J Fail. Anal. and Preven.

(2011) 11:56–70
DOI 10.1007/s11668-010-9410-3

TECHNICAL ARTICLE—PEER-REVIEWED

Reliability Assessment of Cogeneration Power Plant in Textile

Mill Using Fault Tree Analysis
Vallem Ramesh • R. Saravannan

Submitted: 2 June 2010 / in revised form: 27 September 2010 / Published online: 2 December 2010
 ASM International 2010

Abstract The Combined Heat and Power (CHP) Systems r Mean failure rate of a component, h1
are systems that simultaneously generate both electricity P Availability
and useful heat. It is important to analyze the reliability of Q Mean unavailability of a component
these systems to ensure the lowest level of life cycle cost. [Q-Factor = (1  Availability)]
A CHP system installed in a textile mill is considered as a q Demand failure probability
case study to assess the reliability through fault tree anal- Tr Time taken to repair a component, h
ysis (FTA). The common cause failures (CCFs) are MTBF Mean time between failures
evaluated using the b-factor model with the available data MTTR Mean time to repair
on the failure of the plant. On a detailed analysis, it is CCF Common cause failure
found that the unavailability of the plant is 8.50E03, MCS Minimum cut set
which is predominantly caused by the problems related to Ti Testing Frequency
mechanical system, subsystems of boiler, and turbine. The Tf Time to first failure, h
repair and the restoration times for these components used BE Basic events
in the fault tree analysis (FTA) are 48 and 8 h, respectively. VCB Vacuum circuit breaker
Hence, faster restoration of these components affected by MVA Mega volt ampere
shutdown/failure and implementation of reliability-cen- CHP Combined heat and power
tered maintenance (RCM) features will significantly FTA Fault tree analysis
improve the reliability of the system, thereby reducing the RCM Reliability-centered maintenance
time with respect to return on the investment. PM Preventive maintenance

Keywords Fault tree analysis  Common cause failures 

Minimal cut set
Introduction
Nomenclature
k Failure rate Cogeneration is defined as the combined generation of
b Parametric factor electric (or mechanical) and thermal energy from the same
c Component having two modes, i.e., working or initial energy source, and total fuel consumption is sig-
failed nificantly reduced when ‘‘cogeneration’’ or ‘‘combined heat
lc Component failure rate and power’’ (CHP) is applied. The overall performances of
mc Component repair rate any engineering systems depend on the operating condi-
tions of their constitutive individual elements. The
economic, eco-friendly and energy-efficient solutions are
V. Ramesh  R. Saravannan (&)
the key design constraints based on which one should
Refrigeration and Air Conditioning Laboratory, Department
of Mechanical Engineering, Anna University, Chennai, India aim for an optimum solution for any engineering problem.
e-mail: [email protected] The attainment of an optimum solution corresponds to a

123
J Fail. Anal. and Preven. (2011) 11:56–70 57

compromise reached from among different criteria and operation of CHP [11]. One such parameter for evaluat-
reflects a sustainability of engineering decisions [1]. For ing the cost effectiveness is by determining the reliability
instance, consider a CHP system, which can generate of plant in terms of CHP from the experience of its
energy and heat simultaneously, besides reducing green- manager. Such an evaluation would help in determining
house gas emissions [2]. An engineering model of such a the effectiveness with respect to the point of view of the
system includes boilers, extraction-cum-condensing tur- investor.
bines, type of fueling device, etc. There are a variety of In order to carry out the failure and repair time analysis
uncertainties that exist in the parameters governing energy as envisaged in this study, the raw data from a textile mill
production, supply, conversion, transmission, and con- in terms of statistical parameters are adopted for the reli-
sumption. Obviously, the decisions made based on the past ability calculations. A reliability analysis on this CHP plant
experience or by an explicit analysis through an expert is performed using FTA to identify the important contrib-
consultation would not be satisfactory to effectively uting factors to the unavailability of the plant. The problem
address the complexities of the system and the uncertain- of quantifying independent failure and common cause for
ties in the design parameters. Therefore, development of an failure are well researched, and analyses techniques such as
implicit approach to integrate a variety of objectives, the b-factor, binomial failure rate, and Bayesian inference,
constraints, and other activities into a general modeling to name a few, are well entrenched in the literature [12].
framework is desired [3]. Their operative analysis should For more accurate results, the common cause failure (CCF)
entail the development of new representation, modeling, analysis with the b-factor method has been considered [13].
and quantification techniques [4]. Eti et al. [5] state that a Safety and economic return is maximized in this plant
model for reliability can be developed using actual repair with process controllers that continually regulate the
and failure data, to predict operational availability, reli- operation of the plant. The controller performance assess-
ability, maintainability, and other essential operating ment techniques provide value by analyzing how well each
parameters of a system. The term ‘‘reliability’’ is measured of the individual controllers in the plant behaves relative to
as the mean time between failures (MTBF) for a system, a desired or target performance [14]. To weed off the
and the term ‘‘availability’’ is the probability that an item unfounded fears in the minds of the plant operators with
will be available as and when required or as a proportion of regard to the ability of the CHP plant to cope with the
total time that an item is available for use. tripping of the grid, the plant is controlled by a distributed
The fault tree analysis (FTA) [6] can be applied either to control system. In view of human interference in an auto-
an existing system in operation or to a new system. When mated control system, there is a need to create awareness
applied to an existing operating system, FTA can be among mill engineers about the importance of instrumen-
adopted to identify weaknesses, evaluate possible tation in the operation and maintenance of the CHP.
upgrades, monitor, and predict behavior. Furthermore, FTA The concept of supportability includes considerations of
can be used to diagnose the causes and identify the participation, autonomy, education, training, reward, rec-
potential corrective measures for an observed system fail- ognition, performance measurement, and management
ure. The fault tree is a tool used to identify and assess the information systems, against which training and trouble-
combinations of the undesired events in the context of shooting are implemented for the engineers of the plant by
system operation and its environment, which can lead to the respective distributed control system suppliers as stated
the undesired state of the system. FTA is recognized by earlier investigators [15]. For such an analysis, there are
worldwide as an important tool for evaluating safety and models that can be developed in general terms that can
reliability in system design [7]. This is an analytic tech- easily be adapted to various system configurations without
nique used for identifying and classifying hazards, and the need for major modifications [16].
calculating system reliability for both simple and complex
engineering systems [8]. By modeling the combinations of
failures, the probability of common mode failures could be Description of the Plant
predicted. Even if components are repairable, there are a
few cases where the assessment of system reliability can be Salient Features of the Plant
reduced to the assessment of system availability. This is,
for instance, the case, when systems under study are peri- The schematic representation of the CHP plant installed at
odically tested [9]. An important reason for conducting a textile plant, with all the technical data of the overall
FTA is that it is a disciplined approach enabling complete system considered for this study is depicted in Fig. 1. The
transformation, which occurs from the usual engineering power plant has a rated capacity of 4.5 MW, which is
drawings [10]. Almost many of the economic parameters operated with a biomass-fired boiler, produces steam at
can be evaluated to find the solution for the successful 26.5 TPH (tonnes per hour) at a pressure of 62 bar and at a

123
58 J Fail. Anal. and Preven. (2011) 11:56–70

STEAM PR = 62 bar
STEAM TEMP = 480°C
TURBINE ALTERNATOR 11KV EB GRID
11KV, 4.5MW, 297A
11KV, 4.5MW
1500 RPM
8250 RPM
DM PLANT PRE HEATER SUPER
HEATER

BOILER VCB
11KV, 800A
VCB VCB
11KV, 800A 11KV, 800A
TEXTILE
PROCESS
SV1
PUMP1 V1
G. TRF
11KV / 11KV
VCB
SV2 CONDENSER 11KV, 800A
V2 (SHELL & TUBE TYPE)
PUMP2
TANK

SV3 S.D. TRF

PUMP3 V3 11KV / 433V
SV4

MV PANEL
INNER FEED
PUMP

INNER
RESERVIOR OUTER
RESERVIOR

V - VALVE
OUTER FEED SV - SOLENOID VALVE
PUMP

Fig. 1 Schematic diagram of cogeneration power plant in textile mill

temperature of 480C for the purpose intended. To mini- Fault Tree Analysis
mize the ash content from the burnt out fuel, a traveling
grade type boiler is opted. The excess power which is Fault tree analysis (FTA) is a systematic methodology used
generated is evacuated through the public utility grid. The in reliability assessment to describe the combination of
substation system comprises a generator breaker with a failure events that lead to a defined system failure and
capacity of 11 kV, 800 A, four VCBs (vacuum circuit identify the weak links present. The analysis starts by
breakers) with connected control, relay panel, and a power defining an undesired event as the top event and graphi-
transformer of capacity 11 kV, 6 MVA. Other protective cally constructing a tree-like structure connecting the
devices, such as lighting arrester, surge capacitor, and events that could cause the top event. The tree is con-
metering panel with potential/current transformer have structed using logical gates such that the inputs below the
been incorporated in the circuit. gates represent failures, and outputs at the top of gates
represent a propagation of failure depending on the nature
General Causes for Failure of the Power Plant of the gate. The fault tree is completed when the basic
failure events are identified. More than one failure mode
Past experience has revealed that the turbine for a CHP/ of a single component can also be modeled wherever
cogeneration power plant application should be rugged, required [8].
preferably with ability to run at slow speeds. The problems In view of the strong modeling ability, FTA is widely
faced in maintaining the steam purity in the boilers do used in reliability analysis for both qualitative and quan-
affect the turbine having solid deposits over its blades. At titative assessment. When the fault tree is evaluated using
higher boiler pressure, one of the major contaminants, Boolean algebra to identify the combination of basic
silica, gets carried over as vapor to settle over the surface components that lead to the failure of top event, sensitive
of the turbine blades. Further, a practical reason for failure components are identified. The principal qualitative results
arises out of faulty design/manufacture, poor service net- obtained are the Minimal Cut Sets (MCS) of the top event.
work, and low inventory of essential spares. This could A cut set is a combination of basic events (BE) that can
pose major problems, specifically after the warranty peri- cause the top event. For quantitative evaluation, the basic
ods. From the failure report, it is noted that there were components are assigned a probabilistic estimate for their
initial problems in the lube oil system, and these could be occurrence and solved to obtain the probabilistic estimate
resolved by proper pre-commissioning checks. of the top event. Various sources such as plant experience,

123
J Fail. Anal. and Preven. (2011) 11:56–70 59

test data, or log reports are used to arrive at the failure system failure, the analyst proceeds down the system to
distribution of the BE. trace possible failure paths. The analyst assigns to each one
In this article, for the model developed for the cogene- of these basic failure events a failure rate. Usually, the
ration power plant, the top event is defined as total plant assigned failure rates are in terms of statistical parameters
unavailability and the component failure rate and their taken from a variety of sources. The general assumption is
restoration time observed from the log book in the plant is that each component considered ‘‘c’’ (‘‘c’’—system com-
used. ponent) has two modes (working and failed), at a failure
rate ‘‘lc’’ and repair rate ‘‘mc’’. In locations, where com-
Symbols and Logic Gates ponents are independent, i.e., both their failures and repairs
are statistically independent, some components are as good
The fault tree model developed for the cogeneration power as new after repair and at the commencement of operation
plant is given in Figs. 1–9. A number of symbols are used of such repaired component can be considered with initial
in the fault tree: time as zero. The failures of systems under study are
modeled by means of coherent fault trees.
• Basic Event (BE): A basic initiating fault requiring no
In the model developed, the failure rate of the compo-
further development, indicated in the form of circle.
nents and their restoration time has been determined from
• Transfer in and Transfer out: As the fault tree for the
the log book on the basis of the two modes referred above.
whole plant is large, the continuations are represented
Based on the failure data, the fault tree is constructed by
as ‘‘Transfer out’’ with a triangle symbol at the bottom,
identifying which of system/subsystem/components could
and the corresponding continuation in the next sheet are
have caused the failure leading to the top most event. It is
represented as ‘‘Transfer in’’ with a triangle symbol at
very important to understand the system to build an unbi-
the top of the fault tree.
ased fault tree, thus allowing an analysis of detail plant
• Common Cause Failure symbol: Redundant compo-
inspection and performance indices [7].
nents for which CCF analysis is performed by grouping
of components and is denoted by a double triangle.
Data Type and Quantification
Two commonly used logical gates are used in the fault
tree model: To carry out quantifications of a fault tree, quantitative
data need to be the inputs for the BE. In the simplest
• The AND gate ( ) whereby all inputs need to occur
form, the input data consist of probabilities for the BE of
for the output to occur.
the fault tree. Also, the need for long-term field data is of
• The OR gate ( ) whereby any input causes the output
great importance for the evaluation of technical and
to occur.
economical performances [17]. Most industrial plants
The AND gate models the redundant case and is thus have been acquiring equipment failure data for several
equivalent to the parallel block diagram. The OR gate years, but seldom have those data been analyzed in a
models the series case whereby any failure causes the top scientific manner [15]. In order to prepare the data for a
event. In probability terms, the AND gate involves multi- reliability analysis, the analyst must convert the infor-
plying probabilities of failure and the OR gate the addition mation in the equipment down-time logs into time-to-
rules. failure and time to repair from which MTBF (mean time
between failures) and MTTR (mean time to repair) can be
Failure Data determined. Equipment down-time logs may be con-
structed in a variety of formats, and the type of data in
The faults can be a chain of events that are associated with the log determines the process that must be used to
component hardware failures, human errors, software convert the log entries to life data [15]. After due plant
errors, or any other pertinent events, which can lead to inspection and detailed interaction with the engineer of
undesired events [6]. The fault tree is aimed at depicting the plant, the failure data registered in the plant log book
the logical interrelationships of BE that lead to an adverse has been further authenticated. The data on component
event, i.e., the top event of the fault tree. Further, the failure and respective restoration times to service has
analysis is aimed at assessing the reliability indices in the been sourced from the textile mill for the period from
power system with inclusion of the major components of Feb. 2002 to Dec. 2008. Original data registered in the
the system. log book with respect to time and date have been con-
The analyst repeatedly asks: ‘‘what will cause a possible verted into number of hours by simple arithmetic
failure to occur?’’ The analyst views the system from a top- calculations. Mean probability rate of failure in number
down perspective, i.e., by looking at a possible high-level of hours has been calculated and indicated in Table 1.

123
60 J Fail. Anal. and Preven. (2011) 11:56–70

Table 1 Component failure data (as observed from Feb. 2002 to Dec. 2008)
Mean failure
System/subsystem name Description Model prob./rate, h

52Low tension breaker Breaker open Repairable 3.91E04

AB mill AB mill trip—electrical Repairable 6.52E05
Alternator Alternator coolant jammed Probability 1.63E05
Plant annual shut down Unavailability due to annual shutdown Repairable 6.52E05
Automatic voltage regulator panel AVR panel maintenance Repairable 6.52E05
Boiler Bed drain line leak Repairable 7.82E04
Boiler Shutdown work Repairable 1.63E03
Boiler Boiler trip Repairable 7.82E04
Transfer bus Bus failure Repairable 6.52E05
Boiler Failure of one compartment air nozzle Repairable 2.61E04
Condenser Condenser fail Probability 1.63E05
Distributed control system-automation DCS module CI 830 card problem Repairable 6.52E05
De-aerator tank DM tank 1 fail Probability 1.63E05
De-aerator tank DM tank 2 fail Probability 1.63E05
De-aerator tank DM tank 3 fail Probability 1.63E05
Electrical panel Electrical problems Repairable 2.61E04
Steam feed pipe line Leak in feed line gasket Repairable 2.61E04
Boiler Fuel Jamming in inline Repairable 6.52E05
Boiler Furnace brick work in progress Repairable 7.82E04
Pump Feed water pump trip Repairable 6.52E05
Control & relay panel Current transformer star point melted Repairable 6.52E05
Control & relay panel High voltage trip Repairable 6.52E05
Turbine Lube oil tank valve jammed/leak Repairable 7.82E04
Control & relay panel Low voltage relay trip Repairable 2.61E04
May shut down Unavailability due to shutdown in May Repairable 3.91E04
Miniature circuit breaker Electrical contact problem Repairable 1.30E04
Motor control centre Moulded case circuit breaker trip Repairable 1.30E04
Turbine Nozzle chest leak Repairable 1.30E04
Power transfer grid Grid failure Repairable 1.96E04
Plant Plant Unavailability due to shutdown Repairable 1.63E03
Instrument transformer Potential transformer 1 fail Repairable 1.30E04
Instrument transformer Potential transformer 2 fail Repairable 1.30E04
Fuel pump Leak Probability 5.00E05
Water pipe line Pump valve leak Probability 5.00E05
Water pipe line Pump valve leak Probability 5.00E05
Control motor Rotor bearing fail Probability 1.63E05
Control motor Rotor winding failure Probability 1.63E05
Control motor Speed variation Probability 1.63E05
Control motor Stator fail Probability 1.63E05
Turbine Thrust wear trip in turbine Repairable 7.82E04
Relay circuit Under voltage relay 1 actuation Repairable 6.52E05
Relay circuit Under voltage relay 2 actuation Repairable 6.52E05
Instrument transformer Oil leakage Repairable 6.52E05
Instrument transformer Transformer changed Repairable 6.52E05
Turbine Turbine vibration probe failure Probability 1.63E05

123
J Fail. Anal. and Preven. (2011) 11:56–70 61

The following factors have been envisaged for carrying method, the CCF failure rate kc is defined as ksb, where ks
out the analysis: is the total failure rate, and b is a small fraction to account
for any implicit dependence between components which
(i) The plant performance log book does not reflect on
cannot be modeled in a fault tree.
the large number of data on simultaneous failures.
(ii) Here, the b-factor of 10% for de-mineralized tanks,
Minimal Cut Set-Based Fault Tree Analysis
pumps for various applications, instrument transform-
ers, and under-voltage relays have been assumed. A
The qualitative evaluations basically transform the fault
value of 10% assigned to the b-factor is to mainly
tree logic into logically equivalent forms that provide more
establish as to whether common cause contribution
focused information. The principal qualitative results that
does have any significant contribution on the suc-
are obtained are the minimal cut sets (MCSs) of the top
cessful operation of the plant.
event. A cut set is a combination of BE that can cause the
This would be evident from the MCS indicated in top event and is the smallest combination of BE that result
Table 2 and other figures depicting the fault tree. in the top event [6]. The minimal cut set and the frequency
duration method are used for the planning and design of
Modeling of Common Cause Failures in a Fault Tree industrial and commercial electric power distribution sys-
tems and their reliability evaluation [7]. The majority of
The CCFs are two or more component failures that occur at FTAs are based on the MCS approach. It is stated that the
the same time or occur within a relatively short interval MCS-based FTAs, the number of possible cut sets grows
and, which are attributed to a common cause. Neglecting exponentially with the size of the fault tree [20]. The MCS
these CCF contributions can result in a significant under- applied for the failure data reflects on the mechanical
estimation of the probability of the top event. The key in components as being the major contributors as indicated in
including CCFs in a fault tree is to identify the components Table 2.
that are susceptible to CCF and then properly model them
in the fault tree [6]. From a practical point of view, the
CCF events included in a plant logic model represent the FTA Modeling
inter-component dependencies, which are considered to be
potentially significant. Many of the systems require System Logical Modeling Approach
redundant components to achieve reliability specifications,
and thus, CCF is a critical factor in system-safety and The plant is modeled using the immediate cause approach,
reliability analyses. Any CCF event consists of component wherein the immediate failures that could lead to the sys-
failures that meet four criteria [12]: tem or subsystem component failure are logically
developed by using logic gates as illustrated above. Based
(1) Two or more individual components fail or are
on the component failure data collected from the plant, the
degraded;
mean probability rate of failure has been computed and is
(2) Components fail within a selected period of time;
indicated in Table 1. In this study, the fault tree is devel-
(3) Component failures result from a single common
oped considering three models based on the operational
cause or coupling mechanism; and
status of the system and sub-system component, which are
(4) Component failure occurs within the established
detailed as follows:
component boundary.
(a) Repairable component: Component which can be
The above stated criteria have been carefully considered
restored to operation from failure mode after
while developing the fault tree referred in the Figs. 2–9.
rectification.
Quantitative screening includes implementation of CCF
(b) Component that fails on demand (mentioned as
events for each of the components in a CCF group,
probability in Table 1): Repairable or Non-repairable
assignment of screening numerical values to each contri-
component that fails when required to function
bution (e.g., by the use of a presumably conservative
leading to failure of the top event.
b-factor) and solution of the fault trees [18]. The qualita-
(c) Tested component: After the failure, the component
tive aspect includes a system-logic development, where
which is rectified or replaced is tested before
potential CCF events are identified along with the mecha-
integrating the component into main operation.
nisms within the system, which can lead to CCF. A
quantitative screening is then used in deciding which of the Based on the component status model, computational
failure scenarios are relevant for the reliability analysis formulae are derived and indicated in the following sec-
[19]. As per Senthil Kumar et al. [13] in the b-factor tion. Risk Spectrum FTA software enables us to develop

123
62 J Fail. Anal. and Preven. (2011) 11:56–70

Table 2 List of minimal cut set

Sl. No. Prob. of failure Minimal cut sets

1 1.63E03 Plant shutdown

2 1.63E03 Boiler shutdown
3 7.82E04 Boiler trip
4 7.82E04 Furnace brick work
5 7.82E04 Bed drain leak
6 7.82E04 Lube oil tank
7 7.82E04 Turbine trip
8 3.91E04 May shutdown
9 2.61E04 Compartment air nozzle
10 2.61E04 Feedline gasket
11 1.31E04 Nozzle chest
12 6.52E05 FWP trip
13 6.52E05 Fuel jam
14 6.52E05 Annual shutdown
15 1.63E05 Alternator coil
16 1.63E05 Rotor winding
17 1.63E05 Variable speed drive
18 1.63E05 Stator winding
19 1.63E05 Vibrator probe
20 1.63E05 Rotor bearing
21 1.63E05 Condenser
22 5.00E06 CCF2-all
23 1.63E06 CCF4-all
24 7.65E08 LT breaker LT panel & grid
25 5.10E08 Low voltage relay Control & relay panel & grid
26 5.10E08 Electrical Grid
27 2.55E08 LT breaker Busbar
28 2.55E08 MCC breaker problem LT panel & grid
29 2.55E08 MCB trip LT DB & grid
30 1.70E08 Bus trip Electrical
31 1.70E08 Bus trip Low voltage relay
32 1.28E08 Other grid related Instrument transformer
33 1.28E08 Control & relay panel Grid related
34 1.28E08 DCS module Grid related
35 1.28E08 High voltage trip Other grid related
36 1.28E08 AVR panel maintenance Other grid related
37 1.28E08 Other grid related Instrument transformer oil
38 1.28E08 AB mill trip Other grid related
39 8.51E09 Busbar MCC panel breaker problem
40 8.51E09 Busbar MCB DB problem
41 4.26E09 AVR panel maintenance Transfer bus
42 4.26E09 Bus trip Instrument transformer oil
43 4.26E09 AB mill trip Bus
44 4.26E09 Bus trip Instrument transformer
45 4.26E09 Bus trip Control & relay panel
46 4.26E09 Bus trip DCS module
47 4.26E09 Bus trip High voltage trip
47 4.26E09 Bus trip HV trip

123
J Fail. Anal. and Preven. (2011) 11:56–70 63

Table 2 continued

Sl. No. Prob. of failure Minimal cut sets

48 2.55E09 CCF-all Other grid related
49 2.03E09 Pump1 Pump2
50 1.28E09 CCF3-all Other grid related
51 8.51E10 Bus trip CCF-all
52 4.26E10 Bus trip CCF3-all
53 2.69E12 Other grid related Instrument transformer 1 (PT) PT2
54 8.98E13 Bus trip Instrument transformer 1 (PT) PT2
55 6.74E13 Other grid related Under voltage relay 1 UV Relay2
56 2.25E13 Bus trip Under voltage relay 1 UV Relay2

Fig. 2 Unavailability of plant fault tree diagram

and model everything from the basic fault tree with logical analysis. With the models considered, the input parameters
gates to advanced fault trees with boundary conditions and as a function of time are derived accordingly:
CCF events and analyze them [21]. When the component is repairable, the input parameters
are
Fault Tree Modeling for Power Plant Q ¼ r  Tr ðEq 1Þ

The fault tree module represented in the figures is devel- When the component is a standby, i.e., it calls for
oped based on major mechanical and electrical system and demand operation, the input parameter is
subsystem components as per schematic drawing reported
earlier. While developing the module, adequate care has Q ¼ q ðEq 2Þ
been taken to ensure that the entire plant function and When the component is tested periodically, the input
failure sequence are analyzed critically. As stated above, parameter is
the three models for the components, namely, repairable,
failure on demand, and tested have been considered for the Q ¼ r  Tr þ r  ðTi þ Tf Þ ðEq 3Þ

123
64 J Fail. Anal. and Preven. (2011) 11:56–70

The ‘‘q’’ factor is derived from actual failure data event. The generated solutions provide an optimal pattern
sourced from the textile mill log book which is based on of energy resource/service allocation and facility for
the recorded observation of the engineers for a period of 5 capacity expansion options with a minimized system cost,
years. This is an input parameter taken up for analysis from maximized system reliability, and a maximized energy
the actual registered field data. security. As stated above, the evaluation of the cost
effectiveness by determining the reliability of plant in
terms of the point of view of the manager of CHP is
essential. The determination of reliability in terms of the
Reliability-Centered Maintenance (RCM)
CHP manager’s point of view using FTA has necessarily
reflected the plant unavailability to be 8.50E03.
According to the Electric Power Research Institute (EPRI),
Owing to the delay in the restoration of the plant and
RCM is a systematic consideration of system functions, the
other connected failures, there has been an increase in
way functions can fail, and a priority-based consideration
terms of payback period which prompted the investigators
of safety and economics that identifies applicable and
to carry out the reliability assessment. The consolidated
effective preventive maintenance (PM) tasks. The main
outcomes of the results on the plant unavailability and
focus of RCM is, therefore, only on the system functions,
restoration time in respect of both mechanical and elec-
and not on the system hardware. The RCM was designed to
trical components are shown in Table 3. It may be noted
balance the costs and benefits, to obtain the most cost-
that unavailability and restoration/replacement times are
effective PM program [22]. The RCM analysis is a sys-
the same for all the electrical systems and subsystem
tematic engineering methodology to identify PM
components in the plant and found to be unique as
requirement for equipments. The RCM analysis process is
indicated.
a much-repeated task, and is dependent on the RCM
From the Fig. 2, it is observed that the top event of the
analysis experience [23]. The reliability of data is neces-
plant is 8.50E03 which is the sum of MCS with a res-
sary to decide the criticality, to mathematically describe the
toration time of 48 h. Such delays can be averted by
failure process, and to optimize the time between PM-
maintaining effective inventory of critical components.
tasks. Reliability data include
While analyzing the functioning of the boiler and sub-
• Mean time to failure (MTTF); systems in Fig. 3, it is observed that the Boiler Furnace
• Mean time to repair (MTTR); Brick Work and Pipeline Leakage are the contributing
• Failure rate function k(t). factors for the plant unavailability at 7.82E04 with a
restoration time of 40 h. Such inordinate time loss could be
Based on the developed fault tree and evaluated data, the
averted by effective planning and implementation of reli-
RCM process is recommended for improving on the reli-
ability centered maintenance approach, namely, preventive
ability of the CHP plant.
and periodic maintenance techniques. The unavailability of
turbine and subsystems is also found to be very high for
such critical plant operations.
Results and Discussion The Fig. 4 reflects to lube oil leakage contributing to the
unavailability of the turbine at 7.82E04 with restoration
The essential information gathered from the FTA estab- time of 80 h which is very high for any of such critical
lishes the prioritization of the contributing factors to the top plant operation. It is recommended that adequate care is

Table 3 Components unavailability and restoration time

Sl. No. Description of failed component Q-Factor (1  P) Tr-repair time factor

Mechanical component
1. Total plant unavailability 8.50E03 …
2. Plant unavailability—annual shut down 6.52E05 4.00E00
3. Boiler unavailability 7.82E04 4.80E01
4. Unavailability of turbine 7.82E04 8.00E00
Electrical component
5. Voltage relay, instrument transformer grid 6.52E05 4.00E00
6. Motor control centre’s 6.52E05 4.00E00

123
J Fail. Anal. and Preven. (2011) 11:56–70 65

Fig. 3 Representation of the fault tree for boiler and subsystems

Fig. 4 Scheme of the fault tree for turbine and subsystems

123
66 J Fail. Anal. and Preven. (2011) 11:56–70

Fig. 5 Deaerater unavailability fault tree diagram

Fig. 6 Total power plant representation by fault tree

taken during foundation casting and alignment of the tur- remains high. As such the multiple pumps which are con-
bine and connected subsystems. nected to the deaerator ought to be checked for CL
The deaerator unavailability is at 5.00E05 as indicated alignment and the connected protective circuit ought to be
in Fig. 5. Since multiple pumps are connected to the sys- calibrated on periodical basis.
tem, the analysis is carried out with CCF mode. Here, the In the Fig. 6 representing the FTA for Mill and Power, it
demand failure probability which is an optional parameter is observed that the major cause for unavailability of the

123
J Fail. Anal. and Preven. (2011) 11:56–70 67

Fig. 7 Representations of mill and power fault tree analysis

Fig. 8 Loss of power fault tree diagram

mill is due to high level of replacement of electrical utility is required to gather advance intimation on grid
components, the ‘‘Q’’ factor being 6.52E05, which is due isolation to avert excess power generation.
to relays and control gear connected to the system. It is observed from Fig. 8 that the motor control centers
From the Fig. 7, it is found that the ‘‘Q’’ factor is (MCCs) which effectively controls the motors connected to
6.52E05. The major contributing factors in terms of com- the pumps has a ‘‘Q’’ factor of 6.52E05. This factor is
ponents are under-voltage relay, instrument transformer, and primarily due to the high-voltage connectivity in the sys-
tripping of Grid. A close liaison with power-evacuating tem. Incorporation of isolation device is warranted for

123
68 J Fail. Anal. and Preven. (2011) 11:56–70

Fig. 9 Fault tree analysis for MCC panel

effective and instantaneous isolation of faulty motors. The reduction in the running cost of the plant. Based on the
Fig. 9 represents fault tree developed for the overall plant. FTA, the following conclusions are drawn and recom-
On further analysis, it is noticed that control and relay mended for implementation:
components known for their sensitivity has significantly
• The mechanical systems’/subsystems’ components con-
contributed to the unavailability of the electrical system of
nected with boiler such as Brick Work, Bed Drain, Feed
the plant. Evacuation of generated power could not be
water Pump trip, and Jammed Fuel Line are the major
effected because of grid failure, and hence contributed to
contributing factors compared to the failure of other
the loss of power. It is recommended that a close liaison
electrical system and subsystem components.
with the public utility (Power Evacuating Body) is required
• The availability features like an effective inventory
to gather advance intimation on grid isolation.
level reduce the long downtime with respect to
From the results in Fig. 10, it is observed that the major
electrical system.
contributing factors for plant unavailability are the
• Optimizing the power output from the CHP plant
mechanical system and subsystem components, with the
extends value addition for the entire mill and therefore
other factor being the failure of the grid. Since the other
effective RCM approach and availability features so
factors’ contribution amounts to less than 9%, they do not
recommended are implemented to derive tangible
contribute phenomenally toward the optimized perfor-
financial benefits.
mance of the plant.
• The major restoration time factor is attributed to
mechanical components only, which is primarily
caused by unscheduled shut down and failures arising
Conclusions out of nonexistent maintainability techniques.
• While considering the annual benefits with respect to
The objective of promoting a CHP plant in a textile mill investment against returns, the investment proposition
is to limit the overall power consumption from the gov- in the plant under study does not meet out the projected
ernment feeders, thereby, accomplishing a substantial financial target level. Thus, the long gestation time for

123
J Fail. Anal. and Preven. (2011) 11:56–70 69

Fig. 10 Pie chart for plant unavailability

the plant restoration does have a phenomenal impact on 6. Vesely, W.E., Dugan, J., Fragola, J., Minarick, J., Railsback, J.:
the time frame for financial returns, and hence there has Fault Tree Handbook with Aerospace Applications. National
Aeronautics and Space Administration, NASA (2002)
been a spill over on the payback period by over 4 7. Volkanovski, A., Čepin, M., Mavko, B.: Application of the fault
months. tree analysis for assessment of power system reliability. Reliab.
• The financial implications which are in excess of 6.3% Eng. Syst. Saf. 94(6), 1116–1127 (2009)
on payback period could have been averted had there 8. Shalev, D.M., Tiran, J.: Condition-based fault tree analysis
(CBFTA): a new method for improved fault tree analysis (FTA).
been an effective scheduled overhauling. Reliab. Eng. Syst. Saf. 92, 1231–1241 (2007)
9. Dutuit, Y., Rauzy, A.: Approximate estimation of system reli-
ability via fault trees. Reliab. Eng. Syst. Saf. 87(2), 163–172
(2005)
References 10. Churchley, A.R.: A rationale for the reliability assessment of high
integrity mechanical systems. Reliab. Eng. 19, 59–71 (1987)
1. Mazur, V.: Fuzzy thermo economic optimization of energy- 11. Giaccone, L., Canova, A.: Economical comparison of CHP sys-
transforming systems. Appl. Energy 84, 749–762 (2007) tems for industrial user with large steam demand. Appl. Energy
2. De Paepe, M., Mertens, D.: Combined heat and power in a lib- 86, 904–914 (2009)
eralised energy market. Energy Convers. Manag. 48, 2542–2555 12. Ramirez-Marqueza, J.E., Coit, D.W.: Optimization of system
(2007) reliability in the presence of common cause failures. Reliab. Eng.
3. Cai, Y.P., Huang, G.H., Yang, Z.F., Tan, Q.: Identification Syst. Saf. 92, 1421–1434 (2007)
of optimal strategies for energy management systems plan- 13. Senthil Kumar, C., John Arul, A., Singh, O.P., Suryaprakasa Rao,
ning under multiple uncertainties. Appl. Energy 86, 480–495 K.: Reliability analysis of shutdown system. Ann. Nucl. Energy
(2009) 32, 63–87 (2005)
4. Zio, E.: Reliability engineering: old problems and new chal- 14. Howard, R., Cooper, D.J.: Performance assessment of non-self-
lenges. Reliab. Eng. Syst. Saf. 94(2), 125–141 (2009) regulating controllers in a cogeneration power plant. Appl.
5. Eti, M.C., Ogaji, S.O.T., Probert, S.D.: Reliability of the Afam Energy 86, 2121–2129 (2009)
Electric Power Generating Station, Nigeria. Appl. Energy 77(3), 15. Eti, M.C., Ogaji, S.O.T., Probert, S.D.: Integrating reliability,
309–315 (2004) availability, maintainability and supportability with risk analysis

123
70 J Fail. Anal. and Preven. (2011) 11:56–70

for improved operation of the Afam thermal power-station. Appl. 19. Kvam, P.H.: A parametric mixture—model for common-cause
Energy 84(2), 202–221 (2007) failure data. IEEE Trans. Reliab. 47(1), 30–34 (1998)
16. Thorin, E., Brand, H., Weber, C.: Long-term optimization of 20. Choi, J.S., Cho, N.Z.: A practical method for accurate quantifi-
cogeneration systems in a competitive market environment. Appl. cation of large fault trees. Reliab. Eng. Syst. Saf. 92(7), 971–982
Energy 81(2), 152–169 (2005) (2007)
17. Arifujjaman, Md., Iqbal, M.T., Quaicoe, J.E.: Reliability analysis 21. RiskSpectrum FTA advanced fault tree software tool
of grid connected small wind turbine power electronics. Appl. 22. Rausand, M.: Reliability centered maintenance. Reliab. Eng.
Energy 86, 1617–1623 (2009) Syst. Saf. 60, 121–132 (1998)
18. IAEA-TECDOC-648: Procedures for Conducting Common 23. Chenga, Z., Jiab, X., Gao, P., Wua, S., Wanga, J.: A framework
Cause Failure Analysis in Probabilistic Safety Assessment. for intelligent reliability centered maintenance analysis. Reliab.
International Atomic Energy Agency (1992) Eng. Syst. Saf. 93, 784–792 (2008)

123