acunto Organic
1D - Not heterotrophic Not Not
and Frunzo No apply Simulated carbon | Not simulated No apply Not specified -
continuum simulated and simulated specified
(2011) Ammonia
autotrophic
bacteria

Chapter 2. Literature review 39

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
AMD Opteron
3174
Length: Processors (48
Single cell Laminar | cores) and 256
Taherzadeh 12,000 m; Dissolved Not
2D No apply with moving No apply Not simulated Re 33- No apply GB RAM | -
et al. (2012) Height: oxygen simulated
tail 150 Comp. time: 6
3,000 m days using 4
cores of code
running
This study did not
Simulated |
Length: consider biofilm
Biocides:
Not specified | growth because
Zhang, 2D - 1000 m; Not Active and Not Not Not Hypochlorous Laminar |
No apply Comp. time 60 there was not
2012 continuum Height: simulated inert cells simulated simulated simulated acid, Re 20
min substrate after
1000 m glutaraldehyde,
rinsing biocide in
QAC, Nisin
experimental tests
Simulated |
Length: Not
Hump
Laminar specified
Lindley et 2D - 1000 m; shape and Dissolved
Single species Simulated Simulated Not simulated Re 1x10-3 | Time of Not specified -
al. (2012) continuum Height: effect of oxygen
|1 growth
1000 m shear
3.5 days
stress
Standard
Autotrophs:
laptops |
1D, 2D Simulated | active and
Clarelli et Length: 1 Not Not 200 | 60 Comp. time
and 3D - Hump inert Light Simulated Not simulated -
al. (2013) cm simulated specified days 1D: 3.25 min |
continuum biofilm Cyanobacteria
2D: 40 min |
cells
3D: 1.84 hours

Chapter 2. Literature review 40

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
Bacillus Colonies
Ghosh et 2D - Not Simulated | substilis: Not Not Not formation was
Simulated Not simulated No apply Not specified
al. (2013) continuum specified Colonies Active and simulated specified simulated simulated up to 24
inert cells hours
Length: 119 Biofilm and flow
63 | 1.3 were fluids with
m; Height: Intel Core Duo
hours two different
179 m CPU P9600
900 | 1.8 viscosity and both
machine at
Simulated | hours of them satisfied
2.67 GHz |
Finger- Dissolved 1000 | the Hele-Shaw
Cumsille et 2D - Not Not Re Comp. time:
Length: and Single species oxygen | Not simulated 13.7 equations. BT
al. (2014) continuum simulated simulated Negligible Several hours
1333 m; mushroom- Glucose hours corresponded to
(e.g., 1440
Height: like shape several biofilm
iterations took
2000 m 1100 | shapes: compact,
25 min for the
17.6 mushroom, finger,
finest grid)
hours semi-circular
biofilms
Length: 600
Simulated | 8 days were
m; Depth: Laminar |
Coroneo et 2D - Circular Not Dissolved Not Not required to reach
300 m; Not specified Not simulated Re not Not specified
al. (2014) continuum and finger simulated oxygen simulated specified finger-like shape
Height: 10 specified
shape of biofilm
m
Dissolved
Multi-species:
1D and oxygen | Biofilm growth
D’Acunto et Length: Not autotrophs, Not Not Not
3D - Simulated Organic Not simulated Not specified was simulated up
al. (2015) 2000 m simulated heterotrophs, simulated specified specified
continuum carbon | to 5 days
inert cells
Ammonia

Chapter 2. Literature review 41

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Model components Max. BT
System / Flow (m) Computer and
Type of Spatial
Reference Domain Bacteria Biomass EPS regime and computational Observations
model character. Substrate Biocide effect
dimensions species detachment production and Re Time of time
of biofilm growth
Simulated |
Macías- 2D - Not Active and Not Not Not Not
Mushroom Simulated Not simulated Not specified -
Díaz (2015) continuum specified inert cells simulated specified specified specified
shape
Biofilm growth
was not included:
Length: Laminar | time scale
Simulated |
Tierra et al. 2D - 1000 m; Re 5.6 | deformation and
Finger-like No apply Simulated No apply Simulated Not simulated No apply Not specified
(2015) continuum Height: 27.8 | detachment vs
shape
1000 m 111.1 biofilm growth
(seconds vs
hours)
Length: Simulated | CPU-GPU Biofilm growth
Simulated | Persister and Laminar |
Zhao et al. 3D - 1000 m; Not Dissolved Antimicrobial Not hybrid was reported for
Mushroom susceptible Simulated Re not
(2016) continuum Depth: 2000 simulated oxygen agent not specified environm. High 1, 10, 55, 27.8
shape bacteria specified
m specified perf. comp. hours
Length: 100 Organic Simulated
Multi-species:
m; Dept: Simulated | Simulated: substrate: | Flow
Jayathilake 3D - aerobic, 48 | 4.6
400 m; Mushroom deformation oxygen, Simulated Not simulated regime Not specified -
et al. (2017) Discrete anaerobic, days
Height: 100 shape and erosion nitrates, not
anoxic growth
m nitrites specified

Chapter 2. Literature review 42

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
In general, modelling evolves according to the interests and needs of researchers and consultants,
advance on numerical methods and development of software and hardware. It can be observed in
Table 2-3 that such research have followed a cyclic pattern on time. Models start with great
simplifications such as ignoring the spatial characteristics of biofilm in order to evaluate the effects
of substances like disinfectants on certain populations of bacteria. Particularly, biofilm modelling
reported in the early 90’s corresponded to one-dimension models of biofilm disinfection, and the
spatial characteristics of the biofilm were not considered (Stewart, 1994; Stewart and Raquepas,
1995). Then, other features were included in the models as progress on numerical methods,
software and hardware is being done. Colonies, finger-like, and mushroom shape biofilms are then
modelled in 2D and 3D. Models also become more sophisticated by including several species in
the biofilm according to their metabolism and resistance to biocides. Similarly, EPS production has
also been included in these models, especially when they were aimed to assess the viscoelastic
properties of biofilms and effects of shear stress. In the last case, biofilm detachment has also been
simulated. Despite of the evident evolution of biofilm modelling, mass transport by coupled flow-
biofilm is limited to laminar flow with low Re.

Biofilm modelling is computationally expensive, depending on the characteristics to be included.

Ideal models in drinking water pipes should include biofilm growth, multi-species, detachment, EPS
production, multi-substrates, disinfection, DBP formation, and mass transport under turbulent flow.
However, including all those characteristics in one model might not be feasible in reasonable times
for practical applications. Therefore, simplifications must be done. For instance, less than 60 min
of computational time have been reported for 1D and 2D models of disinfectant effects on colonies
growth (Eberl and Sudarsan, 2008), antibiotic impact on initial concentration of biomass (spatial
characteristics of biofilm were not simulated) (Zhang, 2012), growth of humps of autotrophic
bacteria (Clarelli et al., 2013), and growth of finger-like and mushroom shape of biofilms

Other 2D model of disinfection of persister and susceptible bacteria, hump-shaped biofilm, lasted
nine hours, using a 21-node cluster of GPU. More expensive computations were run for 2D models
of mushroom-shaped biofilm under laminar flow and Re<7, with eight or 16 processors from a
parallel machine; simulations lasted one day (Picioreanu et al., 2000a). Similarly, Taherzadeh et
al. (2012) used CFD to solve the flow field under laminar regime and Re=33-150 and calculate the
mass transport between bulk and an spherical cell with a moving tale. 48 processors and 6 days
were required for those simulations (Taherzadeh et al., 2012).

Regarding the effects of flow on biofilm dynamics, Lehtola et al. (2006) stated that it is commonly
believed that increasing water flow velocity can increase the bacterial numbers in biofilms due to

Chapter 2. Literature review 43

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
better mass-transfer of growth-limiting nutrients at the higher flow velocity of water, but there are
conflicting results about this matter. The studies developed by Donland and Pipe (1988) and
Ragazzo (2002) (cited by Lehtola et al. (2006)) have reported that the number of bacteria in biofilms
correlated negatively with water flow rate; shear stress appeared to have no interaction on biofilm
formation and no significant effects of flow rate were identified on biofilm thickness. As a result,
Lehtola et al. (2006) studied the changes in water quality and the formation of biofilms in copper
and plastic pilot systems in response to gradual increase of water flow (0.2-1.5 L/min) and velocity
(0.04-0.28 m/s). Pilot systems were composed by 2 parallel loops ( 50 mm), each one connected
to two parallel copper-loops ( 10 mm) and plastic-loops ( 12 mm), respectively. Water and biofilm
samples were taken, the latter one by using collectors made of pieces of pipes of each material.
Biofilm was characterized by heterotrophic plate count (HPC) and total bacteria measurements
using culture and dyeing methods. This study reported that increasing flow rate did not significantly
affect HPC in biofilms in copper pipes, but this parameter increased in plastic pipes.

Therefore, this confirmed that number of bacteria in biofilm increased with higher flow rate, resulting
from better mass-transfer of growth-limiting nutrients. This result is attributable to velocity changes
and not to nutrients concentrations and dilution changes since the experiments considered water
recirculation. However, it is necessary to examine the methods for characterizing biofilm growth
since bacteria culture has recognized limitations regarding the ability to generate accurate results.
In addition, HPC count only represent a limited fraction of the whole microbial community
(Douterelo et al., 2014a; Ren et al., 2015), when used to estimate bacterial loads in water samples.
Consequently, this study might have included non-realistic estimation of bacteria count in biofilm
and water (Lehtola et al., 2006). This study also found that biofilms detachment occurred in copper
pipes during the experiments. A recent work by Abokifa et al. (2016a), who built and validated a
model to predict THM concentrations in DWDNs including the contribution of biomass, found that
the flow velocity significantly affects the THM formation from biomass as it controls bacterial
regrowth by enhancement of mass transfer of solutes across the bulk/biofilm interface caused by
higher velocity.

In relation to the effects of antimicrobial agent on biofilms, Cogan (2008) developed a two-
dimension continuum model of biofilm disinfection, including coupled motion, external fluid, and
viscoelastic properties of the biofilm by considering this as a fluid with viscosity notably higher than
viscosity of bulk water. To determine the various model components for each time-step, the
researchers determined the fluid and biofilm velocities by solving incompressible Stokes equations;
computed the advection, diffusion and reaction of the chemical substances; and the chemical
concentrations were assumed to be at quasi-steady-state, since these equilibrate rapidly. One of
Chapter 2. Literature review 44
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
the key aspects of modelling two fluids is to define the interface between them and solving the
respective equations. Cogan (2008) used the boundary integral method (BIM) which consists of
transforming the equation governing both materials in each subdomain into a single integral
equation, whose solution is the velocity at each point in the domain. Cogan (2008) compared the
disinfection curves for varying biofilm viscosity and they found that flow regimes and high viscosity
of biofilm have very little difference in disinfection using a dynamic interface as compared to the
fixed simulations considered by Cogan (2006). Therefore, assuming a fixed interface may be a
reasonable simplification for a range of flow regimes.

It has been identified that biofilms have physiological protection, which consists of spatially
dependent nutrient consumption leading to regions of lowered biocide effectiveness. Because of
this, typical antimicrobial agents and antibiotics are most effective at killing respiring bacteria
(Costerton et al., 1995). Therefore, disinfection rate must be dependent directly on the nutrient
availability by assuming that disinfection rate is proportional to the bacterial growth rate. This delays
the action of an antimicrobial within the biofilm since the bacteria near the fluid interface consume
the nutrient leading to nutrient-depleted regions. These regions will not be susceptible to
antimicrobial agents since they have zero growth rate. This makes the process transient where
cells near to interface are also more susceptible to disinfection and they are killed quickly, nutrients
can penetrate further into the biofilm reducing the nutrient depleted zones inducing susceptibility
(Cogan, 2008).

The previous explanation agrees with the experiments run by Xue et al. (2014) with pure cells of
P. aeruginosa and P. putida in order to study reactivity between EPS and monochloramine. They
found that reactivity between protein-EPS and monochloramine is eight times higher than
polysaccharides-EPS. High reaction rate reduced monochloramine concentrations at the biofilm
surface and lower concentrations were found in deeper layers of biofilm. According to the nutrient
and oxygen gradients, cells enclosed in the interior of a biofilm may be in a dormant state (Xue et
al., 2014). According to the research of Cogan (2008) and Xue et al. (2014), modelling biofilm
disinfection considering physiological tolerance is a good approach to study resistance
mechanisms of bacteria. However, modelling DBP formation from biofilm disinfection should also
include the transformation of dead bacteria into DBPs since they are still organic matter which
reacts with disinfectant. Similarly, the biomolecules embedded in EPS reacts with disinfectant,
forming DBPs, which may contribute to increase its concentrations in bulk water. In this line,
disinfection rate may not be fully dependent of bacteria growth rate, which may lead to a
simplification of the model.

Chapter 2. Literature review 45

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
2.5 WATER DISINFECTION
Water supply history started with the access to safe water. Due to increasing contamination of
water sources, the efforts were then addressed to improve treatment processes. Chlorine was
introduced to urban drinking water supply systems for first time in the USA at the beginning of the
20th century and it has been used to control pathogenic bacteria in drinking water around the world
(Nieuwenhuijsen et al., 2000). As a result of the discovery of DBP formation from oxidation of
organic matter by chlorine and extensive research, other chemical disinfectants such as
chloramine, chlorine dioxide, ozone, and ultraviolet light (UV) have raised as alternatives to
chlorine. However the first three previous disinfectants also lead to DBP formation (Chowdhury et
al., 2009) and the last two disinfectants only work at point-of-use since they do not produce residual
concentrations in the DWDNs (Betancourt and Rose, 2004). Since a “perfect” disinfectant has not
been discovered yet, chlorine is still used widely by virtue of it is easy to use, low cost, and
especially for its residual effect, which is an important factor to protect drinking water against any
potential microbial contamination in the distribution networks (WHO, 2005). Chlorine is an oxidizing
substance and it is relatively unstable in water; therefore, chlorine decays in DWDNs since it reacts
with particulate and dissolved substances in bulk water, biofilms, materials of the components of
the network, incrustations, and sediments (Al-Jasser, 2007).

2.5.1 Chlorine reactions in water

2.5.1.1 Hydrolytic reactions

When chlorine gas (Cl2) is added to water, hypochlorous acid (HOCl) and hypochlorite ion are
formed (OCl-). These compounds are known as free residual chlorine and they are produced in two
phases:

𝐶𝑙2 + 𝐻2 𝑂 → 𝐻𝑂𝐶𝑙 + 𝐻 + + 𝐶𝑙 − (𝐻𝑦𝑑𝑟𝑜𝑙𝑦𝑠𝑖𝑠)

𝐻𝑂𝐶𝑙 ↔ 𝐻 + + 𝑂𝐶𝑙 − (𝐷𝑖𝑠𝑎𝑠𝑠𝑜𝑐𝑖𝑎𝑡𝑖𝑜𝑛)

The ratio HOCl:OCl- depends directly on pH. When dissolved free chlorine is considered, the
predominant chlorine species present in the pH 1-3 region is chlorine (Cl2), whereas HOCl
predominates in the pH 5-7 region and the hypochlorite ion is the major species present above pH
8 (Gordon and Tachiyashiki, 1991). This indicates that pH is an important variable in disinfection
processes since HOCl is a more powerful disinfectant while hypochlorite ion is weaker (Scarpino
et al., 1972).
Chapter 2. Literature review 46
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
2.5.1.2 Redox reactions
Chlorine can react with either organic or inorganic substances present in water, especially with
those containing nitrogen. Chloramines are produced rapidly when chlorine reacts with ammonium;
other non-disinfectant chlorinated substances are formed when chlorine reacts with organic
nitrogen and other substances. These reactions are slow and can last days or even weeks
(Arboleda, 2000).

2.5.2 Disinfectant decay models

Disinfectant decay is probably the process most commonly simulated in DWDNs due to its direct
relationship with other important parameters such as bacteria regrowth, cells inactivation, and DBP
formation. Water quality models including disinfectant, coupled with network hydraulics (dead/low
velocities zones, water age), are regarded as a proxy model for this inactivation and are used to
identify bacteria regrowth episodes; explain water quality complaints; select water quality
monitoring points, among others (Clark, 2015). To model the disinfection decay process,
parameters such as decay coefficients in bulk water and at pipe walls have been calculated from
experimental data and represented by first and second order kinetic equations.

Chlorine decay in DWDNs is influenced by several factors such as corrosion processes, chlorine
mass transport between bulk water and pipe walls, and reactions with both organic and inorganic
substances in bulk water and with biofilms attached to the wall of pipes, tanks, and accessories of
the system (Vasconcelos et al., 1997). A general first-order expression can be used to describe the
variation of chlorine at different residence times in the distribution network (Equation (2-20)).

𝜕𝐶
= −𝑘𝐶 (2-20)
𝜕𝑡

Here C is the chlorine concentration and k is the first-order decay constant of chlorine. Integrating
Equation (2-20) and defining C = Co at t=0, Equation (2-21) is obtained to describe chlorine decay.

𝐶 = 𝐶𝑜 𝑒 −𝑘𝑡 (2-21)

According to Vasconcelos et al. (1997), k reflects the factors mentioned previously and, as
consequence, it is site-specific and must be verified by field measurements. Therefore, chlorine

Chapter 2. Literature review 47

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
decay constant is including simultaneously the reactions in bulk water and pipes walls (Equation
(2-22)).

𝑘 = 𝑘𝑤 + 𝑘𝑏 (2-22)

Here kb is the reaction coefficient in bulk water and k w is the reaction coefficient at pipe walls.
Chorine can quickly oxidize certain substances such as sulphides, ferrous iron, manganese, and
humic material (Vasconcelos et al., 1997). The reaction rate of the remaining chlorine is frequently
well-characterized as a simple first-order decay process or, in some cases, a second-order rate
process (Clark, 1998). Pipe materials and biofilms play an important role in the quantification of k w.
The two-steps experimental test is usually used to determine k b and kw; kb is determined by testing
chlorine decay in water stored in glass flasks while overall k is calculated by testing chlorine decay
in aged pipes. The difference between k and k b results in kw (Vasconcelos et al., 1997).

2.5.2.1 Wall reaction kinetics

In relation to wall reaction kinetics, chlorine is transported from bulk water to the wall reacting
according to either first-order or zero-order kinetics. According to Vasconcelos et al. (1997), first-
order wall reaction is more appropriate when chlorine is the limiting reactant, as might be the case
with reactions involving biomolecules and EPS inside of biofilms (see more details in Section 2.7).
Zero-order wall reactions can be used to represent the case in which chlorine oxidizes immediately
some reductant as ferrous compounds and the rate is dependent on how fast the reductant is
produced by the pipe; this representation would be more likely applied to corrosion-induced
reactions. Several studies have identified the chlorine wall decay by laboratory or in-situ tests,
analysing the demand exerted by either only biofilms or all the components of pipe walls
(sediments, corrosion by-products, biofilms, loose deposits, etc.); Table 2-4 summarises such
values of kw.

Table 2-4. Disinfectant decay constants reported in experimental studies

Chlorine wall decay constant
Reference Description
(m3/g-day) (1/day)
Growth of biofilm (7 months) in polystyrene beads 5.10 5.00
with sand filtered water, ozonated water and
6.29 6.16
Lu et al. finished water. Temperature was maintained at 18
(1999) °C ±1 (reference water). Tests were run with 2.48 5.11
finished water (reference water was also used and
decay rates are much lower). 2.19 4.52
(1) CI: 0.72 – 39.36

Chapter 2. Literature review 48

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Chlorine wall decay constant
Reference Description
(m3/g-day) (1/day)
Determination of chlorine wall decay by in-situ and DICL: 3.12 (2)
laboratory experiments. In-situ tests: measurements
of chlorine decay between two points, considering PVC: 2.16 (2)
residence time, uniform pipe material and diameter.
Hallam et al.
Laboratory tests: measurements carried out in PVC
(2002)
instrument, including new pipe sections, new pipes
installed in the DWDN for three weeks to allow MDPE: 1.20 (2)
biofilm growth, and pipe sections belonging
originally to the DWDN.
Chlorine decay in bulk water and overall decay were
5.50 6.00
tested in PVC, asbestos and cast iron aged pipes. 
= 152 mm. Modelled drinking water was used in the 4.18 4.56
Buamah et al.
tests. Wall decay was calculated by the difference
(2014)
between overall and bulk water decay constant. Clo
= 0.74 / 1.44 mg/L (Water temperature = 26-30 °C; 8.37 9.12
pH = 6.5-8.5).
Growth of biofilm (10 days) in PVC reactors,  = 77, 2.30 4.59
Lee et al.
98, 145 mm, reclaimed water, disinfection with 1.63 3.27
(2014)
chlorine (Clo = 1, 2, 4 mg/L) (pH = 6.84-6.93). 1.04 2.07
Specific initial concentrations of chlorine were not reported | (2) Average | CI: Cast iron | DICL: cement-lined ductile iron | MDPE:
(1)

medium density polyethylene

Disinfectant decay constant in pipe walls has been attributed to reactions with the pipe material,
biofilms and deposits (Lu et al., 1999; Buamah et al., 2014). This demonstrates that biofilms and
deposits represent an important source of disinfectant decay, especially in smaller diameter pipes
(Lu et al., 1999; Buamah et al., 2014; Lee et al., 2014) as the interactions between the wall and
bulk water are more significant when the diameter is small. For instance, Lu et al. (1999) studied
the influence of biofilms on chlorine demand by growing biomass in polystyrene beads and
analysing S/V1 ratios. These researchers found that diameters smaller than 40 mm (1.5 inches)
and biodegradable dissolved organic carbon higher than 0.6 mg/L are more relevant for chlorine
decay. Such pipe dimensions correspond to that used in building plumbing systems. On the other
hand, Buamah et al. (2014), by testing and simulating PVC, cast iron, and asbestos aged pipes,
found that diameters of three and six inches influenced more the wall chlorine decay in comparison
to nine inches. In contrast, Lee et al. (2014) tested reclaimed treated wastewater by growing
biofilms in PVC reactors. By assessment of the S/V ratios, they found that wall chlorine decay in
=77 mm (3 inches) is about 4–5 times larger than =145 mm (6 inches).

1 Pipe inner wall surface area per unit volume of pipe

Chapter 2. Literature review 49
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
2.5.2.2 Mass transfer kinetics

Based on the transport process described in Section 2.3, Rossman et al. (1994) represented
chlorine reaction with pipe walls by a film-resistant model of mass transfer, along the pipe. The rate
at which chlorine is hydrodynamically transported to the wall is proportional to the difference
between the bulk concentration and the concentration at the wall (Equation (2-23)).

𝑑𝐶 𝑘𝑓
| 𝑚𝑎𝑠𝑠 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 = (𝐶 − 𝐶𝑤 ) (2-23)
𝑑𝑡 𝑟ℎ

Here rh is the hydraulic radius and kf is the mass transfer coefficient, which is usually expressed in
terms of Sh (Equation (2-8)).

2.5.2.3 Overall rate

Disinfectant decay is the primary water quality parameter monitored by water operators and
effective efforts have been done to represent this process by mathematical models, which can be
used in the routine operation of the DWDNs. The overall rate of chlorine loss in a pipe is the key
kinetic variable for models of DWDNs containing this disinfectant; it combines the effects of bulk
reaction, wall reaction, and mass transfer. Equations (2-24) and (2-25) show the expressions for
first-order and zero-order wall reactions, respectively (Vasconcelos et al., 1997).

𝜕𝐶 𝑘𝑤,1 𝑘𝑓
= − (𝑘𝑏 + )𝐶 (2-24)
𝜕𝑡 𝑘𝑤,1 + 𝑘𝑓

𝜕𝐶 𝑘𝑤,0 𝑘𝑓 𝐶
= −𝑘𝑏 𝐶 − 𝑚𝑖𝑛 ( , ) (2-25)
𝜕𝑡 𝑟ℎ 𝑟ℎ

2.6 DBPS IN DRINKING WATER

2.6.1 Disinfection approaches

DBPs are the result of the reaction between disinfectants (chlorine, chlorine dioxide, chloramines,
and ozone) and NOM, biofilms, anthropogenic contaminants, bromide, and iodide, during the
production and/or distribution of drinking water. According to Hrudey (2009), the publication by
Rook (1974) on the discovery of THMs in drinking water changed the perspective that water safety
was only related to waterborne diseases. Since then, extensive research has been conducted to
determine the mechanisms of DBP formation, characterize their precursors, identify different types
Chapter 2. Literature review 50
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
of DBPs formed according to several variables and evaluate exposure routes and the effects on
human health. In addition, the applicability of chlorine as the main water disinfectant has been
questioned because of DBP formation, its inefficiency for inactivation of Giardia lamblia and
Cryptosporidium (Roccaro et al., 2013), the inability to prevent biofilm formation (Wang et al., 2014),
and the incapacity to fully penetrate biofilms in order to deactivate bacteria embedded in the EPS.
As a result, other disinfectants such as chloramine, ozone, chlorine dioxide, and ultraviolet light
(UV) have been proposed but other DBPs, such as haloquinones, can be formed, which can be
even more dangerous than THMs (Chowdhury et al., 2009).

2.6.2 DBP species

NOM, precursor of DBPs, is highly complex in drinking water; as a consequence, the multiple
products of any chemical reaction with NOM are largely unknown (Weinberg, 2009). As a
consequence of the diverse nature of DBP precursors, fewer than half of the halogenated by-
products resulting from the chlorination of drinking water has been identified, and even less is
known about the by-products in water treated with other disinfectants (Weinberg, 2009). According
to Hrudey (2009), over 600 DBP species have currently been identified in chlorinated drinking
waters; they are grouped as THMs, HAAs, haloacetonitriles (HANs), haloketones, chloropicrin,
chloral hydrate, cyanogen halides, halonitromethanes, oxyhalides, bromate, aldehydes,
aldoketoacids, carboxylic acids, maleic acids, chlorophenoles, chloroanisoles, haloacids,
haloacetates, halonitromethanes, iodoacids, iodo-THMs, halo-ketones, halo-aldodehydes,
haloamides, halopyrrole, halofuranones, nitrosamines, carbonyls, volatile organic compounds,
methyl DBPs, and other halomethanes. Table 2-5 presents a summary of the chlorination DBP
groups and their distinctive functional group.

Table 2-5. Distinctive functional groups of chlorination DBPs

DBP group Distinctive functional group DBP group Distinctive functional group

1. THMs 2. Haloacetates

3. HAAs 4. Halonitromethanes

Chapter 2. Literature review 51

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 DBP group Distinctive functional group DBP group Distinctive functional group

5. Haloacetonitriles 6. Iodoacids

7. Haloketones 8. Iodo-THMs

9. Miscellaneous
chlorinated
organics 10. Halo-ketones
(Chloropicrin and
Chloral hydrate)

11. Cyanogen
12. Haloacids
halides

14. Halo-
13. Oxyhalides
aldodehydes

15. Aldehydes 16. Haloamides

17. Aldoketoacids 18. Halopyrrole

19. Carboxylic
20. Halofuranones
acids

21. Maleic acids 22. Nitrosamines

23. Chlorophenoles 24. Carbonyls

26. Volatile organic

25. Chloroanisoles compounds and
methyl DBPs

Chapter 2. Literature review 52

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 DBP group Distinctive functional group DBP group Distinctive functional group

27. Other
halomethanes

Adapted from Hrudey (2009) | X, X’, X’’: halogen atom (chlorine -Cl-, bromine -Br-, and iodine -I-) | R: hydrocarbon | R2: different
chemical combinations

Despite that diversity, only a small number has been assessed either in quantitative occurrence or
health-effects studies. The most studied DBPs are those derived from halogens such as THMs,
HAAs, HANs, and haloketones (Karanfil et al., 2008). According to Richardson et al. (2007) and
Roccaro et al. (2013), the highest occurrence among regulated and non-regulated DBPs is for
chlorate (high g/L levels), followed by chloroform, chlorite, dichloroacetic acid, dribromoacetic
acid, and trichloroacetic acide (low-to mid- g/L levels). In relation to nitrogenous and unregulated
DBPs, HAN occurrence corresponds to sub- to low- g/L levels. These DBPs can include chlorine
and bromine atoms or both of them and brominated species result more toxic and carcinogenic
than chlorinated species (Roccaro et al., 2013; Bond et al., 2014). The current research project is
focused on chloroform and DCAN, which belongs to the groups THMs and HANs, respectively, in
order to cover regulatory and toxicity aspects of DBPs.

 Chloroform

Chloroform is also known as trichloromethane, methane trichloride, trichloroform, methyl trichloride,

and formyl trichloride. Its molecular formula is CHCl 3, and its relative molecular mass is 119.37
g/mol. At room temperature, chloroform is a clear, colourless, volatile liquid with a pleasant etheric
odour. Sources of chloroform in the air can be natural and anthropogenic such as algae and fungi
production in marine and soil environments, respectively (WHO, 2004). Anthropogenic sources
include industrial production and as a result of its formation from other substances such as
chlorination of paper pulp and water for human consumption and swimming pools (WHO, 2004).
Other sources include hazardous waste sites, and sanitary landfills. Due to the volatility properties
of chloroform, inhalation is an important exposure route.

The major effect from acute (short-term) inhalation exposure to chloroform is central nervous
system depression. Chronic (long-term) exposure to chloroform by inhalation in humans has
resulted in effects on the liver, including hepatitis and jaundice, and central nervous system effects,
such as depression and irritability (USEPA, 2000). Chloroform has been shown to be carcinogenic

Chapter 2. Literature review 53

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
in animals after oral exposure, resulting in an increase in kidney and liver tumours (USEPA, 2000).
Chloroform is generally considered to be a no genotoxic carcinogen whose mechanism of action
involves cytotoxicity and regenerative cell proliferation (IARC, 1999). EPA has classified chloroform
into the Group B2, as a substance possibly carcinogenic to humans. The reference dose for chronic
oral exposure of chloroform is 0.01 mg/Kg-day (USEPA, 2016a).

 Dichloroacetonitrile

DCAN belongs to the HAN group, which includes seven species: bromoacetonitrile,
dibromoacetonitrile, bromochloroacetonitrile, chloroacetonitrile, DCAN, trichloroacetonitrile, and
iodoacetonitrile. According to WHO (2017), DCAN is by far the most predominant halogenated
acetonitrile species detected in drinking-water. Similarly, this species has also been identified as
predominant in biomass disinfection tests (Wang et al., 2012c). It has also been reported that DCAN
degrades over a period of hours or days, depending on pH and chlorine concentration (Reckhow
et al., 2001). In order to reduce the occurrence of chlorinated DBPs, water managers are switching
from chlorine to chloramine disinfection (Muellner et al., 2007), but this disinfectant increases the
nitrogenous DBPs, including HANs. In addition, HANs are also formed when disinfectants such as
chlorine, chloramine, chlorine dioxide, or ozone disinfection are applied in the WTPs (McGuire et
al., 2002). In a review work, researchers found that every HAN species induced DNA damage in
mammalian cells of Chinese hamster ovary (Richardson et al., 2007). By cell assays, Muellner et
al. (2007) tested the cytotoxicity and genotoxicity of HANs. In descending order of cytoxicity
dibromoacetonitrile (2.8 M) > iodoacetonitrile  bromoacetonitrile > bromochloroacetonitrile >
DCAN > chloroacetonitrile > trichloroacetonitrile (0.16 mM). Genotoxicity potency, associated with
DNA damage, resulted in descending order highest for iodoacetonitrile (37 M) > bromoacetonitrile
 dibromoacetonitrile > bromochloroacetonitrile > chloroacetonitrile > trichloroacetonitrile > DCAN
(2.7 mM).

2.6.3 DBP regulation

Even though a wide variety of DBPs exists, regulation or guidance values of these substances in
drinking water has been applied mainly to few DBP species of the groups THMs, HAAs and HANs
(see Table 2-6). In addition, maintaining residual disinfectant in DWDNs is required to protect
drinking water against microbiological contamination. Accordingly, the Water Health Organization
recommends residual disinfectant concentrations between 0.2-1.0 mg/L (WHO, 2008) and

Chapter 2. Literature review 54

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Colombian regulation establishes 0.3-2.0 mg/L range (Ministerio de la Protección Social and
Ministerio de Ambiente Vivienda y Desarrollo Territorial, 2007); USEPA (1998) regulates 0.8-4.0
mg/L in the United States (depending on the type of secondary disinfectant), and European Union,
England and Wales do not establish specific values for residual disinfectant (European Union,
1998; UK Parliament, 2000). However, biofilm research has proved that pathogenic bacteria grow
within biofilms, have active metabolism and are protected against disinfectants since EPS exerts a
demand, thus reducing the effective disinfectant concentration to inactivate the cells (Gomez-
Alvarez et al., 2012; Xue et al., 2013).

Table 2-6. Regulation of concentration limits of DBPs in drinking water

Regulations (g/L)
Main Group Compounds WHO USA
European UK Colombia
(WHO, (USEPA,
Union (3) (4) (5)
2017) 2006)
Chloroform 300 - - - -
Bromodichloromethane
60 - - - -
(BDCM)
THM
Bromoform (TBM) 100 - - - -
Dibromochloromethane
100 - - - -
(DBCM)
Total THMs (2) 100 100 80 200
Dichloroacetic acid
50 - - - -
(DCAA)
HAAs
Chloroacetic acid
20 - - - -
(CAA)
HAA5 (1) - - - 60 -
Dibromoacetonitrile
Haloacetonitriles 70 - - - -
(DBAN)
(HANs)
(DCAN) 20 - - - -
Source: Adapted from Chowdhury et al. (2009)
(1) HAA5 (sum of monochloroacetic acid, dichloroacetic acid, trichloroacetic acid, monobromoacetic acid, and dibromoacetic acid).
(2) The sum of the ratio of the concentration of each to its respective guideline value should not exceed 1.
(3) European Union (1998)
(4) UK Parliament (2000)
(5) Ministerio de la Protección Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial (2007)

2.6.4 DBP precursors

NOM, e.g., humic substances, is present at various degrees in all water supply systems and
constitute the major component of the total organic carbon (TOC) concentration in most waters and
it has been identified as the principal precursor in the formation of THMs and HAAs (Liang and
Singer, 2003). NOM contains both hydrophobic and hydrophilic fractions; the hydrophobic fractions
are generally composed of the higher molecular weight NOM with activated aromatic rings, phenolic
hydroxyl groups and conjugated double bonds, and are considered the major precursors of THMs
and HAAs; the hydrophilic fractions are typically composed of the lower molecular weight NOM with
Chapter 2. Literature review 55
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
aliphatic ketones and alcohols (Liang and Singer, 2003). The hydrophobic fractions of NOM exhibit
higher ultraviolet absorbance (UV254)2 and higher specific ultraviolet absorbance (SUVA), defined
as 100 UV254/DOC3. Hydrophobic fractions tend to be more reactive with chlorine than bromine,
while the hydrophilic fractions of NOM exhibit lower UV254 and lower SUVA, and are generally more
reactive with bromine than chlorine (Liang and Singer, 2003).

The presence of bromide ions (Br −) represents further challenges to the control of DBP formation
(Barrett et al., 2000). Waters without Br− mainly form chlorinated THMs (e.g, chloroform) due to
reactions between HOCl and the hydrophobic fractions of NOM. As such, a significant fraction of
the hydrophilic NOM may be left unreacted in these waters. Conversely, in waters with Br −, the
hydrophilic fractions of NOM form brominated THMs through reactions with hypobromous acid
(HOBr), while these brominated THMs may not be adequately characterized by the low SUVA or
low UV254 values (Chowdhury et al., 2009). The removal of hydrophilic NOM through coagulation
processes is often difficult because of their low molecular weights. As such, hydrophilic NOM tends
to remain in finished waters. As a result, formation of brominated THMs in finished water in the
presence of bromide ions is more likely to occur (Chowdhury et al., 2009). Additionally,
experimental evidence from chlorine disinfection of single-species biofilm suspensions has shown
that these can also be important DBP precursors resulting from the oxidation of EPS and lead to
the formation of carbonaceous DBPs and nitrogenous DBPs (Wang et al., 2012c).

2.6.5 Factors influencing DBP formation

During disinfection of drinking waters, most of the chlorine demand is exhausted by reactions with
NOM. Chlorine also reacts with various inorganic compounds in the WTPs and distribution systems
(e.g., ammonia, Fe+2, Mn+2, S-2, Br-, pipe materials, biofilms) (Chowdhury et al., 2009). It has been
shown that increasing pH, within the range 6-8, can increase THM formation (Chowdhury and
Champagne, 2008) but reducing pH can lead to an increase in HAA formation, within the same
range (Liang and Singer, 2003). Considering that pH is also an operational variable to control
dominant chlorine species, the determination of the optimum pH range for disinfection is often
necessary during the operation of water supply systems in order to balance the microbiological and

2 Ultraviolet light at the 254 nm wavelength is passed through a quartz cell containing the sample water and it provides an indication
of the concentration of organic matter, specifically those that contain aromatic rings or unsaturated bonds (double and triple) in their
molecular structures. (http://realtechwater.com/uv254/)
3 SUVA is defined as the UV absorbance of a water sample at a given wavelength normalized for dissolved organic carbon (DOC)

concentration. Weishaar, J.L. et al. 2003. Evaluation of Specific Ultraviolet Absorbance as an Indicator of the Chemical Composition
and Reactivity of Dissolved Organic Carbon. Environmental Science & Technology. 37(20), pp.4702-4708.
Chapter 2. Literature review 56
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
chemical safety of the water. Increase in contact time has also been noted to increase THM
formation; on the other hand, reduction of HAA concentrations has been found with higher retention
times (Tung and Xie, 2009). This has been associated with the biodegradation of dichloroacetic
acid by methylotroph bacteria such as Methylobacterium and Delftia (Hozalski et al., 2008; Zhang
et al., 2009). However, controlling contact time in DWDNs is often a challenge due to variable
hydraulic conditions and water demands imposed in distribution systems.

Temperature and seasonal variability has also been reported to affect THM formation, where the
formation of THMs in warmer waters or during summer seasons has been reported to be higher
than colder waters or winter months (Rodriguez and Sérodes, 2001; Wei et al., 2010; Parvez et al.,
2011). However, other studies have reported no seasonal influence on variations of THM
concentrations (Charisiadis et al., 2015). Wei et al. (2010) also analysed the temporal variation of
other DBP species, finding that HAAs, HANs, and haloketones presented the highest
concentrations in winter, while chloral hydrate were observed in autumn, and chloropicrin
concentrations were the highest in summer. This represents that control of DBP formation is highly
complex since every species seems to have a particular relationship with environmental and
operational variables.

Because organic/inorganic substances act as DBP precursors, their removal prior to disinfection
has proven to be an effective approach for reducing DBP formation potential. Treatments prior to
chlorination can partially remove NOM, which can be reinforced by using granular activated carbon,
enhanced coagulation and membrane filtration; but these may significantly increase O&M costs
(Chowdhury et al., 2007). Formation of DBPs can also be reduced by introducing alternative
disinfectants or a combination of disinfectants, including chloramine, ozone, chlorine dioxide and
ultraviolet radiation followed by post chlorination to protect water quality against microbiological
recontamination in bulk phase. However, the use of these alternative disinfectants can still lead to
the formation of the more toxic DBPs. For instance, Muellner et al. (2007) compared the toxicity
index among four DBP groups and found that genotoxicity and cytotoxicity indices were of
magnitude order of 2x104 and 1x105 for HANs, respectively, while those indices were of magnitude
order of 1x103 and 6x102 for HAA5, respectively. Higher toxicity index indicates higher toxic effects.

Taking into account that biofilms are formed in chlorinated and chloraminated distribution systems
(Gomez-Alvarez et al., 2012; Wang et al., 2014; Mi et al., 2015) and that they can potentially be
DBP precursors as well, it is necessary to develop a decision-making scheme for each water supply
system in order to balance the microbiological and chemical safety of drinking water. For instance,
Raseman et al. (2017) recommended improved decision support systems for adaptation of WTPs

Chapter 2. Literature review 57

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
to extreme weather events. Such tool may be expanded to O&M of DWDNs under normal and
climate change conditions. Protecting water sources and improving water treatment processes to
minimize the content of NOM and nutrients in drinking water seems to be the first and most urgent
procedure to be applied by water managers, since it is well known that they are the primary source
for DBP and biofilm formation, respectively. As a consequence, governmental institutions, water
companies and community in general should be prepared for increasing costs if better drinking
water quality is desired (Dearmont et al., 1998; Yorkshire Water, 2013; Hutton and Varughese,
2016). In addition, proper O&M of DWDNs is also required to avoid that external substances enter
to the system, directly or indirectly leading to increase in the content of TOC and nutrients in the
distribution system, then losing the effort made by the water operators in the WTP.

2.6.6 DBPs and public health

The DBP exposure routes of humans include inhalation, dermal contact, and ingestion by
showering, bathing, and cooking. The study developed by Chowdhury (2016) found that ingestion
route was the highest contributor of risk from THMs and HAAs; in addition, due to the volatile
properties of chloroform, inhalation while showering represents an important exposure pathway
(Villanueva et al., 2007). It is widely accepted that DBPs have negative effects on human health as
they have been identified as potential carcinogenic, teratogenic and mutagenic substances (WHO,
2008). Carcinogenic properties of DBPs refer to the ability to produce cancer tumours and can be
by either cytoxicity or genotoxicity, which are related to cell or genetic material damage,
respectively (Humpage, 2012; Kristiana et al., 2012). Teratogenicity is related to disruption of the
development of the embryo or foetus, causing birth defects. Mutagenicity is a subtype of
genotoxicity in which single DNA bases are substituted or deleted, changing the “code” contained
in the DNA (Humpage, 2012).

In line with that, DBP control in water treatment works has improved considerably, particularly in
developed countries, but many concerns still remain with regard to the formation of DBP within
DWDNs, due to the multiple operational variables involved in the system. Despite the great variety
of DBP species, only THMs and HAAs are regulated by most of the water authorities worldwide
since they are the most abundant species in drinking water (Chowdhury et al., 2009; Hrudey, 2009;
Bull et al., 2011). In contrast, THMs and HAAs are now considered as largely unrelated to public
health risk and are currently considered as primarily surrogates or indicators for other DBPs rather
than being a likely causal agent for the adverse health outcomes suggested by some
epidemiological studies (Hrudey, 2009; Bull et al., 2011).
Chapter 2. Literature review 58
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
The maximum permissible limits regulated for THMs and HAAs have been stablished following
precautionary considerations; however, epidemiological evidence found to date is not conclusive
yet to link a specific DBP species to certain type of cancer (Hrudey, 2009; Hrudey et al., 2015a).
To date, the epidemiological studies by Villanueva et al. (2007) and Cantor et al. (2010) offer the
most substantive evidence for significant association between THM exposure and bladder cancer
risk, but the causative DBP agent has not been identified yet (Hrudey et al., 2015a). Therefore,
more research efforts still must be done in order to clarify the wide spectrum of negative health
effects of DBPs.

Similarly, emerging DBPs have been increased with the changes of disinfection process and some
of them are substantially more toxic than the THMs. In this line, Hrudey (2009) states that the risk
management associated with the control of DBP formation should be addressed to reduce the
precursors of these substances, which may reduce other conceivable DBP formation and
consequently should not create an alternative DBP risk.

2.6.7 Disinfection by-products modelling

Modelling is an important tool used in hydraulic and water quality assessment in water industry.
Chowdhury et al. (2009) presented a chronological review of models for prediction of DBP formation
in drinking water, in the period 1983-2009. Table 2-7 presents the review of the models reported
by this author and was updated to the year 2017. According to this review, there are 52 studies
reporting 179 models. Few models included kinetic rates (16%) and most of them are empirical
models based either on laboratory or field data, which make them site-specific. Due to the high
complexity of the process of DBP formation, a general model to predict their concentrations for
every type of raw/drinking water, treatment, and climate has not been created yet, to the author’s
knowledge. The most simulated DBP group or species are those related to THMs and chlorine as
disinfectant followed by HAAs since these are the regulated DBPs, so there is interest on
developing tools to manage regulation compliance. Sixteen models were built for other disinfectants
such as ozone, chloramine, and chloride dioxide (see Table 2-7).

The models were constructed using commercial humic/fulvic acids and raw waters; despite of this,
the researchers applied them to predict DBP concentrations in drinking water (outlet of the WTP or
in distribution network). Only 12 models were based on drinking water concentrations, therefore
making them applicable to distribution networks. Most of the models reported in Table 2-7 consider
the main influencing factors such as contact time, pH, temperature, precursor indicator, and

Chapter 2. Literature review 59

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
chlorine dose. However, any of the models included neither the contribution of biofilms to DBP
concentrations in bulk water nor conditions of building systems, despite of the experimental
evidence on biological material leading to DBP formation (more details are described in Section
2.7). However, four models developed by Sung et al. (2000) included the variable algae. This can
be considered the first attempt to acknowledge the contribution of biological substances to DBP
formation in drinking water. After, Golfinopoulos and Arhonditsis (2002a) included chlorophyll-a as
indicator of algae presence in four empirical models.

In addition, a recent version of the popular hydraulic and water quality model EPANET named
EPANET multi-species extension (USEPA, 2008) includes the contribution of biofilms and pipes
wall to disinfectant decay caused by the demand exerted by the biofilm matrix and corrosion scales,
respectively. Then, it can be noted that the interest on contribution to DBP formation from
disinfection of biological substances in drinking water systems is gradually increasing.

Chapter 2. Literature review 60

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 2-7. Models for DBP formation
Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Minear and
River raw
Morrow (1983) Multi Data obtained
water + Br | Cl | pH | T Treatment:
cited by 1 THMs Chlorine Laboratory regression from field- > 0.90 --
commercial | NVTOC prechlorination
Chowdhury et model sampling data
humic acid
al. (2009)
Humic acid
River and lake
extracted from
waters using
Urano et al. river raw water Empirical k | pH | TOC | Treatment:
1 THMs Chlorine Laboratory historical data NR --
(1983) and model Cl | t prechlorination
and laboratory
commercial
experiments
product
Engerholm Synthetic
and water:
Amy (1983) commercial k1a | k2a | Treatment:
1 Chloroform Chlorine Laboratory NS NS NR --
cited by humic acid + TOC | Cl | t prechlorination
Chowdhury et deionized
al. (2009) water
Data of river raw
Nonlinear Model based on
Morrow and River raw Br | Cl | pH | T water and Inlet and outlet
4 THMs Chlorine Laboratory regression NR Minear and
Minear (1987) water | NVTOC finished water of the treatment
models Morrow (1983)
supply systems
Amy et al. Database Linear and
UV254 | TOC |
(1987) cited by River raw from nonlinear Treatment:
1 THMs Chlorine Cl | t | T | pH | NS 0.90 --
Chowdhury et waters laboratory regression prechlorination
Br
al. (2009) experiments models
Fulvic and K1a, K2a, K3a,
Ao | Cl | t | Field survey of 0.90
Adin et al. humic acids Mechanistic Treatment: and K4a: kinetic
1 THMs Chlorine Laboratory K1a | K2a | humic and (model
(1991) extracted from model prechlorination rate constants
K3a | K4a bromide and
lake sediment obtained by
Chapter 2. Literature review 61
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
concentrations field curve fitting to
in lake water data) the
experimental
results
Model based on
Amy et al.
Simulated total (1987).
trihalomethanes Extensive data
Database Linear and (TTHMs) were base was used
UV254 | TOC |
Harrington et River raw from nonlinear compared with Treatment: to predict
1 THMs Chlorine Cl | t | T | ph | NR
al. (1992) waters laboratory regression observed values prechlorination TTHM
Br
experiments models in a limited formation,
database or removal of
water utilities NOM, and
alkalinity and
pH changes
Malcolm Pirnie
Inc. (1992) BDCM, TOC | UV254 | Model for
Empirical Distribution
cited by 3 chloroform, Drinking water Chlorine Laboratory Cl | t | Br | T | NS NR individual
model system
Chowdhury et DBCM pH species
al. (2009)
Malcolm Pirnie
THMs, BDCM,
Inc. (1993) TOC | UV254 | Model for
chloroform, Empirical Distribution
cited by 5 Drinking water Chlorine Laboratory Cl | NH3-N | t NS NR individual
DBCM, model system
Chowdhury et | Br | T | pH species
bromoform
al. (2009)
Montgomery Three TOC | UV254 | Validation using
Empirical Treatment: Database
Watson (1993) 11 Chloroform Raw water Chlorine laboratory Br | pH | Cl | t independent 0.88
models prechlorination included the
cited by data base |T database

Chapter 2. Literature review 62

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Chowdhury et 0.8 (for work of Amy et
Br | pH | Cl | t
al. (2009) BDCM Cl/Br < al. (1987)
|T
75)
0.92
TOC | Br | Cl | (for
BDCM
t|T Cl/Br >
75)
0.82
TOC | Br | Cl | (for
DBCM
t|T Cl/Br <
50)
0.83
UV254 | TOC |
(for
DBCM Br | pH | Cl | t
Cl/Br >
|T
50)
TOC | Br | pH
Bromoform 0.86
| Cl | t
TOC | Br | pH
MCAA 0.82
| Cl | t
TOC | UV254 |
DCAA 0.97
Br | Cl | t | T
TOC | UV254 |
TCAA Br | pH | Cl | t 0.98
|T
TOC | UV254 |
MBAA 0.80
Br | pH | t | T
TOC | UV254 |
DBAA 0.95
Br | Cl | t | T

Chapter 2. Literature review 63

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Lou and Observed data
Chiang (1994) were within
River raw Regression THMo | pH | Treatment:
cited by 1 THMs Chlorine Laboratory 10.9% of the NR --
water model NVTOC | t | Cl prechlorination
Chowdhury et simulated
al. (2009) results
Treatment Treatment:
Multiple
Ibarluzea et al. plant and Flu | pH | T | prechlorination
1 Chloroform Chlorine Field regression NR 0.82 --
(1994) distribution Cl and distribution
model
network network
DOC | pH | O3
Ozekin (1994) | Br | t (for T <
cited by Empirical 20 °C)
1 Bromate NS Ozone Laboratory NR NR NS --
Chowdhury et models ABR | O3 |
al. (2009) B20 | T (for T
> 20 °C)
DOC | pH | O3
Bromoform 0.78
| Br | t
DOC | pH | O3
TOBr | Br | T (time = 0.95
24 h) Comparison
Siddiqui and River and Bench - Empirical DOC | pH | O3 with values Treatment:
6 Bromate Ozone 0.88 --
Amy (1994) groundwater laboratory models | Br | Toz reported in the prechlorination
literature
Bromate DOC | pH | Br 0.64
DOC | pH | Cl
Bromate | O3 | t | Br (0 0.68
< t < 1 h)
Laboratory-
NOM isolated Multiple DOC | NH3-N
Song et al. Batch - scaled and
1 Bromate from several Ozone regression | pH | O3 | Br | 0.93 Treatment --
(1996) laboratory literature-
water sources model IC | t
published data
Chapter 2. Literature review 64
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Chloroform Comparison 0.97
BDCM Multiple 0.86
Rathbun River raw pH | Cl | DOC with Treatment:
4 Chlorine Laboratory regression --
(1996b) DBCM water | Br experimental 0.94 prechlorination
model
Bromoform concentrations 0.78
pH | Cl | UV254 Comparison
THMs Multiple 0.98 Model based on
Rathbun River raw |t with Treatment:
2 Chlorine Laboratory regression Rathbun
(1996a) water pH | Cl | Br | experimental prechlorination
NPTOX model 0.96 (1996b)
UV254 | t concentrations
Chang et al. TOC | t | Cl 0.94
River raw Multiple
(1996) cited by TOC | t |
3 THMs water and Chlorine Laboratory regression NS 0.97 Treatment --
Chowdhury et UV254 | Cl
WTPs model
al. (2009) t | UV254 | Cl 0.95
Garcia- Multiple Comparison
Villanova et al. 1 Chloroform WTPs Chlorine Field regression T | pH with observed 0.65 Treatment --
(1997a) model values
Continued from
Garcia- Multiple Comparison
Distribution Garcia-
Villanova et al. 1 Chloroform Drinking water Chlorine Field regression T | pH with observed 0.86
system Villanova et al.
(1997b) model values
(1997a)
POX Fulvic acid Nonlinear Comparison 0.94
Huixian et al. t | TOC | Cl | Treatment:
2 extracted from Chlorine Laboratory regression with measured --
(1997) NPOX pH | T 0.92 prechlorination
lake water models values
Validation using
Clark and data from two Variable u
Commercial Bench - 2nd order Cl | u | TOC | Treatment:
Sivaganesan 1 THMs Chlorine field studies 0.71 includes kinetic
humic acid laboratory kinetic model pH | T prechlorination
(1998) performed in the rate
past

Chapter 2. Literature review 65

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
82% of
predicted values
Multi
Golfinopoulos Chla | pH | Br were found to Outlet of the
1 THMs WTP Chlorine Field regression 0.98 --
et al. (1998) | T | Cl be within ±20% WTP
model
of the measured
values
Amy et al.
(1998) cited by River raw Empirical DOC | Cl | Br | Treatment:
1 THMs Chlorine Laboratory NS NR --
Chowdhury et water model T | pH | t prechlorination
al. (2009)
The
12 kinetic
fundamental
rates |
premise of this
Concentration
Nokes et al. THMs Distribution Kinetic model is the
4 Chlorine Laboratory s of HOBr and NR NR --
(1999) intermediates system models kinetic-
HOCl |
controlled
Chlorine or
process of the
Bromine atom
THM formation
DOC | t | pH | Models based
0.90
Cl | T Validation using on data from
Database field-database Amy et al.
Rodriguez et River raw from Empirical from small water Treatment: (1987),
2 THMs Chlorine
al. (2000) waters laboratory models utilities in prechlorination Rathbun
DOC | pH | T 0.34
experiments Quebec (1996a,b) and
(Canada) Montgomery
(1993)

Chapter 2. Literature review 66

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Prediction of
possible
Type of
Prediction of Database carcinogenic
treatment |
probabilities from and non-
Geographical
Milot et al. of exceeding Distribution Quebec's Empirical carcinogenic
1 Chlorine region in None -- --
(2000) specified networks Mininstry of model effects to
Quebec |
values of the human as well
Season |
THMs Environment as to perform
Water source
epidemiological
studies
THMs NR These models
Chloroform Semi- 0.93
Sung et al. River raw OH- | Cl | k | Validated with Treatment: incorporate
4 Chlorine Field mechanistic
(2000) HAA5 water UV254 | algae monitoring data 0.74 prechlorination chlorine decay
models
TCAA 0.87 kinetics (k)
Databases
Westerhoff et River raw from WTPs. Regression DOC | Cl | Br | Treatment:
1 THMs Chlorine None NR --
al. (2000) water Laboratory model T | pH | t prechlorination
tests
Development of
Nonlinear
the model
optimization Application to a
Elshorbagy et Distribution involves a
1 THMs Drinking water Chlorine -- in a full THMs-t | Cl portion of a real NR
al. (2000) system kinetic
dynamic DWDN
approach for
DWDN
THM formation
Three species Comparison
Variable u
Clark et al. of THMs and Commercial 2nd order Cl | u | pH | Br with 0.5280- Treatment:
12 Chlorine Laboratory includes kinetic
(2001) nine species of humic acid kinetic model | pH experimental 0.9980 prechlorination
rate
HAAs concentrations

Chapter 2. Literature review 67

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Chla | pH | Br
THMs 0.52
Golfinopoulos | T | Cl
Multiple Comparison
and Chla | pH | T | Outlet of the
3 Chloroform WTP Chlorine Field regression with measured 0.51 --
Arhonditsis Cl WTP
models values
(2002a) Chla | pH | Br
BDCM 0.62
| T | Cl
Golfinopoulos Kinetic model k | TOC |
Lakes and Comparison
and based on Halogen Outlet of the
1 THMs treatment Chlorine Field with measured -- --
Arhonditsis differential concentration WTP
plants values
(2002b) equations |Q
Internal and
external
Ground and validations. The
Korn et al. Chlorine Bench - Linear pH | T | t |
1 Chlorite surface last one using 0.95 Treatment --
(2002) dioxide laboratory regresssions NPOC | UV254
sources independent
experimental
results
THMs River and lake
DBP yield
Gang et al. raw water
2 Chlorine Laboratory Kinetic model coefficient | Cl None NR Treatment --
(2002) HAAs followed by jar
|k|t
tests
0.89
TOC | Cl | t
HAAs 0.92
Treated TOC | Cl | t | T 0.80
Sérodes et al. Bench - Regression Outlet of the
12 waters prior to Chlorine TOC | Cl | t | T None 0.78 --
(2003) laboratory models WTP
disinfection THMo | TOC |
THMs 0.89
Cl | t
Cl | t 0.56
4 THMs Chlorine Laboratory pH | t None 0.53 --
Chapter 2. Literature review 68
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Multiple pH | Cl 0.58
Nikolaou et al. River raw Treatment:
regression pH | t | Cl 0.38
(2004) waters prechlorination
HAAs models pH | Cl 0.28
Comparison
Al-Omari et al. Raw and Regression t | Cl | TOC | with field data
1 THMs Chlorine Laboratory NR Treatment --
(2004) clarified water model Br | pH collected in the
DWDN
Cl | pH | TOC
THMs 0.77
Raw and |t Comparison
Regression Cl | pH | RCl |
Boyalla (2004) 3 DCAN drinking water Chlorine Laboratory with measured 0.69 Treatment --
models t
water values
TCP Cl | pH | t 0.68
Lekkas and THMs pH | t | Cl Comparison 0.87
River raw Bench - Regression Treatment:
Nikolaou 2 Chlorine with measured --
HAAs water laboratory models pH | Br | t | Cl 0.51 prechlorination
(2004) values
DOC | Cl | Br |
0.90
T | pH | t
UV254 | Cl | Br
0.7
| T | pH | t Models
DOC | UV254 | Validation using developed
Database
Raw and Multiple Cl | Br | T | pH data from two 0.81 based on Amy
Sohn et al. from bench-
16 THMs coagulated Chlorine regression |t studies Treatment et al. (1987,
(2004) scale
waters models DOC | Cl | Br | performed in the 1998) and
experiments 0.87
t past Montgomery
UV254 | Cl | Br (1991)
0.90
|t
DOC | UV254 |
0.92
Cl | Br | t

Chapter 2. Literature review 69

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
THM (pH=7.5,
T=20°C) | pH 0.92
|T
DOC | Cl | Br |
0.87
T | pH | t
UV254 | Cl | Br
0.80
| T | pH | t
DOC | UV254 |
Cl | Br | T | pH 0.85
|t
DOC | Cl | Br |
0.92
HAA6 t
UV254 | Cl | Br
0.92
|t
DOC | UV254 |
0.94
Cl | Br | t
HAA6
(pH=7.5,
0.85
T=20°C) | pH
|T
Br | DOC | O3
| pH | TIC | 0.98
TOBr NH3-N | H2O2
Br | DOC | O3
0.95
| pH | T
Validation with
Lake raw, Multiple
Uyak et al. TOC | pH | Cl data collected Treatment:
1 THMs finished, and Chlorine Field regression 0.98 --
(2005) |T from a different prechlorination
drinking water model
WTP

Chapter 2. Literature review 70

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
Lake raw, Comparison Treatment and
Toroz and Regression
1 THMs finished, and Chlorine Field TOC | T | Cl with measured 0.83 distribution --
Uyak (2005) model
drinking water values system
Tyrovola and Multiple
Treatment:
Diamadopoulo 1 Bromate Groundwater Ozone Laboratory regression Br | pH | O3 | t None 0.80 --
preozonation
s (2005) model
THMs Synthetic
Chloroform water
prepared with
Rodrigues et BDCM hydrophobic Regression Treatment:
5 DBCM Chlorine Laboratory FA | Cl | T None NR --
al. (2007) fraction of models prechlorination
fulvic acid,
Bromoform isolated from a
dam
Chloroform
DCBM
Probabilistic
DBCM model of Concentration
Uyak and Lake raw Treatment:
7 Bromoform Chlorine Laboratory THM and s of HOBr and None NR --
Toroz (2007) water prechlorination
DCAA HAA HOCl
speciation
BCAA
DBAA
THMs Multiple t | Cl | DOC | Comparison 0.87
Hong et al. River raw pH | T | Br Treatment:
3 Chloroform Chlorine Laboratory regression with measured 0.86 --
(2007) water prechlorination
BDCM models t | pH | T | Br values 0.87
Cl | SUVA | Cl
Multivariate 0.39 Treatment and
Semerjian et River raw and Laboratory | UV254
3 THMs Chlorine regression Cl | UV254 None 0.33 distribution --
al. (2009) drinking water and field
models system
t | Br | Cl 0.31

Chapter 2. Literature review 71

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Number Precursors / Model description
DBPs
Reference of Water Disinfectant Type of Applicability ** Observations
modelled Scale Variables Verification R2
models sources model
THMs 0.87
Chloroform 0.88
BDCM 0.69
Regression
HAA9 0.84
models: Comparison
Four species DOC | UV254 | 0.81 -
weight- and with an
of HAAs Raw water Br 0.85
Chen and molar-based independent
DCAN from five Chlorine and 0.62
Westerhoff 23 Laboratory models for data source: jar- Treatment --
sources not chloramine 0.62 -
(2010) HAN4 DBP tested
specified
potential wastewater 0.63
NDMA formation effluents 0.77
prediction 0.64 -
HAN4
DOC | UV254 | 0.66
Br | N 0.77 -
NDMA
0.80
Ka: global
Comparison chlorine decay
Di Cristo et al. Distribution
1 THMs Drinking water Chlorine Field Kinetic model Ka | Kb | Cl | t with measured NR rate. Kb: kinetic
(2013) system
values in field constant of
THMs formation
Mukundan and
Empirical TOC | pH | t | Distribution
Van Dreason 1 THMs Drinking water Chlorine Fied None 0.75 --
model T system
(2014)
0.34 -
T | SUVA | Cl
0.37
Multiple T | Cl 0.32
Babaei et al. Distribution
8 THMs Drinking water Chlorine Field regression None --
(2015) SUVA | Br | 0.45 - system
models
RCl | Cl 0.54
T | RCl | Cl 0.52

Chapter 2. Literature review 72

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Adapted and updated from Chowdhury et al. (2009) | ** Applicability was defined according to the current study author’s criteria and considering the precursors/water sources used in the experimental tests.
Ao: humic acid concentration | BDCM: bromodichloromethane | Br: bromide concentration | Chla: Chlorophyll-a | Cl: chlorine dose | DBCM: Dibromochloromethane | DBAA: Dibromoacetic acid | DCAA: Dichloroacetic acid |
DCAN: Dichloroacetonitrile | DOC: Dissolved organic carbon | DON: Dissolved organic nitrogen | FA: Fulvic acid | Flu: Fluorescence | HAA5: group of five species of haloacetic acids (MCAA+DCAA+TCAA+MBAA+DBAA) |
HAA6: HAA5 + BCAA | HAN4: Group of four species of haloacetonitriles | IC: Inorganic carbon | MCAA: Monochloroacetic acid | MBAA: Monobromoacetic acid | N: Ammonia, nitrite, nitrate, and DON | NDMA: N-
nitrosodimethylamine | NPTOX: Non-purgeable organic halide | NS: not specified | NR: not reported | NVTOC: Non-volatile TOC | POX: Purgeable organic halide | Q: inflow/outflow flow rate | O3: Ozone dose | RCl: residual
free chlorine | SUVA: Specific ultraviolet absorption | T: temperature | TCAA: Trichloroacetic acid | TCP: 1,1,1-trichloropropanone | t: time (contact time / water age) | THMs: Trihalomethanes | TIC: Total inorganic carbon |
TOBr: Total organic bromide | TOC: Total organic carbon | UV 254: Ultraviolet absorption at 254 nm wavelength

Chapter 2. Literature review 73

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
In relation to correlation coefficients, approximately 45% of the models reported R 2 values higher
than 0.90, 22% did not report the coefficients, and 33% had R 2 values lower than 0.90. Inclusion of
DBP contribution of pipe walls as a consequence of biofilm disinfection may improve the correlation
coefficients of empirical models, especially in smaller pipe diameters, where wall reactions become
more relevant. To improve DBP prediction may help to better assess the exposure to these
substances, which is an important input in epidemiological studies, as mentioned in Section 2.6.6.

Despite of the important efforts made to develop tools for easily predicting DBP concentrations in
the treatment and distribution of drinking water, it is unknown if the water utilities actually have used
any of the models reported in Table 2-7. Thus, this is an opportunity to bring the convenience of
using DBP predictive models to the attention of water managers and operators as a decision-
making tool to better control the formation of such substances during the treatment processes and
along the distribution networks to identify the worst scenarios in terms of possible health risks. For
researchers, it is also important to create and maintain communication channels with the water
industry in order to offer, explain, and adapt such tools to apply them to real cases and effectively
contribute to the improvement of drinking water quality delivered to customers.

2.7 BIOFILMS AS PRECURSORS OF DBPS

Biofilm formation and control have been studied by medical, dental, food and water sciences.
Flemming and Wingender (2010) pointed out that biofilm matrix is composed by microorganisms
(5%) and extracellular materials (95%), although the study developed by Fish et al. (2012) found
that 28-day old biofilms from a full scale experimental distribution system was composed by 34%
of EPS. Biofilms in DWDNs have mainly been studied from the point of view of serving as shelters
of pathogenic microorganisms (Wingender and Flemming, 2011; Simões and Simões, 2013; Fish
et al., 2016), which may be released to bulk water and reach the consumers. With regards to EPS,
these includes polysaccharides, nucleic acids, surfactants, lipids, proteins (Flemming and
Wingender, 2010), and humic acids (Fang et al., 2014); most of which can be the result of microbial
metabolism, known as soluble microbial products (Wang and Zhang, 2010b).

Then, the biofilm role as precursors of DBPs have recently been studied. Table 2-8 shows the main
features identified in experimental studies on interaction between disinfectants and biofilms, biofilm
clusters, planktonic cells and bacterial EPS. Such findings are evidencing the role of biofilms as
DBP precursors.

Chapter 2. Literature review 74

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 2-8. Main features of the interaction between disinfectant and drinking water biofilms
according to experimental studies
Study Features
Disinfection of suspended culture of P. putida yielded higher concentrations of
Wang et al. chloroform than P. aeruginosa. Nitrogen-based EPS secreted by P. putida may
(2012c) content more aromatic structures that polysaccharides-based EPS secreted by P.
aeruginosa. Aromatic structures are known as the main THM precursors.
EPS secreted by P. aeruginosa may be related to selective removal of dissolved
organic matter molecules with hydroxyl, carboxyl, and ester groups. The presence of
EPS reduced chlorine transport in biofilm. Results of viability ratio of cells suggest
Xue et al.
biofilm can be represented by a mushroom shape: viability ratio was lower at surface
(2012)
and bottom of the biofilm and higher in the middle. Lower biomass density at the
bottom generates void spaces, then promoting disinfectant transport and inactivation
efficiency.
Wang et al. THM formation potentials are higher in disinfection of planktonic cells of E. coli than P.
(2013a) aeruginosa biofilms.
The role of alginate-based EPS secreted by P. aeruginosa in chlorine disinfection is
Xue et al.
related to increasing disinfectant demand and then reducing the concentration for
(2013)
effective cells inactivation.
Wang et al. Amino acid Tyrosine yielded higher concentrations of chloroform, suggesting this
(2013b) substance may constitute an important THM precursor.
The action of typical antibiotics depends on the growth rate: disinfection of bacteria is
Xue et al. proportional to the product of nutrient consumption and the bacterial concentration.
(2014) Thus, bacteria that are exposed to higher nutrient levels are more susceptible to the
antimicrobial agent.
Increase of stiffness and decrease of biofilms outer layer thickness may be related to
consumption of EPS by disinfectants in the outer layer and the lack of EPS production
by the inactivated cells near to such layer. Such EPS layer was refilled after three
Shen et al.
months, under disinfection conditions. Higher biofilms stiffness may be due to higher
(2016)
fraction of EPS protein. No voids or channels were observed in the studied biofilms,
which were one-year old and were continuously disinfected with chlorine and
chloramine for three months.
Characterization of EPS extracted from drinking water biofilms, which were incubated
for 20 months and disinfected for four months with high and low chlorine
concentrations, indicated that EPS were conformed principally of aromatic proteins,
fulvic acid-like substances, soluble microbial proteins, and humic acid-like
substances. Such composition was determined by chlorine concentration; biofilm
Lemus Pérez
disinfected with high chlorine concentration (B1) produced more aromatic proteins in
and Rodríguez
comparison with biofilms disinfected with low chlorine concentration (B2). DBP
Susa (2017)
formation potential tests indicated that HAA were higher followed by THM and
emergent DBP (1,1-dichloro-2-propanone, 1,1,1-trichloro-2-propanone, chloropicrin,
DCAN, dibromoacetonitrile, and bromochloroacetonitrile). DBP formation potential of
HAA and chloroform were higher in B1 than B2. Authors suggests that B2 would have
profited from the low biocide effect of chlorine to adsorb more organic matter and iron.

Xue et al. (2014) investigated the influence of monochloramine reactivity with different EPS
components on the viability of biofilm and detached clusters using pure cells of P. putida and P.
aeruginosa. The authors found that reactivity between EPS and monochloramine is minimal, while
protein-based EPS reacted rapidly with this disinfectant. Disinfection of P. putida (protein-based
EPS) presented higher consumption of monochloramine because it reacted with the biofilm surface
while the middle and deeper layers were protected from disinfectant (Xue et al., 2014). However,
Chapter 2. Literature review 75
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
this study did not measure DBPs resulting from this reaction. In contrast, Wang et al. (2012c)
studied the carbonaceous DBPs and nitrogenous DBP formation from disinfection of extracted EPS
from P. aeruginosa and P. putida variants, testing five different chlorine dosages, two pH values,
six water age values, and bromide influence on DBP yields. These authors found that THM and
HAA yields were higher from chlorine disinfection of extracted EPS from P. putida compared to P.
aeruginosa. This was is associated with the content of more aromatic structures in protein-based
EPS in contrast with polysaccharide-based EPS from these two strains respectively (Wang et al.,
2012c).

Wang et al. (2013b) evaluated the influence of the major biomolecules in EPS and their chemical
compositions on DBP formation and speciation. These authors experimented with the chlorine
disinfection of extracted EPS from P. aeruginosa, P. putida, mixed species of biofilm samples
isolated from a water utility, and EPS surrogate. Their results indicated that the composition of
polysaccharide monomers in the bacterial EPS were different, since bacteria species and bacteria
growth condition may significantly affect chemical composition of EPS. They also found similar
results from Wang et al. (2012c) in relation to carbonaceous DBPs from disinfection of protein-
based EPS, identifying extremely high formation of DCAN and 1,1-dichloro-2-propanone from
isolated EPS from drinking water biofilms, likely associated with nitrogenous DBP precursors.

Additionally, Wang et al. (2013a) explored the effects of bacteria phenotype and type of materials
of drinking water pipes on DBP formation from cellular organic matter, using E. coli K-12 as
planktonic cells and Pseudomonas aeruginosa PAO1 as biofilms grown in PVC and galvanized
zinc. E. coli cells reacted for one hour with chlorine or chloramine, while Pseudomonas aeruginosa
grew on chips and reacted with sodium hypochlorite solution. Results showed that THM
concentrations were higher in bulk-cell experiment than in biofilm experiment; the opposite
occurred in HAN concentrations, which may be related to a higher content of nitrogen-based
compounds in the biofilm matrix. There were not differences of DBP concentrations between pipe
materials (Wang et al., 2013a).

In particular, considering the high content of organic nitrogen in wastewaters and biofilms of
DWDNs and the current trend of changing disinfectant from chlorine to chloramines, Yang et al.
(2010) assessed the formation of organic chloramines, nitrogenous DBPs, and intermediates
during reactions between monochloramine and organic-nitrogen compounds. This study found that
higher reaction times produced higher concentrations of nitrogenous DBPs, and that these by-
products also originate from organic chloramines and organic-nitrogen compounds. The authors
concluded that using chloramines in order to reduce THM and HAA formation should only be

Chapter 2. Literature review 76

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
attempted with carefully consideration since organic chloramines can contribute to nitrogenous
DBPs which are of great concern to public health.

Recently, Lemus Pérez and Rodríguez Susa (2017) studied in detail the characteristics of EPS
extracted from drinking water biofilms. Such biofilms were stabilized for 20 months, then were
chlorinated for four months under high (B1: 2.6 mg/L) and low concentrations (B2: 0.7 mg/L)
conditions to emulate the proximity and most remote locations in a DWDN in relation to the WTP
location. The researchers found that chlorination conditions affected the EPS composition of
biofilms, then influencing the type and amount of DBP formed. Results indicated that B1-EPS were
composed by aromatic proteins and B2-EPS were composed by humic substances. The authors
hypothesized that EPS played a protective role for B1 and were nutrient reservoirs for B2. The
results of Lemus Pérez and Rodríguez Susa (2017) are consistent with those from Shen et al.
(2016); both studies found that EPS decreased in biofilms disinfected with high chlorine
concentrations (7.5 mg/L). In addition, Lemus Pérez and Rodríguez Susa (2017) found that B1-
EPS had greater aromaticity due to proteins, which exert a protective role to cells. This agrees with
results of Xue et al. (2013), who identified that proteins reacted with both chlorine and chloramines,
while carbohydrates only reacted with chlorine.

Applying disinfectants to wastewater as part of the treatment process has also been a concern for
researchers due to the DBP formation as a result of the presence of dissolved organic nitrogen,
synthetic organic compounds, and soluble microbial products (Huang et al., 2012; Doederer et al.,
2014; Liu et al., 2017). The results of these investigations indicated that haloacetamides, THMs
and HAAs can impact water sources and arrays of chloraminating utilities for recycled water,
negatively affecting water quality for human consumption.

2.8 FURTHER RESEARCH NEEDED: IMPROVE THE UNDERSTANDING OF THE

ROLE OF BIOFILMS AS DBP PRECURSORS
Most of the research on microbial communities in real scale DWDNs have been focused mainly on
either bulk water (Henne et al., 2013; Holinger et al., 2014; El-Chakhtoura et al., 2015; Mahapatra
et al., 2015) or biofilms habitat (Kelly et al., 2014; Sun et al., 2014; Ji et al., 2015; Lührig et al.,
2015; Ren et al., 2015; Revetta et al., 2016). Few studies have simultaneously analysed the
microbiome on both habitats (Henne et al., 2012). Additionally the majority of the studies have been
conducted in temperate climate geographic regions where different pipes materials and ages were
considered (Holinger et al., 2014; Kelly et al., 2014; Sun et al., 2014; Wang et al., 2014; Ren et al.,

Chapter 2. Literature review 77

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
2015). Up to date, there are only two research works reported in a tropical country on this topic
(Mahapatra et al., 2015; Ren et al., 2015) 4.

On the other hand, in agreement with Hrudey (2009), who claims that DBP control should be
approached from minimization of precursors concentrations, several studies have evaluated, by
laboratory tests, the DBP formation from the disinfection of single-species culture and
heterogeneous biofilms, planktonic cells, and extracted EPS (Fang et al., 2010a; Fang et al., 2010b;
Wang et al., 2012c; Pu et al., 2013; Wang et al., 2013a; Wang et al., 2013b; Xie et al., 2013; Lemus
Pérez and Rodríguez Susa, 2017). Such experiments were carried out under different conditions
of pH, chlorine doses, and contact time to evaluate the influence of those variables on
concentrations and yields of carbonaceous DBPs and nitrogenous DBPs. Concentrations of
carbonaceous DBPs from disinfection of cellular material have been reported in the range of 2-150
g/L and 0.1-5.0 g/L for nitrogenous DBPs (Fang et al., 2010a; Fang et al., 2010b; Wang et al.,
2012c; Wang et al., 2013a); and carbonaceous DBPs yields have been reported in the range of
0.53-105 g/mg C and 0.1-5.0 g/mg C for nitrogenous DBPs (Hong et al., 2008; Wang et al.,
2012c; Wang et al., 2013a; Wang et al., 2013b). It is important to highlight that the UK and
Colombian regulations establish 100 g/L and 200 g/L, respectively as the maximum
concentrations of total THMs in drinking water (UK Parliament, 2000; Ministerio de la Protección
Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial, 2007), which represents that
disinfection of cellular materials may result in concentrations of THMs higher than the regulated
limit.

In summary, biofilms in DWDNs have mainly been studied from the point of view of serving as
shelters of pathogenic microorganisms (Wingender and Flemming, 2011; Simões and Simões,
2013; Fish et al., 2016), which can be released to bulk water and reach the consumers, if the proper
hydraulic conditions occurred. Investigating the role of EPS in the biofilm formation and
development has also been a crucial step to understand the growth, attachment, protection, and
detachment processes of the biofilm matrix (Tsuneda et al., 2003; Wang et al., 2012b; Xue et al.,
2012; Xue et al., 2013; Shen et al., 2016). Disinfectant demand by biofilms and the consequent by-
products have been recently studied (THMs, HAAs, and HANs), and the reactive components of
the biofilms matrix (cells and biomolecules) with disinfectant have also been identified (Wang et al.,
2013b).

4Research of Mahapatra et al. (2015) applied culture methods by incubation in tubes and subcultures in MaConkey
agar.
Chapter 2. Literature review 78
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
DWDNs are complex systems where multiple biological, chemical, physical, and hydraulic
interactions may occur simultaneously. Thus, further research is required to define the role of
biofilms into the formation of DBPs in DWDNs. This is especially important if it is noticed that
epidemiological studies have not produced conclusive results on the relationship between DBPs
and negative effects on human health (Hrudey, 2009; Bull et al., 2011). The most substantive
evidence was offered for bladder cancer risk and THM exposure (Villanueva et al., 2007; Cantor et
al., 2010) but it is not strong enough to classify these DBPs as causal agents. Additionally, there
are many uncertainties in medical and genetic aspects of human bladder, which make difficult to
study such associations (Hrudey et al., 2015a). On the other hand, exposure assessment is another
limitation of epidemiological studies, since they are usually based on historical data of water utilities,
which only includes routine monitoring of regulated DBPs.

Considering that linking bladder cancer and THMs have been studied but there are not conclusive
results yet, it is logically expected that the information of association between other types of cancer
and other DBP species (among over 600 species) is very limited. Due to the study of DBP effects
on human health is very complex and involves several disciplines like medicine, genetics,
epidemiology, and water engineering, several knowledge gaps still remain and must be addressed
to elucidate the real causal DBP agents of negative effects on human health. This may require an
important investment of time and money (Hrudey et al., 2015b). In this line, biofilms must be studied
from the point of view of its role as DBP precursors and, obviously, from the acute risk they
represent for being potential reservoir of pathogens. In general, the most distinguished
characteristic of the biofilms is that their presence in DWDNs is undesirable and effective
mechanisms to minimize their growth and/or to remove them are still required.

With regards to mathematical modelling, biofilm growth, detachment, and disinfectant penetration
have been simulated in drinking water in order to find the strategies to minimize its presence in
water systems. In this case, the modelling includes the transport of dissolved substances
(disinfectant, substrate, and nutrients) from bulk water towards biofilm matrix, and some of them
included EPS production by cells. In relation to DBP modelling, most of the models developed to
date are empirical, few of them include kinetic parameters, and are site-specific. It is unknown if
such models are or were actually used by water utilities. Furthermore, DBP and biofilm modelling
approaches have been clearly addressed separately. In this line, the current research project
intends to close the gap existing in the understanding of the role of biofilms as DBP precursors in
drinking water, according to biofilm, flow, and water quality characteristics. Furthermore, the role of
pathogen reservoir in a real-scale DWDN located in a tropical-climate country was also studied.

Chapter 2. Literature review 79

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
This was done by characterizing the bacterial communities in bulk and biofilms in a tropical-climate
DWDN, characteristics which have been scarcely explored together; exploring the relationship
between methylotrophic bacteria and THM concentrations in bulk water; and developing a
numerical model based on the physics of the system by the hydraulic, chemical, microbiological
and transport processes (biofilm (cells and EPS) ↔ bulk water) occurring in drinking water pipes.
Such model may represent a progress on the predictive tools, which may improve the exposure
assessments of DBPs and highlight the role of biofilms as DBP precursors and the need to minimize
its growth in DWDNs. The model may also enable to water operators and managers to take
practical decisions to reduce DBP concentrations and to appropriately control biofilm formation.

Chapter 2. Literature review 80

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 3 FIELD ASSESSMENT OF BACTERIAL COMMUNITIES AND
THEIR RELATIONSHIPS WITH ENGINEERED FACTORS

3.1 INTRODUCTION

The previous chapter explained the terms physical, hydraulic and water quality integrity in DWDNs
and their interrelations. A summary of recent studies applying molecular analysis to DWDNs
supplied with treated surface water was presented as well. The recent approach of molecular
methods applied to study microbial communities in drinking water systems was also described. A
review of empirical models for DBP prediction in drinking water and their main characteristics were
also discussed in Chapter 2. The current chapter presents the field assessment of bacterial
communities of biofilm and bulk water from a real-scale DWDN located in a tropical-climate city;
characteristics that have scarcely explored together. Water and pieces of pipes were collected in
nine sampling points; biofilm samples were scrapped from the internal walls of the pipes; and DNA
were extracted from bulk water and biofilms. Additionally, drinking water was characterised by
physico-chemical parameters such as temperature, pH, chlorine, total organic carbon (TOC), and
total trihalomethanes (TTHMs).

This chapter also presents the analysis of statistical correlations between microbiological [unit dry
biomass, TOC, relative abundance (RA) of bacterial communities, and diversity and richness
indices], physical (pipe age and material), hydraulic (water age), and water quality parameters (pH,
temperature, free chlorine, and TTHMs); and explores the relationships between bacteria and
TTHMs. The results reported by previous published investigations are also discussed and
compared in the current chapter. This study aimed to characterise the physical properties, water
chemistry and microbial communities of the DWDN to explore the relationships between biotic and
abiotic factors, and to further understand the involvement of microorganisms in processes such as
DBP formation. This new knowledge is needed to inform operational strategies and to ultimately
protect public health.

3.2 METHODOLOGY

3.2.1 Data collection from Colombian water companies

Initially, in order to identify the O&M framework of DWDNs in Colombia, this study intended to apply
a survey to Colombian water companies. A questionnaire was created and sent to 22 companies,
Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 81
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
which supply drinking water to the capitals of provinces in the country and to 12 companies
corresponding to the second largest cities of the provinces. However, only two of these water
companies completed the questionnaire, which are named WC1 and WC2 in this document. Due
to the low response of the Colombian water utilities, the current study was focused on data
collection from the DWDN of the city of Cali. The main findings of the completed questionnaires are
presented below.

Both water companies are considered large utilities since they have more than 2,500 customers,
according to the Colombian Superintendence of Public Services (Superintendencia de Servicios
Públicos Domiciliarios, 2013). The water sources are rivers for both WC1 and WC2, which
continuously supply potable water to their respective city. In relation to disinfection, both utilities
apply chlorine secondary disinfection and only one of them uses gaseous chlorine for primary
disinfection. In relation to the distribution network, both networks are operated by gravity and one
of them also combines it with pumping. The water consumption is measured by micrometres and
none of the networks include re-disinfection stations. PVC is the most common pipe material in
both networks and asbestos is also present. Both water companies are building a validated
hydraulic model, therefore such models are not used yet for simulating water quality.

In relation to drinking water quality, the maximal, minimal and average of the parameters of sanitary
interest of year 2014 reported by WC1 and WC2 were compared to the Colombian regulation
(Ministerio de la Protección Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial,
2007). Maximal values of free residual chlorine, turbidity, nitrites, and total iron exceeded the upper
permissible limit established by the Colombian standards. Only WC1 reported monitored
concentrations of THMs, being the maximal values equal to 200 g/L. Finally, with regards to O&M
of the distribution networks, few data was reported by both water companies. The indicator of
approximately 0.8 leakage repairs per Km of network was found for both WC1 and WC2 for the
year 2014.

3.2.2 Area of study

The present study was carried out in a DWDN in Colombia (see Figure 3-1), in the city of Cali,
located in the province Valle del Cauca, at 995 meters above sea level, with an annual average

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 82
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
temperature of 24.5 °C (23.8-25.1 °C5) and an average annual precipitation of 1425 mm (1048.8-
1742.7 mm6) (IDEAM, 2017). The water supply system is operated by one utility and is composed
of four sub-networks originated from four surface water sources and five WTPs, which are
described in Table 3-1. These sub-networks operate by gravity, pumping, or by a combination of
both. In total, the distribution network comprises 2977 Km of pipelines (SUI, 2017), 10 service
reservoirs, 28 storage tanks, and 19 pumping stations in order to deliver potable water to 2,400,653
people7 (CCC et al., 2016; DANE, 2017).

Figure 3-1. Location of the study area

5 Period 1981-2010
6 Period 1981-2010
7 Calculated from a coverage of drinking water supply of 99.2% for 2015 (CCC, C.d.C.d.C. et al. 2016. Cali cómo

vamos. Informe de Calidad de Vida en Cali 2015. Santiago de Cali.) and projected population for Cali in 2017 of
2,420,013 people (DANE. 2017. Proyecciones de población. [Online]. [Accessed 24th April]. Available from:
www.dane.gov.co)
Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 83
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 3-1. Water sources and treatment description of the water supply system of the city of Cali
Sub-network Source water WTP Treatment description
1 Pance River La Ribera Filtration in multiple phases
Conventional treatment, including the
2 Meléndez River La Reforma option of direct filtration for raw water with
low turbidity.
3 Cali River Río Cali Conventional treatment
Río Cauca Conventional treatment and additional
4 Cauca River specific processes such activated carbon
Puerto Mallarino adsorption.

Figure 3-2 shows the fraction of the total length of the network corresponding to each pipe material 8.
The DWDN of the city of Cali has mainly been made by PVC (53.3%) and asbestos cement
(30.4%), which represents an important fraction considering that asbestos cement becomes
deteriorated under prolonged exposure to aggressive water due to its ion content such as chloride
and sulphates and bio-deterioration due to biofilm growth (Wang et al., 2011). This may lead to
structural failures of pipes (WHO, 2008), which then increases the leakage rate, therefore raising
the vulnerability of the DWDS. Other pipe materials found in the Cali DWDN are cast iron (9.5%)
and, in less proportion, ductile iron, CCP (concrete cylinder pipe), plastic and carbon steel (6.8%
all together).

Figure 3-2. Fractions of pipe materials in the DWDN of the city of Cali

8 EMCALI EICE ESP, 2014, personal communication, 10th December

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 84
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Sampling was carried out in the sub-network 4, which is the largest one and supplied with treated
water from the second biggest Colombian river by conventional processes including activated
carbon adsorption, primary chlorine disinfection, coagulation, sedimentation, filtration, and
secondary chlorine disinfection. The main treatment facility of this sub-network has two open-air
clarified-water reservoirs to be used as a contingency for supplying drinking water (DW) when
turbidity of raw water is higher than 1,000 NTU. When surface water exceeds such turbidity limit,
the facility work must close the water intake and operate with the reservoirs up to 9 hours. If raw
water does not reach turbidity levels lower than 1,000 NTU after this time, water supply is
interrupted for almost 80% of the served population. The same situation occurs when dissolved
oxygen of raw water in the source drops below 3.0 mg O 2/L and a decreasing tendency of this
parameter has been previously identified in comparison to the normal concentrations in a range of
time of hours. More details about this and other events when water supply is suspended in the
studied network is presented in Section 3.3.1.

3.2.3 Preliminary activities

First, water age and water level data in service reservoirs were collected, processed and analysed.
Raw water age data was delivered by the local water utility in Excel format and was processed in
the software ArcMap® 10.2.2, as it is described in the next section. Database of water levels in five
service reservoirs was also in Excel format, provided by the local water company, corresponding
to the months February and March 2015. The sampling campaign was carried out during this period
of time. Water level in service reservoirs allowed determination of how the DWDN operated during
sampling activities, since if the services reservoirs are empty, the network is supplied entirely by
the WTPs. Otherwise, if the service reservoirs still have water, they are supplying certain sectors
of the network and water age can be higher.

3.2.4 Water age data

Water age was determined from a hydraulic model of the sub-network and provided by the local
water company. This model was mainly implemented in the software Infowater 11.5 and EPANET
2.00.12 is sometimes used. Currently, the subnetwork 4 is being divided by pressure zones in order
to proper manage pressure fluctuations and reduce leakages rate and water loss. The Colombian
standards for DWDNs stablishes 147 KPa as minimum pressure at the household connections

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 85
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
(Universidad de los Andes, 2010). Therefore, the water company expects to maintain this pressure
level in the network by the physical modifications they are currently carrying out. The model was
calibrated in the year 2007 and a new calibration is planned since several operative changes have
occurred since then9.

In order to allocate a water age value to every sampling point, Thiessen polygons were built by
using the software ArcMap® 10.2.2; water age map is included in the Appendix 9-A. This method
has the unique property that each polygon contains only one input point, and any location within a
polygon is closer to its associated point than to the point of any other polygon. The line between
two points is located in the middle of the distance between them. Therefore, each sampling point
is located inside of a polygon, so respective water age value is well defined. According to the
histogram of the water age data (see Figure 3-3), these were classified in four intervals as it follows:

1. Low water age: 0 – 8.5 hours

2. Medium water age: 8.5 – 13 hours
3. High water age: 13 – 68 hours
4. Very high water age: 68 – 437 hours

Figure 3-3. Water age ranges for the sampling sub-network

9 Rojas Ramírez, 2017, personal communication, 28th March. E-mail: [email protected]

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 86
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
3.2.5 Sample collection

Sample collection procedure is illustrated in Figure 3-4. Samples were taken from nine sites over
a 3-weeks period and their location is included in Figure 3-5. In each case, pipe sections were
collected during leakage repairs because the normal operation of the DWDN cannot be interrupted
for this kind of activities. This approach for taking biofilm samples is dominated by randomness due
to O&M staff must prioritize the daily scheduled repairs according to the size and type of affected
population (e.g., big neighbourhoods, areas with hospitals and education institutions). In the case
that researchers want to evaluate more strictly engineered variables such as water age, pipe age,
pipe materials, type of network component (pipelines, valves, water meters), among others,
sampling campaigns must be carried out in a long term. According to the researcher criteria, nine
samples of both bulk water and biofilms was an acceptable amount to characterize the bacterial
communities in both habitats and their relationships with other engineered and physico-chemical
variables. Additionally, field assessment like the current one described here also allows suggesting
practical recommendations to the water operators to preserve the drinking water quality to the point
of use.

Step 1 Step 2 Step 3

Step 4 Step 5 Step 6

Figure 3-4. Procedure for collection of water and biofilm samples

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 87
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 3-5. Location of sampling points in the city of Cali

First, bulk water samples were taken in the appropriate bottles according to the physico-chemical
analysis of interest (step 1); 6 L of DW were collected in sterile plastic bottles for DNA extraction.
Water samples were collected in the nearest household; the taps were flushed for 5 min before
storage in the respective bottles. Then in-situ parameters were measured (step 2). After this, water
supply was interrupted by closing valves in order to isolate and empty that part of the network; this
allows to remove the broken piece of pipe and install the new pipe section (step 3). The broken
pieces of pipe were washed with sterile water to remove portions of soil attached to pipes sections
(step 4). The internal surfaces of pipes were also washed with sterile water to remove loose
particles. The pipe sections were wrapped in polythene to be transported at 4 ºC to the laboratory
(step 5) for subsequent biofilm removal and DNA isolation (step 6). Each sampling point was
characterized by water age and pipe characteristics (material, age, and diameter).

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 88
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
3.2.6 Physicochemical analysis

In-situ water parameters were measured as follows: (a) Temperature was measured by a bulb
thermometer; (b) pH by portable meter kit (HQ40d HACH Cat. No. HQ40D53000000, Loveland,
CO) coupled to a pH electrode; and (c) total and free chlorine by the DPD method using a HACH
colorimeter II (Cat. No. 58700-00, Loveland, CO). Quantification of Total Organic Carbon (TOC)
and total THMs (TTHMs) in bulk water was carried out by an accredited laboratory following
standard methods (5310B and 6232B, respectively) (Eaton et al., 2005). Equipment for TOC and
TTHMs analysis were total carbon analyser (Shimadzu TOC 5050A, article number 3750 K3-2,
Columbia, MD) and gas chromatograph (HP 5890, Wilmington, DE and Agilent Technologies
7890B, Santa Clara, CA.), respectively.

The standard method 5310B for TOC measurement is based on high-temperature combustion in a
total carbon analyser. The water sample is homogenized and diluted and a micro-portion is injected
into a heated reaction chamber packed with an oxidative catalyst. The water is vaporized and the
organic carbon is oxidized to CO2 and H2O. The CO2 from oxidation of organic and inorganic carbon
is transported and measured in the carrier-gas streams. Since total carbon is measured, inorganic
carbon must be removed by acidification and sparging or measured separately and TOC is
obtained by difference (Eaton et al., 2005).

The standard method 6232B for TTHM measurement is based on liquid-liquid extraction in a gas
chromatograph; and it is applicable for measuring chloroform, bromodichloromethane,
dibromochloromethane, and bromoform in drinking water. The sample is extracted once with
pentane and the extract is injected into the gas chromatograph equipped with a linearized electron
detector for separation and analysis. In this equipment, the sample is carried by a gas stream to a
separation tube known as the “column” (Eaton et al., 2005).

TOC and dry-biomass content were measured in biofilm samples by scrapping a defined area on
the pipe surface of 75 cm2 in triplicate, in order to obtain enough biomass for both parameters and
for being able to apply descriptive statistics in the case of dry-biomass parameter. For TOC
measurement in biofilms, scrapped biofilms were resuspended in 250 mL of deionized water. For
dry biomass, scrapped samples were dried at 105 °C, for 24 hours and unit dry biomass was
calculated by dividing dry biomass by the scrapped area. Due to the presence of a high amount of
tubercles in the cast iron pipe of sample point 2, unit dry biomass was not calculated since the
scrapped area could not be determined (see Figure 3-6).

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 89
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 3-6. Inner walls of cast iron pipeline of sampling point 2

3.2.7 Molecular methods

Previous of taken the biofilm samples from pipe walls, the inner walls were washed again with
sterile water to remove any extra loose particles. Biofilm samples were collected by scrapping in
triplicate using a sterile frame with area equal to 25 cm 2 and a sterile spatula. Triplicates allowed
to calculate median value of RA of species. DNA isolation was carried out immediately using the
Power Biofilm DNA Kit (MoBio Laboratories, Carlsbad, CA) according to the manufacturer’s
instructions. In total 6 L of DW were filtered for every sampling point (2 L for each triplicate) through
nitrocellulose filters (0.22 m pore-size); filters were further processed for DNA extraction using
the Power Water DNA Kit (MoBio Laboratories, Carlsbad, CA) according to the manufacturer’s
instructions.

Sequencing refers to the determination of the sequence of the nucleobases (adenosine, guanine,
adenine and thymine), which compose the RNA and DNA of any cells. Sequencing of DNA
extracted from biofilm and water samples was performed by Illumina MiSeq Technology using the
Illumina PE MySeq reagent Kit v3 according to the manufacturer's guidelines (Illumina, USA) and
performed by the Molecular Research DNA Lab (Shallowater, TX, USA). 2-5 ng/L of DNA per
sample (n=53) was used for amplification (no replicates per sample were generated) and the V4
variable region of the 16S rRNA gene was amplified using primers 515F/806R (Caporaso et al.,
2011). Sequence data were processed using Mr DNA analysis pipeline (www.mrdnalab.com, MR
DNA, Shallowater, TX). In summary, sequences were merged, depleted of barcodes and primers,
sequences < 150 bp and with ambiguous base calls were removed from further analysis.
Sequences were denoised and chimeras removed. Operational Taxonomic Units (OTUs) were
Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 90
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
defined by clustering at 3% divergence (97% similarity) and were taxonomically classified using
BLASTn against a curated database derived from Greengenes, RDPII and NCBI (DeSantis et al.,
2006) (http://www.ncbi.nlm.nih.gov/, http://rdp.cme.msu.edu).

The total number of reads generated per sample ranged between 7780-304912 and between
13759-238406, for biofilm and bulk water samples, respectively. The number of reads that passed
quality scores ranged between 7240-256972 for biofilm and between 10257-101379 for bulk water
samples. The data set (number of reads per sample) was not normalised or rarefacted to assess
alpha-diversity, in order to avoid losing information from potential important sequences (McMurdie
and Holmes, 2014).

3.2.8 Data analysis

Physico-chemical data was processed and analysed by descriptive statistics such as minimum,
maximum, average, median, quartiles, coefficient of variation (CV), and number of data. The alpha-
diversity of the samples at 97% sequence similarity cut off was analysed by Margalef and Shannon
community richness and diversity indices, respectively, which were calculated with Primer6
software (PRIMER-E, Plymouth, UK). The alpha diversity was selected because the samples were
collected in the same locality, which is supplied by the same surface water source, which is treated
by two WTPs. The medians and means of such indices were statistically compared by t-test and
Mann Whitney U test using the software IBM SPSS Statistics 21. The associations of RA at species
level (considered at 97% sequence similarity cut of) with environmental factors and characteristics
of the sampling points were determined by multidimensional scaling (MDS), by means of Bray-
Curtis similarity metrics, and analysis of similarities (ANOSIM) using Primer6 software. Spearman
correlations were applied to determine the relationships between biofilm parameters and bulk water
physicochemical factors; Shapiro-Wilk tests were run in IBM SPSS Statistics 21 to determine
normal distribution of variables; results of normality tests are presented in Appendix 9-B. All
statistical results were contrasted with significance level equal to 0.05.

With regard to the analysis of bacterial communities in biofilm and water samples, it was based on
the analysis of similarities using the Bray-Curtis coefficient (S), followed by hierarchical clustering
and ordination of samples by MDS, and finally testing for differences between groups of samples
by ANOSIM test. Figure 3-7 shows the stages in the multivariate analysis based on similarity
coefficients (Clarke and Warwick, 2001).

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 91
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 3-7. Stages in the multivariate analysis based on similarity coefficients

 Similarities between samples: percentage of RA of species identified in every sample were

transformed by square root method in order to build a triangular matrix of the similarities
coefficients of Bray-Curtis, so these can be ordered (ranked) for further analysis. Calculation of
the expression (100-S) allows obtaining the dissimilarities matrix (Clarke and Warwick, 2001).

 Hierarchical clustering: similarities matrix is the start point for building a dendrogram, followed by
successively fuse the samples into groups and the groups into larger clusters with the highest
mutual similarities and the gradually lowering the similarity level at which groups are formed
(Clarke and Warwick, 2001).

 MDS: cluster analysis is often best used in conjunction with ordination methods since agreement
between the two representations strengthens reliability of adequacy of both. A MDS graph
represents the configuration of the samples, in a specific number of dimensions, which attempts
to satisfy all the conditions imposed by the rank of the (dis)similarity matrix. The construction of
a MDS involves a numerical algorithm to produce a sample map whose inter-point distances
have the same rank order as the corresponding dissimilarities between samples (Clarke and
Warwick, 2001).

 ANOSIM: it is a multivariate test for the null hypothesis (Ho) “there are not differences in
community composition at the sampling sites” which is based on the statistical parameter R (-1
< R < 1) whose calculation is based on the rank of similarities. If R=1, all replicates within sites
are more similar to each other than any replicates from different sites. If R=0, the null hypothesis

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 92
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 is true, so that similarities between and within sites will be the same on average. To test the
statistical significance of R, it was verified that the probability that Ho is true was less than 5%
(p-value) (Clarke and Warwick, 2001).

3.3 RESULTS

3.3.1 Suspension of drinking water supply in the studied network

The Colombian regulation (Superintendencia de Servicios Públicos, 2006) stablished the obligation
to report, by the utilities, information related to drinking water supply, sewage, and cleaning services
to SUI (Sistema Único de Información for its Spanish abbreviation) (SUI, 2017). A piece of the
information required by the SUI is the suspension of the drinking water supply, which must be
described by type of event, number of customers affected by each event, and duration in hours of
the event. The description of the code used to classify the type of event is shown in Table 3-2 and
Figure 3-8 presents the characterization of these events. Unfortunately, the description of the type
of event does not allow identifying which ones correspond to leakage repairs, operation interruption
of the WTPs, pipeline replacements, among others.

Table 3-2. Description of the code used to classify the type of suspension of drinking water supply
in Colombia

Code Code meaning

1 Suspension of potable water supply due to maintenance with previous notification to customers
2 Suspension of potable water supply due to repairs with previous notification to customers
Suspension of potable water supply due to operative rationing with previous notification to
3
customers or non-controllable rationing
4 Suspensions of potable water supply without previous notification to customers

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 93
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 3-8. Characterization of events of suspension of drinking water supply in the DWDN of the
city of Cali

The first three types of events of potable water suspension are related to maintenance, repairs and
operative activities with previous notification to customers. The type of event No 1 is the most
common one (65.6%) and its influence is more notably on the total duration of the water supply
suspension (62.3%) rather than on the amount of affected customers (3.9%). The type of event No
2 is associated with repairs but the code description does not specify what kind of repairs. In
general, this event was not frequent (8.8%), affected few customers (13.8%) and the total time of
water supply suspension was also short (8.1%). The type of event No 3 is related to operative and
non-controllable rationing and its occurrence was very low (0.07%); therefore its impacts on
affected customers and total duration of events are also low (0.01% for both criteria). The type of
event No 4 refers to suspensions of potable water supply without previous notification to customers.
This impacted a high amount of customers (82.3%) but the amount of events (25.6%) and the total
time of suspension of potable water supply (29.6%) were relatively low.

By grouping the first three types of events, which include previous notification to customers, most
of the events (74.4%) and the highest total time of suspension of potable water supply (70.4%) are
linked to this group; the highest amount of affected customers (82.3%) is due to suspension of
potable water supply without notification to customers. This is important if it is considered that
previous notification to customers can lead to increase the water storage at households as a
contingency, then water demand increases, and velocity and shear stress may raise suddenly and

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 94
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
being higher than the values found under normal operation of the DWDN. Such hydraulic changes
may lead to biofilm detachment; the consequences of this were already discussed in Chapter 2.

In particular, suspension of drinking water supply associated with deterioration of raw water (high
turbidity and low dissolved oxygen) are specially important due to its relation to extreme weather
events and climate change (Montoya et al., 2011; Khan et al., 2015). The latest management report
of the local water company of the city of Cali (EMCALI EICE ESP, 2016) described the evolution of
the events related to decrease of dissolved oxygen in raw water, between 1 st January 2000 and
25th September 2016, and the impact of the clarified water reservoirs built in the main water
treatment of the system. The data is presented in Figure 3-9; the local water company has named
this type of event as “high contaminant load” due to this causes decrease of dissolved oxygen.

Figure 3-9. Evolution of events of suspension of WTP operation in the period 2000-2016

Before the clarified water reservoirs were built and operated in 2009, all the events of high
contaminant load in raw water caused suspension of operation of the WTP from year 2000;
therefore, customers did not have access to drinking water and hydraulic integrity of the network
was lost during these events. From 2009, the WTP was able to operate by using such stored water,
guaranteeing the supply of potable water to customers. Although the reservoirs are an acceptable
contingency practice, it can be noted in Figure 3-9 that, from year 2012, the number of events
causing suspension of WTP operation has increased from once until a maximum of 13 times in
2015. This allows inferring that the duration of the events of high contaminant load is raising since

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 95
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
the reservoirs are designed to operate for up to 9 hours. This again is causing suspension of
drinking water supply to customers and loss of the hydraulic integrity of the DWDN. The amount of
events has also increased from a minimum of 10 events in 2000 until a maximum of 41 in 2013,
with fluctuations between this numbers in the assessed period (Figure 3-9).

3.3.2 Water levels in service reservoirs in the DWDN

The sub-network 4 has four pairs of service reservoirs and one single reservoir located in strategic
points of the city (higher topographic areas) to compensate the pressure in the network (see Figure
3-5); therefore, the network can be supplied partially by both the WTPs and the reservoirs during
high-water demand times. Figure 3-10 shows the variation of water levels in the service reservoirs
of the sub-network 4 and Table 3-3 presents the operational hydraulic conditions of the sub-network
4 during sampling activities.

(a)

(b)
Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 96
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (c)

(d)

(e)
Figure 3-10. Water levels in service reservoirs during sampling activities (a) La Campiña (b) Normal
(c) Siloé (d) Nápoles (e) Ciudad Jardín

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 97
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Table 3-3 presents the hydraulic conditions of the studied subnetwork during the sampling
campaign. Four pairs of service reservoirs were empty during most of the sampling activities; Siloé
reservoir was supplying its respective area during six sampling journeys but just two of the nine
sampling points are probably located in that area (points 3 and 7 - San Judas); in addition, the
supply region of service reservoir Siloé is not well defined, according to water company information.
Moreover, Ciudad Jardín reservoir was supplying its respective area during the entire sampling
campaign but none of the sampling points were located on it; sample location was dependent on
where leakages occurred in the network during the experimental collection period. According to the
aforementioned, the sub-network 4 was mostly supplied by the WTPs during most of the sampling
journeys. As a result, water age was processed according to that condition through Thiessen
polygons as explained in Section 3.2.4.

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 98
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 3-3. Hydraulic conditions of the sub-network 4 during sampling campaign

Water age of sampling point / Filling cycle of service

Operation of service reservoirs during sampling reservoirs
Sampling Water Biofilm
Neighbourhood
point
Ciudad Value Value
La Campiña La Normal Siloé Nápoles Classification Classification
Jardín (hours) (hours)

1 Jorge Isaacs WTP working WTP working WTP working WTP working Draining cycle 13,95 High 13,99 High
Atanasio
2 WTP working WTP working WTP working WTP working Draining cycle 9,71 Medium 9,71 Medium
Girardot
3 San Judas WTP working WTP working Draining cycle WTP working Draining cycle 12,37 Medium 12,37 Medium

4 Las Ceibas WTP working WTP working Draining cycle WTP working Draining cycle 146,01 Very high 8,12 Low

5 San Joaquín WTP working WTP working Draining cycle WTP working Draining cycle 14,41 High 15,59 High

6 Industrial WTP working WTP working WTP working WTP working Draining cycle 10,06 Medium 10,06 Medium

7 San Judas WTP working WTP working Draining cycle WTP working Draining cycle 11,71 Medium 11,47 Medium

8 Olaya Herrera WTP working WTP working Draining cycle WTP working Draining cycle 13,23 High 13,23 High

9 Alfonso López WTP working WTP working Draining cycle WTP working Draining cycle 8,00 Low 8,26 Low

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors in a tropical-weather DWDN 99
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
3.3.3 Water age

The water age corresponding to every sampling point for both biofilm and water samples is included
in Table 3-3 and Figure 3-11. Thus, there are four pairs of samples water-biofilm with medium water
age, three pairs of samples with high water age and one pair of samples with different classification
of water age; the biofilm sample at point 4 had very high water age and water sample had low water
age. The water age at Point 4 differs because water sample had to be taken two streets away from
the biofilm sampling point and, despite their geographical proximity, the respective areas had
notably different water age values. It can be hypothesised that such notable high water age value
may be related to the presence of a recirculation dead-end point in the DWDN.

Figure 3-11. Water age of sampling points

3.3.4 Physical characteristics of sampled pipes

Materials, diameters and age of collected pipes for biofilm sampling are described in Table 3-4.
The predominant material of sampled pipelines is asbestos cement. One biofilm sample was
collected from a cast iron pipeline. Pipe diameters were predominantly 4” and three biofilm samples
were taken from 3”, 8” and 12” pipelines. The oldest pipeline was installed in February 1958 and
the newest one in August 1990. Therefore, the range of pipe age is 25-57 years at the time of
sampling.

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 100
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 3-4. Network characteristics in sampling points
Sampling Pipe Pipe diameter Installation
Neighbourhood
point material (Inches) date of pipe
1 Jorge Isaacs Asbestos 4 21/09/1958
2 Atanasio Girardot Cast iron 4 03/02/1958
3 San Judas Asbestos 3 18/04/1981
4 Las Ceibas Asbestos 4 11/12/1979
5 San Joaquín Asbestos 4 18/08/1990
6 Industrial Asbestos 8 20/05/1972
7 San Judas Asbestos 12 04/06/1981
8 Olaya Herrera Asbestos 4 17/05/1962
9 Alfonso López Asbestos 4 06/04/1964

3.3.5 Water quality and biotic parameters

Water quality and biotic parameters and descriptive statistics are presented in Table 3-5. Water
quality characteristics including temperature, pH, free residual chlorine and TTHMs were within the
expected ranges, except for the lowest concentration of chlorine (0.12 mg/L), which was measured
at sampling point 4, which corresponded to the highest water age (146 h). Such concentration of
free residual chlorine is considered very low according to the recommended values set for drinking
water quality standards by local regulators (0.3-2.0 mg/L) (Ministerio de la Protección Social, 2007).
TOC measured in biofilm samples presents a lower variation (CV=105.6%) compared to the
variation in biofilm mass (CV=321.5%). All concentrations of TOC in DW were reported as lower
than the detection limit (<0.8 mg/L). Regarding TTHMs, concentrations in all water samples were
lower than 40 g/L, which falls below the maximum concentration of TTHMs allowed in drinking
water according to UK and Colombian regulation (100 and 200 g/L, respectively).

In relation to biotic factors, unit dry biomass presents the highest variation among all the variables
analysed in this study (CV=329.09%; Table 3-5). Although calculation of the unit dry biomass in the
cast iron pipe sample (sampling point 2) was not possible, the second highest content of dry
biomass in the biofilm (233.7 - 3,664.8 mg) and highest TOC in the water sample (10.10 mg/L;
Table 3-5) were found in this point. This result can be related to a significant amount of biofilm and
sediments attached to the pipe walls when compared with other sampling sites.

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 101
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 3-5. Water quality, biotic parameters and descriptive statistics
Biofilm
Water quality
characteristics
Sampling point Free res. Total res. Unit dry
Temperature pH TTHMs TOC
chlorine chlorine biomass
(°C) (Units) (mg/L) (mg/L)
(mg/L) (mg/L) (mg/cm2) *

1 26 7,32 1,20 1,35 30,3 0,819 1,41

2 25 7,16 1,66 1,76 28,9 10,104 -

3 25 7,35 1,28 1,43 23,5 1,210 1,45

4 - 7,04 0,12 1,61 36,7 1,453 0,29

5 25 6,76 1,30 1,45 28,3 1,527 0,38

6 26 7,01 1,12 1,33 35,5 1,739 3,23

7 28 7,02 1,15 1,21 30,8 2,139 0,23

8 26 6,86 0,86 1,02 38,6 1,849 2,09

9 27 6,62 1,31 1,57 33,3 2,157 3,34

Minimum 25 6,62 0,12 1,02 23,45 0,819 0,23

Descriptive statistics

Median 26 7,02 1,20 1,43 30,80 1,74 1,43

Mean 26 7,02 1,11 1,41 31,76 2,56 1,55

Maximum 28 7,35 1,66 1,76 38,6 10,10 3,34

Standard deviation 1 0,23 0,40 0,21 4,47 2,70 1,28

CV 3,85% 3,26% 36,25% 14,76% 14,09% 105,64% 82,70%

Colombian standards (a) -- 6.5-9.0 0.3-2.0 -- 200 -- --

* Median of replicates | Descriptive statistics of all data of unit dry biomass (including replicates - mg/cm2): Min=0.11; Median=1.41;
Mean=5.20; Max=81.45; Stand. Dev..=17.11; CV=329.09% | AC: asbestos cement | CI: cast iron
(a) Ministerio de la Protección Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial (2007)

Several physicochemical characteristics of bulk water were correlated to identify the dynamics
present in the studied network; results are presented in Table 3-6. Significant negative correlations
were found between total residual chlorine and temperature (p=0.019), free residual chlorine and
water age (p=0.004) and free residual chlorine and TTHMs (p=0.017). Weak negative correlations
were identified between temperature and free residual chlorine (p=0.052, slightly higher than the
level of significance) and between pH and TTHMs (p=0.042). A positive correlation was observed
between temperature and TTHMs (p=0.003).

With regards to biofilms, correlations presented in Table 3-7 indicated that there is a strong positive
relationship between unit dry biomass and pipe age (p=0.008). Additionally, water age was
Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 102
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 negatively correlated with TOC in biofilms but the p-value was slightly higher than the level of
significance (p=0.063) and no association was identified between water age and unit dry biomass,
possibly related to the influence of pipe age/material over the later variable.

Table 3-6. Spearman correlation coefficients for bulk water parameters

Richness Diversity Total Free

Water
Variables ↓→ index index pH Temper. residual residual TTHMs
age
(Margalef) (Shannon) chlorine chlorine
Richness index
-
(Margalef)
Diversity index
C.N.T -
(Shannon)
Water age 0.277 0.315 -

pH 0.365 *** 0.414 * C.N.T -

Temperature -0.355 *** -0.238 C.N.T C.N.T -

Total residual
0.074 0.149 -0.067 0.117 -0.476 * -
chlorine
Free residual -0.533
-0.251 -0.273 -0.033 -0.401 *** C.N.T -
chlorine **
-0.594
TTHMs -0.259 0.049 0.060 0.802 ** C.N.T -0.671 * -
*
Correlation is significant at the 0.05* / 0.01** level (2-tailed)
*** Correlation coefficient slightly higher than 0.05 → 0.052 < p-value < 0.089

Table 3-7. Spearman correlation coefficients for biofilm parameters

Richness Diversity
Water Unit dry TOC -
Variables ↓→ index index Pipe age
age biomass biofilm
(Margalef) (Shannon)

Richness index (Margalef) -

Diversity index (Shannon) C.N.T -

Water age 0.364 *** 0.375 *** -

Pipe age -0.404 * -0.512 ** C.N.T -

Unit dry biomass -0.582 ** -0.733 ** -0.196 0.559 ** -

-0.552
TOC - biofilm -0.294 -0.357 0.334 0.259 -
***
Correlation is significant at the 0.05* / 0.01** level (2-tailed)
*** Correlation coefficient slightly higher than 0.05 → 0.059 < p-value < 0.068

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 103
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
3.3.6 Characterization of the bacterial community structure of biofilms and bulk water

The RA to phylum and genera level for water and biofilm samples can be observed in Figure 3-12
and Figure 3-13, respectively; groups with RA lower than 10% were grouped in the category
“Others”. Water samples were dominated mainly by Proteobacteria phyla (43-98%), followed by
Cyanobacteria (0.05-41%), and Firmicutes (0.84–34%). With regard to biofilm samples, the
predominant phyla were Proteobacteria (26-72%), followed by Firmicutes (3–30%), and
Actinobacteria (8-19%). Different genera were dominant in each bulk water sample and those with
higher RA were Bacillus, Brucella, Cyanothece, Methylobacterium, and Phyllobacterium (17.47-
95.91%). With regard to biofilm samples, different genera were dominating each sample and those
with higher RA were Acinetobacter, Alcaligenes, Alcanivorax, Bacillus, Deinococcus, Holophaga,
and Thermoflavimicrobium (4.34–43.92%).

Figure 3-12. RA of bacterial groups to (a) phylum level and (b) genus level in water samples

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 104
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 3-13. RA of bacterial groups to (a) phylum level and (b) genus level in biofilm samples

ANOSIM test was applied to assess the relationships between species RA and engineered factors
(global R and p-value; Table 3-8). With regards to water samples, relationships between species
RA and factors water age, free chlorine, pH, and free chlorine and water age combined were
statistically significant. In relation to biofilm samples, ANOSIM test resulted statistically significant
for the factors pipe age, and water age, and unit dry biomass and pipe age combined. Factor “Pipe
material” was not included in the statistic tests due to only one sample was collected from CI
pipeline, then comparison between CI and asbestos cement would not be statistically strong.
Habitat was also a factor influencing the RA of species. MDS analysis also revealed that habitat
and pipe material were the factors which clustered bacterial species (Figure 3-14a and Figure

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 105
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
3-14b, respectively). The two-dimensional plots of the MDS analysis and dendrograms for
visualization of Bray Curtis similarity index for every parameter considered in this study are included
in the Appendix 9-C.

Importantly, other methanotrophic organisms were observed in biofilm samples such as

Methylobacterium (RA=1.16%) and Methylosinus (RA=3.34%). In bulk water, Spearman
correlations with TTHMs were statistically significant for the genus Methylobacter (=0.437; p-
value=0.023) and Methylobacterium (=-0.417; p-value=0.030).

Table 3-8. ANOSIM test for RA of species

ANOSIM ANOSIM
Habitat Factor Habitat Factor
Global R p-value Global R p-value
Water age 0,441 0.002* Pipe age 0,601 0.001*
Temperature 0,002 0.430 Pipe diameter 0,122 0,113
Bulk
Free chlorine 0,441 0.001* Biofilm Water age 0,601 0.001*
water
pH 0,441 0.001* Unit dry biomass 0,127 0,268
Free chlorine – Unit dry biomass –
0,441 0.001* 0.690 0.013*
Water age Pipe age
Habitat 0,446 0.001*
Biofilm –
Bulk Water age 0,304 0.001*
water Water age –
0,666 0.001*
Habitat

Figure 3-14. Non-metric MDS analysis of bacterial RA. Factors (a) Habitat and (b) and pipe material
- biofilm samples

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 106
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
3.3.7 Microbial richness and diversity

Richness and diversity indices were calculated (Table 3-9) according to factors such as habitat and
water age. t-test indicated that the means of richness and diversity indices of biofilm samples are
higher than those of water samples. Spearman correlations were applied to test the relationships
between indices and engineered factors, for both water (Table 3-6) and biofilm (Table 3-7) samples.
Negative correlations were found between biofilm indices and pipe age and unit dry biomass. Pipe
material is also a factor which is influencing the diversity of species in biofilm samples since median
of diversity index from asbestos pipes was higher than those from cast iron pipes. Results of
comparisons of medians indicated that median of richness and diversity indices of biofilm samples
with high water age was higher than those with medium water age. In water samples, richness
index with very high water age was higher than those with low water age; and median of richness
index with medium water age was higher than those with low water age. Positive correlations
between water age and richness and diversity indices were only found in biofilm samples but the
p-values were slightly higher than the significance level.

Table 3-9. Medians and means of richness and diversity indices

Index → Margalef richness
Habitat → Bulk water Biofilm
Habitat
Factor → Water age Water age
Descriptive
Biofilm Water Low Medium High Very high Low Medium High
statistics ↓
Median 253,5 272,0 272,0 271,0 253,0 319,0 251,0 185,5 594,0
Mean 417,1 279,7 229,3 276,5 284,4 328,7 449,6 303,3 550,7
Standard
276,68 85,80 74,77 195,60 86,24 50,20 365,30 237,16 234,59
deviation
N 26 27 3 12 9 3 5 12 9
Low - Very high water age Medium - High water age
p-value 0.022 (a) *
0.050 (b) * 0.019 (b) *
Index → Shannon diversity
Habitat → Bulk water Biofilm
Habitat
Factor → Water age Water age
Descriptive
Biofilm Water Low Medium High Very high Low Medium High
statistics ↓
Median 4,978 4,647 4,422 4,611 4,647 4,876 4,696 4,442 5,648
Mean 5,129 4,622 4,440 4,581 4,662 4,842 5,172 4,759 5,598
Standard
0,76 0,30 0,26 0,27 0,36 0,09 0,92 0,73 0,46
deviation
N 26 27 3 12 9 3 5 12 9
Medium - High water age
p-value 0.003 (a) ** -
0.013 (b) *
Means / Medians comparisons: (a): t-test; (b): Mann-Whitney test
Correlation is significant at the 0.05* / 0.01** level. Only significant tests are shown

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 107
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
3.4 DISCUSSION

3.4.1 Bulk water quality and biotic parameters and their relationships with engineered
factors

As discussed in Chapter 2, DBP formation is known to be influenced by operational parameters

such as pH, temperature, TOC, chlorine dosage, water age, and type and concentration of
precursors. The interactions observed between parameters such as water age, temperature, pH,
free residual chlorine, and TTHMs confirm the dynamics occurring in a DWDN in relation to THMs
formation: increasing water age promotes loss of free residual chlorine because the disinfectant is
volatile and reacts with organic and inorganic matter, likewise the concentrations of THMs were
increasing. Nescerecka et al. (2014) and Wang et al. (2014) also identified depletion of disinfectant
with higher water age in a real scale and simulated DWDNs, respectively.

THM formation is highly directly influenced by pH and temperature (Liang and Singer, 2003). Such
relationship is evidenced by the current results, which show a strong correlation between TTHMs
and temperature. However a negative relationship between TTHMs and pH was found in the
current study, which may be related to the narrow range of pH data (6.62-7.35; CV=3.26%; see
Table 3-5); higher concentrations of THMs have been identified with higher pH in a range of 5-8
units in laboratory experiments (Liang and Singer, 2003; Wang et al., 2012c). Positive and negative
correlations between pH and THMs in the distribution system have also been reported by
Rodriguez and Sérodes (2001) in three DWDNs located in Canada. With regards to unit dry
biomass, values found in the current study (0.11-81.45 mg/cm2) were similar to data reported by
Ren et al. (2015) in a real scale DWDN (1.96-140.79 mg/cm2), located in a subtropical area in
China, which had operated for 11 years at the time of sampling, and included PVC, steel, and cast
iron pipes.

3.4.2 Identification of the bacterial community structure of biofilms and bulk water

Comparing phylotypes detected in bulk water and biofilm, biofilm samples seem to be more diverse
than water samples (7 vs 4 groups, respectively); this was corroborated by the diversity and
richness indices (417,1 vs 279,7 and 5.1 vs 4.6, respectively), as discussed below. Phylum
Actinobacteria, Firmicutes, and Proteobacteria were the common phylotypes in the two habitats,
with the later community being the dominant group in the entire set of samples. Recent studies
from other geographic regions have reported that both water and biofilm samples were dominated

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 108
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
by Proteobacteria phyla (Douterelo et al., 2013; Holinger et al., 2014; Kelly et al., 2014; Sun et al.,
2014; Wang et al., 2014; Mahapatra et al., 2015; Ren et al., 2015). Particularly, the study of Ren et
al. (2015) was developed in the city of Shaoxing, located at the east of China. This city is
characterized for a subtropical monsoon climate; the annual average low temperature oscillates
between 3-25°C and the annual average high temperature varies between 8-33°C
(TravelChinaGuide, 2017). Other important phylotypes in both habitats were Firmicutes and
Actinobacteria, which have also been observed in other real scale DWDNs (Ren et al., 2015;
Bautista-de los Santos et al., 2016).

The phylum Cyanobacteria was present in all the water samples and it belongs to a diverse group
of photosynthetic microorganisms, widespread in aquatic and terrestrial ecosystems;
representative genus of this phyla is Cyanothece, which was highly abundant in four water samples
and has been little studied (Dechatiwongse, 2015) and are not cytotoxin producers (Jakubowska
and Szeląg-Wasielewska, 2015). The source for the high presence of these organisms in the
analysed samples is likely to be from one of the reservoirs of clarified water located at one of the
WTP of the studied sub-network. Revetta et al. (2011), by analyzing 16S rRNA gene clone libraries
derived from DNA extracts of 12 samples and comparing to clone libraries previously generated
using RNA extracts from the same samples, found that these bacteria may be active in chlorinated
drinking water. Since drinking water pipes are dark environments, how Cyanobacteria survive in
these is not clear yet. However, it is hypothesised that due to the limited action that chlorine has
for the inactivation of cyanobacterial planktonic cells (Fan, 2012), cell material could have been
transported within the distribution network and ultimately detected by DNA extraction.

At class level, Alphaproteobacteria are known for their potential higher resistance to chlorine
(Gomez-Alvarez et al., 2012), which explains their dominance in bulk water from seven sampling
locations. On the other hand, Gammaproteobacteria includes the faecal indicator E. coli and genus
Salmonella, Yersinia, Vibrio, and Pseudomonas (Williams et al., 2010). With regards to bacteria
genera, Bacillus was the unique common genus present in both habitats; it was present mainly in
biofilms but was also detected in bulk water. Members of this group are ubiquitous in the
environment, they can form spores that are advantageous as a survival strategy under
environmental and nutritional stresses (e.g., disinfectant presence). Species of the genus Bacillus
are also known for their ability to form a biofilm in different environments (Checinska et al., 2015).

Another example of adaptive-response strategies is the genus Deinoccocus, which was mainly
present in biofilm samples and shows high resistance to ionizing-radiation (gamma and/or UV
radiation) (Rainey et al., 2005). This ability may not be related to environmental adaptation since
Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 109
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
natural sources of this type of radiation on Earth does not reach the resistant levels showed by
these bacteria; compensation of desiccation has been suggested as an explanation for radio-
tolerance ability (Rainey et al., 2005). Its presence in biofilm samples may also be a strategy to
survive in a hostile environment such as DW pipes.

The water source of the studied DWDN is a river, and its basin supports a diverse range of domestic
and industrial activities, which have led to concerning water quality issues (Ocampo-Duque et al.,
2013). In addition, soil surrounding the DWDN entering to the network during repair activities and
fittings and pipelines replacement may also affect the microbial communities in DWDNs. For
example, genera including Thermoflavimicrobium and Phyllobacterium were highly abundant in
bulk water samples at several sampling points. These organisms have been previously reported in
soil- related habitats (Rojas et al., 2001; Yoon et al., 2005). Then their presence may be explained
by the characteristics of raw water source. Bacteria with anaerobic metabolism were detected only
in biofilm samples. Holophaga genus includes only one validly described species: H. foetida, which
was isolated from a dilution series in liquid media inoculated with black anoxic freshwater mud from
a ditch in Germany (Itoh et al., 2011). The origin of this microorganism may be the water source
which belong to a water basin highly contaminated due to domestic and industrial activities
(Ocampo-Duque et al., 2013).

Genus Desulfovibrio was only present in point 2 (11.89%), which corresponds to a cast iron pipe
sample. These organisms are sulphate-reducing bacteria (SRB), related to iron corrosion.
Desulfovibrio finds a favourable environment in this kind of pipes, most likely promoting its corrosion
and potentially leading to failure. Tubercles found in the sampled piece of pipe (Figure 3-6) may
create limited oxygen environment for the growth of these bacteria (Douterelo et al., 2014b). Similar
high abundance of this genus was detected by Ren et al. (2015) in 11-years old cast iron pipes
(16.90%); however Sun et al. (2014) reported low abundance of Desulfovibrio (0.01-0.19%) in 20-
years old cast iron pipes. This may be related to the temperatures associated to the sampling
campaign; while Ren et al. (2015) carried out their field work between January and September,
which are characterized for temperatures between 3-33 °C (TravelChinaGuide, 2017), Sun et al.
(2014) measured temperatures between 9-20 °C for the seven water supply systems considered
in their study. Temperature indirectly affects pipe corrosion because parameters such as biological
activity, dissolved oxygen solubility, solution properties (e.g., viscosity), ferrous iron oxidation rate,
and thermodynamic properties of iron scale vary with temperature. Such parameters directly
influence corrosion of iron pipes (McNeill and Edwards, 2002). Corrosion of metallic pipes in

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 110
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
DWDNs is inconvenient because tubercles reduce the hydraulic capacity of the pipes due to the
scales formation and the accumulation of iron and manganese particles (Douterelo et al., 2014b).

Methylotrophic bacteria were detected in five bulk water samples. Methylobacterium genus
includes methylotrophic organisms (Leisinger et al., 1994), which are ubiquitous in different
environments including soil, freshwater, drinking water, lake sediments, leaf surfaces, and root
nodules (Leisinger et al., 1994). This genus is able to degrade DBPs such as HAAs (particularly
dichloroacetic acid) (Zhang et al., 2009), and they are biofilm formers with high resistance to sodium
hypochlorite disinfection in single-species biofilm (Simões et al., 2010). Importantly,
Methylobacterium has not been found yet in non-chlorinated DWDNs (Martiny et al., 2005; Liu et
al., 2014). Therefore, the presence of these microorganisms in DWDNs should be considered as a
potential indicator of DBP presence, despite of Methylobacterium presents facultative metabolism
and it is able to use a wide range of organic compounds as sources of carbon and oxygen (Gallego
et al., 2005). Therefore, the presence of methanotrophs in biofilm samples may indicate the
presence of anoxic layers within the biofilms, as discussed previously for the genus Desulfovibrio.
Such anoxic environment inside of the biofilms leads to the formation of methane by methanogenic
organisms, which can be consumed by methanotrophic organisms living inside of the biofilm and
bulk water if methane was transported to this habitat. On the other hand, it has also been identified
utilization of chloroform in co-metabolic processes by Methylobacterium (Patel et al., 1982) and
Methylosinus (Park et al., 1991). The negative association identified between TTHMs and
Methylobacterium may be related to an inverse relationship between THMs and HAAs, since
reduction of HAAs concentrations has been found with higher retention times (Tung and Xie, 2009).

Pseudomonas was common in most of the biofilm samples (8 of 9 points) but was not the dominant
genus. While Pseudomonas was present as the dominant group in the initial process of biofilm
formation in 28-days-old biofilms (Douterelo et al., 2014c), Henne et al. (2012) found that biofilm
communities sampled at nearby points were similar, then the authors hypothesised that during
several years physically related biofilm communities will show similar community structures. In
contrast, the current study found that dominant bacterial communities of the studied DWDN (25-57
years old) were different in each sampling point; therefore, the present results show that biofilms
do not present properties of either young or old biofilms. This may be related to the unstable
hydraulic conditions of this water network, which may partially remove biofilm components, then
altering the structure of bacterial communities. Similarly, in a laboratory-based full scale DWDN,
high flow variations indicated the promotion of young biofilms with more cells and less EPS, by the
potential cyclic removal of the first layers of the biofilms (Fish et al., 2017).

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 111
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Bacteria groups, which include pathogenic and opportunistic organisms, were also observed in
biofilm and bulk water samples. Acinetobacter was detected in four biofilm samples; members of
this genus are ubiquitous in the environment, biofilm formers, and have been found in wastewater
treatment reactors, in infected patients, and contaminated clinical devices (Carr et al., 2003; Lin et
al., 2003). In bulk water, Brucella was also found in four samples; this genus comprises 11 species
and 10 of them are associated with human infections (Scholz et al., 2010; Xavier et al., 2010).

Results of this study indicated the presence of genus Staphylococcus in bulk water (RA = 0.04-
7.07%) and biofilm samples (RA = 0.08-10.91%). Ji et al. (2015) also found this organism in water
samples collected from a building plumbing system. Due to this genus constitute a major
component of the human microflora (Heilmann et al., 1996), Ji and collaborators discarded the
hypothesis of samples contamination since these bacteria were absent in the blanks (Ji et al.,
2015). In the present study, Illumina sequencing was not run in blanks but this genus has been
isolated from a model-laboratory DWDN and cultured in dual biofilms, being classified as a
moderate biofilm former (Simões et al., 2007); this ability is used to colonize hospital devices. In
addition, Staphylococcus aereus was also detected in bulk water from a rural DW system
(Lechevallier and Seidler, 1980). It can be hypothesised that the source of this opportunistic
pathogen could be the surface raw water, since the river basin is highly contaminated due to human
and industrial activities (Ocampo-Duque et al., 2013).

3.4.3 Influence of network characteristics over the bacterial communities and richness
and diversity indices in biofilms and bulk water samples

The higher richness and diversity in biofilms than in bulk water samples can be related to the
favourable conditions offered by this microenvironment for bacteria survival such as protection
against disinfectant and higher availability of nutrients. Douterelo et al. (2013) also found in 28-
days old biofilms higher diversity and richness in biofilms in a chlorinated DWDN, indicating that
only certain bacteria in bulk water have the ability to produce EPS and attach to pipe walls. For
instance, Bacillus was the only common genus detected in the two habitats. Bacillus can form
spores that protect them from disinfection in bulk water and allow them to attach to biofilms, then
when the environmental conditions are favourable they start developing as active cells (Checinska
et al., 2015). Conversely, Henne et al. (2012), based on 16S r RNA fingerprints of extracted DNA
and RNA, found that richness of microbial communities was higher in bulk water than biofilms from
a 20-year-old DWDN. The authors argue that only those bacteria that can actively contribute to the

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 112
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
succession of the biofilm were successful in colonizing biofilms, while bacteria that cannot fill
perfectly the narrow niches in biofilms vanished over time.

Relationship between pipe age and unit dry biomass may be related to the detachment of some
asbestos fibres which was observed during biofilm scrapping from the sampled pipes and is
representative of the potential wear of the pipe material in time, due to the biological activity. The
influence of removal of such fibres was described by Wang et al. (2011), who tested the biological
activity in 53-54 years-old sections of asbestos pipes. By stablishing microbial activity of iron-
reducing bacteria (IRB), SRB and biofilm-former bacteria in the patina layer (porous layer, mainly
composed of microbial biomass along with interwoven asbestos fibres) of those pipes sections,
they established that such microbial activity leads to deterioration of asbestos pipes and potential
leakages (Wang et al., 2011). In this study, IRB were observed in biofilm samples, corresponding
to 24-56 years-old pipe sections, such as genus Geobacter (RA: 1.96-4.51%) and family
Rhodobacteraceae (RA: 3.74-8.87%); SRB such as genus Desulforegula (RA=1.15%),
Syntrophobacter (RA=1.08%), and Clostridium (RA=1.00-3.74%) were also detected.

Although these microbial groups were present with low RA, this fact may indicate the presence of
an anoxic layer in asbestos pipes, which promotes the acidification of the media due to the
production of organic acids from anaerobic metabolism, leading to local pH decrease. This
facilitates the weathering and dissolution of the acid-receptive minerals in hydrated cement matrix,
thus, creating pitting and voids within a pipe wall. In the long term, this process continuously caused
by biological activity can be considered as biodegradation of asbestos material in water pipes
(Wang et al., 2011).

The influence of pipe material on the bacteriological composition of biofilm samples is reflected on
the presence of SRB such as Desulfovibrio, which was present exclusively in CI pipes.
Desulfovibrio finds a favourable environment in this type of pipes, most likely promoting its
corrosion and potentially leading to failure. Similar high abundance of this genus was detected by
Ren et al. (2015) in 11-year old CI pipes however, Sun et al. (2014) reported low abundance of
Desulfovibrio (0.01-0.19%) in 20-year old cast iron pipes. The tubercles found in the sampled piece
of pipe (Figure 3-6) may create a favourable environment for the growth of these bacteria.
Additionally, such tubercles can reduce the hydraulic capacity of the pipes due to the formation of
scales and the accumulation of iron and manganese particles (Douterelo et al., 2014b). Several
studies have confirmed the impact of pipe material over the structure of microbial communities in
biofilm samples collected from simulated DWDNs (Wang et al., 2014), bench-scale pipe section
reactors (Mi et al., 2015), and real-scale DWDNs (Ren et al., 2015). Although there is not an
Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 113
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
absolute consensus about the best material to minimize biofilm growth, in general, plastics appear
to be advisable over metals and cements (Fish et al., 2016).

With regards to water age (8-16 hours), the effect of this factor on the biofilms bacterial communities
may be associated with the relationship between this parameter and physicochemical
characteristics of DW, which was previously discussed. In addition, chlorine low concentrations,
stagnation and low velocities conditions associated with high water age lead to increase the cells
count in bulk water (Nescerecka et al., 2014) and favour biofilm formation (Fish et al., 2016). Water
age is considered as a factor influencing the biological stability of DW (Prest et al., 2016). This was
corroborated by Wang et al. (2014) who established in simulated-DWDN biofilm samples that water
age (2.3 and 5.7 days), disinfectant, and pipe material interact with each other to create distinct
physicochemical conditions and ecological niches, in which various microbes can be selected and
enriched in DW systems. Results of Ji et al. (2015) also confirmed that water age (3, 4.5 and 6.5
days) can influence the shifts in microbial communities in building plumbing and obviously affected
free chlorine, and selectively influenced lead, copper, and zinc concentrations for certain pipe
materials.

In this line, the main purpose to maintain residual disinfectant in a DWDN is to minimise any
potential microbiological contamination, taking advantage of its fast oxidation ability, resulting in
microorganisms’ inactivation, despite the protective effect caused by EPS in the biofilm matrix (Xue
et al., 2013). Spearman’s correlations showed no associations between indices and concentrations
of free chlorine. Due to continuous variation of free chlorine in bulk water, hourly monitoring of this
variable assisted by online devices in the network is recommended in order to test further
correlations. Points with potential high water age are suggested to monitor residual disinfectant
such as extreme points of the DWDN, service reservoirs, storage tanks, and dead-end sections
(Montoya et al., 2009).

ANOSIM indicated statistical significant differences among species for the factor pH, which was
also correlated positively with both richness and diversity indices. Due to the relationship between
pH and alkalinity and the governance of this factor over the relative proportions of hypochlorous
and hypochlorite, which present different disinfection efficacies, pH is impacting the variability in
the water bacterial community as found by Sun et al. (2014); Ji et al. (2015) also determined higher
individual association of pH with the bacteria and archaea organisms in water samples.

Temperature and richness index were negatively correlated; similar results were found by Henne
et al. (2013) by comparing microbial communities of cold and hot water (T=41 °C approximately);

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 114
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
cold water presented higher community diversity and high stability over time. This study considered
T=3 °C, which corresponds to typical temperature values for tropical cities with hot weather.
Deeper analysis should be done in relation to the influence of temperature variation in DWDNs in
tropical weather.

3.5 SIGNIFICANCE AND LIMITATIONS OF THE STUDY

The current study collected samples, one sample in time, of bulk water and biofilm in nine points of
a DWDN located in a tropical-climate city, where warm temperatures dominate. The bacterial
communities were assessed in both habitats, and other physico-chemical parameters were used
to characterise them, with special focus on THMs. It is important to acknowledge that size of data
is small and temporal variation was not included in this study. It is important to highlight that the p-
value of some correlations was slightly higher than the level of significance, which may be
associated with the limited number of data collected in this study. The current procedures for
attention of leakage repairs also represents a challenge on developing demanding field work
studies like the current one, considering that the normal operation cannot be interrupted for
sampling activities, consequently, then randomness governs a sampling plan. O&M staff must
prioritize the repairs according to the size and type of affected population (e.g., big neighbourhoods,
areas with hospitals and education institutions); therefore sampling activities must be carried out in
a long term if the scope included the evaluation of more strictly engineered variables such as water
age, pipe age, pipe materials, type of network component (pipelines, valves, water meters), among
others.

The field assessment approach presented here represents a progress in the study of the bacterial
composition in Cali’s DWDN, which exhibits several O&M challenges in relation to the loss of
hydraulic integrity. To the author knowledge, this is the first time molecular analysis are applied to
study, at full scale, the microbiological drinking water quality in this water network. This allowed to
correlate biotic parameters with engineered characteristics, then some O&M recommendations
were produced, which may be useful to the water utility. The author is aware that some
communication mean must be identified in order to present and explain the findings of this study to
the water managers of the studied water network.

The bacterial composition of habitats bulk water and biofilm in the studied DWDN was identified by
DNA extraction, which only allows indicating the presence of particular bacteria, but it is not possible

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 115
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
to determine when they entered to the system, for how long they have been there, what the exact
source is, and if they were metabolically active, viable and/or cultivable. As economical, technical,
and trained human resources improves in developing countries such as Colombia, the research
and knowledge transfer with regards to biofilms and their interactions with other variables in
DWDNs can be promoted. For instance, the recent work by Lemus Pérez and Rodríguez Susa
(2017) used biofilms grown with drinking water treated by a WTP in Bogotá (Colombia) to
quantitatively assess the influence of the type of EPS in the potential formation of several species
of DBPs.

Further studies should incorporate the analysis of RNA and active genes in order to improve the
characterization of the role played by microorganisms and their interaction with engineered factors
of DWDNs. Additionally, the temporal variation of bacterial communities in DWDNs can also be
assessed; Bautista-de los Santos et al. (2016) found that bacterial community composition is
significantly different across sampling sites for time-periods during which there are typically rapid
changes in water use, which suggests hydraulic changes (driven by changes in water demand)
contribute to shaping the bacterial community in bulk drinking water over diurnal time-scales. In
relation to hydraulic changes, monitoring turbidity also contributes to identify the relationship
between this parameter and the presence of particles, cells, and biofilm detached clusters in bulk
water (Douterelo et al., 2013; Douterelo et al., 2014b).

3.6 CONCLUSIONS

The application of sequencing analysis represents a step forward in the study of microbiological
aspects of DWDNs in tropical-climate countries. The key findings from the study are:

 Most of the bacterial communities identified in this work have also been found in temperate-
weather water systems. This may indicate that some drinking water bacteria are ubiquitous and
that treatment and engineered environments shape the bacterial communities in a specific way.
 Similarly to temperate-climate DWDNs, bacterial communities in sampled biofilms are different
from those in bulk water, with the former more diverse and richer.
 Pipe age, pipe material, water age, free chlorine, pH and temperature were associated with
microbiological parameters indicating that these are key to the composition of bacterial
communities. Deeper analysis should be done in terms of the influence of temperature variation
in DWDNs in tropical climates.

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 116
in a tropical-weather DWDN
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Pipe material influences the microbial ecology of DWDNs. Desulfovibrio was identified exclusively
in the CI pipe.
 Methylotrophic bacteria were found in biofilms and bulk water; these microorganisms are known
to be able to degrade DBPs as HAAs.
 Significant correlation was found between RA of Methylobacter and Methylobacterium and TTHM
concentrations in bulk water.
 Design and O&M of DWDNs should consider all the possible procedures to minimise biofilm
growth to manage both biological and chemical stability of drinking water

Chapter 3. Field assessment of bacterial communities and their relationships with engineered factors 117
in a tropical-weather DWDN
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 4 A SIMPLE BIOFILM CHLORINATION MODEL

4.1 INTRODUCTION

Modelling is an increasingly important tool for O&M activities in the drinking water industry. It is an
especial useful tool to help cope with the complexity of DWDNs. Modelling in this context can be
classified in two major groups: hydraulic and water quality modelling, with the later one dependent
on the first one. Modelling offers useful information using relatively few resources, however it
presents challenges mostly related to obtaining reliable primary information (inputs) and validation.
In addition, qualified personnel are required to use the model depending on the software and its
level of elaboration, and interpret results.

As was discussed in Chapter 2, chlorine decay and DBP formation are the most common targets
for water network operators, while biofilm modelling has been addressed mainly for research
purposes. Recent laboratory studies have determined that components of the biofilm matrix such
as cells, EPS, trapped organic matter, and biomolecules act like DBP precursors (Hong et al., 2008;
Fang et al., 2010a; Fang et al., 2010b; Wang et al., 2012c; Wang et al., 2013a; Wang et al., 2013b;
Lemus Pérez and Rodríguez Susa, 2017). Despite this, the contribution of biofilms to DBP
formation has scarcely been considered in DWDNs models. To the author’s knowledge, there is
only one study published that includes the contribution of planktonic cells, biofilms, and detached
biofilm clusters in the THM formation in main trunk pipes and dead-end zones (Abokifa et al.,
2016a).

The literature review of Chapter 2 indicates that DBP and biofilm modelling have been developed
separately. Empirical DBP models have been proposed to predict DBP concentrations in drinking
water considering the main influencing factors such as time, pH, temperature, precursors,
disinfectant dose, and bromide concentration. In relation to precursors, variables such as TOC,
DOC, algae, chlorophyll, UV254, and SUVA have been used. The last two have been applied as
precursor indicators, but biofilms have not been included yet in empirical models. These models
are site-specific since they include the peculiar conditions of each studied DWDN such as seasonal
variations, water quality parameters and type of water treatment process or technology.

DBP models have been mainly developed to offer water quality information in the DWDNs, while
drinking water biofilm models have been aimed at avoiding, minimizing and/or controlling biofilm
growth. Considering the important role played by EPS in initial formation, growth, and survival of
biofilms; several models have included the EPS production by bacteria (Kommedal et al., 2001;

Chapter 4. A simple biofilm chlorination model 118

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Xavier et al., 2005a; Xavier et al., 2005b; Duddu et al., 2008; Zhang et al., 2008; Duddu et al., 2009;
Cogan, 2010; Cogan, 2011; Lindley et al., 2012; Clarelli et al., 2013; Ghosh et al., 2013; Macías-
Díaz, 2015; Tierra et al., 2015; Zhao et al., 2016).

The model proposed here aims to improve the DBP prediction in water systems by better
comprehension of the role of biofilms as DPB precursors in plumbing systems, where stagnation
conditions are predominant and DBP exposure occurs. The model predicts the DBP formation
potential in one-dimension and involves chlorine as disinfectant and chloroform and DCAN as
DBPs. Chloroform and DACN were selected because the first is the most abundant and
predominant DBP in drinking water and the second is the most predominant haloacetonitrile
species detected in drinking water (WHO, 2017) and in biomass disinfection tests (Wang et al.,
2012c), and it has been recognized as a more toxic substance than HAAs (Muellner et al., 2007).

To the author’s knowledge, models representing the contribution of the biofilms to the bulk water
have been developed only for particulate substances such as biofilm erosion or detachment, but
not for dissolved substances. Following this line, the current model represents an original approach
in biofilm and DBP modelling by considering the biofilm matrix as an organic matter reservoir
penetrated by chlorine, which oxidizes EPS and cells, then transforming a fraction of them into
chloroform and DCAN, separately, which are then transported by diffusion from the biofilm to the
bulk water. The influence of pertinent parameters on bulk water concentrations is discussed, and
the model applications and limitations are also explained.

4.2 MODEL DEVELOPMENT

4.2.1 Model definition

According to Vannecke et al. (2015), it is likely a general rule that increased complexity concerning
microbial diversity and/or meso-scale aggregate architecture will be more useful when the focus is
on understanding fundamental micro-scale outputs. When the focus is on macro-scale outputs
(e.g., substrate removal rates, optimal bulk condition), this complexity is clearly not always
necessary. However, under some conditions, such additional model features can be critically
informative for bulk reactor behaviour prediction or understanding.

The main purpose of the model developed in this chapter is to represent the formation potentials
of chloroform and DCAN by the oxidation of the biofilm matrix with HOCl. As the diffusion processes
of simulated substances occur in minutes, biofilm growth, decay and detachment and EPS

Chapter 4. A simple biofilm chlorination model 119

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
formation occur in hours (Picioreanu et al., 2000a), initial constant concentration of cells and EPS
and their decay due to the reaction with chlorine within the biofilm were considered. A 1D model
was produced to initially represent the experimental results from Wang et al. (2013a), and then the
model was extrapolated to plumbing pipes under conditions where stagnation prevails. A schematic
representation of the model is presented in Figure 4-1.

Figure 4-1. Schematic representation of the 1D model

The main assumptions made when developing the current model are:

 Closed system: Glass vessels used in laboratory experiments and drinking water pipes under
normal operation can be considered closed systems, where there is not an exchange of
materials with the surrounding environment. Therefore, the model presented here also
represents a closed system.
 Stagnant fluid: During the evenings, stagnation conditions can be present in DWDNs, especially
in residential areas. In plumbing systems, water can be stagnant during periods of no
consumption of water such as the evenings, weekends, or holiday seasons.
 Flat biofilm: A biofilm can be compact or present voids and channels within itself. Compact
biofilms are the result of systems limited by the biomass growth rate. Finger-like and mushroom-
like biofilms are generated in systems limited by the substrate transport (Picioreanu et al.,
1998b). Biofilms with irregular surfaces gradually shift to compact biofilms with smooth surface
when nutrient availability increases due to higher flow velocities (Picioreanu et al., 2000a).
Therefore, it is expected that, under static flow conditions, biofilm should develop with channels,
voids, and rough surface. However, biofilms in drinking water pipes of plumbing systems are
exposed to both turbulent flow and stagnant conditions but there is not yet evidence on the
morphology characteristics of biofilms growing under these conditions. Considering that
numerical simulation of complex morphologies of biofilms such as mushroom-like shape is

Chapter 4. A simple biofilm chlorination model 120

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 computationally expensive, flat biofilms were assumed here to reduce the complexity of the
calculations.
 Biofilm is a continuous medium; cells and EPS are homogeneously mixed in biofilm matrix: The
biofilm is treated here as a homogeneous system. Voids and channels as part of the morphology
of the biofilm and the protective role of EPS against cell disinfection are not simulated in the
current model to reduce geometric elaboration complexity, which would introduce further
solution complexity. Since this model simulates the DBP formation potentials, this assumption
means that chlorine reacts at the same time with both cells and EPS.
 Constant biofilm thickness: As described in Table 2-3, Table 2-8, and Table 4-2, biofilm growth
can take several hours or days. Consequently, the computational time required to simulate
biofilm growth can be in the order of hours, days or weeks, which is subject to the computational
resources available. This also implies that internal boundary represented by biofilm surface is
fixed, i.e., it does not change as cells and EPS are being transformed into DBPs.
 Fixed initial concentration of cells (X) and EPS (E): This was assumed as the current model
simulates the DBP formation potentials and it does not include the biofilm growth.
 Bulk water is a complete mixed reactor: Laboratory experiments of biofilm disinfection have
been carried out in small glass vials (40 mL, 65 mL) (Wang et al., 2012c; Wang et al., 2013a).
Therefore, it is reasonable to assume this type of reactor, given the small volumes. However,
plumbing systems can represent volumes of the order of litres (e.g., 6.3 L for  = ½ inches and
L = 50 m) and the flow is usually represented by plug flow. Such simplification must be carefully
considered for risk assessment since DBP exposure at the point of use depends on which tap
is first opened.
 Chloroform (S) and DCAN (S) are modelled separately according to fraction of cells and EPS
converted into the respective DBP: The current model was developed to simulate individual
species of DBPs in order to make it flexible. The amount of DBP species formed relies on the
fraction of cells (Fx) and EPS (Fe) transformed into DBPs; then, as the experimental data are
available, it is relatively easy to predict concentrations of individual DBP species from biofilm
chlorination.

4.2.2 Model equations

Figure 4-2 sketches the 1D-stagnation model and the corresponding equations are included in
Table 4-1. The diffusion of two dissolved substances, chlorine and DBP, through the biofilm matrix
was simulated, considering cells and EPS decay. The formation of chloroform and DCAN is

Chapter 4. A simple biofilm chlorination model 121

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
expressed in terms of yield coefficients from disinfection of cells and EPS and their fraction
transformed into DBP. Chloroform and DCAN were simulated separately according to each liquid
diffusion coefficient and the fraction of cells and EPS transformed into the respective DBP. Bulk
water concentrations of dissolved substances are defined by the boundary condition in the biofilm
surface, assuming that bulk water is a complete mixed reactor (Equations (4-8) and (4-14)). Under
stagnation conditions, diffusion processes are dominating the substance transport; therefore Fick’s
law is applied in the biofilm and source/sink terms are also included.

Figure 4-2. 1D mass balance equations for soluble and particulate components in biofilm and
biofilm surface

Table 4-1. Model equations in biofilm

Equation Description of variables Subscripts
C: Disinfectant concentration
X: Cell concentration
E: EPS concentration
C: Chlorine
k1: Disinfectant decay rate – cells
S: DBP
𝑑𝐶 𝑑 𝑑𝐶 disinfection
= (𝐷𝐶−𝐵 ) − 𝑘1 𝐶𝑋 − 𝑘2 𝐶𝐸 X: Cells
𝑑𝑡 𝑑𝑧 𝑑𝑧 k2: Disinfectant decay rate – EPS
E: EPS
oxidation
B: Biofilm
(4-1) DC-B: Molecular diffusion coefficient
L: Bulk liquid
of chlorine in biofilm
Bs: Biofilm surface
Ab: Biofilm area
V: reactor volume
t: Time

Chapter 4. A simple biofilm chlorination model 122

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Equation Description of variables Subscripts
X: Cell concentration
𝑑𝑋 k1: Disinfectant decay rate – cell
= −𝑌1 𝑘1 𝑋𝐶
𝑑𝑡
disinfection
(4-2)
Y1: Cell-chlorine yield coefficient

E: EPS concentration
𝑑𝐸 k2: Disinfectant decay rate – EPS
= −𝑌2 𝑘2 𝐸𝐶
𝑑𝑡
oxidation
(4-3)
Y2: EPS-chlorine yield coefficient

S: DBP concentration
DS-B: Diffusion coefficient of
chloroform/DCAN in biofilm
𝑑𝑆 𝑑 𝑑𝑆 X: Cell concentration
= (𝐷𝑆−𝐵 ) + 𝐹𝑋 𝑘1 𝐶𝑋 + 𝐹𝐸 𝑘2 𝐶𝐸
𝑑𝑡 𝑑𝑧 𝑑𝑧 E: EPS concentration

(4-4) FX: Fraction of cells transformed

into DBP
FE: Fraction of cells transformed
into DBP

4.2.3 Parameters selection

4.2.3.1 Biofilm thickness (BT)

Considering that the current model does not include bacteria growth, biofilm thickness (BT) arises
as an important input parameter, since it defines the domain size and other parameters such as
initial concentration of EPS (Eo). In order to define appropriate values of BT, several drinking-water-
related studies were reviewed to identify the reports of this parameter. BT values and the
experimental conditions of these studies are described in Table 4-2. It should be noted here that
conditions such as inoculum, incubation time, method of BT measurement are diverse and it is also
not possible to identify any correlation between biofilm and thickness and biofilm age (see Table
4-2).

Chapter 4. A simple biofilm chlorination model 123

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 4-2. Review of biofilm thickness data reported by drinking-water-related studies
Tests description
Type of Method for Biofilm
Study Inoculum / thickness
reactor for Incubation Substrate measuring
Bacterial (m)
biofilm growth time medium biofilm
community
/ experiments thickness
773
428
Chen Bacteria P. Optical
Artificial 476
and aeruginosa ERC1 method of
biofilms growth N.S Rich medium
Stewart dispersed in Bakke and 771
in flow cell
(1996) agarose gel slabs Olsson
456
526
Optical 6.2
Minimal-salts
Peyton Bacteria P. method of
Annular reactor 14.6 days medium with 17.61
(1996) aeruginosa Bakke and
glucose 31.85
Olsson
1 day 0.27
2 days 0.78
4 days Bacteria P. putida 3.03
7 days 4.36
10 days 7.46
1 day 2.33
2 days 2.41
Bacteria P.
4 days 3.75
aureofaciens
7 days CLSM 5.17
Modified FAB
images
Heydorn Laboratory 10 days medium 7.06
processed
et al. channel flow supplemented
1 day using the 0.66
(2000) cells with sodium
program
2 days citrate 1.37
COMSTAT
Bacteria P.
4 days 3.21
fluorescens
7 days 5.61
10 days 8.47
1 day 3.43
2 days 2.49
Bacteria P.
4 days 4.96
aeruginosa
7 days 4.52
10 days 5.65
CLSM
images 30
Xue et Laboratory
Bacteria P. 0.02 strength processed
al. channel flow 6 days
aeruginosa LB broth using the
(2012) cells
program 40
COMSTAT
Bacteria P. CLSM
Xue and Laboratory Nutrient
aeruginosa - images
Seo channel flow 6 days growth 159
algT(U) - no processed
(2013) cell medium: 0.02
disinfection using the
Chapter 4. A simple biofilm chlorination model 124
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Tests description
Type of Method for Biofilm
Study Inoculum / thickness
reactor for Incubation Substrate measuring
Bacterial (m)
biofilm growth time medium biofilm
community
/ experiments thickness
Bacteria P. strength LB program
aeruginosa - PAO1 broth COMSTAT 162
- no disinfection
Bacteria P.
aeruginosa -
164
mucA22 - no
disinfection
Bacteria P.
aeruginosa -
algT(U) - chlorine 44
disinfection (0.5
mg/L)
Bacteria P.
aeruginosa - PAO1
- chlorine 85
disinfection (0.5
mg/L)
Bacteria P.
aeruginosa -
mucA22 - chlorine 149
disinfection (0.5
mg/L)
Bacteria P.
aeruginosa -
algT(U) - disinfected 38
with
monochloramine
Bacteria P.
CLSM
aeruginosa - PAO1
images 47
Xue et Laboratory - disinfected with
0.02 strength processed
al. channel flow 6 days monochloramine
LB broth using the
(2014) cells Bacteria P.
program
aeruginosa -
COMSTAT
mucA22 - 46
disinfected with
monochloramine
Bacteria P. putida -
disinfected with 48
monochloramine
PVC coupons 1 year 120
Shen et Optical
in CDC Groundwater source for drinking
al. coherence
laboratory 1 month water 93
(2016) tomography
reactors
N.R: not reported | CLSM: confocal laser scanning microscopy | CDC: Centres for disease control and prevention

From Table 4-2, 43 BT values were reported in seven studies and their descriptive statistics are
included in Table 4-3. Figure 4-3 revealed that 50% of biofilm thickness values varied between 4
and 107 m (quartiles 1 and 3 -Q1, Q3-). This range was used to randomly vary this parameter in
Chapter 4. A simple biofilm chlorination model 125
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
simulations with the current model. It is worth noting that all the reviewed studies (Table 4-2) used
bacteria Pseudomonas for their experiments since they are EPS producers, then biofilm formers,
and have been identified in DWDNs, in both temperate and tropical climate locations. As described
in Chapter 3, Pseudomonas genus was common in most of the biofilm samples (8 of 9 points)
collected in Colombia. Douterelo et al. (2014c) identified Pseudomonas genus as the dominant
group in the initial process of biofilm formation in 28-days-old biofilms in a laboratory-based full
scale DWDN fed with drinking water supplied to the city of Sheffield (UK).

On the other hand, the study of influence of long-term disinfection on mechanical properties of
biofilms (Shen et al., 2016) found that BT is recovered after one month of continuous chlorination
(4 mg/L), in relation to initial BT before tests. It is therefore reasonable to assume that the thickness
of laboratory-incubated biofilms over two weeks is fairly presentative of more established biofilms.

Table 4-3. Descriptive statistics of biofilm

thickness reported in Table 4-2
Descriptive Biofilm thickness
statistic (m)
Minimum 0.27
Q1 4
Median 18
Q3 107
Maximum 773
Average 111
Std. deviation 199
CV 178%
Number of data 43
Figure 4-3. Box plot biofilm thickness data

4.2.3.2 Initial volumetric biofilm density (Xo)

The DBP potential formation experiments developed by Wang et al. (2013a) managed cells density
6.5 ± 0.6 CFU/cm2, on a PVC surface, in glass vial volume (65 mL), chip area (1x10 -4 m2), and
number of chips per disinfection tests (10) 10. In order to convert the cell density data into volumetric
density, the relationship between cell counting units and biomass for P. aeruginosa developed by
Kim et al. (2012) was used (Equation (4-5)). They cultured bacteria on Luria-Bertani medium, in 1
L flask, and harvested them for further colony counting and measurement of dry weight of biomass
(Kim et al., 2012). The authors recommended application of this relationship in transport models in

10 J.J. Wang, 2016, personal communication, 19th January 2017

Chapter 4. A simple biofilm chlorination model 126
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 aquifers since cell concentrations in water samples are usually determined by counting methods
(Kim et al., 2012). The initial volumetric biofilm density was then established equal to 40.87 mg/L.

𝑌 = 1.8𝑥108 𝑋 − 3.6𝑥106 (4-5)

Here Y is colony forming unit (CFU/mL) and X is biomass (mg/mL). In order to simulate the
formation of DBPs in pipe/plumbing systems, cell density data was reviewed in studies previously
published, which consisted of biofilm monitoring in real-scale DWDNs by installation of coupons or
biofilm reactors. The cell density was determined by the methods 4',6-diamidino-2-phenylindole
staining and epifluorescence microscope; data and further details of the reviewed studies are
presented in Table 4-4.

Table 4-4. Review of cell density data in drinking water biofilm reported by other researchers
Tests description Total cell
Author Biofilm Pipe/coupon count
Type of reactor Water quality Area (cells/cm2) (a)
age material
Glass 7.20E+05
35 days
Two reactors were PE 4.30E+06
Kalmbach installed in a Glass 7.20E+05
Cl: 0 mg/L, T: 12.2
et al. domestic DWDN at 56 days 2 cm2
PE °C, TOC: 3.3 mg/L 1.80E+06
(1997) U. of Berlin (water
taps) Glass 7.00E+05
70 days
PE 1.50E+06

Two biofilm reactors

3.60E+06
located at the outlet
of WTP in Germany 18 Steel, PVC, 1650
T: 11.7 °C
(no disinfected months and PE mm2
drinking water) and
3.10E+07
Wingender within the DWDN
and
Flemming CI 7.00E+06
(2004) Cement-lined 3.00E+05
Galvanized
(2-99 2.00E+08
steel Cl: <0.01 - 0.13
18 pipe sections years Several
mg/L 4.80E+05
old)
5.00E+07
PVC
8.00E+05
1.60E+07
Cl < 0.3 mg/L, T:
Lee and Semi-pilot  13 mm,
Galvanized 16.0 - 25.1 °C, 975
Schwab L=8.8 m connected 84 days 7.00E+06
iron TOC: 1.27 - 2.65 mm2
(2005) to a DWDN in Seoul
mg/L
Långmark Pilot plant prior to Cl: 0.26 mg/L,
80-90
et al. the distribution in PE AOC: 34 mg/L, T N.R 3.00E+04
days
(2005) two systems, outlet (winter): 6.6 °C

Chapter 4. A simple biofilm chlorination model 127

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Tests description Total cell
Author Biofilm Pipe/coupon count
Type of reactor Water quality Area (cells/cm2) (a)
age material
of the WTP, PE
tubing, system A
T (autumn): 7.4 °C 6.00E+04
System A - DWDN T (summer): 18.8
4.00E+04
°C
Cl: 0.24 mg/L,
AOC: 36 mg/L, T 1.00E+04
(winter): 8.5 °C
System A - Pilot
T (autumn): 11.5
plant 8.00E+04
°C
T (summer): 13.2
1.00E+05
°C
Coupons were
located post Cl: 0.09 mg/L,
treatment and at the AOC: 15 mg/L, T 9.50E+04
end of the DWDN. (winter): 8.8 °C
System B
T (autumn): 10.7
9.00E+04
°C
System B - DWDN
T (summer): 20.8
Glass 3.00E+05
°C
Cl: 0.02 mg/L,
AOC: 27 mg/L, T 3.00E+05
(winter): 9.4 °C
System B - Pilot
T (autumn): 14.4
plant 3.50E+05
°C
T (summer): 15.2
7.00E+05
°C
Three reactors:
Batch, Propella,
Cl: 0.15 mg/L, T:
Flow cell connected 4.9 cm2 8.00E+06
Manuel et 15.9 °C, TOC: 2.32
to Porto's DWDN. 56 days PVC
al. (2007) mg/L, DOC: 2.12
Batch
mg/L
Propella and Flow
2 cm2 5.00E+07
cell
Flow cell reactor
connected to
Porto's DWDS at
1.10E+07
different flow
regimes. Stagnation
Cl: 0.15 mg/L, T:
Manuel et - turbulent flow
20 days PVC 20 °C, TOC: 2.3 2 cm2
al. (2009) Stagnation - laminar
mg/L 2.60E+07
flow
Turbulent - laminar
7.40E+06
flow
Continuous -
7.80E+06
laminar flow
The methods 4',6-diamidino-2-phenylindole staining and epifluorescence microscope were used in every studies to
(a)

count total cells

Chapter 4. A simple biofilm chlorination model 128

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Table 4-5 and Figure 4-4 include the respective descriptive statistics of the cell density data. Ten
values of Xo were randomly selected within the range defined by Q1 and Q3 (3x10 5 – 7.8x106
cells/cm2) for the model developed here. By using the weight of single cell bacterium E. coli (6.65
x 10-10 mg), biofilm area and reactor volume, cell density count values were transformed into
volumetric concentration (mg/L).

Table 4-5. Descriptive statistics of cell

density in drinking water biofilm reported
by other researchers
Descriptive Total cell count
statistic (cells/cm2)
Minimum 1.00E+04
Q1 3.00E+05
Median 7.20E+05
Q3 7.80E+06
Maximum 2.00E+08
Average 1.31E+07
Std. deviation 3.55E+07
CV 271.83%
Number of data 33
Figure 4-4. Box plot cell density data

4.2.3.3 Initial concentration of EPS (Eo)

Experimental tests of mechanical properties of drinking water biofilms indicate that EPS
concentration depends on biofilm thickness. Kundukad et al. (2016) found that polysaccharides
contribute to the stiffening of P. aeruginosa biofilms. A positive correlation was found between the
biofilm surface stiffness and the micro-colony size, which was attributed to the differential
expression of the biofilm’s EPS according to the stages of biofilm growth. Similarly, Shen et al.
(2016) concluded that long-term disinfection of one-year-old drinking water biofilms may not
significantly remove net biomass; after one month of disinfection, bacteria adapt to disinfectant,
produce EPS and replenish the outer layer.

Celmer et al. (2008) grew biofilms from autotrophic denitrifiers in two laboratory-scale membrane
biofilm reactors, at temperatures 19.61 and 19.18 °C, and pH 7.09 and 7.02. These authors
reported data of biofilm thickness, total and volatile solids, and EPS concentrations, which were
used to define the linear correlation presented in Figure 4-5. This equation was included in the
current model as an indirect way to involve the EPS production by cells.

Chapter 4. A simple biofilm chlorination model 129

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 7000

EPS concentration (mg/L)

6000

5000

4000

3000

2000 y = 5.0295x + 1884.3

R² = 0.9803
1000

0
0 200 400 600 800 1000
Biofilm thickness (m)
Figure 4-5. Linear correlation between biofilm thickness and EPS concentration (Celmer et al.,
2008)

4.2.3.4 Effective diffusion coefficient of dissolved substances (  C and  S)

Taking into account that DBP formation from biofilm disinfection under stagnation conditions is
represented by the Fick’s law for the diffusive processes of chlorine and DBP that are involved,
diffusion of substances reduces within the biofilm matrix in comparison to the liquid phase as a
consequence of the presence of EPS (Westrin and Axelsson, 1991). Therefore, it is necessary to
define a ratio of diffusion coefficients, D liquid / Dbiofilm for each substance. This ratio is referred to as
C for chlorine and S for DBPs. Chlorine diffusion coefficient in liquid and biofilm reported by Chen
and Stewart (1996) were used in the present model; resulting in C = 0.94. The value of S was
defined by using several values within the range 0.016-1.0. The diffusion coefficient in liquid for
chloroform and DCAN were 9.288x10-10 m2/s and 2.67x10-5 m2/s, respectively (Buzatu et al., 2007;
Poling et al., 2007).

4.2.3.5 Reaction rates (k1 and k2) and yield coefficients (Y1 and Y2)

Yield coefficients represent the mass of cells or EPS in which reactive sites have been depleted
per mass or chlorine consumed (Chen and Stewart, 1996). Chlorine-cell and chlorine-EPS reaction
rates and yield coefficients found by Chen and Stewart (1996) in the short-duration batch
experiment (20 min) were used in the present 1D model (see Table 4-6). However these authors
did not specify the temperature at which tests were carried out; for the current study, it was
assumed to be 20 °C as was used by Abokifa et al. (2016a). The pH of these experiments were
maintained in 7.2.

Chapter 4. A simple biofilm chlorination model 130

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 4-6. Values of reaction rates (k1 and k2) and yield coefficients (Y1 and Y2)
Parameter Value
Chlorine-cell reaction rate (k1) (m3/g-sec) 1.1x10-3
Chlorine-EPS reaction rate (k2) (m3/g-sec) 3.7x10-6
Chlorine-cell yield coefficient (Y1) (g/g) 1.85
Chlorine-EPS yield coefficient (Y2) (g/g) 540

4.2.3.6 Factors FX and FE

Factors FX (0.7) and FE (0.3) were used in the DBP formation equation to represent the portion of
cells and EPS transformed into DBPs. Experimental data reported by Fang et al. (2010b) were
processed to find the fractions of chloroform formed from disinfection of extracellular organic matter
and algal cells. M. aeruginosa laboratory tests consisted on chlorination and chloramination of 100
mL of solutions of extracellular organic matter, intracellular organic matter, and NOM, with TOC
content 5 mg/L, temperature 22 °C, pH = 7.0, during three days (Fang et al., 2010b). For DCAN,
such factors were adjusted to find similar values of this DBP reported by Wang et al. (2013a) with
the same values of BT resulting in similar values of chloroform reported by these authors. Such
adjustment resulted in FX = 0.3 and FE = 0.17. These factors do not necessarily sum to one, since
several DBPs may be formed simultaneously and the current model is simulating chloroform and
DCAN separately.

4.2.3.7 Temperature

Temperature is widely recognized as one of the most important factors in chemical reactions by
increasing reaction rates at higher temperatures. Temperature dependence of reaction rate is
expressed in terms of the Arrhenius equation (Levenspiel, 1999). Having this in mind, the empirical
relationship derived by Kiéné et al. (1998) was used to indirectly include temperature within the
model by adjusting the reaction rates k1 and k2 in the range 5-25 °C (Equations (4-6) and (4-7)).

−6050 −6050
𝑘1 = 0.0011𝑥℮(𝑇+273 ) (4-6) 𝑘2 = 3.70𝑥10−6 𝑥℮(𝑇+273 ) (4-7)

Units of constants are: 0.011 m3/g-sec, -6050 K, T+273 K, and 3.70 x 10-6 m3/g-sec.

Chapter 4. A simple biofilm chlorination model 131

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 4.2.4 Boundary conditions

Table 4-7 includes the initial and boundary conditions of the model. At t = 0, chlorine is assumed
to be completely mixed in bulk water and its concentration within the biofilm is zero; DBP
concentration is also assumed to be zero in the entire domain at t = 0. The flux boundary condition
is included in the biofilm surface (Bs) (Equations (4-8) and (4-14)); dissolved substances decay or
increase according to this flux, biofilm area and reactor volume (IWA Task Group on Biofilm
Modeling et al., 2006). On the substratum (pipe wall), there is no exchange of dissolved and
particulate substances (i.e., no flux).

Table 4-7. Initial and boundary conditions in 1D model

Initial
Boundary conditions Description of variables Subscripts
conditions
C: Disinfectant concentration
𝑑𝐶 𝑑𝐶 𝐴𝐵 DC-B: Molecular diffusion
= −𝐷𝐶−𝐵 |𝐵𝑠 ( )
𝑑𝑡 𝑑𝑧 𝑉
coefficient of chlorine in biofilm
t=0 Ab: Biofilm surface
CB = 0 (4-8)
V: Reactor volume
CBs = Co
𝑑𝐶 Co: Initial concentration of
𝑗𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑢𝑚 = −𝐷𝐶−𝐵 =0
𝑑𝑧
chlorine
(4-9)
t: Time
𝑑𝑋
𝑗𝐵𝑠 = −𝐷𝑋−𝐿 |𝐵𝑠 = 0
𝑑𝑧
C: Chlorine
X: Cells concentration S: DBP
t=0 (4-10)
XB = Xo Xo: Initial cell concentration X: Cells
𝑑𝑋
𝑗𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑢𝑚 = −𝐷𝑋−𝐵 =0 E: EPS
𝑑𝑧
B: Biofilm
(4-11)
𝑑𝐸 L: Bulk liquid
𝑗𝐵𝑠 = −𝐷𝐸−𝐵 |𝐵𝑠 = 0
𝑑𝑧 Bs: Biofilm surface

E: EPS concentration
t=0 (4-12)
EB = Eo Eo: Initial EPS concentration
𝑑𝐸
𝑗𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑢𝑚 = −𝐷𝐸−𝐵 =0
𝑑𝑧
(4-13)
S: DBP concentration
𝑑𝑆 𝑑𝑆 𝐴𝐵
= −𝐷𝑆−𝐵 |𝐵𝑠 ( ) DS-B: Diffusion coefficient of
t=0 𝑑𝑡 𝑑𝑧 𝑉
SBs = 0 chloroform/DCAN in biofilm
SB = 0 Ab: Biofilm area
(4-14)
V: Reactor volume

Chapter 4. A simple biofilm chlorination model 132

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Initial
Boundary conditions Description of variables Subscripts
conditions
𝑑𝑆
𝑗𝑠𝑢𝑏𝑠𝑡𝑟𝑎𝑡𝑢𝑚 = −𝐷𝑆−𝐵 =0
𝑑𝑧
(4-15)

4.2.5 Geometry and mesh

The current model was developed in one dimension. In this case the geometry is represented by a
line with length representing the biofilm thickness. The mesh was built by dividing the line into
several elements (see Figure 4-6). It was implemented within the software COMSOL Multiphysics
5.2a.

Figure 4-6. Representation of the 1D model mesh

4.2.6 Analysis of mesh dependence

Figure 4-7 shows the influence of element size on the chloroform concentration in bulk water. As it
is expected, data from coarser grids are more dissimilar that those from finer grids. The percentage
of difference pairs of maximal, median and average values for extremely fine and extremely coarse
meshes is 17.7%, while this parameter resulted in 0.5% among extremely fine and extra fine
meshes. The results presented in this chapter were obtained by employing extremely fine grids,
since the computational time was very low (less than 9 s). The number of elements of every type
of mesh is indicated in Table 4-8.

It is important to mention that, for simulations with the lowest BT (7 m), the length of the elements
was 0.07 m, which may challenge the applicability of the use of continuous equations. However,
implementing such low resolution was necessary in order to solve the equations representing the
fast reaction between chlorine and biomass, as mentioned in section 4.2.7. Furthermore, the results
of the model developed here resulted in agreement with experimental data obtained by other
researchers, as discussed in the section 4.3.5.

Chapter 4. A simple biofilm chlorination model 133

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 4-7. Influence of grid size on simulation results

Table 4-8. Number of elements of every type of mesh

Number of domain Number of domain
Type of mesh Type of mesh
elements elements
Extremely fine 100 Coarse 10
Extra fine 50 Coarser 8
Finer 27 Extra coarse 5
Fine 19
Extremely coarse 3
Normal 15

4.2.7 Processing and numerical methods

Initially, model equations were coded and solved in MATLAB R2015b by discretizing them with the
explicit scheme finite difference scheme, central difference in space, and forward in time. As a
consequence of the small diffusion coefficients, the model is represented by stiff equations.
Therefore, small time step and grid size have to be used in order to keep the stability of the solution.
This resulted in high computational time: up to 6 hours for chloroform and more than one week for
DCAN simulations. For reasons of numerical efficiency, the model was implemented in COMSOL
Multiphysics 5.2a, which also applies finite difference elements as numerical method and employs
stabilization methods to solve the transport of species module. The computational time of COMSOL
is up to 9 s using an extremely fine mesh.

Chapter 4. A simple biofilm chlorination model 134

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
The implementation of the model in MATLAB was an important step in this project because the
author acquired modelling and a deeper understanding of the governing equations during this
phase. Consequently, the discretization of the equations is described as it follows. Equations (4-16)
and (4-17) shows the forward difference in time approximation for the dissolved substances
chlorine and DBP, respectively. For the second order derivative of both chlorine and DBP, central
difference in space is used (Equations (4-18) and (4-19), respectively). Then, by combining both
approximations, the solution for the time and space variation of chlorine and DBP is given by
Equations (4-20) and (4-21), respectively, which also include the reaction terms.

a) Biofilm sub-domain

𝐶 (𝑡+𝛿𝑡) − 𝐶 (𝑡) 𝑆 (𝑡+𝛿𝑡) − 𝑆 (𝑡)

𝛿𝑡 𝐶 = (4-16) 𝛿𝑡 𝑆 = (4-17)
∆𝑡 ∆𝑡

(𝐶𝑖+1 − 2𝐶𝑖 + 𝐶𝑖−1 )(𝑡) (𝑆𝑖+1 − 2𝑆𝑖 + 𝑆𝑖−1 )(𝑡)

𝛿𝑧2 𝐶 = (4-18) 𝛿𝑧2 𝑆 = (4-19)
∆𝑧 2 ∆𝑧 2

𝑛
𝐶𝑖+1 − 2𝐶𝑖𝑛 + 𝐶𝑖−1
𝑛
𝐶𝑖𝑛+1 = 𝐶𝑖𝑛 + ∆𝑡 ∗ [𝐷𝑐 ( ) − 𝑘1 𝐶𝑖𝑛 𝑋𝑖𝑛 − 𝑘2 𝐶𝑖𝑛 𝐸𝑖𝑛 ] (4-20)
∆𝑧 2

𝑛
𝑆𝑖+1 − 2𝑆𝑖𝑛 + 𝑆𝑖−1
𝑛
𝑆𝑖𝑛+1 = 𝑆𝑖𝑛 + ∆𝑡 ∗ [𝐷𝑆 ( ) + 𝐹𝑋 𝑘1 𝑆𝑖𝑛 𝑋𝑖𝑛 + 𝐹𝐸 𝑘2 𝑆𝑖𝑛 𝐸𝑖𝑛 ] (4-21)
∆𝑧 2

To solve the boundary condition in the biofilm surface for dissolved substances, the same
approximations described previously were applied (Equations (4-22) and (4-24) for chlorine and
DBP, respectively). The solution of the boundary condition is given by the Equations (4-23) and
(4-25); the terms with the subscript Bs+1 and Bs-1 refers to the adjacent elements to the biofilm
surface in bulk water and biofilm, respectively.

b) Biofilm surface (Bs)

𝐴𝐵 𝐶𝐵𝑆+1 − 𝐶𝐵𝑆−1 𝐴𝑏
𝐶𝑖𝑛+1 = 𝐶𝑖𝑛 − ∆𝑡 ∗ 𝑗𝐶−𝐵𝑠 ( ) (4-22) 𝑗𝐶−𝐵𝑠 = 𝐷𝐶−𝐿 ( ) (4-23)
𝑉 2∆𝑧 𝑉

𝐴𝐵 𝑆𝐵𝑆+1 − 𝑆𝐵𝑆−1 𝐴𝑏
𝑆𝑖𝑛+1 = 𝑆𝑖𝑛 + ∆𝑡 ∗ 𝑗𝑆−𝐵𝑠 ( ) (4-24) 𝑗𝑅−𝐵𝑠 = 𝐷𝑆−𝐿 ( ) (4-25)
𝑉 2∆𝑧 𝑉

Chapter 4. A simple biofilm chlorination model 135

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
4.2.8 Model implementation in COMSOL Multiphysics

The current model for prediction of DBP in drinking water under stagnation conditions was
implemented in the software COMSOL Multiphysics 5.2a. This software is structured by modules
to represent physical and chemical processes taking place within a system. According to the
specific needs of the modeller, COMSOL allows the addition of the required modules and their
coupling. The modules used for running the current model and the boundary and initial conditions
are described in Table 4-9.

Table 4-9. Main features for model implementation in COMSOL Multiphysics 5.2a
Boundary /
Module Condition Initial condition (t=0)
Domain
Wall No flux
Concentration obtained
Transport of diluted Biofilm surface (Bs) from Boundary ODE
module (a) Co = 0 ; So = 0
species (tds)
Reactions for chlorine
Domain decay and chloroform
formation
Reactions for cells and
Domain ODE Domain Xo = Xo; Eo = Eo
EPS decay
Boundary ODE Biofilm surface (Bs) Source term: Flux Co = Co and So = 0
ODE: ordinary differential equations | (a) Constraint Individual dependent variables was included for coupling tds with Boundary ODE
module

4.2.9 Selection of time step

Time step was determined after a sensitivity study, testing several values (i.e., 1 s, 5 s, 10 s, 30 s,
1 min, 5 min, 10 min, and 15 min). As it can be observed in Figure 4-8, no differences were identified
among the eight sets of data. Percentage of difference between each consecutive pair of maximal
and median values was 0.0% and less than 4.2% for average values. Thus, for short time
simulations (less than 3 hours), time step was 30 s and 15 min for long time simulations (i.e., more
than 8 hours).

Chapter 4. A simple biofilm chlorination model 136

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 4-8. Time step comparison among chloroform concentrations in bulk water

4.2.10 Experimental study used as a reference

The work of Wang et al. (2013a) was used as a reference to compare the current model results
with those obtained in their DBP formation potential tests. Their experiments considered chlorine
and chloramine disinfection of P. aeruginosa (PAO1) biofilms, which were grown on chips with area
equal to 100 mm2. They used 100 mL bacterial solution and 20 mL of nutrient broth solution for
both PVC and galvanized zinc chips, which were incubated at 25 °C in the dark. The cell density
biofilm was monitored at 4, 8, 12, 24, 36 and 60 h; the biofilms from both types of chips reached
maturation after 24 h incubation. Biofilm thickness was not reported in this study. Then, DBP
formation potential was tested by disinfection of chips in 65-mL vials, at pH 8 and temperature of
20 °C, for 24 hours. The cell density was adjusted to 6.5 CFU/m2.

The results for PVC chips were considered here since this material is commonly used in the DWDN
of the city of Cali (53.3% of the total network length), as discussed in Chapter 3. THM and
haloacetonitrile (HAN) yielded 7.7±1.1 g/mg C and 1.9 ± 0.3 g/mg C, respectively. By
considering TOC content of biofilms of  0.1 mg11 and assuming that chloroform represents 71% 12
of total THMs (TTHMs) allowed determining the range of 6.6–9.6 g/L for chloroform and 2.6-3.3

11 Email: J.J. Wang, 2016, personal communication, 25th August 2016

12 This percentage was determined by results of chlorination of E. coli suspension. Wang, J.-J. et al. 2013a. Disinfection byproduct
formation from chlorination of pure bacterial cells and pipeline biofilms. Water Research. 47(8), pp.2701-2709.
Chapter 4. A simple biofilm chlorination model 137
Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
g/L for HANs13. The summary of their experimental conditions and results are presented in Table
4-10.

Table 4-10. Experimental conditions and results of tests developed by Wang et al. (2013a)
Parameter Value
pH (chlorination reactions) 8.0 ± 0.2
Temperature (chlorination reactions) 25 °C
Biofilm incubation 60 hours at 25 °C
Contact time for DBP formation potential 24 hours
0.1 mg C / 65 mL
TOC
1.54 mg/L
Amount of chips used in DBP potential formation 10
TTHMs yields (PVC chips) 7.7 ± 1.1 mg-Cl2/mg-C
HANs yields (PVC chips) 1.9 ± 0.3 mg/mg-C
Chloroform range concentration 6.62-9.46 g/L
DCAN range concentration 2.62-3.22 g/L

4.2.11 Data analysis

The indicator percentage of difference was widely used in order to quantify the similarities and
dissimilarities among results obtained with different features of the 1D model. Percentage of
difference between “value 1” and “value 2” is expressed in the Equation (4-28).

𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 = 𝑉𝑎𝑙𝑢𝑒 1 − 𝑉𝑎𝑙𝑢𝑒 2 (4-26)

𝑉𝑎𝑙𝑢𝑒 1 + 𝑉𝑎𝑙𝑢𝑒 2 (4-27)

𝐴𝑣𝑒𝑟𝑎𝑔𝑒 =
2

𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 (4-28)
𝑃𝑒𝑟𝑐𝑒𝑛𝑡𝑎𝑔𝑒 𝑜𝑓 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 = 𝑥100
𝐴𝑣𝑒𝑟𝑎𝑔𝑒

Here value 1 and value 2 are any number of the statistic descriptors mentioned previously, which
are object of comparison. In some cases, maximal values were compared and, in others, the
median and average of entire sets of data were used to calculate the percentage of difference.
Median was also included here because it was more suitable for comparison purposes since this
descriptor is less sensitive to extreme values. In general, percentage of difference of 5% was used
as criteria to define whether pairs of values were different or similar.

13 100% of reported concentrations of HANs was assumed as DCAN as predominant specie within this group of DBP
Chapter 4. A simple biofilm chlorination model 138
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
4.3 RESULTS AND DISCUSSION

4.3.1 Typical model behaviour

To illustrate the general behaviour of the model, a simulation was run with BT=16 m, Clo=1 mg/L,
Xo=40.87 mg/L, which corresponded to biomass-limiting conditions, i.e., chlorine concentration is
enough to completely oxidize biomass and still residual concentrations remain within the biofilm.
Figure 4-9 presents the variation of dissolved substances within the biofilm, Figure 4-10 shows the
behaviour of the same variables in bulk water, and Figure 4-11 illustrates the decay or particulate
substances within the biofilm. Chlorine rapidly penetrates the biofilm (after 1 s) and chloroform
concentration varies from zero (at t=0) until a maximum value, then it decreases to very small
values due to the decay of cells and EPS and diffusion to the bulk water. In the bulk water, chlorine
decays while chloroform increases, until steady state is reached due to complete depletion of cells
and EPS, when chlorine has fully penetrated the biofilm.

(a)

(b)
Figure 4-9. Dissolved substances within biofilm (a) Chlorine (b) Chloroform

Chapter 4. A simple biofilm chlorination model 139

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 4-10. Dissolved substances in bulk water after 3 hours

(a)

(b)
Figure 4-11. Decay of particulate substances (a) Cells (b) EPS
Chapter 4. A simple biofilm chlorination model 140
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
4.3.2 Chlorine penetration within the biofilm

Chen and Stewart (1996) grew artificial biofilms with bacteria P. aeruginosa and polysaccharide
alginate to experimentally study the chlorine penetration within biofilms and modelled the process
by a mathematical representation of diffusion and chemical reactions. The experiments were
carried out in a flow cell with flow rate equal to 400 mL/min. The first step in the construction of the
current model was to compare the diffusion of chlorine within the biofilm to the results reported by
these researchers. Figure 4-12 indicates that, after 2 min, both models are able to reproduce the
experimental results; chlorine penetrates the biofilm up to 200 m. For t=15 min, both modelled
chlorine profiles also match with experimental data. Chlorine profiles from simulated data with the
current model, after 30 min and up to 1 hour, represents a faster diffusion of the disinfectant and
complete penetration of the biofilm, which may be related to the boundary condition used in both
models. The current model includes a stagnation condition and diffusion starts in the biofilm surface
(Figure 4-12b). Chen and Stewart (1996) was developed under flow condition and used the film
theory in the mass transfer boundary layer (IWA Task Group on Biofilm Modeling et al., 2006), then
diffusion starts in a small layer adjacent to the biofilm surface (Figure 4-12a). In general terms, the
mathematical representation of chlorine diffusion was acceptable to proceed with DBP simulations.

(a)

Chapter 4. A simple biofilm chlorination model 141

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 4-12. Comparison of chlorine penetration within biofilm (a) Chen and Stewart (1996):
experimental vs model data (b) Experimental data from Chen and Stewart (1996) vs simulated data
from current model

4.3.3 Sensitivity analysis

A single parametric sensitivity analysis (Loucks and van Beek, 2005) was done in order to define
the influence of the 14 parameters that compose the proposed model. This analysis was carried
out by varying each parameter one at a time while others were kept constant. Then, such influence
among parameters was compared by plotting the percentage of change of every parameter vs the
chloroform concentration in bulk water. The base values of each parameter used in this sensitivity
analysis are indicated in Table 4-11.

Table 4-11. Base parameters used in the sensitivity analysis

Parameter Base value
S 0.46
Biofilm thickness (BT) 49 m
Initial EPS concentration (Eo) 4775 mg/L
Initial chlorine concentration (Clo) 1 mg/L
Initial cell concentration (Xo) 40.87 mg/L
Temperature (T) 15 °C
Fraction of cells transformed into DBPs (Fx) 0.5
Fraction of EPS transformed into DBPs (FE) 0.5
Cell-chlorine yield coefficient (Y1) 1.85 (g/g)
EPS-chlorine yield coefficient (Y2) 540 (g/g)
Biofilm area (Ab) 10 cm2
Reactor volume (V) 65 mL

Chapter 4. A simple biofilm chlorination model 142

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
The analytical solution to Equation (4-14) is given in Equation (4-29), which allows for calculating
the maximum DBP concentration extractable from the biofilm. According to this, the parameters
biofilm area (Ab), reactor volume (V), biofilm thickness (BT), fraction of cells transformed into
chloroform (Fx), cell-chlorine yield coefficient (Y1), initial cells concentration (Xo), fraction of EPS
transformed into chloroform (FE), EPS-chlorine yield coefficient (Y2), and initial EPS concentration
(Eo) directly influence the DBP formation potential. Here M S is the molar mass of the DBP
substance.

𝐴𝑏 𝐹𝑥 𝐹𝐸
𝑀𝑎𝑥 𝑆𝑒𝑥𝑡−𝑏𝑖𝑜 = 𝑀𝑆 (𝐵𝑇) ( 𝑋0 + 𝐸0 ) (4-29)
𝑉 𝑌1 𝑌2

4.3.3.1 S, reaction rates k1 and k2, and initial chlorine concentration
Similar influence of parameters S, k1, k2, and Clo on bulk water concentrations of chloroform can
be observed in Figure 4-13a, Figure 4-14a, Figure 4-15a, and Figure 4-16a, respectively; it can be
observed that such concentrations do not change for every tested parameter value, after 3 hours
of contact time. However, the significance of such behaviour is different for every case. S is related
to speed at which chloroform is being transported, due to the reduction of the substance diffusion
coefficient within the biofilm. Thus, values of S as small as 0.016 allows obtaining the maximum
concentration of chloroform in bulk water in shorter time in comparison to ratios higher than 0.09.
Such difference is observable in less than 40 min (Figure 4-13b). For values of S higher than 0.09,
there is not obvious difference on the time when maximal concentrations of chloroform are reached.

(a)

Chapter 4. A simple biofilm chlorination model 143

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 4-13. Influence of S on chloroform concentration in bulk water (a) Chloroform
concentrations in bulk water at different values of S after t = 3 h (b) Temporal variation of
chloroform concentrations in bulk water for three values of S

With regard to reaction rates k 1 and k2, the influence of these parameters on chloroform
concentrations is actually related to the speed of the reactions; such influence is more significant
for k1. The time required to reach the maximum potential chloroform concentration is reduced from
eight hours corresponding to the minimum value tested (1.1x10 -4 L/mg-s) down to one hour for the
maximum value considered here (2.2x10 -3 L/mg-s) (Figure 4-14b).

(a)

Chapter 4. A simple biofilm chlorination model 144

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 4-14. Influence of reaction rate k1 on chloroform concentration in bulk water (a) Chloroform
concentrations in bulk water at different values of k1 after t = 3 h (b) Temporal variation of
chloroform concentrations in bulk water for several values of k1

The influence of reaction rate k2 is less pronounced than k1, due to the contribution of EPS to DBP
formation potential is smaller in comparison to cells. The value of k 2 small as 3.7x10-7 L/mg-s
generated complete transformation of cells and EPS into chloroform in four hours. Higher values
of k2 increases the speed of the reactions and the time required to reach the maximal potential
concentration of chloroform is the same (1 h) for k 2 ranging between 2.22x-10-6 and 7.4x10-6 L/mg-
s (Figure 4-15b).

(a)

Chapter 4. A simple biofilm chlorination model 145

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 4-15. Influence of reaction rate k2 on chloroform concentration in bulk water (a) Chloroform
concentrations in bulk water at different values of k2 after t = 3 h (b) Temporal variation of
chloroform concentrations in bulk water for several values of k2

In relation to Clo, influence of this parameter on chloroform concentrations allows identifying

whether the reactions are chlorine or biomass limiting. Values of Clo smaller than 0.3 mg/L are not
able to completely oxidize cells and EPS, then this model turns to be chlorine limiting. For such
chlorine concentrations, the model resulted in minimum remaining concentrations of cells and EPS
between 0.38-40.87 mg/L (Figure 4-16a) and 58.69-2397.3 mg/L (Figure 4-16b), respectively, after
3 hours of contact time. For Clo higher than 0.3 mg/L, the model turns biomass limiting, since the
DBP concentrations depend on the amount of biomass available for reaction with chlorine.

This indicates that, under certain values of Xo and Clo, the model can be either chlorine or biomass
limiting. The same condition was found by Wang et al. (2012c) in batch experiment tests with culture
suspensions of different strains of P. aeruginosa. By using initial cell density of 1.04 and 0.05 mg/L 14
and chlorine dose between 0.5 and 5.0 mg/L, the disinfection experiments were biomass limiting
since free chlorine concentrations were negligible in all tested samples. Abokifa et al. (2016a) also
included in their multicomponent model a second order reaction limited by the biomass
concentration to predict the THM formation from chlorination of biofilm, planktonic cells, and
detached biofilm clusters.

14 Cell density was reported as 108 and 106 CFU/mL, which were transformed into volumetric cell density
Chapter 4. A simple biofilm chlorination model 146
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 4-16. Influence of initial chlorine concentration on chloroform concentration in bulk water
(a) Comparison between Clo and minimum cell concentration (b) Comparison between Clo and
minimum EPS concentration

4.3.3.2 Temperature

The effect of temperature over the chloroform formation potentials is associated with the change of
reaction rates k1 and k2, as explained in the previous section. Despite this, a similar analysis is
presented here however based on the variation of the water temperature. According to the results
in Figure 4-17, temperature does not affect the steady state concentrations of chloroform in bulk
water after three hours of reaction time but does influence the speed of the reactions (Figure 4-17a).
As was expected, the model is able to represent that the time required to obtain the maximal
potential concentration of chloroform increases as temperature rises through the range 5-25 °C
(Figure 4-17b). For instance, at 5 °C, two hours are required for entire reaction between chlorine
and cells and EPS; while, at 25 °C, this happens in 45 min.

Chapter 4. A simple biofilm chlorination model 147

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 4-17. Temperature influence on chloroform concentration in bulk water (a) Chloroform
concentrations in bulk water at different temperatures after t = 3 h (b) Temporal variation of
chloroform concentrations in bulk water at the temperature range 5-25 °C

4.3.3.3 Yield coefficients Y1 and Y2 and reactor volume

Parameters such as cell and EPS yield coefficients and reactor volume result in chloroform in
exponential decay of chloroform in bulk water (Figure 4-18). Therefore, up to certain values of these
parameters, concentration of chloroform is minimum and does not vary significantly if these
parameters increase. Particularly, small values of V resulted in high concentration of chloroform,
which represents that this substance is dissolved in a low amount of solvent. Similarly, low yield
coefficients represent that higher amount of mass of cells or EPS have been depleted per mass of
chlorine consumed; therefore, more chloroform will be formed.

Chapter 4. A simple biofilm chlorination model 148

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)

(c)
Figure 4-18. Influence of (a) Y1, (b) Y2, and (c) reactor volume on chloroform concentration in bulk
water after t = 3 h

4.3.3.4 Biofilm thickness, initial cell and EPS concentration, fraction of cells and EPS transformed
into DBP and biofilm area

Parameters such as biofilm thickness (BT), initial cell and EPS concentration (Eo), fraction of cells
and EPS transformed into DBP (FE) and biofilm area (Ab) have a linear effect on bulk water
concentrations of chloroform (see Figure 4-19).

Chapter 4. A simple biofilm chlorination model 149

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)
(b)

(c) (d)

(f)
(e)
Figure 4-19. Parameters with linear influence on chloroform concentration in bulk water (a) Biofilm
thickness (b) Initial cell concentration (c) Initial EPS concentration (d) F X (e) FE (f) Biofilm area

4.3.3.5 Comparison among parameters

This comparison was carried out in order to identify the parameters with more influence on bulk
water concentrations of chloroform. The 14 parameters assessed in this sensitivity analysis are
compared in Figure 4-20. The percentage of change in relation to the base value of both parameters
and the respective chloroform concentrations were calculated and presented in Table 4-12. Reactor
volume (V) and chlorine-cell yield coefficient (Y1) were the parameters with influenced most the
response variable (Figure 4-20a); reduction on 90% of these values represents an increase of
886% and 649% on chloroform concentrations, respectively (Table 4-12 and Table 4-13).

Chapter 4. A simple biofilm chlorination model 150

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
To clarify which parameters were also influencing in second place, another graph was plotted
eliminating the two parameters mentioned previously (Figure 4-20b). Then, variation among -90%
- 100% of initial cell concentrations, biofilm area, and biofilm thickness, parameters with linear
influence on chloroform concentrations, produces the same degree of variation in such
concentrations (Table 4-13). Chlorine-EPS yield coefficient (Y2) is a parameter with some degree
of influence on bulk concentrations of chloroform; a reduction of this of 90% causes an increase of
this DBP in approximately 60% (Table 4-12). A more detailed sensitivity analysis can help to better
rank the parameter influence on model outputs and identify interrelations existing between
parameters. This is addressed in Chapter 5.

(a)

(b)
Figure 4-20. Comparison of influence on chloroform concentration in bulk water among
parameters (a) 14 parameters tested (b) 12 parameters tested excluding V and Y 1

Chapter 4. A simple biofilm chlorination model 151

Disinfection by-product formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 4-12. Percentage of change of parameters and the respective chloroform concentrations in bulk water
Percentage of change
S CHCl3 k1 CHCl3 k2 CHCl3 Clo CHCl3 T CHCl3 Y1 CHCl3 Y2 CHCl3
-96,52% 0,00% -90,00% -11,29% -90,00% -1,04% -100,00% -100,00% -66,7% -0,1% -90,0% 649,1% -90,0% 60,1%
-96,30% 0,00% -80,00% -1,45% -80,00% -0,14% -95,00% -26,29% -60,0% -0,1% -80,0% 356,6% -80,0% 31,0%
-80,43% 0,00% -70,00% -0,18% -70,00% -0,02% -90,00% -17,11% -53,3% 0,0% -70,0% 211,2% -70,0% 18,4%
-58,70% 0,00% -60,00% -0,02% -60,00% 0,00% -85,00% -7,30% -46,7% 0,0% -60,0% 138,0% -60,0% 11,9%
-39,13% 0,00% -50,00% 0,00% -50,00% 0,00% -80,00% -2,46% -40,0% 0,0% -50,0% 91,1% -50,0% 7,9%
-26,09% 0,00% -40,00% 0,00% -40,00% 0,00% -70,00% -0,29% -33,3% 0,0% -40,0% 61,4% -40,0% 5,3%
0,00% 0,00% -30,00% 0,00% -30,00% 0,00% -60,00% -0,04% -26,7% 0,0% -30,0% 39,0% -30,0% 3,4%
28,26% 0,00% -20,00% 0,00% -20,00% 0,00% -48,00% 0,00% -20,0% 0,0% -20,0% 23,0% -20,0% 2,0%
36,96% 0,00% -10,00% 0,00% -10,00% 0,00% -28,00% 0,00% -13,3% 0,0% -10,0% 9,9% -10,0% 0,9%
67,39% 0,00% 0,00% 0,00% 0,00% 0,00% 0,00% 0,00% -6,7% 0,0% 0,0% 0,0% 0,0% 0,0%
86,96% 0,00% 10,00% 0,00% 10,00% 0,00% 19,00% 0,00% 0,0% 0,0% 10,0% -8,6% 10,0% -0,7%
97,83% 0,00% 20,00% 0,00% 20,00% 0,00% 31,00% 0,00% 6,7% 0,0% 20,0% -15,3% 20,0% -1,3%
117,39% 0,00% 30,00% 0,00% 30,00% 0,00% 57,00% 0,00% 13,3% 0,0% 30,0% -21,4% 30,0% -1,8%
40,00% 0,00% 40,00% 0,00% 70,00% 0,00% 20,0% 0,0% 40,0% -26,3% 40,0% -2,3%
50,00% 0,00% 50,00% 0,00% 90,00% 0,00% 26,7% 0,0% 50,0% -30,8% 50,0% -2,6%
60,00% 0,00% 60,00% 0,00% 100,00% 0,00% 33,3% 0,0% 60,0% -34,5% 60,0% -3,0%
70,00% 0,00% 70,00% 0,00% 40,0% 0,0% 70,0% -38,0% 70,0% -3,3%
80,00% 0,00% 80,00% 0,00% 46,7% 0,0% 80,0% -40,9% 80,0% -3,5%
90,00% 0,00% 90,00% 0,00% 53,3% 0,0% 90,0% -43,7% 90,0% -3,8%
100,00% 0,00% 100,00% 0,00% 60,0% 0,0% 100,0% -46,0% 100,0% -4,0%
66,7% 0,0%

Chapter 4. A simple biofilm chlorination model 152

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 4-13. Percentage of change of parameters and the respective chloroform concentrations in bulk water
Percentage of change
V CHCl3 BT CHCl3 Xo CHCl3 Eo CHCl3 Fx CHCl3 FE CHCl3 Ab CHCl3
-90% 886,2% -85,71% -85,81% -90% -82,86% -80,42% -11,77% -80,0% -71,4% -80,0% -10,0% -90% -90,0%
-80% 396,9% -83,67% -83,78% -80% -73,67% -70,53% -10,33% -60,0% -53,5% -60,0% -7,5% -80% -80,0%
-70% 232,1% -81,63% -81,74% -70% -64,45% -45,13% -6,61% -40,0% -35,7% -40,0% -5,0% -70% -70,0%
-60% 149,4% -79,59% -79,71% -60% -55,24% -32,17% -4,71% -20,0% -17,8% -20,0% -2,5% -60% -60,0%
-50% 99,7% -77,55% -77,68% -50% -46,02% 0,00% 0,00% 0,0% 0,0% 0,0% 0,0% -50% -50,0%
-40% 66,5% -75,51% -75,65% -40% -36,83% 13,34% 1,95% 20,0% 17,8% 20,0% 2,5% -40% -40,0%
-30% 42,8% -73,47% -73,61% -30% -27,62% 29,65% 4,34% 40,0% 35,7% 40,0% 5,0% -30% -30,0%
-20% 25,0% -71,43% -71,58% -20% -18,41% 46,70% 6,84% 60,0% 53,5% 60,0% 7,5% -20% -20,0%
-10% 11,1% -69,39% -69,55% -10% -9,21% 80,19% 11,74% 80,0% 71,4% 80,0% 10,0% -10% -10,0%
0% 0,0% -67,35% -67,51% 0% 0,00% 98,22% 14,38% 100,0% 89,2% 100,0% 12,6% 0% 0,0%
10% -9,1% -65,31% -65,48% 10% 9,21% 10% 10,0%
20% -16,6% -44,90% -45,08% 20% 18,40% 20% 20,0%
30% -23,1% -36,73% -36,91% 30% 27,62% 30% 29,9%
40% -28,5% 0,00% 0,00% 40% 36,83% 40% 39,9%
50% -33,3% 6,12% 6,17% 50% 46,04% 50% 49,9%
60% -37,5% 26,53% 26,78% 60% 55,24% 60% 59,8%
70% -41,1% 51,02% 51,60% 70% 64,45% 70% 69,8%
80% -44,4% 71,43% 72,34% 80% 73,66% 80% 79,8%
90% -47,3% 93,88% 95,24% 90% 82,86% 90% 89,7%
100% -50,0% 108,16% 109,85% 100% 92,07% 100% 99,7%

Chapter 4. A simple biofilm chlorination model 153

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
In relation to reactor volume and biofilm area, these parameters can be easily determined in a
DWDN from geometrical characteristics of the water pipes, or surfaces and reactors used in-situ or
laboratories. Then, low uncertainty is expected from these parameters. Yield coefficients were
determined by Chen and Stewart (1996) from several experimental data sets by the method of
nonlinear least-squares; such experiments consisted of measuring chlorine profiles within artificial
biofilms of P. aeruginosa. Therefore, further adjustments of this parameter can be done by
laboratory tests with heterogeneous biofilms growth under real-scale DWDN conditions.

In relation to initial cell concentration, it is important to remark that the specific chemical, physical
and hydraulic conditions of the DWDNs shape the bacterial communities of biofilms as discussed
in Chapter 3. Consequently, it is necessary to collect cell density data from the biofilms growing in
the DWDN of interest. Despite initial EPS concentration having a lower impact on variation of
chloroform bulk concentrations, it is also important to measure the EPS content in biofilms of real-
scale DWDN to improve the reliability of the model predictions. Biofilm thickness is also an
important parameter influencing the model results and is clearly interacting with parameter Eo,
since the latter is calculated using BT values (Section 4.2.3.3). Additionally, further biofilm research
could led to the formulation of a relationship between EPS concentration and biofilm thickness in
biofilms naturally found in real-scale DWDNs.

Biofilm thickness can be determined by microscopy and imaging techniques (Shen et al., 2016).
As presented in Table 4-2, biofilm thickness for drinking water biofilms have been measured in
those grown in laboratory reactors. In a personal communication with Dr Isabel Douterelo, she
mentioned the difficulty faced by measuring this parameter in biofilms grown on coupons installed
in real-scale DWDNs, since the biomass content was very high and dyes were not able to fully
penetrate the biofilms, then they were not completely observable thorough the microscope 15.

Other factors such as chlorine concentration and temperature can be defined by historical records
supplied by the water utilities or direct measurement in the field (Nescerecka et al., 2014; Dippong
et al., 2016). Parameters such as fraction of cells and EPS transformed into DBPs can be adapted
from experimental studies reporting results of the DBP of interest (Fang et al., 2010b; Wang et al.,
2013a) or adjusted to fit the model, if there were experimental data to compare to or validate. The
diffusion coefficient of DBP within the biofilm may also be adjusted if the model is coupled to cells
growth and EPS production simultaneously. This factor can also be calculated by measuring DBP

15 Technical meeting at the University of Sheffield, 24th March 2017. E-mail: [email protected]
Chapter 4. A simple biofilm chlorination model 154
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
concentration profiles within biofilms; however, it is not clear if the micro-sensors for such
substances are already available for laboratory applications.

4.3.4 Factors affecting the DBP formation

As explained in Chapter 2, the factors which affect the DBP formation are the concentration and
type of NOM, type and concentration of disinfectant, pH, temperature, contact time and bromide
concentration. The current model was developed for water systems using chlorine as disinfectant.
The type of organic matter is included through the parameters corresponding to DBP precursors
such as bacteria and EPS (k1, k2, Y1, Y2, FX, FE). These substances are mainly composed by
biomolecules such as genetic material, proteins, lipids, and polymers (Wang et al., 2012b; Fish et
al., 2015). Particularly, it has been identified that EPS contain aromatic molecules, are hydrophilic
and contain fulvic like substances (Wang et al., 2012c; Lemus Pérez and Rodríguez Susa, 2017).

As also mentioned before, temperature is indirectly included through the reaction rates k1 and k2.
Equations (4-6) and (4-7) can be used to adjust these rates to the local temperature of the DWDN
of interest. Parameters such as S, k1, and k2 affect the time when maximal potential concentration
of DBP is reached. Higher rates and diffusion coefficient of DBP substance accelerate the
reactions; therefore, less time is required to oxidize the entire biomass of the biofilm. This model
did not include pH variable directly; however experimental results and field data matched with
simulated results, as discussed in Sections 4.3.5 and 4.3.6.1. The influence of bromide
concentration is not included in the current model; it is accepted that bromide presence in
disinfected water favours the formation of brominated DBPs (Chowdhury et al., 2009). More
information is required to determine the presence and amount of bromide within biofilms.

4.3.5 Comparison between model results and experimental data

According to Wang et al. (2013a), tests of DBP formation potentials resulted in ranges of 6.6–9.4
g/L and 2.6–3.2 g/L for chloroform and DCAN, respectively. The current model reproduced such
results for BT in ranges of 12–16 m and 12–14 m, for these substances, respectively (Figure
4-21). These outcomes are consistent with the conditions of biofilm growth reported by Wang et al.
(2013a); who incubated biofilm, in an enriched medium, for 60 hours, and the steady state was
reached after 24 hours. Hence, thin biofilms may be expected under such condition.

Chapter 4. A simple biofilm chlorination model 155

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 4-21. (a) Chloroform and (b) DCAN concentrations according to results of Wang et al.
(2013a)

4.3.6 Simulation of scenarios: building plumbing system supplied with warm water

DBPs are monitored in the distribution network by water utilities and regulation agencies, but people
are exposed to them inside of the facilities. Exposure occurs by inhalation, dermal contact, and
ingestion; through domestic activities such as bathing, showering, cooking, and drinking
(Chowdhury, 2016). The same chemical, biological and hydraulic processes occurring in the
distribution network also take place in plumbing systems. In addition, such systems are
characterised by high stagnation times and temperatures, therefore water age increases inside of
the buildings, disinfectant decays faster, and more favourable conditions exist for bulk bacteria
regrowth, biofilm growth and DBP formation (Ji et al., 2015; Chowdhury, 2016). Overnight sleeping
time, working hours, and weekends in commercial buildings are examples of stagnation conditions
in plumbing systems. Stagnation on dead-end sections of DWDNs can also take place during
overnight sleeping time. Higher water temperatures in plumbing systems can occur due to the
heating system used to raise the temperature of the building or only for domestic activities

Chapter 4. A simple biofilm chlorination model 156

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
(showering, bathing, and washing). This causes the temperature of water flowing in the plumbing
pipes to increase.

To analyse the scenario of chloroform and DCAN formation in Colombian plumbing systems, the
parameters included in Table 4-14 were used when running the current model, which yielded 12
scenarios in total (Table 4-16). The scenario 1 is the least critical one, since it includes the lowest
values of Clo, Xo, and BT, according to the ranges defined in Section 4.2.3. On the other hand,
scenario 12 corresponds to the most critical one.

Table 4-14. Parameters used to simulate DBP formation in plumbing systems

Parameters Value Source
Assuming typical domestic activities:
Stagnation times 11 h (morning – afternoon)
overnight sleeping time and working hours
Typical diameter found in houses in
Pipe diameter () 0.5 inches
Colombia
Assuming house depth of 25 m with two
Length of the pipe (L) 50 m
floors
Water temperature in Cali – Colombia
Temperature 25 °C (Table 3-4) and in Dhahran, Saudi Arabia
(Chowdhury, 2016)
Initial chlorine Minimum and maximum values found in
0.12 and 1.66 mg/L
concentration Cali – Colombia (Table 3-4)
Extreme values within the range Q1-Q3
Biofilm thickness 7 and 102 m
(Table 4-3)
Calculated from BT with equation included
Initial EPS concentration 2.397,3 and 1.919,5 mg/L
in Figure 4-5
Initial cell concentration 0.89; 4.03; and 13.98 mg/L Values within the range Q1-Q3 (Table 4-5)

4.3.6.1 Chloroform formation compared with field assessment (Chowdhury, 2016)

Chowdhury (2016) investigated the effects of plumbing systems on human exposure to DBPs in
drinking water by measuring several species of THMs and HAAs, during one year, in two houses
located in the city of Dhahran, Saudi Arabia. The sampling campaign included three points (the
water network at the entry of the houses, the house tap, and the hot water tank). The results of this
study are consolidated in Table 4-15; water pH ranged between 6.89-8.33. According to the author,
there were not any differences between sampling points, thus DBP species data for three sampling
points were presented as averages, resulting in chloroform concentrations in the water network
equal to 5.51 g/L; plumbing systems equal to 9.37 and 12.42 g/L for cold and hot water,
respectively. This represents that 3.86 and 6.91 g/L of chloroform was formed in cold and hot
water, respectively, under stagnation conditions (Chowdhury, 2016).

Chapter 4. A simple biofilm chlorination model 157

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 4-15. TTHMs and free chlorine concentrations measured in plumbing systems (Chowdhury,
2016)
TTHMs (g/L) Free chlorine (mg/L)
Sampling point Stagnation condition
PM-AM AM-Aftern. PM-AM AM-Aftern.
Water network 6.94 6.50 0.33 0.26
House tap 11.1 10.40 0.14 0.14
Hot water tank 14.60 13.20 0.11
TTHMs formed under Cold water 4.16 3.90
0.11 – 0.33 mg/L
stagnation conditions (g/L) Hot water 7.66 6.70
Chlorine demand under Cold water 0.19 0.12
-
stagnation conditions (mg/L) Hot water 0.22 0.14

Figure 4-22 shows the chloroform concentrations under initial concentrations of chlorine measured
in plumbing systems by Chowdhury (2016) (0.33 mg/L; Table 4-15). Due to the average
concentration of chloroform found in plumbing premises was low (3.86 g/L), lowest biofilm
thickness and cell density were tested in the model. Model outputs resulted in similar
concentrations to Chowdhury (2016) for pipe diameter between ¾ - 1 inches. The diameters of the
plumbing networks were not reported by the author. However, these results confirm that the model
is able to reproduce DBP concentrations associated with field conditions.

Figure 4-22. Chloroform concentrations in bulk water for different pipe diameter

Chapter 4. A simple biofilm chlorination model 158

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
4.3.6.2 Chloroform formation

Figure 4-23 presents the chloroform concentrations in bulk water for BT of 7 m and 102 m.
Simulated chloroform concentrations ranged between 6.98-31.62 g/L and 89.29-466.20 g/L for
7 m and 102 m of BT, respectively. For BT=7 m, 5.25 hours were required to achieve the
maximum DBP concentration in bulk water with the lowest chlorine concentration of 0.12 mg/L and
Xo of 0.89 and 4.03 mg/L. In the case of the highest Xo (13.98 mg/L), six hours were needed to
reach the maximal potential chloroform concentration. For chlorine concentration of 1.66 mg/L,
complete transformation of cells and EPS into chloroform occurred in less than 30 minutes (Figure
4-23a).

(a)

(b)
Figure 4-23. Chloroform concentrations in bulk water for several scenarios of initial chlorine and
cell concentration (a) BT = 7 m (b) BT = 102 m

Chapter 4. A simple biofilm chlorination model 159

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
For BT=102 m and Clo of 0.12 mg/L, simulations indicated that the time required for complete
reaction chlorine-biomass reduces as Xo increases (7.0, 5.5, and 4.25 hours for Xo of 0.89, 4.03,
and 13.98 mg/L, respectively). This is related to faster production of DBP when more biomass is
available, which is given by the expression F X × k1 × C × X of Equation (4-4). This expression
indicates that more DBP is formed per unit of time for high cell concentration, then the maximal
potential concentration of DBP is reached in shorter time in comparison to low concentrations of
cells.

The different pattern between two BT values in relation to reaction time is explained by whether the
reactions are chlorine or biomass limiting. For 7 m, reactions can be biomass limiting, residual
concentration is 0.11 mg/L (Figure 4-25), and reaction time is directly proportional to cell
concentration. On the contrary, reactions can be chlorine limiting for 102 m, residual
concentrations are almost negligible, and reaction time is inversely proportional to cell
concentration. For Clo of 1.66 mg/L, complete transformation of cells and EPS into chloroform also
occurred in less than 30 minutes (Figure 4-23b).

In comparison to reference values, for BT=7 m, the highest Xo led to a maximal potential
concentration of chloroform higher than the concentration found in the field work developed in the
city of Cali (average 31.62 g/L vs 22.55 g/L16 - Section 3.3.5). For BT=102 m, it is worth noting
that most of the scenarios considered here resulted in potential concentrations of chloroform higher
than the limit defined in the UK regulations (100 g/L) (UK Parliament, 2000). In relation to
Colombian standards (200 g/L) (Ministerio de la Protección Social and Ministerio de Ambiente
Vivienda y Desarrollo Territorial, 2007), only the scenario of Clo=1.66 mg/L and Xo = 13.98 mg/L
resulted in notably greater concentration (466.2 g/L). Such value can be perceived as unrealistic
or improbable, but it is difficult to confirm due to the lack of field data to use as inputs of the model.
Therefore, it is important to highlight the importance of studying biofilm properties in DWDNs of
developing countries for collecting reliable field data and define real scenarios of these networks.
In general, biofilms can be understood as potential chronic risk sources in DWDNs, which may
even lead to non-compliance of local regulations for chloroform and TTHMs, according to model
results.

16 71% of TTHMs represent chloroform – section 4.2.10

Chapter 4. A simple biofilm chlorination model 160
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
4.3.6.3 Dichloroacetonitrile formation

As expected, the model suggests that thicker biofilms promote the formation of higher potential
DCAN concentrations. For BT=7 m and 102 m, the simulated DCAN concentrations ranged
between 3.43-13.16 g/L and 44.3-195.94 g/L, respectively (Figure 4-24a and b). In relation to
reaction time needed to get to the maximal potential DCAN concentration, the same pattern of
chloroform was found for this substance and the two values of BT tested and Clo=0.12 mg/L. For
BT=7 m, reaction time increases as Xo increases, while this time is shorter as Xo increases for
BT=102 m. As mentioned previously, this is related to faster chlorine depletion when more
biomass is available, since more DBPs are being formed. For chlorine concentration of 1.66 mg/L,
complete transformation of cells and EPS into DCAN occurs in less than 30 minutes (see Figure
4-24a and b). For BT=102 m, following DCAN guidelines (WHO, 2017) becomes more
challenging, since all the tested scenarios produced concentrations higher than 20 g/L (Figure
4-24b) and a maximum of 195.94 g/L is obtained for the most critical conditions (Clo=1.66 mg/L,
Xo=13.98 mg/L).

It is important to highlight that DCAN is an unstable DBP species, which degrades under certain
conditions of free chorine and pH. DCAN degrades in the absence of chlorine above pH 7and below
pH 6.5. In the presence of free chlorine, DCAN degradation can be much faster in the condition of
a pH of about 6–8.5 under low to moderate chlorine residuals (Reckhow et al., 2001). Therefore,
despite of the formation of this DBP under stagnation conditions, DCAN concentrations may reduce
if conditions of pH and free chlorine are the appropriate for its degradation. Finally, it is important
to remark that HANs can be more toxic than THMs and its control must follow the WHO
recommendation of 20 g/L (WHO, 2017), while further research leads to a precautionary threshold
based on health effects.

(a)
Chapter 4. A simple biofilm chlorination model 161
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 4-24. DCAN concentrations in bulk water for several scenarios of initial chlorine and cell
concentration (a) BT = 7 m (b) BT = 102 m

4.3.6.4 Chlorine demand

Figure 4-25 shows chlorine decay in bulk water during the formation of DBPs and Table 4-16
includes the chlorine demand exerted by biomass for each scenario. While Chowdhury (2016)
found chlorine demands of 0.19 and 0.12 mg/L in morning and afternoon measurements,
respectively (Table 4-15), the chlorine demands exerted by cells and EPS in the current model
resulted in much lower values (0.009-0.42 mg/L) (Table 4-16). Since chlorine decay was calculated
based on the reaction rates and yield coefficients found by Chen and Stewart (1996), chlorine
demand does not change with the amount of chloroform and DCAN produced. In addition, the
recent experimental study by Lemus Pérez and Rodríguez Susa (2017) determined that all the
chlorine consumed by the biofilm matrix was not transformed into DBPs. The current model
approach considers the consumption of chlorine only by cells and EPS and separately simulates
the formation potentials of two DBP species. In real systems, chlorine reacts with organic and
inorganic substances contained in the biofilm matrix such as iron, adsorbed organic matter, and
soluble microbial products; and several DBP species are simultaneously formed according to the
physico-chemical conditions, chlorine dose, and amount and type of precursors (Wang et al.,
2012b; Xue et al., 2013; Lemus Pérez and Rodríguez Susa, 2017).

Chapter 4. A simple biofilm chlorination model 162

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a) (b)
Figure 4-25. Chlorine concentrations in bulk water, under several scenarios of initial chlorine and
cell concentration (a) BT = 7 m (b) BT = 102 m

Table 4-16. Chlorine demand on plumbing system scenarios

Scenario Chlorine demand (mg/L)
1 BT = 7 μm | Clo = 0.12 mg/L | Xo = 0.89 mg/L 0.009
2 BT = 7 μm | Clo = 0.12 mg/L | Xo = 4.03 mg/L 0.013
3 BT = 7 μm | Clo = 0.12 mg/L | Xo = 13.98 mg/L 0.025
4 BT = 7 μm | Clo = 1.66 mg/L | Xo = 0.89 mg/L 0.012
5 BT = 7 μm | Clo = 1.66 mg/L | Xo = 4.03 mg/L 0.016
6 BT = 7 μm | Clo = 1.66 mg/L | Xo = 13.98 mg/L 0.028
7 BT = 102 μm | Clo = 0.12 mg/L | Xo = 0.89 mg/L 0.120
8 BT = 102 μm | Clo = 0.12 mg/L | Xo = 4.03 mg/L 0.120
9 BT = 102 μm | Clo = 0.12 mg/L | Xo = 13.98 mg/L 0.120
10 BT = 102 μm | Clo = 1.66 mg/L | Xo = 0.89 mg/L 0.204
11 BT = 102 μm | Clo = 1.66 mg/L | Xo = 4.03 mg/L 0.256
12 BT = 102 μm | Clo = 1.66 mg/L | Xo = 13.98 mg/L 0.423

4.3.6.5 Influence of S/V ratio on DBP formation

One of the main characteristics of plumbing systems is the high S/V ratio, as explained in Section
2.2.4. The S/V ratio is included in the current biofilm chlorination model in the boundary condition
at the biofilm surface, which represents the variation of dissolved substances in the bulk water
(Equations (4-8) and (4-14)). Results for 12 scenarios corresponding to pipe diameter of ½ inches,
which is commonly used in small building facilities in Colombia, were presented previously. In order
to illustrate the influence of this parameter of DBP concentrations, simulations for the least and
most critical scenarios were run (1 and 12, respectively) (Table 4-16).

Chapter 4. A simple biofilm chlorination model 163

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Considering that S/V ratio is inversely proportional to pipe diameter (4/), increasing the pipe
diameter resulted in reduction of DBP concentrations in bulk water. Simulated chloroform ranged
between 0.30-6.91 g/L (scenario 1) and between 6.55-466.2 g/L (scenario 12), for between ½
- 9 inches (Figure 4-26a). Similarly, modelled DCAN ranged between 0.16-3.39 g/L (scenario 1)
and between 3.27-195.94 g/L (scenario 12), for the same range (Figure 4-26b). The analysis
of chlorine wall decay has determined that, as pipe diameter increases, the wall reactions becomes
less important and chlorine decay mainly relies on bulk water reactions (Lu et al., 1999; Hallam et
al., 2002; Buamah et al., 2014; Lee et al., 2014). For DBP analysis carried out here, pipe diameter
as big as 9 inches may lead to potential concentrations of 0.30-6.55 g/L and 0.16-3.27 g/L of
chloroform and DCAN, respectively, which represent small fractions of the UK and Colombian
regulation and WHO guidelines.

(a)

(b)
Figure 4-26. Influence of S/V ratio on DBP concentrations in bulk water (a) Chloroform (b) DCAN
Chapter 4. A simple biofilm chlorination model 164
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
In order to improve the comparison of DBP concentrations to regulation compliance and guidelines,
several simulations for both chloroform and DCAN were run for several values of BT and S/V ratio.
This was done in order to find the values of this parameter which resulted in lower concentrations
than the suggested or regulated ones. Figure 4-27 shows the results of DBP concentrations for
Clo=1.66 mg/L and Xo of 1.28x10 -4 mg/cm2 and 4.44x10-3 mg/cm2 for DCAN and chloroform,
respectively. Due to Xo is expressed in terms of surface density, the volumetric concentrations are
specified in Figure 4-28, according to the pipe diameter.

(a)

(b)
Figure 4-27. DBP concentrations in bulk water and regulation limits / guidelines (a) Chloroform, Clo
= 1.66 mg/L, Xo = 4.44x10-3 mg/cm2 (b) DCAN, Clo = 1.66 mg/L, Xo = 1.28x10-4 mg/cm2

Chapter 4. A simple biofilm chlorination model 165

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 4-28. Cell concentration according to pipe diameter

Particularly, for the given Clo and Xo, chloroform concentrations fall below the Colombian and UK
regulated values (200 and 100 g/L, respectively), for BT values smaller than 49 m and 16 m,
respectively, and pipe diameter of ½ inches. However, by increasing the pipe diameter to only ¾
inches, chloroform concentrations reduce around 51%. Therefore, thicker biofilms result in
chloroform concentrations lower than the regulated thresholds. For instance, 84 m potentially
yields 188.5 g/L and 27 m potentially yields 59.3 g/L in  of ¾ inches. The same occurs with
DCAN concentrations; BT smaller than 40 m yields potential concentrations lower than the value
suggested by the WHO (20 g/L) (WHO, 2017) for every  considered here. Increasing diameter
from ½ inches to ¾ inches, DCAN concentrations reduce around 36%. Such values of pipe
diameter may offer a guideline for future recommendations for design and installation of plumbing
pipes in small buildings, where low water consumption is expected. In order to properly propose a
minimum pipe diameter for plumbing facilities, further hydraulic analysis must be done to determine
the influence of such change on pressure, velocities, and residence time.

4.3.6.6 Discussion of DBP formation potentials simulated under stagnation conditions

The discussion presented in Sections 4.3.6.2 and 4.3.6.3 was based on regulation limits or
guideline values; however, such thresholds are more related to precautionary limits rather than
health-based recommendations. The toxicity of some DBPs, from around 600 DBPs species
discovered to date, has been tested on bioassays with mice and cells using concentrations higher
that those found in drinking water (Hrudey, 2009). Overall, the epidemiological evidence is
Chapter 4. A simple biofilm chlorination model 166
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
insufficient to declare chlorinated DBPs as carcinogenic in humans or a cause of birth defects
(Hrudey, 2009; Hrudey et al., 2015a). On the other hand, the presence of pathogenic
microorganisms in tap water will certainly cause negative public health impacts (Hrudey, 2009). In
this line, the risk management alternative is to reduce the precursors of DBPs in the water source,
maintaining efficient treatment processes and proper O&M of distribution networks. For plumbing
systems, consumers should be advised on flushing the tap water, for the first use of water after a
period of stagnation.

In relation to reduce DBPs precursors, it is in this point that biofilm formation and disinfection
becomes relevant. Biofilms in drinking water have only been approached in research as potential
reservoirs of pathogens, which can be released to bulk water and reach the consumers (Wingender
and Flemming, 2011). Biofilm monitoring is not included neither in routine O&M of distribution
networks by water utilities nor by regulatory agencies. The current model, based on experimental
data reported by other researchers, does acknowledge the role of biofilms as DBP precursors and
present the potential concentrations of chloroform and DCAN under certain scenarios. The model
presented here allows identifying the most significant parameters and better understanding of the
interaction of those involved in the formation of two DBP species formed from the chlorination of
drinking water biofilms. In addition, the model gives a better understanding of the rates at which
chloroform and DCAN are formed; which is not always evident with solely laboratory data.

4.4 MODEL APPLICATIONS AND LIMITATIONS

The current model can only predict the chlorine decay according to the demand exerted by EPS
and cells. The reactions of this disinfectant with other substances contained in the biofilm matrix
and attached to the pipe walls such as sediments and loose deposits and bulk water reactions are
not included; therefore, chlorine is exclusively available for biofilm reactions. This means that
reactions occur faster compared to inclusion of total demand of chlorine in water pipes.

The proposed model does not include neither cell growth nor EPS production. For long term
assessment, i.e., days and weeks, biofilm growth should be incorporated in order to completely
analyse the dynamics between disinfectant and biofilm matrix; and influence of hydraulics in mass
transfer and biofilm detachment. For instance, Abokifa et al. (2016a) found that maximal bacteria
growth (biofilm and planktonic cells together) occur after six hours, while cell death and inactivation
was reduced. Then, mortality rate increased for the next 12 hours. As a consequence, THM
formation reached a plateau for the next 30 hours and constant THM concentration was produced

Chapter 4. A simple biofilm chlorination model 167

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
during this period of time (Abokifa et al., 2016a). Further re-chlorination led to extra reaction with
remaining biomass and increasing THM formation; then another plateau was observed until 60
hours of simulation. This indicates that THM formation reached a peak and remained constant until
re-chlorination was applied, then another peak was found and remained constant until 60 hours
(Abokifa et al., 2016a).

It is known that biofilm growth is a slow process (e.g., hours or days; see modelling and
experimental data in Sections 2.4.3 and 4.2.3.1) compared to oxidative reactions with chlorine.
Despite this the current model did not include biofilm growth and chlorine demand exerted by other
substances present in water pipes, the proposed model developed here is applicable to determine
the DBP formation potentials under different of scenarios of biofilm properties and drinking water
quality. Model applications also include improving the understanding of field and experimental tests,
which are carried out under controlled conditions, as demonstrated by the comparison of results to
other studies. Additionally, this model can be applied to plumbing systems and dead-end sections
of DWDNs under stagnation conditions in order to predict the potential DBP formation under
specific conditions of the assessed system.

On the other hand, this model does differentiate between cells and EPS contribution to DBP
formation, which allows the incorporation of kinetic variables resulting from experimental studies
carried out for both particulate substances. It is important to highlight that the interest on studying
the role of EPS as a protective barrier for bacteria within the biofilm and as DBP precursor is
increasing (Fang et al., 2010b; Huang et al., 2012; Wang et al., 2012c; Xue et al., 2012; Wang et
al., 2013b; Lemus Pérez and Rodríguez Susa, 2017). Other advantages of the proposed model is
the low computational cost due to it is simulating in one dimension; and is able to predict individual
DBP species by incorporating their respective diffusion coefficients. Further development of this
model may support the improvement of exposure assessment in humans and comparison to
maximum concentrations established by toxicological studies. Particularly, this model simulates the
formation of DCAN, which can be more toxic substance than other DBPs, and there has not been
regulated yet for any national regulatory agency. Long term prediction of DCAN should include a
sink term since it degrades over a period of hours or days, depending on pH and chlorine
concentration (Reckhow et al., 2001).

The current model relies on microbiological variables such as parameters of biofilm thickness, and
initial cell and EPS concentration. As discussed above, biofilms are not frequently monitored in
DWDNs by neither water utilities nor regulatory agencies. Therefore, data of these parameters are
not often available in real-scale water networks; and to interpret idealized laboratory data can be

Chapter 4. A simple biofilm chlorination model 168

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
challenging in the context of real DWDNs. As a first step, initial EPS concentration was calculated
in this study with a linear equation originated from an experimental study with municipal
wastewater. Field-work information collected in Cali’s DWDN related to water quality was used as
inputs for the several scenarios simulated with the current model. However, to the author’s
knowledge, biofilm thickness, and initial cell and EPS concentration data for this network have not
yet been published. This indicates that more effort is required in promoting drinking-water-biofilm
research in developing countries by improving international collaborations, laboratory facilities,
molecular analysis techniques, protocols for sampling in real-scale DWDNs, among others.

Finally, as human exposure to water substances occur within the facilities, drinking water quality in
plumbing systems becomes relevant in public health topics. The epidemiological study by Wright
et al. (2017) found that DBP exposure classification may be improved if models were available for
the studied network. On the other hand, DBP monitoring take place in the distribution network,
which is characterized for flow rates leading to slow velocities in dead-end zones or normal
velocities in the main pipes. The current model is applicable under stagnation conditions, which are
present in dead-end sections of a network or in plumbing facilities, where water consumption
actually occurs. To analyse the mass transfer rate of dissolved substance in the biofilm surface,
hydrodynamics are included in a 2D model for water pipes. The definition, development, results
and discussion of this model can be found in the following chapter.

4.5 CONCLUSIONS

Biofilms attached to the pipe walls have been acknowledged as a cause for chlorine decay in water
pipes but they have not yet been included in DBP models for DWDNs. To the author’s knowledge,
this is the first model including biofilm EPS oxidation as precursor of DBPs together with cell
disinfection. The current study led to the following conclusions:

 Under stagnation conditions and small pipes, biofilm-chlorine reactions at the pipe walls are
important for DBP contribution to the bulk water.
 Increasing the complexity of a model is not necessarily directly related to a better approximation
of the process. The current 1D model of chlorine disinfection of flat biofilms is able to predict the
chloroform and DCAN formation potentials in bulk drinking water, under stagnation conditions.
 Chloroform and DCAN formation from biofilm chlorination can be expressed in terms of cell and
EPS concentrations, chlorine yields, reaction rates, and fraction of cells and EPS transformed
into DBP.

Chapter 4. A simple biofilm chlorination model 169

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Due to the slow process of biofilm growth and EPS production, fixed initial cell and EPS
concentration are an appropriate approach to simulate DBP formation potentials occurring in
few hours.
 Whether the reactions are chlorine or biomass limiting depends on the parameters biofilm
thickness and initial chlorine and cell concentrations together.
 In the current model, DBP diffusion coefficient within the biofilm is relevant only for analysis on
short time as 40 minutes.
 Temperature effects are indirectly included in the model by adjusting the reaction rates of
chlorine with cells and EPS. Higher temperature leads to acceleration of reactions.
 Parameters biofilm thickness, initial cell and EPS concentration, fraction of cells and EPS
transformed into DBP and biofilm area have a linear effect on bulk water concentrations of
DBPs. Reactor volume and chlorine yield coefficients have a decreasing exponential effect of
DBP concentrations in bulk water.
 Reactor volume and chlorine-cell yield coefficient are the most influencing parameters on bulk
water concentrations of DBPs, followed by biofilm area and initial cell concentration. Biofilm
thickness also has an important effect on the model results by the direct relationship with cell
and EPS concentrations and the definition of the domain dimension.
 To promote biofilm research in real-scale drinking water systems will increase the availability of
data useful for DBP models, considering biofilms as precursors.
 Plumbing systems can favour the DBP formation from disinfection of biofilms due to the high
stagnation time and high S/V ratios.
 Pipe diameters as small as ½ inches in plumbing facilities may lead to DBP concentrations
higher than the regulated or guidance values. Increasing the diameter to ¾ inches can
significantly reduce the concentrations even to values under such thresholds.

Chapter 4. A simple biofilm chlorination model 170

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 5 MODELLING DBP FORMATION FROM BIOFILM CHLORINATION
UNDER HYDRODYNAMIC CONDITIONS

5.1 INTRODUCTION

This chapter aims to explain the process of developing a 2D model, coupled to a pipe flow, in order
to assess the influence of hydraulics over mass transport and formation potentials of chloroform
and DCAN from the chlorination of biofilms in drinking water. This study aimed to improve the DBP
prediction in water pipes under hydrodynamics conditions. While the simple model developed in
Chapter 4 can be applied in water systems under stagnation conditions, the coupled-flow model
can be used, in general, in DWDNs with bulk flow. By the current model, chlorine and DBP mass
transfer were assessed under different scenarios according to the flow regime, Reynolds number,
pipe diameter; and initial chlorine, cell, and EPS concentrations. A sensitivity analysis based on
Morris method was also carried out to screen the parameters influence on model outputs.

5.2 MODEL DEVELOPMENT

5.2.1 Model definition

The model described here simulates the formation potentials of chloroform and DCAN from
chlorination of flat biofilms in drinking water pipes, in two dimensions, under bulk flow. This model
is based on the 1D model previously developed; initial constant concentration of cells and EPS and
their decay due to the reaction with chlorine within the biofilm were considered. The schematic
representation of the model is presented in Figure 5-1. The starting point for developing the current
model was the simple model presented in Chapter 4 and the physico-chemical data collected in
the field work discussed in Chapter 3 (see Figure 5-2).

Figure 5-1. Schematic representation of the 2D model

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 171
Disinfection by-products formation from disinfection of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 5-2. Process for building a flow-coupled model to predict DBP formation potentials from
chlorination of biofilms

The main assumptions made to develop the current model are:

 Flow field solution: DWDNs are built to transport water from the WTPs to the consumers.
Therefore, pipelines are characterized by bulk flow and the flow rates depend on the water
demand. Transitional and turbulent flow were considered here because the first one can be
present in dead-end sections of DWNs (Abokifa et al., 2016b) and the latter one is predominant
in water pipes under normal operation (Cogan, 2010). Laminar flow was also included for
comparison purposes, since the biofilm modelling was developed under this flow regime to
reduce computational cost.
 Flat biofilm: As explained in Chapter 4, a biofilm can be compact or present voids and channels
within in, according to the nutrients availability. This means that biofilm growth under limited
transport of nutrients (slow flow) leads to thicker biofilms due to the presence of voids channels
rather than higher cell density (Picioreanu et al., 1998b; Picioreanu et al., 2000b). Considering
that the current model is mainly focused on transitional and turbulent flow (Re: 2,300-10,000),
flat biofilm surface is considered here as a reasonable approximation to the biofilm morphology
in drinking water pipes.
 Sherwood number and mass transfer rate of dissolved substances (chlorine and DBPs) are
used to characterize the mass transport at the biofilm surface: similar approach has been
applied for biofilm modelling, which includes biofilm growth and transport of substrates from bulk

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 172
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 water to the biofilm matrix, through the biofilm surface (IWA Task Group on Biofilm Modeling et
al., 2006).

Other assumptions were also made for the 2D model, which were also presented in Chapter 4:

 Biofilm is continuous medium

 Constant biofilm thickness
 Cells and EPS are homogeneously mixed in biofilm matrix
 Fixed initial concentration of cells (Xo) and EPS (Eo)
 Chloroform (S) and DCAN (S) are modelled separately according to fraction of cells and EPS
converted into the respective DBP

5.2.2 Model equations and initial and boundary conditions – Reactions

Figure 5-3 sketches the 2D-bulk flow model and the corresponding equations are included in Table
5-1. Under isothermal and isotropic conditions (i.e., constant temperature and constant diffusion
coefficient), chlorine diffuses through the biofilm matrix, reacting with cells and EPS (Equation
(5-1)). Since this model includes fixed initial concentrations of cells and EPS, the variation of both
particulate substances is given by their decay according to reactions with chlorine (Equations (5-2)
and (5-3)). Such reaction leads to the formation of DBPs (Equation (5-4)). Parameters Fx and FE
in Equation (5-4) allows simulating the formation of individual species of DBPs, according to
experimental data reported by other researchers, as was explained in Section 4.2.3.6. In bulk water,
transport of dissolved substances is including both convection and diffusion (Equations (5-5) and
(5-6)). The source term in both equations is defined by the consumption of chlorine by biofilm
reactions and the formation of DBPs (expression in the right side of the equations).

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 173
Disinfection by-products formation from disinfection of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 5-3. 2D mass balance equations for soluble and particulate components in biofilm, biofilm
surface, and bulk water

Regarding the initial conditions, at t=0, chlorine is completely mixed in bulk water with concentration
Co, which was also specified at the inlet. Within the biofilm, chlorine concentration is zero. DBP
concentration is zero everywhere at t=0. Initial concentration of cells and EPS (particulate
substances) are Xo and Eo at t=0. When reactions start, these substances decay until complete
depletion, according to contact time with chlorine.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 174
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 5-1. Mass balance equations and initial conditions in 2D model – Cylindrical pipes
Initial
Phase Equation Description of variables Subscripts
conditions
C: Disinfectant concentration
X: Cell concentration
𝑑𝐶 E: EPS concentration
= 𝐷𝐶−𝐵 ∇ ∙ ∇𝐶 − 𝑘1 𝐶𝑋 − 𝑘2 𝐶𝐸 C: Chlorine
𝑑𝑡 k1: Disinfectant decay rate – cells
S: DBP
(5-1) disinfection
t=0 X: Cells
1 𝜕 𝜕 CBo = 0 k2: Disinfectant decay rate – EPS oxidation
Divergence of a vector: [∇]𝑟 = 𝑟 𝜕𝑟 (𝑟) [∇]𝑧 = 𝜕𝑧 E: EPS
DC-B: Molecular diffusion coefficient of
𝜕 𝜕 B: Biofilm
Gradient of a scalar: [∇]𝑟 = [∇]𝑧 = chlorine in biofilm
𝜕𝑟 𝜕𝑧 L: Bulk liquid
Biofilm

Co: Initial concentration of chlorine

Bs: Biofilm surface
t: Time
r: Pipe radius, r-
X: Cell concentration
coordinate
C: Disinfectant concentration
z: z-coordinate
𝑑𝑋 k1: Disinfectant decay rate – cells
= −𝑌1 𝑘1 𝑋𝐶 along the pipe
𝑑𝑡 t=0 disinfection
XBo = Xo length
Y1: Cell-chlorine yield coefficient
(5-2)
Xo: Initial concentration of cells
t: Time

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 175
Disinfection by-products formation from disinfection of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Initial
Phase Equation Description of variables Subscripts
conditions
E: EPS concentration
C: Disinfectant concentration
𝑑𝐸
= −𝑌2 𝑘2 𝐸𝐶 k2: Disinfectant decay rate – EPS oxidation
𝑑𝑡 t=0
EBo = Eo Y2: EPS-chlorine yield coefficient
(5-3) Eo: Initial concentration of EPS
t: Time
S: DBP concentration
X: Cell concentration
E: EPS concentration
k1: Disinfectant decay rate – cells
𝑑𝑆 disinfection
= 𝐷𝑆−𝐵 𝛻 ∙ 𝛻𝑆 + 𝐹𝑋 𝑘1 𝐶𝑋 + 𝐹𝐸 𝑘2 𝐶𝐸 t=0
𝑑𝑡 SBs = 0 k2: Disinfectant decay rate – EPS oxidation
SBo = 0 DS-B: Diffusion coefficient of DBP in biofilm
(5-4)
FX: Fraction of cells transformed into DBP
FE: Fraction of EPS transformed into DBP
So: Initial concentration of DBP
t: Time

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 176
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Initial
Phase Equation Description of variables Subscripts
conditions
C: Chlorine concentration

𝑑𝐶 𝑑𝐶 ⃗ : Velocity vector
𝑢
+ ∇(𝑢
⃗ 𝐶) = 𝐷𝐶−𝐿 𝛻 ∙ 𝛻𝐶 − |𝐵𝑖𝑜𝑓𝑖𝑙𝑚 t=0
𝑑𝑡 𝑑𝑡 DC-L: Molecular diffusion coefficient of
CLo = Co
chlorine in liquid
(5-5) Co: Initial concentration of chlorine
Bulk water

t: Time
S: DBP concentration
𝑑𝑆 𝑑𝑆 X: Cells concentration
+ ∇(𝑢
⃗ 𝑆) = 𝐷𝑆−𝐿 𝛻 ∙ 𝛻𝑆 + |𝐵𝑖𝑜𝑓𝑖𝑙𝑚 t=0
𝑑𝑡 𝑑𝑡 E: EPS concentration
SLo = 0
(5-6) DS-L: Diffusion coefficient of DBP in liquid
t: Time

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 177
Disinfection by-products formation from disinfection of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Table 5-2 presents the boundary conditions for the transport of dissolved substances. Continuity
flux boundary condition is included at the biofilm surface (Bs) for chlorine (Equation (5-7)) and DBP
(Equation (5-8)). This boundary condition means that substance concentrations and flux are the
same on both sides of the interface between bulk water and biofilm (IWA Task Group on Biofilm
Modeling et al., 2006). On the other hand, there is no exchange of particulate substances, cells
and EPS, between biofilm and bulk flow (no flux condition – Equation (5-9)). At the pipe inlet, the
boundary condition is the concentration of chlorine and DBP (Equations (5-10) and (5-11),
respectively). At the pipe outlet, the boundary condition is represented by the outflow, which
assumes that there is zero gradient in the normal direction, of neither chlorine (Equation (5-12)) nor
DBP (Equation (5-13)) at this boundary. At the substratum (pipe wall), there is no exchange of
neither particulate nor dissolved substances (no flux condition – Equations (5-14) and (5-15),
respectively).

Table 5-2. Boundary conditions for transport of dissolved substances in 2D model – Cylindrical
pipes
Boundary Boundary conditions Description of variables Subscripts
𝑑𝐶 𝑑𝐶
𝑗𝐶−𝐵𝑠 = −𝐷𝐶−𝐵 |𝐵 = −𝐷𝐶−𝐿 |𝐿
𝑑𝑟 𝑑𝑟 j: Flux C: Chlorine
(5-7) C: Chlorine concentration S: DBP
Biofilm surface (Bs)

S: DBP concentration X: Cells

𝑑𝑆 𝑑𝑆
𝑗𝑆−𝐵𝑠 = −𝐷𝑆−𝐵 |𝐵 = −𝐷𝑆−𝐿 | 𝐿 DC-B: Molecular diffusion E: EPS
𝑑𝑟 𝑑𝑟
coefficient of chlorine in biofilm B: Biofilm
(5-8)
DC-L: Molecular diffusion L: Bulk liquid
𝑗𝑋−𝐵𝑠 = 𝑗𝐸−𝐵𝑠 = 0 coefficient of chlorine in liquid Bs: Biofilm

(5-9) surface
Cinlet: Chlorine concentration at r: Pipe
𝐶𝑖𝑛𝑙𝑒𝑡 = 𝐶𝑜
the pipe outlet radius, r-
(5-10) Co: Initial concentration of coordinate
Inlet

chlorine z: z-
𝑆𝑖𝑛𝑙𝑒𝑡 = 0
Sinlet: DBP concentration at the coordinate
(5-11)
pipe outlet

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 178
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Boundary Boundary conditions Description of variables Subscripts
n: Normal vector
DC-L: Molecular diffusion

𝑛 ∙ −𝐷𝐶−𝐿 ∇𝐶 = 0 coefficient of chlorine in liquid

DS-L: Diffusion coefficient of
(5-12)
Outlet
DBP in liquid
𝑛 ∙ −𝐷𝑆−𝐿 ∇S = 0 C: Gradient of chlorine

(5-13) concentration
S: Gradient of DBP
concentration
𝑗𝑋−𝑤𝑎𝑙𝑙 = 𝑗𝐸−𝑤𝑎𝑙𝑙 = 0

(5-14)
Walls

j: Flux
𝑗𝐶−𝑤𝑎𝑙𝑙 = 𝑗𝑆𝑤𝑎𝑙𝑙 = 0

(5-15)

5.2.3 Model equations and initial and boundary conditions – Bulk flow

Table 5-3 shows the Navier-Stokes equations for solving the flow field for incompressible flow
(mass-continuity and momentum). Here, drinking water is treated in its liquid phase and energy
conservation equation was not solved because the model was developed for isothermal conditions.
Navier (1827; cited by Canuto et al. (2007)) must be credited with the first attempt at deriving the
equations for homogeneous incompressible viscous fluids on the basis of considerations involving
the action of intermolecular forces. Stokes (1845; cited by Canuto et al. (2007)) who, under the sole
assumption that the stresses are linear functions of the strain rates, derived the equations in the
form that is currently in use. Equation (5-16) is the equation of mass conservation, also known as
the continuity equation. The equation for momentum conservation is given by Equation (5-17). The
expression on the right-hand side of this equation ((𝑢⃗)) corresponds to the viscous stress
tensor.

Table 5-3. Navier-Stokes equations for incompressible flow

Component Equation Description of variables
𝝏 : Fluid density
+ 𝛁(𝝆𝒖
⃗ )=𝟎
Mass 𝝏𝒕 ⃗ : Velocity vector
𝑢
(5-16) t: Time

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 179
Disinfection by-products formation from disinfection of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Component Equation Description of variables

𝝏(𝝆𝒖⃗)
+ 𝛁(𝝆𝒖
⃗ ⃗𝒖) = −𝛁𝒑 + 𝛁(𝛍𝛁𝒖
⃗) p: Pressure
Momentum 𝝏𝒕
: Molecular viscosity
(5-17)

Table 5-4 presents the boundary conditions for the flow field. Normal velocity condition is specified
at the inlet (Equation (5-18)), pressure at the outlet (Equation (5-19)), and no slip at the walls
(Equation (5-20)).

Table 5-4. Boundary conditions for flow field in 2D model – Cylindrical pipes
Boundary Boundary conditions Initial conditions Description of variables
Uo: parabolic profile of Vo u: Normal inflow velocity
𝑢 = −𝑛𝑈𝑜 Turbulence conditions: magnitude
Inlet
Turbulent intensity = 0.1 Uo: Velocity magnitude
(5-18)
Turbulent length scale =  Vo: Average velocity
𝑝=0
Outlet - p: Pressure
(5-19)
Walls and 𝑢
⃗ =0
Biofilm surface - ⃗ : Velocity vector
𝑢
(Bs) (5-20)

The model SST (Shear Stress Transport) was chosen for solving the transitional and turbulent flow.
Details about the process to select the appropriate turbulence model are presented in the Section
5.2.4.3. Turbulence is a property of the flow and it is measured by the Reynolds number (Re). In
this study, the range Re = 2300 – 10,000 was used as the range for the transition regime between
laminar and turbulent flow (Abraham et al., 2009). Turbulence is characterized by a wide range of
flow scales: the largest occurring scales, which depend on the geometry, the smallest, quickly
fluctuating scales, and all the scales in between (COMSOL, 2013). The Navier-Stokes equations
can be used for turbulent flow simulations, although this would require a large number of elements
in order to capture the turbulent flow fluctuations in time and space. A strategy to tackle this issue
is to divide the flow into large, resolved scales and small, unresolved scales. The small scales are
then modelled using a turbulence model to reduce the numerical cost of resolving all present scales
(COMSOL, 2013).

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 180
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
In this line, the Navier-Stokes equations for incompressible flow are adapted to include the flow
quantities into an averaged value and a fluctuating part. Then, 𝑢
⃗ is replaced by the averaged
velocity field U (COMSOL, 2013) (Table 5-5 and Table 5-6). The term ∇ ∙ (𝜌𝑢′ ⨂𝑢′ ) specified in
the Equation (5-22) is called the Reynolds stress tensor (COMSOL, 2013).

Table 5-5. Navier-Stokes equations for turbulent incompressible flow

Component Equation Description of variables
𝝏 : Fluid density
+ 𝛁(𝝆𝑼) = 𝟎
Mass 𝝏𝒕 U: Averaged velocity field
(5-21) t: Time

𝝏(𝝆𝑼)
+ 𝛁(𝝆𝑼) + 𝛁 ∙ (𝝆𝒖′ ⨂𝒖′ ) = −𝛁𝒑 + 𝛁(𝛍𝛁𝑼) p: Pressure
Momentum 𝝏𝒕
: Molecular viscosity
(5-22)

The equations of the turbulence model SST are included in Table 5-6. The SST turbulence model
combines the κ-ε and κ-ω turbulence models. The model equations are formulated in terms of
turbulent kinetic energy (κ) (Equation (5-23)) and the dissipation per unit turbulent kinetic energy
(ω) (Equation (5-24)). The turbulent viscosity is expressed by the Equation (5-25). E is the
characteristic magnitude of the average velocity gradients (Equation (5-26)). P is a function of Pk
(Equation (5-27)), which is the rate of production of the turbulent kinetic energy κ (Equation (5-28)),
and the terms κ, ω, ω2 are Prandtl number-like parameters for the transport κ and ω (Abraham
et al., 2009). The model constants are defined through interpolation of appropriate inner and outer
values (Equation (5-29)). F1 is a blending function (Equation (5-30)) that facilitates the combination
of the standard κ-ε and κ-ω models. The  terms are model constants.

The factor , which multiplies the production term P in Equation (5-24), has the role of diminishing
the rate of turbulence production in flows that are not fully turbulent (Abraham et al., 2009). The
values of  range between 0 and 1. COMSOL interpolates  by Equation (5-29) with default values
1 = 5/9 and 2 = 0.44 (Table 5-6). On the other hand, Abraham et al. (2009) presented a transition
model to adapt the SST turbulence model to non-fully turbulent flows for the simulation of heat
transfer in round pipes. The adaptation was made via the damping factor , which depends also on
the local stability status of the flow in the near-wall region. Complete development of the transition
model can be found in Abraham et al. (2009). Despite of some Re of the simulations presented in

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 181
Disinfection by-products formation from disinfection of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
this study fall into the transitional flow regime, the SST model included in COMSOL Multiphysics
5.2a was used (see Section 5.2.4.3).

SST model uses the wall distance parameter (l w; Equations (5-31) and (5-34)), which is provided
by a mathematical Wall Distance interface. Wall distance is included when using the SST model
implemented with COMSOL Multiphysics 5.2a. The solution to the wall distance equation is
controlled using the parameter lref. The distance to objects larger than lref is represented accurately,
while objects smaller than lref are effectively diminished by appearing to be farther away than they
actually are. This is a desirable feature in turbulence modelling since small objects would get too
large impact on the solution if the wall distance were measured exactly (COMSOL, 2013). Finally,
the model SST in COMSOL Multiphysics 5.2a includes the default constants specified in Table 5-6,
which remained unchanged in the application to the current biofilm chlorination model.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 182
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 5-6. Equations of the turbulent model SST
Equations Description of variables
𝜕(𝜌𝑘)
+ ∇(𝜌𝑘𝑢⃗ ) = 𝑃 − 𝜌𝛽𝑜∗ 𝑘ω∇ ∙ [(𝜇 + 𝜎𝑘 𝜇 𝑇 )∇𝑘] (5-23)
𝜕𝑡 t: Time
𝜕 (𝜌𝜔) 𝜌𝛾 𝜌𝜎𝜔2
+ ∇(𝜌𝜔𝑢
⃗)= P − ρβω2 + ∇ ∙ [(𝜇 + 𝜎𝑘 𝜇 𝑇 )∇𝑘] + 2(1 − 𝐹1 ) ∇𝜔 ∙ ∇𝑘 (5-24) k: Turbulent kinetic energy
𝜕𝑡 𝜇𝑇 𝜔
ω: Dissipation per unit turbulent
𝜌𝑎1 𝑘
𝜇𝑇 = (5-25) kinetic energy
max(𝑎1 𝜔, 𝐸𝐹2 )
turbulent kinetic energy
𝐸 = √2𝐸𝑖𝑗 𝐸𝑖𝑗 (5-26)
ρ: Fluid density
𝑃= min(𝑃𝑘 , 10𝛽𝑜∗ 𝑘ω) (5-27)
: Molecular viscosity
2 2
𝑃𝑘 = 𝜇 𝑇 (∇𝑢 ⃗ )𝑇 ) − (∇ ∙ 𝑢
⃗ + (∇𝑢
⃗ : (∇𝑢 ⃗ )2 ) − 𝜌𝑘∇ ∙ 𝑢
⃗ (5-28)  : Turbulent viscosity
3 3 

Constants of the model: 𝜆 = 𝐹1 𝜆1 + (1 − 𝐹1 )𝜆2 for 𝜆 = 𝛽, 𝛾, 𝜎𝑘 , 𝜎𝜔 (5-29) F1, F2: Interpolation functions
4
𝐹1 = 𝑡𝑎𝑛𝑔ℎ(𝜃1 ) (5-30) E: Characteristic magnitude of the
√𝑘 500𝜇 4𝜌𝜎𝜔2 𝑘
𝜃1 = 𝑚𝑖𝑛 [𝑚𝑎𝑥 ( ∗ , ), ] (5-31) average velocity gradients
𝛽𝑜 𝜔𝑙𝑤 𝜌𝜔𝑙𝑤 2 𝐶𝐷𝑘𝜔 𝑙𝑤 2
T: Temperature
2𝜌𝜎𝜔2
𝐶𝐷𝑘𝜔 = 𝑚𝑎𝑥 ( ∇𝜔 ∙ ∇𝑘, 10−10 ) (5-32) Pk: Production term
𝜔
𝐹2 = 𝑡𝑎𝑛𝑔ℎ(𝜃2 2 ) (5-33) lu: Distance to the closest wall
2√𝑘 500𝜇 1, 1, *0, 1, 2, k1, k2, ω1,
𝜃2 = 𝑚𝑎𝑥 ( ∗ , ) (5-34)
𝛽𝑜 𝜔𝑙𝑤 𝜌𝜔𝑙𝑤 2 ω2, a1: constants of the model
Turbulence model parameters: The default model constants are given by:
1=0.075 | 2=0.0828 | *0=0.09 | 1=5/9 | 2=0.44 | k1=0.85 | k2=1.0 | ω1=0.5 | ω2=0.856 | a1=0.31
Source: adapted from COMSOL (2013)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 183
Disinfection by-products formation from disinfection of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.2.4 Pre-processing

5.2.4.1 Geometry and mesh

The current model presents some mathematical challenges at the biofilm surface; fluxes vary from
zero to a finite value in few seconds. Several mesh and geometry configurations were tested and
it was observed that, when the mesh size in this region was higher than the length scale of the
mathematical solution, negative concentrations of the dissolved substances arose at the pipe wall
and/or outlet. In order to avoid this, the optimal geometry and mesh were selected in function of
obtaining positive concentrations of chlorine and chloroform and positive fluxes, under laminar and
transition flow regimes. Therefore, five configurations were tested by a combination of several
geometry and model characteristics, which are specified in Table 5-7.

Table 5-7. Geometry configurations tested in the current model

Main bulk Full Stabilization domain with no reaction
Configuration Surface
water biofilm Before main bulk After main bulk
Nr. reactions
domain thickness water domain water domain
1 X X
2 X X
3 X X X
4 X X X X
5 X X X

The main bulk water domain represents the portion of the pipe where wall reactions were occurring
along its length, from the beginning to the end. Surface reactions indicated the solution of wall
reactions in COMSOL by assuming that the biofilm is completely mixed, then simulations of the
reactions along the total biofilm thickness is not necessary. This represented a simplification of the
model because specifying reactions at the wall as boundary was sufficient. Full biofilm thickness
means that wall reactions were specified along the total biofilm thickness, then constructing a
subdomain for this was necessary. A stabilization domain with no reactions refers to the
construction of a subdomain for bulk water, where no reactions were taking place. This was made
in order to allow the development of the flow and the transport of dissolved substances without
steep changes of their concentrations due to reactions with biomass.

The meshes were created with COMSOL Multiphysics 5.2a. The first attempt was made with a
triangular mesh refined at the wall and at the boundary between the stabilization and bulk water
domains; biofilm reactions were included as surface reactions. However, fluxes at the wall
presented notably oscillations. This mesh was discarded and effort was focused on rectangular
element meshes. Finally, the configuration number 5 was selected together with a mapped mesh,
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 184
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
including boundary layers at the biofilm surface and in between stabilization and bulk water
domains in order to smooth the transition between elements of different size. Figure 5-4 presents
the geometry configuration and Figure 5-5 shows the mesh details.

Figure 5-4. Geometry of the water pipe

Figure 5-5. Mesh details of the water pipe

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 185
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
The software COMSOL offers the parameter “mesh quality” to assess the degeneration of the
elements, according to their shapes. Quality varies between 0 and 1; 0 represents the presence of
degenerated elements and 1 perfectly regular elements. Low mesh quality leads to inverted
elements and convergence issues (COMSOL, 2012); therefore, it is preferable to obtain mesh
quality near to 1 at the time of constructing the mesh. For the stabilization domain, high mesh
quality was not necessary because this domain was only used for development of the flow and
transport of chlorine with no reactions. The mesh quality of the biofilm subdomain is very low due
to the aspect ratio of the elements; thickness varied between 7-102 m while length was 1 m. In
order to minimise this difference, a boundary layer was included at the biofilm surface to improve
the transition of the element size between biofilm and bulk water domains.

The highest mesh quality was established for bulk water in order to better predict the transport of
dissolved substances along the pipe. Another boundary layer was included between the
stabilization and bulk water domain. Overall, the selected geometry and constructed mesh were
appropriate for the scope of the current model, obtaining reasonable chlorine and chloroform
concentrations and fluxes at the pipe wall and outlet, within an acceptable computational time (see
Sections 5.2.4.2 and 5.3.1). Table 5-8, Table 5-9 and Table 5-10 describe the main mesh
characteristics for  of 10, 6, and 3 (inches), respectively. The number of elements was changed
according to the pipe diameter in order to keep the same mesh quality. The length of the pipe was
kept equal to 1 m to ease the interpretation of the results.

Table 5-8. Characteristics of the mapped mesh for pipe diameter of 10 inches
Element distribution Number of Average
Domain / Region Size elements
on edges elements quality
 Length: 0.5 m  Wall: 150
Stabilization 34,200 0.350
 Predefined size: normal  Inlet: 150
 Length: 1.0 m  Biofilm surface: 800
Bulk water 152,000 0.740
 Predefined size: fine  Wall: 150
 Length: 1.0 m
 Biofilm surface: 800
Biofilm  Thickness: 7-102 m 4,000 0.002
 Wall: 5
 Predefined size: fine
Whole mesh statistics: Total number of elements / Average quality 190,200 0.651
 Number of boundary layers: 40
Boundary layer in biofilm
 Stretching factor: 1.1
surface
 Thickness of first layer: BT/5
Boundary layer in  Number of boundary layers: 30
between stabilization  Stretching factor: 1.5
and bulk water domain  Thickness of first layer: L/800

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 186
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 5-9. Characteristics of the mapped mesh for pipe diameter of 6 inches
Element distribution Number of Average
Domain / Region Size elements
on edges elements quality
 Length: 0.5 m  Wall: 150
Stabilization 21,600 0.340
 Predefined size: normal  Inlet: 100
 Length: 1.0 m  Biofilm surface: 800
Bulk water 96,000 0.740
 Predefined size: fine  Wall: 100
 Length: 1.0 m
 Biofilm surface: 800
Biofilm  Thickness: 7-102 m 4,000 0.002
 Wall: 5
 Predefined size: fine
Whole mesh statistics: Total number of elements / Average quality 121,600 0.646
 Number of boundary layers: 20
Boundary layer in biofilm
 Stretching factor: 1.2
surface
 Thickness of first layer: BT/5
Boundary layer in  Number of boundary layers: 30
between stabilization  Stretching factor: 1.5
and bulk water domain  Thickness of first layer: L/800

Table 5-10. Characteristics of the mapped mesh for pipe diameter of 3 inches
Element distribution Number of Average
Domain / Region Size elements
on edges elements quality
 Length: 0.5 m  Wall: 150
Stabilization 10,800 0.340
 Predefined size: normal  Inlet: 50
 Length: 1.0 m  Biofilm surface: 800
Bulk water 48,000 0.741
 Predefined size: finer  Wall: 50
 Length: 1.0 m
 Biofilm surface: 800
Biofilm  Thickness: 7-102 m 4,000 0.002
 Wall: 5
 Predefined size: fine
Whole mesh statistics: Total number of elements / Average quality 62,800 0.625
 Number of boundary layers: 40
Boundary layer in biofilm
 Stretching factor: 1.1
surface
 Thickness of first layer: BT/5
Boundary layer in  Number of boundary layers: 30
between stabilization  Stretching factor: 1.4
and bulk water domain  Thickness of first layer: L/800

5.2.4.2 Mesh convergence and selection of time step

Once the base mesh was defined for 10-inch pipes (mesh 1, 190.200 elements, Table 5-8), the
next step was to determine the mesh convergence. This was done by refining that mesh in both
directions (r, z), including the boundary layers; such refinement consisted on doubling the number
of elements. The mesh 2 resulted in 752.800 elements. Taking into account that the current model
was applied for prediction of bulk concentrations of DBPs and characterisation of mass transport
of dissolved substances at the biofilm surface, velocity at the centre of the bulk water domain along
r axis and the parameter Sh
¯ (Equation (2-11)) for both chlorine and chloroform were used to
compare both meshes. These parameters are representing the flow regime and mass transport

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 187
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
simulations of the model. Several time steps (i.e., 1s, 5 s, 10 s, 30 s, 1 min, 5 min) were also
compared for Sh
¯ in order to identify the appropriate time step simultaneously to mesh convergence
test.

Results of velocity at the centre of the bulk water domain and the parameter Sh
¯ are presented in
Figure 5-6. The conditions of the simulations are also specified in Figure 5-6a. Result comparisons
allowed determining that results of velocity and Sh
¯ obtained with both meshes are similar.
Percentage of difference for maximal, median, and average velocity were 0%, 0%, and 0.11%,
respectively. In the case of Sh
¯ for chlorine, percentage of difference ranged between 1.46-3.18%
for the assessed time steps. Finally, Sh
¯ for chloroform, percentage of difference ranged between
0.16-1.10% for the assessed time steps. Due to small differences between two meshes, the mesh
1 was selected for running the subsequent simulations.

(a)

(b)
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 188
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (c)
Figure 5-6. Comparison of velocity and Sh
¯ between mesh 1 and 2 (a) Velocity monitored in the
centre of bulk water domain along r axis (b) Sh
¯ for chlorine (c) Sh
¯ for chloroform

In relation to the time step, the selected ones resulted in dissimilar results between consecutive
pairs for every mesh. The percentage of difference ranged between 19.38-121.47% and 2.54-
97.87% for chlorine and chloroform, respectively with mesh 1; and between 19.07-121.26% and
2.63-98.35% for chlorine and chloroform, respectively with mesh 2. For this reason, a time step of
0.1 s was also tested with mesh 1. Figure 5-7a presents the comparison of Sh
¯ for both chlorine and
chloroform and Figure 5-7b shows the comparison of average chloroform concentrations at the
pipe outlet for several time steps. In relation to the latter, percentage of difference varied between
0.0-3.98%, 0.15-86.47% and 0.02-6.33% for maximal, median and average concentrations,
respectively, for consecutive pairs of time steps. In particular, percentage of difference between
values corresponding to 0.1 s and 1 s resulted in 0%, 0.15% and 0.02% for maximal, median and
average concentration of chloroform, respectively. In the case of Sh
¯ , percentage of difference was
reduced to 4.78% and 0.08% for chlorine and chloroform, respectively between time steps 0.1 s
and 1 s. Due to small differences between time steps 0.1 s and 1 s, the latter one was selected for
the subsequent simulations.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 189
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 5-7. Comparison of Sh
¯ (a) and average chloroform concentration at the pipe outlet (b) in
mesh 1

5.2.4.3 Selection of turbulence model

Considering that drinking water pipes are characterized by turbulent flow regime (Cogan, 2010), it
is important to analyse the mass transport of chlorine and DBPs in the current 2D model under high
Re. In addition, the current model involves reactions at the pipe wall; therefore, the flow near to it
becomes important to simulate correctly. In this line, a brief background on turbulence models is
presented here in order to better explain the selection of the turbulence model for the subsequent
simulations. According to Frei (2013), the turbulent flow near a flat wall can be divided up into four
regimes (Figure 5-8). At the wall, the fluid velocity is zero, and for a thin layer above this, called the
viscous sublayer or laminar sublayer, the flow velocity is linear with distance from the wall. Further
away from the wall, the region buffer layer is located. In the buffer region, the flow begins to
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 190
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
transition to turbulent and it eventually transitions to a region where the flow is fully turbulent and
the average flow velocity is related to the log of the distance to the wall. This is known as the log-
law region. Even further away from the wall, the flow transitions to the free-stream region. The
viscous and buffer layers are very thin and, if the distance to the end of the buffer layer is δ, then
the log-law region will extend about away from the wall.

Figure 5-8. Regimes of the turbulent flow near to a flat wall

Source: Frei (2013)

Three Reynolds-averaged Navier-Stokes turbulence models from seven models available in

COMSOL Multiphysics 5.2a were tested in order to select the most appropriate one to simulate the
mass transfer at the biofilm surface. Table 5-11 presents a brief description of the models SST,
yPlus, and Low Reynolds number κ-ε. These models were selected as they solve the flow
everywhere, therefore a detailed comparison on velocity magnitude and concentrations of DBPs at
the pipe outlet could be carried out. In the case of simulations of heat transfer in round pipes,
versions of low Reynolds number κ-ε and low Reynolds number Reynolds stress turbulence models
(Thakre and Joshi, 2000), κ-ε with standard wall functions (Jayakumar et al., 2008), and SST
(Abraham et al., 2009; Di Piazza and Ciofalo, 2010) have been used in comparison with
experimental data, for transitional and fully developed turbulent flows.

Table 5-11. Description of five turbulence models included in COMSOL Multiphysics 5.2a
Regions where
Turbulence Variables
the flow is solved Applications Limitations
model computed
/ not solved
It is a It tends to be most accurate
Everywhere
combination of when solving the flow near
SST the κ-ε in the the wall --
It does not use wall
free stream
functions
and the κ-ω Flow over an aerofoil

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 191
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Regions where
Turbulence Variables
the flow is solved Applications Limitations
model computed
/ not solved
models near
the walls
Everywhere It provides good
Turbulence approximations for internal The least
yPlus
viscosity It does not use wall flow, especially in electronic accurate model
functions cooling applications
κ: turbulent
Everywhere
Low kinetic energy Lift and drag forces and It uses more
Reynolds ε: rate of heat flux can be modelled memory than
It does not use wall
number κ-ε dissipation of with higher accuracy standard κ-ε
functions
kinetic energy
Adapted from Frei (2013)

To select the appropriate turbulence model for the subsequent simulations, simulations with the
five turbulence models described in Table 5-11 were run. Velocity magnitude at the centre of bulk
water domain along r axis (see Figure 5-6a) and chloroform concentrations at the pipe outlet were
used to compare the performance of these models (Figure 5-9). The three models solved the flow
in this region and the velocity magnitude is zero at the pipe wall. In order to quantify the similarity
between turbulence models, percentage of difference was calculated for pairs or maximal (Table
5-12), median (Table 5-13), and average (Table 5-14) velocity. Pair of models yPlus/SST was the
only one which resulted in percentage of difference lower than 5% for every velocity value
compared.

(a)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 192
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-9. Comparison of five turbulence models included in COMSOL Multiphysics 5.2a (a)
Velocity monitored at the middle of bulk water domain (b) Average chloroform concentrations at
the pipe outlet

Table 5-12. Percentage of difference between pairs of turbulence model – Maximal velocity
Low Re
Turbulence model yPlus SST
κ-ε
yPlus -
Low Re κ-ε 20.53% -
SST 4.77% 15.80% -

Table 5-13. Percentage of difference between pairs of turbulence model – Median velocity
Low Re
Turbulence model yPlus SST
κ-ε
yPlus -
Low Re κ-ε 7.14% -
SST 4.54% 2.60% -

Table 5-14. Percentage of difference between pairs of turbulence model – Average velocity
Low Re
Turbulence model yPlus SST
κ-ε
yPlus -
Low Re κ-ε 8.05% -
SST 2.24% 5.81% -

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 193
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
The effect of turbulence on concentrations of chloroform is observed in Figure 5-9b. Concentration
curves are very dissimilar. Percentage of difference of pairs of turbulence models were higher than
5% for all the values assessed here, except for average values of the pair Low Reynolds/SST.

Up to this point, it is clear that solution of the flow near to the wall is crucial as the current model
involves reactions at the pipe wall and mass transport at the biofilm surface. For instance, Di Piazza
and Ciofalo (2010) found that models SST and Reynolds stress model – ω were in excellent
agreement with direct numerical simulation results in relation to pressure drop in helically coiled
heat exchanger and Re of 14,000 and 80,000. With regards to heat transfer, the predicted
temperature with both models were in satisfactory agreement (Di Piazza and Ciofalo, 2010).

Taking into account that there are no experimental data to compare with in order to select the most
appropriate turbulence model, comparisons between results obtained with the software Ansys
Fluent 16.0 were made. Simulations using Ansys Fluent 16.0 used an axisymmetric mesh, with
64,135 rectangular elements, = 10 inches and L = 1 m. The selected models were laminar and
SST; laminar was included because low Re (1248) was used in this analysis. Laminar flow regime
was included in order to analyse a wider range of Re. Figure 5-10a shows the comparison of
velocity magnitude monitored in the middle of bulk water domain for models laminar and SST run
in both software COMSOL and Ansys Fluent. Similarly, maximal (Table 5-15), median (Table 5-16),
average (Table 5-17) velocities were compared for each turbulence model between both software.

Velocity profile of laminar and SST model resulting from both software are similar for maximal and
median velocity (percentage of difference lower than 5%), but percentage of difference for average
velocity was higher than 23%. This similarity between both software for laminar and SST turbulence
models represents that the solution is solver independent and that the model can be replicated in
both software. In conclusion, the turbulence model selected for the subsequent simulations in this
study was SST; the curves of this model obtained from COMSOL and Ansys Fluent are included
in Figure 5-10b for better illustration. It is important to highlight that SST tends to be more accurate
solving the flow near the wall because it combines κ-ε and κ-ω models (Frei, 2013).

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 194
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 5-10. Comparison of velocity monitored at the middle of bulk water domain between
software COMSOL and Fluent (a) Laminar and SST (b) SST

Table 5-15. Percentage of difference between COMSOL and Fluent for the same turbulence model –
Maximum velocity
Software Fluent
Turbulence model Laminar SST
COMSOL Laminar 0.02% -
SST - 0.47%

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 195
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Table 5-16. Percentage of difference between COMSOL and Fluent for the same turbulence model –
Median velocity
Software Fluent
Turbulence model Laminar SST
COMSOL Laminar 0.88%
SST - 4.52%

Table 5-17. Percentage of difference between COMSOL and Fluent for the same turbulence model –
Average velocity
Software Fluent
Turbulence model Laminar SST
COMSOL Laminar 24.12%
SST - 23.71%

Due to the importance of coupling flow and mass transport, further analysis of the influence of
turbulence on mass transport was carried out by comparing Sh
¯ of chlorine and chloroform resulting
from conditions described in Figure 5-9 and Figure 5-10. Results are shown in Table 5-18; the
typical behaviour of the model described in Section 5.3.1 presents the normal variation in time of
Sh. The Sh
¯ calculated for the turbulence models presented in Table 5-18 are dissimilar; percentage
of difference for every pair of Sh
¯ were as high as 173%. Sh
¯ for chloroform with laminar flow and
SST turbulence model corresponding to simulation with inlet velocity equal to the normal velocity
magnitude were the only pair of similar values (0.0011925 vs. 0.0012109; percentage of difference
= 1.53%).

Table 5-18. Sh
¯ for different turbulent models
Inlet velocity: Normal
Turbulence Inlet velocity: Parabolic profile
velocity magnitude
model
Chlorine Chloroform Chlorine Chloroform
Laminar N.A N.A 0.0011925 0.0056078
yPlus 0.0014186 0.082039 N.A N.A
Low Re 0.0041935 0.170580 N.A N.A
SST 0.0027529 0.116310 0.0012109 0.076996
N.D: Not Defined | N.A: Not Apply

The previous analysis was undertaken to select the most appropriate turbulence model for coupling
flow and mass transport during the formation of chloroform from chlorine disinfection. This enabled
the conclusion to be drawn that DBP concentrations are highly sensitive to the turbulence model
used in the simulations. Therefore, further field or laboratory research regarding DBP formation

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 196
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
from biofilms under hydrodynamic conditions should consider turbulence as an important factor
influencing the results.

5.2.5 Implementing the model in COMSOL

As explained previously, rectangular meshes (Table 5-8, Table 5-9, and Table 5-10), time step of
1 s, and SST turbulence model were chosen to predict the DBP formation potentials, in drinking
water pipes, by chlorination of biofilms, under hydrodynamic conditions. First, the flow field was
solved under steady state, then transient simulations for substance transport from this converged
steady flow field were performed; it was assumed dissolved substances (chlorine and DBPs) do
not affect the flow field. The discretization approach was P1+P1 (linear elements for both the
velocity components and the pressure field) for solution of flow, linear for solution of transport of
dissolved substances, and quadratic discretization for solution of decay of cells and EPS.

The solver was “Direct” for both transitional flow and transport of dissolved substances simulations.
Direct methods find an approximation of the solution A-1 f = u by matrix factorization in a number of
operations that depend on the number of unknowns. Factorization is expensive, but once it has
been computed, it is relatively inexpensive to solve for new right-hand sides. Since direct methods
are expensive in terms of memory and time intensive for CPUs, they are preferable for small- to
medium-sized 2D and 3D applications (Marra, 2013). This is the case of the current model;
therefore, direct solver was selected.

The numerical simulations were run in a computer with a processor Intel® Core™ i5-4430 CPU
3.00 GHz, and RAM memory of 16.0 GB. The model was implemented in the software COMSOL
Multiphysics 5.2a, by a 2D axisymmetric geometry of a cylindrical pipe. Computational time ranged
between 1-10 hours, according to domain size, flow regime, and simulation time. For instance,
each simulation for  = 3 inches, transitional flow and simulation time of one hour lasted around 1-
2 hours. Each simulations of the sensitivity analysis lasted around 3 hours. Table 5-19 presents
the modules, boundary and initial conditions used in each simulated domain.

Table 5-19. Main features for model implementation in COMSOL Multiphysics 5.2a
Module Domain Boundary Boundary condition Initial values
Normal velocity –
Inlet
Parabolic profile
Laminar flow Bulk water 0
Outlet Pressure. Po = 0
Bs No slip

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 197
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Module Domain Boundary Boundary condition Initial values
Centre of the
Axial symmetry
pipe
Normal velocity –
Parabolic profile
Inlet
Transitional Turbulent intensity = 0.1
and turbulent Turbulent length scale = 
Bulk water 0
flow Outlet Pressure. Po = 0
SST model Wall No slip
Centre of the
Axial symmetry
pipe
Inlet Inflow. Co = Co; Ro = 0
Outlet Outflow
Transport Bulk water Bs Continuity Co = Co; Ro = 0
diluted Centre of the
substances Axial symmetry
pipe
(tds) Wall No flux
Biofilm Co = 0; Ro = 0
-- Reactions
Reactions for
Domain ODEs Biofilm cells and EPS -- Xo = Xo; Eo = Eo
decay

5.2.6 Scenarios simulated

The scenarios simulated with the model proposed here are based on the variables flow regime,
Reynolds number, pipe diameter, and drinking water quality, in order to identify their influence on
response variables such as average DBP concentration at the pipe outlet, Sh
¯ and mass transfer
rate of DBPs. The conditions of the simulated scenarios were as it follows:

 Laminar, transitional, and turbulent flow

 Re: 1,202 – 50,030
 Pipe diameters of 3, 6 and 10 inches
 BT of 7 m and 102 m, Clo of 0.12 mg/L and 1.66 mg/L, and Xo of 2.82 x 10-4 mg/cm2 and
4.44 x 10-3 mg/cm2

5.2.7 Parameters selection

The selection of parameters was explained in Chapter 3. Table 5-20 presents the consolidation of
the microbiological and kinetic parameters; and hydraulic-related parameters such as flow rate,
velocity and pipe diameter considered for simulations of the scenarios described in the previous
section.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 198
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 5-20. Parameters used in the 2D model simulations
Parameter Value Reference
Biofilm thickness (BT) 7 and 102 m Table 4-3
Initial biofilm density (Xo) 2.82x10-4 and 4.44x10-3 mg/cm2 Table 4-5
Initial EPS concentration (Eo) 𝐸𝑜 = 5.0295𝐵𝑇 + 1884.3 Celmer et al. (2008)
Temperature (T) 25 °C Section 3.3.5
Diffusion coefficient of chlorine in
2.66 x 10-9 m2/s Chen and Stewart (1996)
liquid
Diffusion coefficient of chloroform in
9.29 x 10-10 m2/s Buzatu et al. (2007)
liquid
Diffusion coefficient of DCAN in
1.66 x 10-9 m2/s Poling et al. (2007)
liquid
Chlorine-cell reaction rate (k1) 1.1x10-3 m3/g-sec Chen and Stewart (1996)
Chlorine-EPS reaction rate (k2) 3.7x10-6 m3/g-sec Chen and Stewart (1996)
Chlorine-cell yield coefficient (Y1) 1.85 g/g Chen and Stewart (1996)
Chlorine-EPS yield coefficient (Y2) 540 g/g Chen and Stewart (1996)
Portion of chlorine reacting with
0.70 Section 4.2.3.6
cells - chloroform (Fx )
Portion of chlorine reacting with
0.30 Section 4.2.3.6
cells - DCAN (Fx )
Portion of chlorine reacting with
0.30 Section 4.2.3.6
EPS - Chloroform (FE )
Portion of chlorine reacting with
0.17 Section 4.2.3.6
EPS - DCAN (FE )
S 0.50 Section 5.2.8
260 – 1080 L/h Abokifa et al. (2016a)
Flow rate (Q)
2170 – 2820 L/h -
Re 1,202 – 50,030 -
Pipe diameter () 3, 6, 10 inches -

5.2.8 Effective diffusion coefficient of DBPs

In Chapter 4, the influence of s on chloroform concentrations was observed for time less than 60
min and values as small as s = 0.016, under stagnation conditions. Differences were noted on the
speed of the reactions but not in the potential concentrations of chloroform. Taking into account
that there are not experimental measurements of diffusion of chloroform and DCAN within biofilms,
the influence of s on concentrations of both substances were examined again under transitional
flow. Figure 5-11 shows the chloroform concentrations in bulk water and Sh
¯ for the same
substance. Results indicate there is no significant difference among concentrations at the pipe
outlet (percentage of difference 0.00-0.21%, 0.01-0.59%, and 0.00-0.04% for maximal, median and
¯ for chloroform is very dissimilar among the three values of s tested.
average values). However, Sh
Percentage of difference ranged between 38.14-135.31% for consecutive pair of Sh
¯ values and it
was 187% for s values of 0.016 and 1.000. Such difference is due to calculation of Sh
¯ (Equation
(2-11)), which relies on concentration gradients of dissolved substances at the biofilm surface; then,

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 199
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
changes of diffusion coefficient of these substances within the biofilm in relation to bulk water
coefficients leads to different gradients for each s tested.

(a)

(b)
Figure 5-11. Influence of S on (a) chloroform concentrations in bulk water at the pipe outlet and (b)
Sh
¯

Figure 5-12 shows the DCAN concentrations in bulk water and the Sh
¯ for the same substance.
Similarly to chloroform, results for DCAN evidence there is no difference among concentrations at
the pipe outlet (percentage of difference 0.00-0.32%, 0.24-0.56%, and 0.00-0.04% for maximal,
median and average values). However, Sh
¯ for DCAN is very dissimilar among the three values of
s tested. Percentage of difference ranged between 38.17-135.29% for consecutive pair of Sh
¯
values and it was 187% for s values of 0.016 and 1.000. As previously mentioned, there is not
information regarding experimental studies to calculate the effective diffusion coefficients of
chloroform and DCAN within biofilms. Therefore, s=0.5 was arbitrarily chosen as an intermediate
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 200
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
value within the possible range. This parameter was included in the sensitivity analysis and it was
possible to determine its degree of influence on the response variables (see Section 5.3.7).

(a)

(b)
Figure 5-12. Influence of S on (a) DCAN concentrations in bulk water at the pipe outlet and (b) Sh
¯

5.2.9 Mass transport

Mass transport of dissolved substances chlorine and DBPs was characterized by the local Sh
¯ (Equation (2-11)), and ̅̅̅
(Equation (2-9)), average Sh 𝑘𝑓 (Equation (2-11)).

5.2.10 Sensitivity analysis

The aim of a sensitivity analysis in a model is to establish what factors need better determination,
and to identify the weak links of the assessment chain (i.e those that propagate most variance in

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 201
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
the output) (Saltelli et al., 2004). For the particular model presented here, carrying out a sensitivity
analysis is important because the model is based on microbiological data, which are not frequently
available for full scale DWDNs. Therefore, to determine which parameters influence most the
outputs is crucial for offering recommendations on further research and on where the efforts on
field work and laboratory assessment must be put in. According to Menberg et al. (2016), there are
three distinctive sensitivity analysis methods: i) Morris method for parameter screening (low
computational cost), ii) linear regression analysis (medium computational cost), and iii) Sobol
method (high computational cost).

The results of these three methods applied to an energy building model were compared and
researchers concluded that, by using the median value for measuring the parameter influence,
Morris method yielded robust results for evaluations with small sample size and comparable to the
other two method results (Menberg et al., 2016). For instance, researchers used 10 independent
evaluations and 10 trajectories and found the same information regarding the differentiation of
influential and negligible parameters (Menberg et al., 2016). To investigate higher-order effects and
parameter interactions, Menberg et al. (2016) used 50-150 trajectories and found that correlation
of elementary effects (EEs) and parameter values in Morris method can also provide basic
information about parameter interactions.

Taking into account that the proposed model here has a high computational cost (1.5-3.0 hours per
each run), Morris method was chosen to perform a sensitivity analysis in order to assess the level
of influence of each model parameter and define basic interactions between parameters. This was
done based on the same approach of Menberg et al. (2016); median value of elementary effect
measurement was used. The application of the method is presented in the following section.

5.2.10.1 Parameter screening with Morris method

Morris (1991) introduced a method for parameter screening in combination with a factorial sampling
strategy in order to identify parameters that can be fixed at any value within their range without
affecting the variance of the model outcome. Morris’s method is based on the model output
represented by a mathematical function Y(Z) with Y as a vector of one model output. The outputs
for the current analysis were the average of the average DCAN concentration at the pipe outlet and
Sh
¯ for chlorine and DCAN. Z is a N*m matrix of model inputs (Z) with N samples of m parameters
defined within the lower and upper bounds for each parameter Z min and Zmax, respectively (Menberg
et al., 2016).

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 202
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
For sampling, the parameter space (Zmin - Zmax) is discretized by transforming the input parameters
into dimensionless variables in the interval (0;1) and dividing each parameter interval into a number
of p levels, which form a regular grid in the unit-length hypercube Hm (Menberg et al., 2016). The
first sample is randomly chosen and each sample differs only in one coordinate from the preceding
one (Morris, 1991). A sequence of m+1 points is called a trajectory; in each trajectory, parameter
changes only once by a pre-defined value i. One point in a trajectory represents one evaluation
run of the model. The magnitude of variation in the model output due to the pre-defined variation
of one parameter Z is called elementary effect (EE) (Equation (5-35)) (Morris, 1991).

𝑌(𝑍 + 𝑒𝑖 ∆𝑖 ) − 𝑌(𝑍)
𝐸𝐸𝑖 = (5-35)
∆𝑖

Here ei is a vector of zeros, except for the i-th component that equals ±1 and represents an
incremental change in parameter i (Garcia Sanchez et al., 2014). While one trajectory allows the
evaluation of one elementary effect for each parameter i, a set of t trajectories enables statistical
evaluation of the finite distribution of the EEs. Absolute average (μ*) (Equation (5-36)) and standard
deviation (SD) (Equation (5-37)) are the statistical measures commonly used to assess the EEs.

𝜇𝑖∗ = 0.5 ∑|𝐸𝐸𝑖𝑡 | (5-36)

𝑡=1

𝑟
1
𝑆𝐷 = √ ∑(𝐸𝐸𝑖𝑡 − 𝜇𝑖 )2 (5-37)
(𝑟 − 1)
𝑡=1

Here r is the number of random trajectories (with index t), and t is a set of multiple trajectories. The
absolute average (μ*) indicates the magnitude of influence of a parameter on the model outcome
and is often used to rank the parameters according to their importance. Menberg et al. (2016), by
10 independent evaluations of the Morris method each with 10 trajectories, found that using μ* can
result in inconsistent ranking of parameters influence, due to some bias by the occurrence or
absence of outliers and the resulting skewness in the EE distributions. In order to overcome this
issue, the researchers proposed the application of the absolute median *, because this measure
is less influenced by the type of distribution of the EE and by outliers, then negligible parameters
can be identified (Menberg et al., 2016).

The SD is a measure for the spread in the model outcome due to changes in a specific parameter.
It indicates that the magnitude of influence of a parameter is dependent on the values of the
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 203
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
remaining parameters, and can be interpreted as a measure for non-linearity and parameter
interactions (Morris, 1991). The computational cost of the Morris method depends on number of
parameters m and number of t trajectories and is given by t*(m+1).

5.2.10.2 Higher-order parameter interaction effects

Menberg et al. (2016), by applying an alternative strategy, investigated the interaction effect
between two parameters by utilising the results generated from the Morris method. The strategy
consisted on evaluating the correlation coefficient of the first parameter (its value) with the EE of
the other parameter as an indication of second-order effects identified (Menberg et al., 2016).
Despite of this technique is relatively easy, large number of trajectories are required in order to
interpret the correlations meaningfully. The study of Menberg et al. (2016) used the input parameter
matrices and their corresponding EEs obtained from sensitivity analysis runs ranging from 50 to
150 trajectories to compute the corresponding Pearson’s correlation coefficients. For each set of
trajectories, correlation coefficients were calculated between the EEs for each parameter i and the
corresponding parameter values resulting in a matrix of correlation coefficient for each sensitivity
run (Menberg et al., 2016). Then, coefficients with p-value lower than 0.005 were selected to
calculate the sum of absolute correlation coefficients for each parameter. Ranking the parameters
according to such sum resulted in identical to the ranking based on Sobol indices (Menberg et al.,
2016).

5.2.10.3 Implementation of Morris method

The Morris method was applied by following the same procedure adopted by Menberg et al. (2016),
i.e., * instead of μ* for screening parameters; 10 trajectories and 12 parameters, which resulted
in 130 simulations, corresponding to 33 days of computational time. All simulations were run for
t=3 hours. Higher order interactions between parameters were not evaluated due to the
computational cost. For instance, Menberg et al. (2016) used 50-150 trajectories, which represents
for the current model a computational time of 108 days and 325 days. The method was
implemented in MATLAB 2015b, using the adapted version of toolbox SAFE developed by Pianosi
et al. (2015). Simulations of sensitivity analysis were run by using the tool “Batch sweep” in
COMSOL Multiphysics 5.2a. Influence of parameter variation was assessed for the response
variables Sh
¯ for chlorine, Sh
¯ for DCAN, and average and median value of average DCAN
concentration at the pipe outlet. Average and median concentrations of DCAN were chosen
because the typical behaviour of those curves are characterized by increasing values, reaching a
maximal value and then decreasing to almost negligible concentrations. Then, average
concentration at the pipe outlet may be affected by the maximal value in each run; then, it is
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 204
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
important to consider the peak concentrations effects. The ranges of parameters included in the
sensitivity analysis are shown in Table 5-21, which are the same ranges used in the parametric
sensitivity analysis applied in Chapter 4, Section 4.3.3.

Table 5-21. Ranges of parameters considered for sensitivity analysis

Range s Xo Eo Clo Fx FE
Units  mg/cm2 mg/L mg/L  
Min 0.016 6.65x10-6 935 0.12 0.00 0.00
Max 1.00 1.33x10-1 9465 2.00 1.00 1.00
Range Y1 Y2 k1 k2 Q BT
Units g/g g/g L/mg-s L/mg-s L/h m
Min 0.19 54 1.1x10-4 3.7x10-7 600 7
Max 3.70 1080 2.2x10-3 7.4x10-6 1100 102

5.2.11 Post-processing

The results processing included the description of the typical behaviour of the model by explaining
the flow field (contour plots of velocity and pressure magnitude in the pipe and velocity profiles);
the contour plots of concentrations of substances in bulk water and biofilm; fluxes at the biofilm
surface; and curves of concentrations of dissolved substances at the pipe outlet. The transport of
dissolved substances was characterized by the average global Sh
¯ for chlorine and DBPs and
calculation of mass transfer rates.

5.3 RESULTS AND DISCUSSION

5.3.1 Typical model behaviour

5.3.1.1 Flow field

The flow field of the simulations of the 2D model is generally explained by the comparison of
laminar, transitional and turbulent flow in Figure 5-13. The pressure is higher near to the pipe inlet
and decreased until zero at the outlet, according to the boundary conditions. For transitional flow,
pressure near to the inlet is around 0.6 Pa and it reaches the maximal value when the transitional
flow is fully developed. As expected for higher flow rates, maximum pressure magnitude is higher
in turbulent flow (20.5 Pa), followed by transitional flow (1.3 Pa) and laminar flow (0.14 Pa).

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 205
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a) Q = 260 L/h; Re = 1,202

(b) Q = 1080 L/h; Re = 4,994

(c) Q = 5425 L/h; Re = 25,084

Figure 5-13. Pressure (Pa) surface plot – Inlet: left; Outlet: right (a) Laminar flow (b) Transitional
flow (c) Turbulent flow

Velocity profiles of three flow regimes simulated here are observed in Figure 5-14. With regards to
laminar flow, velocity profiles are uniform along the pipe (Figure 5-14a), which is characteristic of
this flow regime. For transitional (Figure 5-14b) and turbulent flow (Figure 5-14c), the velocity
magnitude reached a maximum near to the inlet, which has a boundary condition defined by a
parabolic profile. Once the flow is developed along the pipe, maximal velocities in the centre of the
pipe are reduced, then the velocity profiles are more similar.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 206
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)

(c)
Figure 5-14. Velocity profiles (a) Laminar flow (b) Transitional flow (c) Turbulent flow

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 207
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.1.2 Concentrations of substances in biofilm and bulk water

Considering the current model corresponds to the simulation of the formation potentials of DBPs
from chlorination of biofilms, all the simulations were run until concentrations of DBPs at the pipe
outlet were almost zero. Due to the disproportion between pipe radius and biofilm thickness, results
are presented in non-scaled plots for better explanation. To ease the results interpretation, Figure
5-15 shows a sketch on the non-scaled geometry and its configuration.

Figure 5-15. Sketch of the non-scaled geometry

The typical behaviour of dissolved and particulate substances in the biofilm corresponds to
transitional flow regime (Re = 2774) and is presented in Figure 5-16. Chlorine penetration occurred
in the first seconds; at short time as 0.1 min (6 sec), chlorine within biofilm had already reached
half of the inlet concentration and DCAN was already formed. Higher concentrations of DCAN were
at the bottom of the biofilm most likely due to diffusion occurring in all directions while transport to
the bulk occurs at the biofilm surface, then DCAN formed near to this boundary was being
evacuated first. At t=0.1 min, no significant decrease of cells and EPS was observed in Figure 5-16.

At t=1 min, chlorine in biofilm was already reaching the inlet concentration in the section near to
the inlet. It is important to mention that chlorine first penetrated the biofilm thickness, then continues
filling this region in the flow direction (z). Cells and EPS already showed a slight decrease and
DCAN concentration reached the maximum value. At t=5 min, a notable reduction of cells and EPS
was observed, chlorine concentration was higher within the biofilm and DCAN concentration still
exhibited the highest value. At t=15 min, chlorine concentration within biofilm was equal to the

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 208
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
concentration at the pipe inlet, there were negligible concentrations of cells and EPS, and DCAN
was already disappearing from the biofilm. From t=30 min and until the end of the simulations (1
hour), chlorine reached the maximum concentration possible and there were negligible
concentrations of cells, EPS, and DCAN.

As described in Section 4.3.2, experimental data reported by Chen and Stewart (1996) indicated
that complete penetration of a biofilm as thick as 428 m occurred after t=60 min, with flow rate of
450 L/min. Then, it is reasonable that complete penetration of biofilm by chlorine with thickness of
102 m and higher flow rate than 450 L/min takes place in a shorter time (e.g., 15 min). These
results indicate that the current model is able to represent the general pattern of the reactions
between chlorine and biofilm for the subsequent formation of DBPs. Reactions occur fast, in the
first 5 min, when chlorine penetrates the thickness of the biofilm, reacting with cells and EPS, which
are transformed into DCAN.

With regards to dissolved substances in bulk water, Figure 5-17 shows the typical behaviour of
chlorine and DCAN in this region. At t=0.1 min, chlorine was completely mixed in the bulk and
concentration was equal to the inlet concentration, as specified in the initial conditions. DCAN was
absent in the bulk at this time. After 1 min, consumption of chlorine near to the biofilm surface was
observed and DCAN transport was also occurring. Chlorine concentration reached the lowest
concentration at the pipe outlet at t=1.67 min (Figure 5-20a) and DCAN reached the highest
concentration at the pipe outlet at t=4.3 min (Figure 5-20b). However, the surface plot of the bulk
at this time did not show any evident differences in relation to these substances.

For t=5 min, while chlorine decreased in bulk water near to the biofilm surface, DCAN increased in
bulk water. Then, at t=15 min, chlorine was again filling the bulk domain and DCAN concentration
was decreasing because it was leaving the pipe. It is important to mention that cells and EPS were
completely depleted within the biofilm (Figure 5-16) at t=15 min, but there was still DCAN within
this region. Then, it is reasonable to still find DCAN in the bulk water, near to the biofilm surface at
t=15 min and 30 min. For times of 45 min and 60 min, DCAN was completely evacuated from the
pipe and chlorine concentration is equal to that specified at the inlet.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 209
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 5-16. Contour plots of substance concentrations in biofilm – BT = 102 m, Clo = 1.66 mg/L, Xo = 2.33 mg/L , Re = 2774,  = 3”

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 210
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Figure 5-17. Contour plots of concentrations of dissolved substances in bulk water - BT = 102 m, Clo = 1.66 mg/L, Xo = 2.33 mg/L , Re = 2774,  = 3”

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 211
Disinfection by-products formation from chlorination of biofilms in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.1.3 Fluxes of dissolved substances at biofilm surface

Figure 5-18 presents the variation of the DCAN (a) and chlorine flux (b) at the biofilm surface. Since
chlorine was being transported from bulk water to the biofilm, fluxes were positive; the opposite
occurred for DCAN because this substance was being transported from biofilm to bulk water.
Taking into account that chlorine decay occurred faster, highest variation of its flux took place in
terms of seconds. In contrast, DCAN flux variation occurred place in terms of minutes. It is also
important to note that absolute values of fluxes resulted in higher values for DCAN in comparison
to chlorine. This is also reflected in the Sh
¯ for both substances; Sh
¯ of chlorine was lower than Sh
¯
of DCAN (0.13962 vs 0.23260). Finally, it is also worth noting that the flux curves exhibited a
curvilinear shape at the beginning of the biofilm (in the flow direction), when reactions chlorine-
biomass were occurring at the beginning of the reactive length of the pipe.

(a)

(b)
Figure 5-18. Flux of dissolved substances at the biofilm surface (a) DCAN (b) Chlorine
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 212
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.1.4 Variation of Sh
¯ at the biofilm surface

Variation of Sh
¯ of chlorine and DCAN is included in Figure 5-19. Sh
¯ was high at the beginning of
the reactions between chlorine and biomass. Then, Sh
¯ rapidly decreased and became steady after
few seconds (around 86 seconds) for chlorine and few minutes (around 5 minutes) for DCAN.

Figure 5-19. Variation of Sh

¯ at the pipe outlet

5.3.1.5 Concentrations of dissolved substances at the pipe outlet

Variation of concentrations of DCAN and chlorine at the pipe outlet is included in Figure 5-20. As
chlorine concentrations were decreasing, DCAN was increasing. Minimal and maximal values of
chlorine and DCAN, respectively were reached at different times. Chlorine decayed faster in
relation to the formation of DCAN. However, such decay was almost negligible for this case. The
chlorine demand was 1.3x10-3 mg/L and maximum DCAN concentration was 0.34 g/L, for
reactions occurring in a pipe with length of 1 m.

Figure 5-20. Average concentration of chlorine and DCAN at the pipe outlet
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 213
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.2 Influence of flow regime on DBP formation and transport

To understand how flow regime influences DBP formation and transport at the biofilm surface,
results from laminar, transitional and turbulent flow simulations were assessed under the same
initial conditions of chlorine concentration, cell density, and biofilm thickness. This analysis was
based on the average concentration of chloroform and DCAN at the pipe outlet and Sh
¯ of these
three substances. Average concentrations of DBPs at the pipe outlet are presented in Figure 5-21
and average Sh
¯ of chlorine and DBPs are included in Figure 5-22.

(a)

(b)
Figure 5-21. Average concentration of chloroform at the pipe outlet according to flow regime (a)
Chloroform (b) DCAN

(a) (b)
Figure 5-22. Variation of local Sh according to flow regime (a) DBPs (b) Chlorine
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 214
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Under the three flow regimes, concentrations of chloroform and DCAN were decreasing as Re
increased, due to dilution effect caused by the increased of the flow rate. Similarly, DBP transport
occurred faster under turbulent flow conditions, followed by transitional flow and laminar flow. This
is observed by the narrowing of the concentration curves as Re increased (Figure 5-21). This is
also reflected in the values of Sherwood number, which were higher for the three dissolved
substances simulated for higher Re (Figure 5-22).

Although flow in drinking water pipes is frequently turbulent (Cogan, 2010), it is important to show
the differences among results from three flow regimes, as laminar flow is commonly assumed in
biofilm modelling, as explained in Section 2.4.3 (Table 2-3, Chapter 2), and transitional flow can be
found in dead-end sections of DWDNs (Abokifa et al., 2016b). Depending on the biofilm
characteristics that a modeller wants to represent, biofilm modelling is computationally expensive
and improved hardware is required to efficiently carry out this task. Therefore, researchers have
decided to assume laminar flow under low Re (0.32x10-5 - 150) to simplify the calculation of the
flow next to the biofilm surface (Eberl et al., 2000; Picioreanu et al., 2000a; Eberl and Sudarsan,
2008; Zhang et al., 2008; Duddu et al., 2009; Cogan, 2010; Lindley et al., 2012; Taherzadeh et al.,
2012; Zhang, 2012; Cumsille et al., 2014; Tierra et al., 2015).

Most of the previous cited studies focused on substrate transport towards the biofilm to model its
growth and influence on its morphology. Particularly, Cogan (2010); Cogan (2011) and Zhang
(2012) simulated the disinfection of biofilms under laminar flow to analyse the behaviour of
susceptible and persistent bacteria and the disinfection effectiveness by simulating active and dead
cells. Eberl and Sudarsan (2008) presented a case similar to the one presented here; they modelled
the disinfection effects on multiple colonies growing in a narrow channel, under slow flow. These
authors found that hydrodynamics played an important role for the transport of substrate (dissolved
oxygen); increasing the flow velocity led to faster a more stable growth. In the case of the
disinfectant, the colonies located near to the inlet of the channel took the hardest hit, biomass was
quickly inactivated and limited in the downstream region (Eberl and Sudarsan, 2008). The model
developed here showed that, under continuous chlorine supply and transitional and turbulent flow,
pipe was refilled with chlorine after few minutes ( 15 min, Section 5.3.1.2).

For the case of the formation potentials of DBPs from chlorination of biofilms, it is important to
consider the real flow conditions found in drinking water pipes due to hydraulic variables such as
water demand, velocity, and pressure are not constant along the day. Therefore, transport
parameters such as Sherwood number and mass transfer rate can be affected by such variation.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 215
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.3 Influence of Re on DBP transport

Once it was defined that the flow regime actually affects the transport of DBP potentially formed in
drinking water pipes, several Re corresponding to transitional and turbulent flow were simulated to
analyse in more detail the likely influence of flow rate on DBP formation and transport. Chlorine,
chloroform and DCAN were assessed for biofilm thicknesses of 7 m and 102 m.

5.3.3.1 Chloroform and dichloroacetonitrile formation

Average concentrations of chloroform, DCAN and chlorine at the pipe outlet are shown in
Figure 5-23, Figure 5-24, and Figure 5-25, respectively, for ten values of Re, corresponding to
transitional flow. Similarly for turbulent flow, average concentrations of chloroform, DCAN and
chlorine at the pipe outlet are shown in Figure 5-26, Figure 5-27, and Figure 5-28, respectively, for
ten values of Re. Increasing Re was simulated by increasing velocity and keeping the same pipe
diameter (3 inches).

(a)

(b)
Figure 5-23. Average chloroform concentration at the pipe outlet for several Re – Transitional flow
(a) BT = 7 m (b) BT = 102 m
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 216
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 5-24. Average DCAN concentration at the pipe outlet for several Re – Transitional flow (a)
BT = 7 m (b) BT = 102 m

(a)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 217
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-25. Average chlorine concentration at the pipe outlet for several Re – Transitional flow (a)
BT = 7 m (b) BT = 102 m

(a)

(b)
Figure 5-26. Average chloroform concentration at the pipe outlet for several Re – Turbulent flow (a)
BT = 7 m (b) BT = 102 m
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 218
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 5-27. Average DCAN concentration at the pipe outlet for several Re – Turbulent flow (a) BT =
7 m (b) BT = 102 m

(a)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 219
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-28. Average chlorine concentration at the pipe outlet for several Re – Turbulent flow (a)
BT = 7 m (b) BT = 102 m

As it is expected, DBP concentrations were decreasing as Re increased, due to dilution effect

caused by the increase of the flow rate (Figure 5-23, Figure 5-24, Figure 5-26, and Figure 5-27).
Chlorine demand increased with higher Re, which is probably associated with faster mass transport
towards the biofilm (Figure 5-25 and Figure 5-28). This is also observed in the shape of the
concentration curves. Minimum chlorine concentrations occurred faster as Re increases.
Figure 5-23, Figure 5-24, Figure 5-26, and Figure 5-27 also allows observing the influence of
increase of Re on mass transport of DBPs. Curves of chloroform and DCAN concentrations are
slightly closer to the vertical axis of the plots and are narrower as Re rises, which apparently reflects
the faster evacuation of both DBP species from the pipe, when convective transport became more
important in relation to diffusion transport. For instance, for BT=102 m, DCAN concentration of
0.05 g/L is obtained at t=15 min for Re=2774, while the same concentration is found at t=11 min
for Re=4994.

Finally, concentrations of both DBP species were higher for BT=102 m in comparison with BT=7
m, as expected when more biomass is available to react with chlorine. This is in agreement with
the structure of the model, since higher biofilm thickness represents higher biomass available for
reaction with chlorine, when the pipe is continuously fed with disinfectant. However, this has not
been tested yet by experimental studies, to the author’s knowledge. Published research studied
other variables influencing the DBP formation from biomass disinfection such as pH, temperature,
reaction time, disinfectant dose, pipe material, bacteria species, and type of biomass organic
matter; then concentration of DBP precursors were maintained constant during the experiments

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 220
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
(Fang et al., 2010a; Fang et al., 2010b; Wang et al., 2012c; Pu et al., 2013; Wang et al., 2013a;
Wang et al., 2013b). On the other hand, the current model did not consider the protective role of
EPS for cells against disinfection. Such protective role consists on increased chlorine demand by
reaction with EPS and dissolved organic matter, and more compact structure at the bottom of the
biofilm. This could reduce the availability of disinfectant in the deeper layers and cells located in
this region are able to survive (Xue et al., 2012; Xue et al., 2013).

In this line, DBP risk assessment associated to biomass disinfection must be analysed carefully,
since the simplifications of the current model may lead to overestimation of the risk, considering
that the model presented here simulate the DBP formation potentials. Avoiding EPS protective role
leads to complete oxidation of cells by chlorine; this means that biofilm would not exist in real scale
DWDNs, since disinfectant would oxidize every cell produced. However, this is not true; biofilms do
grow in drinking water pipes and are more diverse than bacterial communities in bulk water, as
discussed in Chapter 3. The current model used reaction terms associated to EPS according to
experiments carried out by Chen and Stewart (1996), who used agarose as surrogate of EPS
produced by P. aeruginosa. However, biofilms in DWDNs are heterogeneous and EPS may be
diverse as well (Lemus Pérez and Rodríguez Susa, 2017). Then, chlorine demand exerted by EPS
may be different than the calculated by Chen and Stewart (1996). It is important to highlight that
factors Fx and FE corresponds to experimental tests carried out with extracted and suspended
intracellular and extracellular organic matter from algae, by separate (Fang et al., 2010b); this
differed from the real physical configurations of biofilms. To the author’s knowledge, work by Wang
et al. (2013a) is the only published study which includes DBP formation from biofilm chlorination,
under the natural conditions (i.e., biofilm growing attached to a surface).

Despite of this, thicker biofilms, either related to higher cell density or EPS concentration, may
increase both microbial and chemical risks. This means that increased biomass is available for
reaction with the disinfectant and major bulk water concentrations of DBPs may be expected.
Additionally, the secreted EPS protect the cells against the action of the disinfectant, then allowing
them to grow and reproduce. If pathogenic microorganisms were present in biofilms, these can
survive within this habitat and be released to bulk water if the proper hydraulics conditions were
present. Then, they could reach the consumers, which represents an important microbial risk
depending on the pathogen dose, level of exposure and susceptibility of the consumer. Water
operators must pay special attention to dead-end branches, where slow flow is frequent (Abokifa
et al., 2016b), since thicker biofilms may be present there due to limited transport of nutrients
(Picioreanu et al., 1998b; Jayathilake et al., 2017). Therefore, it is necessary to design and apply

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 221
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
a plan for O&M of DWDNs, which includes cleaning of pipes, flushing dead-end sections, and
opportune replacement of old pipelines.

5.3.3.2 Mass transport

¯ and ̅̅̅
In order to examine transport of dissolved substances in more detail, Sh 𝑘𝑓 for chloroform,
DCAN and chlorine were analysed. Results are presented in Figure 5-29, Figure 5-30, and Figure
¯ and ̅̅̅
5-31, respectively, for transitional flow. For turbulent flow, Sh 𝑘𝑓 for chloroform, DCAN and
chlorine are shown in Figure 5-32, Figure 5-33, and Figure 5-34. The behaviour of both parameters
is the same for the three dissolved substances analysed here, for both flow regimes: Sh
¯ and
̅̅̅
𝑘𝑓 increases as Re rises.

(a) (b)
Figure 5-29. Sh
¯ (a) and mass transfer rate (b) of chloroform for several Re – Transitional flow

(a) (b)

Figure 5-30. Sh
¯ (a) and mass transfer rate (b) of chloroform for several Re – Turbulent flow

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 222
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a) (b)
Figure 5-31. Sh
¯ (a) and mass transfer rate (b) of DCAN for several Re – Transitional flow

(a) (b)
Figure 5-32. Sh
¯ (a) and mass transfer rate (b) of DCAN for several Re – Turbulent flow

(a) (b)
Figure 5-33. Sh
¯ (a) and mass transfer rate (b) of chlorine for several Re – Transitional flow

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 223
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a) (b)
Figure 5-34. Sh
¯ (a) and mass transfer rate (b) of chlorine for several Re – Turbulent flow

¯ was notably higher for BT=102 m than for BT=7 m (Figures 5-29a – 5-34a),
Particularly, Sh
which is logic since the parameter BT was included in the numerator of Equation (2-9) as the
characteristic length. This also corresponds to the nature of the model, considering that higher
biofilm thickness included more biomass available for reaction with chlorine, then chlorine demand
and transport of chlorine and DBPs increased through the biofilm surface boundary.

Simulated results suggest that the difference among BT was compensated in the calculation of ̅̅̅
𝑘𝑓
(Equation (2-11)). Therefore, the difference between ̅̅̅
𝑘𝑓 for both BT values and for the three
dissolved substances was significantly reduced, and such difference reduced even more when Re
increased (Figures 5-29b – 5-34b). For turbulent flow, ̅̅̅
𝑘𝑓 was similar for both BT values analysed
here. Percentage of difference of ̅̅̅
𝑘𝑓 for chlorine and DBPs reduced from 7-12% for Re=2774 to
0.05-0.5% for Re=50,030. Simulated values of ̅̅̅
𝑘𝑓 were in the order of 10-6-10-5 m/s and always
exhibited and increasing pattern. These results are not surprising since higher velocities (i.e., higher
¯ and ̅̅̅
Re) correlates with thinner boundary layers, then higher Sh 𝑘𝑓 could be obtained (IWA Task
Group on Biofilm Modeling et al., 2006; Taherzadeh et al., 2012).

Another important result of this study is the behaviour of Sh

¯ . As explained in Section 2.3.2.2,
several empirical correlations have been proposed for Sh
¯ under laminar and turbulent flow, which
represents a polynomial relationship with Re and Sc. The behaviour of Sh
¯ found with the current
model was far from such approach. The relationship between Sh
¯ and Re was clearly linear, which
may be related to the assumption of a flat biofilm surface. It is known that irregular biofilm surface
¯ and ̅̅̅
results in non-uniform boundary layer thickness, then different values of Sh 𝑘𝑓 can be obtained
along each point of the biofilm surface (Picioreanu et al., 1998b). Taherzadeh et al. (2012) also

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 224
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
found a linear relationship between Sh
¯ and Re for a single cell with a moving tail, under laminar
flow conditions (Re up to 150).

5.3.4 Influence of S/V ratio on DBP transport

5.3.4.1 Different flow rates and equal Re

Results from Chapter 3 explained how S/V ratio influenced the DBP formation under stagnation
conditions. A similar analysis is included here for chloroform and DCAN, under transitional flow,
and three different pipe diameters (3, 6, and 10 inches), keeping the same Re in each pipe. Figure
5-35 presents the average concentrations of chloroform and DCAN at the pipe outlet. Figure 5-36
shows the average concentrations of chlorine at the pipe outlet. Figure 5-37 shows Sh
¯ for the three
dissolved substances simulated. As expected, higher concentrations of chloroform and DCAN and
higher chlorine demand were found in smaller pipe diameters.

For transitional flow, maximal concentration of chloroform increased from 1.17x10 -2 g/L for =10”
to 4.06x10-2 g/L for =3” (Figure 5-35a). Minimal chlorine decreased from 1.65997 mg/L for =10”
to 1.65988 g/L for =3” (Figure 5-35b). The increased DBP concentrations found in smaller
diameters is in agreement with the findings of other researchers in relation to the increase of
chlorine demand when pipe diameter decreases (Lu et al., 1999; Buamah et al., 2014; Lee et al.,
2014). This is related to interactions bulk-wall become more significant in smaller pipe diameters.

(a)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 225
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-35. Average chloroform (a) and DCAN (b) concentrations at the pipe outlet for several pipe
diameters – Transitional flow

Figure 5-36. Average chlorine concentrations at the pipe outlet for several pipe diameters –
Transitional flow

Similarly, mass transport increased in smaller pipe diameters, which is reflected in the shape of
concentration curves, since curves are narrower and closer to the vertical axis, which suggests that
substances were leaving the pipe faster (Figure 5-35). Sh
¯ could have also implied faster mass
¯ of chloroform increased from 0.0059 for =10”
transport in smaller pipes. For transitional flow, Sh
to 0.0226 for =3” (Figure 5-37a). Sh
¯ of DCAN increased from 0.0052 for =10” to 0.0199 for =3”
¯ of chlorine increased from 0.0028 for =10” to 0.0121 for =3” (Figure 5-37b).
(Figure 5-37a). Sh

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 226
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
(a)
Figure 5-37. Average Sh
¯ for DBPs (a) and Chlorine (b) for several pipe diameters – Transitional flow

5.3.4.2 Equal flow rates and different Re

A similar analysis to the previous one is included here for chloroform and DCAN, under transitional
and turbulent flow, by keeping the same flow rates and different Re in each pipe of diameters 3, 6
and 10 inches. Figure 5-38 presents the average concentrations of chloroform and DCAN at the
pipe outlet. Figure 5-39 shows the average concentrations of chlorine at the pipe outlet. Figure 5-40
shows Sh
¯ for the three dissolved substances simulated. As expected, higher concentrations of
chloroform and DCAN and higher chlorine demand were found in smaller pipe diameters.

(a)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 227
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-38. Average chloroform (a) and DCAN (b) concentrations at the pipe outlet for several pipe
diameters – Turbulent flow

Figure 5-39. Average chlorine concentrations at the pipe outlet for several pipe diameters –
Turbulent flow

(a) (b)
Figure 5-40. Average Sh
¯ for DBPs (a) and Chlorine (b) for several pipe diameters – Turbulent flow

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 228
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.5 DBP formation under several conditions of drinking water quality

5.3.5.1 Transitional flow

Two scenarios of drinking water quality were selected to test the DBP formation under those
conditions. The least critical scenario corresponded to the lower bound of the range for parameters
BT, Clo, and Xo (Table 5-22). The most critical scenario corresponded to the upper bound of the
range for parameters BT, Clo, and Xo (Table 5-22). Range of chlorine concentration was defined
by the minimal and maximal values found in the field work developed in Cali’s DWDN (Chapter 3).
Figure 5-41 presents the chloroform and DCAN concentrations at the pipe outlet and Table 5-22
¯ and ̅̅̅
includes Sh 𝑘𝑓 for the three dissolved substances analysed here for transitional flow. As it is
expected, DBP concentrations were higher under the most critical conditions in relation to the
lowest BT, Clo, and Xo. While least critical scenario exhibited maximal concentrations of 0.0037
g/L of chloroform (Figure 5-41a) and 0.0020 g/L of DCAN (Figure 5-41b), the most critical
scenario resulted in maximal concentrations of 0.58 g/L (Figure 5-41a) of chloroform and 0.34
g/L of DCAN (Figure 5-41b).

(a)

(b)
Figure 5-41. Average concentrations of (a) chloroform and (b) DCAN at the pipe outlet for two
scenarios of drinking water quality – Transitional flow
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 229
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
On the other hand, the shape of concentration curves showed that mass transport of DBP is
different in both scenarios. Both substances left the pipe in five hours under the least critical
conditions, while both substances left the pipe in only one hour under the most critical conditions,
due to gradient concentrations as explained below. Faster mass transport under the most critical
conditions of water quality was evidenced by values of Sh
¯ included in Table 5-22. Sh
¯ of chloroform,
DCAN, and chlorine increased around 171% under the most critical conditions in relation to the
least critical conditions.

¯ and ̅̅̅
Table 5-22. Sh 𝒌𝒇 for dissolved substances for two scenarios of drinking water quality –
Transitional flow
Scenario Sh
¯ - Chloroform Sh
¯ - DCAN Sh
¯ - Chlorine
Least critical scenario: BT = 7 m | Clo =
0.0212 0.0178 0.0109
0.12 mg/L | Xo = 2.82x10-4 mg/cm2
Most critical scenario: BT = 102 m | Clo =
0.2680 0.2326 0.1396
1.66 mg/L | Xo = 4.44x10-3 mg/cm2
Mass transfer Mass transfer Mass transfer
Scenario
(m/s) - Chloroform (m/s) - DCAN (m/s) - Chlorine
Least critical scenario: BT = 7 m | Clo =
2.81 x 10-6 4.23 x 10-6 4.10 x 10-6
0.12 mg/L | Xo = 2.82x10-4 mg/cm2
Most critical scenario: BT = 102 m | Clo =
2.44 x 10-6 3.79 x 10-6 3.64 x 10-6
1.66 mg/L | Xo = 4.44x10-3 mg/cm2

In contrast, despite of Sh
¯ is directly proportional to the biofilm thickness and flux at the biofilm
surface (Equation (2-9)), ̅̅̅
𝑘𝑓 resulted higher under the least critical conditions in relation to the most
critical conditions, for the three dissolved substances simulated (Table 5-22). For instance, flux of
DCAN was in the order of 10-3 - 10-4 g/m2-s under the least critical conditions, while such flux was
in the order of 10-1 g/m2-s under the most critical conditions (Figure 5-42). The same pattern
applies for flux of chlorine; values of magnitude order of 10 -6 – 10-5 mg/m2-s resulted under the
least critical conditions, in comparison to values of magnitude order of 10 -2 – 10-3 mg/m2-s under
the most critical conditions (Figure 5-43). Then, Sh
¯ was lower when concentrations of chlorine and
DBPs were lower, which may indicates that diffusive transport is predominant over convective
transport at the biofilm surface under those conditions.

On the other hand, calculation of ̅̅̅

𝑘𝑓 (Equation (2-11)) is inversely proportional to the characteristic
𝑘𝑓 resulted lower for BT=102 m in relation to BT=7 m. Higher mass
length (BT here), then ̅̅̅
transfer rate at biofilm surface occurred in the least critical scenario due to smaller BT, which is
related to faster consumption of biomass. This was also confirmed by the time when DCAN left the
biofilm and reached very low concentrations in this domain (10-9 mol/m3); DCAN evacuated the

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 230
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
biofilm domain at t=60 min under the most critical conditions, while DCAN left the biofilm domain
at t=7 min under the least critical conditions.

(a)

(b)
Figure 5-42. Flux of DCAN at the biofilm surface for two scenarios of drinking water quality (a)
Least critical scenario (b) Most critical scenario

(a)
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 231
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-43. Flux of chlorine at the biofilm surface for two scenarios of drinking water quality –
Transitional flow (a) Least critical scenario (b) Most critical scenario

5.3.5.2 Turbulent flow

Average concentrations of chloroform and DCAN at the pipe outlet are observed in Figure 5-44, for
the least and most critical scenario regarding water quality parameters. Similarly to transitional flow,
simulations for turbulent flow indicate that concentrations of both DBP species were higher for the
most critical scenario in comparison to the least critical scenario, which is related to more biomass
available for reaction with chlorine when biofilm thickness is higher.

(a)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 232
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-44. Average concentrations of (a) chloroform and (b) DCAN at the pipe outlet for two
scenarios of drinking water quality – Turbulent flow

¯ and ̅̅̅
Table 5-23 shows the Sh 𝑘𝑓 for DBP species and chlorine corresponding to turbulent flow.
Similar to transitional flow, Sh
¯ was higher for the most critical conditions, which is related to the
higher biofilm thickness, as explained previously and suggests that convection effects increases
when higher concentrations are present in bulk water. This was also observed when comparing the
time required for DBPs to evacuate the pipe: one and five hours for the most and least critical
conditions, respectively. In relation to ̅̅̅
𝑘𝑓 , mass transfer also resulted slightly lower for higher BT
(most critical scenario), due to calculation of this parameter is inversely proportional to BT (Equation
(2-11)). This is reflecting that lower biomass is consumed faster, as explained in the previous
section.

¯ and ̅̅̅
Table 5-23. Sh 𝒌𝒇 for dissolved substances for two scenarios of drinking water quality –
Turbulent flow
Scenario Sh
¯ - Chloroform Sh
¯ - DCAN Sh
¯ - Chlorine
Least critical scenario: BT = 7 m | Clo =
0.1027 0.0901 0.0566
0.12 mg/L | Xo = 2.82x10-4 mg/cm2
Most critical scenario: BT = 102 m | Clo =
1.4733 1.2929 0.8107
1.66 mg/L | Xo = 4.44x10-3 mg/cm2
Mass transfer Mass transfer Mass transfer
Scenario
(m/s) - Chloroform (m/s) - DCAN (m/s) - Chlorine
Least critical scenario: BT = 7 m | Clo =
1.36 x 10-5 2.14 x 10-5 2.15 x 10-5
0.12 mg/L | Xo = 2.82x10-4 mg/cm2
Most critical scenario: BT = 102 m | Clo =
1.34 x 10-5 2.10 x 10-5 2.11 x 10-5
1.66 mg/L | Xo = 4.44x10-3 mg/cm2

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 233
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.6 Discussion

The current 2D model presented here predicted the formation potentials of chloroform and DCAN
from biofilm chlorination in a drinking water pipe of length equal to 1 m. The simulations were
undertaken for several scenarios in order to evaluate the influence of flow regime, Re, S/V ratio,
and drinking water quality on DBP formation and mass transport. In terms of absolute
concentrations, some extrapolations can be made with regards to UK and Colombian standards for
chloroform (100 g/L and 200 g/L, respectively) (UK Parliament, 2000; Ministerio de la Protección
Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial, 2007) and WHO guideline for
DCAN (20 g/L) (WHO, 2017). According to the maximum concentrations presented in the Section
5.3.5, for transitional flow, DCAN concentrations might be higher than 20 g/L for pipelines length
of 9.85 Km and 0.06 Km, if the least and most critical conditions uniformly prevailed along the pipes,
respectively. In the case of chloroform, lengths of 27 Km and 54 Km would be necessary to exceed
the UK (100 g/L) and Colombian (200 g/L) regulation, respectively (UK Parliament, 2000;
Ministerio de la Protección Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial,
2007), if the least critical conditions prevailed along the pipes. For the most critical conditions, 0.17
Km and 0.34 Km would be required to surpass the regulations mentioned previously, respectively.

In the case of turbulent flow, DCAN concentrations might be higher than 20 g/L for pipelines length
of 60.79 Km and 0.22 Km, if the least and most critical conditions uniformly prevailed along the
pipes, respectively. In the case of chloroform, lengths of 154 Km and 309 Km would be necessary
to exceed the UK (100 g/L) and Colombian (200 g/L) regulation, respectively (UK Parliament,
2000; Ministerio de la Protección Social and Ministerio de Ambiente Vivienda y Desarrollo
Territorial, 2007), if the least critical conditions prevailed along the pipes. For the most critical
conditions, 0.55 Km and 1.10 Km would be required to surpass the regulations mentioned
previously, respectively.

This simple analysis suggests that the higher risk is related to DCAN under transitional flow (lower
velocities). As explained in Chapters 2 and 4, the experimental evidence on cells indicates that
toxicity of nitrogenous DBPs is higher in comparison to carbonaceous DBPs (Muellner et al., 2007).
The most clear epidemiological evidence in relation to THMs, but not conclusive, is related to
bladder cancer (Hrudey et al., 2015b). Therefore, for higher biomass content and chlorine residuals,
the potential risk may be present even in small DWDNs, specially is small pipes. For bigger water
systems, even low biomass content and chlorine residuals may also represent potential health
risks. However, it is important to highlight that DCAN is an unstable DBP since it is degraded
depending on the pH and chlorine concentration. DCAN degrades in the absence of chlorine above
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 234
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
pH 7 and below pH 6.5. In the presence of free chlorine, DCAN degradation can be much faster
from a pH of about 6–8.5 under low to moderate chlorine residuals (Reckhow et al., 2001).
Therefore, higher risk may be present in building premises and in the portion of the network
supplying directly the consumers (smaller pipes), since less opportunities may occur for changing
the physico-chemical conditions of drinking water, then reducing the probability of degradation of
DCAN.

On the other hand, there is mass transport. The present results showed that mass transport is
faster for turbulent flow, higher Re, smaller pipe diameters, and higher Clo, higher BT, and higher
Xo together. Thus, a balance between control of acute and chronic risk must be considered by
water operators. Slow mass transfer favour the reduction of DBP concentrations, since chlorine is
slowly penetrating the biofilm thickness and DBP formed within it are being slowly transported to
the bulk. However, this slow mass transport may also favour the growth of microorganisms living
within the biofilm, even pathogenic microbes. If there was enough contact time between biomass
and chlorine, the disinfectant can be completely depleted and DBP concentrations can reach the
maximum potentials. Absence of chlorine would also favour the biofilm regrowth. As mentioned in
Chapter 3, the best approach for water operators is to minimize the biofilm formation, which can
host pathogenic microorganisms and may be significant DBP precursors. A deeper analysis of the
health implications of biofilm presence in drinking water pipes is presented in Chapter 6.

5.3.7 Sensitivity analysis

Sensitivity analysis was carried out based on the Morris method adapted by Menberg et al. (2016)
and calculations were performed using the MATLAB toolbox developed by Pianosi et al. (2015).
The influence of 12 parameters was evaluated for response variables Sh
¯ for chlorine, Sh
¯ for DCAN,
average of average concentrations of DCAN at the pipe outlet, and median of average
concentrations of DCAN at the pipe outlet.

5.3.7.1 Sh
¯ - chlorine

Table 5-24 shows the median and standard deviation of EE for each parameter assessed.
Parameters were sorted in descendent order according to the level of influence; i.e., from highest
to lowest median of EE. Figure 5-45 shows the points corresponding to both values of EE from
Table 5-24 (a) and the area covered by each parameter according to the lower and upper bounds
of median and standard deviation of EE (b). Then, the most influencing parameters on Sh
¯ - chlorine
were biofilm thickness and flow rate, which is a logical result since the Sherwood number is a
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 235
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
measure of the diffusion and convective transport at the biofilm surface. The influence of the
remaining parameters was very low; they are located near to the origin in Figure 5-45a.

Table 5-24. Median and standard deviation of EEs on response variable Sh

¯ - Chlorine
Standard Standard
Parameter Median of EE Parameter Median of EE
deviation of EE deviation of EE
BT 0.173 0.024 Y2 0.001 0.002
Q 0.046 0.023 k1 0.001 0.002
Xo 0.003 0.003 k2 0.001 0.002
Clo 0.003 0.003 s 0.000 0.000
Eo 0.001 0.002 Fx 0.000 0.001
Y1 0.001 0.006 FE 0.000 0.000

Convective transport depends on the hydrodynamics of the bulk flow, which is directly linked to the
flow rate. Additionally, calculation of Sh
¯ (Equation (2-9)) includes the characteristic length assumed
here as the biofilm thickness. This parameter is more influencing than flow rate; Figure 5-45b shows
that standard variation of EEs for biofilm thickness presented more variation in comparison to its
median value, while the median of EEs for flow rate varied more in comparison to its standard
variation. Therefore, flow calculation is important in modelling, then it is necessary to appropriate
define the turbulence model and flow rate.

(a)

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 236
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (b)
Figure 5-45. EEs on response variable Sh
¯ - Chlorine (a) Median vs Standard deviation (b) Median vs
Standard deviation including upper and lower bounds

5.3.7.2 Sh
¯ - DCAN

Similarly to Sh
¯ of chlorine, Table 5-25 shows the median and standard deviation of EE for each
parameter assessed for Sh
¯ of DCAN. Figure 5-46 shows the points intersecting the median and
standard deviation of EEs (a) and the area covered by each parameter according to the lower and
upper bounds of median and standard deviation of EEs (b). These results allowed identifying that
the most influencing parameters were biofilm thickness, followed by s, and Q (Table 5-25). As
explained in the previous section, BT and Q are influencing Sh
¯ of DCAN because the calculation
of Sherwood number (Equation (2-9)) includes the biofilm thickness as the characteristic length
and it is a measure of the diffusive and convective transport at the biofilm surface. In addition and
as described in Section 5.2.8, the effective diffusion coefficient within the biofilm is an important
parameter for determining the mass transport of DBPs at the biofilm surface because this model
included only diffusive transport within the biofilm. Then, the gradient concentration at this region
is determined by the diffusion of the dissolved substances within the biofilm.

Table 5-25. Median and standard deviation of EEs on response variable Sh

¯ - DCAN
Standard Standard
Parameter Median of EE Parameter Median of EE
deviation of EE deviation of EE
BT 0.343 0.307 Xo 0.004 0.010
s 0.229 0.881 k1 0.003 0.015
Q 0.074 0.144 k2 0.003 0.004
Clo 0.008 0.011 Eo 0.002 0.001
Y1 0.005 0.021 Fx 0.001 0.002
Y2 0.005 0.004 FE 0.000 0.002

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 237
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
As it is observed in Figure 5-46a, the influence of other parameters assessed is very low since they
are located near to the origin of the plot. While BT is the most influencing parameter on Sh
¯ of
DCAN, s presented the highest variation in both median and standard deviation of EEs. This
represent that this parameter must be carefully defined in case of the interest is mainly on the study
of the mass transport of DBPs at the biofilm surface.

(a)

(b)
Figure 5-46. EEs on response variable Sh
¯ - DCAN (a) Median vs Standard deviation (b) Median vs
Standard deviation including upper and lower bounds

5.3.7.3 Average DCAN

Table 5-26 presents the list of parameters sorted according to the level of influence over the
response variable average DCAN. Figure 5-47 shows the points corresponding to both values of
EE from Table 5-26 (a) and the area covered by each parameter according to the lower and upper
bounds of median and standard deviation of EE (b). For the average DCAN at the pipe outlet, the
most influencing parameters were Xo, BT and Fx. This is related to the effect caused by the
maximal concentrations on the average DCAN. When higher values of these three parameters are
present together, more DBPs are formed. Then, average concentrations are higher.
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 238
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 5-26. Median and standard deviation of EEs on response variable average DCAN
Standard Standard
Parameter Median of EE Parameter Median of EE
deviation of EE deviation of EE
Xo 0.170 0.122 Q 0.048 0.076
BT 0.136 0.264 Y2 0.040 0.105
Fx 0.092 0.284 Clo 0.005 0.189
Y1 0.075 0.109 s 0.000 0.001
FE 0.061 0.076 k1 0.000 0.046
Eo 0.055 0.078 k2 0.000 0.074

In contrast to the Sherwood number, other parameters such as yield coefficients Y 1 and Y2, Eo,
and Q are also influencing the average DCAN at the pipe outlet but in less degree than the other
three parameters mentioned previously (Figure 5-47a). Parameters k1, k2, and s have no influence
on the average DCAN at the pipe outlet (Figure 5-47a). Xo was the most influencing parameter but
variation of average and standard variation of EEs is low compared to the parameters BT and Fx.
Particularly BT exhibited higher variation of the median of EEs, while Fx exhibited higher variation
of the standard variation of EEs (Figure 5-47b).

(a)

(b)
Figure 5-47. EEs on response variable average DCAN (a) Median vs Standard deviation (b) Median
vs Standard deviation including upper and lower bounds
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 239
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
5.3.7.4 Median DCAN

Table 5-27 presents the list of parameters sorted on descending order of influence over the
response variable median DCAN. Figure 5-48 presents the intersection points of median and
standard variation of EEs for the same response variable and the area covered by the upper and
lower bounds or such values. For the case of the median DCAN at the pipe outlet, the most
influencing parameter is Clo, followed by k 1, BT, Y1, Xo, and Fx. This represents that all the
parameters involved in the reaction with cells are the most influencing ones over the median of
DCAN at the pipe outlet. This means that the influence over the median DCAN is distributed among
more parameters, which is the result of eliminating the effect of outliers over the measures of central
tendency as the median value. The parameters associated to EPS (i.e Eo and F E) and s have little
or no influence on the median DCAN.

Table 5-27. Median and standard deviation of EEs on response variable median DCAN
Standard Median of Standard
Parameter Median of EE Parameter
deviation of EE EE deviation of EE
Clo 0.039 0.185 k2 0.007 0.033
k1 0.034 0.086 Q 0.006 0.084
BT 0.023 0.371 Y2 0.004 0.121
Y1 0.022 0.115 Eo 0.002 0.090
Xo 0.015 0.133 s 0.000 0.000
Fx 0.012 0.269 FE 0.000 0.079

The spread of parameters influencing the median DCAN is more clear in Figure 5-48a. While BT is
the third parameter most influencing the median DCAN, its variation of median and standard
variation of EEs is higher than the other two parameters ranked in the two first places. Since higher
BT represent more biomass available for reaction with chlorine and further transformation into
DBPs, it is logical to find that variations on this parameter led to variations on the median
concentrations of DCAN at the pipe outlet.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 240
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 (a)

(b)
Figure 5-48. EEs on response variable median DCAN (a) Median vs Standard deviation (b) Median
vs Standard deviation including upper and lower bounds

5.3.7.5 Summary

Table 5-28 presents the ranking of the parameters included in the sensitivity analysis according to
their influence on response variables. In general, BT resulted the most important parameter for the
characterization of the mass transport of DBPs through the biofilm surface and their concentrations
in bulk water. Other important parameters are Xo, Clo, Y 1, Q, and k1. Parameters related to EPS
(Eo, k2, FE, Y2) exhibited medium and low influence on the response variables. This may be related
to the current model assumed that cells contributed more to DBP concentrations, EPS and cells
were completely mixed within the biofilm and the protective role of EPS to cells against disinfection
was not considered. Refinement of the model should include this feature of EPS to model how
chlorine reacts first with it, then reducing its availability for disinfection in the deeper layers of the
biofilm. Q is globally representing medium influence of response variables, but especially important
for the mass transport through the biofilm surface; similarly, s is also very important for this.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 241
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Experimental calculation of s involves the development of micro-sensors to measure
concentrations of the target DBPs within the biofilm.

Table 5-28. Ranking of parameters according to their influence on response variables

Parameter ranking
Parameter
Sh – Chlorine Sh – DCAN Average DCAN Median DCAN
BT 1 1 2 3
Q 2 3 7 8
Xo 3 7 1 5
Clo 4 4 9 1
Eo 5 10 6 10
Y1 6 5 4 4
Y2 7 6 8 9
k1 8 8 11 2
k2 9 9 12 7
s 10 2 10 11
Fx 11 11 3 6
Fe 12 12 5 12

As previously mentioned and discussed in Chapter 3, more efforts must be made in order to
improve laboratory techniques to properly monitor microbiological parameters of biofilm such as
thickness and cell density in real scale DWDNs. This will contribute to improve the estimation of
DBP formation potentials from biofilm chlorination in water pipes and increase the availability of
data for further calculations of microbial risks related to biofilms. Data of flow rates and chlorine
concentrations are available from historic records of the water utilities or can be directly measured
in the DWDN.

5.4 MODEL APPLICATIONS AND LIMITATIONS

The 2D model developed here presents several features that contribute to new knowledge in the
field of biofilm–DBP modelling in drinking water. This model was formulated for prediction of
formation potentials of two DBP species (chloroform and DCAN), under transitional and turbulent
flow, which are the flow regimes commonly found in drinking water pipes. The flow field was
calculated and the turbulent model SST was selected to solve the flow near to the pipe wall. This
allowed calculating the local and global Sh at the biofilm surface boundary; then, mass transport
rates were also available for the scenarios considered in this study.

Mass transport characterization is important in the current model since the interest of researchers
on biomass (cells and EPS) as DBP precursors is increasing from the point of view of experimental

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 242
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
works (Hong et al., 2008; Fang et al., 2010a; Fang et al., 2010b; Wang et al., 2012c; Pu et al.,
2013; Wang et al., 2013a; Wang et al., 2013b; Lemus Pérez and Rodríguez Susa, 2017).
Therefore, this model can be used to compare to and explain experimental data obtained under
hydrodynamic conditions. Experimental studies for mass transport in biofilms have been developed
for substrate consumption such as glucose and oxygen (Zhang and Bishop, 1994; Guimerà et al.,
2016). The researchers have made attempts to represent such process by engineered correlations
among Sh, Re, and Sc. The current model showed that Sh for the substances chloroform, DCAN,
and chlorine is linearly correlated with Re for a flat biofilm, in the range 2,774-50,030. To the
author’s knowledge, this is the first study reporting such analysis for these substances.

Coupling the flow field to the transport of dissolved substances by CFD is a computationally
expensive task. Therefore, the current model included several simplifications such as flat biofilm,
no biofilm growth in order to only simulate DBP formation potentials, and flow in steady state. It is
known that biofilms can have irregular surfaces as a result of their internal morphology, depending
on the substrate concentration (Picioreanu et al., 1998b). The dynamics between residual
disinfectant and planktonic cells and biofilms in drinking water pipes is characterized by depletion
of the disinfectant by reaction with biomass and other inorganic and organic substances. During
this process, microorganisms are inactivated; once disinfectant is low, bacteria regrowth occurs.

Abokifa et al. (2016a) included this dynamics in their 1D model, under transitional flow, and found
that THM contribution from biomass occurs in pulses. While microorganism inactivation were taking
place, chlorine was decaying and THMs were increasing until reaching a plateau in the curve. When
bacteria regrowth occurred, the same previous pattern was repeated, increasing the absolute THM
concentration from biomass chlorination (Abokifa et al., 2016a). Furthermore, flow in water pipes
varies according to changes on customer demand or due to operative procedures such as closing
valves or intermittent supply (NRCNA, 2006). Therefore, steady state would not be the predominant
hydraulic condition. Variation of flow leads to changes on shear stress and biofilm detachment can
occur.

The current model is also limited by the absence of experimental data to compare to. Previous
experimental studies were carried out under static conditions or laminar flow regime. However, this
model can help to design new experiments on DBP formation in drinking water pipes under
transitional and turbulent flow. In addition, the sensitivity analysis of the model was carried out by
using the method with the lowest computational cost (Morris method) and by evaluating the method
only once. This represented 33 days to complete the sensitivity analysis. The methodology applied
by Menberg et al. (2016) included 10 evaluations of the model with the Morris method. The high
Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 243
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
computational cost of CFD models limited the application of more thorough sensitivity analysis
methods and this may represent that parameter influence was not properly ranked. However, the
results of the sensitivity analysis presented here agreed with the variables involved in the equations
included in the model. Furthermore, this analysis offered a general idea of the parameters which
must be determined with more accuracy and frequency in real scale DWDNs to be used as inputs
of the current model.

As hardware, software and more efficient numerical models are developed, more characteristics of
the real biofilm and flow in drinking water pipes can be included in future modelling studies. In the
meantime, the current model is an appropriate tool to evaluate the factors that influence the DBP
formation potentials resulting in certain initial characteristics of the biofilm, which can be relative
easy to measure. As mentioned in Chapter 4, some improvements on microscopy techniques must
be done to measure biofilm thickness from real scale DWDNs. On the other hand, the water industry
must do more efforts to incorporate biofilm research results into routine O&M of DWDNs. In this
line, an effective and close relationship academy, water industry and regulatory agencies must be
promoted.

5.5 CONCLUSIONS

This chapter presented the development of a model, in two dimensions, for predicting the formation
potentials of chloroform and DCAN, under bulk flow.

 Biofilms should be actively acknowledged as DBPs precursors. This may lead to improve the
prediction of DBPs in distribution networks. Improved models of DBP formation coupled with
validated hydraulic models in water networks may help to improve the assessment of chronic
risks in drinking water.
 Turbulence model SST is appropriate to solve the flow in the whole domain, especially near to
the pipe wall, where wall reactions take place. This model is also solver independent as shown
through solution in both COMSOL 5.2a and Ansys Fluent 16.0.
 The concentrations of chloroform and DCAN are mainly dependent of water quality parameters
such as initial chlorine concentration, initial cell concentration, and biofilm thickness. Higher
values of such parameters can lead to higher concentrations of both DBP species.
 Pipe diameter was also an important parameter influencing the DBP concentrations. Such
concentrations were higher with small pipe diameter such as 3 inches in relation to 10 inches.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 244
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Faster mass transport at the biofilm surface was found for higher Re, being faster for turbulent
flow, followed by transitional and laminar. Similarly, faster mass transport may occur in smaller
pipe diameter, and with higher initial concentrations of chlorine and cells and higher biofilm
thickness.
 Mass transfer rates of magnitude order of 10 -6 – 10-5 mg/m2-s were found for chloroform and
DCAN under transitional and turbulent flow regime.
 A linear relationship between Re and Sh was found with the simulations run by the model
developed in this study.
 From the point of view of microbiological risk, slow flow represents the most critical condition to
properly disinfect biofilms and control bacteria growth within them. Similarly, chemical risk may
be also high due to lower flow rates led to higher concentrations of DBPs, despite of the reduced
mass transfer.
 Field and laboratory assessment of the biofilm properties must be focused primarily on biofilm
thickness and cell density. Reaction rates and yield coefficients can still rely on published data
in the scientific literature but further experimental studies may refine those results by using other
types of disinfectant, multispecies biofilm, and environmental temperature.
 Water operators must pay especial attention to control the chronic risk associated with DBPs
and human health in the smaller pipes of the DWDNs, where lower flow rates can be present,
then major interactions bulk-biofilm are expected and more DBPs can be formed. Furthermore,
for those portions of the networks are immediately close to the customers, then the formed
DBPs will enter the plumbing systems and may increase their concentration under stagnation
conditions, which prevail within the building facilities.

Chapter 5. Modelling DBP formation from biofilm chlorination under hydrodynamic conditions 245
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 6 GENERAL DISCUSSION, IMPLICATIONS AND APPLICATIONS

6.1 GENERAL DISCUSSION

The current research project analysed the impact of the presence of biofilms in DWDNs by two
approaches: microbial and chemical risk, with major emphasis on the last one, due to the growing
research interest on biomass as DBP precursors. Figure 6-1 shows how the dual role of biofilms
was studied in the current research project. Such impact was studied by a fieldwork carried out in
a full scale DWDN located in a tropical-weather country. Such fieldwork allowed identifying the
predominant bacterial communities in biofilm and bulk water habitat, in nine points of the DWDN.
Correlations between biotic parameters and engineered factors were determined and relationships
between bacterial communities and THMs were also explored. Following the field study, two
models were developed as tools to predict the formation potentials of chloroform and DCAN from
chlorination of biofilms in drinking water pipes. Such models allowed determination of the
concentrations of such DBPs under stagnation and hydrodynamic conditions, according to biofilm
characteristics, pipe diameter, water quality, and hydraulic conditions.

Figure 6-1. Dual role of biofilms studied in the current research project
Chapter 6. General discussion, implications and applications 246
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
As can be observed in Figure 6-1, by the field assessment in a tropical-weather DWDN, correlations
between TTHMs and methanotrophs in bulk water samples and the presence of these organisms
in biofilm samples may confirm the interrelations between microorganisms and chemical
substances such as DBPs in real scale DWDNs. Concentrations of free residual chlorine measured
in the Cali’s distribution network were used as inputs of the several scenarios simulated with the
two models developed in the current research project. In this line, Figure 6-2 presents other
parameters to be considered in further field assessments of biofilms, in order to collect more field
data to be used as inputs of models of disinfectant decay and DBP formation including biomass
contribution. Parameters to be considered in future field works should include biofilm thickness,
chlorine and DBP concentrations, flow rate, pipe diameter, cell density, EPS concentration and
identification of microbial communities. This will contribute to improve definition and validation of
hydraulic and water quality models focused on analysis of balance between microbial and chemical
risk in drinking water systems.

Figure 6-2. Two approaches to study the microbial and chemical significance of biofilms in
drinking water

Biofilms in drinking water have been mainly studied for research purposes, especially due to its
capacity to host pathogens. These organisms can be released to the bulk, reach the consumers,
and cause infection if the biofilm is detached and disinfectant is insufficient or unable to inactivate
them. In plumbing systems, opportunistic pathogens (e.g., Legionella spp., Mycobacterium spp.,

Chapter 6. General discussion, implications and applications 247

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Pseudomonas aeruginosa) generate special concerns due to they find the perfect conditions there
for survival and growth. Opportunistic pathogens are disinfectant resistant, biofilm formers, and
able to grow within amoeba, under oligotrophic conditions and low oxygen levels (Falkinham et al.,
2015). Additional to the ingestion exposure route, opportunistic pathogens can also be inhaled from
aerosols formed during showering. These pathogens affect to individuals with predisposal
conditions such as advance age (older than 70 years), cancer or immunodeficiency (Falkinham et
al., 2015). Therefore, special attention must be paid to safety of drinking water supplied particularly
in medical institutions and care homes.

On the other hand, investigation on biomass as DBP precursors has been continuously developed
in the last decade. Researchers have been interested on quantifying, by laboratory tests, the
contribution of algae, EPS, planktonic cells of single species, both heterogeneous and
homogeneous biofilms, detached clusters biofilms, and biomolecules to the formation of
carbonaceous and nitrogenous DBPs, by both chlorination and chloramination (Hong et al., 2008;
Fang et al., 2010a; Fang et al., 2010b; Wang et al., 2012c; Pu et al., 2013; Wang et al., 2013a;
Wang et al., 2013b; Lemus Pérez and Rodríguez Susa, 2017). According to this, a recent study
focused on modelling the formation of THMs in a DWDN, for turbulent flow regime, included the
contribution of fixed and suspended biomass (Abokifa et al., 2016a). The researchers estimated
that biomass contributed with around maximum 12% of the total THM concentrations simulated
(Abokifa et al., 2016a).

According to Lee and Schwab (2005), DWDNs from developing countries are characterized for
several failures such as low or absence of residual disinfectant, low water pressure, intermittent
service, and high rate of leaks; poor basic sanitation is also a common situation in such countries.
Everything together represents that a big portion of the population of these countries are exposed
to microbial risks by water consumption (Lee and Schwab, 2005). As described in Chapter 3, the
DWDN of the city of Cali presents all the failures mentioned previously, in certain portions of the
network and during specific times. A special feature of this network, which has been reported
frequently in communication media, is the intermittent supply due to the contamination of the water
source. In addition, the ageing infrastructure (30.4% of the pipelines are made of asbestos) and
the hydraulic changes cause a high rate of leaks (81/100 Km for year 2013) 17. High rate of leaks
represents that service must be frequently interrupted in the affected region to repair the leak, then
disturbing the hydraulics of the circuit. According to USEPA’s guidelines (USEPA, 2002c) and
during the field work, it was also observed that the new piece of pipe installed did not contain any

17 EMCALI EICE ESP, 2014, personal communication, 10th December

Chapter 6. General discussion, implications and applications 248
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
protection against contamination and it is unknown if this was disinfected prior to installation.
Neither disinfection nor flushing prior to putting the pipe into service was confirmed. Repair staff
neither take any preventive procedure to protect themselves to avoid passing contamination to the
pipelines. The affected portion of the network was not flushed either prior to reinitiate water supply.

Thus, it is clear that several processes are simultaneously occurring in Cali’s DWDN that may
threaten the microbial drinking water quality delivered to the consumers. With regards to chemical
risk, DBP concentrations were in average 31.76 g/L, which is below the UK and Colombian
standards (100 and 200 g/L, respectively) (UK Parliament, 2000; Ministerio de la Protección
Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial, 2007). In this line, the models
developed here showed that the more risky situation is under stagnant conditions, particularly in
plumbing premises, where DBP exposure occurs. Similarly, other studies have also stablished that
households are exposed to microbial risk in the building facilities (Falkinham et al., 2015; Ji et al.,
2015; Ji et al., 2017). Under no flow situation, chlorine may decay by reaction with pipe walls, then
DBPs are formed and transported to bulk water, and bacteria regrowth occurs in both bulk water
and biofilm. After a period of stagnation, householders use stagnant water and can be exposed, by
mainly ingestion or inhalation, to DBPs and bacteria.

Under turbulent flow situation, it is clear that the major chemical risk associated to DBPs is in
smaller pipes, i.e., those around the point of use, where bulk-wall interactions are more important,
and flow rates are lower, then dissolved substances as DBPs are more concentrated. Furthermore,
mass transfer is enhanced when Re is higher. Under the same Re, low chlorine concentrations led
to reduced mass transfer, and DBP formation and transport is reduced. However, low disinfectant
residuals also poses a microbial risk, because low mass transfer at the biofilm surface reduces the
penetration of the biofilm by chlorine and, consequently, the inactivation of potential pathogenic
microorganisms may decrease.

Disinfectant resistance by bacteria, protection by biofilm of microorganisms, DBP formation by

biofilm chlorination, and disinfectant decay occur in real scale DWDNs. Then, exclusively relying
on disinfection for biofilm control with focus on pathogen control is not sufficient. In this line, it is
necessary to stablish a trade-off between the microbial and chemical risk and minimize the biofilm
formation to guarantee the delivery of safe water to the consumers, which is the final goal of water
utilities, as it has been stablished by Hrudey and collaborators (Hrudey, 2009; Hrudey and Charrois,
2012). In order to present the practical implications of both approaches, the next sections include
the discussion of the microbial risk associated with bacteria identified in the field assessment, the

Chapter 6. General discussion, implications and applications 249

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
chemical risk analysed with the mathematical models, their implications on O&M of DWDNs, and
the applicability of this research.

6.2 BACTERIAL COMMUNITIES AND PUBLIC HEALTH

Potential pathogenic bacteria were identified in several bulk and biofilm samples in Cali’s DWDN.
Due to such identification was based on DNA sequencing of a low number of base pairs, it can only
be established the presence of the bacteria to genus level. The establishment of species and
pathogenic strains and their viability must be done by other microbiological techniques such as
quantitative and real-time PCR, staining-microscopy methods, measurement of ATP, and
incorporation of bromodeoxyuridine, microautoradiography combined with FISH (Douterelo et al.,
2014a). Due to the limitations of this study, it is not possible to calculate quantitative microbiological
risk assessment. However, the results presented here are crossed with the map risks determined
by Pérez-Vidal et al. (2012) for the same DWDN.

The city of Cali is administratively divided into 22 communes, which are groups of neighbourhoods
that share similar socio-economic conditions. Pérez-Vidal et al. (2012) generated maps of
hydraulic, physical and water quality integrity, according to criteria such as minimal pressure,
number of leakages in aqueduct, number of leakages in sewage, number of leakages in household
connections, percentage of material in the DWDN, historical records of water quality, and water
quality complaints. The map risk was the result of the weighed combination of previous maps and
population density.

Table 6-1 presents the sampling points and communes where DNA of potential pathogenic bacteria
was identified and the respective risk classification obtained by Pérez-Vidal et al. (2012). The
crossed information presented in Table 6-1 shows that sampling points where pathogenic bacteria
were identified are located in the areas of high and very high risk. This is not a coincidence, since
samples were collected during leak repair activities and map risk was created using the number of
leaks in the aqueduct. Pérez-Vidal et al. (2012) assigned the highest weight to this criterion together
with minimal pressure in the map of physical and hydraulic integrity (30% each). Similarly, this type
of integrity had the highest weight together with water quality (40% each). Finding pathogenic
bacteria in those communes represents that choosing the criterion number of leakage in a DWDN
is reasoning, due to the loss of physical integrity during repair activities promotes the introduction
of external contaminants to the system. This is probably the cause of the presence of bacteria listed
in Table 6-1.

Chapter 6. General discussion, implications and applications 250

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 6-1. Potential pathogenic bacteria identified in Cali’s DWDN and their corresponding risk
area defined by Pérez-Vidal et al. (2012)
Bacteria genus / Sampling Commune risk classification
Commune
Habitat points according to risk maps
Acinetobacter
Point 8 4 Very high risk
Biofilm
Point 1 4 Very high risk
Brucella Point 2 8 Very high risk
Bulk water Point 4 7 High risk
Point 6 8 Very high risk
Mycobacterium
Point 1 4 Very high risk
Bulk water
Point 6 8 Very high risk
Staphylococcus
Point 8 4 Very high risk
Biofilm
Point 9 7 High risk
Point 3 10 High risk
Pseudomonas
Point 5 4 Very high risk
Biofilm
Point 8 17 High risk

Acinetobacter, Staphylococcus and Pseudomonas are biofilm formers, opportunistic pathogens,

have been identified in infected patients and are able to colonize medical devices (Heilmann et al.,
1996; Stover et al., 2000; Carr et al., 2003). Therefore, it is important that medical institutions and
care homes incorporate additional remediation measures to avoid potential interaction between
opportunistic pathogen and immune compromised patients. In order to reduce opportunistic
pathogens in this type of institutions, Falkinham et al. (2015) recommended to increase the
temperature of hot water and avoid its recirculation; clean and disinfect shower heads; replace
shower heads with large hole; install single unit hot water systems to reduce the travel time of hot
water; remove aerators from taps and faucets; carry out building wide disinfection; reduce aerosols
by filter entrapment; and install microbial filtration devices in specific locations. It is important to
mention that these organisms may be present in other regions of Cali’s DWDN, which may be
transported from one point to others during hydraulic changes.

To the author’s knowledge, this is the first study reporting Brucella in a DWDN. This genus was the
predominant group in Point 1. Brucella comprises 11 species, ten of them are associated with
human infections (Scholz et al., 2010; Xavier et al., 2010) and the hosts of these pathogenic
bacteria are commonly mammals (Eisenberg et al., 2012). Human brucellosis is considered as a
life-threatening debilitating disease and sources of infection include inhalation of contaminated dust
or aerosols, ingestion of contaminated milk and unpasteurized dairy products, and exposure of
mucosa or skin abrasions to fluids and tissues from aborted foetuses of infected animals or carcass
(Xavier et al., 2010). Despite of there is not conclusive evidence on microbial risk in the studied
DWDN, the results presented here indicate that water managers should adopt improved O&M
Chapter 6. General discussion, implications and applications 251
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
measures to prevent deterioration of drinking water quality delivered to the consumers.
Furthermore, another study should be carried out to confirm the presence of potential pathogenic
bacteria belonging to the genus identified in this project.

6.3 DBPS AND PUBLIC HEALTH

DBP formation is a very complex process occurring in drinking water systems because many
factors participate in it. Factors include type and dose of disinfectant, type and amount of
precursors, pH, temperature, residence time, and presence of bromide. This complexity is reflected
on the identification of around 600 DBPs, classified into 26 groups, but only four groups are
currently subject of regulation or recommendation of health-based thresholds (Richardson et al.,
2007). Certainly, the presence of disinfectant in drinking water is the main driver of DBP formation.
Non-disinfected systems do not deal with this problem, but this occurs in cities, where water
sources such as groundwater are properly preserved. Supply systems from Netherlands,
Switzerland and Germany do not rely on residual disinfectants, but do on advanced treatment
processes, improved physical integrity of the distribution system, and careful management of
distribution system operations (Rosario-Ortiz et al., 2016).

When the water sources are contaminated, water utilities must rely on disinfection to inactivate
pathogenic microorganisms in raw water and protect the distributed water from further
contamination. Since the discovery of THMs in 1974, great amount of research has been conducted
and regulations around the world have been designed to protect the consumers from chemical risk
associated with DBPs. In this line, water operators have faced the dilemma about either maintaining
disinfectant residuals in the water network or minimize the DBP formation. When disinfection
cannot be evaded, chemical safety of drinking water is supported by the regulation, then water
utilities focuses on reducing the DBP precursors. It is clear then that regulation of DBPs have
promoted the improvement of drinking water quality by advancement on treatment processes and
better practices of O&M of DWDNs (Humpage, 2012).

The previous analysis is the ideal approach in the context of the uncertainty of the association
between mutagenic, carcinogenic and teratogenic effects on human health. According to Hrudey
and Charrois (2012); Hrudey et al. (2015b), the epidemiological evidence is not yet conclusive to
directly associate the most abundant and regulated DBPs such as THMs and HAAs with such
negative health impacts. To date, the study developed by Cantor et al. (2010) has provided the
strongest indication of potential association between bladder cancer and THMs in drinking water

Chapter 6. General discussion, implications and applications 252

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
(Hrudey et al., 2015b). However, the relationship between DBPs and negative health outcomes is
still an open question. While the answers to this question arrive, it is necessary to put in proper
perspective the management and control of DBPs in drinking water. In this line and according to
the aim of this project, it is necessary to assume the biofilm risk as a trade-off and assimilate the
biofilms as DBP precursors and potential reservoirs of pathogens.

In this context, the models developed in this project represent a tool to improve the prediction of
DBP concentrations in plumbing systems and DWDNs, under both stagnation and hydrodynamic
conditions. Proper prediction of DBPs would help to improve the assessment exposure, which is a
key component on epidemiological studies related to these substances. The hydrodynamic model
also contributes to understand the mass transfer of dissolved substances like chlorine and DBPs
at the biofilm surface, then scenarios where more or less chlorine is desirable in certain areas of
the network can be identified, in order to find the balance between bacteria and DBP control.

Consolidated results of both 1D and 2D models are included in Table 6-2 and Table 6-3,
respectively. These results evidence that the most critical conditions in relation to DBP
concentrations are under stagnation conditions, followed by transitional and turbulent flow.
Diffusion is the predominant mass transport mechanism when there is not flow. For contact time of
11 hours, which can normally occur even in single building premises, all the biomass and chlorine
are available for reacting, and formed DBPs are transported to the bulk water and diluted in small
volumes.

Table 6-2. Main results of 1D model of potential of DBP formation from biofilm chlorination
Initial conditions (Contact time = 11
Potential DBP
hours,  = ½ inches)
concentrations
BT (m) Clo (mg/L) Xo (mg/L)
Chloroform = 6.98 g/L
0.89
0.12 and DCAN = 3.43 g/L
7
1.66 Chloroform = 31.62 g/L
13.98
DCAN = 13.16 g/L
Chloroform = 89.29 g/L
0.89
DCAN = 44.30 g/L
0.12
Chloroform = 146.25 g/L
13.98
DCAN = 61.42 g/L
102
Chloroform = 117.40 g/L
0.89
DCAN = 58.26 g/L
1.66
Chloroform = 195.94 g/L
13.98
DCAN = 466.20 g/L
Initial conditions
Potential DBP
BT and
(inches) Xo (mg/L) concentrations
Clo
7 m | Chloroform:
½-9 0.89 – 0.05
0.12 mg/L 6.91 – 0.30 g/L
Chapter 6. General discussion, implications and applications 253
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 DCAN:
3.39 - 0.16 g/L
Chloroform:
13.98 – 102 m | 466.20 – 6.55 g/L
0.78 1.66 mg/L DCAN:
195.94 – 3.27 g/L

Chapter 6. General discussion, implications and applications 254

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Table 6-3. Main results of 2D model of potential of DBP formation from biofilm chlorination
Initial conditions (Contact time = 1 hour, Pipe length = 1 m)
Potential DBP concentrations
BT Clo Xo  (maximal value)
Sh
¯ - DBPs Mass transfer (m/s)
Re Other
(m) (mg/L) (mg/cm2) (inches)
Chloroform Chloroform Chloroform
Laminar: 4.06 x 10-2 g/L Laminar: 0.0085 Laminar: 1.13 x 10-6
Flow regime: Transitional: 3.46 x 10-2 g/L Transitional: 0.0226 Transitional: 5.34 x 10-6
Laminar, Turbulent: 1.42 x 10-2 g/L Turbulent: 0.1016 Turbulent: 2.14 x 10-5
1200 | 4161 |
7 1.66 4.44 x 10-3 3 transitional
25084
and turbulent DCAN DCAN DCAN
flow Laminar: 2.16 x 10-2 g/L Laminar: 0.0054 Laminar: 1.28 x 10-6
Transitional: 1.97 x 10-2 g/L Transitional: 0.0023 Transitional: 5.44 x 10-6
Turbulent: 6.82 x 10-3 g/L Turbulent: 0.0895 Turbulent: 2.12 x 10-5
Chloroform: Chloroform:
Chloroform = 4.72x10-2 – 3.76x10-2 g/L 0.0163 – 0.0174 2.16x10-6 – 3.46x10-6
7 1.66 4.44 x 10-3
DCAN = 2.55 x 10-2 – 1.97 x 10-2 g/L DCAN: DCAN:
Transitional 0.0144 – 0.0229 3.43x10-6 – 5.44x10-6
3 2774 - 4994
flow Chloroform: Chloroform:
Chloroform = 0.58 – 0.47 g/L 0.2680 – 0.4012 2.44x10-6 – 3.65x10-6
102 1.66 4.44 x 10-3
DCAN = 0.34 – 0.25 g/L DCAN: DCAN:
0.2326 – 0.3462 3.79x10-6 – 5.63x10-6
Chloroform: Chloroform:
Chloroform = 7.83x10-3 – 2.68x10-2 g/L 0.0462 – 0.1863 6.13x10-6 – 2.47x10-5
7 1.66 4.44 x 10-3
DCAN = 3.70 x 10-3 – 1.35 x 10-2 g/L DCAN: DCAN:
10034 - 0.0405 – 0.1646 9.61x10-6 – 3.90x10-5
3 Turbulent flow
50030 Chloroform: Chloroform:
Chloroform = 0.10 – 0.32 g/L 0.6654 – 2.7275 6.06x10-6 – 2.48x10-5
102 1.66 4.44 x 10-3
DCAN = 5.15 x 10-2 – 0.17 g/L DCAN: DCAN:
0.5792 – 2.3994 9.43x10-6 – 3.90x10-5
Chloroform = 4.06 x 10-2 g/L Chloroform = 0.0226 Chloroform = 2.99x10-6
7 1.66 4.44 x 10-3 3 S/V = 52.5 m-1
DCAN = 2.15 x 10-2 g/L DCAN = 0.0199 DCAN = 4.71x10-6
4161
Chloroform = 2.07 x 10-2 g/L Chloroform = 0.0120 Chloroform = 1.59x10-6
7 1.66 4.44 x 10-3 6 S/V = 26.3 m-1
DCAN = 1.50 x 10-2 g/L DCAN = 0.0137 DCAN = 3.24x10-6
Chapter 6. General discussion, implications and applications 255
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Initial conditions (Contact time = 1 hour, Pipe length = 1 m)
Potential DBP concentrations
BT Clo Xo  Re Other (maximal value)
Sh
¯ - DBPs Mass transfer (m/s)
(m) (mg/L) (mg/cm2) (inches)
Chloroform = 1.17 x 10-2 g/L Chloroform = 0.0059 Chloroform = 7.84x10-7
7 1.66 4.44 x 10-3 10 S/V = 15.8 m-1
DCAN = 7.10 x 10-3 g/L DCAN = 0.0052 DCAN = 1.24x10-6
Chloroform = 2.67 x 10-2 g/L Chloroform = 0.0467 Chloroform = 6.20x10-6
7 1.66 4.44 x 10-3 3 10173 S/V = 52.5 m-1
DCAN = 1.33 x 10-2 g/L DCAN = 0.0410 DCAN = 9.72x10-6
Chloroform = 2.46 x 10-2 g/L Chloroform = 0.0274 Chloroform = 3.63x10-6
7 1.66 4.44 x 10-3 6 5086 S/V = 26.3 m-1
DCAN = 1.29 x 10-2 g/L DCAN = 0.0242 DCAN = 5.74x10-6
Chloroform = 1.15 x 10-2 g/L Chloroform = 0.0049 Chloroform = 6.44x10-7
7 1.66 4.44 x 10-3 10 3051 S/V = 15.8 m-1
DCAN = 7.10 x 10-3 g/L DCAN = 0.0049 DCAN = 1.15x10-6
Chloroform = 3.71 x 10-3 g/L Chloroform: 0.0211 Chloroform: 2.81x10-6
7 0.12 2.82 x 10-4
Transitional DCAN = 2.03 x 10-3 g/L DCAN: 0.0178 DCAN: 4.23x10-6
3 2774
flow Chloroform = 0.58 g/L Chloroform: 0.2680 Chloroform: 2.44x10-6
102 1.66 4.44 x 10-3
DCAN = 0.34 g/L DCAN: 0.2326 DCAN: 3.79x10-6
Chloroform = 6.48 x 10-4 g/L Chloroform: 0.1027 Chloroform: 1.36x10-5
7 0.12 2.82 x 10-4
DCAN = 3.29 x 10-4 g/L DCAN: 0.0901 DCAN: 2.14x10-5
3 25084 Turbulent flow
Chloroform = 0.18 g/L Chloroform: 1.4733 Chloroform: 1.34x10-5
102 1.66 4.44 x 10-3
DCAN = 9.17 x 10-2 g/L DCAN: 1.2929 DCAN: 2.10x10-5

Chapter 6. General discussion, implications and applications 256

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
The major concern in building facilities is that formed DBPs are ready to be consumed once the
householders use again drinking water. Therefore, consumers may be exposed to high
concentrations of DBPs by either ingestion or inhalation, after a period of stagnation. On the
contrary, the quantification and impact of DBPs formed under transitional and turbulent flow regime
depends on the pipeline length, flow rate and pipe diameter. The results of the 2D model allow
inferring that the major risk can be in small pipes due to the high S/V ratio and lower flow rates.
Such pipes can be those in the proximity to the consumers; then exposure probability to such DBPs
by customers may increase. Subnetworks installed within residential neighbourhoods are also
prone to stagnation conditions overnight; then the DBP concentrations would increase, as
explained for 1D model results.

With regards to mass transport, mass transfer is faster when Re is higher and pipe diameter
smaller. Decreasing chlorine concentrations in bulk water reduces the mass transport of this
disinfectant and DBP formed at the biofilm surface. This is an important outcome of the model
because it allows assessing the balance between microbial and chemical risk. Low concentrations
of chlorine leads to slow penetration of the biofilm thickness and slow formation of DBPs, which are
transported to the bulk water and leave the pipe slower than when high chlorine concentrations are
present. From the point of view of chemical risk, this would be a desirable condition; however, slow
penetration of the biofilm thickness also represents that potential pathogens are not being
inactivated. The 2D model and further improvements can be useful to evaluate when it is necessary
to increase disinfectant concentrations to inactivate microorganisms and when it must be reduced
to minimise DBP exposure to customers. Such improvements may be achieved by further
availability of microbial characteristics of biofilm in full scale DWDNs; important variables are biofilm
thickness, cell density, and EPS concentration. Measuring DBP concentrations within biofilms by
micro-sensors is also a crucial step for model validation.

The aforementioned leads to recommend close collaboration between water industry and research
laboratories to improve the analytical and microbiological techniques that promote better monitoring
of biofilm roles as both pathogen hosts and DBP precursors. Combined data from models and field
assessments in DWDNs should lead to improve O&M if they are properly interpreted to protect
drinking water quality. Additionally, including wall contribution of DBPs to bulk water concentrations
may also improve the performance of current DBP models for DWDNs.

Chapter 6. General discussion, implications and applications 257

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
6.4 IMPLICATIONS

6.4.1 Implications for water quality in Cali’s DWDN in the context of extreme weather
events

According to IPCC (2014), one of the clear consequences of climate change is the increase in
frequency and severity of a variety of extreme weather events, including heavy rainfall, floods and
droughts. Similarly, the overall frequency in climate phenomena such as El Niño and La Niña is
expected to increase under global warming (Cai et al., 2014; Cai et al., 2015). Additionally, climate
change is projected to increase water scarcity in urban areas, and rural areas are expected to
experience major impacts on water availability and supply (IPCC, 2014). To cope with the urban
risks associated with water supply, IPCC (2014) includes adaptation options such as changes to
network infrastructure as well as demand-side management to ensure sufficient water supplies and
qualities; and increased capacities to manage reduced freshwater availability, and flood risk
reduction. For the present times, IPCC (2014) have calculated water supply-associated urban risk
as medium, medium-very high risk with potential to reduce to medium for the near term (2030-
2040), medium-very high and medium risk for the long term (2080-2100), under the scenarios of 2
°C and 4 °C of increase of global temperature, respectively.

In this line, Cann et al. (2012) documented 87 and 304 waterborne outbreaks from medical and
meteorological databases and the global electronic reporting system ProMED, respectively. Such
waterborne outbreaks were commonly preceded by extreme weather events such heavy rainfall
and flooding. The most common pathogens reported in these outbreaks were Vibrio spp. and
Leptospira spp. The events were often the result of contamination of the drinking water system
(Cann et al., 2012). According to data from medical and meteorological databases, most of the
Vibrio-associated outbreaks occurred in Asia, followed by Africa and South America and most of
the Leptospira-associated outbreaks took place in North America or Asia. Causes of contamination
of the drinking water system included failure to cope by the WTP after the extreme weather events,
failure or inability to cope by sewage systems, and resulting in contamination of drinking water. In
developed countries, route of infection was mainly through the DWDNs. In developing countries,
the main cause was contamination of the water supply.

The information collected from ProMED by Cann et al. (2012) evidenced that the majority of events
were in Africa, followed by Asia and North America. The likely causes of the waterborne outbreaks
were contamination of water, shortage of clean drinking water, poor sanitation and hygiene
following the extreme weather event. In conclusion, less developed countries are more vulnerable
to the consequences of extreme weather events (Cann et al., 2012; Rataj et al., 2016). However,

Chapter 6. General discussion, implications and applications 258

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
developed countries are also susceptible to suffer the negative effects of these events. In both
cases, upgrading the drinking water supply systems, changing water use patterns, explore new
water sources are some adaptation measures, which can be implemented by water utilities and
planners (WHO, 2017).

Additionally to the increased frequency of extreme weather events due to climate change, which
will affect water supply, global warming per se is also expected to change the surface water quality,
consequently affecting the performance of WTPs. Global warming refers to the increase of
environment temperature. Temperature is the main factor affecting almost all physico-chemical
equilibriums and biological reactions (dissolution, solubilisation, complexation, degradation,
evaporation, etc.) (Delpla et al., 2009). Therefore, the impacts of climate change on water quality
parameters include increase of temperature, reduction of dissolved oxygen; rise of DOC, alkalinity,
and pH; enhancement of nitrogen mineralisation; release of nitrogen, phosphorus, and carbon from
soil to water bodies; loss of dilution capacity of water bodies during droughts; rise of organic and
inorganic pollutants; spread of waterborne pathogens and cyanobacteria blooms (Delpla et al.,
2009).

Delpla et al. (2009) established that such changes on surface water quality due to climate change
impact the drinking water quality produced by WTP due to the increase of stable DBP
concentrations, as a consequence of higher concentrations of precursors such as DOC, algae, and
biomass in general. It is expected to have moderate impact for unstable DBPs such as DCAN and
1,1-DCP since they decompose with higher temperatures and HAAs are biodegradable. On the
other hand, spread or waterborne pathogens challenge the operation of WTPs, since disinfection
must be maintained in the treatment train and more efforts must be done on removal of DBP
precursors to minimize DBP concentrations in the DWDNs.

However, Colombian DWDNs may be poorly operated and maintained, even in centralised systems
such as in Cali. Unfortunately, this is a common situation in developing countries due to limited
resources (Lee and Schwab, 2005). One of the causes of deficiencies in DWDNs of developing
countries is the leakages. In the year 2013, Cali’s DWDN reported 81 leakages per 100 Km 18. In
addition, Cali’s water utility loses 53% of the DW produced; in order to reduce loss of water, it is
required to replace around 1200 Km of the oldest pipelines, corresponding to pipe materials CI and
asbestos, which represent 40% of the total network length 19. Additionally, the two main WTPs of

18 EMCALI EICE ESP, 2014, personal communication, 10th December

19 Rojas Ramírez, 2017, personal communication, 10th July. E-mail: [email protected]
Chapter 6. General discussion, implications and applications 259
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
the city of Cali suspended their operation during events of water source contamination in a range
of 1-13 times in the period 2009-201620. The high number of leakage repairs and water supply
interruptions due to deterioration of raw water quality lead to unstable hydraulic conditions in the
DWDN, which can promote the uncontrolled release of biofilm, sediments, loose deposits, and
other biological material. These elements can reach the customers and may represent complaints
due to poor aesthetic conditions or, even worse, public health risks if pathogens were present.

One of the most common recommendations to adapt to climate change, especially during extreme
weather events, is to adjust chlorine doses in the WTPs and maintain residual chlorine in the
network. Such recommendation obeys to the concern of acute risk due to the potential presence of
pathogens, with the subsequent negative impacts on public health, which can be worse in
developing countries with low adaptation and response capacity. From the results presented in this
research project, it is advisable that modelling tools be used to evaluate the trade-off between
microbial and chemical risk in drinking water pipes. In the meantime, DBP research should focus
on providing conclusive evidence on which specific DBP species produce negative health
outcomes, in which degree, which minimal concentrations, and what type of diseases. It is also
advisable that O&M of DWDNs consider the bio-stability approach to preserve the drinking water
quality to the customers.

6.4.2 Implications for O&M activities in DWDN

Bacteriological composition of drinking water biofilms have been mainly studied because of the
protective effects offered to pathogens against disinfectants; such pathogens can be released to
bulk water if hydraulic changes occur. Experimental studies have shown that biomass can act as
DBP precursors. Prevention and removal of biofilms must be a key concern for water utilities;
flushing water pipes has been proved as a suitable technique to remove material attached to
internal pipe surfaces but it is inefficient to completely detach biofilms (Abe et al., 2012; Douterelo
et al., 2013; Fish et al. 2016). Advanced water treatment processes such as membrane filtration
has been proved successful in highest reduction of number of microorganisms in biofilms collected
at the inlet of a DWDN (Shaw et al., 2014). However, a recent study argues that is impossible to
prevent biofilm accumulation but high flow variation could be used to promote young biofilms, which
are more vulnerable to disinfection (Fish et al., 2017).

20 Events counted until September 2016

Chapter 6. General discussion, implications and applications 260
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
In the case of this studied network, avoiding uncontrolled biofilm detachment and contamination of
bulk water is particularly difficult, as it exhibits specific O&M challenges associated with emptying
of the network due to the interruption of operation of the WTPs, pumping operation, closing/opening
valves during leakages repairs and pipelines and accessories replacement. This may lead to favour
the formation of young biofilms, however it is important to consider that the biofilm detachment may
occur in an uncontrolled way, then biofilm clusters containing potential pathogens may reach the
customers.

Samples collected in this study potentially contained pathogenic and opportunistic bacteria but,
whether they were metabolically active or not, is unknown since DNA was used to characterise the
samples. However, these results can motivate the application of proper practices of O&M of the
DWDN in order to avoid uncontrolled biofilm detachment and contamination of bulk water.
Furthermore, CI pipes represent 10% of the total length of the pipelines and asbestos 30%; and
2,400 leakages were repaired in 201321. These O&M activities cause uncontrolled and partial
removal of sediments and biological material and allow the entrance of external particles, which all
together could be promoting microbial growth in the network. Future plans for pipeline replacements
should avoid the use of metal and cement pipes and instead promote the use of pipe materials with
more stable bio-chemical and physical conditions. It is also advisable to maintain stable hydraulic
conditions to avoid biofilm detachment; controlled cleaning procedures of pipes such as flushing
should be carried out to reduce the amount of nutrients available for microorganisms in bulk water
and biofilms and avoid alterations of the organoleptic conditions of DW for the consumers. Special
attention must be put in dead-end zones to flush stagnant water to reduce microorganism regrowth
and DBP formation. More importantly, the efforts carried out in protecting water sources and
improving water treatment could be useless if suitable O&M practices are not applied in the DWDNs
in order to preserve the safety of drinking water delivered to the customers.

6.5 APPLICABILITY OF THE CURRENT RESEARCH PROJECT

Despite of the research evidence in relation to microbial and chemical risk associated to the
presence of biofilms in drinking water pipes, there is not knowledge to date about introduction of
biofilm monitoring in O&M of DWDNs by water utilities. In this line, the tools and results of the
current research project become relevant with regards to the supply of new information of the

21 EMCALI EICE ESP, 2014, personal communication, 10th December

Chapter 6. General discussion, implications and applications 261
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
bacterial communities that may be present in the studied full scale DWDN, and their relationship
with engineered factors and water quality characteristics. In addition, the models developed here
are relatively easy to use and implement, especially for stagnation conditions, if the appropriate
knowledge, trained staff, and software and hardware requirements are available. In this line, more
efforts must be done in order to transfer the practical knowledge to the water industry and
strengthen their capacity to apply such knowledge, especially in the context of developing
countries, where economical limitations exist and are more vulnerable to the impacts of climate
change. The author suggests applying the results of this research project to inform better practices
of O&M of Cali’s DWDN, improve prediction of DBPs in DWDN, assess the balance between
microbial and chemical risk, and make more progress on DBP exposure assessment in
epidemiological studies.

Chapter 6. General discussion, implications and applications 262

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 7 KEY FINDINGS, CONCLUSION AND FUTURE RESEARCH

7.1 GENERAL CONCLUSION

Biofilms play an important role in DWDNs as pathogen reservoirs and DBP precursors. Additionally,
diverse bacterial communities exists in this habitat. Such communities can lead to diverse physico-
chemical and biological processes in this type of engineered systems, by the influence of pipe
materials, hydraulic factors, and drinking water characteristics. One of such processes is the
formation of DBPs by reactions between biofilm matrix and disinfectants. Then, mathematical
models developed in the current research project represent a progress to analyse the chemical and
microbiological risks in drinking water pipes. The models are also useful for prediction of DBP
formation potentials, designing laboratory experiments, and the evaluation of regulation compliance
scenarios. Special attention must be put on large DWDNs in order to control the DBP formation in
the sectors where small pipes and low flow rates or stagnation conditions are frequent.

7.2 KEY FINDINGS

This research potentially informs to water engineers and managers on the impact of the presence
of biofilms in drinking water pipes. The following key findings correspond to the objectives
formulated in the current research project.

7.2.1 Bacterial communities in a tropical-climate DWDN

The field study enabled analysis of the bacterial communities in the DWDN of the city of Cali, and
the following are the key findings:

 Biofilm samples were more diverse in relation to bulk water samples.

 Proteobacteria phyla was also the predominant group in all the samples collected in the tropical-
weather DWDN.
 Genus bacteria associated with soil, surface water, pathogenicity-related, and methyl radicals
like THMs were identified in both habitats.
 Significant correlations were found between TTHM concentrations and RA of methanotrophs in
bulk water.

Chapter 7. Key findings, conclusion and future research 263

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
7.2.2 Relationships between abiotic and biotic parameters in a tropical-weather DWDN

The field study enabled the identification of relationships between biotic, physico-chemical, and
engineered factors determined in the DWDN of the city of Cali, and the following are the key
findings:

 Pipe age, pipe material, water age, free chlorine, pH and temperature were associated with
microbiological parameters indicating that these are key to the composition of bacterial
communities.
 Pipe material influences the microbial ecology of DWDNs. Desulfovibrio was identified
exclusively in the cast iron pipe.

7.2.3 DBP formation potentials from biofilm chlorination under stagnation conditions

The development and analysis of the one-dimensional model enabled the study of the parameters
influencing the formation of chloroform and DCAN under stagnation conditions. The key findings
are the following:

 High stagnation times and S/V ratios favour DBP formation. Therefore, plumbing systems, which
have these characteristics, may be more likely at risk. However, this study did not obtain direct
evidence for this type of system.
 Plumbing systems can favour the DBP formation from disinfection of biofilms due to the high
stagnation times and S/V ratios.
 Concentrations of DBP and DCAN may exceed the UK and Colombian regulated thresholds,
under certain conditions of water quality and biofilm properties.
 Pipe diameters as small as ½ inches may lead to DBP concentrations higher than the regulated
or guidance values. Increasing the diameter to ¾ inches can significantly reduce the
concentrations, even to values under such thresholds. Such reduction are around 51% and 36%
for chloroform and DCAN, respectively.
 DCAN is the more challenging substance in terms of guidance following, compared to
chloroform, because of its potential higher toxicity, which originated a lower reference value: 20
g/L.

Chapter 7. Key findings, conclusion and future research 264

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
7.2.4 DBP formation potentials from biofilm chlorination under hydrodynamic conditions

The development and analysis of the two-dimensional model enabled the study of the parameters
influencing the formation of chloroform and DCAN under bulk flow. The key findings are the
following:

 Higher chlorine concentration, biofilm thickness, cell density, and smaller pipe diameter led to
higher concentrations of DBPs.
 Faster mass transport at the biofilm surface was found for higher Re, smaller pipe diameter,
higher initial concentrations of chlorine and cells, and higher biofilm thickness.
 A linear relationship was identified between Re and Sh.
 Compliance of chloroform and DCAN regulations is linked to the length of the water network the
pipe diameter, and flow regime. Higher chemical risk may occur in long networks, smaller pipes
and lower flow rates related to transitional flow.

7.2.5 Practical recommendations on biofilm and disinfection by-products control in

drinking water networks

The practical recommendations below are the result of the analysis and discussion of the results
included in Chapters 3, 4, and 5.

 To adapt the water industry to climate change from the management, technical, and
infrastructure points of view.
 To develop and implement the Water Safety Plans for water supply systems, which considers
that O&M and these systems should be based in risk management, from the catchment to the
tap.
 To keep fluent communication between treatment and distribution departments in water utilities
in order to coordinate activities, optimize resources and make decisions together. This could
improve the service provided to customers in terms of drinking water quality and service
continuity.
 To upgrade the treatment processes for minimization of organic matter, nutrients and biomass
in treated water in order to reduce the biofilm growth and DBP formation in the DWDNs.
 In order to control both microbial and chemical risks in the DWDNs, it is necessary to design
and apply an integral plan of O&M of DWDNs. Such plan must include replacement of ageing
pipelines, flushing pipes, cleaning storage tanks, reduction of water age, maintaining positive
pressures, and guaranteeing continuous water supply.

Chapter 7. Key findings, conclusion and future research 265

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 To avoid metal and cement materials in new pipelines. Bio-stability concept should be
considered when pipe materials are being chosen for drinking water supply.
 To use predictive tools to analyse the balance between microbial and chemical risks in the
DWDNs and improve DBP estimation and exposure assessment in DWDNs. Biofilm contribution
to DBP concentrations in bulk water must be included in such tools.
 To promote closer relationship among academy, water industry and regulatory agencies in order
to improve communication, transfer knowledge and impact of water quality research in the water
industry.

7.3 FUTURE RESEARCH

 A long-term study to characterize pathogenic microorganisms by molecular methods, in both

bulk water and biofilms, in Cali’s DWDN. Techniques such as q-PCR and microscopy should be
applied to identify the presence and viability of pathogenic bacteria. Biofilms should be growth
in coupons installed directly in the network in order to monitor several points of the network,
without affecting the normal operation of it (Douterelo et al., 2014a). This is important because
previous study and the current one have identified a potential risk due to the high leak rate this
network presents. Furthermore, the DNA of potential pathogenic bacteria was found in either
biofilm or bulk water sampled in Cali’s DWDN.
 To identify the limiting-nutrient for bacterial regrowth in Cali’s DWDN. This would help to water
managers to define the goal of future plans for upgrading the WTPs in Cali’s supply system.
 To continue investigation on laboratory techniques to improve measuring DBPs and monitoring
biofilms. Better microscopy techniques are needed to properly measure thickness of biofilms
found in real scale DWDNs. Moreover, to develop micro-sensors are required to measure DBP
concentrations at different depths within biofilms. Improved laboratory analytical capacities are
fundamental for the progress of research on DBPs and biofilms in drinking water supply.
 To assess the high flow variation in DWDNs on the detachment of biofilms, and to evaluate the
impact of detached biofilm clusters on the transport and fate of pathogens and DBP formation.
Loss of hydraulic integrity may increase in the future due to more frequent interruption of normal
operation of the DWDNs. Therefore, recurrent hydraulic changes may lead the detachment of
biofilm clusters, which can react with free chlorine and increase the DBP concentrations.
Furthermore, clusters mobilization may also favour the biofilm colonization of clean pipe walls
and transport pathogens to the water point of use.

Chapter 7. Key findings, conclusion and future research 266

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 To include the contribution of DBP formation in DWDNs as concentration pulses, lasting the
same time than bacterial regrowth takes place. This would allow the indirect incorporation of
biofilm growth simultaneously with DBP formation from disinfectant-biomass interactions, with
potential reduction of computational cost.

Chapter 7. Key findings, conclusion and future research 267

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 8 REFERENCES

Abe, Y. et al. 2012. Cohesiveness and hydrodynamic properties of young drinking water biofilms.
Water Research. 46(4), pp.1155-1166.

Abokifa, A.A. et al. 2016a. Investigating the role of biofilms in trihalomethane formation in water
distribution systems with a multicomponent model. Water Research. 104, pp.208-219.

Abokifa, A.A. et al. 2016b. Water quality modeling in the dead end sections of drinking water
distribution networks. Water Research. 89, pp.107-117.

Abraham, J.P. et al. 2009. Heat transfer in all pipe flow regimes: laminar, transitional/intermittent,
and turbulent. International Journal of Heat and Mass Transfer. 52(3), pp.557-563.

Adin, A. et al. 1991. Trihalomethane formation in chlorinated drinking water: A kinetic model. Water
Research. 25(7), pp.797-805.

Ahmadi, M. et al. 2013. Failures analysis of water distribution network during 2006-2008 in Ahvaz,
Iran. Journal of Advances in Environmental Health Research. [Online]. 1(2), pp.129-137.
Available from:
http://jaehr.muk.ac.ir/index.php/jaehr/article/view/article_40134_004e6287758dbc4a925c4feadf
4b9522.pdf

Al-Jasser, A.O. 2007. Chlorine decay in drinking-water transmission and distribution systems: Pipe
service age effect. Water Research. 41(2), pp.387-396.

Al-Omari, A. et al. 2004. Modeling trihalomethane formation for Jabal Amman water supply in
Jordan. Environmental Modeling & Assessment. 9(4), p245.

Antoun, E.N. et al. 1999. Unidirectional flushing: A powerful tool. American Water Works
Association Journal. 91(7), pp.62-71.

Arboleda, V.J. 2000. Teoría y Práctica de la Purificación del Agua. Bogotá: Mc Graw Hill.

Babaei, A.A. et al. 2015. Trihalomethanes formation in Iranian water supply systems: predicting
and modeling. Journal of Water and Health. 13(3), pp.859-869.

Badar, M. et al. 2015. Removal of cyanobacterial toxins from drinking water sources by aluminium
sulphate treatment. Brazilian Journal of Biological Sciences. [Online]. 2(3), pp. 135-145.
Available from: http://revista.rebibio.net/v2n3/v02n03a14.html

Barbeau, B. et al. 2005. Dead-end flushing of a distribution system: Short and long-term effects on
water quality. Journal of Water Supply: Research and Technology - Aqua. 54(6), p371.

Barrett, S.E. et al. 2000. Natural Organic Matter and Disinfection By-Products: Characterization
and Control in Drinking Water—An Overview. Natural Organic Matter and Disinfection By-
Products. American Chemical Society, pp.2-14.

Bautista-de los Santos, Q.M. et al. 2016. The impact of sampling, PCR, and sequencing replication
on discerning changes in drinking water bacterial community over diurnal time-scales. Water
Research. [Online]. 90, pp.216-224. [Accessed 3/1/]. Available from:
http://www.sciencedirect.com/science/article/pii/S0043135415304048

Chapter 8. References 268

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Berry, D. et al. 2006. Microbial ecology of drinking water distribution systems. Current Opinion in
Biotechnology. 17(3), pp.297-302.

Betancourt, W.Q. and Rose, J.B. 2004. Drinking water treatment processes for removal of
Cryptosporidium and Giardia. Veterinary Parasitology. 126(1–2), pp.219-234.

Bond, T. et al. 2014. Examining the interrelationship between DOC, bromide and chlorine dose on
DBP formation in drinking water — A case study. Science of The Total Environment. 470–471(0),
pp.469-479.

Boyalla, R.L. 2004. Formation and modeling of disinfection by-products in Newfoundland

communities. Master of Engineering thesis, Memorial University of Newfoundland.

Brettar, I. and Höfle, M.G. 2008. Molecular assessment of bacterial pathogens - a contribution to
drinking water safety. Current Opinion in Biotechnology. 19(3), pp.274-280.

Buamah, R. et al. 2014. Modelling the chlorine decay process in a distribution network using a pilot
system. Water Practice & Technology. 9(4), pp.534-550.

Bull, R.J. et al. 2011. Potential carcinogenic hazards of non-regulated disinfection by-products:
Haloquinones, halo-cyclopentene and cyclohexene derivatives, N-halamines, halonitriles, and
heterocyclic amines. Toxicology. 286(1–3), pp.1-19.

Buse, H.Y. et al. 2013. Eukaryotic diversity in premise drinking water using 18S rDNA sequencing:
implications for health risks. Environmental Science and Pollution Research. 20(9), pp.6351-
6366.

Buzatu, D. et al. 2007. Diffusion Coefficients for the Ternary System Water + Chloroform + Acetic
Acid at 25 °C. Journal of Solution Chemistry. 36(11), pp.1373-1384.

Cai, W. et al. 2014. Increasing frequency of extreme El Nino events due to greenhouse warming.
Nature Clim. Change. 4(2), pp.111-116.

Cai, W. et al. 2015. Increased frequency of extreme La Nina events under greenhouse warming.
Nature Clim. Change. 5(2), pp.132-137.

Cann, K.F. et al. 2012. Extreme water-related weather events and waterborne disease.
Epidemiology and Infection. 141(4), pp.671-686.

Cantor, K.P. et al. 2010. Polymorphisms in GSTT1, GSTZ1, and CYP2E1, disinfection by-products,
and risk of bladder cancer in Spain. Environmental Health Perspectives. 118(11), pp.1545-1550.

Canuto, C. et al. 2007. Spectral Methods. Evolution to Complex Geometries and Applications to
Fluid Dynamics. New York: Springer-Verlag Berlin Heidelberg.

Caporaso, J.G. et al. 2011. Global patterns of 16S rRNA diversity at a depth of millions of
sequences per sample. Proceedings of the National Academy of Sciences. 108(Supplement 1),
pp.4516-4522.

Carr, E.L. et al. 2003. Seven novel species of Acinetobacter isolated from activated sludge.
International Journal of Systematic and Evolutionary Microbiology. 53(4), pp.953-963.

CCC, C.d.C.d.C. et al. 2016. Cali cómo vamos. Informe de Calidad de Vida en Cali 2015. Santiago
de Cali.
Chapter 8. References 269
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
CCWater, C.C.f.W. 2013. Research into customer perceptions of leakage: Report. London.

Celmer, D. et al. 2008. Impact of shear force on the biofilm structure and performance of a
membrane biofilm reactor for tertiary hydrogen-driven denitrification of municipal wastewater.
Water Research. 42(12), pp.3057-3065.

Charisiadis, P. et al. 2015. Spatial and seasonal variability of tap water disinfection by-products
within distribution pipe networks. Science of The Total Environment. 506–507, pp.26-35.

Checinska, A. et al. 2015. Bacillus and Other Spore-Forming Genera: Variations in Responses and
Mechanisms for Survival. Annual Review of Food Science and Technology. 6(1), pp.351-369.

Chen, B. and Westerhoff, P. 2010. Predicting disinfection by-product formation potential in water.
Water Research. 44(13), pp.3755-3762.

Chen, X. and Stewart, P.S. 1996. Chlorine Penetration into Artificial Biofilm Is Limited by a
Reaction−Diffusion Interaction. Environmental Science & Technology. 30(6), pp.2078-2083.

Chowdhury, S. 2016. Effects of plumbing systems on human exposure to disinfection byproducts

in water: a case study. Journal of Water and Health. 14(3), p489.

Chowdhury, S. and Champagne, P. 2008. An investigation on parameters for modeling THMs

formation. Global NEST Journal. [Online]. 10(1). [Accessed 10th June 2016]. Available from:
http://journal.gnest.org/sites/default/files/Journal%20Papers/80-91_518_Chowdhury_10-1.pdf

Chowdhury, S. et al. 2007. Fuzzy risk-based decision-making approach for selection of drinking
water disinfectants. Journal of Water Supply: Research and Technology - Aqua. 56(2), pp.75-
93.

Chowdhury, S. et al. 2009. Models for predicting disinfection byproduct (DBP) formation in drinking
waters: A chronological review. Science of The Total Environment. 407(14), pp.4189-4206.

Clarelli, F. et al. 2013. A fluid dynamics model of the growth of phototrophic biofilms. Journal of
Mathematical Biology. 66(7), pp.1387-1408.

Clark, R.M. 1998. Chlorine Demand and TTHM Formation Kinetics: A Second-Order Model. Journal
of Environmental Engineering. 124(1), pp.16-24.

Clark, R.M. 2015. The USEPA's distribution system water quality modelling program: a historical
perspective. Water and Environment Journal. 29(3), pp.320-330.

Clark, R.M. and Sivaganesan, M. 1998. Predicting chlorine residuals and formation of TTHMs in
drinking water. Journal Environmental Engineering. 124(12), pp.1203-1210.

Clark, R.M. et al. 2001. Predicting the Formation of Chlorinated and Brominated By-Products.
Journal of Environmental Engineering. 127(6), pp.493-501.

Clarke, K.R. and Warwick, R.M. 2001. Change in Marine Communities: An Approach to Statistical
Analysis and Interpretation. 2nd ed. Plymouth, UK: PRIMER-E.

Codony, F. et al. 2005. Role of discontinuous chlorination on microbial production by drinking water
biofilms. Water Research. 39(9), pp.1896-1906.

Chapter 8. References 270

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Cogan, N. 2010. An Extension of the Boundary Integral Method Applied to Periodic Disinfection of
a Dynamic Biofilm. SIAM Journal on Applied Mathematics. 70(7), pp.2281-2307.

Cogan, N.G. 2006. Effects of persister formation on bacterial response to dosing. Journal of
Theoretical Biology. 238(3), pp.694-703.

Cogan, N.G. 2008. Two-Fluid Model of Biofilm Disinfection. Bulletin of Mathematical Biology. 70(3),
pp.800-819.

Cogan, N.G. 2011. Computational exploration of disinfection of bacterial biofilms in partially blocked
channels. International Journal for Numerical Methods in Biomedical Engineering. 27(12),
pp.1982-1995.

COMSOL. 2012. COMSOL Multiphysics User's Guide Version 4.3. p.1292. Available from:
http://people.ee.ethz.ch/~fieldcom/pps-
comsol/documents/User%20Guide/COMSOLMultiphysicsUsersGuide.pdf

COMSOL. 2013. CFD Module User's Guide Version 4.4. p.654. Available from: https://lost-
contact.mit.edu/afs/pdc.kth.se/roots/ilse/v0.7/pdc/vol/comsol/4.4.248/doc/pdf/CFD_Module/CF
DModuleUsersGuide.pdf

COMSOL. 2017. What Is Mass Transfer? [Online]. [Accessed 3rd May]. Available from:
https://www.comsol.it/multiphysics/what-is-mass-transfer

Coroneo, M. et al. 2014. Biofilm growth: A multi-scale and coupled fluid-structure interaction and
mass transport approach. Biotechnology and Bioengineering. 111(7), pp.1385-1395.

Costerton, J.W. et al. 1995. Microbial biofilms. Annual review of microbiology. 49, pp.711-745.

Cumsille, P. et al. 2014. A novel model for biofilm growth and its resolution by using the hybrid
immersed interface-level set method. Computers & Mathematics with Applications. 67(1), pp.34-
51.

D’Acunto, B. and Frunzo, L. 2011. Qualitative analysis and simulations of a free boundary problem
for multispecies biofilm models. Mathematical and Computer Modelling. 53(9), pp.1596-1606.

D’Acunto, B. et al. 2015. Modeling multispecies biofilms including new bacterial species invasion.
Mathematical Biosciences. 259, pp.20-26.

DANE. 2017. Proyecciones de población. [Online]. [Accessed 24th April]. Available from:
www.dane.gov.co

Dearmont, D. et al. 1998. Costs of water treatment due to diminished water quality: A case study
in Texas. Water Resources Research. 34(4), pp.849-853.

Dechatiwongse, P. 2015. A Study of the Growth and Hydrogen Production of Cyanothece sp. ATCC
51142. Doctor of Philosophy thesis, Imperial College London.

Deen, W.M. 1998. Analysis of Transport Phenomena. New York: Oxford University Press.

Delpla, I. et al. 2009. Impacts of climate change on surface water quality in relation to drinking water
production. Environment International. 35(8), pp.1225-1233.

Chapter 8. References 271

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
DeSantis, T.Z. et al. 2006. Greengenes, a Chimera-Checked 16S rRNA Gene Database and
Workbench Compatible with ARB. Applied and Environmental Microbiology. [Online]. 72(7),
pp.5069-5072.

Di Cristo, C. et al. 2013. Modelling trihalomethanes formation in water supply systems.

Environmental Technology. 34(1), pp.61-70.

Di Nardo, A. et al. 2013. Water Network Protection from Intentional Contamination by Sectorization.
Water Resources Management. 27(6), pp.1837-1850.

Di Piazza, I. and Ciofalo, M. 2010. Numerical prediction of turbulent flow and heat transfer in
helically coiled pipes. International Journal of Thermal Sciences. 49(4), pp.653-663.

Dippong, T. et al. 2016. Mathematical modeling of the variation in water quality along the network
of water supply of Satu Mare municipality, Romania. Studia Universitatis Babes-Bolyai, Chemia.
61.

Doederer, K. et al. 2014. Factors affecting the formation of disinfection by-products during
chlorination and chloramination of secondary effluent for the production of high quality recycled
water. Water Research. 48(0), pp.218-228.

Douterelo, I. et al. 2014a. Methodological approaches for studying the microbial ecology of drinking
water distribution systems. Water Research. 65(0), pp.134-156.

Douterelo, I. et al. 2014b. The bacteriological composition of biomass recovered by flushing an

operational drinking water distribution system. Water Research. 54, pp.100-114.

Douterelo, I. et al. 2014c. Bacterial community dynamics during the early stages of biofilm formation
in a chlorinated experimental drinking water distribution system: implications for drinking water
discolouration. Journal of Applied Microbiology. 117(1), pp.286-301.

Douterelo, I. et al. 2013. Influence of hydraulic regimes on bacterial community structure and
composition in an experimental drinking water distribution system. Water Research. 47(2),
pp.503-516.

Duddu, R. et al. 2008. A combined extended finite element and level set method for biofilm growth.
International Journal for Numerical Methods in Engineering. 74(5), pp.848-870.

Duddu, R. et al. 2009. A Two-Dimensional Continuum Model of Biofilm Growth Incorporating Fluid
Flow and Shear Stress Based Detachment. Biotechnology and Bioengineering. 103(1), pp.92-
104.

Eaton, A.D. et al. 2005. Standard Methods for the Examination of Water and Wastewater 21st
edition. 21st edition ed. Baltimore: American Public Health Association.

Eberl, H.J. and Demaret, L. 2007. A finite difference scheme for a degenerated diffusion equatio
arising in microbial ecology. Electronic Journal of Differential Equations. [Online]. 15, pp.77–95.
[Accessed 4th June 2014]. Available from:
http://www.mat.ub.edu/EMIS/journals/EJDE/Monographs/Monographs/Volumes/conf-
proc/15/e1/erbel.pdf

Eberl, H.J. and Efendiev, M.A. 2003. A transient density-dependent diffusion-reaction model for the
limitation of antibiotic penetration in biofilms. Electronic Journal of Differential Equations. 10,
pp.123-142.
Chapter 8. References 272
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Eberl, H.J. et al. 2001. A New Deterministic Spatio-Temporal Continuum Model for Biofilm
Development. Journal of Theoretical Medicine. 3(3).

Eberl, H.J. et al. 2000. A three-dimensional numerical study on the correlation of spatial structure,
hydrodynamic conditions, and mass transfer and conversion in biofilms. Chemical Engineering
Science. 55(24), pp.6209-6222.

Eberl, H.J. and Sudarsan, R. 2008. Exposure of biofilms to slow flow fields: The convective
contribution to growth and disinfection. Journal of Theoretical Biology. 253(4), pp.788-807.

Eisenberg, T. et al. 2012. Isolation of Potentially Novel Brucella spp. from Frogs. American Society
for Microbiology. 78(10), pp. 3753–3755.

El-Chakhtoura, J. et al. 2015. Dynamics of bacterial communities before and after distribution in a
full-scale drinking water network. Water Research. 74(0), pp.180-190.

Elshorbagy, W.E. et al. 2000. Simulation of THM species in water distribution systems. Water
Research. 34(13), pp.3431-3439.

EMCALI EICE ESP. 2016. Informe de Gestión. Cali.

Environment Agency. 2010. The Microbiology of Drinking Water (2010) - Part 2 - Practices and
procedures for sampling. Available from:
https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/316769/MoDW-
2-232.pdf

Council Directive 98/83/EC of 3 November 1998 on the quality of water intended for human
consumption 1998.

European Union. 2015. 2015. COMMISSION DIRECTIVE (EU) 2015/1787 of 6 October 2015
amending Annexes II and III to Council Directive 98/83/EC on the quality of water intended for
human consumption.

Falkinham, O.J. et al. 2015. Opportunistic Premise Plumbing Pathogens: Increasingly Important
Pathogens in Drinking Water. Pathogens. 4(2).

Fan, Y. 2012. Chlorination of toxic cyanobacterial cells and their associated toxins. Maîtrise ès
Sciences Appliquées thesis, Université de Montréal.

Fang, F. et al. 2014. Characteristics of extracellular polymeric substances of phototrophic biofilms

at different aquatic habitats. Carbohydrate Polymers. 106(0), pp.1-6.

Fang, J. et al. 2010a. Formation of carbonaceous and nitrogenous disinfection by-products from
the chlorination of Microcystis aeruginosa. Water Research. 44(6), pp.1934-1940.

Fang, J. et al. 2010b. Characterization of algal organic matter and formation of DBPs from
chlor(am)ination. Water Research. 44(20), pp.5897-5906.

Farley, M. 2015. Water leakage trends for 2015: the changing face of water loss management.
Water21. February(1), pp.36-38.

Fish, K. et al. 2012. The structure and stability of drinking water biofilms. In: WDSA 2012: 14th
Water Distribution Systems Analysis Conference, 24-27 September 2012, Adelaide, South
Australia. Barton, A.C.T.; Engineers Australia, pp.455-467.
Chapter 8. References 273
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Fish, K. et al. 2017. Biofilm structures (EPS and bacterial communities) in drinking water distribution
systems are conditioned by hydraulics and influence discolouration. Science of The Total
Environment. 593–594, pp.571-580.

Fish, K.E. et al. 2015. Characterisation of the Physical Composition and Microbial Community
Structure of Biofilms within a Model Full-Scale Drinking Water Distribution System. PLoS ONE.
10(2), pe0115824.

Fish, K.E. et al. 2016. Characterising and understanding the impact of microbial biofilms and the
extracellular polymeric substance (EPS) matrix in drinking water distribution systems.
Environmental Science: Water Research & Technology.

Flemming, H.-C. et al. 2002. Contamination potential of biofilms in water distribution systems.
Water Science and Technology: Water Supply. 2(1), pp.271-280.

Flemming, H.C. and Wingender, J. 2010. The biofilm matrix. Nature Reviews Microbiology.
[Online]. 8(9), pp.623-633. [Accessed 18 January 2014]. Available from:
http://www.nature.com/nrmicro/journal/v8/n9/abs/nrmicro2415.html

Frei, W. 2013. Which Turbulence Model Should I Choose for My CFD Application? [Online].
Available from: https://www.comsol.com/blogs/which-turbulence-model-should-choose-cfd-
application/

Gagnon, G.A. et al. 2005. Disinfectant efficacy of chlorite and chlorine dioxide in drinking water
biofilms. Water Research. 39(9), pp.1809-1817.

Gallego, V. et al. 2005. Methylobacterium isbiliense sp. nov., isolated from the drinking water
system of Sevilla, Spain. International Journal of Systematic and Evolutionary Microbiology.
55(6), pp.2333-2337.

Gamito, S. 2010. Caution is needed when applying Margalef diversity index. Ecological Indicators.
10(2), pp.550-551.

Gang, D.D. et al. 2002. Using chlorine demand to predict TTHM and HAA9 formation. American
Water Works Association Journal. 94(10), pp.76-86.

Garcia-Villanova, R.J. et al. 1997a. Formation, evolution and modeling of trihalomethanes in the
drinking water of a town: I. At the municipal treatment utilities. Water Research. 31(6), pp.1299-
1308.

Garcia-Villanova, R.J. et al. 1997b. Formation, evolution and modeling of trihalomethanes in the
drinking water of a town: II. In the distribution system. Water Research. 31(6), pp.1405-1413.

Garcia Sanchez, D. et al. 2014. Application of sensitivity analysis in building energy simulations:
Combining first- and second-order elementary effects methods. Energy and Buildings. 68, Part
C, pp.741-750.

Ghosh, P. et al. 2013. Modeling cell-death patterning during biofilm formation. Physical Biology.
10(6), p066006.

Golfinopoulos, S.K. and Arhonditsis, G.B. 2002a. Multiple regression models: A methodology for
evaluating trihalomethane concentrations in drinking water from raw water characteristics.
Chemosphere. 47(9), pp.1007-1018.

Chapter 8. References 274

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Golfinopoulos, S.K. and Arhonditsis, G.B. 2002b. Quantitative assessment of trihalomethane
formation using simulations of reaction kinetics. Water Research. 36(11), pp.2856-2868.

Golfinopoulos, S.K. et al. 1998. Use of a multiple regression model for predicting trihalomethane
formation. Water Research. 32(9), pp.2821-2829.

Gomez-Alvarez, V. et al. 2012. Metagenomic Analyses of Drinking Water Receiving Different

Disinfection Treatments. Applied and Environmental Microbiology. [Online]. 78(17), pp.6095–
6102. Available from: http://aem.asm.org/content/early/2012/06/17/AEM.01018-12.short#cited-
by

Gordon, G. and Tachiyashiki, S. 1991. Kinetics and mechanism of formation of chlorate ion from
the hypochlorous acid/chlorite ion reaction at pH 6-10. Environmental Science & Technology.
25(3), pp.468-474.

Grayman, W.M. et al. 2009. Effects of Redesign of Water Systems for Security and Water Quality
Factors. In: World Environmental and Water Resources Congress 2009, Kansas City, Missouri.

Guimerà, X. et al. 2016. Dynamic characterization of external and internal mass transport in
heterotrophic biofilms from microsensors measurements. Water Research. 102, pp.551-560.

Hallam, N.B. et al. 2002. The decay of chlorine associated with the pipe wall in water distribution
systems. Water Research. 36(14), pp.3479-3488.

Hammes, F. et al. 2010. Assessing biological stability of drinking water without disinfectant
residuals in a full-scale water supply system. Journal of Water Supply: Research and
Technology—AQUA. 59(1), pp.pp 31–40.

Harrington, G.W. et al. 1992. Developing a Computer Model to Simulate DBP Formation During
Water Treatment. American Water Works Association Journal. 84(11), pp.78-87.

Heilmann, C. et al. 1996. Molecular basis of intercellular adhesion in the biofilm-forming

Staphylococcus epidermidis. Molecular Microbiology. 20(5), pp.1083-1091.

Henne, K. et al. 2012. Analysis of Structure and Composition of Bacterial Core Communities in
Mature Drinking Water Biofilms and Bulk Water of a Citywide Network in Germany. Applied and
Environmental Microbiology. 78(10), pp.3530-3538.

Henne, K. et al. 2013. Seasonal dynamics of bacterial community structure and composition in cold
and hot drinking water derived from surface water reservoirs. Water Research. 47(15), pp.5614-
5630.

Heydorn, A. et al. 2000. Quantification of biofilm structures by the novel computer program comstat.
Microbiology. 146(10), pp.2395-2407.

Holinger, E.P. et al. 2014. Molecular analysis of point-of-use municipal drinking water microbiology.
Water Research. 49(0), pp.225-235.

Hong, H.C. et al. 2007. Modeling of trihalomethane (THM) formation via chlorination of the water
from Dongjiang River (source water for Hong Kong's drinking water). Science of The Total
Environment. 385(1–3), pp.48-54.

Chapter 8. References 275

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Hong, H.C. et al. 2008. Yield of trihalomethanes and haloacetic acids upon chlorinating algal cells,
and its prediction via algal cellular biochemical composition. Water Research. 42(20), pp.4941-
4948.

Hozalski, M.R. et al. 2008. Degradation of Halogenated Disinfection Byproducts in Water

Distribution Systems. Disinfection By-Products in Drinking Water. American Chemical Society,
pp.334-348.

Hrudey, S.E. 2009. Chlorination disinfection by-products, public health risk tradeoffs and me. Water
Research. 43(8), pp.2057-2092.

Hrudey, S.E. et al. 2015a. Evaluating Evidence for Association of Human Bladder Cancer with
Drinking-Water Chlorination Disinfection By-Products. Journal of Toxicology and Environmental
Health, Part B. 18(5), pp.213-241.

Hrudey, S.E. et al. 2015b. Evidence for Association of Human Bladder Cancer With Chlorination
Disinfection By-Products. USA: Water Research Foundation and American Water Works
Association.

Hrudey, S.E. and Charrois, W.A. 2012. Concluding thoughts on DBPs, water quality and public
health risks. In: Hrudey, S.E. and Charrois, W.A. eds. Dinsinfection by-products and human
health. London, UK: IWA Publishing, p.304.

Huang, H. et al. 2012. Dichloroacetonitrile and Dichloroacetamide Can Form Independently during
Chlorination and Chloramination of Drinking Waters, Model Organic Matters, and Wastewater
Effluents. Environmental Science & Technology. 46(19), pp.10624-10631.

Huixian, Z. et al. 1997. Formation of POX and NPOX with chlorination of fulvic acid in water:
Empirical models. Water Research. 31(6), pp.1536-1541.

Humpage, A.R. 2012. Mutagen X: The evolving story of an extremely potent mutagen, its toxicology
and human health risk assessment. In: Hrudey, S.E. and Charrois, W.A. eds. Dinsinfection by-
products and human health. London, UK: IWA Publishing, p.304.

Hutton, G. and Varughese, M. 2016. The Costs of Meeting the 2030 Sustainable Development
Goal Targets on Drinking Water, Sanitation, and Hygiene. Summary Report. Water and
Sanitation Program and World Bank Group.

IARC. 1999. Monographs on the Evaluation of Carcinogenic Risks to Humans. Some Chemicals
that Cause Tumours of the Kidney or Urinary Bladder in Rodents and Some Other Substances.

Ibarluzea, J.M. et al. 1994. Trihalomethanes in water supplies in the San Sebastian area, Spain.
Bulletin of Environmental Contamination and Toxicology. 52(3), pp.411-418.

IDEAM. 2017. Tiempo y Clima. [Online]. [Accessed 24th April]. Available from:
http://www.ideam.gov.co/web/tiempo-y-clima/clima

IPCC. 2014. Climate Change 2014 : Synthesis Report. Contribution of Working Groups I, II and III
to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva,
Switzerland: IPCC.

Itoh, T. et al. 2011. Aciditerrimonas ferrireducens gen. nov., sp. nov., an iron-reducing
thermoacidophilic actinobacterium isolated from a solfataric field. International Journal of
Systematic and Evolutionary Microbiology. 61(6), pp.1281-1285.
Chapter 8. References 276
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
IWA Task Group on Biofilm Modeling et al. 2006. Mathematical Modeling of Biofilms. London, UK:
IWA Publishing.

Jakubowska, N. and Szeląg-Wasielewska, E. 2015. Toxic Picoplanktonic Cyanobacteria—Review.

Marine Drugs. 13(3), pp.1497-1518.

Jayakumar, J.S. et al. 2008. Experimental and CFD estimation of heat transfer in helically coiled
heat exchangers. Chemical Engineering Research and Design. 86(3), pp.221-232.

Jayathilake, P.G. et al. 2017. A mechanistic Individual-based Model of microbial communities.

PLOS ONE. 12(8), pe0181965.

Ji, P. et al. 2015. Impact of Water Chemistry, Pipe Material and Stagnation on the Building Plumbing
Microbiome. PLOS ONE. 10(10), pe0141087.

Ji, P. et al. 2017. Impact of water heater temperature setting and water use frequency on the
building plumbing microbiome. ISME J. 11(6), pp.1318-1330.

Kalmbach, S. et al. 1997. Dynamics of biofilm formation in drinking water: phylogenetic affiliation
and metabolic potential of single cells assessed by formazan reduction and in situ hybridization.
FEMS Microbiology Ecology. 22(4), pp.265-279.

Karanfil, T. et al. 2008. Recent Advances in Disinfection By-Product Formation, Occurrence,

Control, Health Effects, and Regulations. Disinfection By-Products in Drinking Water. American
Chemical Society, pp.2-19.

Kelly, J.J. et al. 2014. Temporal Variations in the Abundance and Composition of Biofilm
Communities Colonizing Drinking Water Distribution Pipes. PLoS ONE. [Online]. 9(5),
pep98542. Available from:
http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0098542

Kettler, A.J. and Goulter, I.C. 1985. An analysis of pipe breakage in urban water distribution
networks. Canadian Journal of Civil Engineering. 12(2), pp.286-293.

Khan, S.J. et al. 2015. Extreme weather events: Should drinking water quality management
systems adapt to changing risk profiles? Water Research. 85, pp.124-136.

Kiéné, L. et al. 1998. Relative importance of the phenomena responsible for chlorine decay in
drinking water distribution systems. Water Science and Technology. 38(6), pp.219-227.

Kim, D.-j. et al. 2012. Relation of microbial biomass to counting units for Pseudomonas aeruginosa.
African Journal of Microbiology Research. 6(21), pp.620-4622.

Kirmeyer, G.J. et al. 2001. Practical guidelines for maintaining distribution system water quality.
American Water Works Association Journal. 93(7), pp.62-73.

Kommedal, R. et al. 2001. Modelling production of extracellular polymeric substances in a

Pseudomonas aeruginosa chemostat culture. Water Science and Technology. 43(6), p129.

Korn, C. et al. 2002. Development of chlorine dioxide-related by-product models for drinking water
treatment. Water Research. 36(1), pp.330-342.

Chapter 8. References 277

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Kristiana, I. et al. 2012. Research overview, regulatory history and current worldwide status of DBP
regulations and guidelines. In: Hrudey, S.E. and Charrois, W.A. eds. Dinsinfection by-products
and human health. London, UK: IWA Publishing, p.304.

Kumpel, E. and Nelson, K.L. 2016. Intermittent Water Supply: Prevalence, Practice, and Microbial
Water Quality. Environmental Science & Technology. 50(2), pp.542-553.

Kundukad, B. et al. 2016. Mechanical properties of the superficial biofilm layer determine the
architecture of biofilms. Soft matter. 12(26), pp.5718-5726.

Långmark, J. et al. 2005. Biofilms in an urban water distribution system: measurement of biofilm
biomass, pathogens and pathogen persistence within the Greater Stockholm area, Sweden.
Water Science and Technology. 52(8), p181.

Lau, H.Y. and Ashbolt, N.J. 2009. The role of biofilms and protozoa in Legionella pathogenesis:
implications for drinking water. Journal of Applied Microbiology. 107(2), pp.368-378.

Lechevallier, M.W. and Seidler, R.J. 1980. Staphylococcus aureus in Rural Drinking Water.
[Online]. 30(40), pp.739-742. [Accessed 1st June 2016]. Available from:
http://aem.asm.org/content/39/4/739.full.pdf

Lee, E.J. and Schwab, K.J. 2005. Deficiencies in drinking water distribution systems in developing
countries. Journal of Water and Health. 3(2), p109.

Lee, H.D. et al. 2014. A study on evaluation of the pipe wall decay constants of residual chlorine
and affecting factors in reclaimed water supply system. Desalination and Water Treatment.
53(9), pp.2378-2387.

Lehtola, M.J. et al. 2006. The effects of changing water flow velocity on the formation of biofilms
and water quality in pilot distribution system consisting of copper or polyethylene pipes. Water
Research. 40(11), pp.2151-2160.

Lehtola, M.J. et al. 2004. Removal of soft deposits from the distribution system improves the
drinking water quality. Water Research. 38(3), pp.601-610.

Leisinger, T. et al. 1994. Microbes, enzymes and genes involved in dichloromethane utilization.
Biodegradation. 5(3-4), pp.237-248.

Lekkas, T.D. and Nikolaou, A.D. 2004. Development of predictive models for the formation of
trihalomethanes and haloacetic acids during chlorination of bromide ion rich water. Water Quality
Research Journal of Canada. 39(2), pp.149-159.

Lemus Pérez, M.F. and Rodríguez Susa, M. 2017. Exopolymeric substances from drinking water
biofilms: Dynamics of production and relation with disinfection by products. Water Research.
116, pp.304-315.

Levenspiel, O. 1999. Chemical reaction engineering. Third edition ed. NJ, USA: John Wiley & Sons.

Liang, L. and Singer, P.C. 2003. Factors Influencing the Formation and Relative Distribution of
Haloacetic Acids and Trihalomethanes in Drinking Water. Environmental Science & Technology.
37(13), pp.2920-2928.

Chapter 8. References 278

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Lin, C.K. et al. 2003. The characteristics of the bacterial community structure and population
dynamics for phosphorus removal in SBR activated sludge processes. Water Research. 37(12),
pp.2944-2952.

Lindley, B. et al. 2012. Multicomponent hydrodynamic model for heterogeneous biofilms: Two-
dimensional numerical simulations of growth and interaction with flows. Physical Review E.
85(3), p031908.

Lipponen, M.T.T. et al. 2002. Occurrence of nitrifying bacteria and nitrification in Finnish drinking
water distribution systems. Water Research. 36(17), pp.4319-4329.

Liu, G. et al. 2014. Pyrosequencing Reveals Bacterial Communities in Unchlorinated Drinking

Water Distribution System: An Integral Study of Bulk Water, Suspended Solids, Loose Deposits,
and Pipe Wall Biofilm. Environmental Science & Technology. 48(10), pp.5467-5476.

Liu, G. et al. 2013. A comparison of additional treatment processes to limit particle accumulation
and microbial growth during drinking water distribution. Water Research. 47(8), pp.2719-2728.

Liu, Y. et al. 2017. Formation of disinfection byproducts from accumulated soluble products of
oleaginous microalga after chlorination. Frontiers of Environmental Science & Engineering.
11(6), p1.

Loucks, D.P. and van Beek, E. 2005. Water resources systems Planning and management. Turin:
UNESCO.

Lu, W. et al. 1999. Chlorine demand of biofilms in water distribution systems. Water Research.
33(3), pp.827-835.

Lührig, K. et al. 2015. Bacterial Community Analysis of Drinking Water Biofilms in Southern
Sweden. Microbes and Environments. 30(1), pp.99-107.

Macías-Díaz, J.E. 2015. Positive computational modelling of the dynamics of active and inert
biomass with extracellular polymeric substances. Journal of Difference Equations and
Applications. 21(4), pp.319-335.

Mahapatra, A. et al. 2015. Study of Biofilm in Bacteria from Water Pipelines. Journal of Clinical and
Diagnostic Research : JCDR. 9(3), pp.DC09-DC11.

Manuel, C.M. et al. 2007. Dynamics of drinking water biofilm in flow/non-flow conditions. Water
Research. 41(3), pp.551-562.

Manuel, C.M. et al. 2009. Unsteady state flow and stagnation in distribution systems affect the
biological stability of drinking water. Biofouling. 26(2), pp.129-139.

Marra, V. 2013. On Solvers: Multigrid Methods. [Online]. Available from:

https://uk.comsol.com/blogs/on-solvers-multigrid-methods/

Martiny, A.C. et al. 2005. Identification of Bacteria in Biofilm and Bulk Water Samples from a
Nonchlorinated Model Drinking Water Distribution System: Detection of a Large Nitrite-Oxidizing
Population Associated with Nitrospira spp. Applied and Environmental Microbiology. 71(12),
pp.8611–8617.

McGuire, M.J. et al. 2002. Information Collection Rule Data Analysis. Denver, CO.

Chapter 8. References 279

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
McMurdie, P.J. and Holmes, S. 2014. Waste Not, Want Not: Why Rarefying Microbiome Data Is
Inadmissible. PLOS Computational Biology. 10(4), pe1003531.

McNeill, L.S. and Edwards, M. 2002. The Importance of Temperature in Assessing Iron Pipe
Corrosion in Water Distribution Systems. Environmental Monitoring and Assessment. 77(3),
pp.229-242.

Menberg, K. et al. 2016. Sensitivity analysis methods for building energy models: Comparing
computational costs and extractable information. Energy and Buildings. 133, pp.433-445.

Mi, Z. et al. 2015. Impact of disinfection on drinking water biofilm bacterial community. Journal of
Environmental Sciences. 37, pp.200-205.

Miettinen, I.T. et al. 1997. Phosphorus and Bacterial Growth in Drinking Water. Applied and
Environmental Microbiology. 63(8), pp.3242–3245.

Milot, J. et al. 2000. Modeling the susceptibility of drinking water utilities to form high concentrations
of trihalomethanes. Journal of Environmental Management. 60(2), pp.155-171.

Ministerio de la Protección Social. 2007. 2007. Decreto 1575 de 2007: Por el cual se establece el
Sistema para la Protección y Control de la Calidad del Agua para Consumo Humano.

Ministerio de la Protección Social and Ministerio de Ambiente Vivienda y Desarrollo Territorial.

2007. 2007. Resolución número 2115: Por medio de la cual se señalan características,
instrumentos básicos y frecuencias del sistema de control y vigilancia para la calidad del agua
para consumo humano.

Montoya-Pachongo, C. et al. 2016. Effects of water inlet configuration in a service reservoir

applying CFD modelling. Revista Ingeniería e Investigación. 36(1), pp.31-40.

Montoya, C. et al. 2012. Evaluacion de las condiciones de mezcla y su influencia sobre el cloro
residual en tanques de compensacion de un sistema de distribucion de agua potable. Revista
Ingeniería y Ciencia. 8(5), pp.9-30.

Montoya, C. et al. 2011. Efecto del incremento en la turbiedad del agua cruda sobre la eficiencia
de procesos convencionales de potabilización. Revista EIA. 8(16), pp.137-148.

Montoya, C. et al. 2009. Propuesta metodológica para localización de estaciones de monitoreo de

calidad de agua en redes de distribución utilizando sistemas de información geográfica. Revista
Facultad de Ingeniería de la Universidad de Antioquia. 2(49), pp.129-140.

Morris, M.D. 1991. Factorial Sampling Plans for Preliminary Computational Experiments.
Technometrics. 33(2), pp.161-174.

Morrow, C.M. and Minear, R.A. 1987. Use of regression models to link raw water characteristics to
trihalomethane concentrations in drinking water. Water Research. 21(1), pp.41-48.

Muellner, M.G. et al. 2007. Haloacetonitriles vs. Regulated Haloacetic Acids: Are Nitrogen-
Containing DBPs More Toxic? Environmental Science & Technology. 41(2), pp.645-651.

Mukundan, R. and Van Dreason, R. 2014. Predicting Trihalomethanes in the New York City Water
Supply. Journal of Environmental Quality. 43(2), pp.611-616.

Chapter 8. References 280

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Nescerecka, A. et al. 2014. Biological Instability in a Chlorinated Drinking Water Distribution
Network. PLoS ONE. 9(5), pe96354.

Nieuwenhuijsen, M.J. et al. 2000. Chlorination disinfection byproducts in water and their association
with adverse reproductive outcomes: a review. Occupational and Environmental Medicine. 57(2),
pp.73-85.

Nikolaou, A.D. et al. 2004. Kinetics of the formation and decomposition of chlorination by-products
in surface waters. Chemical Engineering Journal. 100(1–3), pp.139-148.

Nokes, C.J. et al. 1999. Modelling the formation of brominated trihalomethanes in chlorinated
drinking waters. Water Research. 33(17), pp.3557-3568.

NRCNA, N.R.C.o.t.N.A. 2006. Drinking Water Distribution Systems: Assessing and Reducing
Risks. [Online]. Washington D.C: National Research Council of the National Academies.
[Accessed 7 Juanuary 2013]. Available from: http://www.kysq.org/docs/DWDS_NAS.pdf

Ocampo-Duque, W. et al. 2013. Water quality analysis in rivers with non-parametric probability
distributions and fuzzy inference systems: Application to the Cauca River, Colombia.
Environment International. 52, pp.17-28.

Park, S. et al. 1991. Batch cultivation of Methylosinus trichosporium OB3b. I: Production of soluble
methane monooxygenase. Biotechnology and Bioengineering. 38(4), pp.423-433.

Parvez, S. et al. 2011. Temporal variability in trihalomethane and haloacetic acid concentrations in
Massachusetts public drinking water systems. Environmental Research. 111(4), pp.499-509.

Patel, R.N. et al. 1982. Microbial Oxidation of Hydrocarbons: Properties of a Soluble Methane
Monooxygenase from a Facultative Methane-Utilizing Organism, Methylobacterium sp. Strain
CRL-26. Applied and Environmental Microbiology. 44(5), pp.1130-1137.

Peng, C.-Y. et al. 2010. Characterization of elemental and structural composition of corrosion
scales and deposits formed in drinking water distribution systems. Water Research. 44(15),
pp.4570-4580.

Perelman, B.L. and Ostfeld, A. 2013. Operation of remote mobile sensors for security of drinking
water distribution systems. Water Research. 47(13), pp.4217-4226.

Pérez-Vidal, A. et al. 2012. Identificación y priorización de peligros como herramientas de la gestión

del riesgo en sistemas de distribución de agua potable. Ingeniería y Universidad. 16(2), pp.449-
469.

Peyton, B.M. 1996. Effects of shear stress and substrate loading rate on Pseudomonas aeruginosa
biofilm thickness and density. Water Research. 30(1), pp.29-36.

Pianosi, F. et al. 2015. A Matlab toolbox for Global Sensitivity Analysis. Environmental Modelling &
Software. 70, pp.80-85.

Picioreanu, C. et al. 1998a. A new combined differential-discrete cellular automaton approach for
biofilm modeling: application for growth in gel beads. Biotechnology and bioengineering. 57(6),
pp.718-731.

Picioreanu, C. et al. 1998b. Mathematical Modeling of Biofilm Structure with a HybridDifferential-

Discrete CellularAutomaton Approach. Biotechnology and Bioengineering. 58(1), pp.101-116.
Chapter 8. References 281
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Picioreanu, C. et al. 2000a. Effect of diffusive and convective substrate transport on biofilm
structure formation: A two-dimensional modeling study. Biotechnology and Bioengineering.
69(5), pp.504-515.

Picioreanu, C. et al. 2000b. A theoretical study on the effect of surface roughness on mass transport
and transformation in biofilms. Biotechnology and Bioengineering. 68(4), pp.355-369.

Plewa, M.J. et al. 2008. Comparative Mammalian Cell Toxicity of N-DBPs and C-DBPs. Disinfection
By-Products in Drinking Water. American Chemical Society, pp.36-50.

Poling, E.B. et al. 2007. Physical and Chemical Data. In: Perry, R.H. and Perry, J.H. eds. Perry’s
chemical engineers’ handbook. New York: Mc Graw Hill, pp.2-456.

Prest, E. et al. 2016. Biological stability of drinking water: controlling factors, methods and
challenges. Frontiers in Microbiology. 7.

Propato, M. and Uber, J.G. 2004. Vulnerability of Water Distribution Systems to Pathogen
Intrusion: How Effective Is a Disinfectant Residual? Environmental Science & Technology.
38(13), pp.3713-3722.

Pu, Y. et al. 2013. Formation of THMs and HANs during bromination of Microcystis aeruginosa.
Journal of Environmental Sciences. 25(9), pp.1795-1799.

Rainey, F.A. et al. 2005. Extensive Diversity of Ionizing-Radiation-Resistant Bacteria Recovered

from Sonoran Desert Soil and Description of Nine New Species of the Genus Deinococcus
Obtained from a Single Soil Sample. Applied and Environmental Microbiology. 71(9), pp.5225–
5235.

Raseman, W.J. et al. 2017. Emerging investigators series: a critical review of decision support
systems for water treatment: making the case for incorporating climate change and climate
extremes. Environmental Science: Water Research & Technology. 3(1), pp.18-36.

Rataj, E. et al. 2016. Extreme weather events in developing countries and related injuries and
mental health disorders - a systematic review. BMC Public Health. 16(1), p1020.

Rathbun, R.E. 1996a. Regression equations for disinfection by-products for the Mississippi, Ohio
and Missouri rivers. Science of The Total Environment. 191(3), pp.235-244.

Rathbun, R.E. 1996b. Speciation of trihalomethane mixtures for the Mississippi, Missouri, and Ohio
Rivers. Science of The Total Environment. 180(2), pp.125-135.

Reckhow, D.A. et al. 2001. Formation and degradation of dichloroacetonitrile in drinking waters.
Journal of Water Supply: Research and Technology - Aqua. 50(1), p1.

Ren, H. et al. 2015. Pyrosequencing analysis of bacterial communities in biofilms from different
pipe materials in a city drinking water distribution system of East China. Applied Microbiology
and Biotechnology. 99(24), pp.10713-10724.

Revetta, R.P. et al. 2016. Changes in bacterial composition of biofilm in a metropolitan drinking
water distribution system. Journal of Applied Microbiology. pp.n/a-n/a.

Revetta, R.P. et al. 2011. 16S rRNA Gene Sequence Analysis of Drinking Water Using RNA and
DNA Extracts as Targets for Clone Library Development. Current Microbiology. 63(1), pp.50-59.

Chapter 8. References 282

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Richardson, S.D. et al. 2007. Occurrence, genotoxicity, and carcinogenicity of regulated and
emerging disinfection by-products in drinking water: A review and roadmap for research.
Mutation Research/Reviews in Mutation Research. 636(1–3), pp.178-242.

Roccaro, P. et al. 2013. Modeling bromide effects on yields and speciation of dihaloacetonitriles
formed in chlorinated drinking water. Water Research. 47(16), pp.5995-6006.

Rodrigues, P.M.S.M. et al. 2007. Factorial analysis of the trihalomethanes formation in water
disinfection using chlorine. Analytica Chimica Acta. 595(1–2), pp.266-274.

Rodriguez, M.J. and Sérodes, J.-B. 2001. Spatial and temporal evolution of trihalomethanes in
three water distribution systems. Water Research. 35(6), pp.1572-1586.

Rodriguez, M.J. et al. 2000. Estimation of water utility compliance with trihalomethane regulations
using a modelling approach. Journal of Water Supply: Research and Technology - Aqua. 49(2),
p57.

Roeder, R.S. et al. 2010. Long-term effects of disinfectants on the community composition of
drinking water biofilms. International Journal of Hygiene and Environmental Health. 213(3),
pp.183-189.

Rojas, A. et al. 2001. Synergism between Phyllobacterium sp. (N2-fixer) and Bacillus licheniformis
(P-solubilizer), both from a semiarid mangrove rhizosphere. FEMS Microbiology Ecology. 35(2),
pp.181-187.

Rondon, M.R. et al. 2000. Cloning the Soil Metagenome: a Strategy for Accessing the Genetic and
Functional Diversity of Uncultured Microorganisms. Applied and Environmental Microbiology.
66(6), pp.2541-2547.

Rook, J.J. 1974. Formation of haloforms during chlorination of natural water. Water Treatment and
Examination. 23(1), pp.234-243.

Rosario-Ortiz, F. et al. 2016. How do you like your tap water? Science. 351(6276), p912.

Rossman, L. et al. 1994. Modeling Chlorine Residuals in Drinking‐Water Distribution Systems.

Journal of Environmental Engineering. 120(4), pp.803-820.

RSPH, R.S.f.P.H. et al. 2017. Principles of Water Supply Hygiene.

Saltelli, A. et al. 2004. Sensitivity Analysis in Practice. Chichester, England: Wiley.

Sanger, F. et al. 1977. DNA sequencing with chain-terminating inhibitors. Proceedings of the
National Academy of Sciences. 74(12), pp.5463-5467.

Scarpino, P.V. et al. 1972. A comparative study of the inactivation of viruses in water by chlorine.
Water Research. 6(8), pp.959-965.

Scholz, H.C. et al. 2010. Brucella inopinata sp. nov., isolated from a breast implant infection.
International Journal of Systematic and Evolutionary Microbiology. 60(4), pp.801-808.

Semerjian, L. et al. 2009. Modeling the formation of trihalomethanes in drinking waters of Lebanon.
Environmental Monitoring and Assessment. 149(1), pp.429-436.

Chapter 8. References 283

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Sepkoski, J.J. 1988. Alpha, beta, or gamma: where does all the diversity go? Paleobiology. 14(3),
pp.221-234.

Sérodes, J.-B. et al. 2003. Occurrence of THMs and HAAs in experimental chlorinated waters of
the Quebec City area (Canada). Chemosphere. 51(4), pp.253-263.

Shaw, J.L.A. et al. 2014. Assessing the impact of water treatment on bacterial biofilms in drinking
water distribution systems using high-throughput DNA sequencing. Chemosphere. 117, pp.185-
192.

Shen, Y. et al. 2016. Response of Simulated Drinking Water Biofilm Mechanical and Structural
Properties to Long-Term Disinfectant Exposure. Environmental Science & Technology. 50(4),
pp.1779-1787.

Siddiqui, M. and Amy, G. 1994. Empirically and Theoretically–Based Models for Predicting
Brominated Ozonated By–Products. Ozone: Science & Engineering. 16(2), pp.157-178.

Simões, C.L. and Simões, M. 2013. Biofilms in drinking water: problems and solutions. The Royal
Society of Chemistry. (3), p14.

Simões, C.L. et al. 2007. Biofilm Interactions between Distinct Bacterial Genera Isolated from
Drinking Water. Applied and Environmental Microbiology. 73(19), pp.6192–6200.

Simões, C.L. et al. 2010. Influence of the Diversity of Bacterial Isolates from Drinking Water on
Resistance of Biofilms to Disinfection. Applied and Environmental Microbiology. 76(19),
pp.6673–6679.

Sohn, J. et al. 2004. Disinfectant decay and disinfection by-products formation model development:
chlorination and ozonation by-products. Water Research. 38(10), pp.2461-2478.

Song, F. et al. 2015. Effects of Material Properties on Bacterial Adhesion and Biofilm Formation.
Journal of Dental Research. 94(8), pp.1027-1034.

Song, R. et al. 1996. Empirical modeling of bromate formation during ozonation of bromide-
containing waters. Water Research. 30(5), pp.1161-1168.

Srivastava, S. and Bhargava, A. 2015. Biofilms and human health. Biotechnology Letters. 38(1),
pp.1-22.

Stevens, M. et al. 2004. Risk management for distribution

systems. In: Ainsworth, R. ed. Safe Piped Water Padstow: IWA Publishing, pp.121-138.

Stewart, P.S. 1994. Biofilm Accumulation Model That Predicts Antibiotic Resistance of
Pseudomonas aeruginosa Biofilms. Antimicrobial Agents and Chemotherapy. 38(5), pp.1052-
1058.

Stewart, P.S. and Raquepas, J.B. 1995. Implications of reaction-diffusion theory for the disinfection
of microbial biofilms by reactive antimicrobial agents. Chemical Engineering Science. 50(19),
pp.3099-3104.

Stover, C.K. et al. 2000. Complete genome sequence of Pseudomonas aeruginosa PAO1, an
opportunistic pathogen. Nature. 406(6799), pp.959-964.

Chapter 8. References 284

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
SUI, S.Ú.d.I.d.S.P. 2017. Servicio de Acueducto. [Online]. [Accessed 24th April]. Available from:
www.sui.gov.co

Sun, H. et al. 2014. Bacterial community of biofilms developed under different water supply
conditions in a distribution system. Science of The Total Environment. 472(0), pp.99-107.

Sung, W. et al. 2000. Modeling DBP formation. American Water Works Association Journal. 92(5),
pp.53-63.

Superintendencia de Servicios Públicos. 2006. 2006. Por la cual se establece la obligación de

reportar información por los prestadores de los servicios públicos de acueducto, alcantarillado y
aseo, a través del Sistema Unico de Información, para el cálculo de los indicadores de riesgo
de primer nivel de la Resolución CRA 315 de 11 de febrero de 2005.

Superintendencia de Servicios Públicos Domiciliarios. 2013. Informe Sectorial de los Servicios

Públicos Domiciliarios de Acueducto, Alcantarillado y Aseo - 2012 Grandes Prestadores. Bogotá
D.C.

Taherzadeh, D. et al. 2012. Mass Transfer Enhancement in Moving Biofilm Structures. Biophysical
Journal. 102(7), pp.1483-1492.

Thakre, S.S. and Joshi, J.B. 2000. CFD modeling of heat transfer in turbulent pipe flows. AIChE
Journal. 46(9), pp.1798-1812.

The British Geographer. 2017. The climate of tropical regions. [Online]. [Accessed 13th
September]. Available from: http://thebritishgeographer.weebly.com/the-climate-of-tropical-
regions.html

Theron, J. and Cloete, T.E. 2000. Molecular Techniques for Determining Microbial Diversity and
Community Structure in Natural Environments. Critical Reviews in Microbiology. 26(1), pp.37-
57.

Tierra, G. et al. 2015. Multicomponent model of deformation and detachment of a biofilm under
fluid flow. Journal of The Royal Society Interface. 12(106).

Toroz, I. and Uyak, V. 2005. Seasonal variations of trihalomethanes (THMs) in water distribution
networks of Istanbul City. Desalination. 176(1), pp.127-141.

Tosun, I. 2007. Modeling in Transport Phenomena. Second ed. The Netherlands: Elsevier.

TravelChinaGuide. 2017. Shaoxing Weather. [Online]. [Accessed 28th April]. Available from:
https://www.travelchinaguide.com/cityguides/zhejiang/shaoxing/weather/

Tsuneda, S. et al. 2003. Extracellular polymeric substances responsible for bacterial adhesion onto
solid surface. FEMS Microbiology Letters. 223(2), pp.287-292.

Tung, H.-h. and Xie, Y.F. 2009. Association between haloacetic acid degradation and heterotrophic
bacteria in water distribution systems. Water Research. 43(4), pp.971-978.

Tyrovola, K. and Diamadopoulos, E. 2005. Bromate formation during ozonation of groundwater in

coastal areas in Greece. Desalination. 176(1), pp.201-209.

UK Parliament. 2000. 2000. The Water Supply (Water Quality) Regulations 2000.

Chapter 8. References 285

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
United Nations. 2015. Sustainable Development Goal 6. [Online]. [Accessed 30th March 2017].
Available from: https://sustainabledevelopment.un.org/sdg6

Universidad de los Andes, D.d.I.C.y.A.C.d.I.e.A.y.A. 2010. 2010. Reglamento Técnico del Sector
de Agua Potable y Saneamiento Básico: Título B. Bogotá D.C.

Urano, K. et al. 1983. Empirical rate equation for trihalomethane formation with chlorination of
humic substances in water. Water Research. 17(12), pp.1797-1802.

USEPA. 1998. 1998. National primary drinking water regulations: Disinfectants and Disinfection
Byproducts.

USEPA. 2000. Chloroform. [Online]. [Accessed 11th July]. Available from:

https://www3.epa.gov/airtoxics/hlthef/chlorofo.html

USEPA. 2002a. Effects of Water Age on Distribution System Water Quality. Washington DC.

USEPA. 2002b. Finished Water Storage Facilities. Washington DC.

USEPA. 2002c. New or Repaired Water Mains. Washington DC.

USEPA. 2006. 2006. National primary drinking water regulations: stage 2 disinfectants and
disinfection by products rule: final rule.

USEPA. 2008. EPANET multi-species extension. User’s manual.

USEPA. 2016a. IRIS (Integrated Risk Information System). The Integrated Risk Information System
online database. [Online]. [Accessed 13th June 2016]. Available from: https://www.epa.gov/iris

USEPA. 2016b. Quick guide to drinking water sample collection. Golden, CO. Available from:
https://www.epa.gov/sites/production/files/2015-
11/documents/drinking_water_sample_collection.pdf

Uyak, V. and Toroz, I. 2007. Investigation of bromide ion effects on disinfection by-products
formation and speciation in an Istanbul water supply. Journal of Hazardous Materials. 149(2),
pp.445-451.

Uyak, V. et al. 2005. Monitoring and modeling of trihalomethanes (THMs) for a water treatment
plant in Istanbul. Desalination. 176(1), pp.91-101.

Vannecke, T.P.W. et al. 2015. Considering microbial and aggregate heterogeneity in biofilm reactor
models: how far do we need to go? Water Science and Technology. 72(10), pp.1692-1699.

Vasconcelos, J.J. et al. 1997. Kinetics of chlorine decay. American Water Works Association.
Journal. 89(7), p54.

Villanueva, C.M. et al. 2007. Bladder Cancer and Exposure to Water Disinfection By-Products
through Ingestion, Bathing, Showering, and Swimming in Pools. American Journal of
Epidemiology. 165(2), pp.148-156.

Vreeburg, I.J.H.G. and Boxall, D.J.B. 2007. Discolouration in potable water distribution systems: A
review. Water Research. 41(3), pp.519-529.

Chapter 8. References 286

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Vreeburg, J.H.G. et al. 2008. How Effective Is Flushing of Cast Iron Pipes? In: Van Zyl, J.E., et al.,
eds. 10th Annual Water Distribution Systems Analysis Conference WDSA 2008, August 17-20,
2008, Kruger National Park, South Africa. American Society of Civil Engineers, pp.559-570.

Wang, D. et al. 2011. Biodeterioration of asbestos cement (AC) pipe in drinking water distribution
systems. International Biodeterioration & Biodegradation. 65(6), pp.810-817.

Wang, H. et al. 2012a. Effects of disinfectant and biofilm on the corrosion of cast iron pipes in a
reclaimed water distribution system. Water Research. 46(4), pp.1070-1078.

Wang, H. et al. 2014. Effect of Disinfectant, Water Age, and Pipe Materials on Bacterial and
Eukaryotic Community Structure in Drinking Water Biofilm. Environmental Science &
Technology. 48(3), pp.1426-1435.

Wang, J.-J. et al. 2013a. Disinfection byproduct formation from chlorination of pure bacterial cells
and pipeline biofilms. Water Research. 47(8), pp.2701-2709.

Wang, Q. and Zhang, T. 2010a. Review of mathematical models for biofilms. Solid State
Communications. 150(21–22), pp.1009-1022.

Wang, Z.-P. and Zhang, T. 2010b. Characterization of soluble microbial products (SMP) under
stressful conditions. Water Research. 44(18), pp.5499-5509.

Wang, Z. et al. 2013b. Relative Contribution of Biomolecules in Bacterial Extracellular Polymeric

Substances to Disinfection Byproduct Formation. Environmental Science & Technology. 47(17),
pp.9764-9773.

Wang, Z. et al. 2012b. The role of extracellular polymeric substances on the sorption of natural
organic matter. Water Research. 46(4), pp.1052-1060.

Wang, Z. et al. 2012c. Influence of Bacterial Extracellular Polymeric Substances on the Formation
of Carbonaceous and Nitrogenous Disinfection Byproducts. Environmental Science &
Technology. 46(20), pp.11361-11369.

Water Research Foundation. 2014. Water Quality Impacts of Extreme Weather-Related Events
[Project #4324].

Wei, J. et al. 2010. Spatial and temporal evaluations of disinfection by-products in drinking water
distribution systems in Beijing, China. Science of The Total Environment. 408(20), pp.4600-
4606.

Weinberg, H.S. 2009. Modern approaches to the analysis of disinfection by-products in drinking
water. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and
Engineering Sciences. 367(1904), pp.4097-4118.

Weishaar, J.L. et al. 2003. Evaluation of Specific Ultraviolet Absorbance as an Indicator of the
Chemical Composition and Reactivity of Dissolved Organic Carbon. Environmental Science &
Technology. 37(20), pp.4702-4708.

Wen, G. et al. 2014. Using coagulation to restrict microbial re-growth in tap water by phosphate
limitation in water treatment. Journal of Hazardous Materials. 280(0), pp.348-355.

Westerhoff, P. et al. 2000. Applying DBP models to full-scale plants. American Water Works
Association Journal. 92(3), pp.89-102.
Chapter 8. References 287
Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Westrin, B.A. and Axelsson, A. 1991. Diffusion in gels containing immobilized cells: A critical
review. Biotechnology and Bioengineering. 38(5), pp.439-446.

Wheeler, Q.D. 2008. Introductory toward the New Taxonomy. In: Wheeler, Q.D. ed. The New
Taxonomy. London: CRC Press.

Whiley, H. et al. 2015. Opportunistic Pathogens Mycobacterium Avium Complex (MAC) and
Legionella spp. Colonise Model Shower. Pathogens. 4(3).

Whittaker, R.H. 1975. Communities and Ecosystems. New York: MacMillan.

WHO, W.H.O. 2004. Concise International Chemical Assessment Document 58: Chloroform,
WHO, W.H.O., [Exhibition catalogue]. Geneva, Switzerland.

WHO, W.H.O. 2005. Water Safety Plans. Managing drinking-water quality from catchment to
consumer. [Online]. Switzerland: WHO. Available from:
http://www.who.int/water_sanitation_health/dwq/wsp170805.pdf

WHO, W.H.O. 2008. Guidelines for drinking-water quality: incorporating 1st and 2nd addenda,
Vol.1, Recommendations. – 3rd ed. [Online]. 3rd ed. Geneva: WHO Press. [Accessed 28th July
2014]. Available from: http://www.who.int/water_sanitation_health/dwq/gdwq3rev/en/

WHO, W.H.O. 2017. Guidelines for drinking-water quality. 4th Editon. Incorporating the 1st
addendum. [Online]. 4th ed. Geneva: WHO Press. [Accessed 20th February 2017]. Available
from: http://apps.who.int/iris/bitstream/10665/254637/1/9789241549950-eng.pdf?ua=1

Williams, K.P. et al. 2010. Phylogeny of Gammaproteobacteria. Journal of Bacteriology. [Online].

192(9), pp.2305–2314. Available from: http://jb.asm.org/content/192/9/2305.short

Wingender, J. and Flemming, H.-C. 2011. Biofilms in drinking water and their role as reservoir for
pathogens. International Journal of Hygiene and Environmental Health. 214(6), pp.417-423.

Wingender, J. and Flemming, H.C. 2004. Contamination potential of drinking water distribution
network biofilms. Water Science and Technology. 49(11-12), p277.

Wolmarans, E. et al. 2005. Significance of bacteria associated with invertebrates in drinking water
distribution networks. Water Science and Technology. 52(8), p171.

Wright, J.M. et al. 2017. Disinfection By-Product Exposures and the Risk of Specific Cardiac Birth
Defects. Environmental Health Perspectives. 125(2), pp.269-277.

Wright, R. et al. 2014. Adaptive water distribution networks with dynamically reconfigurable
topology. Journal of Hydroinformatics. 16(6), p1280.

Xavier, J.B. et al. 2005a. A framework for multidimensional modelling of activity and structure of
multispecies biofilms. Environmental Microbiology. 7(8), pp.1085-1103.

Xavier, J.d.B. et al. 2005b. A general description of detachment for multidimensional modelling of
biofilms. Biotechnology and Bioengineering. 91(6), pp.651-669.

Xavier, M.N. et al. 2010. Pathogenesis of Brucella spp. The Open Veterinary Science Journal. 4,
pp. 109-118

Chapter 8. References 288

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
Xie, P. et al. 2013. Comparison of Permanganate Preoxidation and Preozonation on Algae
Containing Water: Cell Integrity, Characteristics, and Chlorinated Disinfection Byproduct
Formation. Environmental Science & Technology. 47(24), pp.14051-14061.

Xue, Z. et al. 2013. Pseudomonas aeruginosa inactivation mechanism is affected by capsular

extracellular polymeric substances reactivity with chlorine and monochloramine. 83(1), pp.101-
111.

Xue, Z. et al. 2014. Selective Reactivity of Monochloramine with Extracellular Matrix Components
Affects the Disinfection of Biofilm and Detached Clusters. Environmental Science & Technology.
48(7), pp.3832-3839.

Xue, Z. et al. 2012. Multiple Roles of Extracellular Polymeric Substances on Resistance of Biofilm
and Detached Clusters. Environmental Science & Technology. 46(24), pp.13212-13219.

Xue, Z. and Seo, Y. 2013. Impact of Chlorine Disinfection on Redistribution of Cell Clusters from
Biofilms. Environmental Science & Technology. 47(3), pp.1365-1372.

Yang, X. et al. 2010. Nitrogenous disinfection byproducts formation and nitrogen origin exploration
during chloramination of nitrogenous organic compounds. Water Research. 44(9), pp.2691-
2702.

Yoon, J.-H. et al. 2005. Proposal of the genus Thermoactinomyces sensu stricto and three new
genera, Laceyella, Thermoflavimicrobium and Seinonella, on the basis of phenotypic,
phylogenetic and chemotaxonomic analyses. International Journal of Systematic and
Evolutionary Microbiology. 55(1), pp.395-400.

Yorkshire Water. 2013. Our Blueprint for Yorkshire. The right outcome for Yorkshire. Our Wholesale
Water Business Plan. Yorkshire Water.

Zhang, C. et al. 2017. Effect of pipe materials on chlorine decay, trihalomethanes formation, and
bacterial communities in pilot-scale water distribution systems. International Journal of
Environmental Science and Technology. 14(1), pp.85-94.

Zhang, P. et al. 2009. Biodegradation of Haloacetic Acids by Bacterial Isolates and Enrichment
Cultures from Drinking Water Systems. Environmental Science & Technology. 43(9), pp.3169-
3175.

Zhang, T. 2012. Modeling of Biocide Action Against Biofilm. Bulletin of Mathematical Biology. 74(6),
pp.1427-1447.

Zhang, T. et al. 2008. Phase-Field Models for Biofilms II. 2-D Numerical Simulations of Biofilm-Flow
Interaction. Communications in Computational Physics. 4(1), pp.72-101.

Zhang, T.C. and Bishop, P.L. 1994. Experimental determination of the dissolved oxygen boundary
layer and mass transfer resistance near the fluid-biofilm interface. Water Science and
Technology. 30(11), pp.47-58.

Zhao, J. et al. 2016. Modeling antimicrobial tolerance and treatment of heterogeneous biofilms.
Mathematical Biosciences. 282, pp.1-15.

Chapter 8. References 289

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 9 APPENDICES

Appendix 9-A. Water age map

Figure 9-1. Water age for subnetwork 4 at the DWDN of the city of Cali - Colombia

Chapter 9. Appendices 290

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Appendix 9-B. Normality tests of correlated parameters

Table B-1. Normality tests for water and biofilm parameters

Significance Shapiro-Wilk
Variable Classification
test
Temperature 0.001 Data do not have normal distribution
pH 0.053 Data have normal distribution
Free residual chlorine 0.318 Data have normal distribution
Total chlorine 0.189 Data have normal distribution
Total trihalomethanes 0.930 Data have normal distribution
TOC in biofilms 0.000 Data do not have normal distribution
Water age (water sampling
0.000 Data do not have normal distribution
point)
Water age (biofilm sampling
0.754 Data have normal distribution
point)
Pipe age 0.315 Data have normal distribution
Pipe diameter 0.000 Data do not have normal distribution
Dry biomass per surface area
0.014 Data have normal distribution
(minimal values)
Dry biomass per surface area
0.180 Data have normal distribution
(median values)
Dry biomass per surface area
0.000 Data do not have normal distribution
(maximal values)
Unit dry biomass (all the
0.000 Data do not have normal distribution
values)
Margalef richness index
0.001 Data do not have normal distribution
(biofilm)
Shannon diversity index
0.008 Data do not have normal distribution
(biofilm)
Margalef richness index
0.310 Data have normal distribution
(water)
Shannon diversity index
0.313 Data have normal distribution
(water)

Chapter 9. Appendices 291

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Appendix 9-C. Two-dimensional plots of the Multi-Dimensional Scaling (MDS)
analysis and dendrograms for visualization of Bray Curtis similarity index

Chapter 9. Appendices 292

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Two-dimensional plot of the Multi-Dimensional Scaling (MDS) analysis Dendrogram for visualization of Bray Curtis similarity index
Parameter: Temperature
Parameter: Water age

Chapter 9. Appendices 293

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Parameter: Free residual chlorine
Parameter: pH

Chapter 9. Appendices 294

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Parameters: Free chlorine – Water age

Figure 9-2. Bray Curtis similarities of the relative abundance percentage of species in biofilm samples – Water species

Chapter 9. Appendices 295

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Two-dimensional plot of the Multi-Dimensional Scaling (MDS) analysis Dendrogram for visualization of Bray Curtis similarity index
Transform: Square root
Resemblance: S17 Bray Curtis similarity

2D Stress: 0.12 Pipediameter

4
3
8
Parameter: Pipe diameter

12
Parameter: Unit dry biomass

Chapter 9. Appendices 296

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Two-dimensional plot of the Multi-Dimensional Scaling (MDS) analysis Dendrogram for visualization of Bray Curtis similarity index
Transform: Square root
Resemblance: S17 Bray Curtis similarity

2D Stress: 0.12 Pipeage

56.45
57.08
33.88
35.24
Parameter: Pipes age

24.55
42.81
33.77
52.85
50.96
Parameter: Pipes material

Chapter 9. Appendices 297

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Two-dimensional plot of the Multi-Dimensional Scaling (MDS) analysis Dendrogram for visualization of Bray Curtis similarity index

Parameter: Water age

Parameters: Unit dry biomass - Pipes age

Figure 9-3. Bray Curtis similarities of the relative abundance percentage of species in biofilm samples – Biofilm species

Chapter 9. Appendices 298

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Parameter: Water age
Parameter: Habitat Two-dimensional plot of the Multi-Dimensional Scaling (MDS) analysis Dendrogram for visualization of Bray Curtis similarity index

Chapter 9. Appendices 299

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering
 Two-dimensional plot of the Multi-Dimensional Scaling (MDS) analysis Dendrogram for visualization of Bray Curtis similarity index

Parameters: Water age - Habitat

Figure 9-4. Bray Curtis similarities of the relative abundance percentage of species in biofilm samples – Water and biofilm species together

Chapter 9. Appendices 300

Disinfection by-product formation from biofilm chlorination in drinking water pipes
Carolina Montoya Pachongo. School of Civil Engineering