Entwicklung und Charakterisierung von zementgebundenen

Baustoffen für den extrusionsbasierten 3D-Druck

Development and characterisation of cement-based

materials for extrusion-based 3D-printing

An der Fakultät Bauingenieurwesen der Technischen Universität Dresden zur Erlangung

der Würde eines

Doktor-Ingenieurs (Dr.-Ing.)

eingereichte

DISSERTATION

Vorgelegt von
Venkatesh Naidu Nerella, M. Sc.
aus Ongole

eignereicht am 12. November 2018

Tag der mündlichen Prüfung: 05. Februar 2019

Gutachter:
Univ.-Prof. Dr.-Ing. Viktor Mechtcherine
Prof. Dr. Ir. Geert De Schutter
Asst. Prof. Dr. Arnaud Perrot
Herausgeber: Prof. Dr.-Ing. Viktor Mechtcherine

Institut für Baustoffe

Fakultät Bauingenieurwesen
Technische Universität Dresden
01062 Dresden

Telefon: (0351) 463 36311

Telefax: (0351) 463 37268
E-Mail: [email protected]

Alle Rechte, auch des auszugweisen Nachdruckes, der auszugweisen oder vollständigen
Wiedergabe, der Speicherung in Datenverarbeitungsanlagen und der übersetzung, sind
vorbehalten.

Druck: Druckerei & Verlag Fabian Hille, Boderitzer Straße 21e, 01217 Dresden

CONPrint3D® image on the cover page is with courtesy of Chair of Construction

Machines (CCM), TU Dresden. Copy right belongs to CCM, and to Medienzentrum der
TU Dresden.

ISBN: 978-3-86780-618-3
 ☼
To the two Lakshmis ♥ of my life and to my sun, Arka
Abstract

Extrusion-based 3D-printing — also referred to as digital construction (DC) — involves

automated layer-by-layer deposition of 3D-printable cement-based materials (3PCs). DC
enables speedy, economic and formwork-free construction in addition to quasi-unlimited
architectural design flexibility. To ensure these advancements with DC, the rheological
and mechanical properties of 3PCs need to be accurately measured and optimised
according to the technological requirements. Inherently, the 3D-printed elements can
reveal weak layer-to-layer interfaces, which form the weakest links in the entire structure
and imply pronounced anisotropy. To fulfil the process requirements and to prevent weak
bonds, the 3PCs should exhibit particular, precisely controllable rheological properties,
such as static yield stress and its development in time due to structural build-up. The
complex and partly contradictory material requirements of 3PCs necessitates compre-
hensive experimental investigations. This is a great challenge since there are no standard
test methods for the material characterisation of 3PCs.
In this research, while following an overarching aim of developing high-performance
3PCs for on-site DC, the author focused on material characterisation. The primary
requirements to 3PCs — pumpability, extrudability, buildability and layer-to-layer bond
— are investigated. The deliverables are methods for a) strain-based measurement
of structural build-up, b) computationally estimating pumping pressures, c) inline
extrudability characterisation, d) practice-oriented determination of buildability test
parameters and e) micro- and macroscopic insights on layer-to-layer interface proper-
ties. A 3D-printing test device with a progressive cavity pump, a concentric-cylinder
Couette rheometer, ram extruder and scanning electron microscope were the major
devices applied to systematically characterise the printability of developed fine-grained
high-strength 3PCs.
The extensive analyses of the experimental results revealed that structural build-up
can be characterised with constant effective strain approach. To do so, this effective
strain must be high enough to ensure flow-onset. Simulation results obtained using
the developed CFD (computation fluid dynamics) model called virtual Sliper (slid-
ing pipe rheometer) correlated with experimental measurements. Lubricating layer
was implemented into the CFD model and its viscosity could be calculated using
Krieger-Dougherty model. Comparative analyses of ram-extrusion and viscometer tests
substantiated the significance of the proposed inline extrudability quantification method
for 3PCs. The unit extrusion energy, i.e. energy consumed per extruded unit volume of
3PC, increased with increasing yield stress and plastic viscosity of 3PCs. Buildability
tests’ parameters were identified with a practice-oriented approach. The proposed
model takes machine, labour, material costs and DC approach type into consideration.
Layer-to-layer interface properties were shown to be highly sensitive to time interval
between subsequent layers, binder composition, and substrate porosity/roughness. An
improper choice of parameters can lead to fatal consequences. Furthermore, it was
demonstrated that partial replacement of Portland cement with supplementary cemen-
titious materials causes high structuration rate (structural build-up), superior interface

i
properties and satisfactory mechanical performance.
Developed test methods and the obtained results establish a basis for accurate and reli-
able material characterisation of cementitious materials in the context of extrusion-based
digital construction.

ii
Kurzfassung

3D-Druck durch computergesteuerte selektive Betonablage ist der am häufigsten ver-

wendete Ansatz im Kontext der additiven, digitalen Fertigung im Betonbau. Dieses
schalungsfreie digitale Verfahren ermöglicht ein schnelles, wirtschaftliches Bauen und
bietet einen großen architektonischen Gestaltungsspielraum. Um die Möglichkeiten
der 3D-Druck-Technologie im Betonbau zu entfalten, müssen die rheologischen und
mechanischen Eigenschaften von Beton exakt charakterisiert werden und so eingestellt,
dass sie mit aus den Prozessparametern resultierenden Anforderungen an das Mate-
rial übereinstimmen. Des Weiteren ist die inhärente Anisotropie der 3D-gedruckten
Elemente aufgrund der schichtweisen Herstellung und Bildung von Arbeitsfugen zu
beachten. Um die aus den Prozessparametern resultierenden Anforderungen zu erfüllen
und Schwachstellen zu vermeiden, sollten die druckbare Betone bestimmte, genau kon-
trollierbare rheologische Eigenschaften wie statische Fließgrenze und ihre zeitliche En-
twicklung (aufgrund von strukturellen Aufbau- und Hydratationsreaktionen) aufweisen.
Die komplexen und teilweise widersprüchlichen Materialanforderungen erfordern um-
fangreiche experimentelle Untersuchungen. Dies ist eine große Herausforderung, da es
keine Standardtestmethoden zur Materialcharakterisierung von druckbarem Beton gibt.
Die vorliegende Arbeit fokussiert auf die Erarbeitung einer praxistauglichen Vorge-
hensweise zur Festlegung, Einstellung und Prüfung der erforderlichen Materialeigen-
schaften von frischem und erhärtendem Beton. Folgende wesentliche Anforderungen
an druckbare Betone wurden untersucht: die Pumpbarkeit, die Extrudierbarkeit, die
Verbaubarkeit sowie die Leistungsfähigkeit der Verbunds zwischen den Schichten. Im
Ergebnis der Arbeiten wurden Methoden zur a) dehnungsbasierten Messung des Struktu-
raufbaus, b) rechnerischen Abschätzung von Pumpendruck, c) Inline-Charakterisierung
der Extrudierbarkeit, d) praxisorientierten Ermittlung der Verbaubarkeit und e) mikro-
und makroskopischen Untersuchungen der Grenzflächeneigenschaften der Schichten
entwickelt. In den experimentellen Untersuchungen wurden u.a. ein Beton-3D-Drucker
mit einer Exzenterschneckenpumpe, ein Couette-Rheometer mit zylindrischer Mess-
geometrie, ein RAM Extruder und ein Rasterelektronenmikroskop eingesetzt, um die
entwickelten druckbaren, feinkörnigen, hochfesten Betone systematisch zu charakter-
isieren.
Die Analysen der umfangreichen experimentellen Ergebnisse zeigten, dass der Struk-
turaufbau des frischen Zementleims mit dem Ansatz der konstanten effektiven Dehnung
ermittelt werden kann, diese hinreichend groß sein muss, um den statische Fließ-
grenze zu erreichen (Strukturbruch). Die Simulationsergebnisse aus dem entwickel-
ten CFD-Modell (computational fluid dynamics) für Virtual SLIPER (sliding pipe
rheometer) korrelierten mit experimentellen Messungen. Im CFD-Modell wurde eine
Gleitschicht zwischen Wandung und Kernbeton implementiert und deren Viskosität
mit dem Krieger-Dougherty-Modell berechnet. Die Gültigkeit der vorgeschlagenen
Inline-Charakterisierung der Extrusion wurde durch den Vergleich der Ergebnisse
von RAM-Extrusion- und Viskometerprüfungen nachgewiesen. Die spezifische Extru-
sionsenergie, d.h. der Energieverbrauch pro extrudierte Einheitsvolumen, stieg mit

iii
zunehmender Fließgrenze und plastischer Viskosität des Betons. Die Parameter zur
Beschreibung der Verbaubarkeit wurden mit einem praxisorientierten Ansatz identi-
fiziert. Das vorgeschlagene Modell berücksichtigt Maschinen-, Arbeits-, Materialkosten
sowie die Art des 3D-Drucks. Die Eigenschaften des Verbunds zwischen den Schichten
hängen stark vom Zeitintervall zwischen der Ablage der Schichten, der Bindemit-
telzusammensetzung und der Porosität/Rauheit des Substrats ab. Eine ungünstige Pa-
rameterwahl kann zu fatalen Folgen führen. Der teilweise Ersatz von Portlandzement
durch puzzolanische Zusatzstoffe führte zu einem schnelleren Strukturaufbau, eines
verbesserten Zwischenschichtverbundes und günstigen mechanischen Eigenschaften.
Die in dieser Arbeit entwickelte Werkzeuge zur Materialcharakterisierung und die
damit ermittelten Ergebnisse schaffen eine Grundlage für eine genaue und zuverläs-
sige Materialcharakterisierung zementgebundener Materialien für den Einsatz beim
3D-Druck durch selektive Ablage.

iv
 Acknowledgements

My deepest gratitude to:

Prof. Dr.- Ing. Viktor Mechtcherine

My special thanks to:

Prof. Dr. Ir. Geert De Schutter
Asst. Prof. Dr. Arnaud Perrot

I thank:
Mathias Näther ∗ Martin Krause ∗ Marko Butler ∗ Christof Schröfl ∗ Christian Stahn
Kai-Uwe Mehlisch ∗ Tilo Günzel ∗ Fabian Israel ∗ Markin Viacheslav ∗ Praful Vijay

and
Martina Awassi ∗ Mirella Kratz ∗ Simone Hempel ∗ Łukas Dudziak ∗ Sergiy Shyshko
∗ Michaela Reichardt ∗ Martina Götze ∗ Steffen Müller and other colleagues of the
Institute of Construction Materials, TU Dresden, who supported my research.

Further, for their student project/thesis works:

Mirza A. B. Beigh ∗ Arsalan Iqbal ∗ Paul Busse ∗ Paula Reiher ∗ Egor Ivaniuk,

for the fruitful industrial collaborations:

Opterra Zement Karsdorf GmbH ∗ MC-Bauchemie Müller GmbH Co. KG ∗ Putzmeister
Engineering GmbH ∗ BAM Deutschland AG ∗ Kniele GmbH ∗ Cervenka Consulting,

and the following German organisations, for their sponsorship of my research:

Federal Ministry of the Interior, Building and Community ∗ Federal Ministry of
Education and Research ∗ Federal Ministry of Food and Agriculture ∗ German Research
Foundation.

Vielen herzlichen Dank!

Dresden, November 2018

V. N. Nerella
Vorwort des Herausgebers

Das Interesse am digitalen Betonbau steigt rapide an und damit auch die Erwartungen
an diese neue, vielversprechende Technologie. Die Methoden der digitalen Fertigung mit
Beton gewinnen immer weiter an Reife, sodass weltweit bereits mehrere Pilotprojekte
realisiert werden konnten. Jedoch sind noch etliche Hürden zu nehmen, bevor der
3D-Druck mit Beton und andere digitale Bauverfahren in die Baupraxis überführt
werden können. Eine der gröen Herausforderungen ist die Erarbeitung einer praxis-
tauglichen Vorgehensweise zur Festlegung, Einstellung und Prüfung der erforderlichen
Materialeigenschaften von frischem und erhärtendem Beton. Die Schwierigkeiten bei
der Gestaltung der rheologischen Eigenschaften von Frischbeton ergeben sich aus unter-
schiedlichen, zum Teil entgegengesetzten Anforderungen, welche durch die wesentlichen
technologischen Schritte — Materialförderung, Formung, Ablage und Belastung durch
darauffolgende Betonschichten — bestimmt werden. Die einschlägigen Methoden der
Materialcharakterisierung sind ebenfalls zu durchdenken und ggf. neu zu definieren.
Die vorliegende Dissertation befasst sich mit diesen Fragestellungen im Kontext der mit
grom Abstand am häufigsten verwendeten digitalen Fertigungsmethode, dem 3D-Druck
mittels selektiver, auf Extrusion basierender Betonablage. Herr Nerella spannt sowohl
theoretisch als auch experimentell einen weiten Bogen — von der Formulierung der
vielfältigen Anforderungen an das rheologische Verhalten von Frischbeton, über die
Entwicklung der geeigneten Betonzusammensetzungen und mehreren neuen Unter-
suchungsmethoden bis zu einer umfassenden Charakterisierung von Eigenschaften von
druckbaren Betonen in frischem und erhärtetem Zustand. Diese Breite der Forschung
innerhalb einer Doktorarbeit ist vor allem der Neuheit des Themas geschuldet. Nicht
nur am Institut für Baustoffe der TU Dresden sondern weltweit wurden erst in den
vergangenen vier bis fünf Jahren die wissenschaftlichen Grundlagen des 3D-Drucks
mit Beton nach und nach erarbeitet, wobei Herr Nerella zu einem überaus aktiven
Teilnehmer dieses Prozesses gehörte und gehört. So musste er als einer der weltweit
ersten Promovenden auf diesem Gebiet verschiedene Herausforderungen nahezu parallel
meistern und sich dazu in sehr unterschiedliche relevante Fragestellungen vertiefen, um
ein schlüssiges Gesamtergebnis zu erreichen.
Im Rahmen seiner Promotion hat Herr Nerella eine ganze Reihe von wichtigen Erken-
ntnissen erarbeitet und damit viele Kenntnislücken geschlossen. Der Kenntnisstand in
Bezug auf die additiven Fertigungsverfahren durch selektive Ablage von Betonfilamenten
wurde intensiv erweitert und vertieft. Die entwickelten Prüftechniken und Algorithmen
sowie die erarbeiteten Erkenntnisse und entwickelten theoretischen Ansätze schaffen
eine überaus solide Grundlage für viele weitere Forschungsaktivitäten auf dem Gebiet
des digitalen Betonbaus und leisten einen sehr bedeutenden Beitrag zur Weiteren-
twicklung der Beton-3D-Drucktechnologie. Sie haben damit sowohl einen wichtigen
wissenschaftlichen Mehrwert als auch klare praktische Bedeutung.
Die Arbeiten, die zur vorliegenden Dissertation führten, wurden durch die intensive
Kooperation mit der Professur für Baumaschinen (Prof. Frank Will) und dem Institut für
Baubetriebswesen (Prof. Jens Otto) der TU Dresden im Rahmen gemeinsamer Projekte

vii
befruchtet. In diesem Zusammenhang sei es den fördernden Institutionen
• Zukunft Bau im Bundesministerium des Innern, für Bau und Heimat
• Bundesministerium für Bildung und Forschung,
• Bundesministerium für Ernährung und Landwirtschaft,
• Deutsche Forschungsgemeinschaft,
und den beteiligten Firmen
• Opterra Zement Karsdorf GmbH,
• MC-Bauchemie Müller GmbH Co. KG,
• Putzmeister Engineering GmbH,
• BAM Deutschland AG,
• Kniele GmbH,
• ECT-KEMA GmbH,
• Cervenka Consulting
für die groügige und tatkräftige Unterstützung herzlich gedankt.
Vorteilhaft für die Dissertation war auch die aktive Mitarbeit von Herrn Nerella
in internationalen Gremien, die sich mit dem Thema Digitaler Betonbau befassen:
RILEM Technical Committee RILEM TC 276-DFC “Digital fabrication with cement-based
materials” (Chair: Prof. Nicolas Roussel) und fib Task Group 2.11 “Structures made by
digital fabrication” (Convenor: Asst. Prof. Costantino Menna). Des Weiteren profitierte
sie von der Kooperation von Herrn Nerella mit den Gruppen um Prof. Uwe Füssel
(Institut für Fertigungstechnik, TU Dresden), Prof. Jay Sanjayan (Swinburne University of
Technology, Melbourne, Australien) und Dr. Hiroki Ogura (Shimizu Corporation, Japan).
Für mich war die Zusammenarbeit mit Herrn Venkatesh Naidu Nerella sehr angenehm
und überaus bereichernd. Ich freue mich über deren Fortsetzung und wünsche ihm viel
Erfolg in seiner weiteren wissenschaftlichen und professionellen Entwicklung!

Viktor Mechtcherine

viii
Contents

Abstract i

Vorwort des Herausgebers vii

Notations and abbreviations xiii

1 Introduction 1
1.1 Extrusion-based digital concrete construction . . . . . . . . . . . . . . . . . 1
1.2 3D-printable cement-based composites . . . . . . . . . . . . . . . . . . . . . 2
1.2.1 Various stages in the processing of 3PCs . . . . . . . . . . . . . . . . 2
1.2.2 Properties defining printability of 3PCs for extrusion-based DC . . 5
1.3 Goal, objectives and research questions . . . . . . . . . . . . . . . . . . . . . 7
1.4 Research concept, and outline of this dissertation . . . . . . . . . . . . . . . 7
1.5 Purpose, scope and target audience . . . . . . . . . . . . . . . . . . . . . . . 10

2 Literature and theoretical research 13

2.1 Motivation and context . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1.1 The impeding aspects of conventional construction . . . . . . . . . 14
2.1.2 Additive manufacturing and the fourth industrial revolution . . . . 15
2.2 Digital construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.1 Definition and brief evolution . . . . . . . . . . . . . . . . . . . . . 16
2.2.2 Principles and process . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.3 The potential opportunities of DC . . . . . . . . . . . . . . . . . . . 22
2.2.4 Applications of DC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.2.5 Material technological aspects of DC . . . . . . . . . . . . . . . . . . 26
2.2.6 Machine technological and process aspects of DC . . . . . . . . . . . 28
2.2.7 Cost structure in DC . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.8 Challenges in large-scale implementation of digital construction . . 34
2.2.9 Concrete on-site 3D-Printing — CONPrint3D® . . . . . . . . . . . . 40
2.3 Structural build-up: importance and characterisation . . . . . . . . . . . . . 41
2.4 Pumping of concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.4.1 General introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.4.2 Analytical description of concrete pumping flow . . . . . . . . . . . 45
2.4.3 Experimental characterisation and prediction of discharge pressures 48
2.4.3.1 Pumpability investigations on 3PCs . . . . . . . . . . . . . 49
2.4.4 Numerical modelling of fresh concrete flow and pumping . . . . . . 50
2.4.5 Influence of pumping on rheological properties . . . . . . . . . . . . 51
2.5 Extrusion of concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.5.1 General information . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2.5.2 Analytical description . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.5.2.1 Progressive cavity pumps . . . . . . . . . . . . . . . . . . . 55

ix
 2.5.2.2 Ram extruders . . . . . . . . . . . . . . . . . . . . . . . . . 56
2.5.3Experimental characterisation . . . . . . . . . . . . . . . . . . . . . . 58
2.5.3.1 Extrudability investigations on 3PCs . . . . . . . . . . . . 59
2.5.4 Limitations of previously methods . . . . . . . . . . . . . . . . . . . 59
2.6 Buildability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.6.1 General introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2.6.2 Buildability investigations on 3PCs . . . . . . . . . . . . . . . . . . 61
2.6.3 Challenges in practical implementation of proposed approaches . . 62
2.7 Layer-interface properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
2.7.1 Formation of weak bonds between distinct concrete layers . . . . . . 64
2.7.2 Research on anisotropic properties of 3PCs . . . . . . . . . . . . . . 64
2.8 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

3 MATERIALS, TESTING EQUIPMENT AND GENERAL METHODS 69

3.1 Materials and mixture compositions . . . . . . . . . . . . . . . . . . . . . . 69
3.2 3D-concrete-printing test device . . . . . . . . . . . . . . . . . . . . . . . . . 70
3.2.1 Synchronization of printhead motion with concrete flowrate . . . . 70
3.3 Ram extrusion test device . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.4 General approaches used for fresh-state characterization . . . . . . . . . . . 73

4 RHEOLOGY AND STRUCTURAL BUILD-UP 77

4.1 Experimental program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
4.1.1 Mixture compositions and preparation . . . . . . . . . . . . . . . . . 77
4.1.2 Rheometer setup and testing protocol . . . . . . . . . . . . . . . . . 78
4.1.3 Applied shear rates and applied shear strain . . . . . . . . . . . . . 80
4.1.4 Calculating rheological parameters . . . . . . . . . . . . . . . . . . . 81
4.2 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.2.1 Measurements of structural build-up by means of CSR test . . . . . 82
4.2.2 Characterizing SBP applying varying shear rates . . . . . . . . . . . 85
4.2.3 Influence of the cement paste composition on structural build-up . 88
4.2.3.1 Effect of the superplasticizer dosage . . . . . . . . . . . . . 89
4.2.3.2 Influence of adding accelerator . . . . . . . . . . . . . . . . 89
4.2.3.3 Influence of replacing cement with SCM . . . . . . . . . . 91
4.3 Universal approaches for structural build-up . . . . . . . . . . . . . . . . . 92
4.4 Chapter summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . 93

5 CONCRETE TRANSPORT AND PUMPING 95

5.1 Various approaches for placement of 3PCs . . . . . . . . . . . . . . . . . . . 95
5.2 Testing pumpability of concrete with Sliper . . . . . . . . . . . . . . . . . . 98
5.2.1 Significance of developing a Sliper numerical model . . . . . . . . . 99
5.2.2 Experimental background . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3 Virtual Sliper: single-phase numerical model . . . . . . . . . . . . . . . . . 100
5.3.1 Modelling schema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5.3.2 Implementation of the Bingham model . . . . . . . . . . . . . . . . 102
5.3.3 Geometry, boundary conditions and model parameters . . . . . . . 103
5.4 ISPM results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
5.4.1 Single-phase model results . . . . . . . . . . . . . . . . . . . . . . . . 104
5.5 User-defined single-phase model . . . . . . . . . . . . . . . . . . . . . . . . 107
5.5.1 Modelling schema . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
5.5.2 UDFM results and discussion . . . . . . . . . . . . . . . . . . . . . . 109
5.6 Chapter summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . 113

x
6 EXTRUDABILITY 115
6.1 Attributes of an optimal testing methodology . . . . . . . . . . . . . . . . . 115
6.2 Proposed method for extrudability characterisation . . . . . . . . . . . . . . 117
6.3 Experimental program . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6.3.1 Mixtures and materials . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6.3.2 Production of mixtures and testing procedure . . . . . . . . . . . . . 118
6.3.3 Additional tests for comparative assessment . . . . . . . . . . . . . . 119
6.4 Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
6.4.1 Extrudability assessment from 3DPET results . . . . . . . . . . . . . 120
6.4.2 Extrusion force measurements from ram extruder tests . . . . . . . 123
6.4.3 Comparative analyses: rheological properties ⇔ extrudability . . . 124
6.5 Chapter summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . 126

7 BUILDABILITY OF 3PCS 127

7.1 Buildability criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
7.1.1 Suggested buildability criteria . . . . . . . . . . . . . . . . . . . . . . 127
7.1.2 Complementary considerations . . . . . . . . . . . . . . . . . . . . . 130
7.1.2.1 Scope of the proposed approach . . . . . . . . . . . . . . . 130
7.1.2.2 Downscaling and surrounding conditions . . . . . . . . . . 131
7.2 Application on CONPrint3D® . . . . . . . . . . . . . . . . . . . . . . . . . . 132
7.2.1 Height of the wall specimen and number of layers to be tested . . . 133
7.2.2 Time interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
7.2.3 Experimental verification . . . . . . . . . . . . . . . . . . . . . . . . 138
7.3 Buildability requirements for FP and FWP . . . . . . . . . . . . . . . . . . . 139
7.4 Chapter summary and conclusions . . . . . . . . . . . . . . . . . . . . . . . 144

8 Interface and hardened properties of 3D-printed elements 145

8.1 Materials and methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
8.1.1 Compositions, time intervals and specimen production . . . . . . . 146
8.1.2 3D-printing and mechanical testing . . . . . . . . . . . . . . . . . . 147
8.1.3 Optimized printing sequence . . . . . . . . . . . . . . . . . . . . . . 148
8.1.4 Specimen preparation for microscopy . . . . . . . . . . . . . . . . . 149
8.2 Investigations on the macroscale: results and discussion . . . . . . . . . . . 149
8.2.1 Evaluation of fresh concrete properties . . . . . . . . . . . . . . . . . 149
8.2.2 Interface properties and flexural strength . . . . . . . . . . . . . . . 150
8.2.3 Compressive strengths . . . . . . . . . . . . . . . . . . . . . . . . . . 153
8.3 Microscopic investigations: results and discussion . . . . . . . . . . . . . . . 158
8.3.1 General observations and classification . . . . . . . . . . . . . . . . . 158
8.3.2 Microstructural differences between the 3PCs C1 and C2 . . . . . . 158
8.3.3 Microstructural changes over time . . . . . . . . . . . . . . . . . . . 160
8.3.4 Process aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8.4 Chapter summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

9 Complementary research 165

9.1 On incorporating reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . 165
9.1.1 3D-printing with SHCC . . . . . . . . . . . . . . . . . . . . . . . . . 165
9.1.2 3D-printing of steel reinforcement . . . . . . . . . . . . . . . . . . . 167
9.2 On interface tensile strength and distinct layer casting . . . . . . . . . . . . 170

Conclusions and further research 173

xi
Bibliography 177

A Additional figures and tables 199

List of publications 202

xii
Notations and abbreviations

Greek symbols

αgeom Geometric factor which depends on the geometry of the de- [−]
posited layer/printed element
γap Applied shear strain [−]
γef ,min Minimum effective shear strain [−]
γef Effective shear strain [−]
γlag Shear strain lag = γap − γef [−]
γ̇ Shear rate [s−1 ]
γ̇ap Applied shear rate [s−1 ]
γ̇c Critical shear rate [s−1 ]
γ̇ef Effective shear rate [s−1 ]
∆P Pressure drop in a pipe flow [kP a]
∆Ppcp Differential pressure in a progressive cavity pump [kP a]
∆r Ram displacement [mm]
υ Velocity of extrudate in the die-land of a ram-extruder [mm/s]
η Dynamic viscosity [P a · s]
λ Wavelength of the inner filament [m]
µ, µpl Plastic viscosity [P a · s]
µf Viscosity factor [−]
µi Viscosity of the material in LL [P a · s]
τ Shear stress [P a]
τ0 Yield stress [P a]
τ0,0 Static yield stress of the material when resting time is zero [P a]
τ0,i Yield stress of the material in LL [P a]
τf Yield stress factor [−]
τ(t) Shear stress at material age t [P a]
τ0 (t) Static yield stress of the material when resting time is t [P a]
τs (t) Static yield stress [P a]
φ Solid/coarse particle volume fraction [−]
φm Maximum volume fraction [−]
∅as HFT spread diameter after strokes [mm]
∅bs HFT spread diameter before strokes [mm]
∅max Maximum aggregate size [mm]

Latin letters

Athix Constant rate of increase in the static yield stress [P a/s]

Dst Minimum stator diameter [mm]

xiii
Dstm Maximum stator diameter [mm]
Db Barrel diameter [mm]
Dr Rotor diameter [mm]
dl = d Die land, orifice diameter [mm]
E Eccentricity [mm]
e Thickness of LL [mm]
Ff ,b Force needed to overcome friction between barrel-concrete [N ]
Ff ,b0 Initial frictional force in billet [N ]
Ff ,d Force needed to overcome friction between dead-zone-concrete [N ]
Ff ,l Force needed to overcome friction between die-land-concrete [N ]
Fp,l Force needed for the plastic deformation in shaping zone [N ]
k Concrete filling coefficient in pipe flow [−]
L Pipe length in a pumping circuit [mm]
LB Billet length [mm]
Lb Barrel length [mm]
Ld Dead-zone length [mm]
Ll Die-land length [mm]
Np Rotational velocity of the pump motor [rot/s]
P Discharge pressure [kP a]
Pe Electric power consumed [W ]
Pr e Ram-extrusion pressure [kP a]
Q Flow rate [m3 /h]
Qe , Qe,pcp Extrudate flow rate [cm3 /s]
R Pipe radius in a pumping circuit [mm]
Re Reynold’s number [−]
T Torque [N m]
tage The time since addition of cement to water [min]
tc Characteristic time [s]
td,ef Effective test duration = td − tc [s]
td The time duration in which the sample is sheared during a single [s]
measurement
te Elapsed time taken by γ̇ef to reach γ̇ap [s]
tLL Layer thickness [mm]
U Voltage [V ]
u Velocity [mm/s]
VDC Average printing velocity [m/s]
Vef f ective,FP Effective horizontal velocity of printing for FP [m/s]
Vef f ective,FW P Effective horizontal velocity of printing for FWP [m/s]
Vprinthead,FP Printhead velocity for FP [m/s]
Vprinthead,FW P Printhead velocity for FWP [m/s]
−−−→
Vap Applied velocity, i.e., the velocity with which the printhead is [m/s]
commanded to traverse
−−−−→
Vef f Effective velocity with which the printhead actually traversing at [m/s]
a given point in time
−−−−→
Vmax Maximum velocity with which the printhead can be moved [m/s]
−−−→
VP H Printhead velocity [m/s]
ve,pcp Effective volume of displaced material by a PCP [m3 /s]
w Stator-rotor interference/overlap [mm]

xiv
 Acronyms

%bwob % by weight of binder

3DPET 3D-Printer Extrudability Test
3DPTD 3D-Printing Testing Device
3PC 3D-Printable Cement-based material or 3D-Printable Concrete
AS HFT spread diameter After Strokes [mm]
BKI Baukosteninformationszentrum Deutscher Architektenkam-
mern
BS HFT spread diameter Before Strokes [mm]
CC Conventional Construction
CCE Conventional Construction Equipment
CFD Computational Fluid Dynamics
CSR Constant-Shear-Rate
DC Digital Construction / Digital concrete Construction
DEM Discrete Element Method
EMC Experimental and Model Correlation
FA Fly Ash
FP Filament Printing
FWP Full-Width Printing
HFT Hägermann Flow Table
HRWRA High-Range, Water-Reducing Admixture
ISPM Initial Single-Phase CFD Model without simulating LL
LL Lubricating Layer
LVDT Linear Variable Differential Transformers
MSS Micro-Silica aqueous Suspension
OC Ordinary Concrete
PCE Polycarboxylate Ether
PCP Progressive Cavity Pump
PR Penetration-Resistance
PV Plastic viscosity µ [P a · s]
RCP1 Rotational Control Parameter when printhead is at rest [−]
RCP2 Rotational Control Parameter when printhead is moving [−]
REF Ram Extrusion Force [N ]
RET Ram-Extruder Test
RILEM RILEM Technical Committee on Digital Fabrication with
TC-DFC Cement-based materials
SAOS Small Amplitude Oscillatory Shear
SCC Self-Compacting Concretes
SCM Secondary Cementitious Materials
SEM Scanning Electron Microscope
SIPM Shear-Induced Particle Migration
SP Superplasticizer
STAR State of The Art Report
SYS Static Yield Stress [P a]
TI Time Interval [min]
TSP Traveling Salesman Problem
UCE Unconventional Construction Equipment
UDF User Defined Functions

xv
UDFM CFD Model with User-Defined Functions for simulating LL
UEE Unit Extrusion Energy [J/cm3 ]

xvi
Chapter 1

Introduction

In this chapter, digital construction (DC) and 3D-printable cement-based material

(3PC) are introduced. In addition, the research concept, scope and limits are outlined.
Furthermore, various stages of processing 3PCs in extrusion-based DC are delineated and
the corresponding engineering properties1 are elaborated; see Section 1.2.1 and Table 1.2.

1.1 Extrusion-based digital concrete construction

Extrusion-based digital concrete construction, hereafter simply referred to as Dig-
ital Construction (DC), refers to an automated construction process by means of
layer-by-layer deposition with cement-based materials; see Figure 1.1. DC heralds
the future of construction technology, since it not only enables speedy, economic,
formwork-free construction, but also provides more flexible architectural design and
higher construction safety [1,2]. Note that the term digital construction is commonly used
to broader fields of architecture, construction and fabrication where digital technologies
are utilised.

Figure 1.1: Conceptual illustration of a digital construction site. (Graphics: U. Staiger,

concept: V. Mechcherine, M. Haist, Garrecht).
The initial demonstrations, in a form close to the current DC technologies, are
published in the late 1990s [3]. However, mechanised construction can be traced back to
at least as early as the 1940s [4]. A brief history of DC and a literature report are presented
1 Non-intrinsic material properties, composed of intrinsic material properties and process parameters.

1
Introduction 1.2. 3D-printable cement-based composites

in Section 2.2.1. Most DC processes share some common attributes and differ in others,
however, often the unifying aspect among all of them is the absence of conventional
formwork. In a way, the absence of form-giving formworks is at the root of many
advantages and challenges of DC technologies. In DC the stresses due to hydro-static
pressure of concrete in plastic state — which are conventionally taken by the formworks
— have to be surpassed by its yield strength. This requirement is often met by setting
high initial static yield stress and/or high structural build-up rate of 3PCs. Some of the
material requirements on 3PCs contradict with each other. The formwork-free nature and
conflicting material properties demand material compositions with high performance,
reliability and processes with high precision. Despite such complex problem description,
recent breakthrough advancements in construction technology, chemicals and machinery
make it feasible to realise DC, at least in lab/pilot scale. However, there are numerous
open challenges, which must be addressed prior to wide-ranging and large-scale indus-
trial implementation of DC; see Section 2.2.8. Systematic material characterisation is one
of the first steps towards robust DC processes. The absence of standard test methods and
limited earlier research set the stage for this research work.

1.2 3D-printable cement-based composites

Processing stages and relevant properties of 3D-printable concrete or broadly
3D-printable cement-based material (3PC) are introduced in this section.

1.2.1 Various stages in the processing of 3PCs

Various process stages2 during DC are: production of 3PCs, fresh-state testing, pumping,
layer-deposition and hardened-state testing; see Table 1.1. As a general chronological
reference, the age of 3PCs (from the time of water addition) at various stages is also
presented in Table 1.1.
The first stage in processing of 3PCs corresponds to production and quality control
in fresh-state, and is denoted by Qf . The second to seventh stages correspond to trans-
portation T, pumping/placement P, extrusion E, deposition/printing D, reinforcement
integration R, and quality control in hardened state Qh, respectively.
Processing of 3PCs in Stage Qf is broadly similar to contemporary concrete types. It is
common, but not mandatory, to have a continuous mixing and pumping system that uses
batch produced dry-mixes. The continuous mixing/pumping approach is particularly
useful for small to moderate (up to 0.5 m3 ) concrete production and is applicable for
pre-cast implementation of DC. In such cases, the distance between the mixer (Stage Qf )
and nozzle (Stage D) is relatively low (up to 20 m); thus enabling to design 3PCs that
have high initial yield stresses, in the order of few kP a. Such concretes [7, 8], due to their
non-flowable nature, are also referred to as zero-slump concretes. An essential task of
the Stage Qf is characterisation of the 3PCs rheological properties, following stringent
quality-control protocols. For this purpose, applicability of empirical methods such as
slump test (even with certain adaptations) is limited; see Chapter 2 and Section 3.4. The
time-dependent static yield stress (SYS) and dynamic yield stress, structuration rate and
plastic viscosity (PV) are instrumental for pumping, extrusion and deposition of 3PCs.
The early age (first three hours) properties, such as compressive strength and elastic
modulus need to be characterised to assure continuous deposition without significant
2 The
order and number of stages in 3PCs processing may vary depending on the exact DC technology.
The descriptions given in this chapter are generalised versions with hints of vision.

2
 Table 1.1: The stages in processing 3PCs; the indicated 3PCs ages may vary for certain approaches.
Age Process stage Symbol General description
Production, quality control (QC), rheometry: Production can be similar to contemporary
concretes. Higher precision in dosing and intensive mixing is necessary due to higher
0-120
usage of admixtures and fine additives. QC with empirical methods is often inadequate.
min
Introduction

Rheometry, calorimetry and ultrasonic methods can be used to characterise fresh-state and
transient properties.
Transportation: For large-scale on-site applications, ready-mix trucks are used for
0 - 45
non-accelerated concrete transportation. For off-site and lab-scale applications, this stage
min
is less relevant.
Pumping: Pumping is usually necessary for large-scale applications to deliver concrete to
20 - 90 printhead. It is distinct from conventional pumping due to lower flow-rate requirements
min and discontinuous nature (if no second-stage reservoir is used). High yield stress and
thixotropy complicate the pumping of 3PCs.
Extrusion: Distinct from conventional extrusion. With PCPs or ram extruders, the acceler-
20-90 ated/thixotropic 3PC is extruded from a second-stage reservoir on to the printing platform.
min In small-scale applications, the printhead is often a mere nozzle without a reservoir. In such
cases, “extrusion” occurs due to the concrete pumping pressure.
Deposition: Most crucial stage of DC where 3PCs layers are sequentially deposited, one on
top of the other. Extruded layers must possess sufficient shape stability to sustain gravi-
20-90 tational forces due to self-weight and vertical stresses from the upper layers. Dimensions
min of the layers can be, but not necessarily, identical to those of nozzle outlet and of element
cross-section; see Figure 2.7. Immediately after placing, 3PCs needs to be cured to prevent
plastic shrinkage and early drying shrinkage.
Reinforcing: Depending upon the followed approach, reinforcement can be placed post or
pre-deposition, e.g., post-tensioning with steel reinforcement. Short fibre reinforcement can
Variable
be mixed in 3PC. In pre-printing case, concrete is extruded around (enveloping) [5] or into
pre-placed [6] reinforcement.

Post-processing, QC for mechanical properties and anisotropy: Post-processing can be applied

1 h -
for enhancing durability and aesthetics. Furthermore, mechanical properties and durability
28 D

3
1.2. 3D-printable cement-based composites

related properties are investigated before the structure is taken into service.
Introduction 1.2. 3D-printable cement-based composites

deformation of the printed element in the semi-solid (transient) state. Note that printed
elements could fail in early age due to plastic deformation and/or buckling [8]. Most
of the subsequent process stages and short, long-term properties of DC structures are
dominated by the rheological properties of 3PCs that are set, tested and adjusted in the
Stage Qf .
In large-scale on-site applications, the 3PCs could be transported to the construction
site from the production plant, e.g., with ready-mix trucks. In this transportation stage T,
the 3PCs are designed to stay dormant by the addition of superplasticisers with extended
effective window and by continuous shearing of the 3PCs. In majority of the DC projects
reported so far, Stage T is non-existent. This is due to Qf and D being in the same location
for the small-scale and off-site applications. However, for large-scale applications, Stage
T makes it necessary to quantify time-dependent variation of 3PCs’ properties that
play a significant role in pumping pressures. These properties include plastic viscosity,
dynamic yield stress, air-content and stability when subjected to high pressure. In on-site
applications, acceleration of concrete setting to ensure faster deposition rates cannot be
achieved by the addition of accelerators in the Stage Qf . Instead, a second-stage mixing
after transportation and preferably after pumping stage but before the deposition is to
be followed. Homogenising accelerators into dormant 3PCs, directly in the printhead
shortly before material leaves nozzle will be more suitable. In the pumping stage P,
3PCs are delivered from a material reservoir to the printhead. For lab-scale miniature
prototype production, manual transfer of the material from the mixer to the printhead is
also followed. In order to be pumped without blockages, the 3PCs need to have sufficient
pumpability in the Stage P; see Table 1.2 and Chapter 5 [9, 10]; also note the differences
in conventional pumping and pumping in DC; Table 1.1 and Section 5.1. It is here
noteworthy that the pumpability is so far not specifically addressed in most of the DC
research projects. This is due to the fact that majority of those projects are so far focused
on small-scale objects which need pumping of relatively lower volume, often less than
0.1 m3 , of concrete over a short distance. For large-scale applications, for both on-site
and off-site alike, pumpability will play an essential role as large volumes of materials —
few m3 per batch — need to be efficiently transferred from mixer to printhead.
Once 3PC reaches the printhead, it will be extruded in Stage E, in a (deliberately) dis-
continuous but blocking-free process on the printing platform. The 3PC may be treated
with a phase changing agent in the second-stage reservoir that is on the printhead, shortly
before the nozzle. The deposition/extrusion is carried out for, e.g., with a progressive
cavity pump (PCP). Alternatively, screw extruders, peristaltic pumps and piston pumps
can be used. Similar to pumping, the extrusion process depends highly on yield stress
and plastic viscosity of the concrete and requires adequate extrudability; see Table 1.2
and Chapter 6. Stage E is interconnected with the most critical stage of 3PCs processing
— the deposition stage, D. In this stage the target structure is built layer-by-layer by
the deposition of 3PC at the predefined co-ordinates at a predefined extrusion rate.
The deposited layers must retain their geometries sustaining self-weight and induced
loads. For this reason, the 3PCs in Stage D must possess high static yield stress as well
as should develop mechanical strength proportionally to the printing rate, in order to
ensure buildability; see Table 1.2 and Chapter 7. Note that buildability depends also on
the layer geometry, process parameters and surrounding conditions. Therefore, in Stage
D, the plastic deformation of 3D-printed structure due to self-weight, vertical loading
and plastic shrinkage shall be accurately monitored. In this stage, measures for curing
the printed elements are implemented to mitigate plastic, drying shrinkage and thermal
stresses as well as early-decay of inter-layer bonds that may occur, e.g., due to early
carbonation [11].

4
Introduction 1.2. 3D-printable cement-based composites

Following D, the 3D-printed structures need to be inspected for process accuracy and
quality control — Stage Qh. The characterisation of 3PCs’ properties in the hardened
stage, generally 1 days to 28 days is an essential part of Stage Qh. Specifically, compressive
strength, tensile strength, Young’s modulus, density, porosity and anisotropy due to
inter-layer interfaces have to be systematically characterised. Further aspects of interest
are the surface quality and microstructure of printed elements. Note that in the Stage
Qh the surfaces of printed elements may be coated with plasters or grouts for aesthetic
reasons as well as to enhance durability over long periods. External curing of printed
elements over the first few days could be considerable for off-site production but is,
often, not practical for on-site applications. Incorporating reinforcement is a prerequisite
for many structural applications of DC. This process stage, R, varies considerably both
in terms of method of implementation and age of 3PCs at the time of reinforcement
integration. For instance, if short fibres are used, then the Stage R takes place as a sub
process in Stage Qf . In contrast, if reinforcement is placed between printed layers after D
(eventually to be post-tensioned), then the Stage R takes place after Stage D.

1.2.2 Properties defining printability of 3PCs for extrusion-based DC

Overall, the 3PCs for extrusion-based large-scale DC should fulfil often conflicting spec-
ifications and satisfy numerous dualities [12]. Autonomous and discontinuous nature of
DC demands materials which can be seamlessly transported and extruded over long time
periods. In this regard, 3PCs should have low viscosity, moderate yield stress and low rate
of structural build-up (SBP)3 . In a conflicting requirement, the 3PCs must exhibit higher
yield stress, plastic viscosity and SBP in order to be continuously buildable. To ensure
adequate inter-layer bonds, among others, short deposition time intervals (TIs), lower
yield stress and plastic viscosity [12] might be necessary. However, short TIs complicate
buildability and necessitates high yield stress. As a consequence, the printability of
3PCs can only be defined as a compound property, considering various properties that
influence 3PCs at various process stages; see Figure 1.2 and Table 1.2. Within the scope
of this doctoral research, the printability is defined as “a time-dependent compound
property of pumpability, extrudability, buildability and inter-layer cohesion”.

Figure 1.2: Important material properties in an example DC process: 1 Pumpability, 2

extrudability and 3 buildability in fresh state and inter-layer bond strength in hardened
state.
The definitions of pumpability, extrudability and buildability are presented in Ta-
ble 1.2. The optimum rheological properties, i.e., magnitude of yield stress and plastic
viscosity, are not identical for achieving each of these three engineering requirements. In
general, to fulfil the above specified requirements 3PCs should be thixotropic, cohesive,
quick-setting (with or without addition of a phase changing agents on the printhead),
3 Combining the effects of hydration and flocculation.

5
 Table 1.2: Definition of selected process specific properties of 3PCs.
1.2. 3D-printable cement-based composites

Stage Relevant properties

• Static yield stress: shear stress at the flow initiation after 3PCs are subjected to rest; τ0,s [P a].
• Structural build-up: rate of increase in static yield stress; Athix [P a/min].
• Plastic viscosity: flow resistance characterised with the slope of downward curve of a hysteresis test results; µ [P a · s].
• Transient-state properties: material properties in the semi-solid state such as compressive strength and elastic modulus,
which are necessary for analytical description and prediction of buildability.
• Open-time: the time period in which a 3PC is 3D-printable, i.e., pumpable, extrudable and buildable. Generally, the first
two properties diminish with 3PC’s age. Retarders and superplasticisers are used to maintain long open-time.
• Pumpability: the property of concretes to be pumped with minimal pumping pressures and with minimal variation in
properties due to pumping.
• Structural build-up: SBP plays a crucial role in pumpability of 3PCs. The discharge pressures could increase dramatically
after long pumping interruptions as the material yield stress develops exponentially [13].
• Extrudability: the ability of a material to pass through 3D-printer’s extruder and nozzle configuration with minimal energy
consumption. This definition diverges from the earlier literature on conventional extrusion, in which extrudability is often
defined as a compound property which accounts not only for the extrusion-ability of a material but also its shape stability.
• Drainage-resistance: resistance of 3PCs against liquid-filtration under high pressures in an extruder. If deficient, the
extrudate properties (density, w/c etc.), will vary from the initial 3PCs.
• Buildability: the property of extruded layer to retain its geometry under self-weight and sustained load from upper layers.
• Structural build-up: buildability — in terms of yield-failure — correlates with static yield stress development in time [7].
• Transient-state properties: buildability — in terms of buckling-failure — depends on geometry and transient properties
including: elastic modulus, Poisson’s ratio, the cohesion etc. [8].
• Extrudability: if short fibres are dispersed in 3PCs, then extrudability is the primary concern, provided screw extrud-
ers/PCPs are used.
• Enveloping ability: in case of using conventional reinforcement, the ability of the high-yield stress 3PCs to enveloping
steel bars is critical. Note that external vibration is typically not feasible.
Introduction

• Interface-strength: depending on composition and time intervals 3PCs can be highly anisotropic. Interface strengths could
be significantly lower than corresponding casted concrete strengths.
• Compressive, flexural and tensile strengths: in optimal directions mechanical properties are comparable to those of other
modern concretes. In critical loading-to-interface orientations, alarmingly low strength values can occur.

6
Introduction 1.3. Goal, objectives and research questions

densely packed (to prevent the ingress of external elements due to lack of formwork)
and should possess precisely controllable4 rheological properties [14–16]. Extensive
experimental studies shall be carried out to investigate above mentioned material
parameters and to develop appropriate cement-based materials with optimum properties
in their fresh and hardened states for digital construction applications.

1.3 Goal, objectives and research questions

The overarching objectives of this PhD research are providing a comprehensive ba-
sis for further research on 3PCs and to accelerate the large-scale implementation of
extrusion-based DC. The means of contribution is the development and characterisation
of high-performance printable cement-based materials using novel and contemporary
methodologies.
Specific objectives of this doctoral research are:
a) Identification of the essential engineering properties of 3PCs and their description
based on physical mechanisms.
b) Adaption and development of test methods for process-specific characterisation of
3PCs; see Section 1.4.
c) Systematic development of 3PCs with varying constituents and properties so that
they can be applied in varying application scenarios.
The following are some of the research questions that this PhD strives to answer:
a) Is it possible to produce autonomously constructed concrete structures with the
help of common construction materials? Would it be possible to produce DC
structures without chemical acceleration of 3PCs (on printhead)?
b) If contemporary characterisation methods are inadequate, what are the new process
specific test methods and their corresponding theoretical frameworks?
c) It is known that structural build-up is a crucial requirement for 3PCs. How to
quantify SBP of non-flowable and highly-thixotropic 3PCs?
d) Would it be possible to quantitatively evaluate extrudability and buildability of
3PCs inline and by means of the 3D-printer itself? If so, how?
e) Which are the parameters influencing the 3PCs’ properties? And to what extent?
f) What are the potentials and limitations of ram extruder tests in DC, considering the
geometrical differences in ram extruder (a combination of plug flow and shear flow)
and progressive cavity pump (shear flow at high shear rates)?
g) What is a representative piece of wall (height, breadth) to be tested in buildability
tests by direct printing? And what are the time intervals to be tested to designate a
material printable?
h) To what-extent the layering nature causes anisotropy in DC structures? Would it be
possible to mitigate strength reduction due to layer-interfaces by optimising 3PC’s
composition, despite long time intervals?

1.4 Research concept, and outline of this dissertation

This PhD forms the foundation and conceptual basis for the research on DC and 3PCs
at the Institute of Construction Materials, TU Dresden; and it is carried out at the
initial stage of research on 3PCs. The research began in the year 2014, only with the
general objectives and without a specific work program of methods and parameters to
4 Through the constituent such as admixtures [12] and/or through external activation [14].

7
Introduction 1.4. Research concept, and outline of this dissertation

be investigated. Considering the novelty of DC, limited prior (to 2014) research on 3PCs
and absence of suitable test methods, a dynamic5 approach was chosen for the detailed
research execution. This means the research program has been adapted at various stages
of the research progress based on the renewed assessment of a subtopics’ significance
and available scientific insights. This strategy made it feasible to extensively investigate
topics which were initially not the focus of this research. The investigations on the topics
of a) strain-based SBP measurement approach (Chapter 4), and b) layer-to-layer interface
(Chapter 8) are two examples of such adaptations. Figure 1.3 illustrates the investigated
subjects. For each subject, the primary investigated parameters, applied methods and
final deliverables are presented. Each subject is marked with a different colour and
the notations that are introduced in Table 1.1. The conducted research can be broadly
conceptualised as two parts:
• Fresh state engineering properties: pumpability, extrudability and buildability; in
coherence with the printability definition presented in Section 1.2.2,
• Layer-to-layer interface, compressive, flexural and tensile properties in hardened
state.
The research on the above aspects was complemented by the investigations on
time-dependent fundamental rheological properties in the fresh state (structural
build-up, dynamic yield-stress) and physical properties in the hardened state (porosity,
density, microstructure). Note that most of the topics presented in Figure 1.3 are inter-
linked with each other. The Bingham properties from Qf influences pumping pressures
simulated by virtual Sliper, the deliverable of P. The SBP of Qf influences all the stages
and specifically D, P, E. The time intervals determined in Stage D are directly linked to the
anisotropy investigated in Qh. The inter-dependencies are not showed in the Figure 1.3
but are elaborated in the state-of-the-art (Chapter 2) and in the respective chapters of a
particular topic (Chapters 4 to 9).
Theoretical, experimental and computational techniques were systematically chosen
according to the topic under investigation and missing scientific knowledge, based on
the state-of-the-art in research and technology, which was assessed at various stages. The
geometrical dependency and process-induced changes on the properties of cement-based
materials were known in the literature. To minimise error in material characterisation
inline approaches were emphasised in this research. Specifically, a 3D-printing test
device (3DPTD) with a progressive cavity pump (PCP) was developed for material
characterisation. Utilising the 3DPTD novel inline approaches were developed, such
as the 3D-printer extrudability test. At the same time, conventional test methods such
as ram extruder and viscometer were also applied as links to the preceding research.
Along with the engineering properties, intrinsic material properties such as the structural
build-up were investigated with concentric-cylinder Couette rheometer. Considering
the high yield stress and non-flowing nature of 3PCs, the protocols based on "applied"
(shear rate) values were scrutinised and alternative protocols based on "effective" values
were proposed. This accurate quantification of rheological properties enables numerical
simulation and prognosis of processes involved in digital construction. One such nu-
merical tool was realised in this research for testing pumpability for which there has
been extensive experimental research already available. In light of parallel developments
on fundamental approaches for evaluating and modelling of buildability [7, 8, 17], a
practice-oriented method for calculating buildability test parameters was proposed. The
proposed method directly links to the economic viability of DC processes. Corresponding
to hardened state, the crucial aspect of anisotropy in printed structures was focused upon
with micro and macroscopic investigative methods.
5 As opposed to working on a pre-determined (structured) research program.

8
 Introduction

Figure 1.3: Outline of the research concept, methods used and deliverables. 1 fresh state and 2 hardened state. Stage D is lies in both fresh and
hardened states.

9
1.4. Research concept, and outline of this dissertation
Introduction 1.5. Purpose, scope and target audience

This dissertation is organised into nine chapters. Chapter 1 presents introduction,

research objectives, purpose and the scope of this thesis.
Chapter 2 presents relevant literature on DC; cover detailed description of motivation,
various DC processes, opportunities and challenges. From Section 2.3 to Section 2.7
relevant literature report on structural build-up (SBP), pumpability, extrudability, build-
ability and mechanical properties of 3PCs are covered. These subsections cover various
influencing factors, underlying mechanisms and open questions related to the primary
research aspects of this PhD that are detailed in; see Chapters 4 to 8. Furthermore,
Section 2.2.9 presents a brief gist of CONPrint3D® technology.
Chapter 3 presents the materials, compositions and applied devices. Chapter 4 ex-
tensively deals with the structural build-up of cement-based materials and presents
developed approach for characterising SBP of a wide range of binder pastes based on
strain-based and single-specimen approach. The effect of cement paste composition on its
SBP is detailed and challenges for SBP quantification are summarised. Finally, a universal
strain-based protocol is proposed.
In Chapter 5, the second major stage of processing 3PCs — placing — is addressed.
After analysing the possibilities and challenges of placing 3PCs, major emphasis is laid
on the testing and prediction of pumpability of concrete. Specifically, the virtual Sliper —
a CFD model for computationally determining pumpability of concrete — is delineated.
Extrudability, the third stage of processing 3PCs is addressed at length in Chapter 6.
Extensive experimental investigations rely on simplified ram-extrusion tests and an
inline extrudability characterisation approach for 3PCs — the 3D-printer extrudability
test (3DPET). The chapter discusses in detail the investigated parameters’ influence
on extrudability and comparative analysis made with 3DPET, ram-extrusion tests and
viscometer tests.
Chapter 7 is dedicated to the fourth stage of processing 3PCs — buildability. Two
fundamental questions — the element height and layer-layer time interval to be followed
in buildability tests — are identified. They are answered with help of a practice-oriented
procedure, taking the machine, labour and material costs into consideration. Since
buildability requirements for a material depends not only upon the target structure
but also on the applied printing process/approach, the full-width printing (FWP) and
(filament printing) FP approaches are mathematically described, as a provision to adapt
proposed procedure for FP.
Mechanical properties of 3PCs are addressed in Chapter 8. Special focus is laid on
the macro-, microscopic investigations of layer-layer interface properties. In Chapter 9
additional aspects of DC, including the integration of reinforcement, are concisely
addressed. The most significant findings of this research and ideas for further research
are summarised in Conclusions and further research. The main matter is followed by
appendices; see Contents for details.

1.5 Purpose, scope and target audience

This research work differs with the general approach for doctoral research in the sense
that it addresses multiple complicated research challenges at once. While attempting to
solve open questions, emphasis was laid on systematically seeking new questions. The
general process of scientific research on construction materials can be described as:

Goal → seek the appropriate questions → convert them into a scientific

problem → identify the influencing parameters → investigate → hypothesise a
theory → prove with experimental evidence → predict with a mathematical model.

10
Introduction 1.5. Purpose, scope and target audience

The novel nature of DC restricted the full realisation of the above process for all the
investigated aspects, for which even the questions were not identified clearly when this
research work began. Many years of collaborative effort with extensive experimental
investigations on 3PCs are needed to fully develop the emerging DC technologies.
Upon this background, the purpose of this doctoral work is to comprehensively
address the concrete technological and rheological challenges in the context of DC,
with appropriate interdisciplinary links to process and machine technology wherever
necessary. A special emphasis is laid on developing tools/methods for characterising
printability. The aspects investigated quantitatively include pumpability, extrudability,
buildability, structural build-up and anisotropy. Thus this dissertation may serve as a
precursor for numerous future PhD works and industrial research, focusing solely on
each of these topics. The materials, methods and the 3D-printing process were developed
with consideration of industrial relevance; see Section 2.2.9. Therefore, direct utilisation
of deliverables (listed in Abstract) from this research for industrial practice is anticipated.
Developed tools and obtained experimental data also accelerate realisation of (much
needed) standards and guidelines for characterising cement-based materials applicable
to digital construction.
The CONPrint3D® technology served as underlying process for various scientific
and technical decisions of this research; Section 2.2.9. However, most of the resulting
solutions are applicable also for other extrusion-based DC processes.6
Extensive experimental research on concrete pumpability has been already conducted
in the past decades [9, 10, 18–25]. Thus, concerning placing of 3PCs, only theoretical and
computational techniques were applied in the research at hand. Scope of pumpability
investigations is limited to theoretical scrutiny of various processes possibilities for plac-
ing 3PCs and a computational fluid dynamics(CFD) model for estimating pumpability of
concrete Chapter 5.
Considering the exponentially growing research interest and the finite time available
for this research work, two time constraints were set. The literature research presented
in this thesis was bound to the time-limit of May 2018. In designing the research concept
and selection of primary methodologies 15th January 2017 was set as deadline, which
coincided with the 2nd meeting of RILEM TC-DFC. The set deadlines were decided
considering the on-going collaborative effort as part of RILEM TC-DFC, which will
deliver a comprehensive state-of-the-art report on the subject at hand.

6 Some properties of 3PCs are independent of the DC process while the others aren’t. Extrudability in a
ram extruder can be very different from extrudability in a PCP. However, SBP is a universal requirement
for all extrusion-based DC processes.

11
Chapter 2

Literature and theoretical research

This chapter presents context and literature review on the research aspects relevant to
this doctoral work1 . General aspects about additive manufacturing (AM) and digital
construction (DC), with detailed description of selected DC processes and their op-
eration principles are presented. Furthermore, crucial aspects of the Concrete on-site
3D-printing (CONPrint3D® ) technology are summarised in Section 2.2.9. From Sec-
tion 2.3 to Section 2.7 state-of-the-art (STAR) in rheology and numerical simulation of
cement-based materials (CM), pumpability, extrudability, buildability and properties of
3D-printable cementitious composites (3PCs) in hardened state are covered.

2.1 Motivation and context

The construction industry is one of the pillars of modern civilisation. A glance at the
statistics elucidates the significance and scale of this sector in terms of economic aspects,
environmental impact and social significance:
a) Concrete is the most used man-made material in the world. In the year 2006 alone
an estimated 30 billion tonnes of concrete were consumed [32]. In 2016, cement
production worldwide was 4.65 billion tonnes. Currently, the majority of the cement
is produced in emerging counties [33].
b) The construction industry in the European Union (EU) constitutes 9.7 % of the gross
domestic product and provides 6.6 % of Europe’s total employment [34].
c) Cement and concrete generated over a million jobs in Europe based on the data
from 2012 [33].
d) “Buildings consume 36 % of total energy use, 30 % raw material use, and 12 % of
potable water in the U.S.” [35, 36]
e) “By 2006, cement production contributed to roughly 8 % of worldwide anthro-
pogenic CO2 emissions” [37]. Müller [37] estimated that by 2030 global cement
related CO2 emissions will be higher than the total CO2 of the entire EU.
Despite this staggering scale and impact, the current construction sector is unable
to match the demands in terms of volume of infrastructure developed, sustainability,
productivity and architectural freedom. Furthermore, disappointingly, there are still
large masses of people lacking a safe, durable yet affordable house. Many of the impeding
aspects of current, conventional construction are detailed in Section 2.1.1. Upon this
background, DC is causing a much needed paradigm shift in the construction technology
and materials; see Section 2.2.2. By doing so, DC provides numerous opportunities, which
may potentially solve most of the current construction challenges. Especially, DC strives
1 This chapter contains excerpts from journal publications co-authored by author [12, 26–31].

13
Literature and theoretical research 2.1. Motivation and context

at delivering the much needed productivity, reliability, labour and resource efficiency to
the construction processes. At the same time, DC enables significant improvements in
creative thinking in structural design; see Sections 2.2.3 and 2.2.4.
However, extensive research is necessary, in the first place due to the complex
requirements on rheological, micro-structural, mechanical, morphological properties and
hydration kinetics that 3PCs must fulfil. These properties dictate the entire DC process
by affecting not only the properties of the final 3D-printed structure but also the DC
process parameters and economic viability.

2.1.1 The impeding aspects of conventional construction

The processing of cementitious materials is the technological core of construction
technology. There exist nearly unlimited possibilities of shaping concrete to construct
versatile structures. Notwithstanding, we currently use only a tiny fraction of the
enormous potential inherent to the formability of concrete in fresh state. Majority of
contemporary construction processes are, in principle, still using an evolved version of
a technique which was developed nearly 2000 years ago, i.e., formwork filling [38]. The
use of formworks often leads to high material, labour and machinery costs in addition
to considerable time delays and negative environmental impact [39]. Formwork costs
amount up to 28 % according to [39,40], but can be in some cases half of the total concrete
structure cost [41]. The problematic role of formwork is even more significant in case of
constructing unique objects or complex structures, where the use of formwork is on the
one hand challenging (limited geometrical flexibility) and on the other hand expensive
plus time-consuming. In addition to the costs and architectural limitations, construction
formwork also leads to negative environmental impact. However, usage of formwork is
essential for placing most types of concrete, especially those of ready-mix concrete.
Another important limitation caused by formwork-based construction is the limited
geometrical flexibility. Formworks are generally planar and are primarily responsible for
the globally followed rectangular layouts with straight walls. Construction of curved,
spherical and other non-standard structure with conventional formworks is possible,
however, comes at an elevated cost in terms of time and money. Most importantly, the
geometrical constraints of formwork-based construction limit realisation of advanced
approaches including a) structural principle of form follows force and b) expedite
changes to design with the help of parametric design tools. Since the design of the
structure is defined by the formworks, making changes to an established design involves
in expensive (time, cost) re-fitting of the existing formworks.
Speed of construction and productivity have also plagued construction industry
[34, 42, 43]. Though advancements in concrete technology such as development of
self-consolidating concrete addressed the low rate of productivity in construction to some
extent, construction processes are often categorised into the least productive processes
[44]. The productivity problem is even more pronounced in conjunction with lack of
skilled labour. Many approaches and tools are developed to increase labour productivity
on a construction site. However, without significant change in the construction process
itself, these approaches have limited effect. Fully or semi-autonomous DC technologies
will bring this fundamental change to current construction routines and significantly
increase productivity.
Today, increasingly, more than half of the world’s population lives in urban areas
[45, 46]. Future population growth will be concentrated in developing countries with
close to 90 % growth in Asia and Africa [46]. Yet the growth of urbanisation in the
developing world has not been matched by a commensurate growth in the supply of

14
Literature and theoretical research 2.1. Motivation and context

decent housing [47]. On the one hand time-taking practices such as usage of formwork
and on the other hand lack of automation are the primary reasons for this supply-demand
imbalance. Labour safety is another concern for conventional construction. An aspect
which impacts greatly the feasibility of construction in remote areas and under dangerous
conditions: nuclear radiation, natural catastrophes and industrial applications. Further
challenges in conventional construction can be referred to in [28, 34, 42, 43, 48].

2.1.2 Additive manufacturing and the fourth industrial revolution

Additive manufacturing (AM) can be defined according to ASTM F2792 as “the process
of joining materials to make objects from 3D-model data, usually layer upon layer, as
opposed to subtractive manufacturing technologies” [49]. There exist numerous types
of AM technologies, all of them have the common attribute of producing objects by
addition2 including a) Steriolithiography (SL or SLA), b) Fused Deposition Modelling
(FDM), c) Binder Jetting, d) Selective Laser Sintering (SLS), e) Selective Laser Melting
(SLM), f) Electron Beam Melting, g) Layer Laminated Manufacturing.
AM is not synonymous to any single process listed above, rather it is a hypernym for
many sub-processes and technologies. Generative manufacturing could also be used as
an alternative hypernym, which is defined according to DIN 8580 as “manufacturing
of products from shapeless/shape-neutral materials without using special tools, which
reflects the geometry of the product, such as mould” [50].
Plastics, metals and ceramics have been used as materials in various AM applications.
Usage of AM for not only prototyping but also for real application is already in
industrial practice. For example, in the aerospace industry AM methods are used to
quickly produce components with new and complex designs, to reduce the weight of
the elements and to increase the durability. In medical applications, AM is successfully
used for manufacturing prosthetics. Details on functioning principles, mechanisms and
applications for various AM methods could be referred to in literature [51–54]. Within
the scope of this chapter, a very brief and non-comprehensive review is presented. The
SLA approach was patented by Charles Hull in the year 1986. In SLA, the top surface of
a liquid photopolymer resin is exposed to and solidified (cured) by selectively projected
ultraviolet laser pattern. The sold object is then “lowered” into the resin and the solid-
ification process repeats. Eventually, three-dimensional object is formed layer-by-layer.
The SLA uses CAD tools in a process involving 3D-CAD models “sliced” into layers
resulting in Standard Tessellation Language (STL) data files that are comparable with
steriolithiography machines. In this regard, the SLA forms the basis for modern DC
and other AM-based production processes. That being said, the current most widely
used AM techniques is FDM. FDM works on basis of heating and extruding/depositing
thermoplastic filaments in the specified co-ordinates, layer-by-layer. Thus, FDM differs
with SLA as it does not use a liquid “batch” of resin, does not need laser projection, but
needs thermoplastic filament or pellet. However, FDM shares the CAD models, slicing
process and layered manufacturing process aspects with SLA. The FDM technology was
developed by Scott Crump of Stratasys in the 1980s. Note that the majority of DC
approaches are similar to FDM in the basic principle.
AM gains pronounced significance in the context of the potential fourth industrial
revolution. The fourth industrial revolution is characterised by cyber-physical systems
which unify digital systems with physical and biological objects/spheres; see Figure 2.1.
The first three industrial revolutions are propelled by mechanisation (steam power),
electrification (electric energy) and automation (using information technology and
2 As opposed to traditional manufacturing by subtraction: cutting, milling, drilling, grinding etc.

15
Literature and theoretical research 2.2. Digital construction

electronics). Digitalisation and automation at its core, the fourth industrial revolution
accelerates wide-spread implementation of AM technologies and their rapid advance-
ment. “Industrie 4.0” part of the High-Tech Strategy 2020 action plan in Germany and
“Horizon 2020” in the USA are two ambitious initiatives for the future-proof automation
in the manufacturing sector [55]. The fourth industrial revolution is going to change
the way we conceive, communicate, produce (construct) and purchase things in its
entirety through integrated digital and physical processes. Autonomous organisation of
intelligent production procedures through Internet of Things increases productivity up
to 50 % [56]. Many paradigm shifts that will be caused by the fourth industrial revolution
and Internet of Things could be referred to in [56–58].

1st 2nd 3rd 4th

Mechanization, Mass production,
Computer, Cyber Physical
water power, assembly line,
automation Systems
steam power electricity
Figure 2.1: The four industrial revolutions. The fourth industrial revolution relies on data
transfer and cyber physical systems. (Adapted from: AllAboutLean.com/Christoph Roser).

2.2 Digital construction

The impeding aspects of the conventional construction and the advancements offered by
the additive manufacturing creates an opportunity to overcome the former with the latter.
The construction industry is generally characterised as conservative, due to its so-far
relatively low-extent utilisation of advancements in modern automation and additive
manufacturing technologies. However, DC is a promising solution to this critical issue.

2.2.1 Definition and brief evolution

Digital construction refers to an automated layer-by-layer deposition with 3D-printable
cementitious materials. The earlier definitions of AM may not completely hold when
it comes to digitalised construction processes. It is very essential here to explain the
author’s rationale for defining the term digital construction, especially so, considering
the emergence of numerous approaches and processes including Contour Crafting,
3D-concrete-printing, selective activation etc. Construction for many centuries have been
additive and not subtractive. Construction has also been generative as raw materials such
as clay, concrete are either shapeless or shape neutral. Especially, automated masonry
construction — robotic brick laying, if built using 3D-model data — is very qualified
to be called an AM or GM process considering its additive, layer-by-layer nature. To
avoid the ambiguity, author takes the liberty to categorise various construction processes
that are inspired by AM methods under a single hypernym specific for construction
applications: digital construction. When the material used is concrete, then DC refers to
digital concrete construction. DC is a process in which complete or parts of construction

16
Literature and theoretical research 2.2. Digital construction

elements are produced, without conventional formworks, layer-by-layer from digital

model data with largely autonomous robots. Variants of DC could further be specified
with terms such as extrusion-based digital construction (EBDC) [2, 59] and selective
activation based digital construction (SBDC). Since this doctoral work deals only with
extrusion-based DC, henceforth, the term DC is used to refer to EBDC, unless otherwise
specified.

Brief history: Some of the crucial moments in the evolution of DC are listed in Table 2.1;
this list is not exhaustive and also considers only projects where cementitious materials
(CM) are used, except the D-Shape technology. Pegna [3] is credited as initiator of
additive manufacturing with cementitious materials. In a manual deposited selective
activation based DC technique, Pegna produced 1:87 scaled-down models of a house
with sand, Portland cement and water. At first thin layers of sand and cement were
deposited followed by water deposition for the selective activation. Eventually, water
deposition is depreciated and the dry specimens were steam cured at 300° C temperature
and 1 atm pressure. However, the roots of mechanised formwork-free construction can
be traced back to as early as 1941. Urschel [4, 60] patented and demonstrated an
approach which resembles formwork-free construction and utilises a simple traversable
form; see Figure 2.2. This technique avoids usage of common formworks, needs less
labour and is considerably faster than fully manual construction. Despite so many shared
attributes with modern DC, the Urschel approach lacks, understandably, digital control.
Among the commendable innovation of Urschel’s device are a) the novel arrangement
of mechanised tamping members to feed-and-compact the earth-moist cementitious
material and b) cable reinforcement that is simultaneously laid with the material layer.
The resistance offered by the already embedded part of the reinforcement is expected
to be sufficient for unwinding the cable-winding during material deposition, without
human involvement.
Khoshnevis [59] accomplished pioneering work in DC with the Contour Crafting
technology — an extrusion-based DC technique. Khoshnevis experimented with ceramics
as well as cementitious materials with fine ( ≤ 2 mm) sand as aggregate. With the
primary objective of producing large-scale construction component the researchers of the
University of Loughborough developed Concrete Printing technology [1], which is similar
to Contour Crafting. Another early work in DC is by Enrico Dini of D-Shape [61]. This
technology is similar to Pegna’s, however, largely focused on non-cementitious materials.
DC has witnessed enormous growth in the research as well as industrial activity from the
year 2014; see Figure 2.3 and Table 2.1. Currently, there are over 30 projects worldwide,
ranging from lab-scale feasibility studies to large-scale house constructions. Details on
various DC processes and applications are available in [1, 6, 12, 51, 59, 62–64].

Cocnrete
hopper
Reinforcement

Figure 2.2: The mechanised house printing machine by Urschel, along with various stages in
the construction of a building [4, 60].

17
Literature and theoretical research 2.2. Digital construction

35
Printed, pre-stressed bridge, TU/e, NL

Cummulative number of projects

30
Sinusoidial wall, XtreeE, France
25
3D-printed hotel suite, TotalKustom, USA
20
6-storey building, WinSun, China
15 CONPrint3D, TU Dresden, Germany

10 Freeform Construction, Loughborough Univ., UK

Contour Crafting
5 Univ. of Southern California, USA

0
1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
Year
Figure 2.3: Evolution of projects on digital construction (adapted from [65]).

2.2.2 Principles and process

In this section, various processes of DC are presented and their underlying principles are
described.
Depending on the subjective perspective DC technologies can be classified into various
systems. For example, Duballet [63] classified extrusion-based DC systems depending on
the parameters object scale Cxo , extrusion scale Cxe , printing environment Ce , printing
support Cs , and assembly parameter Ca . Detailed classification of all DC techniques is
not in the scope of this research work. However, based on the general process, the DC
processes can be broadly subdivided into six categories:
a) Extrusion-based DC (EBDC) [2,59]: in this approach, premixed concrete is extruded
layer-wise at the specified coordinates through a nozzle at the determined printing
rates; no vibration of concrete during and after extrusion is involved; no conven-
tional formwork is needed; similar to fused deposition modelling; see Figure 2.4a.
b) Hybrid approach of DC for 3D-printing integrated formwork and conventional
casting of the infill (in the core). These are partial DC technologies, e.g., variants
of Total Kustom [77] and Apis Cor [87], where the outer formwork shells are
3D-printed and the infill is cast with SCC or vibrated concrete; see Figure 2.5a.

a) b)
Figure 2.4: a) EBDC with single-nozzle setup, an example from CONPrint3D® , b) SBDC
using multiple nozzle setup, an example from D-Shape [61].
c) Vibration-based 3D-concrete-printing (e.g., the technology implemented by
HuaShang Tengda Ltd [5]): in this approach a more or less ordinary concrete is

18
 Table 2.1: Some moments in DC evolution.
Timeline Organisation Description Scientific publication Others
Primary Sec. Peer-reviewed Article
1941 Urschel Machine for building walls [4]
1930s House demonstration [60]
1995 Pegna Solid free form construction [66]
1997 [3]
1998 Khoshnevis Contour Crafting [67]
2004 [59, 68, 69]
2007 Loughborough U. Concrete Printing [1, 2, 70, 71]
2008 D-Shape The Radiolaria Pavilion [61, 72]
2012 TU Münich 3D-printing with wood-cement composite [52, 73]
Literature and theoretical research

2013 ETH Zürich Concrete without formwork [74]

2013 Mesh Mould [75]
2018 DFAB House [6, 15, 76]
2014 Total Kustom 3D-printed Hotel [77]
2014 WinSun 3D-printed house components [78]
2014 TU Dresden CONPrint3D® [79]
2018 Sliper, Approach for SBP, PSHCC; 3D-p. steel reinforcement [27, 29, 80, 81]
2015 Perrot Structural built-up of 3PCs [7]
2015 TU Eindhoven 11x5x4 m concrete printer commenced [82]
2016 [83]
2018 3D-printed bridge, Model for early-age mech. behaviour [8, 84],
2015 X-TreeE 3D-printed concrete wall [85]
2016 Large-scale 3D-printing with high performance concrete [62]
2016 RILEM TC-DFC RILEM technical committee on digital fabrication with CM [86]
2017 Apis Cor Portable 3D-concrete-printer [87]
2017 NTU Singapore Meas. of tensile bond strength of 3D-p. geopolymer [88]

19
2.2. Digital construction

2018 Pierre Theoretical framework for penetration depth in SBDC [89]

2018 Digi. Con., CCR 1st RILEM Conference on Concrete and Digital Fabrication [28, 65, 90–92]
Literature and theoretical research 2.2. Digital construction

a) b)
Figure 2.5: Process stages in a hybrid approach of formwork printing and filling with SCC or
vibrated concrete [87], b) the forked-nozzle approach of Huaschang-Tengda [5].

deposited slowly by vibration through a twin-set of nozzles which deliver material,

facing each other; see Figure 2.5b.
d) Robotically fabricated functional formwork, which is post-filled with concrete: the
formwork can be made of steel or polymer and serve also as reinforcement as
demonstrated by Mesh Mould [6]; see Figure 2.6a.
e) Autonomous adaptive formwork: an adaptive formwork is controlled numeri-
cally, depositing special, accelerated concrete step-wise in a process similar to
slip-forming; the shape of formwork can be dynamically changed in the process
resulting in DC of complex slender structures as demonstrated by Smart Dynamic
Casting [64]; see Figure 2.6b.
f) Methods based on selective activation (Dini and Monolite-UK-Ltd, 2015): thin
layers of dry materials are first placed on a platform and binder or activator is
then delivered to specified coordinates; applicable for DC of complex structures;
see Figure 2.4.
In all the approaches, digital data from 3D-CAD model along with printing parameters
such as “velocity of printhead” and “rate of extrusion” are delivered to a 3D-printer from
a computer control unit, in a machine-readable language [12]. The desired structure is
then built layer-by-layer using the defined coordinates and the given printing rates. The
most widely used approaches for DC are extrusion-based DC and selective activation
based DC. In the case of EBDC, premixed material is extruded at the specified coordinates
through a nozzle, in a process similar to FDM. In case of selective activation based DC,
one material or a mixture of several different materials is first placed on a platform (bed)
and binder or activator is then delivered to specified coordinates. Thus, only material
at the required coordinates will be bonded together to form the desired structure.
Once a layer is hardened, binder deposition will continue on to the next layer. Further
classification of selective activation based DC technique can be referred to in [72].
Among the EBDC and selective activation based DC, the latter approach seems most
promising for the construction of large-scale, possibly on-site, structural elements. The
dimensions of the printing device for selective activation based DC techniques must
be bigger than those of the printed element [12, 61]. This is not the case for many
extrusion-based DC technologies, as the examples of CONPrint3D® [12] and Apis Cor
[87] demonstrate. Furthermore, in the case of selective material deposition by extrusion,
the material is delivered only where it is needed permanently and should, therefore,
be sufficiently “buildable”. In contrast, in selective activation based DC techniques,
the support material surrounding the material to be bonded is crucial to keeping
its shape while it is still in plastic state. The need for support material has both

20
Literature and theoretical research 2.2. Digital construction

a) b)
Figure 2.6: a) Robotically fabricated functional reinforcement, an example from Mesh
Mould [6], b) Smart Dynamic Casting process [64].

positive and negative consequences: structures of any geometrical shape can be produced
(positive), but all the non-bonded material must eventually be removed (negative). The
selective activation based DC processes are also apparently more sensitive to surrounding
conditions than the selective material deposition by extrusion. Furthermore, the vertical
deposition rates in selective activation techniques are constrained by the necessity for low
thickness of each layer. The liquid binder must be able to penetrate completely through
the dry powder [89]. As a consequence, the thicker the granular layer the challenging is
the intrusion of binders which are often yield-stress fluids. In the case of extrusion, the
vertical deposition height is constrained by the static yield stress and geometry of the
layer/element. In addition, since the material is deposited only on the element geometry
layout, the overall execution times are faster [89] than selective activation. Thus, at this
stage the application of selective activation techniques seems to be more practical for
off-site production of complex elements having relatively small dimensions only. In this
thesis exclusively the extrusion-based techniques are further addressed, considering:
• the associated DC technique CONPrint3D® is extrusion-based,
• for large-scale on-site application extrusion-based approaches are more suitable,
• the major properties investigated in this thesis are either not relevant or must be
addresses differently in the case of selective activation. For example. due to the
mandatory presence of support material, buildability is not really a challenge in
the context of selective activation technology.
For the purpose of this research, the author further distinguished between full-width
printing (FWP) and filament printing (FP) based on layering technique. In FWP the
breadth of the extrudate is equal to that of the target element, as in CONPrint3D® ; see
Figure 2.7a. In case of FP the breadth of the extrudate is many times smaller than the
breadth of the target element, as in Contour Crafting [59, 77] (Figure 2.7b) or even more

21
Literature and theoretical research 2.2. Digital construction

pronounced in fine filament approach called Concrete Printing [1, 2, 93]; see Figure 2.7c.
Consequently, a cross-section of FP elements consists of outer layers (shell/mould) and
inner layers or fillings.

a) b) c)
Figure 2.7: a)CONPrint3D® technology as an example of full-width printing [courtesy:
Chair of Construction Machines, TU Dresden], b) Contour Crafting as example of filament
printing [courtesy: Contour Crafting], c) Wonder Bench at Loughborough University as
example of fine filament printing (photo by V. Mechtcherine).

2.2.3 The potential opportunities of DC

This section presents the potential opportunities of DC techniques in general.

Productivity: The foremost opportunity that DC processes bring to the construction

sector is productivity. The synergistic interaction of various aspects of the process
chain makes it possible to realise cost, time-effective construction with drastically
reduced chance of error. General connectivity and dynamic interaction among managers,
machines and elements will increase the production efficiency as much as 25 % [27,28].
Improvement of productivity finds its roots specifically in terms of construction time
reduction, higher precision with fewer chances for errors and dynamic monitoring of the
construction process.

Geometrical flexibility and cost-efficiency: In conventional construction, the formworks

define the limits of the shape and geometries. In DC, limits of geometries are not defined
by the formworks, but rather by the operating limits of robot. The absence of, mostly
planar and rectangular, conventional formwork results in quasi-unlimited architectural
possibilities. For example, usage of parametric design tools and robotic construction pro-
cesses enable not only realisation of creative and/or complex architectural forms but also
rapid modifications of the designs [94]. Note that In DC, various stages of construction
are seamlessly connected through digital systems. From a structural perspective, the
higher geometrical flexibility makes it possible to exploit the advancements in structural
engineering such as form follows force principle, topologically optimised structures [95]
and bionic structures. Few examples of synergistic structures with topology optimisation
and digital construction can be referred to in [95, 96].
On one hand, DC approaches make it feasible to produce complex structures at
previously non-feasible construction rates and precision; see the double-curved panel
and complex pillar in Figure 2.6. On the other hand, mass-produced but custom-designed

22
Literature and theoretical research 2.2. Digital construction

housing can is more accessible with DC. The latter application does not necessarily need
to have complex geometries but would make possible to adjust the houses according
to the needs and wishes pertinent to individual end-users. Furthermore, with DC it is
possible to produce functional components with added value as detailed in [1]. Thanks
to the independence of and often complete elimination of formwork, the DC methods
also significantly reduce costs of construction through reduction in material and labour.
Many claims have been made with respect to disrupting cost-reductions with help of DC.
Win Sun claimed up to 70 %, 50 % and 80 % reduction of construction time, materials
and labour, respectively [97]. Furthermore, they have claimed of producing houses with
a cost of 24 U SD/m2 . Apis Cor has claimed of producing a house at 223 U SD/m2 .
Furthermore, numerous scientific articles [62, 98, 99] have stated cost reductions with
DC [100], [15], [101], [102].

Amalgamation and acceleration of science and technologies: The biggest opportunity

DC creates is the necessity to test the extremes of the present technologies. For example,
the conflicting requirements that DC puts on the material rheological behaviour leads
to accelerated development of advanced cementitious materials whose rheological
properties can be steered on-demand. Specifically, admixtures to control the setting and
hydration kinetics will play increasingly prominent role. There has been a significant
amount of research done recently on developing high-range water reducing agents,
nano additions as rheology modifiers, and cellulose ethers as viscosity modifying agents,
to name a few. At the same time, the fields of automation and robotics are advancing
at a rapid rate. On the management and processing fronts, digital-centric approaches
such as Building Information modelling behaviour (BIM) are becoming increasingly
important by hosting not only geometric information but also materiel properties and
process parameters. Enormous data flow from the digital CAD models to complex
kinematics of on-site construction robotics coupled with autonomous measurement and
adjusting of material properties sets the necessary platform for paradigm shifts in all
involved sciences and technologies. DC, essentially, creates the necessity and provides
the opportunities to test all these advancing technologies in a multi-disciplinary process.
The results of this development, independent of the eventual success prospects of DC,
will be synergistic technological and scientific solutions that would benefit concrete
construction, in general. Digital construction alike additive manufacturing [57, 103],
upon successful development, has the potential to revolutionise the construction
industry by converging various aspects of construction processes including design,
planning, material production, transport, element fabrication coupled with management
of labour and resources into one unified seamless digital construction process.

Societal relevance and construction in dangerous locations: DC approaches are also

promising solutions for the challenge of rapid urbanisation in developing countries as
they reduce construction costs as well as dependency on skilled labour. Mass-produced
housings fully utilises the benefits of DC, as once digitalised (3D-model, slicing, BIM · · ·
) the same data can be used multiple times irrespective of number of units produced.
Construction in remote areas, under dangerous conditions (nuclear radiation, natural
catastrophes and industrial applications) and especially in space applications are yet
another opportunity brought by autonomous DC methods. Interplanetary or lunar
settlements either for scientific exploration or for permanent colonisation is a definite
eventuality. Robotic construction of necessary structures prior to the arrival of humans
could play vital role in such endeavours. There is already on-going research on applica-
tion of DC for space applications [59, 104, 105].

23
Literature and theoretical research 2.2. Digital construction

2.2.4 Applications of DC
This subsection was originally co-written by the author and published in [28]. Digital
fabrication has been applied to 3D-freeform architectural design and model prototyping
for many years, some of such applications of digital fabrication and the companies
involved can be referred to in [1, 51, 106]. The scope of this section is limited to con-
struction applications with cementitious materials through DC, specifically those based
on extrusion. Architectural model creation and prototyping are not explicitly addressed.
Pegna [3] introduced selective activation of reactive bulk material (Portland cement) as
a DC technique with potential for large-scale solid freeform structures. Advances in the
selective activation and similar techniques such as D-Shape [61] along with applications
can be referred to in Lowke et al. [72]. As early as 2004, Khoshnevis et al. [59] had
presented the idea of using Contour Crafting, to produce structures, in two process
variants: a) producing outer “shell” elements through automated extrusion and then
filling the inner space with a conventionally cast concrete b) producing outer “shells” as
well as inner structure (which follows a sinusoidal path). Complex and unique structures
have also been proven to be possible applications of DC with considerable potential;
see Figure 2.8b. Energy-efficient wall structures and multi-functional building elements
are of high significance for potential applications of DC [59, 70]. This can be achieved
through a) topology optimisation and b) material development. For example, Buswell et
al. [65] reported of two wall panel designs with the same external geometry and varying
internal cross-section and achieved thermal conductivity much lower than concrete
block works. Added functionality can also be achieved by DC through purposeful
choice of material used, e.g., cementitious materials with very low thermal conductivity
such as light-weight aggregate concrete (LWAC), foam concrete, concrete with wood
shavings/particles as aggregate/filler (Figure 2.8a) and thermally activated concrete (e.g.,
enriched with phase change materials). Usage of multiple materials, each responsible for
a specific function such as compression load-bearing, tensile load-bearing, insulation etc.
has been demonstrated and is envisioned to offer a plethora of opportunities to digitally
fabricate smart structures.
4m

1.52 m
a) b) c)
Figure 2.8: a) 3D-printed LWC with wood as filler [52] as an example of low thermal
conductivity element, partly following earlier works [1, 59] and b) a complex truss-shaped
structure according to the principle “form follows force” [107] and c) part of a residential
building with non-planar walls [5].
Khoshnevis [59] has conceptualised that DC can be utilised to produce entire struc-

24
Literature and theoretical research 2.2. Digital construction

tures with all functions such as utility conduits for plumbing and electric installations,
included. “Wonder bench” by Lim et al. [70] demonstrated the feasibility of functional
integration through DC to produce light-weight, acoustic and post-reinforced structures.
Utilising the high geometrical flexibility topology-optimised structures according to
the principle “form follows force” can be produced by 3D-printing which are well
load bearing yet not massive. A complex truss-shaped 4 metre high post (Figure 2.8b)
supporting the roof of a playground at a school in Aix-en-Provence, France is an example
for this [107]. In addition to the large-scale prototypes, a 6.5 m x 3.5 m compression-only
bridge was designed, following the principle “design by testing”, and produced with
DC. The bridge, is deemed structurally safe for public utilisation and serves currently
on a bicycle-track [84]. Multi-material structures with load-bearing concrete as outer
shell and insulating material as infill are also envisioned to reach marketability with DC
in the near future. DC is also expected to facilitate construction of exotic architectural
geometries, that were not feasible with conventional construction techniques. When the
scale of the elements is concerned, the majority of the early examples of DC such as
Contour Crafting [108] and Concrete Printing focused on producing elements whose size
is sufficient to be implemented in a pre-cast industry. However, on-site applications of
DC — e.g., with mobile 3D-printers — are already proposed and are in the pilot stage
currently [12, 87, 109].
It is relatively straightforward to recognise the applicability of DC for producing
complex and novel structures. Few research works even argued that DC “technology will
only be of interest if it can bring novelty in building performances, that is to produce
constructive elements of novel interest that were too difficult to make before” [110].
However, it is unclear, why DC is not “of interest” for non-novel structures. In fact, the
research at the TU Dresden identified that there is potential and necessity for DC in
common construction applications such as housing (mass-customised or not) [12, 39].
Systematic market and economic evaluation of the construction market in Germany
revealed that, by applying CONPrint3D® , considerable amount of time and 25 % of
the cost savings can be achieved in residential building construction [12, 39]. According
to statistics for the year 2014, the construction of 163,844 residential buildings was
permitted in Germany [111]. When analysed specifically from material perspective, 75.3
% of the materials contemporarily used for manufacturing walls of such buildings (31.1
% brick, 22.3 % limestone, 18.2 % aerated concrete, 3.7 % lightweight concrete) are
proved to be potentially replaceable with DC applicable materials without the use of
reinforcement. Upon this background, it can be concluded that DC has also tremendous
potential for conventional structures such as residential buildings; see Section 2.2.9. On
the one hand, 3D-printed houses offer the social benefit of providing customised yet
affordable housing for masses in developing countries. On the other hand, DC provides a
promising alternative construction method for developed countries.
The applicability of DC for any construction process depends upon both technological
fulfilment (such as the need to use reinforcement for ceilings) and economic viability.
While both of these requirements are fulfilled for non-reinforced walls construction for
residential buildings, this is not the case with many structural applications. Indeed, one
of the limiting aspects for wide ranging application of DC is lack of satisfactory solutions
for integrating vertical reinforcement. This issue, along with possible approaches is
detailed in Section 2.2.8. The author opines that the construction applications of DC will
increase exponentially in the near future, thanks to the wide variety of technologies under
development pursuing vivid applications including: geometrically complex structures,
elements with integrated functionality, topologically optimised structures and solutions
for mass-customisation and cost-effective housing.

25
Literature and theoretical research 2.2. Digital construction

2.2.5 Material technological aspects of DC

This subsection presents some general material aspects in the context of DC. Due to
the conflicting requirements and high precision expected concerning the properties
of 3PCs, the choice of constituent materials for printable concrete compositions may
lead to contradicting conclusions. To achieve quick-setting and high early strengths
higher amount of cement and cement of rapid type has to be used. Higher content
of cement leads to higher shrinkage and thermal stresses which have to be mitigated.
Any irreversible damage due to shrinkage or thermal stresses would lead to a drastic
deterioration of concrete and in the case of form-work free construction this would be
critical. When concrete contains higher fines, generally the (cement/binder) paste content
is increased to meet the rheological requirements. Conversely, compositions with more
coarse aggregates could be formulated with lower paste content and often with lower
cement content. Upon this premise, it could be argued that usage of higher content of
coarse aggregates helps to mitigate shrinkage and thermal stresses, however, such usage
may lead to lower structural build-up (SBP) and very early strength. SBP and specifically,
thixotropic behaviour is highly advantageous in the case of DC as it, inherently, provides
high deformability – thus pumpability – under external force and low deformability –
thus buildability – at rest. The degree of thixotropy is linked to the packing fraction of
composition and often high thixotropic behaviour is observed when more fines are used.
To master these dualities, cementitious compositions should possess in addition to
generally needed properties additional requirements such as on-demand controlled
quick-setting and steerable rheological properties. Overall, the 3PCs so far can be char-
acterised as fine-grained cementitious materials with high volume of paste content and
binder volume. Paste volume in 3PCs is generally above 50 % by volume [2, 27, 68, 112]
and in some cases up to 60 % by volume. Despite the associated negative effects, higher
paste volume is necessary to ensure pumping, extrusion and deposition of 3PCs. For
instance, Ogura et al. [81] showed that mixtures with high sand-to-binder ratio needed
high extrusion force despite relatively high flowability and low yield stress. Generally,
the binder consists of Type 1 Portland cement with 95 % clinker content Rapid type. To
reduce the risk of thermal stresses while improving the rheological behaviour, partial
cement replacements in the form of pozzolanic or inert additives such as fly ash and
limestone powder [2, 8, 27, 68, 112] are often added. In 3Pcs, the ranges for the additives
in terms of wt.% are 10-15 % for micrsosilica [2, 12, 62], 17-30 % for fly ash [2, 12]
and up to 20 % for limestone powder [8, 62]. In addition to these “micro” additives,
additions of “nano” constituents such as nano-clay are also used at low dosages to
enhance thixotropy [68]. That being said, very high amount of cement dosages ranging
above 600 kg/m3 are not rare [2, 62, 112]. The primary objective using additives and
additions is enhancing rheological properties for pumping [10,12], extrusion [2] and high
shape-stability [68]. The associated secondary benefits are improvement of mechanical
properties, for example, due to denser matrix. So far in 3PCs, the size of the aggregate is
constrained by the machine and process limitations and not by the material or structural
specifications. In fact, usage of larger aggregates can be beneficial for DC applications,
as they enable: reduction of required paste volume, minimising shrinkage and thermal
stresses and CO2 footprint. At the same time, aggregates can also be used as a tool
to alter rheological properties, specifically yield stress. Increasing the volume fraction
of aggregates in cement suspension increases the yield stress exponentially [113–115];
an effect that is not be confused with increasing aggregate size at a constant aggregate
volume fraction. Aggregate dosage ranges typically between 40 % and 50 % by volume.
Generally, maximum aggregate size ∅max is less than 2 mm and in a majority of the
applications ∅max < 1mm [2, 8, 27, 68, 112, 116].

26
Literature and theoretical research 2.2. Digital construction

The much needed setting and control of the rheological properties are achieved with
help of superplasticisers, accelerators and retarders, often in an exquisite combination.
Superplasticisers such as polycarboxylate ethers and retarders such as methylenephos-
phonic acid, citric acid and formaldehyde [2] are used to ensure the long open time in
which the 3PCs are pumpable and extrudable. To achieve rapid development of yield
stress, accelerators are added, especially if retarders are used in the first place. Also
commonly used are viscosity modifying agents and thickeners such as cellulose ether.
For this purpose, there are numerous approaches available, all of them attempting to alter
the structuration and hydration kinetics. Detailed description of various approaches and
their underlying mechanisms can be referred to in [14–16, 117, 118].
Depending upon the number of hypothetical stages by which phase change of the
3PCs is achieved the DC process can be categorised into single-stage, two-stage or
multi-stage phase change process; see Figure 2.9. In the single-stage approach, mixing
or activation of the phase change agent (PCA) is carried out only once when concrete is
produced; for example, by adding PCAs which show a delayed response. This means, once
initiated the phase change process cannot be altered at a later point. As a consequence,
if the 3D-printing process is interrupted severe difficulties may occur to re-initiate the
pumping/extrusion process. This is true not only if the phase change is achieved by
chemical acceleration through admixtures but also by thixotropy/SBP [13]. Moreover,
by single-stage approach it is not possible to ensure uniform structuration rate for all the
layers in a single structure as detailed by Reiter et al. [15].
In contrast, in two stage process, a mixture optimised for pumping and extrusion
process is produced and transported to the printhead. Such non-accelerated mixture
is then dosed with a PCA for chemical acceleration. Alternatively, in the multi-stage
process, a PCA which is mixed in the 3PC during production is activated externally when
the phase change is needed. The active rheology and active setting control approaches
are relevant here [28]. The multi-stage approaches, in theory, allow on-demand control
of rheological properties at all stages of 3D-printing. This means, the PCAs are always
present in the 3PCs and activated/deactivated whenever necessary to obtain the required
properties corresponding to the 3DP process stage. If second-stage mixing or external
activation is not possible, then composing a mixture with high structuration rate is
the only possibility. Nano-clay and pozzolanic additives can be used to achieve high
structural build-up. Wangler et al. [101] argued that “for objects with height above
1-1.5 m simply relying on thixotropy for structural build-up will not be very effective”.
However, definitive proof for the upper limits of thixotropy as a means for ensuring
buildability is still due.
Production phase Pumping phase Printing phase

M
Aggregates

Stage 1 Stage 2 Stage 3

Dry mix Wet mix Pump Stage 4
Cement

Printhead

M M M

Figure 2.9: Stages in processing 3PCs depicting positions to alter the 3PC rheology (Source:
W. L. R. Da Silva).

27
Literature and theoretical research 2.2. Digital construction

Though less common, short polymer fibres are used in 3PCs to achieve better build-
ability and lower plastic shrinkage cracking in fresh state. It is known that usage of short
fibres brings many enhancements to the mechanical behaviour of concrete including
higher tensile strengths, enhanced post-crack behaviour and high residual strengths.
However, usage of short fibres complicates the extrusion process and increases the prone-
ness of blockages during pumping of concrete. Example applications of polypropylene,
high-density polyethylene, basalt and carbon fibres with lengths 3-12 mm in 3PCs can be
referred in [2, 81, 119].

2.2.6 Machine technological and process aspects of DC

Every DC approach needs four basic sets of “machines”, responsible for 3PC production,
3PC transportation, positioning the printhead and 3PC deposition, respectively. For
positioning portal robots or multi-axis industrial robots are generally used; also refer to
Section 1.2.1 for discussion on various stages in processing 3PCs. The common practice
so far is to produce 3PCs in a single-stage mixing and transport them with a progressive
cavity pump (PCP) to the printhead nozzle. This is followed by extruding 3PCs,
unaltered, through the nozzle under the pressure of pumping. This common approach is
represented by the process chain B-Ib-1 in Figure 2.10, which illustrates different process
chains of DC. Examples of B-Ib-1 are demonstrated by Contour Crafting [120], Concrete
Printing [2] and also in [8, 121]. Note that often a single mixing-plus-pumping unit, such
as those used in mortar and screed applications, is used for the stages of production and
transportation of 3PCs, for example, refer [8]. Other approaches for the production and
transport phases are using ready-mixed concrete or dry-mix/constituents. The ready-mix
concrete variant is preferable for on-site applications where the printing site is far
from the concrete production plant. A base mix whose hydration is either slowed-down
with the help of superplasticisers or diminished with retarders is used with rheological
properties tailored for pumpability. Therefore, if ready-mix concrete is used in the
production and transport phase, a chemical [122], mechanical [123] or external [14]
interference to alter rheology of the 3PC in the printhead-stage is necessary. To facilitate
this, a “reservoir” and second-stage mixer is typically needed. Upon this background, a
process route of A-I-1 is not feasible. Another compelling process variant is transporting
individual constituents or dry-mix to the printhead and wet-mixing it shortly prior
to the deposition. This C-IC-3 process is highly applicable for applications such as
on-site printing of wind-mill towers. If surrounding conditions such as temperature,
wind are volatile the C-IC-3 approach can be very effective. Since the wet-mixing is
taking places shortly prior to the deposition, it is possible to produce compositions
with high yield stresses. This approach also makes it possible to produce structures
with gradient/tailored concrete properties at various parts of the structures. So far, no
practical implementation of this approach is reported.
The most common pump type used for transport and extrusion phase in DC is PCPs.
Their ability to process dense suspensions with a wide range of viscosities and high
precision discharge with relatively low back-flow make PCPs technically most suitable for
DC applications. PCPs are applicable not only for pumping 3PCs to printhead but also
as extruders and/or second-stage mixers on the printhead. An example printhead with
PCP is used in this research; see Section 3.2. Alternatively, piston-pumps could also be
used to deliver material to the printhead. Unlike PCPs, the piston pumps cannot be used
on the printhead due to their relatively larger sizes as of yet. As not the entire material
mass is sheared (plug-flow), the piston pumps could allow blocking free transport of
3PCs with fibre additions. See Section 2.5 for additional discussion on various type

28
Literature and theoretical research 2.2. Digital construction

of pumps/extruders that are used in DC and their operation principle and theoretical
descriptions.
Ground stage Transport stage Printhead stage

A Readymix plant 1 Nozzle outlet

I Cocnrete pump 2 Printhead

reservoir
Ia Piston pump
Material
B On-site/lab mixer Ib Screw pump optimization

Ic Conveyer
Nozzle outlet

3
Dry mix
C Dry material or II Manual
dry-mix transport
Wet mixing

Nozzle outlet

Figure 2.10: Various possible DC processes. The CONPrint3D primarily follows “A-Ia/1b-2”.
Note that for the chemical alteration of 3PCs rheological state in the printhead second
stage mixers are needed. Though it is possible to use static mixers for the printhead-stage
mixing, a non-stationary mixer can ensure thorough homogenisation of the PCA into the
high yield stress concrete which is under high pressure in the printhead extruder. It is
also essential to have a dosing and pumping unit on the printhead do precisely input
the accelerators or other PCAs. The author tested the feasibility of this approach with an
electro-magnetic pump for dosing accelerator through an adapter at the entrance of the
PCPs stator-rotor assembly. The study revealed the following critical results:
a) The pumping-and-mixing unit needs to be light-weight as it needs to be mounted
on the printhead and is traversed by the robot. Therefore, it is preferable to use the
PCP (which is necessary for concrete deposition) as multi-functional mixing unit.
b) It is possible to easily integrate the second-stage pump into the central control
system of the 3D-printer and automate its operation. The homogenisation of the
pumped liquid is sufficient and an instantaneous setting of 3PCs could be achieved.
c) The primary challenge is to maintain the pressure in the PCA pump higher than
the pressure in the PCP, thus to ensure differential pressure needed for the PCA to
flow into the concrete stream. The pressure in PCP is transient and increases with
increasing flow rate of 3PCs. Therefore it is not-trivial to maintain a constant dosage
of PCA, thus resulting in heterogeneous extrudate.
d) Since in DC material discharge is discontinuous, when PCP does not move the 3PC
in the vicinity of PCA outlet hardens and blocks the flow of PCA when the printing
process is resumed again. Further research is necessary to develop non-bulky
second-stage pumping and mixing units for PCA dosage in the printhead. Note
that, such a unit shall also regulate the amount of PCA being pumped precisely.
A further significant component of a 3D-concrete-printer is the forming elements that
are mounted at the nozzle. The forming elements act as temporary formworks and are
responsible for giving the shape and surface finish to the extrudate. Depending on the
setup, the forming elements can be classified into static or dynamic forming elements;
see Figure 2.11.

29
Literature and theoretical research 2.2. Digital construction

a) b) c)
Figure 2.11: Forming element and nozzle setups in DC systems: a) stationary, circular nozzle
for vertical deposition [77], b) static, rectangular nozzle for horizontal deposition, c) dynamic
setup with a rectangular nozzle and movable forming element [59].

For both variants, extrudate can either retain the dimensions of the nozzle outlet or can
spread to a much larger size. To achieve the latter either low yield-stress 3PC is used or
higher volume of 3PC is pumped than the volume flow rate obtained by multiplying the
nozzle cross-section with the nozzle-layer separation distance and horizontal printhead
velocity3 . As a consequence, extruded material is forced to spread, taking a larger
cross-section than that of the nozzle; see Figure 2.11a. Excluding this size variation, in
stationary forming elements, the shape and geometry of the extruded filament is fixed
and cannot be dynamically altered. Example of static forming elements can be referred to
in [2,8,12,62,121]; see also Figures 2.11a and 2.11b. A dynamic forming element (trowel)
is shown in Figure 2.11c; the trowel can be deflected dynamically to obtain free-form
elements with smooth surface finish [120]. Note that none of the examples provided in
Figure 2.11 is capable of producing elements with varying cross-section and producing
right-angled layers; functions with a multitude of practical benefits. Figure 2.12 presents
a dynamic forming element concept for producing right-angled walls.
At this stage the concept of the printhead is clear, a printhead contains the nozzle
and forming elements of a 3D-printer, and may contain a reservoir as well as a PCA
dosage-and-mixing unit. In addition, the printhead generally also is equipped with
various sensors, for example, distance, temperature, PCP discharge rate. CNC portal
(gantry) systems or multi-axial robotic arms are used commonly for positioning the
printhead. Examples for portal systems are Contour Crafting [120], Concrete Printing,
TU/e [8] project and examples for robotic arm systems are Smart Dynamic Casting [64]
and XtreeE [107]. Note that both portal printhead and robotic arm can also be placed on
a rail-system to extend the printing-volume; see Figure 2.19b. The advantages of portal
systems are high load-carrying capacity and high dimensional accuracy when subjected
to external influences such as wind. They are also more established than the industrial
robots in common practice. Industrial robots enable production of elements with very
complex geometries as they have more degrees of motion. Detailed descriptions of various
machine systems for pumping 3PCs, positioning the printhead and depositing 3PCs are
presented in Näther et al. [124].
3 The height from which 3PCs are deposited on to already deposited layers.

30
Literature and theoretical research 2.2. Digital construction

Figure 2.12: Forming elements with multiple deployable side-planes that enable production of
rectangular layers and right-angled corners (Source: M. Näther, TU Dresden).

2.2.7 Cost structure in DC

This subsection is part of De Schutter et al. [28] and was originally co-written by the
author. In the case of DC, with the help of assured rheological properties and control of
their evolution in time [14,16], placed cementitious material can retain its shape without
any formwork. This can lead to substantial cost reduction. Furthermore, cost reductions
could be achieved by means of the quasi unlimited geometrical freedom, which enables
realisation of smart structures with minimal usage of material, without compromising
structural performance.
The cost structure for DC processes can differ significantly in terms of detailed aspects
and their respective magnitudes when compared to conventional construction. At this
point in time, it is too ambitious to quantitatively present the cost structure of DC since
real scale projects are still very rare and mostly only prototypes are produced for which
cost structure may not be identical to that of large scale implementation. Subjected
to qualitative scrutiny, the four broad components of any construction process are
a) labour, b) machinery, c) material as well as d) design and planning costs (3D-models,
BIM) ; see Figure 2.13a. Further costs do exist, for example, research and development,
testing, consulting fees etc. These costs could be much higher in the case of DC than in
conventional construction. However, such intricacies vary for each DC application and
cannot be generally quantified yet.

Labour
Materials

Equipment Design,
BIM...

a) b)
Figure 2.13: a) Primary components of the cost structure in DC, b) prognosis of variation in
component costs for different application scenarios (qualitative estimations).
In case of DC, total labour costs will be significantly lower than that of conventional
construction, thanks to the automation. The costs of machinery, in principle, depend
on the particular DC approach and the applied techniques. For the sake of discussion,
three major sub-processes are identified: a) transporting material to the printhead,
b) precise positioning of the printhead and c) extrusion or activation of parts of the
desired structure’s component. Machinery used for these sub-processes varies depending
on the DC technique and so are the costs. In a different perspective, the machinery used
for DC can be categorised into a) unconventional construction equipment (UCE) and

31
Literature and theoretical research 2.2. Digital construction

b) (adapted) conventional construction equipment (CCE).

For transportation process often CCE such as piston pumps can be used. For extrusion
or activation, however, in general new equipment, generalized as printhead, is needed.
This is however only true if the printhead has a (multi-)functional complex design, or
in other words if the printhead is not a mere nozzle. In fact, for a successful industrial
implementation of DC, the printhead should have size and shape flexibility (adjustable
nozzle outlet with respect to dimensions and shape) and should be equipped with sensors
which measure in-line the evolution of material properties for process monitoring and
active rheology control. In some techniques of DC, printheads should also be equipped
with multiple nozzles for different materials (e.g., concrete shell and insulation infill).
All these sophistications will make print heads a significant cost factor of DC. When
it comes to the sub-process placing, both CCEs and UCEs are utilised already to some
extent. UCEs such as multi-axial robots [2,62,65,125] and gantry or portal-platforms [59]
are often used for DC. At the same time CCEs, such as truck-mounted concrete pumps
with improvements to the controlling, can be used for the new technology as well as in
their original function [12, 39]; see Section 2.2.9. The more CCEs are incorporated into
the machinery of DC sub-processes the faster will be the industrial acceptance, the lower
will be the total costs of DC and also the higher will be the sustainability.
On the one hand, the total material costs for DC can be lower than those of conventional
construction due to topology optimisation, avoidance of over-engineering and waste
material reduction. On the other hand, they could also be higher in the case of DC due to
the utilisation of expensive admixtures. The concretes used for DC are by material design
more complex (with more constituents and higher requirements on raw materials) so that
they can meet the demands of problem-free processing, high early age structural build-up
[7, 27] and transient mechanical strengths [15, 16]; see Table 1.2. Similarly complex is
ensuring adequate mechanical performance and durability, considering the high risk of
anisotropy. This also has a consequence to concrete producers and practitioners in terms
of mixing process (precise dosing and intensive mixing due to often high powder content)
as well as storage of ingredients. It is therefore envisioned, that material researchers and
concrete technologists will address these challenges and develop less complex yet more
robust cement-based materials applicable for DC. Best option would be that the printable
concrete complies with the general specifications according to the valid codes for concrete
design and construction and production of these advanced materials could be handled
with CCE to a great extent.
The costs for the final sub-process design and planning, which includes 3D-modelling,
BIM etc. are expected to gradually decrease considering the advancements in related
fields. This decrease in costs is particularly likely in case of large-scale implementation
(as the same algorithms and database are used numerous times reducing costs per unit).
Considering the seamless planning-to-fabrication principle of DC construction and the
use of well-proofed algorithms, planning will be more precise (low potential for human
error) and less complex with respect to human involvement. The high reusability of
digital data will eventually render the costs for planning in case of DC negligible in
comparison with conventional construction.
The wide-spread industrial implementation of DC for conventional structures strongly
depends on how “cost-effective” the particular DC solution is. The general understanding
is that the unit cost in case of DC is independent of the production — number of
components produced — as well as complexity. Based on this, it has been argued [70]
that DC is very cost-effective if the number of components produced is low. Figure 2.14a
presents how conventional injection moulding (non-cementitious applications) is only
cost-effective at higher number of components and how additive manufacturing is inde-

32
Literature and theoretical research 2.2. Digital construction

pendent of number of components; see 2 , 1 in Figure 2.14a. A more pragmatic analysis

reveals that even in case of AM and DC, the cost-production ratio is non-linear. Higher de-
velopment, component costs for machines and materials, and high costs for digitalisation
and process monitoring, do require high initial investments in case of DC. Thus, if only a
few components are to be produced the costs in DC could be high. However, the costs per
component are expected to decrease at a more rapid rate with increase in production,
when compared to conventional construction processes; see 3 in Figure 2.14a. The
common consensus is that DC processes have higher construction rates (height or length
per hour) than conventional construction, consequently, reducing construction time and
costs. When various DC approaches are compared with each other, it can be observed
that the construction times vary widely based on the DC technology and application.
Geometrically complex structures with high spatial resolution need longer production
times than comparable non-complex structure. This can be demonstrated by considering
the three examples given in Figure 2.7 with varying complexity. In the full-width-printing
CONPrint3D® approach just one passage of the printhead is needed to create a thick
horizontal “slice” of the wall, while in case of Contour Crafting or even more so in case
of the high-resolution Concrete Printing method developed in Loughborough, multiple
passes are needed to produce the same volume, which means longer printing time. It
also should be considered that costs for finer material to be used for structures with
high spatial resolution are higher. This relationship between construction rate, total costs
and complexity / spatial resolution is illustrated in Figure 2.14b. This topic is examined
further in conjunction with buildability tests in Chapter 7.
A way to overcome the high initial costs in case of DC is to use CCE with minimal
utilisation of UCEs such as industrial robots. One example of such concept is the
CONPrint3D® (Section 2.2.9), which utilises established truck-mounted boom pumps for
transportation, pumping and deposition of concrete. Another way to minimise the costs
is by using “ordinary” construction materials with coarse aggregates and lower cement
content; best materials according to the existing codes. Two material categories are
identified: a) Category Mat1: The material costs are currently higher than conventional
materials, due to expensive admixtures as well as high research, development and testing
costs. b) Category Mat2: With principles such as topological optimisation volume of
material can be drastically reduced, thus total material costs can be lower. Alternatively,
3PCs with similar costs of current general concrete, prospectively, with “ordinary” and
inexpensive components.
The above presented sub-categories of equipment (unconventional, conventional) and
materials (Mat1, Mat2), leads to four application scenarios of DC; see Figure 2.13b: a) un-
conventional equipment (e.g., robots) + expensive high-performance materials; named
DC-UCE-Mat1 b) conventional (at least partially) equipment + expensive high-perfor-
mance materials; named DC-CCE-Mat1 c) unconventional equipment (e.g., robots) +
low-cost materials; named DC-UCE-Mat2 d) conventional (at least partially) equipment
+ low-cost materials; named DC-CCE-Mat In conclusion, while most DC cases lead to
lower costs due to reduced costs in labour and design, the highest cost reduction can be
seen in combination of the optimal use of material (though material design and topology)
and use of conventional/inexpensive equipment.

33
 Literature and theoretical research 2.2. Digital construction

Unit cost of component

Additive
Injection moulding
manufacturing Construct
ion r

Costs/rate
ate

ts
1 Total cos
2

Number of components Complexity, spatial resolution

a) b)
Figure 2.14: a) Variation of unit cost per component in case of AM and injection moulding
(extended from [70]), 1 AM acc. [70], 2 injection moulding and 3 AM considering high
initial automation and material costs; b) total costs and construction rate (dashed line) as a
function of complexity and spatial resolution.

2.2.8 Challenges in large-scale implementation of digital construction

This section lists some of the current limitations and challenges of DC techniques.
Reliability: For a wide ranging implementation of DC in large-scale applications, the
reliability in terms, of process, functionality and structural durability are essential pre-
requisites. Especially, the rheological properties in fresh state and mechanical properties
in hardened state need to be guaranteed in a similar, if not superior, degree in comparison
to current construction standards. Contemporary DC techniques are characterised by
inconsistency due to the complex material requirements and lacking process-specific
standardisation. For many of current DC techniques expertly skilled work-force, e.g.,
robot operators are needed and extreme care in the construction execution is mandatory
[65].
2 3

Figure 2.15: Illustration of some process challenges: 1 curing and post-processing, 2

“waving” of the layers due to excessive material extrusion, 3 a minor “feeding” interruption
in the extruder resulting in structural irregularity for all the subsequent layers, 4
unintended printhead collision with printed wall, 5 achieving sharp and straight edges (note
that in this case, the edges are manually altered).
Minor variations in material properties can lead to total process break down in critical

34
Literature and theoretical research 2.2. Digital construction

cases and can lead to errors such as local geometrical deviations in less critical cases. In
case of on-site construction, the influences of temperature, wind and relative humidity
can be significant, if not mitigated. It is noteworthy that the degree of difficulty in
achieving reliability varies depending on the particular target application of DC. If the
business model for a particular DC is final "pre-fabricated" elements competing with
pre-cast ones, then higher degree of reliability will be needed than, for example, creating
temporary structures or printing integrated formwork.

Lack of adequate on-site DC technologies: According to [34, 43], Eastmann et al. [126]
reported that “studies of off-site production of building components (i.e. prefabrication)
have found a distinct labour productivity advantage in comparison to related on-site
activities” and that the rate of productivity growth is greater compared to on-site
sectors. Similar observations are also reported in [127]. Despite the necessity for on-site
approaches, a vast majority of DC technologies reported so far are focused on off-site
construction. On-site construction is more complex due to the external influences on the
material and printed structures. In addition, safety requirements for using industrial
robots on construction sites is a concern yet to be addressed.

Layer-to-layer bond strengths: The compressive strengths of DC structures are

comparable to contemporary cement-based materials such as OC, HPC, SCC,
FRC etc. Compressive strengths over 100 MP a were reliably achieved in many
investigations [62, 71, 119], including this one [12, 124]. However, the primary concern
with respect to mechanical properties of 3PCs is anisotropy caused by the layer-to-layer
interfaces. Commonly referred to as cold-joints, the weak-interfaces often are the
weakest links in 3D-printed structures and could lead to drastic reduction of their
flexural strength; see Chapter 8 and Section 2.7, for more details.

Shrinkage: Due to their formwork-free nature, extruded DC structures are exposed to

surroundings at a very early age, right after extrusion. As a consequence, significant water
migration to the surface and evaporation to surroundings occur, resulting in plastic and
drying shrinkage. Depending on the temperature, wind velocity, material composition,
and restrain, the shrinkage can be critical, often leading to large cracks, which grow
through the entire height of a specimen/structure; see Figure 2.16.
Furthermore, multiple micro-cracks are often observed on the exterior layers/surfaces.
Often in DC, relatively high amounts of cement are used with no coarse aggregates
increasing the risk of (all forms of) shrinkage. Additionally, the longer the time interval
between layers the higher is the risk of shrinkage and shrinkage cracks due to differential
contraction of neighbouring layers and the difference in their elastic moduli. In addition
to cracking, shrinkage can also lead to weak-interfaces between layers [128] and thus
influences durability of the printed structures. Due to often low w/c, autogenous
shrinkage can play a significant role in 3PCs. Thus, additional to external curing the
possibilities of internal curing such as usage of superabsorbent polymers should be
considered. Despite the utmost significance of this topic, no research is published on this
topic in peer-reviewed journals so far; keeping shrinkage one of the unresolved issues for
on-site DC.

Implementation of reinforcement: The use of reinforcement, often steel reinforce-

ment, is mandatory in most structural applications. Despite the significant progress
in non-reinforced concrete printing [1, 28, 59, 61, 65] the solutions for incorporating
reinforcement into fully automated DC approaches are limited [29, 77, 129]. Most of

35
Literature and theoretical research 2.2. Digital construction

the suggested solutions are still rudimentary [29]. The lack of satisfactory solutions for
integrating reinforcement is one of the important challenges that need to be solved
before wide ranging implementation of DC. The DC approaches across the world
have numerous technological differences; see Section 2.2.2. Therefore, the solutions for
incorporating reinforcement must also consider core principles and process specifics of
various DC technologies and shall be optimised correspondingly. The existing approaches
are summarised as follows:
a) Placement of conventional, cable, fibre, textile reinforcement with concrete deposi-
tion or between 3D-printed concrete layers:
(i) Placing reinforcement such as steel bars or technical textile, horizontally
between 3D-printed concrete layers [5, 77],
(ii) Extruding continuous metal cables/chains along with concrete layer [130],
(iii) 3D-printing integrated concrete formwork and placing vertical reinforcement
in that formwork which will be filled with a flowable or vibrated concrete
[131],
(iv) Pre-stressing with post-placing of reinforcement in 3D-printed conduits [129],
b) “Enveloping” vertical steel reinforcement with concrete during deposition [5],
c) Using robotically produced reinforcement as a functional formwork in which
concrete is placed subsequently [6],
d) 3D-printing with cementitious composites reinforced with dispersed short fibre
[119],
e) Advanced robotic production/placement of reinforcement [6, 29, 59] and concrete,
preferably in parallel.

O
1 O

S S

a) b) c)
Figure 2.16: Shrinkage cracks in extruded concrete elements: a) four layers produced with TI
of 2 min (photographed at age of 1 day), b) overlay printed with TI 1 day, (photographed at
120 min after overlay deposition), c) hardened specimen with overlay-to-substrate TI of 1
day. 1 Dashed lines indicate border between substrate S and overlay O .
Despite the encouraging concepts and prototypes, none of these approaches satisfies
the entire spectrum of relevant requirements for wide ranging implementation of DC;
thus, they should be developed further. The majority of the validated approaches
are discontinuous, meaning that concrete printing and reinforcement placement must
be carried out one after the other. This leads to a) longer construction times with
higher costs and/or b) need of additional equipment for placing reinforcement. The
pre-stressing of post-placed reinforcement is only possible if continuous conduits are
formed during 3D-concrete-printing, effecting the geometrical flexibility. Furthermore,

36
Literature and theoretical research 2.2. Digital construction

the weak layer-to-layer interfaces may also function as easy passages for transport of
aggressive fluids and gases into printed structures. This is an important concern if steel
reinforcement is to be used in DC and the risk of corrosion could be elevated. The
“optimum” solution for incorporating reinforcement in DC shall address compliance to
the legal regulations, ease of technological feasibility, economic viability, portability and
most importantly structural integrity and durability. This necessitates extensive research
and innovative solutions.
Based on the reinforcement integration technology, the categories “a” to “e” can further
be subdivided into continuous or semi-continuous or discontinuous processes. Above
mentioned approaches “d” and “e” can be classified as continuous. In contrast, “b” and
“c” are classified as discontinuous due to the necessity of placing reinforcement either
before or after 3D-concrete-printing. The approach “ai” belongs to the semi-continuous
case [77]. The approach “aii” represents a continuous case [130]. The approaches “aiii”
and “aiv” are the discontinuous case; see Figures 2.5a and 2.7c.
In all the approaches, it is important that the dimensions and degree of reinforcement
should be appropriately designed based on the geometry of a single layer and that of
the target structure [132]. The primary advantage of using conventional reinforcement
is that design and reinforcement assessment can be carried out relying on established
regulations and standards [132]. It must be noted here that though placing horizontal
reinforcement between concrete layers is technologically simpler, vertical reinforcement
is technically more essential in most on-site applications. For the integration of vertical
reinforcement for on-site application, currently the most feasible approach is 3D-printing
formwork with concrete and then placing steel reinforcement — a discontinuous ap-
proach [87]. An alternative approach for using steel reinforcement in DC is pre-stressing.
The Wonder Bench at the University of Loughborough [129] is an example, where 23
voids were designed to form conduits for the post-placement of reinforcement and
installations; see Figure 2.7c. The post-placed reinforcement is then post-tensioned and
enveloped with grout [129]. In a similar fashion, the approach of pre-stressing reinforce-
ment placed through conduits can also be used for DC technologies based on selective
activation. Since in selective activation the “activated” material is always “carried” by the
support material, it is possible to produce free-form structures with complex openings
which can be post-reinforced. Post-reinforcement of elements produced by selective
activation is already demonstrated by the Institute of Advanced Architecture of Catalonia
[133].
Vertical steel reinforcement meshes/mats can be also enveloped with concrete which
is additively deposited. So far two proven DC technologies are following this approach:
a) Smart Dynamic Casting (SDC) and b) forked nozzle approach by HuaShang Tengda
Ltd; see Section 2.2.2. The advanced slip-forming approach such as SDC is applicable
for slender elements with varying cross-section. The target structure’s form-matching
reinforcement can be vertically held in place. The adaptive formwork is robotically
moved vertically while concrete passing this formwork envelops the reinforcement.
In the approach by HuaShang Tengda Ltd, large steel reinforced structures can be
produced, e.g., 400 m2 two-story villa [5]. HuaShang Tengda Ltd uses a forked nozzle
that simultaneously lays concrete on both sides of the steel mats, encasing it securely
within concrete layers [5]; see Figure 2.5b.
Another very promising approach is the Mesh Mould [6], in which, an industrial robot
forms complex, free-form meshes which work both as formwork and reinforcement.
The mesh can consist of robotically bent, cut, and welded steel or 3D-printed polymer.
Eventually, fresh concrete of specific rheology is placed into this formwork and the
surfaces are finished. Khoshnevis et al. [59] conceptualized a system consisting of

37
Literature and theoretical research 2.2. Digital construction

prefabricated steel elements assembled step-by-step by a robot in accordance with the

progress of concrete layers and the filling of the “formwork” with flowable concrete.
Another highly innovative contentious approach is parallel additive manufacturing
of steel or polymer reinforcement and concrete. No practical implementations of this
approach were reported, probably due to the elaborative machine systems that are
needed; and the corresponding expenses.

Process limitations: Production of structures right-angled and sharp-edged elements

is a serious challenge for the current DC techniques. Successful realisation of such
structures needs precise control of printing process, material rheological properties and
most importantly forming elements at the nozzle; see Figure 2.12. Another related issue
is the synchronisation of concrete flow rate and printhead velocity; see Section 3.2.1 and
Section 6.1 for more details. Each electric motor has an acceleration and deceleration
ramp, i.e., the rate at which prescribed velocity is achieved from rest (zero velocity).
In large 3D-concrete-printers, often, the printhead and concrete pump are driven with
different types of motors. A fully foolproof synchronisation of these different motor
systems is not trivial. In addition, the concrete flow rate from extruders, e.g., progres-
sive cavity pumps (PCPs) depends on numerous parameters including time-dependent
rheological properties, extruder geometry, stator-rotor wear state, and surrounding
temperature. As a result, the material-printhead synchronisation is one of the critical
tasks in DC applications. A common consequence of non-optimal synchronisation is layer
inconsistencies occurring when the printing directions are changed [65]; see Figure 2.24.
Furthermore, more critical consequence is insufficient nozzle-filling coefficient, which
results in lower mechanical properties and weak-joints.
Significant development and optimisation are still necessary for the CAM aspects
of DC, especially the optimisation of the algorithms for slicing, print(tool)-path
design. So does the robustness of continuous 3D-printing processes, with numer-
ous pause-and-plays wherever necessary. One of the challenging issues’ geometry in
3D-concrete-printing by extrusion is ensuring consistent layers every time the printing
process paused and resumed or when the printing direction and/or speed is changed,
as emphasized by Buswell et al. [65]. So far no general-purpose slicing software for
3D-printing with 3PCs is available. The material properties play a significant role in the
optimal tool path as well as the element slicing (e.g., the thickness of a layer should
not be smaller than maximum aggregate size). The current common slicing software
tools are inadequate as they were (mostly) developed for polymer materials and lack
functionality that is specific to concrete construction. Since most current slicing software
tools are developed for filament printing, their applicability is yet to be verified for
the full-width printing, the process followed in the CONPrint3D® approach. Due to
the anisotropic mechanical properties of printed concrete, the construction process
and material composition should be carefully decided upon considering the structure
topology and function.
In addition, many of the professed benefits of DC are either yet to be tested on
large-scale or lacking necessary tools for validation. Material savings, for example,
following topology optimisation and form follows force principles is one of the common
claims of DC advocates. DC enables selective deposition of material where it is necessary
and in the exact needed form and quantity. However, implementation and validation of
such concepts are scarce [134]. The present optimisation techniques and corresponding
design constraints, traction compensations need to be reassessed and adapted for
digitally constructed concrete structures. Such adaptation techniques should also
consider the fact that the optimum solution based on established design principles based

38
Literature and theoretical research 2.2. Digital construction

on structural behaviour may not be an optimal solution considering the printing process
and machine constraints.

DC of inclined and horizontal elements: So far, there are no technologically matured

solutions available for 3D-printing of inclined, overhanging or cantilevered structures
and structures with openings (windows, doors, etc.). Some of the proposed approaches
are; see also Figure 2.17: a) using sand as support material, b) using a powder-bed
(selective activation), c) using a recyclable 3D-printed support material, d) gradual
progression, e) placing a thin support structure such as a TRC plate, f) collaborating
robots with at least one robot holding a temporary support.

a) b) c)
Figure 2.17: Various approaches for printing inclined and horizontal elements: a) placing a
support “block” (sketch: M. Näther) b) extruding over a 3D-printed support material
(sketch: J. Kohl), c) gradual progression by purposefully altering deposition coordinates
(photograph: Universität Innsbruck).
The straight forward approach to address this challenge is to use support material
analogues to FDM techniques used for 3D-printing with polymers. However, these
techniques come with the drawback of additional printing time, material, labour costs.
Removal of support material is cumbersome and also restricts geometrical freedom. For
simple structures, however, the geometrical restrictions can be avoided and removal
of support material is relatively simple. Manually placed and formed sand (moist
or dry) could be used to produce horizontal DC elements. Note that, when used in
large-scale the support material approach is practical if the support material is, at least
partly, reusable and is autonomously deposited. This constraint largely limits use of
materials with mineral-based binders as support material. The author co-developed
a reusable, water-solvable and bio-based support material. The feasibility of using
this non-cementitious material is already verified [135]. However, there are still many
open questions concerning a) interactions and reaction between support material and
structural material (concrete), b) influence of support material on costs and time of
execution, c) level of automation, especially removal of support material, d) reusability
and e) adaptation of DC control algorithms and machines for multi-material extrusion,
f) topology of support materials for elements with varying inclinations and spans [135].

Lack of regulations and norms: The increased geometrical complexity of DC also

leads to challenges in testing and characterisation of such structures. Similarly, many
non-standard advancements that DC provides come along with the challenge of being
non-applicable under current norms and regulations. This limits the applicability of
DC only to prototypes and technology demonstrators. New standards or model codes
and regulations are of high practical significance. However, such a statute prerequisites
“reliable” experimental methods and a vast amount of experimental results. The few

39
Literature and theoretical research 2.2. Digital construction

ways to overcome this issue are a) producing hybrid structures such as 3D-printed
formwork and cast core; Figure 2.8b, b) strategic choice of DC process parameters
complying to current regulations; see Section 2.2.9, c) obtaining special permissions with
verification of the technology by extensive testing; refer the 3D-printed bridge designed
by testing in [84].

Economic viability: Total cost savings in comparison with existing technologies is not an
absolute necessity for the success of any new technology. Nonetheless, savings-potential
increases industrial acceptance and accelerates the mass implementation of new tech-
nologies many folds by increasing economic viability. Total costs largely depend on the
machine, labour, material, planning and management costs. All these parameters highly
depend on the choice of applications and construction technology. Majority of the current
DC approaches need expensive machines such as industrial robots and advanced exper-
imental and digital tools. In addition, though low in number, the labour for current DC
must be highly skilled, effectively compromising the economic viability. The materials
used for DC are also generally more expensive than contemporary ordinary concrete due
to the low dosage of sand/aggregates and utilisation of expensive admixtures. Eventually,
the costs of digitisation, printheads, robots and related resources would become more
and more affordable. However, if faster and wide-spread implementation of DC is to be
achieved, specific actions need to be taken in this regard.

2.2.9 Concrete on-site 3D-Printing — CONPrint3D®

To solve some of the challenges described in the previous section, the CONPrint3D®
— Concrete on-site 3D-printing technology — has been developed at the Technische
Universität Dresden; see Figure 1.2. This section presents a few key aspects of the
CONPrint3D® technology. Note that certain research decisions taken for this PhD work
were made considering the long-term objectives of the CONPrint3D® .
The CONPrint3D® focuses on applying 3D-concrete-printing in the field of on-site
concrete construction. The focal points of CONPrint3D® are to develop a time, labour-
and resource-efficient advanced construction process without diminishing the economic
viability and with broader acceptance from the practitioners. This is achieved by:
a) developing process-optimised 3PCs with reliable printability and high mechanical
performance by using common “off-the-shelf” constituents,
b) utilising existing construction and production techniques as much as possible and
by adapting the new process to construction site constraints,
c) selection of target application relying on systematic market research and consider-
ation of the technological feasibility of the CONPrint3D® ,
d) complying to established construction standards and regulations.
e) studying economic potential as well as legal and norm specifications.
In specific terms, above mentioned focal points are realised in the initial variant of the
CONPrint3D® :
a) Using 3PCs composed of Portland cement, fly ash, microsilica, natural sand, PCEs
and other common admixtures. The 3PCs are generally characterised by high
thixotropy, long (90 minutes) “printing window”, high compressive and flexural
strengths. They are tested with help of developed process-specific experimental and
numerical tools for material characterisation [11, 12, 26, 27, 29, 31, 80, 124].
b) Material compositions are developed with wide ranging properties and con-
stituents; see Chapters 4 to 8. Furthermore, mixtures with 8 mm aggregate size
and corresponding printhead were developed. These 3PCs enable applicability of

40
Literature and theoretical research 2.2. SBP: importance and characterisation

CONPrint3D® for wide range of applications.

c) Utilising adapted commercial concrete boom pump as an autonomous robotic ma-
nipulator with the help of a custom-developed printhead attached to the boom; see
Figure 1.2; thus ensuring low equipment, labour costs and faster implementation
into practice.
d) With an extrusion-based process that is distinguished by the layer cross-section
from the Contour Crafting [59]. The CONPrint3D® approach is fit for DC of massive
elements with width of layer = width of the element; see Figure 2.7.
e) Printing structures with established geometries concerning shapes, dimensions of
cross-sections, and connections. In CONPrint3D® layer cross-sections are rectan-
gular and connections are right-angled, as opposed to the common practice of
curved corners/connections in the state-of-the-art. Therefore, established norms
and guidelines could be applied.
f) The primary target applications is multi-story residential buildings. Specifically,
replacement of masonry walls is prioritized; a decision based on market research
(Section 2.2.4), economic viability [12, 39] and state of approaches for integrating
reinforcement in DC [29, 30, 92].
To develop comprehensive, multi-disciplinary solutions, three groups at the TU
Dresden — the Chair of Construction Machines and Conveying Technology (BFT), the
Institute of Construction Materials (IfB), and the Institute of Construction Management
(IBBW) — are working synergistically together on the project. General interdisciplinary
research results from CONPrint3D® feasibility study could be referred to in [12, 39,
136, 137]. The thesis at hand deals with the material aspects of the CONPrint3D® . The
investigations on process and machine aspects will be published elsewhere. Various
stages of the CONPrint3D® research framework have been funded by the German Federal
Institute for Research on Building, Urban Affairs and Spatial Development (BBSR), the
German Research Foundation (DFG) and Federal Ministry of Education and Research
(BMBF).

2.3 Structural build-up: importance and characterisation

This section presents the introductory information with respect to structural build-up
(SBP) of cementitious materials. Detailed descriptions of concrete rheology fundamen-
tals, rheometry, empirical test methods for rheological characterisation of concrete and
numerical simulation of fresh concrete can be referred to in [21, 138–146].
While referring to fresh concrete rheological behaviour, terminology such as “work-
able”, “flowable”, “consistency” and “stiff” have been used in the construction practice.
Researchers concede that these words are subjective and non-scientific [138, 147]. Still,
they can very well be the best choice in specific contexts. While investigating the inherent
changes in the rheology of cement-based materials in DC, the author has had to refer
frequently to the “flowability” of fresh mixes. In this context, the words “less flowable”
and “stiffer” are used when referred to pastes which exhibit higher resistance to flow.
In case of DC, cementitious materials used for extrusion-based processes are inherently
less flowable than vibrated ordinary concrete (OC) or self-compacting concretes (SCC);
see Figure 2.18. High static yield stress is necessary to assure the stability of the extruded
layers after deposition by the printhead. In the absence of formwork, extruded layers
must withstand the gravitational loads of self-weight and vertical hydrostatic pressure
[148] exerted by the layers placed ever upward, one by one. These requirements are often
met by increasing the structuration rate of the material at rest by a) controlling hydration
and/or setting kinetics through the addition of chemical admixtures [68, 101, 149] and

41
Literature and theoretical research 2.3. SBP: importance and characterisation

b) controlling the flocculation rate and, consequently, structural build-up by adding

supplementary cementitious materials and/or other fine constituents such as nano-clay
[150, 151].
When accelerators are used, it is possible to design flowable concretes for DC to ensure
easy, blocking-free pumping and subsequently adding accelerator in the printhead just
before material deposition. Such an approach implies the use of inline, second-stage
mixing techniques to ensure precise dosage and uniform distribution of admixture,
thus often leading to bulky printheads. While novel printheads with inline mixing
techniques are under development, today’s common practice in DC applications and
research is the usage of relatively stiff or even “zero-slump” concrete [8]. Independent of
the initial consistency of concrete and the manner of accelerator addition, concrete after
deposition must have sufficient shape stability to ensure continuous 3D-printing. Thus,
characterising rheological properties of concrete after deposition and their development
in time is of particular interest for DC, independent of individual solutions for concrete
processing before material deposition.

a) b)
Figure 2.18: Examples of a) freshly printed, fine-grained concrete in the fresh state, b) SCC,
showing the fundamental difference in their rheological behaviour at the time of placement [9].
The characterisation of static yield stress is of crucial significance in the context of
DC, not only in ensuring the “shape stability” of printed layers but also in defining the
printing process parameters, such as rate of printing [7] and cross-sectional geometry
of the extrudate. More importantly, the time interval between two layers – which is
crucial to avoid weak layer-interfaces or even “cold-joints” [7, 152] – is also influenced
by the structural build-up of cementitious materials. While there is no standard ap-
proach in quantifying structural build-up, rotational rheometry [113, 114, 139, 153–155],
plate rheometry [156–161] and some simple approaches such as undisturbed slump or
slump-flow test [162–164] and inclined plane [153, 155] methods have been applied
successfully. Considering rotational rheometry specifically, the most common measuring
protocols followed are the stress-growth test [163–165] and alternatively the hysteresis
loop [139, 158, 166, 167]. In stress-growth tests, material such as cement paste at rest
is sheared at a low, constant shear rate and corresponding shear stresses are measured
[155, 156, 164, 168]. The measured shear stress gradually increases to a peak/plateau
before descending to an equilibrium, representing the breakdown of structure built
during the rest period. The peak shear stress, “static yield stress”, increases with rest-time
and age. The rate of increase of static yield stress gives the structural build-up parameter
Athix . Most frequently, the stress-growth test is performed by shearing the test sample
at a very low, constant shear rate [113]. However, the literature shows variations in the
applied strain rate values. The following table summarises the important parameters

42
Literature and theoretical research 2.3. SBP: importance and characterisation

from selected publications.

Table 2.2: Shear rates used in various stress-growth tests.
Shear rate Duration γap Instrument Source
[s1 ] [s] [−]
0.01 100 1.0 Bohlin C-VOR 200 Mahaut et al. [114]
0.001 180 0.18 Anton Paar Perrot et al. [169]
0.005 - 2.5 C-VOR Bohlin® Roussel et al. [156]
0.001 – 0.02 80 0.08 – 1.6 Anton Paar MCR 302 Yuan et al. [170]
0.01 1.5 0.015 - Mostafa et al. [171]
Mahaut et al. [113] used a constant shear rate of 0.01 s−1 for 100 s to measure the
development of the static yield stress of cement pastes. Perrot et al. [169] measured the
static yield stress of cement paste at a shear rate of 0.001 s−1 for 180 s, pointing out that
at such low shear rates, the viscosity effects are neglected. Mahaut et al. [114] compared
the static yield stress of a suspension with 140 µm polystyrene beads in an emulsion at
shear rates of 0.01 and 0.003 s−1 , and found a 10 % difference among the yield stresses
measured. The stress-growth test enables measurement of the minimum shear stress at
which the solid-like behaviour of fresh cement paste diminishes, thus representing the
initiation of flow (flow-onset). Although the flow is very slow and concentrated in a very
narrow area, the structure is still considered to be disturbed [114]. Yuan et al. [165]
studied the sensitivity of measured static yield stress of cement pastes corresponding
to different applied shear rates; see Table 2.2. They suggested that when cement paste is
sheared at varying shear rates, different time periods may be needed to reach the peak
value or plateau shear stress [165].
Additionally, Yuan et al. [170] investigated the structural build-up of cement paste
utilising so-called small amplitude oscillatory shear (SAOS). The sample was subjected to
a continuous sinusoidal excitation with a controlled shear strain (or rather deformation).
The applied strain amplitude was selected to maintain an elastic deformation in the
sample; it was often within the range of 10-6 to 10-4 for cementitious pastes. During
the test, the storage modulus G0 and loss modulus G 00 are measured continuously. The
authors reported that the evolution of G’ measured by means of SAOS and that of static
yield stress follow a similar pattern, even though the material parameters have different
orders of magnitude [165, 170]. It is worth mentioning that during static yield stress
measurements the flow-onset occurs at large critical strains which depend on the network
of colloidal interactions between cement particles, whereas during SAOS, the flow occurs
at the lower critical strain which is linked to the C-S-H formation at the contact point
between aggregated cement grains [172]. The SAOS method enables a non-destructive
measurement and results in a “single batch and single cell” sampling. Therefore, SAOS
has the potential to be used in the context of DC. In this work, however, SAOS approach
is not further discussed, since the author focus on benefits and limitations of constant
shear rate tests.
There are various models describing the structural build-up of cement pastes, e.g.:
Cheng [173], Papo [174], Coussot [175] and Wallevik [176]. The linear model (Equa-
tion (2.1)) of Roussel et al. [156] is applicable in describing structural build-up up to a
characteristic time tc :

τ0 (t) = τ0 + Athix · t (2.1)

43
Literature and theoretical research 2.4. Pumping of concrete

where Athix is the structuration rate constant defined as the rate of increase in static yield
stress over the time at rest and τ0 is the initial yield stress of material prior to rest.
Perrot et al. [7] reported that structural build-up is exponential after tc and introduced
a model Equation (2.2) which asymptotically concurs with Roussel’s linear model prior
to tc [7]:
 
τ0 (trest ) = τ0,0 + Athix · tc etrest /tc − 1 (2.2)

Both Roussel’s and Perrot’s models are used in this thesis for the analysis of measured
static yield stresses; see Sections 4.1.4 and 4.2.3.

2.4 Pumping of concrete

In this section, the second phase in processing 3PCs for large-scale DC, pumping,
is addressed. General introduction to the topic, within the scope of the research at
hand, and brief state-of-the-art along with analytical description of flow of concrete
in a pipe are provided. Experimental characterisation of pumpability and approaches
for prediction of pumping pressure are briefly outlined. The influence of composition,
rheological and tribological properties of concrete, pipeline geometry and surrounding
conditions on the pumpability of concrete were reported extensively in recent literature
[9, 10, 18–22, 25, 177–182] and therefore are not addressed in this dissertation. The topic
of pumping/placing is also addressed in Chapter 5.

2.4.1 General introduction

Developed in the first half of the 20th century, concrete pumping has emerged as one of
the most important processing technologies in the construction industry. It often makes
possible the reduction of construction costs and considerably speeds up the construction
process. Due to expedite discharge rates and higher productivity, pumping of concrete
gradually overtook the classical way of transporting and placing concrete manually
with buckets or by other means. Steady and continuous advancements in the machine
technology (pumps, pipelines) and practitioners understanding of pumping processes
have ensured the success of concrete pumping technology.
Concrete pumping is also crucial to many advancements in related processes such as
DC. In the case of on-site DC techniques, concrete must be transported or pumped over
longer distances, periodically stopping after having pumped each layer, but without any
blockages or circuit breakdowns over longer time durations; see Figure 2.19.
The pressure required to pump concrete depends on its composition as well as on the
pumping specifications such as distance, height to pump, pipeline diameter, and concrete
discharge rate. Any changes in the composition of concrete such as water-to-binder ratio,
aggregate size distributions (grading) or admixtures can exert a pronounced influence
on its rheological properties, indeed its behaviour during pumping. Furthermore, for the
same compositions time-dependent changes in rheological properties alter their pump-
ing characteristics. In addition, variations in the pipeline geometry such as diameter
variation, direction changes (e.g., bends) and surface irregularities affect the pumping
process. The presence and properties of the lubricating layer have the most predominant
influence (excluding the definitive influence of flow rate) on the discharge pressures
[9, 10, 18, 21–23, 178]. The lubricating layer (LL) is formed at the pipe-concrete interface
and is attributed to shear-induced particle migration, wall effects and bleeding water
[10, 23, 24, 178]. The plastic viscosity of the lubricating layer and that of concrete have

44
Literature and theoretical research 2.4. Pumping of concrete

significant influence on pumpability. Whereas the influence of yield stress is in general

understood to be less pronounced [24, 80, 178] in comparison with plastic viscosity.
However, the yield stress influence on pumping pressures increases with decreasing
plastic viscosity of the particular concrete [182]. In other words, the significance of yield
stress for pumpability increases with decreasing yield stress to plastic viscosity ratio
τo /µ. It was proven that pumping pressures could be roughly doubled if pipe diameter
decreases by 20 %. However, the influence of pipe diameter on pumping pressure
loss decreases if concrete flow rate decreases or if τo /µ increases [182]. The shape of
the pipe, bent or straight, also has to be considered in predicting pumping pressure.
Commonly, the pressure losses due to pipe bends are accounted by replacing a bend with
an equivalent pipe length. For example, a 90° bend can be considered to cause pressure
loss equivalent to a straight pipe of 3 m length [179, 183]. Recently, it was also reported
that the influence of bends, and thus the equivalent lengths decrease with increasing flow
rate [25].

a) b)
Figure 2.19: Examples of concrete printing technologies: a) CONPrint3D® : a concrete boom
pump is depicted pumping and precision-placing fresh concrete (source: TU Dresden), b)
Contour Crafting: a concrete pump is depicted pumping concrete to a rail-mounted robotic
printhead (source: University of Southern California).
Due to the large number of influencing parameters, characterising pumpability and
predicting the discharge pressure needed to pump a particular concrete composition
over a given distance or height at a specific discharge rate is not trivial. Moreover, it
is complicated to measure directly the exact variation in the magnitude of discharge
pressure reduction depending on lubrication layer properties. Numerical simulations
offer an alternative to study the lubricating layer phenomena in concrete.

2.4.2 Analytical description of concrete pumping flow

In the light of the inadequacy of empirical methods, approaches based on analytical
models provide more accurate alternative for the prediction of pumping pressures.
The (Hagen-)Poiseuille’s law for laminar, steady state, isothermal, axisymmetric and
one-dimensional pressure-induced flow of a homogeneous Newtonian fluid with a
dynamic viscosity η in a cylindrical pipe of radius R and length L ∀R ≪ L [146, 184, 185]
is:
128ηLQ
P= (2.3)
πd 4
where P is total pressure and Q is discharge rate. Few assumptions that lead to
Equation (2.3) are addressed here, the full derivation and details are described in
[21,186]. For the Poiseuille flow the distribution of shear stress is linear; see Figure 2.20.

45
Literature and theoretical research 2.4. Pumping of concrete

The axial velocity u at a radial position r is described by Equation (2.4) in terms of

pressure drop ∆P (per length L). The velocity profile Equation (2.4) is parabolic with
its curvature increasing with increase in pressure difference and/or increase in Re [186].

∆P (R2 − r 2 )
uz = (2.4)
4ηL

Figure 2.20: 1 Shear stress τ, 2 shear rate γ̇ [187], and 3 velocity uz [187] profiles in a
representative plug plus shear flow during concrete pumping (adapted from [188]). 4 Pipe,
5 lubricating layer, 6 unsheared plug and 7 sheared plug.
The velocity is minimum at the pipe inner boundary and maximum at the central axis;
non-slip condition is assumed at the wall boundary. The shear rate and shear stress are
minimum at the central axis of the pipe and maximum at the pipe wall as shown in
Figure 2.20. The pressure drop occurs only along the flow direction and is assumed linear.
It is assumed that except pressure all other entities are constant along the axial direction.
Both by theory and experiment the (Hagen-)Poiseuille’s law is valid only if the flow is
laminar, for example, defined by the Reynold’s number Re lower than 2000. For a fluid
of density ρ flowing with mean velocity v in a pipeline of hydraulic diameter DH the
Reynold’s number is given by Equation (2.5). For flow of concrete in standard pipelines
the Re is mostly less than 100; see example calculations in [20, 185].

ρvDH
Re = (2.5)
η

The flow of heterogeneous concrete in pumping pipelines does not obey many of the
preconditions for the Poiseuille’s flow. Especially, the viscosity of concrete changes with
shear rate (non-Newtonian behaviour), and can be described in a simplified manner by
Bingham Equation (5.15). Corresponding to a fully-developed flow of a Bingham fluid
with yield stress τ0 and plastic viscosity µ, the Buckingham-Reiner equation describes
the flow rate in terms of pressure drop as [179]

3∆P 4 R4 + 16τ04 L4 − 8τ0 LR3 ∆P 3

Q=π (2.6)
24∆P 3 µL

Equation (2.6) does not consider one important phenomenon that occurs during
concrete pumping flow: the presence of plug and lubricating layer. Generally, the effect
of lubricating layer and shear-induced particle migration (in radial direction of the pipe)
on the pumping pressure is crucial. In fact, a perfect non-slip condition, as assumed in
Equation (2.3) is non-existent in concrete pumping flows, while the presence of high

46
Literature and theoretical research 2.4. Pumping of concrete

shearing lubricating layer or partial slip is evident [9, 18, 20, 21]. This condition can
be denoted by the so-called Navier slip or more accurately with the “lubrication” as
illustrated in Figure 5.2 from [142]. Due to this and other [146] limitations, the analytical
pressure loss predictions based on Equation (2.6), overestimate the pumping pressures
[9, 178]. In his research, Kaplan addressed this limitation extensively and suggested
a relationship between the wall shear stress τi , (Bingham) yield stress τ0,i and plastic
viscosity µi of the lubricating layer:

τi = τ0,i + µi γ̇ (2.7)

As shown in Figure 2.20 the shear stress decreases across the cross-section linearly from
the pipe wall to centre. Kaplan described that if the shear stresses at a radial position
are higher than the concrete bulk yield stress then the shear of bulk takes place. As a
consequence, depending on the properties of concrete and pumping conditions, concrete
can flow as slip (slip + plug) flow or slip + shear (slip + partial shear of bulk + plug)
flow [9, 10, 18]. Therefore, a condition for calculating the critical shear rate γ̇s at which
the transition from plug to shear flow occurs can be obtained:
τ0 − τ0,i
γ̇s = (2.8)
µi

Note that the shear rate γ̇ is related to the concrete flow rate Q. The probability of shear
in the plug increases with decreasing yield stress of the bulk concrete and increasing
flow rate / shear rate. Kaplan et al. [18] further proposed and validated the following
analytical models for pumping pressure using Bingham and interface parameters:

2L
 " #
Qµ i
+ τ0,i for slip flow


R πR2 k




P = (2.9)

 
Q R R

  2 −
2L  πR k 4µ τ 0,i + 3µ τ0 
+  for slip flow with partial shear of bulk
 
µ τ
 
i 0,i

R
1 + 4µ µi

 R 
 


The term k in above equations is a filling coefficient [9, 10, 22]. As shown in Equa-
tion (2.9) the prediction models consider only the properties of LL, if the concrete bulk
is not sheared; and both properties of LL and bulk concrete otherwise. The exact flow
profiles across the pipe cross-section are not identical for various concretes and their
experimental determination is not trivial. However, there are recent advancements with
respect to this issue [21, 179] and it is feasible to experimentally determine the type of
flow. For such purpose ultrasonic velocity profilers [181] and/or Sliper (indirect) can be
used [189, 190]. The Reynold’s number (Re) can also be used as indicator for the type
of flow: pure plug or plug+shear flow [20]. As Re takes viscous and inertial forces into
account, it can indeed be a rough metric that shows tendency of a concrete to flow as
pure plug (more laminar) or with partial shear (more towards turbulent). Additional
information on pressure-induced flow analysis of fresh concrete in a circular pipe can
be referred to in [9, 10, 18, 21, 22, 25, 185].

47
Literature and theoretical research 2.4. Pumping of concrete

2.4.3 Experimental characterisation and prediction of discharge pres-

sures
Various aspects of concrete pumping technology have witnessed continuous improve-
ments over the last century [185]. However, as De Schutter et al. noted, these improve-
ments are mostly based on trial and error approaches and are largely supported by
experiences in the field [185]. Determining the discharge pressure is an essential task
to ensure the success of concrete pumping process. Over the last two decades, much
research has been conducted on this topic, which has led in turn to different prediction
models and testing approaches [9, 10, 19, 22, 23, 183, 191]. A state-of-the-art paper on
prediction of concrete pumping pressures is published recently [24]. Conventionally, this
crucial task has been accomplished using approaches based on empirical test methods
[10] or experience [182]. Generally, organized in the form of monographs or estimation
charts, the conventional approaches take into account pumping circuit geometry and
fresh concrete flow parameters. “Depending on the origin of flow chart, the fresh
concrete property is either the slump or the spread [183]” [185]. However, such empirical
approaches have proven unreliable [10, 18, 22, 178, 185] because of, among others, the
differences in laboratory testing conditions and the corresponding actual situation during
pumping [10, 22, 178].
However, recent studies provided improved analytical models which enable accurate
prediction of pumping pressure [10, 18, 21–23, 25, 178]. Most of these approaches are
based on Kaplan’s models; see Section 2.4.2. In addition, new experimental tools were
developed such as the concrete tribometer [18] and the Sliding Pipe Rheometer (Sliper)
[10, 22]. In the first place, it is essential to consider the concrete flow mechanism
(Figure 2.20) in order to accurately predict pumping pressure. Once the type of flow is
identified, i.e., pure plug flow or plug + shear flow, appropriate models are to be applied.
Quantifying the necessary rheological and model parameters, however, is not trivial and
standardized. The influence of LL, for instance, is decisive for the pumping pressures. By
the least, the LL can have three parameters: yield stress τ0,LL , plastic viscosity µLL and
thickness TLL . One approach to characterise the TLL is using non-destructive techniques
such as ultrasonic velocity profiles [23, 181]. Alternatively, TLL can be computed using
variants of models from pumping pressure or from Tribometer measurements; see [25]
for detailed discussion on this aspect.
Accurately quantifying the rheological properties of LL needs additional experimental
tools. Kwon et al. [24] highlighted the two ways to characterise the lubricating layer: a)
with a tribometer and b) by using (a segment of) a pumping pipe. There are numerous
variants of tribometers proposed and applied for the characterisation of τ0,LL and µLL
[18, 19, 191, 192].
In general, concrete tribometers work on the same principle as co-axial viscometers
with Couette flow and are often developed by modifying an existing concrete rheometer:
Kaplan [9], and Ngo [177] worked with BTRheom, while Chapdelaine [19] used Tattersall
MK-III. However, unlike viscometers [191] where horizontal protrusions (ribs) are
deliberately mounted, tribometers have a smooth wall allowing the measurement of
properties at the cylinder-material interface. Tribometers enable prediction of pumping
pressure by applying Kaplan’s models. The differences in various tribometers may lead
to varying measured τ0,LL and µLL and therefore calculated pressure, at least to some
extent. In addition to the geometrical differences of the tribometers, the type of concrete
can also influence the applicability of a particular tribometer. Indeed, the prerequisite
for using tribometers is that the bulk of concrete is not sheared during the measurement.
As stated by Feys et al. [191], the relatively low yield stresses of highly-workable concrete

48
Literature and theoretical research 2.4. Pumping of concrete

(HWC) and self-consolidating concrete (SCC) result in increased probability of their bulk
yield stress being exceeded during tribometer measurements. Further advantages and
challenges of predicting pumping pressures with help of tribometer measurements can
be referred in [24, 191].
It is also possible to test the pumpability of concretes with full-scale tests. However,
the full-scale tests are not always feasible, e.g., if a large variety of compositions are
to be tested for developing 3PCs. An alternative approach for predicting pumping
pressure is using the Sliper (see Figure 5.2), which is a relatively new device developed
to overcome the conventional problems in testing the pumpability of concrete and
estimating discharge pressures for various concrete types [10,22]. Originally developed as
a practice-oriented and reliable pumpability testing device by Kasten [22], applicability
of Sliper for a variety of concrete types was extensively tested by Nerella et al. [180,193],
Mechtcherine et al. [10] and Secrieru [25].
The Sliper enables reliable estimation of pumping discharge pressures for various
pipeline configurations [10]. In this section, the working principle of Sliper is outlined.
For detailed description of Sliper and its application for concrete compositions, refer
to [10, 22, 25, 180, 190]. The crucial difference Sliper has in comparison with regular
rheometers is its very close adaption to real pumping processes as well as its relatively
simple and robust setup. “By using as its central element a piece of pipe with the actual
geometry of the pipe in operation and by applying a testing procedure which mimics
pumping at various speeds, the physical conditions in the concrete pumping process can
be efficiently reproduced” [10]. In Sliper (Figure 5.2) pumpability is tested by filling the
pipe placed in the topmost position with fresh concrete and eventually letting the pipe
slide downwards under the force of the weights attached to the pipe. Various speeds of
the pipe in the subsequent measurements are achieved by applying various weights. The
speed of the pipe, measured by an ultrasonic sensor, corresponds to the concrete flow rate
Q in the pipe, while the pressure P of concrete, measured at the piston head, is associated
with the pumping pressure. The measurements are then plotted to obtain pressure versus
flow rate relationship P − Q; see Figure 5.2. Analogous to the Bingham model, two
important parameters, here denoted as a and b, are calculated from the pressure and
flow rate values. The parameter a is a function of the intercept of the linear regression
line with the P -axis; it is directly related to the yield stress of the lubricating layer in
the vicinity of the pipe wall. The parameter b is a function of the slope of P − Q curve
related to the plastic viscosity of the material in the same region. A “fully-developed”
LL is a prerequisite for the prediction of pumping pressures using Sliper method; a
condition satisfied by applying so-called “pre-strokes” at the beginning of each Sliper
experiment [10, 180]. Accuracy of pumping prediction based on Sliper measurements
in comparison with large-scale field tests was validated by [10, 22, 25]. The Sliper tests
were also used to estimate the type of flow [25, 189] during pumping and the influence
of LL [190]. Furthermore, a numerical model based on CFD for simulating Sliper tests is
developed in this thesis; see Chapter 5.

2.4.3.1 Pumpability investigations on 3PCs

As mentioned in Section 1.2.1, placing/pumping constitutes an important stage in
processing of 3PCs. Considering the fact that 3PCs are often, have very high yield stress
(in the order of few kPa) and high viscosities, blocking free pumping can be challenging.
In addition, if concrete is dosed with a phase changing agent at the stage of initial mixing,
the risk of blocking can be pronounced. Additional challenges, including that due to high
structural build-up, in pumping 3PCs are described in detail in Section 5.1.

49
Literature and theoretical research 2.4. Pumping of concrete

Despite the significance, by the time of writing this dissertation, no studies were
reported on pumpability investigations on 3PCs. Majority of DC techniques are so far
limited to small scale laboratory tests and prefabrication plants. In such applications,
3PCs are not pumped in neither large volumes nor for long time periods. Thus, the
challenge of pumping this group of high viscous, thixotropic, yield stress cementitious
materials is yet to be confronted. With the advent of on-site applications for DC [84] and
the development of technologies such as CONPrint3D® [12], the scientific and industrial
interest on testing pumpability of 3PCs is expected to increase drastically.
Though pumpability is not investigated directly, many earlier researchers on DC
recognized its importance. Lim et al. identified pumpability as one of the four key
characteristics of DC [70]. They have defined the pumpability of 3PCs as the “The ease
and reliability with which material is moved through the delivery system” [70]. In his
theoretical analysis Roussel restates that there exists prerequisite on printable materials
initial fluidity to be pumped [91]. Furthermore, he [91] underlines that “the ability to
be pumped of a given material (3PCs) depends on their ability to form LL than on the
actual bulk rheology”, since concrete generally flows as plug flow. Though the previous
statement is true to some extent, the bulk rheology of 3PCs is still crucial for their
pumpability characteristics. Since the shear-induced particle migration is often described
as one of the underlying mechanisms for formation of LL, and since the shear rates across
the pipe cross-section also depend on materiel bulk rheology, it can be argued that the
ability to form LL depends on the bulk rheology. Furthermore, depending on at which
process stage the phase changing agents are added, the ability of 3PCs to form LL can
vary significantly during the pumping process. It was numerously emphasized that the
requirements for low plastic viscosity and yield stress to ensure pumpability of 3PCs
conflicts with the requirement of buildability [12, 91, 194]. With no dedicated study on
pumpability investigations in the context of DC, this research field is still very open and
needs imminent attention.

2.4.4 Numerical modelling of fresh concrete flow and pumping

The common numerical approaches in simulating the behaviour of fresh concrete are the
Computational Fluid Dynamics (CFD) [178, 195, 196] and the Discrete Element Method
(DEM) [143, 197]. In the DEM, concrete is modelled by a collection of interacting distinct
particles, which can establish and break contacts dynamically. The method enables
modelling of large strain and stress phenomena of heterogeneous materials. However,
its applicability to large-scale simulations (concrete volume of more than tens of litres)
is strongly hindered by the requirement that millions of particles be simulated and
subsequently DEM needs substantial computational resources.
In CFD, concrete is modelled as a single continuum phase or as multiple continuum
phases, while the modelled domain is discretised as interconnected mesh elements. This
means that the rheological properties of concrete in a phase are constant throughout
the modelled geometry. Numerical solution for fresh concrete’s rheological behaviour is
then made possible by adding apparent viscosity, commonly following the Bingham or
Herschel-Bulkley models, to the Navier-Stokes equations [195, 198, 199]. This method
excludes the interactions among solid particles from the simulation and, hence, is not the
optimal solution to simulate heterogeneous phenomena such as segregation and blocking
directly [146]. Nevertheless, CFD has proven to be a useful tool for simulating fresh
concrete flows where macroscopic phenomena such as pressure vs. flow rate relationships
during pumping are the focus [178, 181]. It is worthy of note that it is possible to couple
CFD calculations, representing the fluid phase, with DEM, which represents the particle

50
Literature and theoretical research 2.4. Pumping of concrete

phase, but such calculation procedures are complex, time-consuming, and still in the
early stages of their development [196].
It is possible to simulate large volume concrete flows such as casting [195] and full scale
pumping [181] using CFD. Thrane et al. [142] simulated mould-filling process of SCC as
well as slump flow with a phenomenological micro-mechanical CFD model. Tichko et
al. [200] developed a CFD model for computing formwork pressure by simulating form
filling through bottom-up concrete pumping. Further literature presents various CFD
modelling using single-phase flow and multi-phase approaches; see [146, 167, 195, 201–
205]. Jo et al. [178] developed a computational approach to estimate lubricating layer
(LL) in concrete pumping. They emphasized the inadequacy of conventional pumping
pressure prediction methods with the help of field measurements and a mock full-scale
pumping test. Tan et al. [206] developed a multi-phase numerical model to investigate
wear mechanisms of concrete piping wall. They combined DEM and CFD approaches:
DEM to model concrete aggregates as discrete particles and CFD to model the continuous
fluid phase.
Choi et al. [181] simulated fresh concrete flow in a 170 m long pumping circuit
using a single-fluid CFD approach. The best-fitting thickness for LL was determined
by comparing calculated pressure with the experimentally measured values at a flow
rate of 50 m3 /h. Furthermore, they analysed pressure profiles comparatively along the
circuit geometry with the pressure profiles measured experimentally and observed that
the deviations between the CFD-calculated results and experimental data were below
7 % [181]. Choi et al. [23] also investigated the formation of the lubrication layer and
simulated the mechanism of shear-induced particle migration (SIPM) during concrete
pumping. Influences of particle shapes were indirectly modelled by solving SIPM
equations, and by implementing a User-Defined Scalar into ANSYS Fluent [23]. It was
concluded that the LL was the dominant aspect in defining the flow of concrete during
pumping. Furthermore, it was found that numerical simulations considering particle
shape correlated well with experimentally measured velocity profiles using an Ultrasonic
Velocity Profiler. Chen et al. [207] performed CFD simulations of wet shotcrete flow in a
pipe, including modelling of LL. In addition to finding the best-fitting thickness of the LL,
the authors also compared numerically determined LL thicknesses with experimentally
measured LL thicknesses and found that LL for tested shotcrete compositions varied from
1.5 mm to 3.5 mm.
The numerical simulations have proven their applicability for analysing flow behaviour
of concrete in long pipe circuits during pumping. However, the research published so far
has not provided numerical models for simulating pumpability testing device, such as
the Sliper.

2.4.5 Influence of pumping on rheological properties

The pumping process induces, especially in large-scale, changes to the rheological and
morphological properties of concrete. For example, differences in the values of yield
stress and plastic viscosity of SCC before pumping and after pumping were recorded
[208, 209]. Variation in rheological properties may result from the “higher shear rates
leading to the dispersion of cement particles and depending on available residual
superplasticizer in the mixing water” [208]. In addition, the pressure-induced pumping
affects air-voids in concrete. The decrease or increase of the air-voids and thereby air
content depends on the mixture composition [183, 208], pipeline geometry [25, 183] and
operating pressures.
The ACI Committee [183] recommends maintaining the pumping pressure to the

51
Literature and theoretical research 2.5. Extrusion of concrete

lowest possible and minimising “free-fall” within a vertically descending pipeline to

reduce loss of entrained air. When it comes to plastic viscosity, in general, a substantial
decrease with increasing pumping (pressure, flow rate and time) was measured. This,
however, cannot be universally valid by the very nature of fluid mechanics, considering
“shear thickening” and “shear thinning” effects. The results published in [25] may
provide evidence for this argument; viscosity parameters were found to decrease for
many but increase for some compositions. Earlier observations on the pumping-induced
changes in yield stress of concrete are also inconclusive. Feys et al. stated that the
yield stress “can decrease, remain relatively constant or even increase dramatically
due to pumping”; attributing to the availability of active superplasticizer in concrete
[208]. Secrieru concluded that “pumping leads to increase in concrete yield stress but
reduces its viscosity” [25]. It is here noteworthy that, while the increase of concretes’
air-content generally leads to decrease in plastic viscosity, the effect on yield stress
depends on the size of air-voids [208]. Additional data on losses in slump, changes
of air-void systems, influence of pressure on rheological properties can be found in
[25, 208–210]. The pumping-induced changes in the rheological behaviour of 3PCs can
have significant consequences to the successful execution of DC. Hence, certain decisions
are taken keeping this into account. For example, Chapter 6 emphasizes inline measuring
techniques for extrudability of 3PCs.

2.5 Extrusion of concrete

Irrespective of the differences in machine configurations, extrudability is a prerequisite
for the success of all extrusion-based DC applications; see Figure 1.2. This section
presents a brief introduction and the essence from the state-of-the-art on concrete
extrusion in the context of DC. In this doctoral work, the conventional extrusion and
DC by extrusion are distinguished. This section does not aim to present a overview of
conventional concrete extrusion.

2.5.1 General information

In the case of applications with cement-based materials, extrusion is typically used for
production of panels and pressure pipes [211, 212]. Extrusion was reported as a process
for enhancing the mechanical properties of the extruded material by decreasing the
porosity of cement-based materials in comparison to conventional casting [213]. Extrud-
ability is not an intrinsic and physical or chemical property of cementitious materials, but
rather a specific property of a complex process. In this dissertation, the author defines
extrudability as “the ability of a material to pass through 3D-printer’s extruder and
nozzle configuration with minimal energy consumption”. This “new” definition diverges
from the earlier literature on conventional extrusion, in which extrudability is often
defined as a compound property which accounts not only for the ability of a material
to pass through an orifice but also its shape stability (the ability of the material to retain
its shape after the extrusion) [214, 215]. It is important to underline that in the context
of DC both extrudability and buildability must be seen as separate properties. Thus,
the extrudability of a cement-based material does not support the notion of its being
buildable as well.
Unlike pumping, not all extrusion processes can be considered as pressure-induced
flows. For instance, displacement of material in case of PCP does not occur based on
induced pressure. That being said, in all extrusion processes a pressure gradient from the
feed side to the discharge side exists, primarily due to a reduction in the diameter of the

52
Literature and theoretical research 2.5. Extrusion of concrete

extruder cylinder (barrel) towards the orifice/outlet. In the context of DC, it is imperative
to distinguish extrusion processes by the extruders’ working principle; see Figure 2.21:
a) Ram extruders: relative motion of a piston (ram) in a hollow cylinder filled with
material 7→ pressure-induced flow
b) Screw extruders: rotation of a helical surface (rotor) on a coaxial cylindrical shaft
(stator), in a hollow cylinder filled with material  mechanical displacement by
helicoid rotation
c) Progressive cavity pumps: rotation of a “helical” cylinder (rotor) in a hollow
cylinder which inner walls are fitted with reciprocating surface (stator). The stator
inner walls are generally made of elastomers. As the rotor rotates, it creates
sequentially opening-and-closing cavities  mechanical displacement by vacuum
generation.

2 2

a) Ram extruder b) Screw extruder c) PCP

Figure 2.21: Schematic representation of various extrusion principles. 1 Inlet, 2
discharge, 3 stator, 4 rotor and 5 cavity. (“b)” is adapted from [216]).
Progressive cavity pumps (PCP) or eccentric screw pumps are a category of positive
displacement pumps originally invented by Rene Moineau in the year 1929/1930 [217–
219]. In the petroleum industry PCPs were used as early as the 1970’s. PCPs’ robustness,
ability to pump heavy high viscous oils, efficiency, low power consumption, ability to
handle higher gaseous volume fractions or sand contents and low initial investment are
few advantages of using PCPs [219]. In the construction industry, PCPs are commonly
found in pumping of fine-grained construction materials such as mortar and plaster.
As indicated in Figure 2.21c, PCPs contain a stator and rotor assembly; and work based
on the principle of progressively opening conveying chambers with rotor’s eccentric mo-
tion in the stator. The rotors are generally single-threaded helical screws made of metal.
The stators are single-threaded and consist of a metal casing with abrasion-resistant
elastomer lining. The inner lining of the stator too has a helical shape similar to that of the
rotor; however, the pitch of stator and rotor can be different. Thanks to the eccentricity,
as the rotor rotates, the conveying chambers open and close progressively from the
inlet/suction side to the discharge side. More complicated configurations of PCPs can
be referred to in [219–221].
PCPs are versatile positive displacement pumps as they can handle a wide spectrum
of fluids with various viscosities, shear dependencies [220] and granular concentrations.

53
Literature and theoretical research 2.5. Extrusion of concrete

In PCP, a cavity closes at an equal volumetric rate as that of subsequent cavity’s opening
rate. As a result (at least in theory) a pulsation-less flow can be achieved [220]. PCPs
have consistent discharge rate and also the ability to pump the material backwards. The
latter ability is particularly useful for DC applications. Shortly before the printhead stops
after printing a layer A controlled reverse rotation can be pre-programmed to prevent
excessive deposition.
One major issue with screw extruders is the back-flow (filtration) of fine paste. The
screw pumps (extruders) have been used for millenniums to displace homogeneous
Newtonian fluids. In such cases, a gap may exist between the outermost point of the
rotor and inner walls of the stator, making construction of screw pumps easier, operation
cheaper and durability longer. The backwards-flow of homogeneous fluids when pumped
with screw pumps is not critical. As long as the fluid displaced by the rotating helicoid is
higher than that flowed back through the gaps, the fluid transport takes place. In contrast,
for non-homogeneous granular suspensions such as concrete, fine paste and fine grains
can filter through the gap. Due to this phase separation, transported material will have
increasingly higher granular concentration, higher yield stress and viscosity, eventually
leading to blockage of extrusion. The separation of concrete and paste phases due to
back-flow in screw extrusion is physically different to the pressure-induced filtration in
ram extrusion and concrete pumping. However, some common consequences due to both
phenomena are heterogeneous extrudate, increase in “work” needed for extrusion and in
some cases complete process disruption.
In contrast to simple screw pumps, a major advantage of PCPs is that lower risk
of counter-flow or slippage. This is enabled by mechanical interference of rotor with
inner walls of stator, dynamically closing the chambers from which pumping medium is
already displaced towards the discharge. The dynamic sealing “requires an elastomeric4
stator; or hydrodynamic sealing, when there is a fluid film” [221] between rotor and
stator. The slippage depends on the viscosity of the material, PCP geometry and
increases with differential pressure between the discharge and suction sides [220]. Usage
dependent wear of the elastomeric stator is another reason for the slippage. Therefore,
independent of the extruder type used in DC, regular measurements of Qe and P siv are
recommended.
Overall, the PCPs are widely applied for concrete pumping and extrusion due to
(among others) their:
a) ability to displace heterogeneous and densely packed granular materials,
b) ability to displace materials over a wide range of viscosities,
c) relatively low counter-flow,
d) high precision,
e) continuous material displacement and
f) light-weight to large scale.
In case of ram extruders, the first three points mentioned above are valid. Major lim-
itations for ram extruders to be incorporated in a freely traversable concrete 3D-printer
printhead are the discontinuous nature and relatively large sizes.

2.5.2 Analytical description

In this section, analytic descriptions of material flow in ram extruders and PCPs are
given. Among other applications, these models are instrumental for efficient design of
the particular extruders and for determining the extruders’ operation variables [221].
4 Stators made of rigid materials also exist, but they lead to excessive slippage due to the necessary

clearance between stator and rotor.

54
Literature and theoretical research 2.5. Extrusion of concrete

However, in the context of DC, the models are essential to compute accurately the
extrudate flow rate Qe . Note that, precise setting of deposition flow rate Qe in DC, also
requires consideration of volumetric efficiency Ψv , e of the extruder.

2.5.2.1 Progressive cavity pumps

Since addressing all type of PCP geometrical configurations [218–223] is not feasible,
only the general model is considered here. The following geometrical parameters are
necessary, all with length dimension in mm:
a) rotor diameter Dr ,
b) eccentricity E,
c) stator pitch Pst ,
d) minimum stator diameter Dst ,
e) maximum stator diameter Dstm ; in the example considered here Dstm = Dst +4E [223]
and
f) stator-rotor interference/overlap w.

Figure 2.22: A sectional view of a PCP geometry, PCP with negative interference is depicted
(reproduced from [223]).
The eccentricity E defines the relative displacement of the rotor centre with respect to
the centre of stator during operation; see Figure 2.22. In more exact terms, eccentricity
is recognized as 2E and is half of the distance between two extremes of rotor centroid.
Another important geometrical aspect of PCPs, the overlap w, can be expressed as w =
(Dr −Dst )/2. Interference was not included in the original concept of PCP by Moineau and
was only considered much later by Vetter [224]. Vetter identified, neglecting interference
resulted in divergence of his experimental results when compared to theoretical values
based on Moineau’s theory [218, 224]. Depending upon the overlap, i.e., algebraic
difference of the rotor diameter and the inner diameter of the stator, the PCPs can be
categorised into a) zero interference PCP, b) negative interference PCP and c) positive
interference PCP.
Moineau originally calculated discharge pressures in the sealing regions of the PCP
using Hagen-Poiseuille model detailed in Section 2.4.2 [184, 221].
The total cross-sectional area of PCP stator is the sum of the area of a rectangle with side
lengths of Dr and Dstm = Dst + 4E plus the areas of two semi-circles with rotor diameter
Dr ⇒ Atotal = Arectangle + Acircle . The area of free space in the cavities is the algebraic
difference of Atotal and area of stator ⇒ Af ree = Arectangle = 4·E·Dr . The volume of material
displaced using free area in the cavities and the pitch of the stator Pst is:

ve,pcp = 4 · E · Dr · Pst (2.10)

55
Literature and theoretical research 2.5. Extrusion of concrete

With the assumption of no slippage, the theoretical volumetric flow rate of PCP Qpcp,t
is calculated by multiplying Vpcp with rotational velocity of the rotor Nr :

Qpcp = 4 · E · Dr · Pst · nr (2.11)

The theoretical volumetric flow rate is also called pump capacity and is expressed
in combination with rotational speed. For a PCP operation time of tpcp , using Equa-
tion (2.11), the theoretical total volume of extrudate (i.e. material displaced) is expressed
as:

Ve−tot,pcp = Qpcp · tpcp (2.12)

The Equation (2.11) is valid when interference w = 0. For the cases of positive or
negative interference, the following equations are to be used [219]:

h 2
i


 4 · E · D r − 8 · E · w − π(w · D r − w ) · Pst · nr ∀w < 0

 4 · E · Dr − 8 · E · w − π(w · Dr + w2 )

h
Qpcp = 


Dr2 2p
 " # #
D
  p
2
 r
 + · arcsin w · Dr − w 2 − 2 w · Dr − w2 · Pst · nr ∀w > 0

 −w
2 2

Dr
(2.13)

Recently there have been CFD models published, simulating Newtonian flows in PCPs.
Gamboa et al. developed a simplified single-phase flow model considering the gap due
to elastomeric stator deformation [218]. Paladino et al. [223] developed a CFD model
for the “full transient 3D Navier-Stokes equations within a PCP, including the relative
motion of stator-rotor”. These models provide insights into the shear-rate, pressure and
flow profiles in the cavities and need to be extended for shear-dependent non-Newtonian
fluids, to be embedded as a virtual tool for DC. Using the above (or similar) models it is
possible to compute the flow rate as a function of geometrical parameters of PCPs. Thus,
computed flow rates shall be “corrected” by subtracting the slippage [218, 223]. When
testing multiple PCP configurations, the volumetric efficiency Ψe can be computed as:

Qpcp,exp
Ψe = (2.14)
Qpcp

The Qpcp,exp is volume flow rate of the extrudate measured in experiments.

2.5.2.2 Ram extruders

In this sub-section, the word extrusion implies extrusion by a ram extruder. Description
of a ram extrusion flow phenomena in terms of fundamental physical quantities facili-
tates prediction of extrusion forces/pressures, if ram extruders are used in production
processes. Furthermore, ram extruders could also be used as rheometers for high yield
stress fluids [225]. Numerous earlier works have dealt with modelling ram extrusion
flows of various yield-stress fluids [225–227]. A simple extruder consists of a barrel
(cylinder), a tapered die-entry (so-called dead-zone), an orifice and a die-land area; see
Section 2.5.3. In such a ram extruder, concrete exhibits as a plug flow in the barrel and
die-land; as a shear plus slip flow at the die entry (shaping zone), where billet diameter
Db reduces to orifice/die-land diameter d.
Therefore, ram extrusion models contain terms corresponding to frictional resistance

56
Literature and theoretical research 2.5. Extrusion of concrete

and bulk plastic deformation, both of which depend on the geometric reduction or
extrusion ratio (f (Db /d)). The frictional force is threefold: between concrete and barrel
walls (cylindrical) Ff ,b , between concrete and unsheared concrete (conical5 , in the
dead-zone) Ff ,d , as well as between concrete and outlet pipe (cylindrical, die-land area)
Ff ,l . Assuming Fp,d is the force needed to achieve bulk plastic deformation in between
the dead-zones (shaping zone), the total ram extrusion force REFt can be formulated as
shown in Equation (2.15).

REFt = Ff ,b + Ff ,d + Fp,d + Ff ,l (2.15)

Since cementitious materials have a yield stress and are non-Newtonian, rate depen-
dency must be considered to accurately describe their extrusion process. Analogous
to Herschel-Bulkley model, the rheological terms viz. initial resistance (yield stress),
viscosity (consistency) parameter and flow index are generally used in mathematical
modelling of ram extrusion flows.
Benbow et al. [226] proposed Equation (2.16) to describe the extrusion pressure in a
uni-directional extruder, where υ is the velocity of the material in the orifice, σ0 , Ff ,l0 are
uniaxial yield stress and wall frictional yield stress, the parameters α, β, n1, and n2 are
material dependent fitting parameters [227], while Ll and d are length and diameter of
die land, respectively. Benbow et al. [226] considered that work done on the material for
its diameter reduction and work done to overcome the frictional resistance at the die-land
constitute the total work done in ram extrusion. Martin et al. extended Equation (2.16)
for the case of multi-directional (multiple dies) extrusion [228]:

Db L
Pr = 2(σ0 + αυn1 )ln + 4(Ff ,l0 + βυn2 ) l (2.16)
d Dl

The parameter α, while not being an intrinsic material property (viscosity), is an

important limitation of Equation (2.16). Horrobin et al. [229], relying on FEM analysis,
reported an error of 21 % in yield stress calculations using the Equation (2.16)6 . To
improve the accuracy of the predictions, Basterfield et al. proposed Equation (2.17)√for
describing Fp,d using Gibson equation [227]. Note that the von Mises’ criterion (τ = σ / 3)
was applied and that ns , ks indicate flow index and flow consistency, respectively. θr is the
inclination angle of conical die-entry/dead-zone.

 √ ns +1 !n
πDb 2  √ Db 2ks 3 2υDb 2 s
 
Fp,d = · 2 3τ0 ln
 +
4 d 3ns 2ns d3
(2.17)
d 3ns
!
n
sin θr (1 + cosθr )s 1 −
Db

Depending on the materials used, the frictional resistance between concrete and barrel
Ff ,b can be significant; Section 6.4.2. Note that Equation (2.16) does not take the friction in
the barrel into account. Perrot et al. [225] developed a methodology to use ram extruder
as a combined rheo-tribometer, in which, both the Ff ,b and the shaping force Fp,d are
quantitatively evaluated. Perrot et al. measured the Ff ,b by fitting load cells on the outer
barrel walls assuming that the “wall friction stress on the inner walls of barrel (induced
5 Assumed conical. Visual observation by [225] on coloured model clay showed that the flow profile is
curvelinear.
6 For an extrusion ratio f (D /d) = 8.3̄.
b

57
Literature and theoretical research 2.5. Extrusion of concrete

by concrete) tends to move the barrel” in the direction of extrusion [225]. They emphasize
that Ff ,b decreases with decreasing billet length in the barrel. When the remaining billet
length equals the dead-zone length Ld , the Ff ,b component of REft equals to Zero, and
therefore measured REFt = Ff ,d + Fp,d + Ff ,l . Thus, to distinguish displacement dependent
Ff ,b from the REFt , quantifying the length of the dead-zone is necessary. The Ld could be
expressed by Equation (2.18).

Db − d
Ld = tan θ (2.18)
2
Perrot et al. [225] derived Equations (2.19) to (2.21) corresponding to the frictional
force required in the dead-zone and in the die-land [225]. Using Equations (2.17)
and (2.19) to (2.21), Perrot et al. [225] computed Herschel-Bulkley parameters of
cementitious paste and clays; the calculated results correlated with the values obtained
from shear vane tests.
1 dREFt nf r
Ff ,b = = Ff ,0 + kf r · υi (2.19)
πDb dLB

!2
τ0 πDb 2  Db 2 n f r
  
d n d
= cot θr ln + kf r υf r 2 1 − (2.20)
 
Fp,d F n
Ff ,b0 4  f ,b0 f r

Db d nf r  D 

Db 2 nf r 
 
n
= πdLl Ff ,b0 + kf r υf r 2 (2.21)

Ff ,l 
d n  fr

where, υi ,kf r and nf r are materiel velocity at the interface, friction factor and friction
index.

2.5.3 Experimental characterisation

From the literature discussing the characterising extrudability of cement-based materials,
three methods can be identified: the ram extruder method [214, 230]; see Section 2.5.3
and Figure 3.3, the penetration-resistance (PR) method [231, 232], and the squeeze flow
method [233]; see Section 2.5.3.
In the case of the ram extruder approach, the force required to extrude material
through a nozzle is a measure of its extrudability. Perrot et al. [215] studied the extrusion
flow conditions and the mechanisms of liquid filtration during the mortar’s extrusion in
ram extruders. They also developed a model to predict extrusion force by considering
the shaping force and frictional resistance acting between cementitious materials and the
cylinder’s inner surface [214, 234]. Furthermore, a criterion for predicting extrudability
was proposed by linking rheological properties with soil mechanics (consolidation theory,
Darcy’s law) [235]. A non-dimensional parameter next was defined as the measure of
extrudability. next considers the effects of both ease of extrusion, expressed by the
numerator (time taken for extrusion: the shorter the better), and the time window for
extrusion, expressed by the denominator (time taken by the extrudate to double its yield
stress at rest: the longer the better).
The penetration resistance (PR) and squeeze flow tests are not treated explicitly in
this dissertation, since both these approaches are not used in the chosen experimental
methodology. The interested reader can refer to with respect to Chen et al. [232] and

58
Literature and theoretical research 2.5. Extrusion of concrete

Barrel Die-entry
Force Die-land

Concrete 1 1

a) b)
Figure 2.23: Schematic of a: a) squeeze flow test., b) ram extruder showing extrusion force
components (adapted from [225]). 1 Plug-flow, 2 unsheared zone, 3 shear+slip flow in
the shaping zone.

Toutou et al. [233] the applicability of these approaches in evaluating the extrudability
of cement-based materials.

2.5.3.1 Extrudability investigations on 3PCs

Despite its significance and the recognition by the DC researchers, quantitative char-
acterisation of extrudability was rarely addressed in the previous investigations. The
extrudability of cement-based materials depends on their composition, the geometry
of the extruder, and the process parameters, such as printing rate and nozzle outlet
geometry (layer cross-section). In the case of DC, if the extrudate’s flow rate is not
accurately synchronized with the velocity of the printhead, the printed structure will
be not continuous or coherent; see Figure 2.24.

a) b)
Figure 2.24: Results of asynchronous 3D-printing: a) printhead is too fast leading to
discontinuous layers, b) printhead is too slow leading to “buckled” layers.

2.5.4 Limitations of previously applied test methods to characterise

extrudability
So far the extrudability measurements published for printable concretes have been
qualitative only [2, 236]. None of the reported methods [232, 233, 235] is completely
adequate for the characterisation of the extrudability of printable concretes. Absent a
standardized methodology, testing material properties away from the process chain of DC

59
Literature and theoretical research 2.6. Buildability

(i.e., offline) using, e.g., ram extruders, is one possible approach. However, concrete in the
screw pump of a 3D-printhead/extruder is subject to different flow fields, shear histories,
and significantly different pressures in comparison to those in a ram extruder. The same
is also valid for the squeeze flow method. The penetration-resistance method is to date
reported only to be able to determine the upper and lower limits of extrudability. Most
importantly, none of the known studies has focused on the direct quantification of material
flow rate from an extruder similar to that used in target DC applications. This parameter
is crucial to the synchronisation of printhead velocity and concrete pump discharge
rate. Thus, new approaches are needed for reliable assessment of the extrudability of
cement-based materials in the context of DC. In the research at hand, the 3D-printer
extrusion test is proposed for characterising the extrudability of cement-based materials
for DC, quantitatively and inline; see Section 6.2.

2.6 Buildability
2.6.1 General introduction
Buildability is defined as the ability of an extruded cement-based material to retain its
geometry (shape and size) under sustained and increasing loads in fresh or transient state.
This definition relates closely to the statement by Le et al. [2] “the printed filaments
should be formed with minimal deformation under the weight of subsequent layers”.
Alternatively, Lim et al. [70] defined buildability as “the resistance of deposited wet
material to deformation under load”. Kazeimian et al. [68] defined shape stability as
“the ability to resist deformations during layer-wise concrete construction” and specified
“three main sources of deformation: self-weight, weight of following layers, and extrusion
pressure”.
Buildability is a complex, process-specific property which depends not only on material
composition but also on process parameters such as layer geometry. If buildability,
printing rate, and related aspects are not in harmony, the 3D-printed structure will
deform or collapse; see Figure 2.25. Buildability depends on, but is not identical to,
the structural build-up of cement-based materials, and this dependence is not exclusive.
A sufficient buildability can be achieved by setting the static yield stress of the 3PCs
high (either at the time of production or later with help of phase changing agents; see
Section 2.2.5). Alternatively, buildability can be ensured by having sufficiently longer
deposition time intervals (TI) between layers. However, longer TIs generally lead to
weaker interface bond strengths. It is noteworthy here that in practical cases a deposited
layer may not necessarily need to have zero deformation when subsequent layer is
deposited. Instead, a layer which deforms within the allowed tolerances is acceptable and
is expected to yield a better inter-layer bond. Overall, buildability defines the (maximum)
rate of printing and thereby the total construction time. Therefore, buildability of 3PCs
is one of the primary defining parameters of the economic viability of DC applications.
3D-printed elements should have consistent and continuous layer geometry over the
entire structure. Figure 2.25 shows an example of printed specimens with varying
numbers of layers and buildability. It is often appealing to print 10, 20 or as many
layers as possible and then “designate” the material as buildable. However, such trivial
approach neither links the material’s fresh properties to the target geometry nor considers
the economic viability of the target application. While it is still useful for relative
comparison of various compositions, printing an arbitrary number of layers with an
arbitrary time interval TI is not a reliable method in characterising buildability. For
example, flowable concrete with low static yield stress, extruded to wide, but thin layers

60
Literature and theoretical research 2.6. Buildability

using a long TI, will likely sustain deposition of a higher number of layers in comparison
to a less flowable concrete, but deposited with a much higher aspect ratio of individual
layers and a short TI.

a) b) c)
Figure 2.25: a) 3D-printed fine-grained concrete specimens (up to five printed layers are
depicted here, TI was 30 s), b) a collapsed 7-layer specimens with buildability deficiency, and
c) collapsed 16 layer specimen with 180 s TI.

2.6.2 Buildability investigations on 3PCs

Le et al. [2] stated that buildability relates “to the capacity to print a certain number
of layers or height”. They have quantified buildability as the “number of filament
layers which could be built up . . . without noticeable deformation of lower layers” [2].
Furthermore, the author suggested a shear strength (maximum torque measured with
a vane rheometer) range of 0.3 to 0.9 kP a, which leads to the successful printing of a
certain number of layers without deformation and extrudability issues. Di Carlo [237]
qualitatively tested shape stability by loading cylindrical specimens in the “fresh” state
using a container filled with sand. In their work Kazeimian et al. [68] proposed a cylinder
stability test for quick comparative evaluation of compositions’ influence on buildability.
They also proposed the layer settlement test in which “layers are printed on top of each
other with the same mechanism as full-scale printer” and suggested using this test for
deciding on “acceptance” or “rejection” of a mixture. In general, two layers and, for
selected compositions, five layers were printed, while the resulting deformations were
measured to evaluate buildability. Panda et al. [88] investigated the influence of the
deposition height or the nozzle stand-off distance7 on the interlayer bond strength. In
relation to buildability, they concluded that if the deposition height is reduced, the
dimensional accuracy of printing is negatively affected while the interface bond strength
is increased [88].
In their commendable work, Perrot et al. [7] considered the following primary
criterion: the flow resistance of a substrate-layer should always be higher than the vertical
loads acting on top of it. The researchers expressed the vertical loads in terms of printing
7 The vertical separation between the nozzle outlet and the top of printing platform or previous layer.
Lower disposition height corresponds to higher deposition pressure provided the flow rate of concrete is
unaltered.

61
Literature and theoretical research 2.6. Buildability

speed and hydrostatic pressure of concrete; see Equation (2.22).

σv = ρ · g · R · t (2.22)

The flow resistance of the substrate-layer was expressed as time-dependent static

yield stress and described by Roussel’s linear [156], see Equation (2.23), or Perrot’s
exponential evolution model for thixotropy [7].

σc (t) = τ0 (t) = αgeom (τ0,0 + Athix · tc ) (2.23)

Subsequently, Perrot et al. [7] proposed a critical failure time tf (see Equation (2.24)) to
predict at what time after deposition, concrete specimen fails if vertical load is increased
at a rate R and the initial (when resting time is zero) static yield stress τ0,0 is evolving
linearly (Roussel, 2006) with a constant slope Athix .
τ0,0
tf = ρ·g·R
(2.24)
αgeom − Athix

where αgeom is a parameter which depends on the geometry of the deposited

layer/printed element [7]. Similar to Perrot, Wangler et al. [101] determined the
minimum time th,min to print a layer of height h, following Von-Miese’s criteria as shown
in Equation (2.25). In other words, the minimum time th,min for producing a layer can
also be termed as the minimum interval between two successive layers needed to ensure
“buildability”.

ρ·g ·h
th,min = √ (2.25)
3 · Athix

Recently, Wolfs et al. [8] developed and validated a numerical model for predicting the
failure of 3D-printed concrete. Attributing early strength (0 to 90 min) of printed concrete
to “combined inter particle friction and cohesion”, they adapted Mohr-Coulomb failure
criteria, also considering time-dependency [8]. Most importantly, Wolfs et al. [8] showed
that the buildability failure of printed elements occurs due to the combined effects of
buckling and (plastic) yielding, usually with the former preceding the latter. The model
accounts for both buckling and yielding by taking a time-resolved elastic modulus E(t),
Poisson’s ratio v(t), cohesion C(t), angle of internal friction ϕ(t), and dilatancy angle ψ(t)
into consideration. Furthermore, the experimental validation of the numerical model
showed good qualitative and acceptable quantitative correlations. This approach, how-
ever, requires extensive experimental studies and, similar to rheology-based approaches,
high precision in execution is needed. In addition, the approach did not address the
economic viability of the target application. More recently, Weng et al. [238] studied the
influence of sand gradation on buildability following the Fuller-Thompson theory and
Marson-Percy model. They examined the buildability by direct printing tests and used
the number of layers as a measure of buildability. Further literature on buildability could
be referred to in [15, 17, 65, 116]. Note that an extensive review of buildability testing
approaches is out of scope for this thesis.

2.6.3 Challenges in practical implementation of proposed approaches

Buildability criteria based on fundamental rheological properties [91], e.g., static yield
stress, and the associated changes over time are still in development; it will take some

62
Literature and theoretical research 2.7. Layer-interface properties

years until development and validation are complete. Generic rheological models which
can consider various process techniques, the shape of the extrudate, and the effects of
temperature and other surrounding conditions will take even longer to be formulated
and proven. Furthermore, the parameter αgeom as utilised in [7] is not generically defined
yet. Perrot et al. [7] computed αgeom from the squeeze flow theory of plastics based on the
work of Roussel and Lanos [239]. However, αgeom used by [7] is only applicable for solid
cylindrical columns. Recently, Weng et al. [238] derived a new, complex equation for
hollow cylindrical columns. A geometric factor for rectangular cross-sections, let alone
for irregularly shaped cross-sections, has to date not been published. Determining the
structuration parameter Athix is also not a trivial task. Currently, there are neither stan-
dard devices nor standard protocols for characterising Athix of cementitious materials.
Even using most modern rheometers different Athix values may be derived for the same
material when different measuring protocols are employed. This implies that for the same
material, different critical failure time or minimum time intervals can be computed using
Equations (2.24) and (2.25).
Process-induced changes in rheological properties comprise another crucial subject.
See Section 2.4.5 for detailed discussion on how pumping process influences rheological
properties of cementitious materials. When it comes to 3D-printing based on extrusion on
a laboratory scale, 3PCs undergo high shear rates and are subjected to high pressure in the
extruder. Therefore, the exact rheological state of the extrudate may vary depending on
the specific extruder and printing-circuit (mixing-transporting-extruding). If large-scale,
on-site applications are realised, a pronounced influence of process/pumping on the
material rheological state is to be expected. Since rheological properties are usually
measured on material taken immediately after mixing, variations of these off-line
measured properties in comparison to the actual rheological properties of the extrudate
are to be expected. This, however, can be solved by carrying out extensive experimental
studies and by the fitting of theoretically predicted “buildability” and experimentally
observed “buildability”. To the best of the author’s knowledge, no such studies have been
reported yet.
As mentioned previously, the economic viability of a DC application is dependent on
the TI, which is directly linked to buildability. Quantitative approaches and systematic
yet comprehensible (for common practitioners) cost-calculation approach for DC ap-
plications are still missing. So far, apparently exaggerated claims have been made by
3D-concrete printing start-ups; see [87, 97]. Furthermore, numerous scientific articles
[15, 62, 98–102] have stated cost reductions with DC, nevertheless without presenting
a reproducible method of cost estimation. Also, the durability of the structures is a
function of the quality of interlayer joints, which depends on TI. Thus, it is crucial
to determine the buildability test parameters based on the intended application and
estimated process parameters. So far no research was published linking economic
viability to material buildability characterisation with clear specifications of “man-hour,
material and machine costs” and their sources. Hence, simple, practice-oriented, yet
rational buildability-assurance criteria are necessary to accelerate the implementation
of digital technologies in construction practice; see Chapter 7.

2.7 Layer-interface properties

Due to the inherent “layered” nature of DC; see Figure 2.7, the 3D-printed structure can
be non-monolithic and weak layer-layer interfaces may occur; see Figure 2.26. Commonly
referred to as “cold-joints” such interfaces may form the weakest links in the entire struc-
ture inducing pronounced anisotropy and therefore are of high research significance.

63
Literature and theoretical research 2.7. Layer-interface properties

The weak interfaces negatively affect not only the mechanical performance but also the
durability of 3D-printed elements [240]. The formwork-free nature of DC means that
the printed structures are more vulnerable to external conditions and more prone to
permanent damage in comparison to conventionally built structures. Therefore, in the
absence of specific bond-enhancing measures, the bond between different 3D-printed
layers is considered critical [2, 11, 62].

2.7.1 Formation of weak bonds between distinct concrete layers

Due to the numerous influencing parameters, description of the formation and me-
chanical properties of layer-to-layer interfaces in physio-chemical terms is difficult.
Earlier researchers [241, 242] and design codes such as Eurocode2 [243], CEB-FIP
Model Code 1990 [244] designated following parameters as decisive for interface bond
strengths: (a) roughness of the substrate, (b) cohesion and friction coefficients of the
interface, (c) moisture level and saturation level of substrate, (d) porosity of the substrate,
(e) ambient and substrate temperatures, (f) stress normal to the interface, (g) composition
of overlay concrete, especially the usage of bonding agent, (h) substrate age, and (i) if any,
degree and orientation of reinforcement connecting substrate and overlay.
In hardened concrete, weak-joints can form due to deteriorated or unhealthy [245]
surface of substrate, due to inadequate roughness of the substrate, lack of chemical
cohesive bond between overlay and substrate, due to surrounding conditions, etc. In case
of both “neighbouring” layers being in a plastic state or one being in a plastic state and
another quasi-plastic, the common case in extrusion-based 3D-printing, the rheological
properties [152] of both substrate and overlay significantly influence the interface
properties. The static yield stress of cement-based materials at rest (upon casting or after
layer depositions) increases linearly up to a characteristic fresh concrete age, usually 45 to
65 min, and increases further exponentially [27, 246]. As a result, if two distinct layers of
concretes are deposited with a time interval (TI) then an interface may form between the
distinct layers. The mechanical performance of layer-to-layer interfaces, especially under
tension and shear, depends on the rheological and morphological properties as well as
the physical and chemical states of both the substrate and the overlay.
Various techniques were applied in conventional distinct layer castings, e.g., in renova-
tion work to enhance concrete-to-concrete bonding. The approaches of sand blasting and
water jetting, in general, are regarded as the most effective surface preparation techniques
in enhancing bond strength [247–252]. Many of the surface treatment approaches rely on
“destructive” approaches for removing unhealthy surface, e.g., pneumatic hammers or
needle-guns, and suffer from the risk of substrate damage due to micro cracking [248].
Other techniques include pre-wetting the substrate and treating it with chemicals. In
fresh-in-fresh casting within an appropriate time interval between the layers, vibration
is used to increase the flowability of concrete temporarily and to enable physical
intermixing of concrete layers. External vibration on one hand assists (active) intermixing
of concrete layers and on the other hand increases plasticity of concrete, resulting in
better overlay enveloping (passive) on the substrate surface.

2.7.2 Reported research on anisotropic properties of 3PCs

Owing to its significance, the research interest on layer-to-layer bonding of 3D-printed
concrete elements is growing [11, 69, 71, 88, 116, 253]. Le et al. [71] reported that average
tensile bond strength was 2.3 MP a for specimens produced within time intervals (TI)
of 0 min and 15 min, but only 0.7 MP a for specimens produced with TI of seven days.

64
Literature and theoretical research 2.7. Layer-interface properties

Zareiyan et al. [69] studied the influence of aggregate size and process parameters on
interlayer adhesion. Relying on prism-splitting tests, they concluded that “bond strength
for 1 inch layer fabrication with 6 min TI between layers decreased by 19 % to 36 % in
comparison with monolithic specimens.” The interlayer bond of 3D-printed elements
was also reported for geopolymers [253]. Panda et al. [88] studied the influence of process
parameters on bond tensile strength and concluded that decreasing the depositing height
(referred to in [88] as nozzle standoff distance) enhanced bond strength significantly.
Sanjayan et al. [116] identified the moisture content of the inter-layer region as one of the
“major factors affecting the inter-layer strength”; higher moisture level on the substrate
surface led to greater bond strength despite the longer time intervals [116]. However,
there are conflicting recommendations by various published standards and technical
committee recommendations on the influence of moisture/water on interface-bond
strength [249, 252].

Figure 2.26: Signs of pronounced local water-intake (highlighted) at the layer interfaces,
presumably due to high porosity of the interlayer, photo by V. Mechtcherine.
Not all established bond-enhancement approaches are applicable for DC. External
vibration of the deposited layers is risky since it would disrupt their structural build-up
and thus compromise buildability. Applying such methods as sand-blasting is not trivial,
but even if technically possible they would unlikely be economical, taking the high
number of layers into account. Water treatment seems to be very practicable; however,
it must be controlled very precisely in order not to influence buildability and other
fresh-state properties. At present, there is ongoing research on measures to improve the
bond strength of printed layers, e.g., by applying a low-viscous mineral-based primer
before deposition of each subsequent layer [254]. Such approaches are not detailed here
since they are out of the scope of the work at hand.
Despite increasing research interest, there is no standard approach to test the interface
strength of 3D-printed concrete layers yet. So far, various studies have followed different
methods. For example, Le et al. [71] printed elements with multiple, side-by-side printed
layers of circular cross-section, i.e., with designed voids between the layers. The specimen
production setup used by Sanjayan et al. [116] consisted of a simple piston-cylinder
extruder (similar to a commercial silicone gun), which was manually filled and operated.
Zareiyan et al. [69] produced specimens layer-by-layer by casting in a mould, a process
which mimicked layer deposition of 3D-printing, but not completely. Each of the above
mentioned approaches has its own advantages and limitations. The findings of one study
may not be transferable to others due to the differences in printing processes and test

65
Literature and theoretical research 2.7. Layer-interface properties

procedures. This makes it necessary at this stage to characterise every 3PC separately to
ensure reliable DC systematically. The intention of the author is not to discredit any of
the approaches followed, but rather to emphasize that the observed influences will differ
greatly depending on the research approach chosen. The general variables are production
approach, shear history due to the particular extruder geometry, type of test for interface
strength, specimen size, loading direction, etc. The relatively simple approaches [69,116]
can be useful in selecting optimum mixtures from extensive mixture groups if multiple
parameters are under investigation.
To the best of author’s knowledge, no dedicated study on comparatively studying
influence of binder composition on mechanical properties of 3D-printed concretes is
reported. Usage of secondary cementitious materials (SCM) is proven to be beneficial
for DC applications, thanks to the enhancement of material thixotropy among others
[2, 12, 27, 62]. At the same time, due to lack of large aggregates and the demand for rapid
initial strength development, mixtures with higher dosage of cement are also commonly
used in DC. Higher cement content can lead to increased risk of shrinkage cracks,
thermal stresses and negative environmental footprint [128]. A one-to-one comparison
of mixtures with sole and high cement content on the one hand and low cement dosage
with SCM on the other hand is of high research significance. This subject is addressed in
the research at hand; see Chapter 8 for more details.

66
Literature and theoretical research 2.8. Chapter summary

2.8 Chapter summary

With the plethora of opportunities (Section 2.2.3), DC has the potential to cause a
paradigm shift in concrete construction. At the same time, due to the complex material,
process and machine requirements and challenges (Section 2.2.8), the promises of
DC are far from complete realisation. The conflicting rheological requirements for
pumpability, extrudability and buildability as well as anisotropic hardened state prop-
erties need extensive experimental investigations. New methods for systematic material
development and characterisation are necessary to address all the described issues. The
developed methods must be based on high scientific advancements and also needs to
be practice-oriented, preferably both aspects combined in one method. The author sees
the largest potential and need for DC in developing countries with limited availability
of skilled labour and expensive, advanced scientific equipment such as high accuracy
rheometers. It is evident that the cost structure in case of DC is going to differ from
conventional construction, with the former being more cost-effective than the latter. At
the same time, not all DC processes share similar costs and economic potential. The DC
processes with high spatial resolution and complexity, which rely on unconventional
equipment need high initial investments. Following further conclusions can be drawn
based on the state-of-the-art:
a) Applicability of the current rheometrical protocols, for measuring the structural
build-up of 3PCs is uncertain and needs critical assessment considering the high
yield stress and structuration rate of 3PCs.
b) So far pumpability is not specifically investigated in the context of DC. 3PCs’
high thixotropy and non-flowable consistency require extensive research on their
pumpability to enable large-scale implementation of DC.
c) Considering the proven applicability of Sliper for pumping pressure prediction,
numerical modelling of Sliper could enable a more purposeful material design and
play a vital role in autonomous concrete construction.
d) Although there are earlier works on numerical simulation of concrete pumping
flow; see Section 2.4.4, so far no numerical model is available to link the resulting
rheological properties as observed in the Sliper tests to any other more commonly
used experimental procedure such as by means of a rheometer or viscometer.
e) So far, the extrudability investigations performed on 3PCs are qualitative only.
Applicability of ram extruders and simple flow tests to characterise extrudability
of 3PCs is an open question. Reliable, quantitative and inline extrudability testing
approaches that also quantify the volumetric flow rate of extrudate are of high
research significance.
f) Buildability of 3PCs is one of their most crucial properties; it depends on yield
stress, structural build-up, layer and structure geometry (buckling) and it also
directly influences the mechanical properties of 3PCs and economic viability of DC
application. A practice-oriented buildability testing approach is still missing.
g) The high potential for weak layer-interfaces makes the durability of printable
structures uncertain, design and structural analysis perplexed. The question of how
binder composition influences interface bond needs to be answered, preferably with
micro- and macroscopic investigations.
h) Before addressing imminent but more complicated challenges of reinforcement
integration and developing standards/guidelines, reliable experimental methods
for the primary issues mentioned above needs to be developed and thoroughly
tested.

67
Chapter 3

MATERIALS, TESTING EQUIPMENT

AND GENERAL METHODS

The generally applied experimental methods, devices and materials are presented in this
chapter. Details of specific methods concerning particular topics are elaborated “locally”
in the corresponding chapters.

3.1 Materials and mixture compositions

A Portland cement CEM I 52.5 R with trade name CEM I 52.5 R ft. (OPTERRA Zement
GmbH, Werk Karsdorf, Germany) according to EN 197-1 [255] was chosen as the
main binder. Such fast-reactive cement was expected to be instrumental in achieving
high early-age strengths necessary for formwork-free digital construction. Two aqueous
high-range, water-reducing admixtures (HRWRA) on polycarboxylate-basis (PCE) were
used as superplasticisers (denoted PCE1 and PCE2). Both PCEs were developed for
ready-mix concrete applications and thus ensured extended consistency retention, i.e.,
> 90 min after water addition. Tap water at sub-room temperature was used for all the
mixtures under investigation. Furthermore, the effect of an alkali-free setting accelerator,
denoted in this thesis as ACC, was also considered. For particular compositions, supple-
mentary cementitious materials (SCMs) were used to reduce total cement content and,
subsequently, the CO2 footprint. At the same time, the SCMs employed were expected
to act as rheology modifiers, increase the structural build-up and reduce “shrinkage”.
These SCMs were a) Class F Fly ash (denoted FA) according to ASTM C618-12a [256]
and b) microsilica suspension (denoted MSS) with a specific surface of 15-30 m2 /g and
a mean particle size of 0.15 µm; 80 wt% of primary particles had diameters less than 5
µm. Trade name of FA is Steament H-4 (STEAG Power minerals GmbH, Germany). Trade
name of MSS is EMSAC 500 SE (Elkem AS, Norway1 ). The utilized MSS contains 50 wt%
microsilica and 50 wt% of water. Furthermore, very fine quartz sand (fraction 1: 0.06 to
0.2 mm) and two fine natural river sands (fraction 2: 0 to 1 mm and fraction 3: 0 to 2 mm)
were used as “aggregates”. Their particle size distributions are plotted in Figure 3.1.
Table 3.1 presents the investigated 3PCs. The differences between the mixtures A and
C2 are the producer and dosage of superplasticizer (PCE). The binder-suspensions, on
which structural build-up measurements were performed, are presented in Chapter 4
and Table 4.2.
1 In Germany supplied by BASF as EMSAC 500 SE

69
Materials and general test methods 3.2. 3D-concrete-printing test device

0.06-0.2mm 0-1mm 0-2 mm

Percentage passing [%]

100
80
60
40
20
0
0.0625 0.125 0.25 0.5 1 2
Sieve size [mm]
Figure 3.1: Sieving curves for the used quartz sands.

3.2 3D-concrete-printing test device

As part of the associated research project CONPrint3D® , the 3D-concrete-printing test
device (3DPTD), prototype 1, was developed at the TU Dresden [12, 124]; see Figure 3.2.
The structure of this test device corresponds to a multi-axis portal system with two
motorized axes and one manually adjustable axis. The printhead can be moved within
a working area of approximately 1.5 m x 3.0 m by means of a dual-axis system
(horizontal) driven by electric motors. The manually operated axis (vertical) is used to
adjust the height of the printing platform, e.g., before printing a new layer. The 3DPTD
control system and software are designed so that, freely programmable trajectories can
be traversed at printhead velocities up to 150 mm/s. The 3DPTD can be controlled
semi-manually with the help of a digital joy-stick and autonomously using G-code or
LUA-script.
The printhead is a modified mortar pump with a frequency-controlled motor, which
is operated digitally via custom-developed control software. A progressive cavity pump
(PCP) mounted on the printhead is employed to extrude concrete from the hopper. The
PCP used in the 3DPTD can rotate at 15-150 rpm, delivering up to 4 l/min concrete
(varies with rheology [26]). The pumping distance is maximum 15 m, with operating
pressure of 20 bar. The concrete container has a capacity of 50 l. Fine-grained concretes
with maximum aggregate diameter of 3 mm can be pumped/extruded. The entire
printhead, with concrete filled in, weighs about 150 kg and is traversed in 3 dimensions.
Both the printhead motion and concrete pump (i.e., PCP on the printhead) discharge
rate are coupled, and the synchronization is steered with help of two parameters viz.
proportionality and correction/offset; see Section 3.2.1 [26]. The nozzle is made of
polyoxymethylene; the primary one has dimensions of 30 mm x 18.72 mm.

3.2.1 Synchronization of printhead motion with concrete flowrate

A 3D-printer should be able to extrude concrete both while the printhead is moving and
−−−→
at rest. To facilitate concrete extrusion at zero printhead velocity Vph [m/s], a rotation
control parameter is necessary, named RCP 1 [−]. While in motion, the rotational velocity
Np [rps] of the motor of the material pump must be synchronized with the printhead
−−−→
velocity Vph :
−−−→
∴ Np ∝ Vph (3.1)

70
Materials and general test methods 3.2. 3D-concrete-printing test device

Table 3.1: Mixture compositions.

Chapter 6 Chapter 6 Chapter 8 Chapter 7, 8
Constituents Density A A-HF C1 C2
[kg/m3 ] Weight per unit volume [kg/m3 ]
CEM I 52.5 R 3100 391 391 627 391
Fly ash 2271 213 213 0 213
MSS 1400 213 213 0 213
Sand 0.06-0.2 2650 252 1260 276 252
Sand 0-1 2650 252 0 276 252
Sand 0-2 2650 756 0 835 756
Tap water 1000 138 138 265 138
Superplasticizer 1010 14 18 8 11
PCE1/PCE2 (PCE1) (PCE1) (PCE1) (PCE2)

A simple proportionality constant RCP 2 , ranging from 0 to 1, alone cannot be employed

due to dimensional imbalance in Equation (3.1). Moreover, there exist three different
velocities of the printhead during 3D-printing:
−−−−→
• Maximum velocity with which the printhead can be moved — Vmax ;
• Applied velocity, i.e., the velocity with which the printhead is commanded to run
−−−→
— Vap ;
• Effective velocity with which the printhead is actually traversing at a given point —
−−−→
Vef .
The applied velocity is the velocity “entered” by the 3D-printer operator. Whereas, the
effective velocity is the actual resulted velocity, subject to acceleration and deceleration of
the motors at the starting and stopping of the printhead motion. Therefore the effective
−−−→ −−−→ −−−→
velocity can differ from the applied velocity; further, ∀ Vap , 0, Vef ≤ Vap . Therefore,
−−−→ −−−→
the velocity of the concrete pump NP is controlled as a function of Vef , Vap and the
proportionality constant RCP 2. Thus resulted synchronization term is:
−−−→
Vef
−−−→ · RCP 2 (3.2)
Vap

The rotational velocity of concrete pump in the 3DPTD is frequency controlled, by

supplying a voltage U [V ] ranging from 0 [V ] to a maximum of Umax = 10 [V ] to the
pump’s frequency converter. U and Umax are related to the rotational velocity of concrete
pump Np as:

U
Np = · Nmax (3.3)
Umax

where Nmax is the maximum rotational velocity of the concrete pump.

71
Materials and general test methods 3.3. Ram extrusion test device

−−−→
If the printhead is in motion, i.e., Vap > 0, the constitutive equation for U is:
  −−−→ 
  Vef 
U = 1 + 2 · 10 (3.4)
 
RCP · RCP
  
 
 −
−−→ 

  V ap


−−−→ −−−→
where, Vef and Vap are the effective and applied vector velocities of the printhead in
[m · s−1 ].
If applied velocity is zero, i.e., if the printhead is stationary, then instead of Equa-
tion (3.4) the following constitutive equation is considered:

U = RCP 1 · 10 (3.5)

In Equation (3.4), RCP1 (rotational control parameter when the printhead is not
moving) and RCP 2 (rotational control parameter when the printhead is moving) are both
dimensionless, ranging from 0 to 1 and are the user end control parameters (to be entered
in the 3D-concrete-printing test device (3DPTD) control software). In general, the higher
the RCP 1 and RCP 2, the higher is the assigned rotational speed. Equations (3.3) and (3.4)
make it possible to:
• Control concrete flowrate in proportionality with printhead velocity by means of
−−−→
Vef
the term −−−→ and proportionality constant RCP 2.
Vap
• Extrude concrete even if printhead velocity is zero, i.e., in the case of a stationary
extrusion via ARP RCP 1.

1
2
5 3

a) b)
Figure 3.2: a) 3D-concrete printing test device (courtesy of M. Näther), b) printhead. 1
concrete hopper, 2 extruder motor, 3 progressive cavity pump, 4 printhead motors, 5
nozzle.

3.3 Ram extrusion test device

A simple custom-built ram extruder was used, in which a metal piston was driven by
a linear actuator; see Figure 3.3. Fresh concrete was filled into a cylinder of diameter
120 mm and height 300 mm and then extruded at a constant defined displacement
rate through a circular opening with 40 mm diameter. During extrusion, the cylinder

72
Materials and general test methods 3.4. Approaches for fresh-state testing

is restrained fully with help of a tensioned belt; see Figure 6 in Ogura et al. [81]. The
ram extrusion force REF at the coupling of the piston and linear actuator was measured
using a load cell with maximum load capacity of 5 kN. The effective ram displacement ∆r
was measured using a displacement measuring system of the linear actuator. Extrusion
velocity could be determined using ∆r and elapsed extrusion time. Both REF and ∆r were
plotted to illustrate visually the extrudability in terms of REF.
1 2 3 4

Figure 3.3: Illustration of principal components of ram extruder for measuring extrusion force
(base sketch: M. Näther). 1 Linear actuator, 2 load cell, 3 piston and 4 barrel.
The measured REF is influenced by friction between the billet, the piston-head and
the inner walls of the cylinder. The frictional force varies depending on the test setup
and must be considered while characterizing extrudability. Specifics of extrusion force
measurements and data analysis are elaborated in Section 6.4.2; see also Figure 6.7.

3.4 General approaches used for fresh-state characteriza-

tion
Haake-Mars rheometer: A coaxial rheometer HAAKE MARS II with a unit cell and vane
rotor of the dimensions presented in Figure 3.4 was employed to investigate structural
build-up and to determine the Bingham rheological parameters. To prevent slippage, the
inner surface of the unit cell was instrumented with steel lamellas.

a) b)
Figure 3.4: a) HAAKE MARS II rheometer and b) schematic of unit cell and vane rotor.
The particular vane-and-cell geometry used in this research is a relative measurement
system as opposed to absolute geometric systems such as cone-plate [257]. To transform

73
Materials and general test methods 3.4. Approaches for fresh-state testing

the geometry dependent measured values into absolute (geometry independent)

rheological entities, a calibration procedure was first carried out. Specifically, the
geometric factor M was determined by testing reference substances (silicon oil and
Carbopol®) with known viscosities in a calibration procedure using standard cone-plate
geometry, previously described in [257, 258].

ConTec Viscometer: The Couette type ConTec 5 Viscometer was employed to investigate
rheological properties of suspension and fine-grained concretes; see Figure 3.5. The
Viscometer has an outer and an inner cylinder mounted coaxially. During measurement,
the outer cylinder is rotated at specified velocities while the inner cylinder is always
stationary. Torque T [N · m] and rotational velocity N [rot/s] were measured and recorded
numerous times each second depending upon the sampling rate assigned. The inner ri
and outer ro radii of the setup were 0.064 m and 0.080 m, respectively. Both cylinders
were equipped with “rib” structures to prevent slippage. Further details regarding the
geometrical conception of the Viscometer can be found in [191, 259, 260]. Measured
torque T [N · m] was plotted against applied rotational velocity N [rot/s]. The parameters
describing the Bingham yield stress τ0 and plastic viscosity µ were computed using linear
regression to measured T −N by means of the Reiner-Riwlin equation [179,259], as shown
in Equation (3.6):

 
1 1
−
R2i R20
τ0 =  G (3.6)
4.π.h.ln RR0i
 
1 1
− 2
R2 R
µ= i 2 0 H (3.7)
8.π .h
where G [N · m] is the torque required to initiate the flow of the mixture (yield stress
parameter) and H [N ·m·s] represents the resistance to deformation (viscosity parameter).
h is the height of material body as measured on the surface of the inner cylinder. In this
work, h was kept constant at 112 mm. The measuring protocols of rheometry, i.e., shear
rate and test durations are described in Sections 4.1.2 and 6.3.3; see also Figures 4.1
and 6.2.

Hägermann flow table tests: A Hägermann flow table (HFT) was used to measure
the spread diameter as an empirical indicator of the 3PCs rheological “behaviour”; see
Figure 3.5c. Thereby, the flowability was characterized according to EN 1015-3:2007-05
[261]. Initially, a conical frustum (dimensions 70 mm top diameter, 100 mm base diameter
(∅0 ) and 60 mm height) is filled with a 3PC accompanied by slight tapping according
to the standard. Then the cone is lifted, allowing the 3PC to spread on the plate
while the plate remains steady. After the spread diameter is measured, the table is
lifted-and-dropped 15 times in 15 s according to [261]. To improve the measurement
accuracy, spread diameter values were measured in four orientations, with a sequential
angular separation of 90 °. Measurements were taken both before strokes (BS) and
after the strokes (AS). Earlier researchers suggested that the existing flow tests on
cement-based materials correlate with their yield values, which can be associated with
their buildability and printability [68, 262]. Indeed, the HFT could be used as a quick
(non-exclusive and non-determinant) indicator of buildability lower limit, i.e., the
minimum yield stress needed to ensure a certain number of layers to be printed without
collapse. This is enabled by HFT sufficiently representing the material behaviour under

74
Materials and general test methods 3.4. Approaches for fresh-state testing

a) b) c)
Figure 3.5: ConTec Viscometer 5, b) static inner (top) and rotating outer (bottom) assemblies,
c) HFT setup (adapted from [261]. 1 conical frustum and 2 movable platform.

self-weight as well as under external force (strokes). From HFT results, the relative slump
before strokes ∅r,bs = ∅∅bs0 and after strokes ∅r,as = ∅
∅as
bs
parameters can be calculated. A low
∅r,bs and high ∅r,as can be preferable for 3PCs as the former indicates better self-stability
(inference to buildability) and latter indicates better flowability under external force
(inference to pumpability and extrudability).
It must be emphasized that the HFT tests cannot be used as the primary char-
acterization technique for printable concretes considering that HFT results does not
indicate 3PCs’ rheological behaviour at high shear rates and/or under high pressures. In
particular, HFT spread values cannot definitively indicate extrudability of a composition.
Non-optimal mixtures with relatively high ∅as could still be deficiently extrudable [81]
or non-extrudable. That being said, a parametric study by author on vast-range of
non-fibre-reinforced 3PCs indicated that mixtures with spread values above 140 mm after
strokes are not buildable2 ; see Figure 3.6 and [124].

2 Forexample, 17 layers could be printed vertically with a time interval below 3 min, as detailed in
Chapter 7.

75
Materials and general test methods 3.4. Approaches for fresh-state testing

230
M1_50 BS
M1_50_AS
210 M1 BS
M1 AS
M2_50 BS
HFT spread diameter [mm]

190
M2_50 AS
M2 BS
170 M2 AS
M3 BS
M3 AS
150
1 M4_LSP BS
M4_LSP AS
130 M4 BS
M4 AS
M5 BS
110
M5 AS
M6 BS
90 M6 AS
0 15 30 45 60 75 90
Age [min]
Figure 3.6: HFT results for various compositions as observed in a parametric study on
printability. 1 140 mm critical line: ∅as above this line indicate non-buildable compositions;
all compositions were extrudable even those with ∅bs = ∅0 ; see [124] for details of mixture
compositions.

76
Chapter 4

RHEOLOGY AND STRUCTURAL

BUILD-UP

This chapter focuses on rheological properties of cementitious materials in the context

of DC. Specifically, the structural build-up (SBP) at rest is addressed. Static yield stress
and its development over time is crucial for buildability [7] as well as pumpability [13]
of 3PCs; see Section 5.1. Further, structural build-up may influence the extrudability of
3PCs provided long process interruptions; however, this aspect was not addressed so far
in literature. To date, the research on quantifying structural build-up has been limited
to non-printable concretes. Since self-consolidating concrete and related cement pastes
have very low yield stress, it is possible to shear the material for a very short test duration
with shear rates as low as 0.01 s−1 [165, 169, 171, 172]. The applicability of low, constant
shear rate tests for printable concretes is yet to be investigated. Considering that DC
cementitious materials are generally rather stiff and accelerating agents are often used,
it is of great significance to re-examine the validity of low, constant shear rate test in the
context of DC.
As a first step toward characterizing structural build-up of printable concrete mixtures,
the structural build-up of cementitious suspensions of varying compositions and rheo-
logical properties is investigated. In that process, the challenges while applying constant
shear rate test for stiff materials are delineated. Furthermore, an improved approach
for characterizing structural build-up is introduced in this chapter; in which strain
rather than shear rate is the decisive parameter. By applying the proposed approach,
the influence of cement pastes’ composition on structural build-up is investigated.

4.1 Experimental program

4.1.1 Mixture compositions and preparation
Cement pastes of various compositions were investigated, while the reference composi-
tion (Ref) was designed in reliance on the high-strength, printable, fine-grained concrete;
see Mixture C2 in Table 3.1. The material constituents are detailed in Section 3.1.
All cement pastes had an equivalent water-to-cement ratio (w/c)eq of 0.42, calculated
according to DIN EN 206-1/DIN 1045-2 with a reactivity parameter K of 1 for microsilica
and 0.4 for fly ash. The reference cement paste contained 0.75 % superplasticizer (PCE2)
by weight of binder (%bwob ). The composition of pastes was varied with the following
objectives:
• to study the validity of proposed test methodology (see Sections 4.1.3, 4.2.1
and 4.2.2) for cement pastes of varying stiffness, and

77
Rheology and structural build-up 4.1. Experimental program

• to investigate the influence of cement paste composition on structural build-up

(Section 4.2.3).
The dosage of superplasticizer was varied with 0.75 %bwob (Ref), 0.50 %bwob (paste
SP-0.50), 0.30 %bwob (SP-0.30) and 0.00 %bwob (SP-0.00). Furthermore, a set-accelerator
was added to Ref at dosages of 0.25 %bwob (paste ACC-0.25), 1.50 %bwob (ACC-1.50)
and 2.50 %bwob (ACC-2.50). Finally, the SCM paste was composed by replacing in Ref
mixture 30 % of cement (by volume) with fly ash (FA) and another 15 % by volume with
microsilica added as MSS, 30 % by volume of cement. For each experiment three litres of
cement paste were prepared, following the preparation methodology given in Table 4.1,
meaning the same applied shear history and good dispersion. The mixed paste was then
filled in the unit cell in three steps, making sure that no/minimal amount of air was
entrapped.
Table 4.1: Cement pastes preparation protocol.
Time Duration Speed Action
[min : s] [min : s] [rpm]
-01:00 to 00:00 01:00 Adding SP to water
00:00 to 02:00 02:00 - Adding cement to water → hand mixing →
adding SCMs, if present
02:00 to 03:00 01:00 139 Mixing
03:00 to 04:00 01:00 - Wall scraping
04:00 to 06:00 02:00 591 Mixing (adding admixtures at 05:00)
06:00 onwards Visual control, if necessary, followed by
scraping and further mixing at 591; up to
08:00 min.

4.1.2 Rheometer setup and testing protocol

If a single rheometer is available, there are three ways to characterize structural build-up
of printable cementitious material using stress growth tests, by:
a) Single-batch: testing a single batch using a rheometer having multiple unit cells,
b) Multi-batch: testing multiple batches using a single rheometer with a single unit
cell, and
c) Single-specimen: testing a single batch using a single rheometer with a single unit
cell.
The first approach, apparently, leads to minimal error in the material characterization.
However, this approach is not feasible if multiple unit cells or multiple identical rheome-
ters are not available. The second approach has as its prerequisite that each batch of tested
cementitious material be produced identically. This is a challenging task considering
the complex composition, preparation, and highly transient rheological behaviour of
cementitious materials and their sensitivity to surrounding conditions. Mahaut et al.
[113] reported an uncertainty of 25 % in yield stress measurements due to the difficult
reproducibility of batches, which in turn would distinctly affect the calculated Athix
values. Moreover, if extensive parameter variation is to be studied using a multi-batch
approach, then the total number of experiments would increase substantially.
In the research at hand, only one rheometer and one unit cell were available, while a
vast variety of parameters had to be investigated to meet the rheological challenges of

78
Rheology and structural build-up 4.1. Experimental program

material design for the targeted DC [12, 101]. Hence, the single-specimen approach was
the only meaningful choice. The author believes that this approach is highly relevant
to the research in the field even though it has not been used extensively yet. However,
without thorough investigation and specific measures to remove sources of error, the
single-specimen approach leads to erroneous characterization of the structuration rate.
In this research, the HAAKE MARS II (see Section 3.4) was employed to investigate
structural build-up and to determine the Bingham rheological parameters. The Häger-
mann flow table (HFT; see Section 3.4) was used to measure the spread diameter as an
empirical indicator of the paste’s rheological behaviour. After preparation, the cement
paste was filled into the unit cell and placed in the rheometer. A hysteresis loop was
performed at 18 min after the addition of water. For this purpose, a shear rate ramp-up
from 0 to 100 s−1 was applied for 30 s, followed by constant shearing at a rate of 100
s−1 for 60 s and a shear rate ramp-down from 100 s−1 to 0 s−1 in the next 30 s; see
Figure 4.1. This procedure is to ensure that the flocculation in the paste attained up to
that point in time is completely broken and the paste has possibly reached steady state,
i.e., the deflocculated state according to Roussel [156]. Tested specimens were kept at rest
following the hysteresis loop until the subsequent constant shear rate tests, which were
addressed hereafter also as constant strain tests considering the fact the applied shear
rates were not constant.
Table 4.2: Compositions of the pastes under investigation, all dosages are for 1 kg/dm3 .
Consti- ρ Ref SP- SP- SP- ACC- ACC- ACC- SCM
tuent [kg/dm3 ] 0.50 0.30 0.00 0.25 1.50 2.50
CEM I 3.10 1.333 1.338 1.341 1.346 1.329 1.307 1.291 0.753
Water 1.00 0.560 0.562 0.563 0.565 0.558 0.549 0.542 0.266
SP 1.01 0.010 0.007 0.004 0.010 0.010 0.010 0.017
ACC 1.37 0.0033 0.020 0.032
FA 2.27 0.411
MSS 1.40 0.411
120 160

100
120
80

60 80

40
40
20

0 0
0 40 80 120 0 20 40 60 80 100

a) b)
Figure 4.1: a) Shearing regime in rheometer tests for hysteresis loop, b) example hysteresis
loop output.
Following the hysteresis loop test, constant strain tests were carried out on the same
sample at paste ages tage of 23, 45, 60, 90, 120 and 150 min; see Section 4.2.2 and Table 4.3.

79
Rheology and structural build-up 4.1. Experimental program

This was made possible by the purposeful choice of γap and td , ensuring that the angular
displacement of rotor and corresponding structural change in the sample were minimal.
For example, combinations of γap = 0.08 s−1 applied for td = 30 s leads to a strain of 2.4
units and angular displacement < 40◦ . Applied shear rates γ̇ap and test time durations td
are chosen such that the resulting applied shear strain γap (deformation) is equal to 2.4
units. Exact values of applied test parameters are presented in Sections 4.2 and 4.2.3.
Table 4.3: Methodology of studying the structural build-up of cement pastes.
Relative time Absolute time Shear rate γ̇ap Action
[min : s] [min : s] [s−1 ]
18:00 -18:30 00:30 0-100 Shear up
18:30 - 19:30 01:00 100 Constant shearing
19:30 - 20:0 00:30 100 - 0 Shear down
23:00, 45:00, 60:00, 00:10 < td <00:30 0.08 < γ̇ap < 0.24 Stress growth test
90:00, 120:00, 150:00

4.1.3 Applied shear rates and applied shear strain

When structural build-up studies are carried out in the context of DC, it is important
to address the shear rates of material at the time soon after the extrusion, i.e., when the
shear rate is zero. Herewith, the flow-onset of the extruded material at rest after printing
and subject to vertical loads of the upper layers must be characterized. Therefore, the
lowest possible shear rate must be chosen to carry out CSR tests in the context of DC.
In most of the reported works on structural build-up of cementitious materials, shear
rates are as low as 0.001 s−1 ; see also Table 2.2. The feasibility of applying such low
shear rates and achieving reliable data depends, however, on the specific instrument used
and the specific material under investigation. For stiff compositions such as those in DC,
applying very low shear rates is more challenging than for flowable mixtures such as
SCC. This was demonstrated by CSR tests performed on non-thixotropic fluids, namely
Carbopol® solutions in water at 0.1 M, 0.2 M and 0.3 M concentrations, which were
studied in preliminary investigations; see Figure 4.2. It could be observed that at very
low shear rates, the instrument needs some time to reach the prescribed shear rate. As a
result, a new parameter, te can be defined as the time elapsed for effective shear rate γ̇ef to
reach applied shear rate γ̇ap . te is proportional to the stiffness of the tested materials and
is inversely proportional to the applied shear rate; see Figure 4.2. For the cement pastes
under investigation, the lowest reliable shear rates were ascertained using a series of
preliminary experiments on each investigated cement paste. The actual values for various
cement pastes are described in Section 4.2. The applied strain γap was calculated as
product of applied shear rate γ̇ap and test duration td : γap = γap ·td . The applied shear rate
is a theoretically constant shear rate assigned for a particular test protocol. The applied
shear rate is fully realized in experiment only if there is no significant delay by rheometric
device in following the prescribed shear regime. The effective strain γef and the
effective shear rate γef were calculated by multiplying experimentally realized angular
displacement ϕ and angular velocity Ω, respectively, with a geometrical shear factor
M. The relative measurements of rheometer are transformed to geometry independent
rheological entities in a procedure described in [257, 258]; see also Section 3.4.
Since the measured rotational displacement and torsional moments were transformed
not by means of a geometric equation but using a conversion parameter, the plug

80
Rheology and structural build-up 4.1. Experimental program

thickness (if any) cannot be directly taken into account. It is noteworthy that for “very
stiff”, printable concretes formation of plug flow is very likely. However, only pastes were
investigated in this study. In fact, the pronouncedly flowable nature of reference pastes
can be seen from the Hägermann flow table spread value of 300 mm before strokes (small
cone slump flow). The formation of plug depends on the material properties, geometry,
effective shear rates and test duration. Considering the large number of mixtures, tested
ages and shear rates under investigation, extensive research on possible (but unlikely)
occurrence of plug flow was not carried out in the thesis at hand.

4.1.4 Calculating rheological parameters

By plotting shear stress as a function of strain, as shown in Figure 4.4a (Section 4.2.1),
it can be observed that the curves increase nearly linearly to a certain limit and then
either transfer into a plateau of constant shear stress or show a slight peak followed by a
plateau. In general, the peak shear stress is considered as the static yield stress. However,
in this research, taking into account the fact that peak stress is not always a distinct
point, static yield stress was calculated as the average of the shear stress values at the
transition where abrupt change in the development of the shear stress–shear strain curve,
into the plateau [165], occurred. The static yield stress values measured at various ages
of the cement pastes under investigation were plotted as a function of tage in Figure 4.11
yielding the information on structural build-up.
Structuration rate parameters Athix,l and Athix,e were calculated by fitting the measured
static yield stress values for various rest times trest to the thixotropy models presented in
Equation (2.1) [156] and Equation (2.2) [7], respectively. Here, it is important to note that
Roussel’s linear model was fitted to the data corresponding only up to the characteristic
time, paste age, tc [7], while Perrot’s exponential model was fitted for the total test
duration. Until tage reaches tc , the growth of hydration products between neighbouring
cement grains results in the linear development of total solid structure in the cement
paste, short-term flocculation [156]. Therefore, the tc is computed as value fitting best the
experimental data in the region where static yield stress evolution changes to exponential
regime. After tage surpasses tc , exponential development of static yield stress takes place
due to the solid volume fraction increase [7], which results from the acceleration of
hydration.
The values describing Bingham yield stress and plastic viscosity are deduced from the
hysteresis loop tests using the downward curve to calculate the respective parameters.

0.1 M 0.2 M 0.3 M 0.1 M 0.2 M 0.3M 0.1 M 0.2 M 0.3M

0.012 0.12 1.2
Shear rate ሶ [s-1]

0.010 0.10 1.0

0.008 0.08 0.8

0.006 0.06 0.6

0.004 0.04 0.4

0.002 0.02 0.2

0.000 0.00 0.0
0 20 40 60 80 100 0 2 4 6 8 10 0 1 2 3 4 5

a) b) c)
Figure 4.2: Effective shear rate measured on various Carbopol ® concentrations with a)
applied shear rate of 0.01 s−1 , b) applied shear rate of 0.1 s−1 , and c) applied shear rate of 1 s−1 .

81
Rheology and structural build-up 4.2. Results and discussion

Since Bingham yield stress was calculated from measurements at the deflocculated state
of the material, it was associated with the static yield stress at zero rest-time trest ; cf.
Section 4.1.2.

4.2 Results and discussion

This section discusses the test methodology using constant shear rate (CSR), which is
further modified to become the constant strain method. The constant strain method is
then extensively scrutinized by considering various cement pastes. Structural build-up
results are detailed in Sections 4.2.1 and 4.2.2 from the point of view of methodology,
and in Section 4.2.3 with respect to the dependency on material composition. Note that,
the markers (measured data points) on γ̇ef −td and γef −td curves were reduced in number
to improve readability.Also note that, γ̇ap − td and γap − td straight lines without markers
represent applied values and not measured data points. Selected Hägermann flow table
test (HFT) results are tabulated in Appendix A; see Table A.2.

4.2.1 Measurements of structural build-up by means of CSR test

As a basis for discussion, CSR tests performed according to the literature and the
experimental results obtained are described in detail for the reference cement paste, Ref,
and a stiffer cement paste, SP-0.30. A shear rate of 0.08 s−1 was applied for 30 s to Ref
and a shear rate of 0.15 s−1 was applied for 16 s to SP-0.30, both regimes leading to a total
applied strain of 2.4 units.
Often in the literature only applied shear rate γap and test duration td are reported
[113, 114, 139, 153–155, 161, 263]. During this study, it was observed that γ̇ef does not
necessarily equal γ̇ap over the total td ; there is a delay time te before the effective shear
rate reaches the intended applied shear rate; see Section 4.1.3. In addition, due to the
differences between effective strain realized in the experiment and the intended applied
strain, there is a “strain lag”:

γlag = γap − γef (4.1)

The “strain lag” is due to the fact that for a constant applied shear rate and given test
duration, the effective shear rates are lower than the anticipated ones over the elapse
time te . This effect increases with increasing stiffness (yield stress) of material under
investigation, which was first demonstrated in the experiments on Carbopol® solutions;
see Section 4.1.3. Also, cementitious suspensions with higher yield stress (which will
further increase with ageing) exhibit higher initial resistance against rotor movement,
thus, leading to a more pronounced strain lag. Furthermore, some instrumental limita-
tions, e.g., inherent acceleration ramp of a motor from zero to the prescribed rotational
velocity, can lead to strain lag. In contrast to stiff mixtures, by theory and experimental
evidence, in case of more flowable mixtures γ̇ap can be equal to γ̇ef , at least at very
early ages, since the consistency of cementitious material is a function of its age, tage ; see
Figure 4.3a. Moreover, as the sample ages, the structural build-up increases, the pastes
stiffen and γap does not equal γef ; see Figure 4.5a. This tendency is more pronounced if
the cement paste is stiffer; see results for SP-0.30 in Figure 4.5b. This makes it necessary
to critically examine the concept of “constant shear rate” tests used to quantify the
structural build-up of pastes aimed at digital construction.
Figure 4.4a shows that the shear stress at all tested ages of Ref reached static yield
stress values, implying that Ref could deform to flow-onset even at a tage of 150 min. In

82
Rheology and structural build-up 4.2. Results and discussion

23 45 60 23 45 60
90 120 150 90 120 150
0.30 0.21

0.18

0.15
0.20
0.12

0.09
0.10
0.06

0.03

0.00 0.00
0 10 20 30 0 3 6 9 12 15

a) b)
Figure 4.3: Shear rate variation with test duration a) Ref, γ̇ap = 0.08 s−1 for 30 s, b) SP-0.3,
γ̇ap = 0.15 s−1 for 16 s.

contrast, the stiffer paste SP-0.30, Figure 4.4b, could deform to flow-onset only up to tage
of 60 min. Thereafter, the shear stress curves at the end of td are below the corresponding
static yield stress values. In this context, by scrutinizing these two samples using their
shear rate and time variation graphs (Figure 4.3), several aspects can be inferred, leading
to important conclusions.
23 45 60
23 45 60
90 120 150
90 120 150
120 120

100 100

80 80

60 60

40 40

20 20

0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5

a) b)
Figure 4.4: Structural build-up of a) Ref using γ̇ap of 0.08 s−1 applied for 30 s and b) SP-0.30
using γ̇ap of 0.15 s−1 applied for 16 s.
Figure 4.3a shows that after tage of 90 min γ̇ef took considerable time to reach γ̇ap .
Nevertheless, since γ̇ef was indeed able to reach γ̇ap , shear stress indeed reached static
yield stress. The same is true for tage of both 120 and 150 min. In contrast, for SP-0.30, γ̇ef
was able to reach γ̇ap only up to tage of 60 min; see Figure 4.5b. Thereafter, γ̇ef was not
able to reach γ̇ap , and for tage 120 and 150 min, γ̇ef is far below γ̇ap . This explains why

83
Rheology and structural build-up 4.2. Results and discussion

static yield stress (Figure 4.4b) could only be reached up to certain tage of a particular
composition, even though all test parameters remained constant. It also implies that
in the ideal case a) γ̇ap and td must be varied not only for different materials but also
for a single material based on its stiffness, which changes with age, and b) for accurate
measurement of the static yield stress value, one should make certain that γ̇ef reaches
γ̇ap .
Figure 4.5 shows the shear-strain variation during testing. In case of Ref, the shear
strain lag γlag is not significant up to 90 min. However, after a tage of 90 min, γlag increases
gradually. Irrespective of this, the shear stress curves (Figure 4.4a) reach static yield stress
values, as γ̇ef reaches γ̇ap (Figure 4.3a). The actual time during experimentation where
γ̇ef equals γ̇ap is denoted by td,ef . In the case of a less flowable mixture SP-0.30, γef is <
γap , even at an early age of 23 min; see Figure 4.5b. However, up to tage 60 min, the paste
appears to be deformable with γef > 1.5. From Figure 4.4b and Figure 4.3b, it can be seen
that for SP-0.30 after a tage of 60 min γ̇ef is much lower than γ̇ap and that τs (t) is not to
be reached.
23 45 60 23 45 60
90 120 150 90 120 150
3.0 3.0

2.5 2.5

2.0 2.0

1.5 1.5

1.0 1.0

0.5 0.5

0.0 0.0
0 10 20 30 0 3 6 9 12 15

a) b)
Figure 4.5: Shear strain variation a) Ref, γ̇ap = 0.08 s−1 for 30 s, b) SP-0.30, γ̇ap = 0.24 s−1
for 10 s.
Based on the observations for Ref and SP-0.30 with respect to the cases where static
yield stress was reached, a minimum of effective shear strain γef ,min can be identified. In
fact, the idea of a critical minimum strain has already been reported in earlier research
[113]; this parameter can be defined as the minimum strain to be applied effectively
during a measurement to ensure flow-onset and, therefore, the successful measurement
of static yield stress. Mahaut et al. [113] mentioned that a strain in the order of 1, induced
by the entire measurement procedure, was sufficient to change the material state. In this
context, the analyses presented above shall be referred, which show that the effective
strain γef is a more appropriate metric in comparison to the applied strain. For the pastes
under investigation, a minimum γef ,min of 1.5 units was needed to ensure flow-onset.
γef ,min appears to be constant for all the cement pastes investigated, except SP-0.00;
see Section 4.2.3.1. Roussel et al. [172] reported that cementitious materials exhibit a
short-term thixotropy due to colloidal flocculation with a characteristic time on the order
of a few seconds and a long-term thixotropy due to the ongoing hydrates nucleation. In
addition, they underlined the existence of two simultaneously existing critical strains

84
Rheology and structural build-up 4.2. Results and discussion

on cement pastes: “the first critical strain is clearly associated with a very stiff elastic
behaviour of the material at low strain, while the second critical strain is associated with
yielding and flow onset.” This implies that the γef ,min is necessary to ensure flow-onset
breaks down the long-term thixotropy. With Equation (4.2) the findings of this section
can be summarized as follows:

td,ef = td − te (4.2)
γlag = γap − γef (4.3)

a) If td,ef > 0 then γ̇ef reaches γ̇ap

b) For appropriate characterization of the structural build-up, the following condition
must be satisfied: γef ≥ γef ,min ≈ 1.5 units.
In such cases, the maximum shear stress measured reaches static yield stresses of
the material tested: maximum of τ (t) = τs (t).
c) In contrast, if td,ef ≤ 0 then γ̇ef does not reach γ̇ap and γef < 1.5 units.
In such cases, the maximum measured shear stress is lower than the static yield
stress of the tested material: maximum of τ (t) < τs (t).
With respect to measurements on the cement pastes presented in this section, in the
case of Ref for all the tested cases td,ef > 0 and γef ≥ γef ,min . In contrast, in case of SP-0.30,
td,ef > 0 and γef ≥ γef ,min ∀ tage ≤ 60 min and td,ef ≤ 0 and γef < γef ,min ∀ tage > 60 min.

4.2.2 Characterizing structural build-up with varying shear rates and

constant applied strain
The approach of studying structural build-up with constant applied strain is presented
in this section. This approach relies on the hypothesis that so long as the applied strain
is constant – even if shear rate varies – with corresponding “compensation” in the test
duration, measured structural build-up remains the same. If proven true, this approach
enables the characterization of the structuration rate of stiff cementitious materials
with varying shear rates. Here, it should be recalled that the time te , time elapsed for
effective shear rate γ̇ef to reach applied shear rate γ̇ap , is lower at higher shear rates and
consequently γlag will also be lower; see Section 4.1.3. The hypothesis above is examined
more closely by testing Ref with a constant applied strain of 2.4 units. This strain is
applied at shear rates of 0.08 s−1 , 0.10 s−1 , 0.12 s−1 , 0.15 s−1 , 0, 20 s−1 and 0.24 s−1 for a
td of 30 s, 24 s, 20 s, 16 s, 12 s and 10 s, respectively.
Figure 4.6 shows measured shear stresses as a function of effective shear strain for
selected γ̇ap and td combinations. In all cases, it can be generalised that for the first 60
min, the shear stress curves evolve gradually up to a limit of approximately 10 Pa, after
which, at tage of 90 min and 120 min, shear stress increases drastically up to about 40
Pa. These results agree well with findings by Perrot et al. [7]. The structural build-up is
linear up until the characteristic time tc , followed by an exponential increase as the solid
volume fraction increases due to hydration; see Figure 4.7.
The influence of varying shear rates, at constant strain, on the measured static yield
stress values becomes first evident after 120 min, for the first 60 min the differences are
minimal; see Figure 4.6. Similarly, concerning the shear strain, it can be seen in all three
cases presented in Figure 4.8 that up to a paste age of 60 min, the difference between
γap and γef is negligible. However, after 90 min, the shear-strain lag γlag starts to grow.
These variations in the effective strain agree with measured shear stresses, proving the
strain-based approach to be appropriate. Applied shear rate γ̇ap and effective shear rate
γ̇ef , for each combination mentioned above, are given as an illustration in Figure 4.9 for

85
Rheology and structural build-up 4.2. Results and discussion

tage of 23 min and 150 min; see also Figure A.1 in Appendix A. At a tage of 23 min, as the
paste is more flowable, the γ̇ef reaches γ̇ap quickly, whereas at tage of 150 min, due to the
structural build-up, γ̇ef reaching γ̇ap is delayed by a certain te . For a lower γ̇ap , a higher
td is needed, and the elapsed time te is higher, while for higher γ̇ap , lower td is required,
and the te is lower. If constant td and variable shear rates are applied, then effective test
duration is higher for the test with higher applied shear rate; note that the effective strain
will then be constant. For example, in Figure 4.9b, for γ̇ap = 0.08 s−1 and td = 30 s, te is
approximately 20 s whereas for γ̇ap = 0.24 s−1 and td = 10 s, te is about 8 s. Furthermore,
for higher shear rates, the impulse delivered is higher, and consequently the shear stress
curve reaches the static yield stress quickly, even though td,ef is relatively low.
23 45 60 23 45 60 23 45 60
90 120 150 90 120 150 90 120 150
50 50 50

40 40 40

30 30 30

20 20 20

10 10 10

0 0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5

a) b) c)
Figure 4.6: Shear stress vs. effective shear strain at applied shear rate γ̇ap of a) 0.08 s−1 for 30
s, b) 0.15 s−1 for 16 s and c) 0.24 s−1 for 10 s.

40

0.24 s-1
30 0.20 s-1
0.15 s-1
0.12 s-1
20
0.10 s-1
0.08 s-1
10

0
0 30 60 90 120 150

Figure 4.7: Development of static yield stress (SYS) for Ref, measured with a constant applied
strain using various combinations of shear rate and test duration.
Therefore, it is preferable to apply a higher γ̇ap over shorter test durations for less
flowable materials in the case of single-batched measurements. In the case of relatively
flowable mixtures or when multi-batch measurements are conducted, lower γ̇ap over
longer durations can be applied, nonetheless keeping γap constant by adjusting td .
Besides, the order of variation for the lower to higher γ̇ap must be within 10-2 . It is
challenging to keep effective strain γef , which is the decisive aspect for the whole
measurement, at an absolute constant for numerous experiments, especially so if the
rheological properties of the materials under investigation vary in a broad range.
Therefore, in all strain-based measurements, the minimum effective strain as defined in

86
Rheology and structural build-up 4.2. Results and discussion

Section 4.2.1 can be followed as a rule of thumb.

Figure 4.7 presents the development of static yield stress as obtained using various
combinations of shear rate and test durations. The curves show similar courses up to
a paste age of 60 min. The Athix values determined are given in Table 4.4. Despite
the differences in γ̇ap and γap at later ages, the calculated Athix with constant strain
approach is statistically identical to that obtained using the constant shear rate approach.
The Athix,l and Athix,e of the reference mixture from four repeated constant shear rate
tests, expressed as mean ± standard deviation, are 0.125 ± 0.023 and 0.069 ± 0.003,
respectively; see Table 4.4. The average values of Athix,l and Athix,e for the reference
mixture obtained from the shear rate variation in constant strain tests are 0.140 ±
0.022 and 0.063 ± 0.006, respectively. These results prove that the difference in the
structuration rate parameters determined as between both approaches is statistically
insignificant. The Bingham parameters, calculated from the hysteresis loop test are also
presented in Table 4.4. They show coherence in the values confirming that the Ref batches
tested in all investigated cases are rheologically very similar.
23 45 60 23 45 60 23 45 60
90 120 150 90 120 150 90 120 150
3.0 3.0 3.0
2.5 2.5 2.5
2.0 2.0 2.0
1.5 1.5 1.5
1.0 1.0 1.0
0.5 0.5 0.5
0.0 0.0 0.0
0 10 20 30 0 4 8 12 16 0 2 4 6 8 10

a) b) c)
Figure 4.8: Development of the effective shear strain during the tests at applied shear rate γ̇ap
of a) 0.08 s−1 , b) 0.15 s−1 and c) 0.24 s−1 .
0.30 0.30

0.27 0.24 s-1 0.27

0.24 0.24 s-1 0.24 0.24 s-1

0.20 s-1
0.21 0.21

0.18 0.18
0.15 s-1
0.15 s-1 0.15
0.15 s-1
0.15

0.12 0.12 s-1 0.12

0.09 0.09 0.08 s-1

0.10 s-1
0.08 s-1
0.06 0.06

0.03 0.08 s-1 0.03

0.00 0.00
0 10 20 30 0 10 20 30

a) b)
Figure 4.9: Applied shear rates γ̇ap (straight lines) and effective shear rates γ̇ef as a function
of test duration for Ref a) at tage of 23 min and b) at tage of 150 min.

87
Rheology and structural build-up 4.2. Results and discussion

As the hypothesis of constant strain tests presented at the beginning of this section
is proved, it is possible to characterize structuration of less flowable or non-flowable
cementitious materials with a combination of varying shear rates and test durations,
keeping applied strain constant. In Section 4.2.3 constant strain methodology is inves-
tigated further by testing different pastes.

4.2.3 Influence of the cement paste composition on structural

build-up
Table 4.5 gives the parameters under investigation as well as the results obtained from the
tests on various cement pastes. Figure 4.10a shows the development of static yield stress
for the Ref mixture; it increases exponentially after a characteristic tage in agreement with
Perrot et al. [148].
Table 4.4: Overview of test parameters and obtained results for investigations on Ref paste
with constant γap . Athix is calculated using Roussel’s and Perrot’s models. τ0,0 , µ are
Bingham’s parameters.
γ̇ap td γap Athix,l (R2 ) Athix,l Athix,e (R2 ) Athix,e τ0,0 µ
[s−1 ] [s] [−] [P a/min] % variation [P a/min] % variation [P a] [P a · s]
0.08 30 2.4 0.11 (0.99) 21.43 0.070 (0.99) 10.53 2.02 1.03
0.10 24 2.4 0.12 (1.00) 14.29 0.059 (0.99) 6.84 2.07 0.81
0.12 20 2.4 0.14 (0.93) 0.00 0.055 (0.98) 13.16 2.06 0.80
0.15 16 2.4 0.15 (0.93) 7.14 0.065 (0.99) 2.63 2.33 0.84
0.20 12 2.4 0.17 (0.98) 21.43 0.070 (0.93) 10.53 2.64 0.99
0.24 10 2.4 0.15 (0.94) 7.14 0.061 (0.99) 3.68 1.71 0.85

Table 4.5: Overview of pastes composition variation, test parameters and measured Athix and
Bingham parameters.
Paste Dosage γ̇ap td γap Athix,l (R2 ) Athix,e (R2 ) τ0,0 µ
%bwob [s−1 ] [s] [−] [P a/min] [P a/min] [P a] [P a · s]
Superplasticizer dosage variation
SP-0.75 0.75 0.08 30 2.4 0.12 (0.98) 0.07 (0.99) 2.13 0.99
SP-0.50 0.50 0.10 24 2.4 0.18 (0.99) 0.13 (1.00) 11.69 1.23
SP-0.30 0.30 0.15 16 2.4 0.40 (0.99) 0.17 (0.87) 21.72 1.55
SP-0.00 0.00 0.20 20 4.0 1.83 (1.00) 0.75 (0.67) 85.96 2.80
Accelerator dosage variation
ACC-0.25 0.25 0.12 20 2.4 0.14 (0.96) 0.07 (0.97) 1.27 0.79
ACC-1.50 1.50 0.15 16 2.4 0.24 (0.84) 0.09 (0.97) 3.99 0.99
ACC-2.50 2.50 0.15 16 2.4 0.56 (0.99) 0.25 (0.79) 12.80 2.20

Table 4.5 also provides the structuration rate parameters Athix,l and Athix,e calculated
according to Roussel et al. [156] and Perrot et al. [7] along with the Bingham parameters
for all the paste. It is to be noted that the linear model overestimates the value of the

88
Rheology and structural build-up 4.2. Results and discussion

structuration rate before tc . Thus, if used for predicting 3D-printing process parameters
[7], Athix,l can be mission-critical by leading to failure of the printed structure. Therefore,
for such applications Athix,e based on the exponential model is more suitable.

4.2.3.1 Effect of the superplasticizer dosage

The dosage of SP was varied gradually from 0.75 %bwob (Ref) to 0.00 %bwob (SP-0.00);
cf. Table 4.5. In the first two variations of SP, the pastes were deformable, and the applied
constant strain was sufficient for the shear stress values to reach the static yield stress
(SYS); see Figure 4.10b. Note that only the static yield stress measurements for mixtures
Ref and SP-0.50 follow the exponential course. In the cases of SP-0.30 and SP-0.00, growth
rate of static yield stress appears to decrease after tage of approximately 90 and 60 min,
respectively. These pastes are stiffer, and thus static yield stress was not realized in the
tests performed at later ages of the respective cement pastes. In fact, expecting such an
outcome, in case of SP-0.00, the applied strain was increased to 4.0 units (from 2.4 units
applied for Ref). However, even then, the shear stresses did not reach static yield stress at
tage of 60 min as the paste was not deformable; see Figure 4.11a.

40 500

400
30

300
20
200

10
100

0 0
0 50 100 150 0 50 100 150

a) b)
Figure 4.10: a) Measured SYS curve for reference cement paste in comparison to models by
Perrot and Roussel (γ̇ap = 0.08 s−1 and td = 30 s), b) SYS development for all dosages of
superplasticizer; dashed lines indicate the expected development of SYS based on [7].
Therefore, the static yield stresses presented for SP-0.00 and SP-0.30 measured at tage
of 45 min and 60 min, respectively, are not true static yield stresses; the true static yield
stresses at these ages are much higher. For this reason, the static yield stress development
appears not to follow Perrot’s model [7]; see Figure 4.10b. Figure 4.11b shows that for
SP-0.00, the presence of γlag is evident at tage as early as 23 min. Irrespective of this, since
the effective strain is higher than 1.50 units, the shear stress reached the static yield stress
at tage of 23 min. Athix values for all four dosages of SP given in Table 4.5.

4.2.3.2 Influence of adding accelerator

Three different dosages of accelerator – 0.25, 1.50 and 2.50 %bwob – were used in this
investigation. Adding 0.25 %bwob ACC to Ref did not show any increase in static yield
stresses. This holds true even in case of 1.50 %bwob up to tage of 120 min. Bravo et al.
[264] reported that addition of certain alkali-free accelerators with dosage < 2.0 % leads
to a retardation effect. This can be explained by a low amount of Al3+ and SO4 2− supplied
by the accelerator, which amounts might contribute to the formation of an additional
protective thin layer of ettringite around cement grains, which in turn can slow down the

89
Rheology and structural build-up 4.2. Results and discussion

23 45 60 23 45 60
90 120 150 90 120 150
250 5.0

200 4.0

150 3.0

100 2.0

50 1.0

0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 5 10 15 20

a) b)
Figure 4.11: Measurements on Ref without superplasticizer: a) shear stress development with
effective shear strain, b) shear strain development with test duration for various paste ages.

hydration of cement phases [264].

However, addition of 2.50 %bwob accelerator enhanced the structuration rate of the
cement paste significantly. The Bingham yield stress for Mixture ACC-2.50 is 6 times
higher than that of Ref. The paste structuration evolves linearly up to 60 min, after which
a surge in the structuration can be observed; see Figure 4.12a. As a result, flow onset could
be effectively applied only up to a tage of 90 min. Similar to mixtures SP-0.00 and SP-0.30,
in case of ACC-2.50, true static yield stress values were not reached at the ages mentioned
above. Hence, the maximum values from shear stress measurements are plotted instead
of static yield stress values; see Figure 4.13a.
23 45 60 23 45 60
90 120 150 90 120 150
100 3.0

2.5
80

2.0
60
1.5
40
1.0

20
0.5

0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 0 4 8 12 16

a) b)
Figure 4.12: Measurements on Ref with various accelerator dosages: a) shear stress
development with effective shear-strain, b) shear-strain development with test duration for
various ages of cement paste.

90
Rheology and structural build-up 4.2. Results and discussion

4.2.3.3 Influence of replacing cement with secondary cementitious materials (SCM)

Replacing a portion of the cement with SCM resulted in a highly thixotropic paste; see
Section 4.1.1 for details. Considering the corresponding surge in structural build-up with
tage , the total applied strain γap in these investigations was increased to 4.5 units at tage of
23 min to 6.0 units at tage of 150 min; see Table A.1 in Section 9.2. The table also gives the
corresponding combinations of γ̇ap and td . Figure 4.14 presents measured shear stresses
as a function of shear strain.
In all of the measured test samples the shear stress did not reach static yield stress.
Even though the applied shear strain γap was increased to a maximum of 6.0 units, in all
the cases, the effective shear strain γef was far below the applied values of γap . In fact, γef
is lower than 0.06 units in all cases, including that with a γap of 6.0 units; see Figure 4.13b.
Since the static yield stresses were not reached for this highly thixotropic composition,
static yield stress development curve is plotted using the maximum values of measured
shear stresses at the respective tage . These maximum static shear stresses lie much below
the actual static yield stresses. Nevertheless, from these measurements, we can conclude
that SCM paste has a more pronounced structural build-up than all other compositions
under investigation. Figure 4.14b presents the development of static yield / maximum
shear stresses for Ref, ACC-2.50, SP-0.00 and SCM (HFT spread diameters after strokes
for these mixtures were 30.25 cm, 19.50 cm, 10 cm and 17.56 cm, respectively (Table A.2).
Although, the SCM-mixture was more flowable than SP-0.00 and exhibited similar
flowability as compared with ACC-2.50, it clearly had more pronounced structural
build-up. Since SCM has a considerably lower cement content than other pastes, it can be
concluded that the higher structuration rate of SCM was primarily due to flocculation
and, secondarily, due to hydration. Thus, replacing cement with SCM appears to be
more appropriate to make cementitious materials “printable” than using accelerators or
reducing SP dosage. In fact, using set-accelerators will also lead to difficult situations if
printing or pumping processes are interrupted, e.g., due to blocking or machine failure.
However, these all hold true only if inline accelerator mixing technique just before the
nozzle is not possible; see also Section 2.3. Furthermore, it must be noted that the degree
of phase change achievable with SCMs is limited when compared to that of setting
and hydration accelerators. Consequently, in applications where faster construction rates
must be achieved, use of accelerators at higher dosages (>3 % bwoc) might be necessary.

10 -1 23
0.1

150 45
10-2
0.01
120
10-3
0.001 60
90
10-4
0.0001 90
60

10-5
0.00001 120
30
-6
0 10
0.000001 150
0 50 100 150 0 10 20 30 40

a) b)
Figure 4.13: a) SYS development for different accelerator dosages, b) the effective and applied
shear strain on logarithmic scale as a function of test duration in the case of Mixture SCM.

91
Rheology and structural build-up 4.3. Universal approach

23 45 60 SCM SP_0.00
90 120 150 ACC_2.50 Ref
300 350

250 300

250
200
200
150
150
100
100

50 50

0 0
0.00 0.05 0.10 0.15 0.20 0 50 100 150

a) b)
Figure 4.14: Measurements from Ref with added SCM as partial cement replacement: a) shear
stress development with effective shear-strain and b) development of SYS with age for different
paste compositions.

4.3 Universal approaches suggested for further studies on

structural build-up
Taking the broad spectrum of concrete compositions and the challenges in measuring
structuration rate into consideration, two universal approaches, i.e., applicable irrespec-
tive of material flowability, can be suggested as described below.

Approach 1 – Constant shear rate method (for multi-batch procedure

only):
Conventional constant shear rate test with a very low γ̇ap and a constant but long td can
be followed with the precondition that flow onset must take place.
In other words, a clear peak or plateau in measured shear stresses must be observed
when plotted against effective strain. This, however, means longer test durations for
stiffer compositions and different test durations for the same composition at various ages.
Therefore, a long enough td must be ensured so that static yield stress can be measured for
all compositions under investigation. An easy way to determine suitable td to be applied
is by testing the least flowable composition of the planned experimental program first
and then applying this for the rest of the testing program. In this way, it can be assured
that irrespective of the flowability of individual mixes and low shear rates, flow onset will
take place within the tested duration.
This approach is only valid if flow onset can be reached with a constant, low γap . If
the structuration rate of the material is much higher than the rate of increase of effective
strain, then flow onset cannot take place. For non-flowable and high Athix,e mixtures if low
shear rate is applied, then the increase in total effective strain will be much lower than
minimum strain needed for flow onset. In such case, extended td may also be inadequate
as structural build-up is progressing at a comparatively faster rate.

92
Rheology and structural build-up 4.4. Chapter summary and conclusions

Approach 2 – Effective strain rate approach (valid both for

single-specimen and multi-batch procedures):
If a single-specimen approach is followed, the constant shear rate method described
above is not applicable. When a longer td is applied, the test sample under investigation
at one measurement could be deformed to an extent that the successive measurement will
not be valid. The only way to avoid such a scenario is taking the effective strain γef into
consideration, thus moving from shear rate based approach to a strain-based approach.
By considering effective strain, the universal approach can be proposed as:
Stress-growth test based on constant strain satisfying the condition: γef ≥ γef ,min .
In such cases, flow onset is assured, and maximum measured shear stress reaches
static yield stress of tested material: maximal of τ (t) = τs (t). So long as the condition
γef ≥ γef ,min is satisfied, γ̇ap , td and consequently γap can be varied according to the
particular material under investigation and its tage . In doing so, the lowest possible γ̇ap
should be applied to minimise viscous effects [161, 168] on the measured shear stresses.

4.4 Chapter summary and conclusions

It is demonstrated that applying low shear rates over short measurement periods does not
always result in the achievement of flow-onset in stiffer cementitious materials commonly
used in digital construction. For such materials, a characteristic delay exists before the
effective shear rate reaches the applied shear rate. This leads to lower effective strains in
the tested materials than the applied strains and consequently to the erroneous character-
ization of structural build-up. This chapter introduced a single-specimen approach based
on single-batch production and variable shear rates for characterizing the structural
build-up of a wide range of cement pastes. Applying this approach, the effect of cement
paste composition on its structural build-up was investigated.
With extensive analyses of experimental results, it was demonstrated that even if the
applied shear rate is varied in the range of 0.08 s−1 to 0.24 s−1 , by adapting corre-
sponding measurement duration td for the sake of compensation, structural build-up
can be successfully measured without significant variation in the calculated structuration
parameter. Based on this research at hand, following further conclusions can be drawn:
a) The constant shear rate test with shear rate γ̇ap as low as 0.001 to 0.01 s−1 can
be applied as a universal approach without taking effective strains into account
only if flow onset can be assured. This generally requires a longer td ; thus, such a
measuring protocol is not applicable to single-specimen approaches.
b) Considering today’s very wide spectrum of concrete compositions, which will surely
expand even further in future, and the corresponding wide range in rheological
properties, protocols for characterizing structural build-up relying on constant
shear rate appear deficient.
c) Irrespective of the type of cement paste being investigated, a stress growth test
based on constant effective strain γef can be followed, satisfying the condition
γef ≥ γef ,min , where γef ,min is the minimum effective strain needed to ensure flow
onset. In such cases, maximum measured shear stress reaches static yield stress of
tested material: maximum of τ (t) = τs (t). For the pastes investigated in this research
γef ,min is ≈ 1.5 units.
d) As long as the condition γef ≥ γef ,min is satisfied, γ̇ap , td and consequently γap
can be varied according to the particular material under investigation and its tage.
However, the lowest possible γ̇ap should be applied to minimize viscous effects on
the measured shear stresses.

93
Rheology and structural build-up 4.4. Chapter summary and conclusions

e) Although γ̇ap was varied in the range from 0.08 s−1 to 0.24 s−1 in the work at hand,
for the majority of the experiments γ̇ap of 0.15 s−1 was used. The effective shear
rates γ̇ef in all these cases were lower than the corresponding applied shear rates
γ̇ap for the major period of the td due to the time elapsed te for effective shear rate
to reach applied shear rate.
f) The variation in the calculated structuration parameter Athix , which was the
primary objective of the present investigation, has been proven statistically insignif-
icant. Potential viscous effects due to the relatively high γ̇ap appear to have no
significant effect on Athix .
g) Based on representative static yield stresses, mixtures with partial cement
replacement by SCM exhibited the highest structuration rate in this study,
primarily due to the increase in flocculation. It appears that using SCM is
appropriate approach to make cementitious materials “printable”. In fact, use of
SCM can be preferred to adding accelerators or reducing dosage of superplasticizer
in case of applications where inline-mixing of accelerator in printhead is not
possible and/or moderate vertical construction rates are targeted.

Acknowledgements: The presented work in this chapter was partially carried out in
the form of a student project work [265] under close supervision of the author. The
work is published in peer-reviewed journal Cement and Concrete Research [27]. All the
acknowledgements and support of the co-authors mentioned in [27] are valid in this
chapter too.

94
Chapter 5

CONCRETE TRANSPORT AND

PUMPING

This chapter addresses another major stage of processing 3D-printable cementitious ma-
terials (3PCs) — transport/placing/pumping. Traditionally, fresh concrete is transported
(excluding motorized vehicle transport such as ready-mix trucks) in many ways including
pumping, crane-bucket, shotcrete, manual transport, etc. Alternatively, raw-materials of
concrete can be moved from a place A to place B, and then, mixed shortly prior to the
actual application. The latter approach has specific advantages with respect to digital
construction (DC); see Section 5.1e. Currently, concrete pumping has emerged as one of
the most widely used approaches for transporting concrete by enabling faster concrete
placement and thus reducing construction costs.

5.1 Various approaches for placement of 3PCs

In the case of on-site DC techniques, concrete must be transported without any blockages
or circuit breakdowns over longer distances and time durations. The pumping specifica-
tions and related material requirements may vary significantly depending on the applied
DC technique. Considering the lack of published research on this topic, a theoretical
perspective is presented in this section. The principal requirement for this process is to
transport 3PCs from the production mixer to the printhead/nozzle-outlet. This task can
be accomplished in:

a) one-step with high yield stress 3PCs: direct pumping of the 3PCs from the mixer
to the nozzle-outlet. In this case, 3PCs must have high-enough initial yield stress
to ensure sufficient “buildability” (see Section 2.6); the high yield stress τ0,0 is
obtained without necessarily using a phase changing agent (PCA), for e.g., by
reducing (w/c)eq or optimizing paste/aggregate ratio. This approach may work for
short transport distances and brief pumping. Achieving high τ0,0 without increasing
plastic viscosity µ0 is not trivial and as a result, the pumpability of this category of
3PCs is problematic.
b) one-step with PCA dosed 3PCs: The initial τ0,0 and µ0 are set low enough to ensure
placing, however, are controlled to increase drastically by the time of deposition,
with help of a PCA. One common example is adding hydration/setting accelerators
to 3PCs, if pumping distance is short. The dosage of PCA is adjusted so that
sufficient “open time” is available for pumping, after which material pumpability
diminishes and buildability increases rapidly. The prerequisite for this approach
is in-depth understanding of various printing process stages and the hydration

95
Concrete transport and pumping 5.1. Various approaches for placement of 3PCs

kinetics resulted of PCA dosage. The dosage and delay time for activation of PCAs
must be synchronized precisely to facilitate the pumping and extrusion phases.
Even then, this case comes with the increased risk of “blockages” and reparation
expenses, due to potential disruption in the printing process.
c) one-step with high structural build-up (SBP) 3PCs: The initial τ0,0 and µ0 are set
sufficiently low to ensure pumpability. The required yield stress at deposition above
the critical yield stress τ0,c is achieved relying on thixotropy and structuration
without using a PCA. The high thixotropic nature of this category 3PCs keeps the
viscosity and yield stress of the 3PCs low as long as the material is being sheared
(structure break down). After extrusion, buildability is enhanced through structural
build-up at rest. This case comes with the limitation of the printing rate, according
to [101].
d) two-steps with inline-mixed 3PCs: The initial τ0,0 and µ0 are set low enough to
ensure placing the 3PCs from the mixer to a “reservoir” on the printhead. The phase
change is achieved by dosing and homogenizing PCA in the printhead followed
by deposition through the nozzle. This promising approach has the prerequisite of
(light-weight) inline mixing tools here termed as “second-stage mixer1 ”, which can
be mounted on moving printhead. To the best of author’s knowledge, there are no
scientific publications on this issue as of yet.
e) two-steps with dry-mix 3PCs: Conveying dry-mix to a mixer that is mounted on
to the printhead where the wet-mixing takes place. Since the transfer distance of
hydrated-material is very short, highly thixotropic and/or accelerated 3PCs can
be produced guaranteeing faster printing rate. This approach can be beneficial
for constructing large structures in remote areas with limited ready-mix concrete
availability and also where surrounding (temperatures, wind, etc.) conditions may
vary significantly. However, heavy capacity robotic arm or manipulator will be
needed to carry the mixing system on the printhead.
f) two-steps with dynamically controlled 3PCs: In this approach, 3PCs are premixed
with PCAs which enable on-demand controlling of rheological properties. In
such approaches, rheology is controlled actively, for example, by mixing magnetic
particles that can be controlled with externally applied magnetic fields to form rigid
or entangled structure and thus increasing the shear resistance of the cementitious
materials [28]. By definition, this process is reversible at least to some extent. Hence,
it is possible to alter the rheology of 3PCs multiple times at various stages of
DC. Yield stress evolution, if two-step variation of this approach is used, can be
represented with the process d) in Figure 5.1.

The evaluation of static yield stress for the above described approaches is represented
with full lines in Figure 5.1. The dashed lines indicate the critical minimum yield stress
that 3PCs must possess to ensure buildability. The approaches a) and b) can be very
critical if pumping process breakdowns occur temporarily. Excluding the dry-mix process
e), the inline-mixed 3PCs d) appears to be the most appropriate among the presented
approaches. The high SBP (thixotropic) 3PCs approach can be employed where an
inline-mixing technique is not feasible and vertical construction rates are moderate; see
Chapter 4. The pumpability requirements vary corresponding to the placing approaches.
At this stage of research on DC, conclusive statements on pumpability of 3PCs cannot
be asserted. However, based on the available literature addressing non-printable types of
concretes and the research carried out in this PhD, the following statements can be made:
1 Could be a static or dynamic mixer.

96
Concrete transport and pumping 5.1. Various approaches for placement of 3PCs

Printhead

a)

b)

c)
Position of PCA addition
d)

e) Dry mix transport

Time, process-progress
M

Production Transport Printhead Deposition

Figure 5.1: Illustration of various approaches for placing 3PCs in DC; full lines show yield
stress development over time, the critical yield stress is shown with dashed lines. Green and
red shades indicate perceived low and high difficulties in pumping, respectively.

(i) Among the pure plug, and plug plus shear flows, for the placing approaches a) as
well as potentially b) and c) pure plug flow would occur. In contrast, for approach
d) plug plus shear flow is more likely. This statements are drawn based on the
influence of τµo on the flow profiles during pumping. Feys [182] detailed in length
how increase of τµo increases the tendency of plug flow and influence of yield stress
on the pumping pressures. In the approaches a), c) and b) the initial yield stress is
expected to be much higher than that in the case of approach d); see Figure 5.1.
(ii) 3PCs inherently requires high structuration rate, as a consequence, the influence
of thixotropy will be critical on the pumpability. So far, the influence of increase
in rheological properties at rest on concrete pumpability have not gained much
attention. In their revealing research, De Schutter et al. [13] emphasized that
“short interruptions during pumping led to major difficulties in resuming pumping
operations due to sometimes tremendous effect of internal structural build-up”.
With large-scale pumping tests, they have identified that a 20 min delay lead to
increase of pressure loss by 50 % for a mixture with Athix of 0.3 P a/s. Despite
increasing the pumping pressures to 350 bar, the pumping operations could not
be resumed after the delay.
(iii) DC by nature is a “step-wise” process; the deposition of the material is often
stopped to move the printhead to a new position. If a second stage reservoir is not
used to “collect” the pumped concrete and achieve phase change in the printhead,
the intricacies of printing process lead to disruption of the pumping process.
(iv) Unlike casted concrete, the rheological properties of the 3PCs need to be precisely
set and controlled to ensure printability. The pumping-induced changes in the
rheological properties and air-content can therefore have more pronounced conse-
quences in form-work free construction. It is technically more accurate to conduct

97
Concrete transport and pumping 5.2. Testing pumpability of concrete with Sliper

any rheological as well as mechanical characterization of printed concrete after

pumping it through the representative pipeline or subjecting to similar pressure
and shearing conditions.

Considering the large variation in concrete transportation and placing processes,

influencing parameters and the lack of dedicated research, a numerical pumpability
testing approach was developed as a part of this dissertation. In the following sections,
the experimental background and the developed numerical model are described.

5.2 Testing pumpability of concrete with Sliper

As the first part of this research, extensive laboratory investigations [10] were conducted
using Sliper; see Figure 5.2. Using the rheological parameters obtained from Sliper
experiments and the specifications of pumping circuit, one can estimate the discharge
pressure P required for a pumping circuit under field conditions using Equation (5.4)
(considering plug flow with no deformation in the plug [10, 18, 190]). Details of Sliper
experimental procedure and derivation of Equation (5.4) are described by the author in
[10], in this section only a brief summary is presented. Among the two models proposed
by Kaplan (see Section 2.4.2), the plug-flow equation is adapted to the Sliper geometry to
obtain Equation (5.1). The physical basis for this equation is that the pumping discharge
pressure is a summation of pressures associated with yield stress Py , plastic viscosity Pv
and dead-weight pressure PH [10].

4l 16l µi
P = Py + Pv + PH = τ0,i + Q + ρgH (5.1)
d πd 3 e
where τ0,i and µi are the yield stress and plastic viscosity of the interface layer, also called
lubricating layer (LL). The parameter e represents the thickness of LL, Q is flow rate, d
and l are the diameter and length, respectively, of the Sliper pipe; see Equation (5.2).
Due to the limitations in determining the thickness of LL accurately, the parameter
µi
e is replaced with an effective viscosity parameter b. The values of P and Q of Sliper
measurements are linearly related with a P -intercept – A; see Figure 5.2c.

P = A+B·Q (5.2)

Analogous to Bingham model the P -intercept A and slope of P -Q curve B are related
to yield stress and plastic viscosity of the material in the vicinity of Sliper pipe. The
atmospheric pressure PH is nullified before every Sliper test [10]. Equations (5.1) and (5.2)
give together Equation (5.3).

d ·A B · π · d3
a= , b= (5.3)
4l 16 · l
Subsequently, Equations (5.1) and (5.3) provide Equation (5.4).

4l 16 · l · Q
P= a+ b (5.4)
d π · d3
For a large-scale pumping circuit under field conditions, pumping pressure necessary for
a desired flow rate Q can be computed by replacing l and d values in Equation (5.4) with
length L and diameter D of the pipeline.
From this point on, this chapter is organized into the following segments: experimental

98
Concrete transport and pumping 5.2. Testing pumpability of concrete with Sliper

background (Section 5.2.2), single-phase numerical model (Section 5.3), and single-phase
model enhanced with user-defined function (Section 5.5), followed by the summary and
conclusions (Section 5.6). In both numerical model sections, the modelling schemas are
presented in detail, followed by results and discussion. In the initial single-phase model,
Sliper simulations were carried out without considering the LL. It was demonstrated
that single-phase numerical models which do not consider the LL are applicable only to
some specific concrete compositions. Upon this background, the single-phase model was
improved by implementing a separate LL using a user-defined function; which properties
were calculated using Chateau-Ovarlez-Trung and Krieger-Dougherty models. Finally,
the applicability of user-defined single-phase model for predicting pressure measured
with Sliper is demonstrated for all concrete compositions under investigation.

5.2.1 Significance of developing a Sliper numerical model

Today’s very broad spectrum of concrete compositions, which will certainly become
even broader in the future, makes it necessary to perform numerous experiments to
characterize their rheological properties and subsequently determine their pumpability.
Moreover, considering on the one hand that concrete properties are time and thixotropy
(resting time) dependent and that on the other hand pumping in the case of modern
processes such as DC demands pumpability for longer durations, extensive experimental
studies are needed to assure the pumpability of concrete, even if its composition is
unaltered. Therefore, a purely experimental approach for assessing pumpability does not
seem feasible in all the cases, since extensive experimental tests demand a great deal of
time and economic investment.
As in many other fields, numerical simulations present a promising tool in com-
plementing experimental studies [146] since they may not only save time and costs
but also increase the accuracy of the predictions, enable better insight into material
behaviour, and support material design as well. With this in mind the experimental
research on the pumpability of concrete [10, 190] has been extended by developing
a numerical model which simulates the Sliper experiment and in this way enables
the estimation of the discharge pressure numerically. These numerically determined
pressures from the virtual Sliper can thereafter be utilized to predict the pumping
pressure for pumping circuits of various lengths and configurations using Equation (5.4)
[10, 22]. The virtual Sliper also makes it possible to investigate numerous concrete
types with wide variations in their rheological properties without conducting physical
experiments. Thus, virtual Sliper can play an important role in establishing intelligent
material design approaches for pumpable concretes. In addition, virtual Sliper becomes
an inherent part of DC approaches, where minimal involvement of human labour is
envisioned with enhanced machine-material integration (autonomous in the lines of
Industry 4.0). Fresh concrete properties can be determined online or “in the process”
during transporting and the printing process, these properties could then be sent to
virtual Sliper. Virtual Sliper predicts the pumping pressures needed to ensure consistent
pumping flow-rates considering the “change” in rheological properties as well as in target
structure geometry. Thus, a numerical model to simulate Sliper experiments or another
adequate technique to predict pumping behaviour is indispensable.
The approach presented in this chapter is significantly different to those previous to
it. In the literature, numerical models were primarily developed to study concrete flow
characteristics, and that of LL, in a long pumping circuit. So far no numerical model is
available to link the resulting rheological properties as observed in the Sliper tests to any
other more commonly used experimental procedure such as the use of a rheometer. This

99
Concrete transport and pumping 5.3. Virtual Sliper: single-phase numerical model

is a rather significant knowledge deficiency that should be overcome with the utmost
urgency, considering the proven applicability of Sliper for pumping pressure prediction
and its newness and unavailability in most concrete laboratories and ready-mix concrete
practitioners. This chapter addresses this challenge by presenting virtual Sliper – a
CFD model which is able to predict Sliper experimental results using rheological input
parameters according to the Bingham model as obtained from Viscometer tests.

Upper pipe

Handles

Lower pipe

Piston

Figure 5.2: Sliper, courtesy of Schleibinger Geräte T. u. G. GmbH.

5.2.2 Experimental background for the virtual Sliper

As the first part of this research, extensive laboratory investigations of the pumpability
of fresh concrete were conducted using Sliper [10]. Twelve concrete mixtures were
systematically investigated, with the major focus on the influence of various constituents
of concrete on its rheological properties and pumpability. The results obtained clearly
demonstrated the links between such mixture parameters as the water-to-binder ratio,
aggregate shape, use of admixtures, and consistency class on the one hand and the
pumpability of concrete on the other. Figure 5.3 gives an overview of the investigated
mixtures. All these concretes were also tested using a ConTec Viscometer 5 and the
flow table. The evaluation of the results emphasised the advantages of using Sliper to
characterise the pumping behaviour of concrete, while a positive correlation was found
between the Sliper viscosity parameter b and the plastic viscosity µ as obtained from the
Viscometer tests. Finally, the predicting capability of Sliper was validated under field
conditions by measuring full-scale pumping pressure. The details of the mixtures tested,
testing procedures, and the results obtained are presented and discussed in [10].

5.3 Virtual Sliper: single-phase numerical model

In the Initial single-phase model (ISPM), concrete was assumed to be a continuous
Bingham fluid.

5.3.1 Modelling schema

Concrete flow was modelled by the conservation equations for mass and momentum of
steady-state, incompressible flow. The continuity equation for the conservation of mass

100
Concrete transport and pumping 5.3. Virtual Sliper: single-phase numerical model

Parameter Concrete composition and specifications

Cement CEM II 42.5

Concrete OC HPC

Aggregate Round Crushed Round Crushed

(w/c)eq 0.45 0.6 0.45 0.3 0.3 0.3 0.3

Consistency F3 F3 F3 F3 F3 F5 F5

SCM Microsilica Microsilica Fly ash

Number 1 2 3 5 7 10 11
Figure 5.3: Overview of concretes investigated experimentally; see [10].

is [266]:

~=0
∇·u (5.5)

where u
~ is the velocity, ∇ is the divergence in tensor notation, and ∇ · u
~ is the time rate of
change of the volume of a moving fluid element in an infinitesimally small cell [198].
The momentum in the numerical model is conserved according to:

∇ · (ρ~
uu~ ) = ∇ · σ + ρ~
g (5.6)

where g~ is the gravitational acceleration, ρ is the density and σ is the stress tensor. σ can
be subdivided into normal stresses σii and the shear stresses τij as:

0 0 σxx + p
    
σxx τxy τxz  p τxy τxz 

  

σ = τyx σyy τyz  = − 0 p 0 +  τyx σyy + p (5.7)
  
τyz 
  
τ
zx τzy σzz 0 0 p  τzx τzy σzz + p
  

σ = −pI + τ (5.8)

where p is an isotropic pressure term, −∇p is the pressure gradient, I is the unit dyadic,
and τ is the shear stress tensor [267] as reported in [146].
From Equations (5.6) to (5.8) we obtain Equation (5.9):

∇ · (ρ~
uu~ ) = ∇ · τ − ∇p + ρ~
g (5.9)

For incompressible flow, shear stress can be expressed as:

∇ · τ = η∆~
u (5.10)

u · (∇ · u
ρ~ ~ ) = −∇p + η∆~
u + ρ~
g (5.11)

here η is the dynamic viscosity; see also Section 5.3.2. The concrete flow was considered
isothermal in accordance with [146, 200, 203].

101
Concrete transport and pumping 5.3. Virtual Sliper: single-phase numerical model

5.3.2 Implementation of the Bingham model

Fresh concrete is a heterogeneous material with constituents of various sizes and shapes.
Its rheological behaviour is that of a yield stress fluid, i.e., it yields non-zero shear stress
at a shear rate of zero. With increasing shear rates, the shear stress increases often
non-linearly, thus accounting for shear thickening or thinning effects. Hence, concrete
can be classified as a non-Newtonian fluid with yield stress. The Hershel-Bulkley model
represents this shear stress-shear rate (τ − γ̇) relationship (Equation (5.12), Figure 5.4aa)
using three parameters: yield stress τ0 , plastic viscosity µpl , and power-law index n. For
the sake of simplicity, the description of concrete as a Newtonian fluid with yield stress is
often applied alternatively, the so-called Bingham model (Equation (5.13), Figure 5.4ab).
This model was proven to represent concrete’s rheological behaviour with adequate
accuracy [197, 268, 269].

τ = τ0 + µpl γ̇ n (5.12)

τ = τ0 + µpl γ̇ (5.13)

n>1 X
1
n=
1
n<1 Y Z

a) b) c)
Figure 5.4: a) Shear stress vs. shear rate curves for the bi-viscous model (if n=1 and γ̇c is close
to zero then theBingham model is represented), b) a part of discretized Sliper quadrant, c) a
detailed top view of the optimized mesh.
Table 5.1 gives a conceptual physical description of the numerical model and experi-
ments. Input and output parameters of the numerical model were chosen to reflect the
corresponding experimental parameters closely.
In the present research, the Bingham model was implemented by adapting the built-in
Herschel-Bulkley model, also called the bi-viscous model [203] of commercial CFD solver
ANSYS Fluent; see Equation (5.14). The bi-viscous model emulates two viscous fluids:
one with a very high, constant dynamic viscosity η0 and one with gradually decreasing
dynamic viscosity ηi . The change from one behaviour to another occurs coincidentally
with the passing of the critical shear rate γ̇c :
" #
τ0 γ̇ γ̇

η0 = (2 − ) + k (2 − n) + (n − 1) γ̇ ≤ γ̇c



γ̇c γ̇c γ̇c


η= (5.14)


" #n−1

 τ 0 γ̇
ηi = γ̇ + k γ̇ γ̇ > γ̇c



c

Assigning the power-law index n equal to 1, Equation (5.14) results in the Bingham
model as long as the shear rate is higher than the critical shear rate, i.e., γ̇ > γ̇c .

102
Concrete transport and pumping 5.3. Virtual Sliper: single-phase numerical model

τ0 γ̇

η = (2 − )+k γ̇ ≤ γ̇c

0


γ̇c γ̇c

η= (5.15)


τ0
ηi = γ̇ + k

γ̇ > γ̇c




To complete the implementation, the critical shear rate γ̇c was assumed to be close
to zero, thus prompting the solver to examine the second case in Equation (5.14)
immediately after flow initiates, i.e., even at very low shear rates. Low γ̇c implies a very
high dynamic viscosity η0 as long as γ̇ ≤ γ̇c .

5.3.3 Geometry, boundary conditions and model parameters

The numerical discretization of the computational domain plays a vital role in the overall
performance of the model. The accuracy of the model significantly increases with more
detailed, finer discretization but leads also to a drastic increase in computational time.
Hence, an optimum is to be strived for, which would represent the physical phenomena
sufficiently while being cost-effective. The Sliper pipe is axisymmetric; see Figure 5.2. The
flow of the concrete can be assumed to be axi-symmetric in a macroscopic perspective.
Therefore, instead of the whole pipe geometry, a quarter of it was modelled using the
meshing tool ICEM CFD; see Figure 5.4. The modelled geometry is created as an assembly
of four surfaces: pipe wall (Sliper pipe), top (top surface of filled concrete), bottom
(piston head of Sliper, where the pressure sensor is located) and internal walls, the inner
boundaries of the Sliper quadrant under consideration. The boundary conditions applied
are a no-slip moving wall for the pipe wall, symmetry for the internal walls, and a fixed
wall for the top and bottom walls. Because of this relatively simple geometry, the use
of the single-phase material model, and a structured grid, the computational time is
relatively low. The boundary layer close to the pipe wall is important in investigating
the lubricating layer phenomenon. The fineness of discretization (dimensions of finite
elements) plays a significant role in numerical simulations. Considering this, a numerical
refinement near the wall surface was employed; see Figure 5.4. Instead of creating fine
mesh throughout the entire modelled domain, which would increase the computational
time drastically, the region close to the pipe wall was modelled using a finer mesh. Since
the observed thickness of the lubrication layer varies between 1 mm and 5 mm (see details
in Section 5.5.2), the size of the elements close to the pipe boundary was chosen as
0.84 mm. This resulted in the total number of 58460 elements in the model.
Table 5.1: Experimental and numerical model parameters.
Experiment Numerical model
Input a) Concrete composition a) Rheological properties (from
b) Velocity of pipe (varies with Viscometer tests)
added weights) b) Velocity of pipe (wall velocity
parameter)
Output a) Pressure measured with the a) Pressure as “measured” by
sensor virtual pressure sensor
b) Pumpability curves (Excel) b) Pumpability curves (Excel)

Concrete composition and pipe velocity V are the varying parameters in the Sliper
experiments. In other words, different rheological properties were attained by mix design,

103
Concrete transport and pumping 5.4. ISPM results and discussion

and different pipe velocities were obtained by changing the weights attached to the
pipe [10]. Similarly, the rheological parameters of the concretes under investigation,
yield stress τ0 and plastic viscosity µ, as measured using Viscometer and pipe velocity
from Sliper tests, were used as input parameters for the numerical model. The output
parameter from the experiments was pressure P measured by a sensor at the bottom
of the Sliper pipe for different discharge rates Q so that pumpability P -Q curves could
be plotted [10]. In the numerical model, the pressure acting at the bottom wall was
“measured” by means of a virtual sensor, which calculates it as area-weighted averages.

5.4 ISPM results and discussion

The experimental results corresponding to the following numerical simulations were
published in [10,180]. Experimental results are referred in this chapter only to the extent
necessary, complete analyses and discussion can be found in [10].

5.4.1 Single-phase model results

Seven concrete mixtures, namely 1, 2, 3, 5, 7, 10 and 11, were chosen from the
experimental study [10] based on the availability of the Viscometer results, but for a
better overview the notations are used unaltered in this thesis. Figure 5.5 shows measured
pressures from Sliper experiments and calculated pressures from ISPM simulations over
corresponding flow rates. Both experimental and numerical results confirm the consensus
that with an increase in flow rate the discharge pressure becomes higher as well, showing
a nearly linear P -Q relationship. The pressure values calculated with ISPM showed
good agreement with the Sliper measurements, both qualitatively and quantitatively,
for mixtures with low water-binder ratio of 0.3, i.e., Mixtures 5, 7, 10 and 11. The
representative results are shown in Figure 5.5a on the example of Mixture 10. In contrast,
for mixtures with high water-binder ratios of 0.45 and 0.6, especially for Mixtures 2 and
3 (see Figure 5.5b), the calculated results showed little agreement with experimental
findings. Correspondingly, the P -Q diagram can be roughly subdivided into two areas
as showed in Figure 5.6.
Pressure P [kPa]
Pressure P [kPa]

Flow rate Q [m3/h] Flow rate Q [m3/h]

a) b)
Figure 5.5: Comparison of experimental results and corresponding numerical simulations: a)
Mixture 10 ((w/c)eq = 0.3), b) Mixtures 1 and 3 (both with (w/c)eq = 0.45).
The primary reason for the poor agreement between the calculated and measured P-Q

104
Concrete transport and pumping 5.4. ISPM results and discussion

curves for concretes with relatively high (w/c)eq appears to be the effect of the lubricating
layer, which forms at the pipe wall and has an immense influence on the discharge
pressures according to [10, 18, 22, 24, 181, 193]. Since ISPM does not consider the LL, i.e.,
constant material properties are assigned throughout the model domain (Sliper pipe),
the pumping behaviour of mixtures with relatively high water-to-binder ratios cannot
be predicted well. The reason behind is that the rheological properties of the LL in such
mixtures differ very pronouncedly from those of bulk concrete and the thickness of the
LL is relatively high in comparison to concretes with lower (w/c)eq . This conclusion is
supported by the results reported by Jo et al. [178], who observed a decrease of LL
thickness from 5 mm to 1 mm with decreasing water-to-binder ratio. Furthermore, Choi
et al. [24] measured increase in LL thickness with increasing w/c. Thus, neglecting LL in
the ISPM makes the model only applicable to concrete with a low (w/c)eq .
35
Mixture 1
11
30
le
5
l i cab able Mixture 2
p c
25
(w/c)eq = 0. 30 7 ap ppli
M a
ISP not Mixture 3
Pressure P [kPa]

M
20
10 ISP
1 Mixture 5
15 3
Mixture 7
10
(w/c)eq = 0. 451 2 Mixture 10
5 (w/c)eq = 0. 60
Mixture 11
0
0 10 20 30 40 50 60 70
3
Flow rate Q [m /h]
Figure 5.6: Applicability of ISPM based on the correlation of simulation results with
experimental results (the base figure is adapted from [10]).
Still further, it can be deduced from the fundamentals of the numerical model
developed that the larger the discrepancies in Sliper and Viscometer measurements for a
concrete mixture, the poorer the experiment-model correlation (EMC) is. Table 5.2 shows
that mixtures with high (w/c)eq , i.e., Mixtures 1, 2 and 3, yielded very low values of
Sliper viscosity parameter b, 0.59 P a·s/mm, 0.14 P a·s/mm and 0.81 P a·s/mm, respectively,
while the corresponding plastic viscosities measured by Viscometer were 76 P a·s, 38
P a·s and 230 P a·s, respectively. To understand the relative sensitivity of both Sliper and
Viscometer to mixture variation, a value denoted as “discrepancy” ∆ (Equation (5.16))
was calculated. When comparing measurements from different instruments, to avoid
the common variables such as geometrical factors and test conditions, a “relative
parameter” can be calculated. Ferraris et al. [149] calculated relative plastic viscosity
to compare measurements from various rheometers. By dividing Sliper and Viscometer
measurements of viscosity viz. bn and µn of any composition n by the same of a reference
composition r relative viscosities were calculated. In this research, Mixture 10 was chosen
as reference considering in the measured spectrum of viscosities its closeness to the
medium viscosity measurements as well as the low influence of the lubricating layer;
see also Section 5.5.2. In the next step, the discrepancy is quantified by dividing the
difference in relative viscosities as obtained from Sliper and Viscometer measurements
with relative viscosity obtained by using Sliper Rb .

105
Concrete transport and pumping 5.4. ISPM results and discussion

The calculated discrepancy values are presented in Table 5.2 as percentages. Very
high discrepancies in cases of Mixtures 2 and 3 indicate very poor EMC. The negative
sign in the discrepancy indicates that Sliper relative viscosity parameter showed higher
reduction than that of relative plastic viscosity from Viscometer.

µn
b 
n
br − µr
∆= (5.16)
Rb

It is noteworthy here that Sliper measurements catch the influence of the LL much
better than does the Viscometer, which originates from the devices’ respective geometries
and functioning principles in both experiments. In the Viscometer, the whole concrete is
sheared with the help of special ribs designed to prevent slippage. Although in Sliper
varying from mixture to mixture, primarily the material in the vicinity of the pipe wall is
deformed. The discrepancy of mixtures with poor EMC, i.e., 2, 3 and 1 all are negative and
high, thus showing higher reductions in relative viscosities measured with Sliper than
those with Viscometer. Thus, it is evident that among the concretes under consideration,
the behaviours of Mixtures 2 and 3 followed by Mixture 1 were significantly influenced by
a prominent LL. Mixture 1 showed a somewhat better EMC than Mixture 3 even though
it had the same (w/c)eq . This can be explained by the fact that the yield stress of M1
was with 245 P a among the lowest of all tested concretes, and much lower than 450 P a
and 533 P a measured for M2 and M3, respectively. The low yield stress of M1 appears
to have resulted in the partial shearing of the plug during Sliper testing; see [18, 189].
In other words, here the shear of the plug affects the measured pressure in addition to
the deformation of the LL. Thus, since ISPM considers only the shear of concrete bulk,
M1 yielded better EMC in comparison to M2 and M3. This view is also supported by the
best-fit results presented in Section 5.5.2. Obviously, the applicability of ISPM is strongly
limited due to its not considering the LL. To overcome this limitation, ISPM was improved
by implementing a user-defined function as described in the next section.
Table 5.2: Experimental results from Sliper and Viscometer tests (adapted from [10]),
showing relative viscosities and discrepancy.
Mixture-Nr Sliper Viscometer Sliper Viscometer Discrepancy
b µ Rb Rµ ∆
[P a·s/mm] [P a·s] [P a·s/mm] [P a·s] [%]
Mixture 1 0.59 76 0.51 0.60 -17.57
Mixture 2 0.14 38 0.12 0.30 -147.73
Mixture 3 0.81 230 0.70 1.83 -159.16
Mixture 5 2.30 236 2.00 1.87 6.35
Mixture 7 1.74 164 1.51 1.30 13.98
Mixture 10 1.15 126 1.00 1.00 0.00
Mixture 11 3.10 291 2.70 2.31 14.32

106
Concrete transport and pumping 5.5. User-defined single-phase model

5.5 User-defined single-phase model

A single-phase model with user-defined functions, hereinafter UDFM, enables the
incorporation of the LL’s influence into the initial single-phase model. Even though it
is possible to model the LL and plug as separate phases using a multi-phase model,
such an approach demands relatively large amounts of computational resources, and
so it increases model complexity and, accordingly, computational time. In contrast, the
single-phase model calculation with UDF took approximately the same time as the
simulations using the initial single-phase model.

5.5.1 Modelling schema

The LL forming during the flow of concrete in a pipe is assumed to be composed of
cement, water, admixtures and fine aggregates; therefore, it can be broadly considered to
be fine mortar. The absence of coarse aggregates in the fine mortar layer reduces its yield
stress τ0 and plastic viscosity µ drastically. From the perspective of numerical modelling,
this means that the fluid elements (cells) near the wall have lower yield stress τ0 and
plastic viscosity µ than more distant elements. To assign varying rheological properties
appropriately across the cross-section of Sliper, the numerical model must identify
the positions of all elements in a specified region near the pipe wall and assign the
corresponding properties. This task was accomplished by implementing “user defined
functions” (UDF) in ANSYS Fluent controlling τ0 and µ values of the “fluid elements”
based on their distance from the pipe wall.
The UDF approach improves model performance by adding four parameters to the
initial model: layer thickness TLL , yield stress factor τf , viscosity factor µf and gradient
factor; see Figure 5.7. The yield stress and viscosity factors are the ratios of yield
stress and plastic viscosities of mortar to the same values for concrete. Note that in
the literature the terms “dimensionless viscosity” and “dimensionless yield stresses”
or “relative viscosity” and “relative yield stress” have been used previously [113, 149].
However, these terms were introduced as the ratio of higher viscous material to the lower
viscous material, while the present case is the inverse, hence the choice of the new terms.
TLL defines the region in which the properties of fine mortar should be applied. Defining
two different sets of material properties in the LL domain on one hand and plug domain
on the other implies an abrupt local variation of the model parameters, causing numerical
instability. Hence, the rheological properties over the TLL are varied with a gradient,
having a lower limit at the pipe wall and an upper limit (equal to plug properties) at the
inner boundary of the LL. Feys et al. [270] observed that the flow of self-consolidating
concrete (SCC) during pumping can be described as plug flow with a uniform velocity
profile and a LL with large velocity gradient. In addition, the results from Jo et al. [178]
and Choi et al. [187] affirmed the existence of a large velocity gradient in the vicinity of
the pipe wall, thus substantiating the approach chosen for UDFM. It is possible to vary
properties within the LL using linear, exponential or any other mathematical function.
For the current model, a linear function was chosen.
Determining the thickness and rheological properties of the LL is not trivial. Ex-
perimental investigation of LL requires special equipment such as Ultrasonic Velocity
Profiler [181] and/or tribometer [189]. Such an examination has not been a part of the
preceding experimental work [10]. Therefore, a theoretical approach for the estimation of
UDFM model parameters τf and µf was followed. There are numerous investigations on
predicting yield stress and plastic viscosity of a suspension based on the corresponding

107
Concrete transport and pumping 5.5. User-defined single-phase model

values of the suspending fluid [115, 271–275].

v
τc (Φ) 1−Φ
t
=  (5.17)
τc (0) 1 − ΦΦ
2.5Φm
m

1 4

2 5

Figure 5.7: Sketch of modelled plug and lubricating layers in UDFM including model
parameters. 1 Plug, 2 LL region, 3 TLL , 4 τ0 , µ and 5 τ0,i = τf · τ0 , µi = µf · µ.
Mahaut et al. [113, 114] reported that the yield stress of suspensions depends only
on coarse particle concentration (suspending phase) and yield stress of the carrying
fluid and is independent of physicochemical interactions. They observed that elas-
tic modulus-to-concentration relationship followed the Krieger-Dougherty Law and
was in agreement with Chateau et al. [274], deriving Equation (5.17). According to
Krieger-Dougherty [271], the viscosity of suspensions can be related to the coarse particle
concentration and viscosity of fluid matrix with the help of Equation (5.18).

µc (Φ) 1
= (5.18)
µc (0) [η]Φm
1 − ΦΦ
m

In these models, τc (Φ) and µc (Φ) are the yield stress and viscosity of the suspension
with particle concentration (packing fraction) of Φ, τc (0) and µc (0) are the yield stress and
plastic viscosity of the suspension with particle concentration of zero, Φm is the maximum
possible packing fraction, and [η] is the intrinsic viscosity [187, 276].
Equation (5.17) and Equation (5.18) are valid for monodisperse suspensions. For
multidisperse suspensions such as concretes, as reported in [187], the Farris [272] model
should be employed:
![ηa ]Φa,m ![ηb ]Φb,m
µc (Φ) Φa Φb
= 1− 1− (5.19)
µc (0) Φa,m Φb,m

where indices and b indicate two phases in the suspension, e.g., fine sand and coarse
sand in concrete.
Following the above models, Jo et al. [178] deduced for concrete Equation (5.20) and
Equation (5.21):
v
1 − Φgravel
v v
τco (Φ) 1 − Φcement 1 − Φsand
t t u
u
u
t
=  (5.20)
τi (0) 1 − ΦΦcement
2.5Φcement,m 
1 − ΦΦsand
2.5Φsand,m 
Φgravel 2.5Φgravel,m

cement,m sand,m 1− Φgravel,m
!−2.5Φcement,m !−2.5Φsand,m !−2.5Φgravel,m
µco (Φ) Φ Φ Φgravel
= 1 − cement 1 − sand 1− (5.21)
µi (0) Φcement,m Φsand,m Φgravel,m

108
Concrete transport and pumping 5.5. User-defined single-phase model

where Φcement , Φsand and Φgravel are the concentrations of cement, sand and gravel in
cement paste, mortar, and fresh concrete, respectively; Φcement,m , Φsand,m and Φgravel,m are
the corresponding maximum concentrations; τco (0) and µco (0) are the plastic viscosity
and yield stress of concrete with solid particle concentration Φ; τi (0) and µi (0) are the
plastic viscosity and yield stress of concrete with zero solid particle concentration.
Within the constraints of the work at hand, upon implementation of LL through UDF,
a parametric study was carried out by varying one model parameter while the other two
were kept constant. The thickness of the LL (TLL ) was varied from 1 mm to 5 mm based
on literature and reported expert opinions [178, 187]. The viscosity factor µf was varied
from 0.001 to 0.8, while yield stress factor τf was varied for a few mixtures from 0.08
to 0.8 and kept constant at τf = 0.2 for the other mixtures; see Section 5.5.2. Finally,
the µf values at which numerical simulations fitted the experimental results best were
compared with µf values calculated using Equation (5.18) and considering the LL to be
equivalent to the representative mortar. It is noteworthy that the rheological properties
of the lubrication layer in pipe flow may not be necessarily equal to those of constituting
mortar belonging to the corresponding concrete. The validity of this condition depends
upon the concrete composition and flow characteristics. For flows with a very thin LL, the
constituting of fine mortar consisting of cement paste and very fine sand is apparently
realistic. However, in this first study, for sake of simplicity, the properties of the LL were
assumed to be equivalent to those of constituting mortars.

5.5.2 UDFM results and discussion

Though ISPM yielded good correlations for mixtures with low w/b, it does not imply the
absence of LL during the pumping of these mixtures. In reality, the existence of LL for
all mixtures is very probable. However, the crucial point to notice is the thickness and
properties of this layer as well as flow characteristics, i.e., plug flow or plug-plus-shear
flow. Kaplan et al. [18] suggested that concrete flows as slip plus shear flow after the
interface shear stress τi exceeds the yield stress τ0 of concrete. Even in case of low (w/c)eq ,
common for SCC, presence of LL with partial shearing of plug region is apparent. Feys et
al. [270] suggested that bulk concrete is partially sheared during the pumping of SCC due
to very low yield stress. Research carried out at the TU Dresden [189, 190] independent
of this work ascertained this phenomenon very clearly with SCC compositions having
(w/c)eq = 0.29 resulting in shear of the bulk concrete. Nonetheless, if the thickness of
the layer is small for the tested mixture or the difference in plug and LL properties is
relatively low, then the simulation results obtained without implementing LL (ISPM)
correlate reasonably well with the measured results.
Disregarding the good correlation of a few mixtures in cases of ISPM, all the mixtures
under investigation were simulated using UDFM. In the first step, a sensitivity analysis
for the model parameters was conducted, documenting the high influence of µf on the
calculated pressure. Figure 5.8a illustrates the comparative results for the calculated
pressure with viscosity factors varying from 0.001 to 0.2, while layer thickness and yield
stress factor were kept constant at 5 mm and 0.2, respectively. It was also observed that
the increment in the viscosity factor can cause drastic drops of the pressure at higher
discharge rates Q and lower pressure drops at lower discharge rates Q. Similarly, layer
thickness TLL also showed considerable influence on the calculated results. Figure 5.8b
gives the comparative results of measured and calculated pressure values for layer
thicknesses of 3, 4 and 5 mm.
In contrast, τf variation yielded minimal influence on calculated pressures. Figure 5.9
shows the comparative results for yield stress factors of 0.08, 0.2 and 0.8 with layer

109
Concrete transport and pumping 5.5. User-defined single-phase model

thickness of 5 mm and viscosity factors of 0.001 (Figure 5.9a) and 0.02 (Figure 5.9b).
Similar low influences of yield stress on rheological behaviour of concrete have been
generally reported, especially in cases of high shear flows where the final shape of the
fluid body is not of primary concern [178, 196, 203]. Here it may probably originate
from the flow characteristics of concrete in the pipe. Prior to a Sliper experiment the
formation of the LL is ensured with help of pre-strokes [10]. During the progress of a
Sliper experiment, due to the presence of LL and ever-increasing pipe velocity, shear rates
in the vicinity of the pipe wall are relatively high. From the fundamentals of the Bingham
model, it is clear that at higher shear rates the influence of yield stress is relatively
low since the plastic viscosity tends towards dynamic viscosity; see also Figure 5.4a.
Considering the low influence of τf on calculated pressures, its value was set constant
at 0.2 for all the subsequent simulations.
35 30
Mix. 3 Mix. 3
30 25
Pressure P [kPa]

Pressure P [kPa]
25 𝜏𝑓 = 0.2,𝑇𝐿𝐿 = 5 mm 𝜏𝑓 = 0.2, = 0.001
20
20
15
15
0.2 10 3 mm
10 0.067
5 4 mm
5 0.02
0.001 5 mm
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Flow rate Q [m /h]
3 Flow rate Q [m /h]
3

a) b)
Figure 5.8: Results of simulations using UDFM; comparison of results calculated a) with
different viscosity factors (layer thickness 5 mm, yield stress factor 0.2), b) with different layer
thicknesses of 3, 4 and 5 mm (yield stress factor 0.2, viscosity factor 0.001).
Table 5.3 presents the final best-fit parameters from the numerical simulations as well
as µf values calculated using Equation (5.18). It is noteworthy that Mixtures 1 and 5 con-
tained round natural quartz aggregates while all the other mixtures contained crushed
basalt aggregates. The intrinsic viscosity [η] for the round and crushed aggregates was
assumed, broadly based on [187], to be 2.5 and 5, respectively. For Mixtures 2, 5 10 and
11, best-fit values matched positively with viscosity factors calculated/predicted using
Equation (5.18); for Mixture 3 the match was very close. The only divergent case in
these results was Mixture 7 for reasons yet to be determined. Mixture 1 in a peculiar
case showed a positive EMC for three different TLL , viz. 5 mm, 4 mm and 1 mm, each
with different µf , viz. 0.50, 0.40 and 0.12. The ambiguous results are rather another
upshot of slip plus shear flow that occurs in case of low yield stress concrete; see
Section 5.4. Nevertheless, acceptable agreement of best-fit values and predicted viscosity
factors makes it possible to estimate µf parameters roughly for further studies, largely
independent of empirical knowledge. In addition, by assigning calculated τf and µf
values, one can determine TLL for a particular composition if experimental pressure and
flow rate results are available. Also, input parameters of UDFM, that Equation (5.18)
does not indicate, are the thickness of LL and the gradient of LL properties. However,
as repeatedly proven, once formed, the thickness of the LL remains constant throughout
the pumping period [18, 23, 24, 189] and the dependency of TLL on concrete composition
is becoming more and more clear [178, 189, 190, 270, 277]. Furthermore, the gradient
of LL properties and, as a consequence, velocity profiles in pipe flow could also be
experimentally deduced or assumed based on the literature [178, 181].

110
Concrete transport and pumping 5.5. User-defined single-phase model

Nonetheless, the dependence of any suspension’s rheological properties on their

suspending fluid is influenced by the
a) Maximum packing fraction of the particles;
b) Packing fraction of particles or paste volume in the concrete;
c) Particles, shape, size distribution and size range: minimum and maximum diame-
ter;
d) Intrinsic viscosity of particles.
Furthermore, in the case of applying particle concentration models for Sliper model,
TLL has also to be considered. Considering the complexity of this phenomenon, the
findings presented in this study should be taken as preliminary observations on the
applicability of particle concentration models for determining properties of the LL.
Further studies are essential to explain how the parameters influence the applicability
of such models in determining the properties of the LL.
20 25
Mix. 3 Mix. 3
20
Pressure P [kPa]

Pressure P [kPa]
15 𝜇𝑓 = 0.001,𝑇𝐿𝐿 = 5 mm 𝜇𝑓 = 0.02,𝑇𝐿𝐿 = 5 mm
15
10
10
0.8
5 0.8
0.2 5
0.08 0.2
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Flow rate Q [m3/h] Flow rate Q [m3/h]
a) b)
Figure 5.9: Results of UDFM simulations for different yield stress factors carried out with
layer thickness 5 mm and viscosity factor a) 0.001 and b) 0.02.
Figure 5.10 brings together the results of experimental measurements and of ISPM
and UDFM simulations. The results of numerical simulation using the calibrated UDFM
model showed very good agreement with experimental results for both mixtures. Similar
results were obtained for other mixtures. The layer thickness in the Sliper experiments
appears to vary from 1 mm for mixtures with low water content (low (w/c)eq ) to 5 mm
for mixtures with high water content, in agreement with earlier research findings [18,
22]. The yield stress of LL is 5 times lower than that of concrete based on assumed τf .
The plastic viscosity factor, which is the ratio of plastic viscosities of LL to the same of
concrete, varied from 1/1.25 to 1/20. For Mixtures 1, 2, 10, and 11 the plastic viscosity of
the LL is approximately 15 times lower than that of concrete; note the different TLL .
During pipe flow, the material shears as long as its yield stress is overcome by the shear
stresses. Consequently, the sheared region in case of low yield stress and low viscous
concretes is larger/thicker than that of less flowable concretes. In the context of UDFM
model, this sheared zone is represented by the LL zone which has lower yield stress
and plastic viscosity than the bulk. Therefore, the mixtures with low water content (low
(w/c)eq ), which also exhibit higher yield stress and plastic viscosity [10] should fit better
with lower TLL values such as 1 mm, indicating very moderate LL formation.
When looked more specifically, Mixture 2 with a (w/c)eq of 0.6 is very flowable and also
has very low plastic viscosity measured by means of Viscometer [10]. Consequently, the
pressures P measured as a function of flowrate Q for Mixture 2 are significantly lower
than all other tested compositions. The (w/c)eq of 0.6 and higher paste content of Mixture
2 ensured a thicker shearing zone with low plastic viscosity in the Sliper experiments, as

111
Concrete transport and pumping 5.5. User-defined single-phase model

demonstrated by viscosity parameter b, which was equal to 0.14 P a·s/mm; see Table 5.2
in [10]. Thus, irrespective of the lower µ for the plug, high TLL and lower plastic viscosity
are to be assigned to the LL for simulating high (w/c)eq mixtures.
Meas. ISPM UDFM Meas. ISPM UDFM

Mix. 10 Mix. 3
30
40
25
Pressure P [kPa]

Pressure P [kPa]
20 30

15 20
10
10
5
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Flow rate Q [m /h]
3 Flow rate Q [m /h]
3

a) b)
Figure 5.10: Experimental and numerical results for a) Mixture 10 having a low water
content and b) Mixture 3 with high water content.

Table 5.3: Viscosity factors µf calculated using Equation (5.18) and best-fit values of
simulations when compared to experiments.
Mixture µf Best fit
1 0.10 0.50/0.40/0.12 for TLL = 5/4/1 mm
2 0.07 0.07 for TLL = 5 mm
3 0.05 0.02 for TLL = 5 mm
5 0.11 0.07/0.10 for TLL = 1/2 mm
7 0.07 0.80 for TLL = 1 mm
10 0.07 0.07 for TLL = 1 mm
11 0.07 0.07 for TLL = 1 mm
In contrast, mixtures with the lowest (w/c)eq of 0.3 seem to develop a very thin LL. The
mixtures 5, 7, 10 and 11, all with lower (w/c)eq 0.3 exhibited the best-fit TLL values of
1 mm. For Mixture 5 best-fit µf is 1/10 to 1/14, depending on layer thickness. Mixture 5
has a (w/c)eq of 0.3 and is without any secondary cementitious materials; it has the lowest
water content among all mixtures having (w/c)eq of 0.3. For Mixtures 1 and 5, the plastic
viscosity of the LL is approximately 10 times lower than plastic viscosity of concrete. For
the best fit of Mixture 3, a 50 times lower plastic viscosity of the LL than that of the
plug was determined. At first glance, this may lead to false conclusions about Mixture
3’s having a 5 mm thick LL of very low µf , lower than that of Mixture 2. However, in
the experimental investigations, it was observed that in the case of Mixture 3, due to a
potential experimental error in the Viscometer tests, the torque-rotational speed curve
T-N did not fit well with the P -Q curves of the Sliper tests. While for all other mixtures
T-N curves correlated well with P-Q curves, the T-N curve measured for Mixture 3 was
very steep, leading to a much higher plastic viscosity measurement from the Viscometer
[10]. Such a high plastic viscosity, when used as input for the numerical model, leads to

112
Concrete transport and pumping 5.6. Chapter summary and conclusions

the prediction of overly high Sliper pressures. Consequently, best fit values for TLL and
µf have to be high and very low, respectively.

5.6 Chapter summary and conclusions

For large-scale implementation of DC, pumping of concrete is an essential part. From this
chapter, following conclusions could be drawn, with respect to pumping in the context
of DC:
a) The placement process for 3PCs can vary widely both in terms of technological
aspects and rheological requirements. There has been no published research on this
aspect.
b) In this work, the various 3PC placement scenarios were analysed theoretically. The
approach d) where a PCA is mixed inline on the printhead was identified as the
most appropriate for DC.
c) Due to the relatively high τµ0 the influence of yield stress is expected to be more
significant for pumpability in case of 3PCs, when compared to, e.g., SCC.
d) Thixotropic effects raise concerns and require urgent research on pumpability of
3PCs.
With respect to virtual Sliper:
a) A numerical tool to simulate Sliper tests — virtual Sliper — was developed
using the single fluid CFD method, short ISPM. ISPM predicted the general trend
of pressure variation with respect to the discharge rate correctly and provided
acceptable agreement with experimental results in the case of concretes with low
water-to-binder ratios.
b) However, since the lubrication layer LL was not considered, the applicability of
ISPM was limited to the scenarios in which the formation of LL was not pronounced.
c) An improved single-phase model with User Defined Function (UDF) was devel-
oped, which can model the lubricating layer. This approach also makes possible the
estimation of the thickness, yield stress and plastic viscosity of the LL for different
concrete compositions.
d) A parameter study performed using the UDF model ascertained that both the
thickness and plastic viscosity of the LL have pronounced influences on concrete
pumpability, while the yield stress parameter showed a negligible effect. Addition-
ally, the best-fit results suggest that the assumption of a yield stress factor parameter
equal to 1/5 is appropriate, while plastic viscosity factor and thickness parameters
varied within the ranges from 1/1.25 to 1/20 and from 1 to 5 mm, respectively.
e) The parameter study and the subsequent empirical fitting of the calculated results
indicate the validity of calculating the viscosity of the lubricating layer as a function
of concrete viscosity using the Krieger-Dougherty model.
The developed virtual pumpability testing tool should enable both a more purposeful
material design of pumpable concrete and estimation of pumping processes.

113
Chapter 6

EXTRUDABILITY

In this chapter, the third major phase in processing 3PCs — the extrusion — is
addressed; see also Section 2.5. Recognizing the absence of experimental methods for
process-specific extrudability characterisation the chapter focuses on test methods. At
first, intricacies and required attributes for extrudability test methodology for 3PCs
are described. Following this, a new methodology — the 3D-printer-extrudability-test
(3DPET) — is proposed for characterising the extrudability of 3PCs, both quantitatively
and inline; see Section 6.2. Furthermore, the results obtained using the proposed
approach are compared with the results of a “simple” ram extruder, Hägermann flow
table test and Viscometer tests; see Section 6.4.3.

6.1 Attributes of an optimal extrudability testing method-

ology
An optimal test method for characterising extrudability in the context of DC shall possess
following attributes:
A1) quantitative,
A2) inline and continuous with high potential for complete automation,
A3) capable to quantify material (extrudate) flow rate,
A4) geometry independent.
On the backdrop of the absence of formwork, conflicting material requirements, and
minimal human involvement, the ease and accuracy of extrusion is very crucial to ensure
reliable DC process and consistent geometry of printed elements. For example, consider a
layer with dimensions width wlayer , height hlayer and length llayer , which is a segment of a
larger wall with width W , height H and length L being 3D-printed with a linear velocity
of llayer [m/s]. Then, the printhead must extrude material at a flow rate of wlayer ·hlayer ·llayer
[m3 /s]. This volume flow rate must be ensured independent of
a) transient rheological properties of concrete such as yield stress and plastic viscosity;
b) changes in the printhead velocity and acceleration, e.g., when it changes the
direction of movement;
c) change in the efficiency of concrete pump at various pumping velocities; which can
be due to motor limitations and shear rate dependent changes in the rheological
behaviour of concrete flowing in a pipe (e.g., the wall friction [228] and the
formation of a lubricating layer in a pipe can vary depending on the shear rate
with which concrete is sheared [18]). Also; see Figure 6.4 for non-linear dependency
between the discharge rate of a progressive cavity pump (PCP) and differential
pressure.

115
Extrudability 6.1. Attributes of an optimal testing methodology

This challenge becomes even more complex if the cross-section of the target wall
geometry is not uniform over the entire length of the layer; a common possibility
considering complex geometries that are anticipated to be 3D-printed. In such cases, the
extrudate flow rate should be varied precisely as the printing process progresses.
To ensure blockage-free extrusion, qualitative methodologies may be sufficient, at least
to some extent. However, to achieve high precision in terms of extrudate flow rate, the
extrudability of applied material must be tested with a method which is quantitative
(attribute A1); a simple pass or fail test of extrudability would not be adequate. For the
reasons detailed above, it would be highly advantageous if a single test would quantify
both extrudability and exact extrudate flow rate (attribute A3) at the nozzle outlet of the
3D-printer for various control parameters; see the proposed method in Section 6.2.
The required attribute A2 “inline, continuous . . .” finds its basis in the autonomous
nature of DC. Offline test methods such as ram extruder test (RET) or squeeze-flow test
are carried out by taking concrete out and away from the DC process. Such approaches
cannot directly take into account the process-induced variation of material rheological
properties. For example, in the case of pumping SCC, differences in values of yield stress
and plastic viscosity were recorded before and after pumping [208]. The variation in
rheological properties may result from the “higher shear rates leading to the dispersion
of cement particles and depending on available residual superplasticizer in the mixing
water” [182] and due to the influence of pressure on rheological properties [209, 210];
See Section 2.4.5 for more details on process-induced changes in rheological properties.
So far, progressive cavity pumps or simple screw pumps are used for extrusion of concrete
in 3D-printing. In such cases, concrete undergoes high shear rates and is subjected to high
pressure in the extruder. Therefore, the exact rheological behaviour of the extrudate may
vary depending on the specific extruder and printing circuit (mixing → transporting
extrusion).
When large-scale, on-site applications are realized, a pronounced influence of process-
ing on the material rheology must be expected. Due to these reasons, characterisation of
extrudability with offline test methods may lead to erroneous results. This issue could be
solved to some extent by carrying out extensive experimental studies and by fitting the
offline extrudability measurements with effective extrudability in a 3D-printer. To the
best of the author’s knowledge, no such studies have been reported yet. Even then, human
involvement would be necessary to conduct offline experiments. Again, considering
the time and shear history dependent variation of concrete rheological properties, the
extrudability tests must be carried out at numerous concrete ages (form water addition).
Therefore, an inline test method that is directly integrated into 3D-printing process-chain
is of high research and industrial significance in the context of DC; since such test method
a) takes process-induced material changes into account,
b) eliminates the necessity to have a separate test device and,
c) drastically minimizes human involvement.
The anticipated test can be autonomously performed and can be carried out continuously
throughout the 3D-printing process.
The attribute A4 “geometry independence” is a well-known requirement in the
rheological characterisation of concretes and in material testing in general. Extrudability,
being a compound engineering property, is dependent on fundamental rheological
properties as well as process parameters such as extruder type, geometry, geometry of
nozzle-outlet, etc. For quantifying the “true” material behaviour, the test method shall
quantify the extrudability as a “geometry independent” quantity. This would enable
process agnostic characterisation of material extrudability, i.e., a material characterised
“extrudable” remains the same whichever 3D-printer is used with whichever nozzle ge-

116
Extrudability 6.2. Proposed method for extrudability characterisation

ometries and motor efficiencies. The geometric independency can be achieved with help
of analytical models and transformation equations. In rotational rheometry, geometry
independent (at least in the theory) values of yield stress and plastic viscosities are
obtained from geometry-dependent measurements of torque and rotational velocities,
using Reiner-Riwlin or extended Reiner-Riwlin transformation equations [179]. Also in
ram extrusion, there are some ways to take  geometry influence into account; for example,
diameter of barrel
the extrusion constant diameter of orif ice and terms representing length and shape of
die-land area are often used in the analytical modelling of ram extrusion force [214].
That being said, the transferability of extrudability measurements from ram extrusion
tests to other types of extruders is yet to be investigated. An alternative approach to
overcome artefacts due to the geometry influence is characterising extrudability with the
same geometry that will be used for 3D-printing; see Section 6.2. The limitations of test
methods previously used for characterising extrudability are outlined in Section 2.5.4.
Most importantly, none of the reported studies has focused on the direct quantification
of material flow rate from an extruder similar to that used in target DC application.
As detailed above, the quantification is crucial to the synchronization of printhead
velocity and concrete pump discharge rate. Thus, new approaches are needed for reliable
assessment of the extrudability of cement-based materials in the context of DC.

6.2 Proposed method for extrudability characterisation

The proposed method quantifies the energy required for extrusion of concrete at a
specified flow rate. The corresponding equipment, the 3D-printing test device, consists
of a Progressive Cavity Pump (PCP) operated by an asynchronous motor; see Figure 3.2.
The flow rate of the extrudate is proportional to the PCP’s rotational velocity, which
depends on the set rotational speed of the motor, the operating point of the motor, and
the gear ratio. The operating speed of the motor for the 3DPTD used in this research
was controlled using custom-built software. The quantities measured in the experiments
are electric power consumption Pe and extrudate flow rate Qe , as measured at different
rotational velocities of the PCP.
3DPTD

Power measuring Extrudate

Electric supply Motor PCP
unit collection

Computer Manual flowrate

Control Data measurement
unit processing
Extrudability

Figure 6.1: Schematic view of the extrudability measurement setup using 3DPTD.
Total electrical power consumed by the concrete pump of 3DPTD while extruding
concrete at a predefined rotational speed (controlled through RCP1) was measured.
For this purpose, a power measuring device was installed between the power input
of concrete pump and the computer-controlled frequency converter; see the schema in
Figure 6.1. Average power Pe , in [W ], based on the values recorded over the period of
measurement, excluding outliers, was calculated and plotted as a function of RCP 1;
see Figure 6.3. In addition, the extruded material was collected for a duration of 90
[s] and its weight was measured on a scale away from the printer. The flow rate of
the extrudate was then determined by converting weight into volume using extruded

117
Extrudability 6.3. Experimental program

concrete density. The rotational velocity of PCP of the 3D-printing test device was
controlled according to the synchronization-schema detailed in Section 3.2.1; see also
Equations (3.4) and (3.5) The experimental results collected on power and flow rate
were used in calculating the extrudability index U EE [J/cm3 ], which was calculated
by dividing average power consumption with flow rate. Therefore, the lower the U EE
of material under investigation, the higher is its extrudability for the extruder setup
utilized.
The extrudability parameters from 3DPET as presented in this thesis may not be
“transferable” to a different 3D-printer setup. However, the core idea of the proposed
approach (measurements of electric power consumption and flow rate) is straightforward
and can be implemented for any type of 3D-printer/extruder. Further development of
the proposed approach, with supporting theoretical framework, may eventually enable
geometry independent characterisation of the extrudability of 3D-printable concretes.

6.3 Experimental program

The primary objective was to validate the proposed approach while investigating its
sensitivity for variation in extrudability with varying concrete composition.

6.3.1 Mixtures and materials

Two 3PCs were investigated, both having a mixture of Portland cement, microsilica
and fly ash as binder, and an equivalent water-to-cement ratio w/ceq of 0.42, calculated
according to EN 206-1; see also Table 3.1. The reference mixture A is a high-strength,
printable fine-grained concrete which has been proven to be consistently extrudable
and buildable. The paste volume was 52.5 %. Fine quartz sand (Fraction 1: 0.06-0.2
mm) and natural river sand (Fraction 2: 0-1 mm and Fraction 3: 0-2 mm) were used as
aggregates. The aggregate fractions were added in the ratios of 0.2:0.2:0.6 for Fractions
1:2:3, respectively; see also Table 3.1. The second mixture, A-HF, was formulated by
replacing Fraction 2 and Fraction 3 in Mixture A with Fraction 1. The particle size
distribution of aggregates in both A and A-HF are presented in Figure 3.1. Due to its
increased fineness and dry surface-to-fluid ratio, Mixture A-HF was stiffer than Mixture
A. To ensure successful extrusion of A-HF, the content of PCE superplasticizer in A-HF
was increased to 2.6 %bmob from 2.0 %bmob in A.

6.3.2 Production of mixtures and testing procedure

Experiments were conducted in two series. In the first series, concrete was produced in
50 litres batches with an Elba mixer, and subsequently, 3DPET tests were conducted. In
the second series 10 litres of concrete were produced with a Hobart mixer, followed by
RET, Viscometer and flow table tests. Table 6.1 gives details on the mixing procedures.
The 3DPET procedure is presented in Table 6.2. After mixing, a portion of concrete
was loaded into the 3DPTD hopper, and the rest was covered with a polymer foil
to prevent water evaporation. By re-filling the hopper promptly, care was taken to
keep the level of concrete in the hopper steady for all the measurements. The actual
extrudability measurement involves extruding the concrete at a fixed rotational control
parameter RCP1 value for 90 s for each individual measurement. Prior to each extrusion
measurement, concrete was extruded without recording for 30 s to ensure “filling” of
the 500 mm long pipe connecting the PCP with the nozzle. The extruded concrete was
collected into buckets which were, according to the procedure, moved under and away

118
Extrudability 6.3. Experimental program

from the nozzle at measurement times of 0 and 90 s. The extruded concrete was then
compacted, the weight and volume of the extrudates were measured. Power consumption
was recorded during the extrusion. The entire measurement process was carried out for
the RCP1 values of 0.3 0.5, 0.6, 0.7 and 0.9.
Table 6.1: Mixing procedure.
Time [min : s] Production steps Speed [rpm]
Elba Hobart
-3:00 - 0:00 Homogenizing dry materials 25 139
-1:00 - 0:00 Mixing water and SP; scrape walls 0 0
0:00 Add liquids to dry mix including MSS 0 0
0:00 - 1:00 Mixing 25 139
1:00 - 3:00 Mixing 45 591
3:00 - 4:00 Pause, scrape walls of the mixer bowl 0 0
4:00 - 6:00 Further mixing 45 591

6.3.3 Additional tests for comparative assessment

One of the objectives of this research was to evaluate the adequacy of conventional
methods for characterising extrudability. Considering that the ram extruder approach
has been used widely for characterising extrudability [169, 214, 234], this method was
applied here as well. The ram extrusion test (RET) device described in Section 3.3 and
illustrated in Figure 3.3 was used for this purpose. Although the used RET test device still
needs further optimization1 , for the comparative purpose of this section it was certainly
adequate; see Section 6.4.3. For comparative analysis, the variation of REF for the tested
mixtures is compared with the variation of U EE measurements from the proposed inline
experimental method. Extrudability as assessed by the ram extrusion tests can also be
quantified as average extrusion force per unit extrusion: unit ram extrusion force REFu
[N /cm3 ]. However, since the experiments conducted in this research were carried out
at constant effective ram velocity, i.e., at a constant flow rate, comparing ram extrusion
forces REF [N ] is sufficient.
In addition to RET, rheological tests were performed with the finite length cylinder
Couette type rotational Viscometer ConTec 5; see Figure 3.5 [144, 179]. For the in-
vestigated 3PCs, hysteresis loop tests were performed at 20 min after water addition;
see Figure 6.2. From the measured torque T and rotational velocity N , the parameters
describing the Bingham yield stress τ0 and plastic viscosity µ, were computed by means of
the Reiner-Riwlin equations and using linear regression; see Equations (3.6) and (5.13).
Further, the Hägermann flow table (HFT) test was also carried out as an empirical
indicator of concrete flow behaviour.
1 Ability to directly measure frictional resistance and using metal barrel, for example.

119
Extrudability 6.4. Results and discussion

Table 6.2: Measuring protocol.

Step No. Activity Relative time [min : s]
1 Production of concrete 00:00 - 10:00
2 Flow table test 15:00 - 20:00
3 Extrusion at RCP1 0.3 20:00 - 21:30
4 Height and weight measurement 22:00 - 23:30
5 Extrusion at RCP1 0.5
6 Height and weight measurement 24:00 - 25:30
7 Extrusion at RCP1 0.6
8 Height and weight measurement 26:00 - 27:30
9 Extrusion at RCP1 0.7
10 Height and weight measurement 28:00 - 29:30
11 Extrusion at RCP1 0.9
12 Height and weight measurement 30:00 - 31:30

1.0
Rotational velocity N [rps]

1.8
Torsional Moment [Nm]

0.8
Torque T [N. m]

1.6
0.6
1.4
0.4
0.2 1.2

0.0 1.0
0 20 40 60 80 100 120 0.0 0.2 0.4 0.6 0.8 1.0
Test duration td [rps] Rotational velocity N [rps]
a) b)
Figure 6.2: a) Hysteresis loop measurement protocol, b) measured torques as a function of the
rotational velocity of Viscometer’s outer cylinder.

6.4 Results and discussion

6.4.1 Extrudability assessment from 3DPET results
As described earlier, it is essential to measure and subsequently to steer how much
material is actually extruded at different values of RCP1 (rotational control parameter
when the printhead is not moving). This knowledge makes it feasible to adjust the control
parameters for the concrete pump according to the geometry of the target construction
element and printhead velocity. The power consumption and flow rate measurements of
Mixture A are shown in Figure 6.3. The power consumption increases almost linearly
from approximately 480 W at 0.3 RCP1 to about 890 W at 0.6 RCP1. Interestingly, at 0.9
RCP1 the power consumption decreases slightly. Similarly, the flow rate also increases
nearly linearly up to 0.6 RCP1 from 34 cm3 /s to 66 cm3 /s with an insignificant increase,
and eventually stagnation, with further increases in RCP1. The measurements on Mixture
A-HF yielded similar dependencies, however, with higher power consumption and lower
flow rate values; see also Section 6.4.3.

120
Extrudability 6.4. Results and discussion

It must be noted that the motor used to run the PCP was a 3-phase asynchronous type.
The non-linear dependence of flow rate and power consumption on the RCP 1 values, i.e.,
stagnation above certain RCP1, could have three origins:

a) Motor slip: the slip of asynchronous motor (concrete pump) can be substantial at
higher RCP1 values and/or if the load on the motor is higher than the nominal
load.
b) Control software implementation: errors in the 3D-printer software, e.g., incorrect
translation of the input control parameter (in this case RCP1) to the analogue
frequency of the motor that defined the rotational velocity of PCP.
c) Nonlinear increase in the slippage of PCP: for elastomeric PCPs (the one used in
this study), the value of the volumetric flow rate Qe,pcp decreases nonlinearly with
increasing ∆Ppcp ; see Figure 6.4.

80 1200
Flow rate [cm3/s]

70
1000

Power [W]
60
50 800
40
600
30
20 400
0 0.2 0.4 0.6 0.8 1.0 0 0.2 0.4 0.6 0.8 1.0
RCP1[-] RCP1[-]
a) b)
Figure 6.3: a) Flow rate and b) power measurements as a function of RCP1 for Mixtures A
and A-HF.
In the first two cases mentioned above, the characteristics of the frequency converter
and the motor play vital roles. The drastic increase of what appears to be slip at higher
RCP1 is probable in the case of Mixture A-HF, as the motor’s nominal power of 1.1 kW was
exceeded. However, this is less probable in the case of Mixture A. Further measurements
and analyses are necessary for a thorough explanation of this phenomenon. The third
case is an unavoidable consequence of using PCPs with elastomeric stators. Pessoa
and Paladino examined the slippage for both elastomeric (interference w > 0; see
Section 2.5.2.1) and metallic (w < 0) PCPs [219, 223]. For elastomeric PCPs the value
of w decreases with increasing internal (differential) pressure due to the increased
deformation of elastomer. In the metallic case, w is constant. Upon this background
performance curves were plotted for (theoretical) displaced volume ve,pcp and differential
pressure ∆Ppcp [219]. The ve,pcp is calculated by dividing Qpcp (see Equation (2.13)) with
rotational velocity of stator. The ve,pcp decreases non-linearly with increasing ∆Ppcp ; see
Figure 6.4.
The physical basis for the increasing slippage with differential pressure can be traced
to the decrease in w due to elastomer deformation. Thus resulted slippage can be very
pronounced if the deformation of elastomer is higher than the interference [219]. In
case of metallic PCPs slippage increases linearly with increasing ∆Ppcp . Above statements
offer a potential explanation for the decrease in the flow rate of 3PC at higher operation
velocities of the PCPs. However, in the author’s opinion, slippage of the PCPs may not be
high enough to cause a significant decrease in the measured flow rates. Rather, either a
combination of all three causes or errors in software implementation could have lead to
the observed stagnation of Pe and Qe .

121
 Extrudability 6.4. Results and discussion

Despite the above mentioned presence of a potential uncertainty, the measured flow
rate values are sufficient to decide which RCP1 value has to be set during extrusion of a
particular concrete composition in achieving a specific flow rate. Figure 6.3a shows also
that flow rate follows the same trend for both mixtures. Furthermore, with increasing
flow rate, power consumption also increases. For the quantitative evaluation of the
extrudability of different compositions, merely comparing flow rate measurements is
inadequate. Hypothetically, if two materials X and Y have the same flow rate for a certain
RCP1 value and a lower power consumption in case of X, then X is more extrudable.
Consequently, a combined parameter such as the unit extrusion energy U EE (energy
per unit volume of extruded material) which takes into account both P and Q is more
appropriate for comparative evaluation of a material’s extrudability. Figure 6.5a presents
the calculated U EE values for Mixtures A and A-HF.
displaced volume displaced volume
sli
pp
ag
e

stator
deformatoin > w

a) b)
Figure 6.4: Expected characteristic performance curves for a) elastomeric and b) metallic
PCPs. The increasing slippage with differential pressure is shown (adapted from [223]).
24
22
UEE [J/cm3]

20
18
16
14
12
0.2 0.4 0.6 0.8 1.0
RCP1 [-]
a) b) c)
Figure 6.5: a) Unit Extrusion Energy (UEE) as a function of RCP1, b) appearance of the
Mixture A-HF and c) appearance of the Mixture A (photos are taken at the time of printing).
Mixture A has a significantly lower U EE in comparison to A-HF at all RCP1 values,
showing A to be the more extrudable material. Average electric power consumption for
extruding unit volume of Mixture A, of all RCP1 measurements, is 13.57 J/cm3 while that
for Mixture A-HF is 21.93 J/cm3 . Mixture A-HF, obtained from Mixture A by replacing
coarser sand with a finer one, is characterised by a higher fine material content and
solid-to-fluid volume ratio, both of which causes higher water demand of the solid
constituents. Therefore, Mixture A-HF was considerably stiffer in comparison to Mixture
A; see also Figure 6.5 and Figure 6.8b.

122
Extrudability 6.4. Results and discussion

6.4.2 Extrusion force measurements from ram extruder tests

The raw data of ram extrusion force (REF) measured at various ram displacements ∆r
are presented in Figure 6.6. The easily distinguishable characteristic of the REF-∆r curves
are the three distinct sections corresponding to three different stages in the experiment, as
was also observed in earlier works on ram extrusion [169,235,278]. These three stages are:
a) pronounced, exponential increase of REF at flow initiation, b) slight, gradual decrease
in REF during the main part of the experiment and c) non-linear increase of REF near
the end of the test. According to Perrot et al. [169] these three parts can be attributed to
a) compaction/consolidation of the tested material, b) decrease in the frictional resistance
with decreasing billet length and c) compacting against static/un-sheared (so-called
“dead-zone”) material when the ram reaches the vicinity of the nozzle.
2.0

1.6
REF [kN]

1.2

0.8

0.4

0.0
0 15 30 45 60 75 90 105 120 135
Ram displacement Δr [mm]
Figure 6.6: Measured ram extrusion force (REF) as a function of ram displacement.
The REF is influenced by friction between the billet, the piston-head and the inner
walls of the cylinder. The frictional force varies depending on the test setup and can be
very different for other extruder types. To capture the intrinsic friction component in the
setup, ram “extrusion” experiments were carried out without any material just before
actual tests with concrete; see Figure 6.7a. Average REF value denoted as REFa for the
second stage of the tests without concrete is 0.017 kN . Interestingly, the first and the
third parts of REF − ∆r curves from tests without concrete are similar in trend to those
observed in the tests with concrete, indicating that a portion of the non-linear behaviour
is originated from the test setup. In the first stage, the increase in force is due to a)
compacting of the tested material and b) activation of friction resistance by the induced
ram displacement. The increase in REF in the third part of the curve could be due to
compressive stresses induced by the belt used to hold the PVC cylinder in position. Since
first and last parts of the REF − ∆r curves are biased by the experimental setup and only
partially originate from the properties of the tested material, the second stage is used
exclusively for the comparative analyses with respect to the mixtures extrudability; in
accordance with Perrot et al. [169]. Figure 6.7b presents resulting “processed” REF values
for the second part of REF − ∆r curves. Surprisingly, Mixture A-HF needed significantly
lower REF in comparison to Mixture A. This is true even though the latter exhibited a
“more plastic” consistency.

123
 Extrudability 6.4. Results and discussion

6.4.3 Comparative analyses: rheological properties ⇔ extrudability

Viscometer torque measurements at various rotational velocities are shown in Fig-
ure 6.8a. Figure 6.8b shows the results of Hägermann flow table tests; the results of all
other tests are presented in Figure 6.9 to assist comparative assessment.
The calculated values of dynamic yield stress are 306.2 P a and 642.7 P a and those
of plastic viscosity are 5.32 P a·s and 5.77 P a·s for Mixture A and Mixture A-HF,
respectively. The large difference in the yield stress is well explicable considering higher
water demand, but the same water content of Mixture A-HF in comparison to Mixture A.
Hägermann flow table test (HFT) spread diameters for A and A-HF were 104 mm and 103
mm before the strokes, respectively, and 111 mm and 109 mm after the strokes.
0.05 1.0
Air before A A
0.04 Air before A-HF 0.8
y=-0.0015x+0.938
REF [kN]

REF [kN]
0.03 0.6

0.02 0.4
A-HF
0.01 0.2 y=-0.0011x+0.347
0.00 0.0
0 30 60 90 120 0 30 60 90 120
Ram displacement Δr [mm] Ram displacement Δr [mm]
a) b)
Figure 6.7: Results of the ram extrusion tests: a) unprocessed extrusion force as a function of
ram displacement in the measurements without concrete, b) processed, the second part of the
REF − ∆r curves from measurements with concrete after subtraction of frictional resistance.
A-HF A Before strokes After strokes
3.5
140
Spread diameter [mm]

3.0
120 111 109
2.5 104 103
Torque [N m]

. . 100
2.0
80
1.5 60
1.0 40
0.5 20
0.0 0
0.0 0.2 0.4 0.6 0.8 1.0 A A-HF
Rotational velocity [rps] Mixture
a) b)
Figure 6.8: a) Torques measured in the Viscometer tests as a function of rotational velocity
obtained from downwards segment of the hysteresis loop, b) Hägermann flow table test results.
The average REF values of the second stage of the ram extrusion test, after subtracting
frictional resistance, are 0.81 kN and 0.26 kN for Mixture A and Mixture A-HF,
respectively. These results apparently contradict those of 3DPET, which shown A-HF
having higher UEE than A; see Figure 6.9. While the results from 3DPET, Viscometer,
HFT test and visual observation indicate higher energy needed to extrude Mixture A-HF,
the REF values imply the opposite; see Figure 6.9.
Though the exact origin of this contradiction cannot be proved with the data available,

124
Extrudability 6.4. Results and discussion

some hypotheses can be outlined. One possible cause is blockage of sand grains between
piston-head and cylinder. Blockage of sand grains is a common artefact, which can also
occur in screw-extruders. Sand grains, with diameter neither much smaller nor much
bigger than the clearance between the piston-head and inner walls of the cylinder, can
block/hinder the ram advancement, leading to higher REF values. While Mixture A-HF
had only very fine sand with maximum diameter of 0.2 mm, Mixture A had three different
sand fractions with diameter varying from 0.06 mm to 2 mm. It can be assumed that
the probability of blockage is higher in the case of Mixture A. The periodic crests and
troughs visible in REF − ∆r curve (see Figures 6.6 and 6.7) may also originate from sand
grains, which are blocked and later released. Another possible cause for the observed
inconsistency between the 3DPET and RET results is the potential influence of slippage
or lubricating layer (LL) [10, 18, 24] on the measurements from the latter. The formation
of plug-flow in a progressive cavity pump (used in 3DPET) is very unlikely. In contrast,
the concrete flow in the “barrel part” of the ram extruder is characterised as “plug-flow”
[169, 214, 234]. As a result, the bulk of concrete does not shear and slides on the thin
(lubricating) layer of “fine-paste” formed between the bulk and barrel’s inner wall.
Consequently, in the case of plug flows, total resistance from a material against flow may
be primarily influenced by the lubricating layer rather than by the non-sheared concrete
bulk. The different geometrical setups in the case of ram extruder and progressive cavity
pump play significant roles with respect to the force and energy needed for extrusion.
The shear rates and occurrence of the LL in RET and 3DPET are likely to be very
different. It is known that the formation and properties of LL depend on the composition
of concrete [25, 80]. Based on the above statements, the variation in the LL properties
among the tested mixtures would influence more the RET results than the 3DPET results;
potentially leading to different observations in different tests. The extent of LL influence
and the differences in LL characteristics such as interface viscosity and interface yield
stress, of both tested mixtures needs further experiments.
REF UEE τ0 µ
2.0 25 750 10
21.93 642 9Plastic viscosity µ [Pa.s]
1.6 20 600 8
Yield stress τ0 [Pa]

7
UEE [J/cm3]
REF [kN]

1.2 13.57 15 5.3 5.7

450 6
0.81 306 5
0.8 10 300 4
3
0.4 0.26 5 150 2
1
0.0 0 0 0
A A-HF A A-HF
Mixture Mixture
a) b)
Figure 6.9: Comparison of results from various tests on mixtures A and A-HF a) ram
extrusion force REF and unit extrusion energy U EE, b) yield stress and plastic viscosity.

125
Extrudability 6.5. Chapter summary and conclusions

6.5 Chapter summary and conclusions

The main findings of this chapter can be summarized as follows:
a) Inline and continuous monitoring of material properties is a prerequisite for
realizing completely autonomous construction technologies. Such a method is
proposed to characterise extrudability quantitatively by measuring the electric
power consumed while extruding a 3PC.
b) It was demonstrated that measuring the flow rate of the extrudate is insufficient in
characterising extrudability comparatively. Thus, unit extrusion energy U EE was
suggested as an appropriate metric of extrudability, which takes into account both
flow rate and electric power consumption during extrusion.
c) Mixture A-HF, with its very high content of finest sand, exhibited a substantially
higher U EE and thus was designated as less extrudable in comparison to Mixture
A, which had a sand with much higher maximum aggregate size. The increase in the
fineness of solid constituents reduces the extrudability if the water content remains
unchanged, based on the 3DPET results.
d) The proposed method can serve purposeful material development for DC and help
in determining the synchronization-control parameters to be set for 3D-printing;
see Section 3.2.1.
Comparative analyses of the applied methods and obtained results elucidate the
suitability of test devices with respect to the characterisation of extrudability.
e) There was a pronounced difference in U EE and minor difference in HFT spread
diameter for Mixtures A and A-HF. Therefore, it can be concluded that the simple
flow tests such as Hägermann flow table test are not appropriate for evaluating the
extrudability of cementitious materials for digital construction, at least not when a
PCP or screw pump is used.
f) On the other hand, Viscometer measurements shown a significant difference in
yield stress and only a minor difference in plastic viscosity for these mixtures, which
corresponds well to the difference in U EE values. Rheometers and Viscometers can
serve as indicators for variation in extrudability.
g) It is important to use a test device similar in its working principle to the device
used in full-scale application to avoid geometry-induced artefacts in extrudability
characterisation.
h) Simple ram extruder tests may be applicable if a ram extruder is to be used in the
printhead. The device seems to be prone to produce spurious results depending on
many geometrical aspects, such as aggregate-size distribution and piston-cylinder
gap.
i) The 3DPET test is promising for inline quantitative analyses of extrudability with
minimal geometry-induced artefacts. This is the only method that can give a direct
measurement on the volume flow rate of extrudate for a given control parameter.
This ability is instrumental in the seamless digital construction of objects, especially
those with varying cross-sections. If the cross-section of the target element varies,
then the flow rate of extrudate has to be adjusted correspondingly, which is only
possible if flow rate is measured.

Acknowledgements: Experimental work presented in this chapter is partially carried out

in the form of a student project work [279] under the close supervision of the author. The
work is published in peer-reviewed journal Cement and Concrete Composites [26]. All
the acknowledgements and support of the co-authors mentioned in [26] are valid for this
chapter too.

126
Chapter 7

BUILDABILITY OF 3PCS

This chapter deals with the fourth phase of processing 3PCs — buildability. A
practice-oriented approach to determine buildability test parameters is presented. The
approach takes various process aspects and construction costs into consideration. In
doing so, direct links between laboratory buildability tests and target applications
are established. The proposed approach is verified by applying it to a high-strength,
printable, fine-grained concrete.

7.1 Determining buildability test parameters considering

target application
The suggested approach is described below taking an example wall structure. From a
practical perspective, there are three primary parameters defining any buildability tests
when applied in laboratory investigations for material characterisation: 1) the height of
the wall to be printed, 2) the section geometry of each layer, thus, the total number of
layers to be printed and the aspect ratios, and 3) the time interval (TI) between subsequent
layers.

7.1.1 Suggested buildability criteria

The first step in the proposed approach is to identify and define the target application.
Thereafter, a detailed printing scenario must be established based on the building design
and project planning. This task is normally seen as part of the “design and process” aspect
of digital construction (DC) work-flow. Generally, it involves “slicing” the 3D geometry,
identifying the optimal travel route of the printhead, and determining layer breadth,
height, and contours. The methodology presented below is for full-width layer printing
(FWP). Methodology for filament printing (see Figure 2.7b) is addressed in Section 7.3.
In the case of massive FWP, layer breadth is equal to the final breadth of the wall to
be printed; see Figure 2.7a. For the convenience of further discussion in this chapter,
relevant aspect ratios are listed in Table 7.1.
If the height and breadth of the wall to be printed (part of the target structure) are Happ
and Bapp , respectively, then the aspect ratio αapp of the continuously produced “target”
element, whose buildability has to be ensured can be expressed as Equation (7.1):
Happ
αapp = (7.1)
Bapp

127
Buildability of 3PCs 7.1. Buildability criteria

Table 7.1: Aspect ratios of various elements produced by means of digital construction.
Element Variables Aspect ratio
Target wall Happ = height and Bapp = breadth αapp = Happ /Bapp
Experimental Hexp = height and Bexp = breadth αexp = Hexp /Bexp
specimen
Target layer HL,app = height and BL,app = breadth αL,app = HL,app /BL,app
Experimental HL,exp = height and BL,exp = breadth αL,exp = HL,exp /BL,exp
layer

A straightforward but often economically infeasible manner in verifying buildability

is to produce a full-scale structure of the targeted application. Alternatively, buildability
can be tested by producing a scaled-down version of the target structure in a laboratory
with an appropriate 3D-printing device. However, printing an arbitrary number of
layers with an even so arbitrary time interval will not prove a material buildable; see
Section 2.6.1. Instead, calculating the minimum height Hmin,exp of the wall to be tested in
the laboratory, using the breadth of the layers printed in laboratory experiments Bexp and
target structure’s aspect ratio αapp is more appropriate; see Equation (7.2).

Hmin,exp = Bexp · αapp (7.2)

If the height of a single printed layer in laboratory experiments is hlayer,exp , then the total
number of layers to be printed is given by Equation (7.3).
Hmin,exp
nlayers = (7.3)
hlayer,exp

Combining Equations (7.2) and (7.3) gives Equation (7.4).

αapp · Bexp
nlayers = (7.4)
hlayer,exp

When downscaling the wall geometry to a laboratory specimen, the limits given by
each particular concrete composition must be considered. Specifically, the maximum
aggregate of mixtures poses a requirement of minimum breadth and height of each single
layer. Choosing the minimum dimension of layer cross-section to be the three times of
the maximum aggregate size seems to be adequate here. Such a ratio ensures that no
pronounced wall effects occur so that the features of the material do not change and the
material can be well extruded. Further comments on downscaling, including possible
changes in maximum aggregate size, follow in Section 7.1.2.
The next open question revolves around determining the time interval between layers,
which depends on the rate of printing. The rate of printing is a very crucial parameter
for DC. The rate of printing can be neither too high (leading to inadequate buildability)
nor too low (leading to weak interface bonds, higher costs). Similar to the specimen
height, the time interval TI to be followed in laboratory tests can be deduced directly
from the target process parameters. The total travel length L of the printhead can be
determined from the layout/floorplan of the target application; see Figure A.3. With an
average (horizontal) printing velocity of VDC , the minimum time interval between two
layers can be expressed by Equation (7.5):

128
Buildability of 3PCs 7.1. Buildability criteria

T Imin = L/VDC (7.5)

In this first approximation VDC is assumed constant, not accounting for a) velocity vari-
ations when printing corners, and b) the acceleration and deceleration at the beginning
and the end of printing one layer, respectively. For TI > L/VDC , the printing process has
to be halted, e.g., to account for possibly insufficient buildability of the applied material,
thus leading to longer construction times and losses in efficiency. For all other cases where
TI < L/VDC concrete buildability is over-engineered, i.e., more than necessary, which may
affect the interlayer bond negatively, but also poses greater challenges in meeting the
requirements of pumpability and extrudability. While defining TI, the economic viability
of the target application shall be considered as well. The corresponding process term, the
average velocity VDC , is already addressed in Eq. 10. It is worthy to note that for a wall of
given gross dimensions, the time interval TI in the case of FP approaches will be higher
than that for FWP approaches due to longer total travel length of the printhead. This
aspect is elaborated in Section 7.3. To be economically viable, the condition according to
Equation (7.6) must be fulfilled for any DC approach. Here, CostsDC and CostsCC are the
total construction costs in the case of the DC approach chosen and in the case of current
corresponding conventional construction (CC) approach, respectively.

CostsDC 6 CostsCC (7.6)

"Replacement of masonry structures for residential buildings" have been identified as

an example application for CONPrint3D® [12, 39]. In this case, CostsCC are the current
total construction costs for masonry in a unit residential building. Expressing the costs of
machine and labour as cost per unit time and material costs as cost per unit volume, the
total costs can be estimated by using Equation (7.7):
(McCoDC + LaCoDC ) · tDC + MtCoDC · volDC + AdCoDC − SaDC 6
(7.7)
(McCoCC + LaCoCC ) · tCC + MtCoCC · volCC + AdCoCC − SaCC

where, McCoDC , LaCoDC and McCoCC , LaCoCC are the machinery/equipment costs
and labour costs for DC and CC, respectively; MtCoDC and MtCoCC are material costs
for DC and CC, respectively; AdCoDC and AdCoDC are additional costs costs for DC
and CC, respectively. Additional costs represent, among others, costs of design, control
measures, concrete curing (more elaborate in DC than CC), additional testing, consulting
fees which may be necessary in the case of DC. SaDC and SaCC are savings that are
exclusive to either DC or CC, respectively. For instance, if structures are produced using
high-performance concrete, then the durability of such elements can be much higher
than that of, as an example, masonry. This can be translated into cost per structure or
representative volume. At the same time, masonry walls can be more energy-efficient
than concrete walls. Such intricacies will vary for each DC application and cannot yet
be quantified generally. tDC and tCC are the times needed for constructing the target
structures in cases of DC and CC, respectively, while voldc and volcc are the total volumes
of material used in case of DC and CC, respectively.
Expressing time in terms of average velocity in the case of DC and inverted construction
rate RI (unit: h/m2 ) in the case of CC, we obtain:

129
Buildability of 3PCs 7.1. Buildability criteria

Lt
(McCoDC + LaCoDC ) · + MtCoDC · volDC + AdCoDC − SaDC 6
VDC (7.8)
(McCoDC + LaCoDC ) · (RI · surf ace area) + MtCoCC · volCC + AdCoCC − SaCC

where surface area is the total “one-side” surface area of the element being constructed;
Lt is the total travel length of the printhead, which is assumed to be equal to the length
for a single layer L multiplied by the total number of layers nlayers . The traversing
of the printhead without printing, i.e., moving it to a new printing position is not
considered. Such traverses are very specific to the target application and the related
process parameters and can be added to Lt , if known.
The net value of additional costs AdCoDC and savings SaDC in the case of CONPrint3D®
is assumed to be a lump sum of costs amounting to 10 % of the total construction costs.
For the masonry application no additional costs other than material, equipment and
labour were considered; see Section 7.2.2.
The minimum average printhead velocity for an economically viable DC application,
Equation (7.9), can be obtained by rearranging Equation (7.8):
VDC,min >
(McCoDC + LaCoDC ) · Lt (7.9)
¯ · ((McCoCC + LaCoCC ) · (RI · suf race area) + MtCoCC · volCC ) − (MtCoDC ) · volDC
0.90

Since it is known that:

a) Lt = the length of the layer multiplied by the number of layers nlayers ,
b) surface area = the length of the layer multiplied by the height of the wall, and
c) total material volume is the length multiplied by the height and breadth of the
walls,
Equation (7.9) can be transformed to Equation (7.10):
VDC,min >
(McCoDC + LaCoDC ) · nlayers (7.10)
¯ · ((McCoCC + LaCoCC ) · (RI · Happ ) + MtCoCC · HCC · BCC ) − MtCoDC · HDC · BCC
0.90

where HCC , BCC , HDC , BDC are the height and breadth of the walls in cases of CC and
DC, respectively. While HCC = HDC is chosen here for ease of comparison, the breadth of
the layers produced in DC can be smaller than that of CC. Since materials used for DC
applications are often superior to masonry in terms of mechanical performance, thinner
walls produced using DC can meet in principle the same design specifications as thicker
walls produced using CC. Equations (7.9) and (7.10) can be adapted also to other DC
applications to compute a minimum average printhead velocity that should be attained
to make the DC application economically viable with respect to the fabrication process
as such. Certainly, there are also other factors, which may influence economic feasibility
to a great extent. Thus, the entire process from planning to actual construction should be
evaluated.

7.1.2 Complementary considerations

7.1.2.1 Scope of the proposed approach
The proposed approach is applicable if DC technology is used for the construction of
common structures such as residential buildings, offices, bridges, etc. The approach

130
Buildability of 3PCs 7.1. Buildability criteria

does not yet address complex and unique structures with non-standard elements and
artworks. In author’s opinion, cost efficiency might not be the first priority for complex
and unique structures with respect to the use of DC. However, for mass-produced
structures such as houses, the cost efficiency is one of the key requirements.
While applying the proposed approach, the following statements should be considered:
a) The economic viability approach under consideration is independent of the pre-
sented “simple buildability” test. Therefore, it can be applied stand-alone to
identify the maximum TI between the layers based on economic viability. Thus, the
approach can be used with any buildability tests or numerical models; see [7, 8, 17].
This enables determination of the minimum yield stress and the rate of structural
build-up that the concrete should have for economically viable DC.
b) Provided the necessary testing equipment, skilled technicians and other resources,
the fundamental approaches published in Wolfs et al. [8], Suiker et al. [17] and
Perrot et al. [7] can be followed to test buildability. Since the approaches in [7,8,17]
do not address economic viability, Equation (7.10) as proposed in this chapter can
be used to complement these approaches.
c) The proposed approach addresses base-structure construction and not the final,
ready-to-commence structure. In other words, the costs of utility installations are
not considered.
d) Often chemical admixtures such as accelerators are used to ensure buildability. In
such cases, a very important question is: in which dosage does the admixture need
to be added? The approach proposed in this work at hand enables determination
of the duration of the maximum economically viable time-interval. This helps in
determining the dosage of the chemical admixture to be introduced so that the
required time-interval is maintained.

7.1.2.2 Downscaling and surrounding conditions

The proposed approach assumes that material behaviour tested on the down-scaled
specimens under laboratory conditions (lab-tests) is representative of the material
behaviour in full-scale structure in real application (full-scale). In reality differences
between material behaviour as in lab-tests and full-scale applications may arise, e.g., due
to variation in the quality of extrusion/compaction (process-induced changes) or in the
evaporation rate related to the volume of the printed element.
Here it is important to emphasize that the challenge posed by some differences in
lab experiments and full-scale applications is a universal issue that is not particular
to buildability tests or DC in general. Taken broadly, such issues can only be resolved
by full-scale tests and direct comparison, followed by error-minimization measures. The
approach proposed in this thesis for laboratory tests already takes into account certain
considerations in mimicking conditions under full-scale applications.
a) Tests are performed in-line, i.e., they are integrated in the 3D-printing process, as
opposed to offline tests where concrete has to be collected separately and tested in a
test device. Thus, time-dependent influences are avoided and errors due to process
induced changes are minimized.
b) Specimens are extruded using an extruder system similar to that foreseen in the
prospective full-scale application, thus subjecting the tested material to a similar
shear history and degree of compaction.
c) The curing of laboratory specimens is adjusted to that planned for the full-scale
application, indeed, if planned at all. If specific temperature, humidity, and wind
conditions are expected for the full-scale application, the laboratory tests should be

131
Buildability of 3PCs 7.2. Application on CONPrint3D®

performed under similar environmental conditions if possible.

d) If admixtures are added in the printhead during full-scale applications, the same
procedure should be implemented in the lab-scale 3D-printing.
Regarding the change of scales, the author does anticipate that the proposed approach
withstands the change in scales barring some limitations, which need further investiga-
tion. So long as the changes are only in dimensions and not in shape of the geometry, the
validity of the proposed approach can be seen as convincing. Nevertheless, a correction
factor may need to be introduced to take dimensional changes into account. Such factors
can be identified theoretically/numerically and/or with the help of full-scale validation
experiments. It goes without saying that in some cases, i.e., full-scale applications or in
large and tall structures, downscaled experiments are the only feasible option in assessing
buildability.
To extend the range of downscaling, the maximum aggregate size used in mixtures
for laboratory experiments may be reduced. It seems feasible to decrease the maximum
aggregate size for 3D-printing in a laboratory since the effect of maximum aggregate
size can be relatively well estimated based on available contemporary knowledge. It is
known from [7] that the buildability of a printed layer can be expressed in terms of,
and proportionally depends on the structural build-up of concrete. Mahaut et al. [113]
related the static yield stress of cementitious suspensions with coarse particles τc (φ, t) to
the static yield stress of the suspending cement paste τc (0, t), to the time t passed at rest,
and to the coarse particle volume fraction φ; see Equation (7.11).

τc (φ, t) = τc (0, t)g(φ) (7.11)

They suggested that the presence of φ volume fraction of coarse particles in a cement
paste will magnify its static yield stress as a function of the volume fraction of coarse
particles g(φ). Here, φm is the maximum volume fraction.
q
g(φ) = (1 − φ)(1 − φ/φm )−2.5φm (7.12)

In addition, Mahaut et al. [113] postulated that “the structuration rate Athix has the same
dependence on the coarse particle volume fraction as the yield stress”. These findings
imply that the parameters identified through a proposed approach on a concrete with
finer aggregates could be applied to concrete with coarse aggregates using a factor similar
to that of Equation (7.12). However, this hypothesis is yet to be verified.

7.2 Applying proposed approach to CONPrint3D®

The generalized approach proposed in Section 7.1.1, for determining buildability test
parameters is applied here to an on-site concrete 3D-printing technology, CONPrint3D® .
On completion of this section, the specifications of buildability tests on cementitious
materials applicable for CONPrint3D® are clearly derived. As elaborated in Section 7.1.1,
the necessary specifications are: height of the experimental wall to be printed, the
number of layers in this wall, and the time TI between the layers. The target full-scale
implementation of CONPrint3D® technology will be primarily with coarse-grained
aggregates, of maximum aggregate size 8 mm to 16 mm. However, the use of fine-grained
concretes should not be excluded at this stage, due to their advantages with respect to
processing and the maturity of the required extrusion machine technology, “lightness” of
the printhead, etc. In this context and as a first step the verification presented below will
consider an application case where fine-grained concrete is used.

132
Buildability of 3PCs 7.2. Application on CONPrint3D®

7.2.1 Height of the wall specimen and number of layers to be tested

One- to multi-story residential buildings [12] make up an important segment of the
primary target applications the CONPrint3D® approach. Table 7.2 presents exemplarily
the dimensions of outer and inner walls for a single story. Also presented are the
nozzle dimensions employed for the lab-scale 3D-printing testing device (3DPTD); see
Section 3.2. Furthermore, Table 7.2 presents various extra cases, namely Inner, Outer2
and Outer3. These cases provide additional information on how Hmin,exp and nlayers
vary depending on the wall and nozzle dimensions. Hmin,exp and nlayers are calculated
using Equations (7.2) and (7.4). These calculations highlight the consideration of layer
geometry in the proposed buildability test methodology. If laboratory nozzle breadth is
increased and laboratory nozzle height remains unchanged, then the height Hmin,exp to be
tested in the laboratory also increases. If the breadth of the target wall decreases, as in the
case of inner walls, then the Hmin,exp increases. Similarly, if the height of the printed layer,
i.e., corresponding opening of the nozzle opening in the lab, increases where its breadth
remains unchanged, then the number of layers to be printed in the laboratory tests will
decrease. These variations of buildability test specifications are crucial, considering that
the nozzle dimensions in the laboratory printers may vary significantly even though the
final targeted element (application) dimensions and the geometry of the real application
printhead do not change. Figure 7.1 illustrates how the number of layers varies a) if nozzle
height is kept constant and breadth is increased and b) breadth is kept constant and
height is increased.

Figure 7.1: Variation of number of layers to be tested with variation in nozzle cross-section for
a constant target application.
The 3PC utilized for the verification tests, Mixture C2, is a high-strength fine-grained
concrete; see Section 8.2.3. This 3PC is suitable in principle for the outer load-bearing
walls of the target residential buildings. Thus, for the laboratory buildability tests, the
case “Outer” presented in Table 7.2 will be considered. The actual nozzle height used for
the experiments is 17 mm. As shown in Table 7.2, to ascertain the buildability of Mixture
C2, 18 layer specimens have to be printed in the laboratory tests. The test specimens shall
comply with maximum time interval and prescribed tolerances, for example, according
to the German standard DIN 18202:2013-04 [280]. The handling of the cases other than
Outer should follow the same routine; note that a perspective 20 mm nozzle height is
assumed for the calculations of all cases except Outer in Table 7.2.

133
Buildability of 3PCs 7.2. Application on CONPrint3D®

Table 7.2: Dimensions of the target application and corresponding calculated parameters for
buildability testing.

Parameter Unit Outer Inner Outer2 Outer3

Wall height Happ [m] 2.500 2.500 3.000 2.500

Application

Wall breadth Bapp [m] 0.240 0.175 0.240 0.240

Aspect ratio αapp 10.417 14.286 12.500 10.417

Layer breadth BL,exp [m] 0.030 0.030 0.030 0.060

Lab (αexp = αapp )

Layer height HL,exp [m] 0.017 0.020 0.020 0.020

Minimum height to be tested Hmin,exp [m] 0.313 0.429 0.375 0.625

No of layers nlayers [#] 18 21 19 31

7.2.2 Time interval

The calculations presented in this section are valid for walls of one floor of a multi-story
house erected using CONPrint3D® . For purposes of comparison, an estimation for walls
made of conventional masonry is provided too. The process parameters of masonry
construction are given in Table 7.3; here the use of sand-lime bricks is assumed. These are
used to calculate the construction costs for the masonry work and to derive VDC,min . The
outer wall length of 43.04 m and the inner wall length of 19.86 m as presented in Table 7.3
are calculated from the floor layout of the target application, presented in Figure A.3. The
openings (e.g., windows) are also included in the travel lengths as the printhead must
traverse this distance, however without depositing material. To calculate the printhead
travel length for each layer to be deposited, it is essential to address “printing scenarios”
or the aspect of “tool-path optimization”.
In the example presented here, it was assumed that one complete layer along the outer
walls is printed first. Subsequently, further layers of the outer walls are added upon
one another. After the completion of the outer walls, the inner walls are produced and
frictionally connected to the outer walls by stainless steel anchors, which are inserted into
each layer. In contrast to this simple scenario, the path of the printhead can be defined
in numerous ways, as shown in Figure 7.2, while various algorithms can be utilized for
determining the optimum printing scenario. In such cases, the wall length in Table 7.3,
must be adjusted according to the printing path finally chosen.
It is noteworthy that the choice of a suitable starting point and the minimization of
idle-traverses (travel times without concrete discharge, shown with the dashed line in
Figure 7.2) are of great importance in any printing strategy.
The optimal printing path depends on numerous boundary conditions. An essential
optimization criterion is the path length. While the shortest printing path should be
generally preferred, it cannot always be used due to other constraints, such as the motion
and clearance profiles of the printhead or on-site construction process conditions. Zhang
et al. [102] adapted the so-called “travelling salesman problem (TSP)” to determine the
optimal tool path for constructing with Contour Crafting technology. They derived the
shortest paths by adding multiple vertexes in the corners and wall connections, and then

134
Buildability of 3PCs 7.2. Application on CONPrint3D®

Table 7.3: Dimensions of the target application and process parameters of example masonry
construction.
Parameter Unit Outer Inner Outer2 Outer3
Wall height [m] 2.500 2.500 3.000 2.500
Application

Wall length [m] 43.04 19.860 43.040 43.040

Wall breadth [m] 0.24 0.175 0.240 0.240

Inverted rate of construction [h/m2 ] 0.6

Masonry

Machine costs [e/h] 14

Labour costs per person [e/h] 35
No. of workers [person] 3
Material costs [e/m3 ] 164

by transforming the optimization problem from the node-oriented to the edge-oriented

view.
The rate at which masonry construction takes place is generally expressed in terms of
the time needed to complete a square metre of a wall, usually an hour. Since this rate
is inverse to the rate generally considered in concrete construction rates, i.e., unit area
per unit time, more specifically square metres per hour, it is termed here inverted rate of
construction with a notation RI. The RI values used in this thesis are according to [281]
and for the case of masonry wall type KS-L-R 8 DF 240 mm. The rate of construction is
given per person. Thus, when deriving VDC using Equation (7.10), one must multiply RI
with am actual number of workers working on the masonry application, which, in this
case is three, leading to an effective RI of 0.2 h/m2 .
Machine costs presented in Table 7.4 for a small crane (Kleinkran C.2.00.0007I) are in
accordance with BGL 2015 [282]. These costs include repair, depreciation and interest
costs. The material costs are also calculated for the masonry wall type Format KS-L-R
8 DF 240 mm and include delivery and mortar costs [281]. Process parameters of the
CONPrint3D® are presented in Table 7.4, which are used to calculate the construction
costs and derive VDC,min .
Table 7.4: Process parameters for the application of CONPrint3D® .
Parameter Unit Outer Inner Outer2 Outer3
Layer height [m] 0.16 0.117 0.16
CONPrint3D®

Layer breadth [m] 0.24 0.175 0.24

Machine costs [e/h] 140
Labour costs (2 persons) [e/h] 70
Material costs [e/m3 ] 130
Since CONPrint3D® is an full width printing (FWP) process, the breadth of the printed
layer is equal to the breadth of the target wall. For the case “Outer” considered here, the
breadth of a printed layer is 0.24 m. The thickness of the layer is a process parameter,
which affects the total number of layers to be printed, total printing time, and thus
construction costs. The maximum feasible layer thickness depends on the material

135
Buildability of 3PCs 7.2. Application on CONPrint3D®

properties. For the case presented here, layer thickness was assumed to be two-thirds
of the layer breadth in accordance with the geometrical proportions of the nozzle of the
lab printer. Numerous scaled-down wall-elements had been already printed using the
aforementioned nozzle-aspect ratio. Machine costs for CONPrint3D® as given in Table 7.4
are, at the current stage, higher than that of masonry construction. They include costs for
a modified concrete boom pump, costs for transporting concrete from mixing plant to a
construction site and costs for adjusting and calibrating the pump on the site. In general,
for DC technologies lower manpower costs are envisioned in comparison to conventional
construction. In the case of CONPrint3D® , two workers would be necessary from today’s
perspective: one for machine monitoring and one for auxiliary works. Based on [281], the
average wage of one person is calculated at 35 e/hour.
2 6

1 5
8 9
6 4

7 10 3 10 7 1
9 8

5 3

4 2

5 1

4 6
9 7
3 5

10 7 6 8 9 2
8 10

2 4

1 3
Figure 7.2: Various scenarios for printing the walls of the considered house. Full lines indicate
printing and dashed lines indicated traversing without extruding concrete (Courtesy of M.
Krause).
The material costs for CONPrint3D® , 130.00 e/m3 , are calculated conservatively for
Mixture C2 used in the experiments. They include material cost for admixtures and
additives (microsilica suspension, fly ash and superplasticizer). In sum, the material costs
are approximately 70 % higher in comparison to the costs for ordinary concrete of the
strength class C25/30 used in conventional construction. The Mixture C2, containing
expensive additives and admixtures, was chosen deliberately for the calculations to
ensure process-safe implementation of the on-site digital construction of load-bearing
elements. Note that printable concretes are, in general, contain fine mineral additives and
chemical admixtures to achieve the required rheological, mechanical and/or durability
characteristics. The fine-grained concrete considered here has a compressive strength of
over 80 MP a at an age of 28 days. However, for the target residential building application,
the required concrete class according to DIN EN 206-1 is C25/30. Thus, a considerable

136
Buildability of 3PCs 7.2. Application on CONPrint3D®

reduction in material costs is feasible in the course of the optimization process. In

addition to material, machine and labour costs, 10 % additional costs are added to the
total costs of CONPrint3D® as detailed in Section 7.1.1.
The minimum average printhead velocity Vdc,min for CONPrint3D® application, to
be equally economically viable with masonry construction, is calculated using Equa-
tion (7.9) and data from Tables 7.2 to 7.4. After knowing Vdc,min the maximum (again
economically viable) TI between layers can be calculated as TI = Lt / Vdc,min . Table 7.5
presents both Vdc,min and maximum TI calculated for various application cases of
CONPrint3D® . For the Outer case, the time interval between extruding subsequent layers
should not exceed 43.9 min. In addition to the maximum TI , it is also possible to define
the minimum TI based on the maximum printing speed of the printer to be used. The
current laboratory printer for CONPrint3D® has Vdc,max = 540 m/h. However, for the
laboratory test presented in this chapter, a printhead velocity of 270 m/h was set, which
is used to calculate the minimum feasible time interval as TI = Lt / Vdc,max ; see Table 7.5.
The minimum TI for the Outer case is 9.56 min; see Table 7.5. It goes without saying
that if printing of the entire construction were to occur at a rate close to Vdc,max then
the costs and construction time in case of CONPrint3D® will be considerably reduced in
comparison to those of conventional construction.
Table 7.5: Process parameters of an example CONPrint3D® application.
Parameter Unit Outer Inner Outer2 Outer3
Vdc,max [m/h] 58.9 72.0 50.0 100.1
Maximum [min] 43.9 16.5 51.6 51.6
TI
Vdc,max [m/h] 270.0
Minimum T I [min] 9.6 4.4 9.6
The minimum TI variation with printing velocities is shown in Figure 7.3 for printing
lengths of 5 m to 100 m. The figure shows that the influence of VDC,max on minimum TI is
significant only below certain printhead velocities. Furthermore, the lower the printing
length the lower the influence of VDC,max on minimum TI.
120
100 m
100 80 m
60 m
80
40 m
Minimum TI [

60 20 m
10 m
40
5m

20

0
0 100 200 300 400 500 600
Maximum printing velocity VDC,max [m/h]
Figure 7.3: Variation of minimum T I with maximum printing velocity.

137
Buildability of 3PCs 7.2. Application on CONPrint3D®

7.2.3 Experimental verification

Wall specimens are produced with the 3D-concrete-printing test device (3DPTD); see
Section 3.2. The buildability of Mixture C2 was verified for a target application of
residential house, specifically for the case Outer; see Tables 7.3 to 7.5.
The design of the buildability experiment is to produce nlayers number of layers
with T Imin of time interval between layers. If the printed wall retains its geometry
and dimensions, then the tested material is applicable for the target application. As
determined in Section 7.2, a total of 18 layers with maximum TI of 43.9 min must be
printed. The TI mentioned here is the upper limit. TI longer than 43.9 min adversely
affects economic viability if all other parameters are assumed constant. Here it needs to
be recalled that such “long” time intervals may reduce inter-layer bond strengths. Hence,
while developing printable concretes, interface properties [71,253] and extrudability [26]
must be considered as well. Interface-bond investigations with Mixture C2 shown that
short, two-minute TI proved to yield quasi-monolithic printed structures with minimal
anisotropy; see Chapter 8. Considering this, it was decided to assess the buildability using
a rather short TI of 2 min.
At a concrete age of 20 min from water addition, printing experiments were started.
The primary wall (Wall 1) consisted of 25 layers, printed at a constant velocity of 75
mm/s; see Figure 7.4. Upon observing no visual deformation of the layers after completion
of required 18 layers, additional 7 layers were printed. Since more than the required
18 layers could be printed with a TI much shorter than the economically required
TI, the Mixture C2 could be designated buildable for the considered application case.
Furthermore, the dimensions of the wall specimen were measured at the age of 1 day with
a precision of 0.1 mm. Note that printed individual layers’ cross-sectional dimensions can
vary from those of the nozzle outlet; not only depending on the buildability but also on
the flowrate of the material. To objectively distinguish the deformation due to the weight
of the upper layers from the deviations due to excessive material pumping, a separate
reference wall (Wall Ref) was also printed with the same printing speed but with only
three layers. Following specific metrics were observed for both walls:
a) Visually detectable deformations;
b) Height of the printed wall;
c) ∆dpt = (breadth of the bottom layer - breadth of the top layer) / breadth of the top
layer.
The results of these measurements are presented in Table 7.6. No visual collapse or
deformations of the layers could be noticed on the printed specimens. When seen from
top-view, all layers appeared consistent and no bulging of lower layers could be detected.
The wall surface looked flat with right-angled layer corners. The height of Wall 1 with
25 layers was only 0.3 mm lower than the designed height of 425 mm. This asserts the
buildability of Mixture C2.
Table 7.6: Dimensions of the specimens produced for buildability tests.
Wall No of layers Height Std. dev. Breadth top Std. dev. Breadth bottom Std. dev. ∆dpt
[mm] [mm] [mm]
1 25 424.7 0.7 32.3 0.1 32.1 0.3 0.0
Ref 3 50.5 0.0 31.9 0.3 32.4 0.2 0.0

The thickness of the layers was measured along the length at each 10 cm. From these
measurements, the average thickness values were calculated. Every time the printhead
accelerated from rest to 75 mm/s and every time it decelerated from 75 mm/s to zero
velocity, minor disharmony occurred in the proportionality of the concrete flowrate and

138
Buildability of 3PCs 7.3. Buildability requirements for FP and FWP

printhead velocity. Thus, the thickness measurements were only considered starting 10
cm away from the edges. The mean breadth of the reference wall’s top and bottom layers
were 32.3 and 32.1 mm, respectively. The observed, maximum 2 mm, deviation from
the designed 30 mm nozzle width does not necessarily is due to the deformation of the
printed layers. For the Wall Ref (only three layers), the breadth of the layers was also
up to 2 mm higher than the nozzle breadth. If the deviation from the nozzle geometry
would be due to material deformations under a load, then in the case of Wall 1 (25 layers)
such deformations must be considerably higher than those of Wall Ref. This observation
confirms that the measured deviations were not due to the layer deformations under load
of subsequent layers. Moreover, if the printed layers would have deformed due to the
weight of upper layers then the breadth of the bottom layer should be wider than that of
the top layer. The ∆dpt values are 0.0 in both walls, indicating no significant differences in
the breadth of the top and bottom layers. The minor deviations from the nozzle breadth
are attributed to the fact that pumping had slightly higher material flowrate than need to
deliver the required volume for printing the cross-section of 30 mm by 17 mm. Overall,
the visual observations as well as height and breadth measurements, confirm the stability
and the buildability of Mixture C2 up to 25 layers with a TI of 2 min, confirming Mixture
C2’s applicability for intended target application.

a) b) c)
Figure 7.4: 700 mm-length wall specimen (Wall 1) printed with time intervals of 2 min
between layers. Photos are taken after printing 23 layers. b) and c) shows top views indicating
no lateral deformation.

7.3 Buildability requirements for filament printing and

full-width printing
The height component of the buildability test parameters remains essentially the same
for both the full-width printing (FWP) and filament printing (FP) approaches. In contrast,

139
Buildability of 3PCs 7.3. Buildability requirements for FP and FWP

the effective length of each layer to be printed varies significantly between FWP and FP,
which directly affects the TI. The effective length refers to total travel distance, which
is often greater than the length of the actual wall. The printhead with a single nozzle
opening (as in [1, 12, 77, 87]) travels approximately twice the distance in the case of FP
when compared to FWP, to complete the deposition of the outer filaments; see Figure 7.5.
Furthermore, additional time is needed to place the inner wave-like filament. Consider
Vprinthead as the velocity of the printhead that is calculated using distance traversed
and Vef f ective as element horizontal printing velocity that is calculated using printhead
displacement. In the case of FWP, the entire layer cross-section is printed in one run.
Thus, the printhead velocity Vprinthead,FWP is equal to the effective horizontal printing
velocity Veffective,FWP . For FP, Vprinthead,FP is lower than the Veffective,FP . Consequently, to
achieve an equal construction rate in the case of FP, printhead velocity Vprinthead,FP has to
be much higher than in the case of FWP. A simplistic approximation for FP can be:

Vprinthead,FP = Vef f ective,FP (2 + k) with k = f (λ, û) (7.13)

with λ is the wavelength and û is the amplitude of the wave depicting the inner filament
of the wall produced by means of FP; see Figure 7.5c.

a)

b)

c)
Figure 7.5: Top sectional views of two walls of identical length and width, produced through
a) full width printing FWP and b) filament printing FP that also illustrates two additional
alternatives for inner filament defined by wavelengths λ; c) a scheme showing wavelength and
semi-amplitude of the sinusoid depicting inner filament in FP.
Since the Veffective of FP is much lower than that of FWP, the TI between layers in the case

140
Buildability of 3PCs 7.3. Buildability requirements for FP and FWP

of FP is also longer. Although it may appear advantageous in terms of “available” resting

time for structural build-up, a longer TI means a reduction in the economic viability and a
higher risk of formation of weak bonds between layers. Savings in material costs through
FP in comparison to FWP are less significant. As Tables 7.3 and 7.4 show, material costs
are not the major contributor to the total construction costs.
Equation (7.13) implies two cases that are summarized in Table 7.7, where FP is
compared to FWP for the same target application.
Table 7.7: Requirements in case of FP depending on the comparative constant with FWP.
Comparative case FP requires
a) Higher Vprinthead,FP
1) Constant economic via- b) More sophisticated 3D-printer setup to han-
bility dle the higher mechanical stresses induced
by faster accelerations and deceleration of the
printhead
a) Lower buildability (T I will be higher)
2) Constant printhead
b) Higher open-time/working-window of the
speed
construction material
c) Longer construction times

It is of interest mathematically to describe the above-mentioned intricacies of build-

ability requirements for FP and FWP in order to facilitate:
• the choice between FP and FWP, if the choice of material is restricted;
• extension of the model presented in Section 7.1.1 to FP cases by providing a math-
ematical description of TI in terms of wall geometry. For instance, the minimum TI
in the case of FP can be calculated using Equations (7.22) and (7.24);
• development of process-agnostic, printable concretes, i.e., concretes applicable for
both FP and FWP processes.
The following is the derivation for the needed Vprinthead,FP in relation to Vprinthead,FWP ,
provided only one nozzle is used and constant economic viability or constant construc-
tion rate is to be achieved (Case 1 in Table 7.7). Assuming
• length of the wall to be produced: L,
• distances that printhead travels in case of FWP (Figure 7.5a): L1,
• distances that printhead travels in case of FP (Figure 7.5b, c): L2 for outer and L3
for inner profile,
we obtain:
• time for printhead travel in FWP: tFWP = L1/Vprinthead,FWP = L1/Veffective,FWP and
• time for printhead travel in FP: tFP = (L2+L2+L3)/Vprinthead,FP .
Assuming equal economic viability in FP and FWP as in Case 1 (Table 7.5), both the times
should be equal, thus:

L1 L2 + L2 + L3
= (7.14)
Vef f ective,FW P Vprinthead,FP

L2 + L2 + L3
Vprinthead,FP = ( )Vef f ective,FW P (7.15)
L1

141
Buildability of 3PCs 7.3. Buildability requirements for FP and FWP

Since the length of the wall L = L1 = L2 , L3:

L3
Vprinthead,FP = (2 + )V (7.16)
L ef f ective,FW P
Expressing the inner filament in FP as a sinusoid y = i sin(jx) from 0 to L; the length
(distance between 0 and L) of the inner filament is:
Z Lq
L3 = 1 + ((i · j)cos2 jx)dx (7.17)
0

here, i = semi-amplitude of the inner filament = û = β/2 − b and j = 2π/λ; β and b are the
breadth of the produced wall and filament, respectively, and λ is the wavelength of the
inner filament; see Figure 7.5c. Therefore,
Z Lr
β 2π 2πx
L3 = 1 + (( − b) · )cos2 dx (7.18)
0 2 λ λ

From Equations (7.16) and (7.18) we obtain:

R Lq
1 + (( 2 − b) · 2π 2 2πx
β
0 λ )cos λ dx
Vprinthead,FP = (2 + )Vef f ective,FW P (7.19)
L
Equation (7.19) enables determination of printhead velocity to be followed while apply-
ing FP to retain economic viability equal to that of FWP processes. Furthermore, if equal
economic viability is not the primary concern and constant printhead speed is followed
for both FP and FWP, i.e., Case 2 in Table 7.7, then a higher minimum TI is valid for FP.
The following relationships between TImin,FP and TImin,FWP can be obtained:

T ravel length in case of FP

T Imin,FP = ( )T Imin,FW P (7.20)
T ravel length in case of FW P

R Lq
1 + (( 2 − b) · 2π 2 2πx
β
0 λ )cos λ dx
T Imin,FP = (2 + )T Imin,FW P (7.21)
L
In addition, with no comparison to FWP, the minimum TI in case of FP can be
calculated for a printhead velocity of Vprinthead :

R Lq
2L + 0 1 + (( 2 − b) · 2π 2 2πx
β
λ )cos λ dx
T Imin,FP = (7.22)
Vprinthead

Equation (7.22) is a generalized formula to determine the minimum time intervals

between layers, which must be followed during laboratory buildability tests if an element
of target length L is produced through FP. It must be noted that Equation (7.22) is valid
only if one nozzle is used to print the filaments; see Figure 7.5a. If a printhead with more
than one nozzle is used for printing, then the following scenarios can be followed:
a) Two passes are needed for one horizontal layer of the target wall. In the first pass,
two nozzles extrude the outer “shells” and in the second pass, one nozzle extrudes
the inner “wave”; see Figure 7.6b. An example of this scenario is demonstrated by

142
Buildability of 3PCs 7.3. Buildability requirements for FP and FWP

Contour Crafting
b) One pass is needed for one horizontal layer of target wall. Two nozzles extrude the
outer “shells” and in parallel, a third nozzle prints the inner “wave”. To the best of
author’s knowledge, this scenario has not yet been demonstrated.
Moreover, in both scenarios, one additional pass of the printhead or another device will
be needed if spaces between “shells” and “wave” need to be filled, e.g., with insulating
materials or (self-compacting) concrete; see Figure 7.6c.
For the scenarios “a” and “b” mentioned above, Equation (7.22) can be transformed to
Equations (7.23) and (7.24), respectively:
R Lq
1 + (( 2 − b) · 2π 2 2πx
β
0 λ )cos λ dx
Vprinthead,FP = (1 + )Vef f ective,FW P (7.23)
L

R Lq
1 + (( 2 − b) · 2π 2 2πx
β
0 λ )cos λ dx
Vprinthead,FP = ( )Vef f ective,FW P (7.24)
L
Similarly, Equation (7.22) can be reformulated according to the number of noz-
zles/robots used. Based on the deduced relationships, it can be concluded that when
testing the buildability of a material for FP processes, generally higher velocities of the
printhead are to be followed during laboratory tests in comparison to corresponding tests
for FWP.

a) b) c)
Figure 7.6: Various scenarios for printing “one set of horizontal layers” in case of FP printing
process: a) single nozzle – three passes are needed [87], b) three nozzles – two passes are
needed [283] and c) representation of single nozzle with “post-filling” of the empty space
between layers [284].

143
Buildability of 3PCs 7.4. Chapter summary and conclusions

7.4 Chapter summary and conclusions

Buildability is the ability of an extruded cement-based material to retain its geometry
(shape and dimensions) under sustained and increasing loads in fresh or transient state. The
loads acting are a function of printed structure’s height h(t). In rheological terms, the
resistance can be expressed as a function of static yield stress τs (t) and a geometric factor
αgeom [7]. The approaches based on rheology did not consider buckling effect on the
buildability [7, 101]. The role of buckling can be critical and increases with increasing
slenderness of the structure. Predicting failure of printed structures necessitates precise
rheometrical characterisation. If combined plastic yielding and buckling failure is to
be predicted1 , extensive experimental and numerical investigations are necessary. For
wide-range implementation of DC, pragmatic test methods without necessity of complex
testing equipment (such as rheometers or numerical simulation tools) are beneficial.
This chapter introduced a practice-oriented buildability test methodology that is
directly linked to their target applications. Important aspects can be summarized as
follows:
a) The height of the wall to be printed and the time interval between layers were
identified as the primary specifications for laboratory buildability tests needed to
validate the choice of material and process parameters.
b) The height of the experimental wall was calculated using the aspect ratio of the
target construction element.
c) Maximum time interval was determined considering the minimum printing ve-
locity needed for DC to be economically viable in comparison with conventional
construction.
d) The proposed approach takes machine, labour, and material costs into considera-
tion. Therefore, the model is applicable for quantifying the economic viability of
DC processes. The model can be extended to take into account digitalization costs,
process interruptions and other DC intricacies.
e) In the case of CONPrint3D® application for a residential building, a lab-scale wall
of 17 layers with a maximum time interval of approximately 52 min was determined
as the buildability test parameter.
f) The buildability of a fine-grained, printable concrete was verified for this applica-
tion scenario.
g) The comparative analyses of full-width printing (FWP) and filament printing
(FP) revealed that, assuming the same rate of construction, the buildability of a
material for FP processes should be tested at higher velocities of the printhead in
corresponding laboratory experiments than in the case of FWP.
h) In addition, a relationship was derived to calculate the printhead velocity needed in
the case of FP related to the velocity in FWP, to achieve an equal construction rate.

1 For example, using Mohr-Coulomb failure criteria [8].

144
Chapter 8

Interface and hardened properties of

3D-printed elements

This chapter addresses the properties of 3D-printable cementitious materials (3PCs) in

the hardened state, in particular, mechanical performance and anisotropy. In the initial
stages of this doctoral research, compressive and flexural strength of the printed elements
were comparatively evaluated with cast specimens. The initial studies revealed that, with
3PCs it is rather trivial to achieve high compressive and flexural strengths and obtained
properties are comparable with cast specimens [12, 112]. However, further examination
confirmed high anisotropy in the mechanical performance of 3D-printed structures due
to the weak-interfaces or so-called “cold-joints”.
So far, no dedicated study on the influence of binder composition on mechanical
performance of 3D-printed concretes has been reported. The use of secondary cemen-
titious materials (SCM) has been proven to be beneficial for DC applications, thanks to
the enhancement of material thixotropy, among other parameters [2, 12, 27, 62]. At the
same time, due to lack of large aggregates and the demand for rapid initial strength
development, mixtures with higher dosages of cement are also commonly used in DC.
Higher cement content increases the risk of shrinkage cracking, thermal stresses, weak
joints [128] and a negative environmental footprint. A one-to-one comparison of mixtures
with high cement clinker content as the dominant binder on the one hand and SCM with
low cement clinker dosage on the other is of high research significance.
Against this background, the author presents in this chapter micro- and macro-level
investigations on mechanical performance and interface properties of 3PCs with varying
binder compositions. On the macroscopic scale, specimens produced by 3D-printing with
varying time intervals were subjected both to compression and 3-point-bending tests.
On the micro-level, scanning electron microscope (SEM) investigations of microstructure
at the interface of substrate and overlay are presented. Considering the influential
parameters mentioned above, test devices, methods and production approaches are
described in detail. The detailed presentation of the chosen approach will help in a)
analysing the experimental results better and b) developing standards/guidelines in the
long term. Both matters are essential for the widespread implementation of DC. The
research attempts to illuminate various potential cases of layer interfaces’ microstructural
formation. In addition to elaborating on how layer interfaces are influenced by mixture
composition, ageing and process aspects, phenomena considered likely in leading to
anisotropy in terms of mechanical performance are analysed as well.

145
Interface and hardened properties 8.1. Materials and methods

8.1 Materials and methods of investigation

8.1.1 Compositions, time intervals and specimen production
Two printable, fine-grained concrete compositions, Mixtures C1 and C2, were investi-
gated. Mixture C1 had CEM I Portland cement (PC) as the sole binder, while Mixture
C2 was given a binder consisting of 55 wt.% PC, 30 wt.% fly ash (FA) and 15 wt.%
microsilica (added as 30 wt.% of microsilica suspension, MSS). A water-to-cement ratio
w/c of Mixture C1 and equivalent water-to-binder ratio (w/c)eq of Mixture C2 were 0.42.
Table 8.1 gives further details of the respective compositions. C1 with 627 kg/m3 cement
needed a lower SP dosage than did C2, which had 391 kg/m3 of cement. In addition, the
paste volume of C1 was reduced to mitigate plastic shrinkage and thermal stresses, both
of which are to some extent higher in the case of higher cement content mixtures. PCE1
was used for Mixture C1 and PCE2 was used for Mixture C2. The dosages of SP in terms
of percentage by weight of binder (% bwob) are presented in Table 8.1. The sand fractions
were added in the ratios of 0.2:0.2:0.6 for fractions 1, 2 and 3, respectively. See Section 3.1
for details of used materials including particle size distributions.
Table 8.1: Mix-compositions under investigation.
Mixture Binder (w/c)eq Paste SP SP dosage Binder content
[vol. %] [%bwob] [kg/m3 ]
C1 CEM I 0.42 47.3 PCE1 0.75 627
C2 CEM I, MSS, FA 0.42 52.5 PCE2 2.00 391, 213, 107

RPM

Mixing SP with water

Adding liquids to dry materials

Scraping
Mixing Mixing walls Mixing
at 25 RPM at 45 RPM No mixing at 45 RPM

-2 -1 0 1 3 4 6 Time [min]
Figure 8.1: Mixing procedure.
Three different time intervals between depositions of 3D-printed layers — 2 min, 10
min and 1 day — were investigated. The (time interval) TI of 2 min was the minimum time
needed for the 3D-printer to produce one layer and the TI of 10 min represents a typical
TI, which is expected to be a common case for the production of full-scale wall structures.
Finally, the TI of 1 day, i.e., 24 hours, represents an interruption of the building process
and later resumption. To cover this case, two separate concrete batches, Batch 1 and Batch
2, were necessary to produce the specimens; see Table 8.2. On the first day, specimens for
2 min and 10 min and bottom (first) half of the 1 day TI specimens were produced. The
top half of 1 day TI specimens were produced on the second day. A single-shaft pan mixer
CEM 60 S Elba with a maximum concrete mixing capacity of 60 L was used, following

146
Interface and hardened properties 8.1. Materials and methods

the mixing regime presented in Figure 8.1.

Table 8.2: Production batches of concrete specimens with various time intervals.
Mixture Batch Day Vol. [L] Time intervals
C1 1 1 35 2 min, 10 min, 1 day (1st half)
C1 2 2 20 1 day (2nd half)
C2 1 3 35 2 min, 10 min, 1 day (1st half)
C2 2 4 20 1 day (2nd half)

8.1.2 3D-printing and mechanical testing

On the macroscopic scale, flexural (3-point-bending) tests and compression tests were
conducted at concrete ages of 1 day and 28 days; see Figure 8.2. Workability of fresh
concrete was characterised using the Hägermann flow table test (HFT) in accordance
with DIN EN 1015-3 [261]. Specimens for testing mechanical properties were 3D-printed
using the 3D-concrete-printing test device 3DPTD [12, 124]; see Figure 3.2.

a)

Figure 8.2: a) Representation of specimen extraction from printed wall, b-d) flexural test
setups, e-g) compression test setups. The dashed insets on prisms indicate prism halves used
for compression tests.
Mechanical tests were carried out on specimens which were saw-cut from a printed
straight wall; see Figure 8.2. For this purpose, nine walls were 3D-printed with a
printhead velocity of 75 mm/s. All processing parameters were kept constant, but TI
between placing a subsequent layer was systematically varied; see Figure 8.3. Specimens
were tested in three different cases:
a) loading perpendicular, abbreviation Perp, to layer interface plane,
b) loading parallel to the interface plane, Par1, in this case the specimen has just one
interface plane and
c) loading is parallel to interfaces and many interfaces exist in the specimen, Par2; see
Figure 8.2.
Compressive and flexural strengths were measured according to EN 1015-11 [261].
However, the dimensions of the saw-cut prism specimens were 120 mm x 25 mm x 25

147
Interface and hardened properties 8.1. Materials and methods

mm, deviating from the recommended 160 mm x 40 mm x 40 mm in [261]. The span of

specimens for the flexural tests was 100 mm. Compression tests were conducted on the
prism halves resulted from specimens tested in the flexural tests.
It should be noted that it is possible to produce specimens of above mentioned geome-
try with just two layers and use them then for interface flexural tests in the directions Perp
and Par1. However, preliminary investigations showed that two-layer, 3D-printed walls
go through high plastic shrinkage, thus leading to cracking, a phenomenon explained
by the high surface-to-volume ratio. Hence, a minimum of four layers were printed
for all walls. The additional layers exert a vertical load on the investigated specimen
in its fresh state as well, thus increasing the “representativeness” of the experiment in
respect of large-scale applications. For specimens to be tested in Par2, eight-layer walls
were produced; see Figure 8.2. All the printed walls were protected from surrounding
influences by covering with wet textile and a polythene sheet until 24 h after printing.
From an age of 1 day printed walls were cured under water for 7 days and in a
climate-controlled chamber from the 8th day to the 28th day. The nomenclature for the
various cases investigated in this study is detailed in Figure 8.4a.

8.1.3 Optimized printing sequence

Concrete produced in various batches may show slight variations in its material prop-
erties. Considering this, all specimens belonging to one type of concrete were produced
from a single batch, except — obviously — the top-halves of the 1 day TI specimens; see
Figure 8.3.

Figure 8.3: Layer-deposition schema (not scaled): Top and front view of the printed walls (L=
length, H= height) are depicted with the age of concrete at the beginning of each layer shown;
also the time intervals, number of layers and printing sequence are presented.
The time-dependent rheological properties of fresh concrete also influence the interface
properties. If walls for each TI were to be produced sequentially, i.e., printing layers of a

148
Interface and hardened properties 8.2. Macroscale: results and discussion

second wall only after fully completing printing the first wall, then the total production
time would exceed many hours. Similarly, if all the layers of a wall were to be printed
with a specified TI, e.g., 1 day, then a total of 8 days and 8 batches of concrete would be
needed for producing walls with TI of 1 day. Considering these challenges, the specimen
production regime was optimized in such a way that the “waiting time” between various
layer depositions of a particular wall was used to produce layers for another wall. Only
one layer interface (central) in the walls was produced with the specified TI, while all
other interfaces were produced with as short a TI as possible. For instance, in the wall
3 of 1 day TI only the fifth layer was produced with 1 day TI and all other layers were
produced with 2 min TI; see Figure 8.3. The optimized production sequence for all the
3D-printed walls is presented in Figure 8.3. Wall numbers are denoted following the
printing sequence as W1, W2 ... W9. Figure 8.3 also gives the total number of layers in
each of the walls, corresponding TI and age of concrete from the time after water addition
at the beginning of each layer.

8.1.4 Specimen preparation for microscopy

The region between two layers was investigated by means of SEM (QUANTA 250 FEG,
FEI) in low vacuum mode. In addition, the microstructure at a reference point in the
core of a layer was captured and compared with the microstructure of the interface
of two printed layers; see Figure 8.4. The samples were examined in the natural
state without application of conductive layers. The cross-sections of printed specimens
prepared through sawing do not represent the interface clearly due to the influence by
quasi-polishing by the saw blades. Hence, the specimens were prepared by splitting.
Splitting force was applied vertically to the printing direction to get a “fresh” broken
surface; see Figure 8.4. Though splitting is expected to exert lower mechanical stresses
than wet-grinding, the microscopic investigations showed that the splitting process led
in some cases, where bonding between the layers was weak, to opening or cracking along
the interface.
C# TI# A#
Mixture Time interval Age 1
#=1 # = 2 = 2 min # = 1 = 1 day
#=2 # = 10 = 10 min # = 28 = 28 day
# = 1d = 1 day 2
a) b) c) d)
Figure 8.4: a) Specimens nomenclature, b) regions investigated under microscope: 1
reference/core, 2 interface, c) saw-cut surface, d) split surface. See also Figure 8.14b.

8.2 Investigations on the macroscale: results and discus-

sion
8.2.1 Evaluation of fresh concrete properties
The rheological properties of concrete in fresh state have a significant influence on the
layer interface properties [11, 152]. With respect to the Bingham parameters — yield
stress and plastic viscosity — the yield-stress is expected to be more significant in the
case of non-vibrated, layer-wise concrete deposition. This can be substantiated by noting

149
Interface and hardened properties 8.2. Macroscale: results and discussion

that there is no mechanical or flow-induced intermixing, i.e., with resulting higher shear
rates occurring between two deposited layers, as in the case of distinct layer casting [152].
In contrast, the flowability under self-weight (flow initiation) plays a more significant role
in digital construction. Hägermann flow table (HFT) spread values for both C1 and C2
measured from various batches are presented in Figure A.4. They correlate with the yield
stress of concrete and thus give an indication of the concretes rheological states at the
time of production. All tested mixtures had a HFT spread value of 100 mm, equal to
base diameter of the frustum before strokes and values under 120 mm after applying
15 strokes [261]. Such a very low spread is a prerequisite to achieve buildability of
3D-printable cementitious materials (3PCs) if no accelerator is dosed in the nozzle. The
HFT spread values after strokes of C2 for both batches and the C1 first batch as well were
almost identical. The HFT spread after strokes of the C1 second batch (1 day TI) was
slightly higher than that of the C1 first batch and, importantly, greater than that for both
batches of C2. This means, from a pure “flowability” point of view 1-day TI specimens
of C1 are more likely to develop a good interface bond than those of C2. Nevertheless,
the mechanical properties testing indicates that C2 has a superior bond to that of C1;
see the following sections. Thus, it is hypothesized that the better interface bond in the
case of C2 did not originate primarily from its rheological properties, in terms of higher
flowability, but rather from its material composition and the physicochemical aspects of
the layer interface.

8.2.2 Interface properties and flexural strength

Bending test results, which are most relevant for the evaluation of the bond strength, are
summarized in Table 8.3 and graphically illustrated in Figure 8.5. The overall flexural
strength of Mixture C1, i.e., the mean of all directions and time intervals tested, at
the age of 28 days is 7.67 MP a with a coefficient of variation of 42.6%. The same for
Mixture C2 is 7.39 MP a, with a coefficient of variation of 10.5%. The very high coefficient
of variation for C1 is explained by the influence of the layer interface on the flexural
strengths detailed below. In Perp and Par1 directions the flexural strengths of the “cement
only” composition — C1, are clearly higher than composition C2 at both the ages tested.
However, in the critical direction of Par2 the flexural strengths of C1 are significantly
lower than that of C2, clearly substantiating the weak interface bond of C1. The strength
loss in terms of flexural strength, ∆f lex , due to layered manufacturing is given as the ratio
of the difference in flexural strength from “Perp” and “Par2” tests to the flexural strength
in “Perp” direction:

flexural strength Perp − flexural strength Par2

∆f lex = (8.1)
flexural strength Perp

It can be anticipated that flexural strength is not significantly affected by the presence of
weak layer interfaces when loading is perpendicular (Perp) to the joints. In comparison
to the Perp series, no negative effect on the flexural strengths was observed if the
loading direction was Par1; see Figure 8.2 for loading cases. Considering the particular
arrangement of joints in this series, this result is well plausible and could be expected.
Again, as expected the specimens tested in the Par2 arrangement yielded considerably
lower values of flexural strength. The tensile stresses act here perpendicular to several
layers interfaces, which in failing at moderate stress levels are definitively the weak links.
From the results presented in Table 8.3 and Figure 8.5, it is evident that concrete C2 has
much lesser anisotropy in comparison to concrete C1; see low values of strength loss
∆ flex . This is despite Mixture C2’s lower flexural strengths in loading cases Perp and

150
Interface and hardened properties 8.2. Macroscale: results and discussion

Par1 than in the cases measured for C1. For both C1 and C2, the clear effect of TI on
the reduction in flexural strengths can be identified. In case of C1, TI of 1 day led to an
alarming 91.9 % reduction in flexural strength with the lowest measured flexural strength
of 0.83 MP a. 1 day TI resulted in a decrease in flexural strength of C2 by 23.1 %, which
seems tolerable, especially considering that the lowest flexural strength measured for C2
was 5.61 MP a. Thus, even for long TI the formation of cold joints can be mitigated to
a great extent by careful selection of material composition. However, it is essential here
to underline the positive effect of keeping time intervals short. In the case of Mixture
C2 specimens produced with 2 min and 10 min TI showed only very moderate loss of
flexural strength, 9.9 % and 14.1 %, respectively. This tendency is also valid for Mixture
C1, where specimens produced with 2 min and 10 min TI showed a decrease in flexural
strength by 47.7 % and 68.2 %, respectively. A comparison of the strength losses for C1
specimens tested at the ages of 1 day and 28 days reveals a minor improvement in the
interface properties over time. Furthermore, in the case of C1 the trend of strength loss
resulting from increasing TI is identical at both ages tested. For C2, flexural strengths in
Par2 direction are not available for the age of 1 day due to a measurement error. However,
the trend is expected to be similar as that observed for 28 days old specimens. The origins
for the weaker interface bond strength in the case of C1 in comparison to C2 could be
attributed to various aspects, including:
a) porosity and saturation state of the substrate,
b) moisture condition of the surface,
c) magnitude of plastic shrinkage,
d) varying yield stress and/or plastic viscosities of the deposited material and sub-
strate.
As detailed in Section 8.2.1, Mixture C1 was slightly more flowable than C2 and so
the yield stress could not be the primary reason for its weaker interface properties.
It was reported that higher moisture levels on the surface increase interface bond
strengths, despite longer TI [116]. Visual examination carried out during the 3D-printing
of the specimens indicated that Mixture C1 had higher surface moisture throughout the
printing duration than the C2 specimens.
A further reflection concerns the porosity of the printed material. Mixture C1 had
only Portland cement as binder while C2 had a considerable portion of pozzolana fines.
It is known that the addition of microsilica usually leads to a denser microstructure.
Additionally, lower binder volume content in C1 may cause some difference in the pore
structure. The porosity of the printed specimens is given in Table 8.4. In the absence of
dedicated porosity measurements, it was quantified as the ratio of “difference between
measured and theoretical densities” to “the theoretical density”, assuming zero % air
content. The measured density is the mean value of the measured densities of specimens
mechanically tested in various directions. The theoretical density was calculated as the
sum of all the constituents’ input weights per unit volume. The porosity of the printed
C1 specimens is, at 6.1 %, nearly 60 % higher than that of the C2 specimens. Microscopic
and naked-eye visual examinations, in case of C1, revealed large pores/cavities both
at the interface and as well as in the core of the printed specimens. Such large pores,
ranging from 50 µm to over 500 µm, reduce drastically the interface contact area of two
3D-printed layers. Due to their large size, these pores can be regarded as micro-cracks.
Thus, on one hand, the resulting reduction in interface bond area; on the other hand, high
stress concentrations on the pore (crack) edges, naturally reduces the overall interface
bond strength. The stress concentrations on the pore edges increase over-proportionally
with increasing flaw size. Moreover, even if the pores are not on the surface (in the
interface zone), but just below it, they still lead to weak lower bond strengths as this

151
8.2. Macroscale: results and discussion

Table 8.3: Flexural and compressive strengths of printed specimens at ages of 1 day and 28 days.
Flexural Compressive
Notation Age Perp Par1 Par2 ∆f lex Mean Co. Vari. Perp Par1 Par2 ∆comp Mean Co. Vari.
Units day [Mpa] [%] [Mpa] [%] [Mpa] [%] [Mpa] [%]
C1-TI1m-A1 1 6.93 6.88 1.11 84 5.0 55.0 41.9 39.3 40.5 3 40.6 2.6
C1-TI1m-A28 28 - 9.68 5.26 48 8.3 26.2 62.6 71.1 80.8 -29 71.5 10.4
C1-TI10m-A1 1 6.06 6.21 1.15 81 4.5 52.5 35.9 38.7 41.9 -17 38.8 6.3
C1-TI10m-A28 28 9.62 9.27 3.06 68 7.3 41.2 60.9 72.8 83.5 -37 72.4 12.7
C1-TI1d-A1 2 8.11 8.07 0.21 97 5.5 68.0 42.3 44.4 39.8 6 42.2 4.5
C1-TI1d-A28 28 - 11.07 0.83 92 7.4 63.0 70.0 74.3 69.8 0 71.4 2.9
C2-TI1m-A1F1 1 4.82 5.40 - - 5.1 5.6 28.3 29.9 - - 29.1
C2-TI1m-A28 28 7.53 8.78 6.78 10 7.7 10.7 98.9 101.6 101.1 -2 100.5 1.2
C2-TI10m-A1 1 4.83 5.32 - - 5.1 4.8 28.2 29.7 - - 29.0
C2-TI10m-A28 28 7.84 8.23 6.73 14 7.6 8.3 98.5 102.6 105.1 -7 102.1 2.7
Interface and hardened properties

C2-TI1d-A1 2 5.08 5.38 - - 5.2 2.8 34.0 34.9 36.8 -8 35.3

C2-TI1d-A28 28 7.29 7.76 5.61 23 6.9 13.4 96.3 93.5 101.8 -6 97.2 3.5
Mean (C1-A28) F2 28 9.97 10.00 3.05 69.3 7.7 42.6 64.5 72.7 78.0 -22 71.8 7.8
Mean (C2-A28) F2 28 7.55 8.26 6.37 15.7 7.4 10.5 97.9 99.3 102.7 -5 99.9 2.0
F1: Results not available for the age of 1 day due to a measurement error.
F2: The mean (C1-A28) and average (C2-A28) values represent the mean values of all the specimens tested at age of 28 days for Mixtures C1 and
C2, respectively. The corresponding strength loss ∆ values indicate the magnitude of anisotropy; e.g., ∆f lex for C1-A28 and C2-A28 are 69.3 % and
15.7 %, indicating the more anisotropic nature of the former mixture. Similar conclusions can be drawn from the coefficient of variation values
presented above.

152
Interface and hardened properties 8.2. Macroscale: results and discussion

area is subjected to higher bending moment than the rest of the specimen during testing.
In contrast to C1, Mixture C2 has a significantly lower number of large pores both at the
interface and as well as in the core, the reasons of which were addressed above already.
Table 8.4: Density ρ and porosity of 3D-printed specimens at concrete age of 28 days.
Property TI C1 C2
Theoretical ρ 2291.4 2230.6
Measured ρ 1 min 2156.6 2134.5
10 min [kg/m3 ] 2143.1 2151.2
1 day 2155.0 2148.7
Mean 2151.5 2144.8
Porosity [%] 6.1 3.8
Plastic shrinkage of the overlay concrete may also have contributed to weaker interface
bond strengths [128]. Beushausen et al. [128] reported that “The overlay is subjected
to shrinkage and thermal movements” and at longer time intervals, “the substrate
deformations are usually minor or negligible”. This implies that even if the freshly
deposited overlay initially envelopes well the terrains of the substrate, subsequent
differential plastic shrinkage strain may lead to separation of the overlay-substrate
connections. Additional investigation will be needed to quantify plastic shrinkage and
estimate the particular effect of differential shrinkage on the interface bond. This subject
cannot be covered in this thesis.

8.2.3 Compressive strengths

The compressive test results are of general interest for classifying the developed printable
cement-based composites, for comparative selection of various mixtures and for eventual
utilization in structural analyses and design. To study the anisotropy, the compressive
strength loss due to layered manufacturing, ∆comp , is calculated as the ratio of the
difference in compressive strengths from “Perp” and “Par2” tests to the compressive
strength in “Perp” direction; see Equation (8.2).

compressive strength Perp − compressive strength Par2

∆comp = (8.2)
compressive strength Perp

As expected, no definite quantitative conclusions could be drawn from the ∆comp values
in respect of weak joints. Remarkably, however, the coefficient of variation for 28 days
compressive strengths (7.8% for C1 and 2.0% for C2) follows the same trend as the
coefficient of variation for flexural strengths at this age (42.6% for C1 and 10.5%),
i.e., Mixture C1 shows higher variation in its strength values depending on testing
direction both in bending and compression. This indicates higher anisotropy in case
of C1 than C2. Remarkable is also the much lower magnitude of variation in the case
of compressive tests, which shows the less pronounced anisotropy with respect to the
material performance under compressive loading. It is noteworthy that the “losses” in
compressive strength are mostly negative, i.e., compressive strength in direction Par2
is actually higher than that measured in Perp direction; see Table 8.3 and Figure 8.8.
Summarizing these observations, it can be concluded that although the compression tests
performed in different directions cannot provide exact data on how the interface bond
varies, they seem to be able to serve well for quick estimations of anisotropy arising out
of various compositions and process parameters.

153
Interface and hardened properties 8.2. Macroscale: results and discussion

Figure 8.5: Results of flexural tests; mixture type and the TI are given in the insets.

154
Interface and hardened properties 8.2. Macroscale: results and discussion

The overall 28 days average compressive strength of Mixtures C1 and C2 for all
directions and time intervals is 71.8 MP a and 99.9 MP a, respectively; see Table 8.3. The
overall 1 day average compressive strength of Mixtures C1 and C2 for all directions and
time intervals is 40.5 MP a with a coefficient of variation of 0.9% and 32.8 MP a with
coefficient of variation of 8.7%, respectively. The compressive strength attained at 1 day
is 56% and 33% of 28 days strengths for Mixtures C1 and C2, respectively. This can be
explained by the comparatively higher cement content of C1 (627 kg/m3 ) than that of C2
(391 kg/m3 ); see Table 8.1. However, it is noteworthy that at the age of 28 days, Mixture
C2 has 30% higher compressive strength than Mixture C1, ascertaining once again the
benefits of using pozzolanic additives in concretes for DC. The author refer here to the
fact that usage of SCMs also improves rheological behaviour of printable cement-based
composites by enhancing thixotropy and therewith the increase in static yield stress [27].
For 1 day compression test results of both C1 and C2, no significant differences were
found in the compression strengths of various directions. However, at an age of 28 days
the compressive strengths in Par2 were generally higher than those of Par1 and Perp.
Some earlier studies reported such phenomena too [116, 253]. In most instances of the
mixtures investigated, the sequence of compressive strengths is in ascending order: Perp,
Par1 and Par2. While the difference in measured compressive strengths is not significant
from a structural point of view [12], the origins of this phenomenon should be critically
addressed.
One hypothesis for the observed influence is the differential water evaporation from
the specimens produced with varying TI. This hypothesis is illustrated in Figure 8.6;
where “o” and “s” indicate the overlay and substrate parts of a specimen. The specific
amount of water evaporated, increases with increasing surface-to-volume ratio As /V ol.
If a single specimen is examined as two parts: substrate and overlay, the As /V ol. of the
individual parts is higher than that of the total specimen; see Figure 8.6. This implies
that if the substrate is exposed longer without protection from drying, the water loss will
be higher. The increased water loss may have led to increased porosity and shrinkage
cracks, both of which can reduce the compressive strength. Alternate explanations
are explored below. Another probable reason for observed differences in compressive
strength measured in different directions can be explained as follows: the specimens and
the loading adapter/plane used for uniaxial compression tests in this study were not
cubical and square shaped, respectively. The compressive tests were conducted on prism
halves with dimensions of 25 mm x 25 mm x 60 mm; see Figure 8.2. The compression load
was applied on an area of 25 mm x 40 mm. Front, side and top views along with stress
distribution in investigated loading cases are shown in Figure 8.8. The blue lines indicate
layer interfaces. The specimens in compression tests typically fail along so-called shear
planes which “separate” the specimen core standing under triaxial compression from the
outer regions which are under a mixed compression-tension mode. The weak interfaces
(cold-joints) are the weakest links in the specimen and can, thus, trigger crack formation.
Both the extent of the (weak) interface area lying outside the triaxial compressive core
and the size of individual defects lying in the crossing of shear planes are factors decisive
for the observed load level at failure.
In the case of the Perp arrangement, the weak interface area outside the triaxial
compressive zone is the largest, which may explain why the strength values measured
in this direction are the lowest. In the cases of Par1 and Par2 the extension of interfaces
outside the triaxial compressive region is more or less the same. The weak interfaces
within this region are aligned with the acting force in both cases, however, in the Par1
arrangement, a longer portion of the interface is involved when compared with the Par2
arrangement. Since the weak interfaces oriented parallel to the load may lead to splitting

155
Interface and hardened properties 8.2. Macroscale: results and discussion

of the specimen and to premature failure, an on-average lower compressive strength

measured for Par1 (larger weak area aligned with the loading direction) seems explicable.
Remarkable in the case of 28 days compression tests on C2, i.e., the case with the strongest
interface among the cases tested, the difference in compressive strengths obtained for all
three directions is the least.
4

h
3 C-o
h/2
C-s C-s
0

As/vol. [mm-1]
Height [mm]

4
h 2 B-o
h/2
B-s B-s
0
4
h
A-o
1
0 A-s
0 10 min 1 day
Time from deposition
Figure 8.6: Three 2-layered printed specimens: 1 specimen A with 0 min TI, 2 specimen B
with 10 min TI, 3 specimen C with 1 day TI; and their 4 surface-to-volume ratio
(As /V ol., notice the secondary axis) as a function of time indicating differential
water-evaporation.
Panda et al. [253] and Sanjayan et al. [116] explained the anisotropic be-
haviour of printed elements under compression loading using a hypothesis based on
direction-dependent compaction. It was postulated that printed layers go through a vary-
ing degree of compression in various directions, resulting in corresponding compressive
strength variation. On closer observation two types of compaction can be identified in
printed structures: a) compaction in the extruder in the direction of concrete deposition
and b) compaction by weight of concrete layers deposited on-top. In this research, the
specimens for Perp and Par1 were taken from 2nd and 3rd layers of 4-layered walls,
whereas the specimens for Par2 were taken from the lower halves of 8-layered walls; see
Figure 8.2 and Figure 8.3. This choice was made since the specimens were cut horizontally
“along the layers” for the Perp and Par1 cases and vertically “across the layers” for
the Par2 case. The likelihood of the Par2 specimens being denser is indeed plausible.
The average densities of specimens (made of both compositions) in Perp, Par1 and Par2
directions are 2143, 2146 and 2156 kg/m3 , respectively. This substantiates the observed
compressive-strength variation’s possible origin in process-induced density variation,
at least to some extent. Summarizing, it can be stated that both the presence of weak
interfaces and potential variation in the compaction degree are the likely causes of the
anisotropic mechanical properties as observed in compression tests on printed concrete
specimens.

156
Interface and hardened properties 8.2. Macroscale: results and discussion

Figure 8.7: Compression test results of Mixture C1 and Mixture C2 for various TIs.

157
Interface and hardened properties 8.3. Microscale: results and discussion

a) b) c)
Figure 8.8: Stress distribution in loading cases investigated: a) Perp, b) Par1 and c) Par2.
Units are in mm. The blue lines and coloured plane indicate layer interfaces and interface
planes, respectively.

8.3 Microscopic investigations: results and discussion

8.3.1 General observations and classification
The presence of weak joints in the majority of the investigated cases, especially at an early
age of 1 day was evident from the microscopic analysis of the specimens. Based on visual
inspection the interfaces could be roughly subdivided in the four cases, namely: Case
A: weakly bonded, Case B: weakly bonded due to processing and curing conditions, Case
C: temporarily weakly bonded and Case D: strongly bonded. Case A is identified when
the interface exhibits long and wide separations between the two neighbouring layers,
where the separations are unlikely to be self-healed or bridged by hydration products
before examination/testing at the age of 28 days; see Figure 8.9a.
Case B is associated with interfaces weakened by the entrapment of air during layer
deposition or by other phenomena such as shrinkage and/or carbonation on the surface
of a layer before deposition of the subsequent layer after an extended time interval;
see Figure 8.14. Thus, by definition, the occurrence of Case B could be mitigated by
changing concrete composition and/or process parameters, such as TI and/or utilizing
a proper curing regime. Sanjayan et al. [116] observed as well that the “moisture level
at the surface between the layers is one of the major factors affecting the inter-layer
strength”. Various solutions are conceivable in providing external or internal curing and,
hence, mitigate the negative consequences of moisture loss as well as plastic and drying
shrinkage. In providing for internal curing one of the seemingly promising examples
in the absence of protecting formwork is the addition of superabsorbent polymer in
printable concrete [240]. Such a measure has already proven itself efficient in reducing
shrinkage-induced cracking in the use of cast concrete.
Case C represents temporarily weak interfaces, whose layer separations are narrow and
less porous in comparison with Cases A and B and which clearly indicate a self-healing
tendency over time due to the formation of hydration products; see Figure 8.9c and
Figure 8.12. Case D means that both layers are bonded/interlinked tightly; actually,
the interface between layers looks very similar to reference regions in the core; see
Figure 8.9d.

8.3.2 Microstructural differences between the 3PCs C1 and C2

Weak joints were observed for Mixture C1 with respect to all TIs under investigation.
In the case of the shortest TI of 2 min, the difference in interfaces between layers of C1
and C2 specimens is very clear. Despite the shortest TI, C1-TI2-A28 probes show layer
separation with horizontal separation length of over 1000 µm and vertical separations of

158
Interface and hardened properties 8.3. Microscale: results and discussion

over 200 µm; see Figure 8.10. In addition to parallel separation, elliptical cavities also
could be observed, which can be explained by air “enclosure” during the upper layer
deposition supported by the surface unevenness of substrate and upper filament. In the
case of C1, however, the air voids occur not only at the interface but also in the core of
the layers, independent of the TI; see the following subsections for further explanation
and Figure 8.13a.

a) C1-TI1d-A28: Very long, wide separation b) C1-TI2-A28: cavities due to “air enclosure”.
between layers.

c) C2-TI1d-A28: narrow interface filled by d) Zoomed pictures indicating interface (right)

delayed hydration products. quality as good as that of the core region (left).
Figure 8.9: SEM images showing the four observed types of layer interface microstructure.
In contrast to Mixture C1, the interface of C2-TI2-A28 appears to be mostly well
bonded as only a few larger pores are visible under microscope; see, for example,
Figure 8.13b. Here, two elongated cavities at the interface can be clearly associated
with “air enclosure”, a process induced defect. Between these large cavities, however,
the layers are very strongly bonded, substantiating the highest flexural strengths in
the Par2 direction measured on C2-TI2-A28; see also Section 8.2.2. The moderate TI
of 10 min specimens showed similar tendencies as of TI 2 min when C1 and C2 are
compared: C2 has narrower layer separations than C1 and C1 has large voids in the
core material as well. Both C1 and C2 specimens produced with 1 day TI showed
separations of concrete layers, however, in case of C1 the separation is pronounced;
see Figure 8.9a,c. This in in agreement with the strength losses as observed in bending
testing: C1-TI1d-A28 tested in the Par2 direction showed a reduction in flexural strength
of over 90%. Specimens made of Mixture C2 with less pronounced interface defects
registered strength loss of over 23% for the case C2-TI1d-A28. It can be concluded that in
addition to material optimization, changes in the processing, i.e., roughening of the layer

159
Interface and hardened properties 8.3. Microscale: results and discussion

top-surfaces, spraying bond-enhancers or controlling surrounding conditions, might be

needed to achieve homogeneous microstructure and isotropic material properties in the
case of extrusion-based material deposition.

a) C1-TI2-A28. b) C1-TI10-A28.
Figure 8.10: Weak joints in C1 specimens depicting very long, wide separations between
layers deposited with TI of a) 2 min, b) 10 min.
Overall, the microstructure of specimens made of C1 and C2 clearly show that for
the same time interval Mixture C2 develops a more homogeneous microstructure at the
layer interfaces. In case of 1 day TI specimens, long needle-shaped ettringite crystals
are found in numerous positions of the layer interface’s “empty” space; see Figure 8.11a.
At the same time, spherical clusters of ettringite are also present in narrower spaces.
Additionally, delayed growth of other hydration products, especially freely hanging
calcium hydroxide in larger voids could also be identified as well. As reported in [285],
among others “carbonation processes, moisture effects and, with that, moisture and
temperature changes such as those which occur under natural conditions can cause
ettringite formation in hardened concrete”. Further investigations are needed to confirm
whether the easier transport of external water and gases leads to delayed preferential
growth of these products at the interfaces and in their vicinity.

a) C2-TI1d-A28. b) C1-TI10-A28.
Figure 8.11: Examples of a) apparent, partial self-healing of separation at the interface, b)
only partial filling of large separation with free-hanging hydration products.

8.3.3 Microstructural changes over time

Two types of time-dependent changes were observed over the 28 days of ageing: a)
self-healing due to the bridging of micro-defects by growing hydration products, mainly

160
Interface and hardened properties 8.3. Microscale: results and discussion

C-S-H phases and b) alteration due to carbonation and filling with non-strength-giving
products. When it comes to self-healing, two parameters are significant: the initial
layer-separation distance and type of delayed hydration. Specimens produced with 2
min and 10 min TI showed potential for self-healing, thanks to their narrower initial
separation. A highly magnified SEM image of C2-TI2-A1 shows that already as early
as after 1 day “some healing” of the interface can be found; see Figure 8.12. The
C1-TI2-A28 of the poorly performing Mixture C1 shows a very tightly bonded section
of the layer interface as well; see Figure 8.12c. However, it is essential to distinguish
if the SEM-capture represents the conditions in the entire specimen or rather in only a
minor part. Figure 8.12d shows a typical specimen of the C1-TI2-A28 series; the observed
self-healing over time in case of C1 is very limited even for specimens produced with the
shortest time interval, apparently concentrating at small segments across interface where
the separation distance is small enough to be bridged. The development of interface
microstructure with increasing age is clearly reflected in the flexural strength values
measured in Par2 direction; see Table 8.3 and Figure 8.5.

a) C2-TI2-A1: Hydration products (mainly b) C2-TI2-A28: Very tightly bonded layers more
C-S-H) of both layers are growing into each other. or less throughout the interface.

c) C1-TI2-A28: Very tightly bonded layers; d) C1-TI2-A28: Investigated specimen.

overall area of such good bond is limited; see d).
Figure 8.12: Self-healing of the interface by layer-bridging with hydration products.
Obviously, the delayed product formation is insufficient for the complete self-healing
of weak interfaces in many instances; see Figure 8.11. Microscopic images for the better
performing Mixture C2 with TI of 1 day at an age of 28 days show only partially
developed bonding microstructure; see Figure 8.11a and Figure 8.9c. The long separation
regions between layers identified at the early age of 1 day are often apparently filled with
some reaction products. However, this filling substance was found to be a combination

161
Interface and hardened properties 8.3. Microscale: results and discussion

of calcite, ettringite and/or portlandite which does not exhibit a good defect bridging
as opposed to C-S-H phases. This observation is again supported by the mechanical
test results; see Table 8.3 and Section 8.2. In the case of C1-TI10-A28 and a few other
specimens, the layer separation is too large to be filled completely, let alone self-healed,
which would mean a gain in flexural strength when tested in the Par2 direction; see
Figure 8.11. While delayed growth of hydration products at the interface can be seen
here, free hanging portlandite crystals at the interface are neither “strength giving”, nor
are they connecting the neighbouring layers.

8.3.4 Process aspects

Another distinctive feature displayed by the SEM images of 3D-printed specimens are
longitudinal cavities and air voids occurring both at the interface and in the “core”
of the layers; see Figure 8.13. They differ distinctly from usual air voids and capillary
pores in concrete by their shape and size. The large air voids were found mostly in the
specimens made of C1, especially in the core. The typical longitudinal cavities induced
by 3D-printing originate from air entrapped between layers, the process likely depending
upon the rheology of the concrete, the speed of printing, the distance between previous
layer and nozzle, the shape of extrudate and the possible vibration of the printhead. The
greater number and larger size of such voids in the case of Mixture C1 can be explained
by the observation that this mixture containing Portland cement as the only binder had a
coarser matrix and was less cohesive than Mixture C2, which contained fine microsilica
and fly ash. Furthermore, certain vertical oscillations of the nozzle could facilitate the
formation of long cavities due to entrapped air between slightly wave-like layer shapes.
Influence of printing process parameters on interface strength was also reported by
Panda et al. [253]. The tensile bond strength of interface was found to be in nearly
inverse-linear proportion to the deposition elevation of the new layer above the substrate:
with an increase in the deposition height from 0 mm to 4 mm, the tensile bond strength
decreased from 2.3 MP a to 1.5 MP a.

a) C1-TI2-A1: cavities at the interface and in the b) C2-TI2-A28: cavities due to “air enclosure” at
core. an otherwise very well bonded interface.
Figure 8.13: Cavities at the interface in both C1 and C2 and, in C1, also away from interface.
It is obvious that a reduction of the deposition height without altering the concrete
discharge rate and printhead velocity should lead to a better “compaction” of the
deposited material against substrate, resulting in enhanced interface properties. The
large cavities observed in the present study can be likely traced back to a high deposition
height of 2 cm, which was chosen in order to achieve higher dimensional accuracy. More
or less as a consequence, cavities could be formed at the interfaces between printed

162
Interface and hardened properties 8.3. Microscale: results and discussion

layers depending on the rheological state of the material at the time of extrusion and
the nozzle-oscillation amplitude. Though large cavities at the interface can be seen both
in C1 and C2 specimens, their frequency was much lower in the case of Mixture C2.
Moreover, curing conditions also exhibit a pronounced influence on specimens’ features
in case of TI of 1 day. Despite the careful protection of printed specimens by wet cloths
and polythene sheets from the surrounding atmosphere, 1 day TI specimens made of
C2 underwent remarkable carbonation. Especially the top surface of the previous layer
(produced one day earlier) showed clear formation of calcites, which may have prevented
it from developing a complete bond with the subsequent layer; see Figure 8.14.

a) C2-TI28-A28. b) General view of the corresponding specimen.

Figure 8.14: Calcite formation at the layer interface of SCM mixture.

163
Interface and hardened properties 8.4. Chapter summary

8.4 Chapter summary

The mechanical properties of 3D-printed, fine-grained concrete specimens were in-
vestigated with respect to the effect of interface quality between individual layers,
accompanied by microscopic analysis, particularly of the interface areas. The 28 days
overall mean compressive strength for Mixtures C1 and C2 were 71.8 MP a and 99.9
MP a, respectively. The changes in the material composition and time intervals between
placements of subsequent layers exhibited a pronounced influence on the interfaces’
microstructure and, therewith, on the degrees of material heterogeneity and anisotropy.
The material composition has been proven decisive in achieving a good bond between
individual layers, even for long time intervals. So, Mixture C2, containing high amounts
of supplementary cementitious materials, showed superior performance in terms of both
homogeneity of microstructure and low dependency of mechanical properties on the
direction of loading when compared to Mixture C1, made with Portland cement as its
only binder. Based on the results and analyses the following statements can be deduced:
a) Mixture C2 showed a relatively moderate reduction in flexural strength of 9.9 %,
14.1 % and 23.1 % for specimens produced with time intervals of 2 min, 10 min
and 1 day, respectively. These results stand for the most crucial testing direction
Par2, which means that the flexural strength values yield information on the tensile
strength of the interlayer bond. It should be noted that no surface treatment to
improve the interfaces was carried out in this study. Hence, it appears well possible
to develop printable compositions with minimal loss, or even no loss at all, of
strength due to the layering process.
b) It is well known that the uniaxial compression tests cannot adequately quantify
interface strength. However, if performed in three different directions, they can still
serve as an indicator of the specimens’ anisotropy.
c) Close examination of various potential mechanisms revealed a) higher porosity, b)
higher amount of large pores/microcracks at the interface (also in the core) and
possibly also c) drying and plastic shrinkage of deposited material as causes for the
observed weak interface bond strengths, in particular, in the case of C1.
d) The presence of weak-interface in the majority of the 3D-printed specimens was
established. Observed interfaces were classified into four categories based on their
visual appearance, separation distance and origin.
e) SEM images clearly indicated the influence of time intervals and material composi-
tion on the interfaces’ quality. The formation of large cavities at the interfaces and
in the cores of C1 specimens explains its weaker interface bond.
f) For some specimens, e.g., for the series C2-TI2-A28, no weak interfaces between
layers were identified. In some other specimens, interface flaws seem to be largely
filled/self-healed with hydration and/or carbonation products.
g) However, such “filled” specimens did not always yield an expected gain in flexural
strength, since the filling products were in some cases C-S-H phases efficiently
bridging the layer-separations, while in other cases calcite, ettringite, and/or
portlandite crystals, which did not exhibit any significant defect-bridging action
with respect to the transfer of tensile forces.

Acknowledgements: The work presented in this chapter was published in peer-reviewed

journal Construction and Building Materials [31]. All the acknowledgements and support
of the co-authors mentioned in [31] are valid for this chapter too.

164
Chapter 9

Complementary research

There are numerous further investigations — with considerable amount of experimental

work — that were carried in complement to the main research presented in the previous
chapters. This chapter presents some of those works in a very concise manner, considering
their research and practice significance.

9.1 On incorporating reinforcement

The requirements with respect to the rheological behaviour of the 3PCs for various
digital construction (DC) processes can vary significantly; so do the printhead design
and other machine specifications. As a consequence, the possibilities for reinforcement
integration also vary depending on the DC process [132]. The advantages of DC such
as formwork-free nature and geometrical complexity, become specific challenges when it
comes to integrating reinforcement; see also Section 2.2.8.
To address this significant and challenging aspect of DC, two reinforcement approaches
were investigated. On one hand, short fibre reinforcement was investigated by de-
veloping printable strain-hardening cement-based composites (PSHCC). On the other
hand, additive manufacturing of steel reinforcement was examined. PSHCC research
work was jointly carried out by Dr. H. Ogura of Shimizu Corporation, Japan and the
author. 3D-printing of steel reinforcement was joint research of Institute of Manufacturing
Technology and Institute of Construction Materials of TU Dresden, the experimental work
was carried out as part of Diploma Thesis of Ms. J. Grafe [286] which was co-supervised
by the author. Both the works have been peer-reviewed and published [29, 30, 81].

9.1.1 Printable strain-hardening cement-based composites (PSHCC)

SHCC are a promising group of fibre-reinforced concretes; they exhibit multiple fine
cracks and strain hardening characteristics under tensile loading. In addition to ex-
tremely high mechanical performance under quasi-static and dynamic loading, SHCC
have a number of further advantages such as 1) narrow cracks that are favourable with
respect to the durability of structural elements, 2) added fine polymeric fibres help to
mitigate plastic, autogenous and drying shrinkage, thus, preventing the formation of
cracks. Since freshly 3D-printed cementitious elements are exposed to drying right from
the moment of extrusion, if no specific curing measures are taken, the mitigation of
shrinkage seems to be a critical issue. Therefore, the possibility of using dispersed fibre
reinforcement is an attractive approach to integrate reinforcement into the 3D-printing
processes.
Four compositions were investigated, out of which two were printable; see Table 9.1.

165
 Complementary research 9.1. On incorporating reinforcement

The binder was composed of 75 wt% cement CEM II/A-M (S-LL) 52.5R, 15 wt%
microsilica and 10 wt% fly ash. The length, diameter, density and tensile strength of the
fibres were 6 mm, 0.012 mm, 0.97 kg/cm3 and 3000 MP a, respectively. Fine sands with
maximum size of 1.0 mm were used. A PCE superplasticizer was dosed at 2.0 %bwoc.
Table 9.1: Compositions and their HFT spread diameters before ∅bs and after strokes ∅as .
Mixture W/B S/B fibre vol.% ∅bs ∅as ∅bs
∅as −1
A 0.24 1.20 0.3 119 142 0.19
B 0.22 0.50 1.0 120 135 0.13
C 0.22 0.20 1.5 119 133 0.12
D 0.24 0.20 1.5 129 153 0.19
The calculated relative spread values from HFT measurements are presented in
Table 9.1; while Figure 9.1a shows the results of ram extrusion tests. From direct printing
(buildability) tests, Mixtures A and D were determined to be buildability deficient.
Mixtures B and C with low relative spread ∅ ∅as − 1 of 0.12 and 0.13 were printable up
bs

to the tested seven layers (TI = 1 min) with consistent filaments without noticeable
deformation of bottom layers. Though the HFT results correlated with buildability
(see also Section 3.4), no direct correlation with ram-extrusion was observed. This can
be explained by the fact that extrudability depends not only on the yield stress but
also on plastic viscosity and friction between barrel and extruder. Mixture A with a
higher sand-to-binder ratio lead to high frictional component of the REF and thereby
requires higher ram-extrusion energy. Note that prognosis of blockages in progressive
cavity pump based on the REF measurements is not trivial considering the geometrical
differences; Section 6.4.3.
1.5
A B C D

1.0
Force [kN]

0.5

0 20 40 60 80 100 120
Displacement [mm]
a) b)
Figure 9.1: a) Ram extrusion force (REF) as a function of ram displacement, b) section of a
printed PSHCC wall.
PSHCC walls of length 1000 mm and height 140 mm were printed at a velocity 3 m/min,
beginning at a concrete age of 20 min after water addition. Printed layers were free of
surface defects and discontinuities and the layer edges were squared; see Figure 9.1b.
Prism shaped specimens of dimensions of 250 mm x 24 mm x 40 mm were saw-cut from
the printed walls for mechanical tests. The longitudinal axis of prim specimens was
in horizontal direction of printed walls, thus, parallel to interfaces between the layers.
Both printable compositions, Mixtures B and C, were used as well to produce mould-cast
specimens for the uniaxial tensile tests.
The uniaxial tension tests were carried out at the ages of 27 or 28 days. The tests were
performed in a 100 kN load capacity testing machine in a deformation-controlled mode

166
 Complementary research 9.1. 3D-printing of steel reinforcement

with a displacement rate of 0.05 mm/s. Non-rotatable boundary conditions were ensured
by glueing the samples at both ends in 20 mm thick steel rings, which were then bolted to
the testing machine. The deformations were measured on a 100 mm gauge length in the
middle of the specimens using two LVDTs; see [81] for additional mixture, production
and testing details.
As shown in Figure 9.2a, all specimens show strain-hardening behaviour, which was
accompanied by the formation of multiple, closely spaced fine cracks. The first-crack
stress and tensile strength are 4.18 MP a, 5.32 MP a for Mixture B and 4.25 MP a, 5.66
MP a for Mixture C, respectively.
The results show that ultimate tensile strain of Mixture C, 3.21 %, exceeds the yield
strain of the steel bars, and that it is possible to enhance tensile characteristic of
3D-printed members considerably. The mechanical properties of PSHCC were found
to be superior to those of cast SHCC; especially in case of Mixture C with 1.5 vol% of
fibre content; see Figure 9.2b. Figure 9.2b shows fracture surface as observed by a digital
microscope. Fibre distribution in printed specimens appears more preferentially1 aligned
than that of cast specimens. Fibre orientation was likely influenced by the printing
process. Further analyses of results are to be referred in Ogura et al. [81].
8 4
Vf = 1.0 % (B-print) Vf = 1.0 % (B)
Ultimate tensile strain [%]

Vf = 1.5 % (C-print) Vf = 1.5 % (C)

Tensile stress [MPa]

6 3

4 2

2 1

0
0 1 2 3 4 5
Printed Mould-cast Printed
Strain [%] specimens specimens specimens
a) b) c)
Figure 9.2: a) Tensile stress-strain curves of PSHCC, b) ultimate tensile strain of PSHCC with
different fibre contents, c) fracture surface of a printed Mixture C’s specimen.

9.1.2 3D-printing of steel reinforcement using gas-metal arc welding

(GMAW)
Autonomous additive manufacturing of steel reinforcement enables the realization of
full potential of DC. 3D-printing steel reinforcement by means of gas-metal arc welding
(GMAW) is one way to realize this; see Figure 9.3. The GMAW with its relatively high
melting point, has the advantage of faster production rates in comparison with beam
based processes and lower costs in comparison with laser-based approaches [29]. In the
process followed in this research, an arc is generated between a work-piece and a wire
electrode. The material feeding and energy input are synchronized and autonomous.
This process is surrounded by a shielding gas to prevent unwanted chemical reactions
between molten metal and the air. The base material (substrate) used in the investigations
presented here was a 3 mm–thick steel sheet made of S235JR. Furthermore, a solid wire
electrode with a diameter of 0.8 mm and a chemical composition, according to ISO
14341-A - G 46 5 M G3Si1 was employed.
1 In the direction perpendicular to crack surface

167
Complementary research 9.1. 3D-printing of steel reinforcement

The application of the argon-rich protective gas M21-ArC-25 (corresponding to EN ISO

14175) with small CO2 additives enabled the achievement of uniform penetration profile
as well as the improvement of the melt flow properties. The produced reinforcement
bar diameter is controlled by adjusting the “stickout” of electrode and the welding time
for each “layer/drop”. This process, however, has to be controlled precisely, otherwise,
cross-section of steel bar will be inconsistent, due to the relatively higher temperatures,
due to increased number of total layers (wide but thin) and longer time at each point.
To precisely control the “stickout” length accurately an adaptive process control is
implemented; which ensured dimensional accuracy. This online process control works by
detecting the drop in the applied open-circuit voltage when the wire electrode touches
the substrate. During this tactile measurement, the welding current source is not released,
ensuring that no arc is ignited. For tension tests, steel rebars of peripheral diameter 8 mm
and reduced diameter of 7.5 mm in the central “tested” length of 65 mm were 3D-printed;
see Figure 9.3c.

a) b) c)
Figure 9.3: Printed steel specimens a) without ribs, b) with ribs, c) with varying diameter.
The 3D-printed bars exhibited approximately 28 % lower yield stress and 16 % lower
tensile strength values in comparison with conventional steel reinforcement (B500), but
a considerably higher ductility, both in terms of pronounced yielding and higher strain
capacity; see Figure 9.4. The yield stress2 and strain capacity of printed specimens are
307 MPa and 22 %; the same for B500 are 425 MPa and 6.28 %. Visual observation of
the fracture surfaces also confirmed more ductile fracture (necking) than B500 (shear
type of fracture). Further, pull-out tests were conducted on specimens in which Mixture
C2 (see Table 3.1) was caste, enveloping the steel reinforcement; following the approach
by Rehm [287]. The concrete-reinforcement bond length was precisely defined to be 16
mm and 32 mm. The bond shear strengths of printed reinforcement were approximately
20% lower than those of ribbed B500. The shear-stress displacement curves indicated
a pronounced displacement-softening of the bond for B500, while the corresponding
curves for 3D-printed bars showed plateau no plateau could be expected. after reaching
the maximum stress level. The plateau indicates ductile bond failure, which is beneficial
with respect to structural design. Full description of the production process, material
specifications, complying standards and further analyses of results are to be referred to
in Mechtcherine et al. [29].

2A stress corresponding to strain of 0.2% is given

168
Complementary research 9.1. 3D-printing of steel reinforcement

a) b)
Figure 9.4: a) Stress–strain curves from uniaxial tension tests: fracture surfaces are shown in
insets, b) shear stress–displacement curves from pull-out tests with reinforcement-concrete
bond length of 32 mm.
Limitations and outlook:
A brief overview of approaches for integrating reinforcement in digitally produced
concrete structures is presented in Section 2.2.8. Various presented concepts and pro-
totypes are encouraging, however, none of these approaches satisfies the entire spectrum
of relevant requirements for wide ranging implementation of DC and needs to be
developed further. This is also true for the approaches of PSHCC and 3D-printing of
steel reinforcement. The experimental investigations briefly presented in this section are
promising, however, following statements must be considered in further research.
• The PSHCC developed was tested only in one loading direction — loading parallel
to fibre, layer orientation. The mechanical performance when loading is acting
perpendicular and/or lateral to the layer orientation is yet to be tested and likely
will not exhibit strain-hardening behaviour.
• Extrusion of PSHCC was carried always at a PSHCC age of 20 to 45 min.
Time-dependent variation of the rheological properties, especially, extrudability is
yet to be investigated.
• In some cases, fibre balling/whirling around the feeding element in the 3D-printer
hopper was observed. Extensive experiments need to be carried out to identify in
which rheological state and at which PCP operating conditions blockages occur.
• Influence of fibre addition on the layer-interface properties is yet to be investigated.
• Using a ram extruder instead of a PCP for the extrusion of PSHCC is a potential
alternative to avoid fibre balling and blockages. However, ram extruders are
discontinuous in nature and generally heavy to be used in a printhead.
• The printed reinforcement and concrete bond can be further improved with process
optimization, e.g., by printing steel bars with ribs or special anchorages.
• The promising performance of printed steel rebar will be further developed as
a solution to automated printing of both steel reinforcement and concrete in
synchronization.

169
Complementary research 9.2. On interface tensile strength and distinct layer casting

9.2 Interface tensile bond strength investigations and dis-

tinct layer casting
Complementary to the research results (Series 1) presented in Chapter 8, the following
study was carried out with the support of two student thesis works [288, 289]:
Series 2: Design and evaluation of a simple casting-based test approach;
Series 3: Uniaxial tensile tests on both the Mixtures C1 and C2.
In Series 2 and Series 3, time intervals (TI) of 2 min, 10 min and 1 day were studied.
Only the SCM mixture (C2) was studied in Series 2 by means of two tensile test setups:
peripheral glueing of metal stamps and lateral glueing of metal plates. However, in Series
3 both C1 and C2 were investigated only with the lateral glueing technique. The Series
2 and Series 3 investigations were carried out at a much later time than Series 1 and the
material batches for all constituents had to be changed. Therefore, the direct comparison
of the Series 2 and Series 3 with Series 1 may not be fully accurate. However, since no
direct tensile tests were carried out in Series 1, a provisional comparison is made here
between flexural bond test results of Series 1 (Figure 8.5) and tensile test results of Series
3 (Figure 9.5a). Tensile bond strength of both tested mixtures decreased with increasing
TI following same trend as observed in the bending tests of Series 1 and in agreement
with the corresponding microscopic observations. The cement only Mixture C1 has more
pronounced decrease of the bond strength with increasing TI when compared to Mixture
C2. In both cases the tensile bond strength could be expressed as a 2nd order polynomial
a · T I 2 − b · T I + c. Here, TI is the time interval between layers and a, b and c are constants.

Notes on test setups

The primary objective to perform Series 2 was to identify reliable testing procedure for
the layer-interface studies in context of DC. Furthermore, a simple specimen production
setup was conceptualized that is sufficiently representative of the 3D-printed specimens,
however, is much less time and resource consuming. Note that in the investigated
3D-printed specimens contained always six layers. The interface-between the third and
the fourth layer is always printed with the target TI, while others were printed with 2
min TI; see Section 8.1.3 for detailed discussion on this choice. Tensile tests on the 25 mm
X 25 mm X 120 mm specimens of better performing mixture, C2, with peripheral glueing
of metal stamps lead to following conclusions:
a) Since the entire length of the specimens is subjected to tensile loads the specimens
often failed at unanticipated locations, likely due to defects in specimen such as
entrapped air voids, micro-cracks due to shrinkage on the top layers.
b) In case where the interfaces were strongly bonded, the glue at the peripheral failed
making the tests inconclusive. Furthermore, failure of specimen close to the glued
area was also observed. This could potentially find its origins in stresses induced by
hardening of the glue.
c) Glued ends of the specimens must be parallel to avoid eccentricity. This necessitates
polishing of both edges, which often lead to failure of specimens at the interface.
d) The extruded layers and their interfaces are not always perfectly horizontal. Since
all interfaces are under stress, the chance of eccentricity is high in the peripheral
glueing setup.
Majority of the above listed issues could be solved with the lateral glueing approach.
In this case, four metal plates are glued laterally to the specimen as shown in Figure 9.5c.
This results in effective failure localization in the target interface. Naturally, with this
approach, the rate of successful measurement of tensile bond strength was 100 %.

170
Complementary research 9.2. Interface tensile strength

F F
C1
Tensile bond strength [MPa] 4 C2

0
2 min 10 min 1 day
Time interval F F
a) b) c)
Figure 9.5: a) Tensile bond strengths, b) peripheral glueing and c) lateral glueing setups for
tensile bond strength test. 1 Target interface, 2 displacement sensors.

The primary approach followed for the interface studies in this research is 3D-printing
large walls and cutting numerous prism specimens, which are tested under flexural and
tensile loads. Specimen printing with the 3DPTD requires on average 5 hours (excluding
preparation and post-processing), 30-50 litres materials and 4-5 workers. Extraction of
prism specimens requires cutting with wet-grinding blade, which often induces “some”
micro-cracks in the printed specimens. If extensive parameter combination is under
investigation, a simpler and expedite test setup is preferable; one such setup is presented
below.
In the proposed setup two distinct layers are cast in each of the two halves of a wooden
mould; see Figure 9.6a. Both mould halves could be easily fixed to a support platform.
The top mould-half is supported by a retractable base-plate, and can slide in the first
step of the production, concrete is cast in the bottom mould-half. Shortly before the
designated time interval, second mould-half is cast with concrete and then placed on top
of the first one. In the third step, exactly when the designated time-interval is reached,
the base-plate is retracted. As a result, simulation of layered-construction is achieved
by casting. Note that various designs of the moulds and also casting procedures were
investigated.
One critical difference between printed and cast specimens is that the process-induced
changes in material rheology and density are different for both cases. Another significant
difference is that in case of casting there exist a restraint (from the inner walls of the
wooden moulds) on the top half of the concrete. The restraint reduces the downward
pressure from the cast concrete and was found to diminish the interface bond strength
significantly. To determine the restraint force the mould filled with concrete was hanged
1 cm above a weight scale. The setup was such that the concrete is “freely-hanged” and
can slide downward from the mould on to the scale. The restraint force is calculated
as the difference in the measured weight and weight of the concrete without mould.
Subsequently, the restraint weight is added on top of the cast specimen. To assess the
representativeness of cast specimens, their tensile bond strengths are compared to the
one from 3D-printed specimens; see Figure 9.6b. The cast specimen inner cross-section
was 40 mm X 40 mm. In case of 3D-printed specimens the cross-section was 47 mm X 30
mm. All other parameters were kept identical for both printing and casting. Production
of the specimens was carried out in parallel from the same batch of 3PC and tensile tests
were carried out at an age of 28 days.

171
Complementary research 9.2. Interface tensile strength

4
3D-printed

Tensile bond strength [MPa]

Step 2
2 3 Cast

1
2

Step 1 1

1
2 0
3 2 min 10 min 1 day
Time interval
a) b)
Figure 9.6: a) Moulds for casting of 3PCs: 1 lower mould-half, 2 upper mould-half, 3
retractable base-plate, b) tensile bond strengths obtained from testing 3D-printed and cast
specimens with lateral glueing setup.

The tensile-bond strength results from printed and cast specimens are presented in
Figure 9.6b. Primary conclusions are:
a) Tensile bond strengths from the cast specimens follow the same trend as those of
printed specimens.
b) Tensile bond strengths of cast specimens are mostly lower than those of printed
specimens.
Observed low strengths of cast specimens may have their origins in applied vibration
and the restraint from the moulds. Controlled vibration at 50 Hz for 30 s was necessary to
ensure mould filling considering the non-flowable consistency of the 3PCs. The vibration
results in formation of smooth paste surface on the surface of bottom mould-half. The
upper mould-half is always cast separately and only placed on the first half — similar
to 3D-printing. The mould-halves were not vibrated after they are connected. Substrate
roughness is one of the key parameters that influences layer-bond strengths [241–243].
The smoother surfaces at the interface is believed to have lead to the weak interfaces.
Figure 9.6b also shows that 1 day bond strengths of cast specimens are same as those of
printed specimens. Remarkably, the strength reduction from 10 min TI to 1 day TI of cast
specimens is only 15 %. Whereas, the same in the case of printed specimens is 62 %. The
water evaporation in printed specimens is significantly higher as they are not protected
by the moulds. The water lost (evaporated ) would be pronounced in the case of 1 day
TI printed specimens. The dry substrate sucks water from overlay leading to differential
shrinkage causing reduction in interface bond area and shrinkage cracks; see Figure 2.16.
As a consequence, in the case of 1 day TI the printed specimens resulted in a higher bond
strength reduction than cast ones.
Overall, the simple distinct layer casting setup showed qualitative agreement with the
printed specimens and therefore can be used as an alternative test method. However,
differences such as that in water evaporation must be considered while using simple test
setup; especially in case of long time intervals. The procedure could be improved by
sliding out the cast concrete from the moulds (right after casting) and by (quantifiable)
roughening the substrate surface. Instead of peripheral glueing, the lateral glueing
technique is recommended for further tensile bond strength investigations.

172
Conclusions and further research

The processing of cementitious materials is the technological core of modern construc-

tion. Construction processes — which are conventionally severely impeded by low pro-
ductivity, scarcity of skilled labour and limited geometrical freedom — are on the verge
of a potential revolution with the emergence of extrusion-based 3D-concrete-printing.
Such advanced digital construction (DC) approaches open up new horizons as they
derive efficient and faster construction processes which are formwork-free, less labour
intensive, autonomous and offer quasi-limitless geometrical possibilities. However, the
major significance and revolutionary potential of DC reveal themselves in the context
of Construction Industry 4.0. DC represents a logical, decisive step arising from the
already well-developed tools of digital design and planning (CAD, BIM, etc.) towards
digital manufacturing.Digitalisation, Internet of Things and automation at its core,
the imminent fourth industrial revolution is going to cause drastic changes across
industries. In effect paradigm shifts are expected in the way we live and the way we
construct buildings. Among current DC techniques, the extrusion-based DC is the most
practical one for large-scale and on-site construction with enormous social, economic and
technical potential. When fully developed, the extrusion-based DC enables cost-effective,
customized housing for the masses in developing countries, significant productivity
improvements in developed countries and autonomous construction in remote and
dangerous locations.
To convert the promises of DC into practical applications, at first, numerous current
challenges need to be solved, specifically those related to 3D-printable cementitious ma-
terials (3PCs). Within the scope of this doctoral research, the printability was defined as
“a time-dependent compound property of pumpability, extrudability and buildability”.
Furthermore, the hardened state property of interface bond strength is crucial. Overall,
the 3PCs for extrusion-based large-scale DC should fulfil conflicting requirements.
Extensive experimental studies are needed to develop 3PCs with optimum properties
in their fresh and hardened states. Conventional experimental methods are inadequate
to verify process-specific properties of 3PCs. Despite the overwhelming research and
academic interest, no standard approaches are available yet for all the sub-properties
of printability.
The primary accomplishment of this PhD research is an approach for comprehensive
and quantitative characterisation of high-performance 3PCs with novel and contempo-
rary methodologies. For this purpose, various experimental and computational tools were
utilised; while links to process and machine aspects were directly established wherever
appropriate. A 3D-printing test device (3DPTD) with a progressive cavity pump (PCP), a
simplified ram extruder and multiple measuring methods and protocols were developed.
The synchronisation of printhead velocity and PCP discharge rates was delineated.
The 3PCs developed in this research are made of common construction materials and
fulfilled the complex process specific requirements. From the extensive experimental
investigations and simulations, following major conclusions can be drawn.

173
Complementary research 9.2. Interface tensile strength

Conclusions
Structural build-up (SBP) investigations: For stiffer pastes, suitable for 3D-printing
with concrete, the use of very low shear rates was found to be much more challenging
than for flowable concretes with a low yield stress, such as SCC, which are commonly
investigated for SBP in the reported literature. With analyses of rheometrical results,
it was demonstrated that if a constant strain-based approach is followed, even if the
applied shear rate γ̇ap is varied and measurement duration td is adapted for the
sake of compensation, the structural build-up can be successfully measured without
significant variation in the calculated structuration rate. Two protocols are suggested for
characterising SBP of high-yield stress 3PCs:
a) The conventional constant shear rate test with a very low γ̇ap and a constant but
long td satisfying the precondition that flow onset must take place.
Note that, a long enough td must be ensured so that static yield stress can be measured
for all compositions under investigation. This approach is not valid if the structuration
rate of the material is very high leading to the rate of increase of effective strain during
the experiment to be insufficient to cause flow onset. The approach is also not valid if a
single specimen (i.e. single batch and single cell) procedure is used.
b) Alternatively, by considering effective strain, the universal stress-growth test based
on constant strain, satisfying the condition: γef ≥ γef ,min , can be followed. In
this novel approach, γ̇ap , td and consequently γap can be varied according to the
particular material under investigation and its tage .
Furthermore, influence of the cement paste composition on structural build-up
was investigated by applying the strain-based protocol. Mixtures with partial cement
replacement by SCM exhibited the highest structuration rate in this study, primarily due
to the increase in flocculation. It appears that using SCM is an appropriate approach to
improve printability of 3PCs.

Concrete placing and pumping: The distinct aspect of pumping in DC in comparison

with conventional pumping is that the former is often discontinuous, periodic3 . The high
τo
µ — which is common if initial yield stress of 3PCs is high — implies that the flow
profiles during pumping would be predominantly plug flow. Since most 3PCs are highly
thixotropic, drastic increase in discharge pressure would occur if the pumping process
is interrupted for a long duration. As a consequence, the two-stage approaches where
low viscous and low yield stress pumpable base mixture is first pumped to printhead
reservoir and then activated (either by mixing a phase changing agent or by external
triggering) are of significance for large-scale DC. So far, pumpability of 3PCs has not
been studied and having a digital pumpability testing tool will accelerate investigations
on this significant topic.
A numerical tool to simulate sliding pipe rheometer tests — virtual Sliper — was
developed using the single phase Computational Fluid Dynamics method. When the
lubrication layer (LL) was not considered, the applicability of the initial model was
limited to the scenarios in which the formation of LL was not pronounced. Model
enhancement by implementing LL, with the User Defined Function (UDF) of ANSYS
Fluent® enabled the estimation of the thickness, yield stress and plastic viscosity of the
LL for different concrete compositions. Both the thickness and plastic viscosity of the LL
have pronounced influences on concrete pumpability, while the yield stress parameter
showed a negligible effect. The parameter study corroborated the followed approach for
calculating the viscosity of the lubricating layer, as a function of concrete viscosity, using
3 For example, stopping after having printed each layer.

174
Complementary research 9.2. Interface tensile strength

the Krieger-Dougherty model. The developed virtual pumpability testing tool should
enable purposeful material design of pumpable concrete.

Investigations on extrudability of 3PCs: An optimal extrudability test method for 3PCs

should be characterised by attributes such as quantitative, inline, continuous, geometry
independent and high potential for complete automation. Furthermore, such a method
should be capable to quantify material (extrudate) flow rate. In light of this, a new
methodology — the 3DPET — is proposed for characterising the extrudability of 3PCs,
quantitatively, inline and with minimal geometry-induced artefacts. Extrudability of two
compositions were investigated combining multiple test methods. While the results from
3DPET, viscometer, HFT tests and visual observation indicated lower extrudability for the
stiffer, high yield stress and more fine-grained Mixture A-HF, than for the low yield stress
Mixture A, the ram extrusion force REF values implied the opposite. Further analysis
confirmed that in case of ram extrusion, due to plug flow, the bulk yield stress played
less significant role. In ram extruders, unlike PCPs the blockage of sand grains between
piston-head and cylinder resulted in higher REF values.
It is important to use a test device similar in its working principle to the device
used in perspective, full-scale application to avoid geometry-induced artefacts in
extrudability characterisation. The occurrence and significance of phenomena such as
filtration/drainage, slippage and consolidation vary depending upon the extruder type
and geometry. Note that, both for PCPs and simple-screw pumps slippage occurs. The
3DPET is the only method that can give a direct measurement of the effective volume
flow rate of extrudate for a given control parameter.

Buildability of 3PCs: Buildability is the most crucial requirement of 3PCs in fresh

and transient states. The rate of vertical progression f (T I) of printed structures and
their final geometric precision depend directly on the buildability. The layer-interface
properties, extrudability and pumpability are influenced by the means of achieving
adequate buildability (e.g. with high initial yield stress). Buildability depends on, but
is not identical to, the structural build-up of cement-based materials. The geometry of
printed structures, ambient conditions and process parameters also have a pronounced
influence on the buildability — all of which are distinguished by the intended DC
application. As a consequence, it is crucial to consider the target application while
characterising buildability of 3PCs. A printed structure is buildable as long as its shear
flow resistance is higher than the shear loads acting on it. In rheology based approach,
the flow resistance is expressed as time-dependent static yield stress in conjunction
with a geometric factor. Note that the role of buckling can be critical and increases
with increasing slenderness of the structure. Predicting failure of printed structures
necessitates precise and extensive experimental and numerical investigations.
State of the art on buildability investigations focuses on determining the extremes
of a 3PCs buildability without considering a particular application. In such cases,
the suitable application has to be selected for the 3PCs. Alternatively, 3PCs could be
specifically developed suiting for a particular application. Thus, materials are selected
and optimised matching the structural, machine and process requirements including
economic viability. Such an application-oriented approach was introduced in the research
at hand. The proposed approach takes various process aspects and construction costs
into consideration. In doing so, direct links between laboratory buildability tests and
target applications are established. A basis for calculating the minimum height and time
interval to be followed during laboratory testing is elaborated. Comparative analyses
of full-width printing (FWP) and filament printing (FP) processes revealed that higher

175
Complementary research 9.2. Interface tensile strength

velocities of the printhead should be established for laboratory buildability tests of

a material for FP processes than those needed for FWP process, providing for equal
construction rates.
The economic viability approach under consideration is independent of the presented
simple buildability test (direct printing test). Therefore, it can be applied stand-alone to
identify the maximum TI between the layers based on economic viability. Furthermore,
the approach can be used with any buildability tests or numerical models, e.g., [7, 8, 17].
It enables determination of the minimum yield-stress and the rate of structural build-up
at which concrete should have for economically viable DC.

Interface and hardened properties: Formulating high compressive strength 3PCs is

trivial, due to their high binder content, low w/c and usually dense matrix. Not so straight
forward is the prevention of weak layer-interfaces. The weak interfaces negatively affect
the printed structures’ mechanical performance, diminish their durability and therefore
threaten their structural integrity. Systematic mechanical tests and SEM investigations
on layer-interface bonds were performed on 3PCs with varying binder composition. The
3PCs’ composition, followed by layer deposition time interval, influenced the interface
bond strength pronouncedly. In addition, due to higher porosity, higher amount of large
pores/microcracks (at the interface and in the core) and possibly also due to higher
drying and plastic shrinkage the mixture with Portland cement as sole binder exhibited
weaker layer-interfaces and considerable reduction in bond strengths with increasing
time intervals. Remarkably, the weak-layer interfaces also influence the compressive
strengths of 3PCs when tested in multiple directions; see Section 8.2.3. Microstructure
inspections corroborated findings from mechanical tests and revealed air cavities at
the interfaces. Overall, the mixture with SCMs yielded superior bond strengths and
self-healing ability.

Further research
Rheology: A comparative study on the rheological behaviour of 3PCs using different
approaches including rotational rheometry, penetration resistance and non-destructive
tests is of high significance. In addition, a comparative study on structural build-up
quantification by single-batch, multi-batch constant-shear rate tests, hysteresis tests and
oscillation tests would be beneficial. Proposed strain-based approach shall be further
extended to stiff, printable concretes containing coarse aggregates. While doing this,
plug flow and shear-localisations shall be investigated using such techniques as magnetic
resonance imaging and computer tomography as well as accompanying image analysis.

Pumpability: Applicability of the developed CFD model for full-scale pumping was
already validated by the subsequent research at the Institute of Construction Materials,
TU Dresden [25] in which experimentally measured properties of LL were used as
input, instead of the theoretically calculated ones (Section 5.5.1). On the one hand, the
single-phase model could be further improved by testing it for high yield-stress and
low viscous 3PCs. On the other hand, an open-source implementation of the developed
model and its subsequent free provision on a public domain would serve the purpose of
a virtual concrete laboratory.

Extrudability: Further work is envisioned to measure torque at the rotor of printhead

motor, and to establish an analytical model linking the extrudability of cement-based
materials to their rheological properties and to geometrical configurations of the

176
progressive cavity pump. It is important to measure the exact rotational velocity
of the PCP resulting from an applied value of 3D-printer control parameter RCP1.
This would enable analytical determination of extrudate flow rate and volumetric
efficiency; see Equation (2.14). In the current approach, the flowrate of the extrudate is
measured manually, this could be replaced with a flow-meter placed in the nozzle-outlet.
This would make the proposed approach fully autonomous and enable continuous
extrudability characterisation, every time the 3D-concrete-printer is operated.

Buildability: Since the developed Mixture C2 could pass the buildability test with a
much shorter TI than the calculated maximum TI, further optimisation of the mixture
can be carried out. A less buildable mixture with, likely, lower cement, additives and
admixtures contents should be composed. Proposed buildability testing approach needs
to be validated by means of full-scale printing tests with printable concretes containing
aggregates with a maximum size of 8 mm. Investigating the influence of various ambient
conditions (temperature, humidity, wind velocity) on the buildability of 3PCs is of high
significance. Continuous measurement of layer deformation with optical sensors or by
image analysis enables precise assessment of layer deformation during printing and
could be implemented as a process monitoring technique.

Interface: More research is needed to be able to develop appropriate strategies for

mitigating the formation of weak interlayer joints by adjusting the material composition
and process parameters as well as by using some additional measures such as internal
or external curing and application of a low-viscosity mineral-based primer before
deposition of the subsequent layer. Further insights into the layer-interface behaviour
can be gained by extending the experimental program to other investigation methods,
including capillary suction quantification [240] and evaporation measurements.
Bibliography

[1] R. A. Buswell, R. C. Soar, A. G. F. Gibb, A. Thorpe, Freeform Construction:

Mega-scale Rapid Manufacturing for construction, Automation in Construction
16 (2) (2007) 224–231. doi:10.1016/j.autcon.2006.05.002.
[2] T. T. Le, S. A. Austin, S. Lim, R. A. Buswell, A. G. F. Gibb, T. Thorpe, Mix design
and fresh properties for high-performance printing concrete, Materials and
Structures 45 (8) (2012) 1221–1232. doi:10.1617/s11527-012-9828-z.
[3] J. Pegna, Exploratory investigation of solid freeform construction, Automation in
Construction 5 (5) (1997) 427–437. doi:10.1016/S0926-5805(96)00166-5.
[4] W. E. Urschel, Machine for building walls, USA Patent US2339892 (1941).
[5] C. Scott, Chinese Construction Company 3D Prints an Entire Two-Story House
On-Site in 45 Days (2016).
URL https://3dprint.com/138664/huashang-tengda-3d-print-house/
[6] N. P. Hack, A robotically fabricated structural stay-in-place formwork system,
Ph.D. thesis, ETH Zurich (2018). doi:10.7134/phlebol.14-06.
[7] A. Perrot, D. Rangeard, A. Pierre, Structural built-up of cement-based materials
used for 3D-printing extrusion techniques, Materials and Structures 49 (4) (2016)
1213–1220. doi:10.1617/s11527-015-0571-0.
[8] R. Wolfs, F. Bos, T. Salet, Early age mechanical behaviour of 3D printed concrete:
Numerical modelling and experimental testing, Cement and Concrete Research
106 (May 2017) (2018) 103–116. doi:10.1016/j.cemconres.2018.02.001.
[9] D. Kaplan, Pumping of concrete, PhD thesis [in French], Ph.D. thesis, Laboratoire
Central des Ponts et Chaussées, Paris (2001).
[10] V. Mechtcherine, V. N. Nerella, K. Kasten, Testing pumpability of concrete using
Sliding Pipe Rheometer, Construction and Building Materials 53 (2014) 312–323.
doi:10.1016/j.conbuildmat.2013.11.037.
[11] V. N. Nerella, S. Hempel, V. Mechtcherine, Micro- and macroscopic investigations
on the interface between layers of 3d-printed cementitious elements, in:
M. Santhanam, R. Gettu, R. G. Pillai, S. K. Nayar (Eds.), International Conference
on Advances in Construction Materials and Systems (ICACMS), Vol. 3, RILEM
Publications SARL, Chennai, 2017, pp. 557–565.
[12] V. Nerella, M. Krause, M. Näther, V. Mechtcherine, 3D printing technology for
on-site construction, Concrete Plant International (4) (2016) 36–41.
[13] G. De Schutter, Thixotropic Effects During Large-scale Concrete Pump Tests on
Site, in: M. Santhanam, R. Gettu, R. G. Pillai, S. K. Nayar (Eds.), International
Conference on Advances in Construction Materials and Systems (ICACMS), Vol. 2,
RILEM Publications SARL, Chennai, 2017, pp. 1–7.
[14] G. De Schutter, K. Lesage, Active control of properties of concrete: a (p)review,
Materials and Structures 51 (5) (2018). doi:10.1617/s11527-018-1256-2.
[15] L. Reiter, T. Wangler, N. Roussel, R. J. Flatt, The role of early age structural
build-up in digital fabrication with concrete, Cement and Concrete Research
(2018). doi:https://doi.org/10.1016/j.cemconres.2018.05.011.

179
[16] D. Marchon, S. Kawashima, H. Bessaies-Bey, S. Mantellato, S. Ng, Hydration and
rheology control of concrete for digital fabrication: Potential admixtures and
cement chemistry, Cement and Concrete Research 112 (December 2017) (2018)
96–110. doi:10.1016/j.cemconres.2018.05.014.
[17] A. S. Suiker, Mechanical performance of wall structures in 3D printing processes:
Theory, design tools and experiments, International Journal of Mechanical
Sciences 137 (January) (2018) 145–170. doi:10.1016/j.ijmecsci.2018.01.010.
[18] D. Kaplan, F. D. Larrard, T. Sedran, Design of Concrete Pumping Circuit, ACI
Materials Journal 102 (2) (2005) 110–117.
[19] C. F, Fundamental and practical study on the pumping of concrete. Ph.D.
Dissertation., Ph.D. thesis, Université Laval, Quebec (2007).
[20] S. Jacobsen, J. H. Mork, S. F. Lee, L. Haugan, Pumping of concrete and mortar –
State of the art report, Tech. rep., Consortium Concrete Innovation Center, Oslo
(2008).
[21] D. Feys, Interactions between rheological properties and pumping of
self-compacting concrete, Ph.D. thesis, Ghent University (2009).
[22] K. Kasten, Gleitrohr – Sliding Pipe Rheometer: A method to establish the flow
properties of high viscous media in pipelines, Ph.d, TU Dresden (2010).
[23] M. S. Choi, Y. J. Kim, S. H. Kwon, Prediction on pipe flow of pumped concrete
based on shear-induced particle migration, Cement and Concrete Research 52
(2013) 216–224. doi:10.1016/j.cemconres.2013.07.004.
[24] S. H. Kwon, P. J. Kyong, J. H. Kim, P. S. Surendra, State of the Art on Prediction of
Concrete Pumping, International Journal of Concrete Structures and Materials
10 (3) (2016) 75–85. doi:10.1007/s40069-016-0150-y.
[25] E. Secrieru, Pumping behaviour of modern concretes Characterisation and
prediction, Ph.D. thesis, Technische Universität Dresden (2018).
[26] V. Nerella, M. Näther, A. Iqbal, M. Butler, V. Mechtcherine, Inline quantification
of extrudability of cementitious materials for digital construction, Cement and
Concrete Composites 95 (1) (2019) 260–270.
doi:10.1016/j.cemconcomp.2018.09.015.
URL https://linkinghub.elsevier.com/retrieve/pii/S0958946518302956
[27] V. Nerella, M. Beigh, S. Fataei, V. Mechtcherine, Strain-based approach for
measuring structural build-up of cement pastes in the context of digital
construction, Cement and Concrete Research 115 (2019) 530–544.
doi:10.1016/j.cemconres.2018.08.003.
URL https://linkinghub.elsevier.com/retrieve/pii/S0008884618301650
[28] G. De Schutter, K. Lesage, V. Mechtcherine, V. N. Nerella, G. Habert,
I. Agusti-Juan, Vision of 3D printing with concrete – Technical, economic and
environmental potentials, Cement and Concrete Research 112 (November 2017)
(2018) 25–36. doi:10.1016/j.cemconres.2018.06.001.
[29] V. Mechtcherine, J. Grafe, V. N. Nerella, E. Spaniol, M. Hertel, U. F 3d-printed
steel reinforcement for digital concrete construction – manufacture, mechanical
properties and bond behaviour, Construction and Building Materials 179 (2018)
125 – 137. doi:https://doi.org/10.1016/j.conbuildmat.2018.05.202.
[30] V. N. Nerella, H. Ogura, V. Mechtcherine, Incorporating reinforcement into digital
concrete construction, in: C. Mueller, S. Adriaenssens (Eds.), Annual IASS
Symposia: Creativity in structural design, International association for shell and
spatial structures, Cambridge, Massachusetts, 2018, pp. 1–8.
[31] V. N. Nerella, S. Hempel, V. Mechtcherine, Effects of layer-interface properties on
mechanical performance of concrete elements produced by extrusion-based
3D-printing, Construction and Building Materials 205 (2019) 586–601.
doi:10.1016/j.conbuildmat.2019.01.235.
URL https://doi.org/10.1016/j.conbuildmat.2019.01.235https:
//linkinghub.elsevier.com/retrieve/pii/S0950061819302843
[32] H. Klee, Recycling Concrete, Tech. Rep. June, World Business Council for
Sustainable Development (2009). doi:10.1680/ensu.2004.157.1.9.
[33] CEMBUREAU, The-European-Cement-Association, Key facts & figures (2018).
URL https://cembureau.eu/cement-101/key-facts-figures/
[34] H. Nasir, H. Ahmed, C. Haas, P. M. Goodrum, An analysis of construction
productivity differences between Canada and the United States, Construction
Management and Economics 32 (6) (2013) 1–13.
doi:10.1080/01446193.2013.848995.
[35] David Malin Roodman, Nicholas Lenssen, A Building Revolution: How Ecology
and Health Concerns Are Transforming Construction – Worldwatch Institute,
Worldwatch Institute, 1995.
URL http://www.worldwatch.org/node/866
[36] L. Klotz, M. Horman, M. Bodenschatz, A lean modeling protocol for evaluating
green project delivery, Lean Construction Journal 3 (1) (2007) 1–18, cited By :45.
[37] N. Müller, J. Harnisch, A blueprint for a climate friendly cement industry, Tech.
rep., WWF International, Ecofys Germany GmbH, Nürnberg (2008).
URL
http://scholar.google.com/scholar?hl=en{&}btnG=Search{&}q=intitle:
A+blueprint+for+a+climate+friendly+cement+industry{#}0
[38] R. Mark, P. Hutchinson, S. The, A. Bulletin, N. Mar, R. Mark, P. Hutchinson, On
the Structure of the Roman Pantheon, The Art Bulletin 68 (1) (1986) 24–34.
doi:10.2307/3050861.
[39] R. Schach, M. Krause, M. Näther, V. N. Nerella, CONPrint3D:
3D-Concrete-Printing as an Alternative for Masonry, Bauingenieur 9 (2017)
355–363.
[40] R. Schmitt, Die Schalungstechnik, 1. Auflage, 1st Edition, Ernst u. Sohn, Berlin,
2001.
[41] W. Chen, J. R. Liew, The civil engineering hand book, 2nd Edition, CRC Press
LLC, New York, 2003.
[42] Henry Mwanaki Alinaitwe, Jackson A. Mwakali, Bengt Hansson, Factors affecting
the productivity of building craftsmen-studies of Uganda – umar muhammad –
Academia.edu, Journal of Civil Engineering and Management XIII (3) (2007)
169–176.
[43] S. G. Naoum, Factors influencing labor productivity on construction sites,
International Journal of Productivity and Performance Management 65 (3) (2016)
401–421. doi:10.1108/IJPPM-03-2015-0045.
[44] B. Green, Productivity in Construction : Creating a Framework, Tech. rep., The
Chartered Institute of Building (2016).
[45] R. Mohan, S. Dasgupta, Urban Development in India in the 21 st Century :
Policies for Accelerating Urban Growth, in: Fifth Annual Conference on Indian
Economic Policy Reform, no. 231, Stanford Center for International Development,
2004, pp. 1–63.
[46] United Nations, World Urbanization Prospects, Department of Economic and
Social Affairs, United Nations, 2014.
URL http://esa.un.org/unpd/wup/Highlights/WUP2014-Highlights.pdf
[47] European Network for Housing Research, Housing in Developing Countries
(2017).
URL https://www.enhr.net/housingindeve.php
[48] Rethinking construction by the construction task force, Tech. Rep. 14,
Department of Trade & Industry, U. K. (1998).
[49] ASTM F2792 – 12a Standard Terminology for Additive Manufacturing
Technologies (2012). doi:10.1520/F2792-12A.
[50] DIN, DIN 8580 – Manufacturing processes – Terms and definitions, division
(2003).
URL https://www.beuth.de/en/standard/din-8580/65031153
[51] P. Wu, J. Wang, X. Wang, A critical review of the use of 3-D printing in the
construction industry, Automation in Construction 68 (2016) 21–31.
doi:10.1016/j.autcon.2016.04.005.
[52] K. Henke, Additive Baufertigung durch Extrusion von Holzleichtbeton, Ph.D.
thesis, Technische Universität München (2016).
[53] Types of 3D printers or 3D printing technologies overview – 3D Printing from
scratch (2015).
URL http://3dprintingfromscratch.com/common/types-of-3d-printers-
or-3d-printing-technologies-overview/
[54] Seven Types of 3D Printers – Different printing and extruder technologies (2017).
URL https://penandplastic.com/3d-printer-types/
[55] Sara Zaske, Germany’s vision for Industrie 4.0: The revolution will be digitised –
ZDNet (2016).
URL http://www.zdnet.com/article/germanys-vision-for-industrie-4-0-
the-revolution-will-be-digitised/
[56] S. Peters, Additive Manufacturing – The path toward individual production, Tech.
rep. (2015).
[57] C. Anderson, S. Schmid, Makers das Internet der Dinge: die nächste industrielle
Revolution, Hanser, 2013.
[58] Germany trade and invest – GTAI, Industrie 4.0 – Smart manufacturing, Tech.
rep. (2016).
[59] B. Khoshnevis, M. Bodiford, K. Burks, E. Ethridge, D. Tucker, W. Kim, H. Toutanji,
M. Fiske, Lunar contour crafting–a novel technique for ISRU-based habitat
development, Automation in construction 13(1) (January) (2004) 5–19.
[60] Urschel wall building machine (Mar 2017).
URL https://www.youtube.com/watch?v=QXqwnJTVSsE
[61] E. Dini, Monolite-UK-Ltd, D-Shape (2015).
URL https://d-shape.com/
[62] C. Gosselin, R. Duballet, P. Roux, N. Gaudillière, J. Dirrenberger, P. Morel,
Large-scale 3D printing of ultra-high performance concrete – a new processing
route for architects and builders, Materials & Design (2016)
102–109doi:10.1016/j.matdes.2016.03.097.
[63] R. Duballet, O. Baverel, J. Dirrenberger, Classification of building systems for
concrete 3D printing, Automation in Construction 83 (2017) 247–258.
doi:10.1016/j.autcon.2017.08.018.
[64] E. Lloret-Fritschi, Smart Dynamic Casting – A digital fabricaiton method for
non-standard cocnrete structures, Phd, ETH Zurich (2016).
doi:10.3929/ethz-a-010782581.
[65] R. Buswell, 3D printing using concrete extrusion: a roadmap for research, Cement
and Concrete Research 112 (October 2017) (2018) 37–49.
doi:10.1016/j.cemconres.2018.05.006.
[66] J. Pegna, Application of Cementitious Bulk Materials to Site Processed Freeform
Construction, in: 6th Solid Freeform Fabrication (SFF) Symposium, Austin, Texas,
1995, pp. 39–45.
[67] B. Khoshnevis, Contour Crafting – State of Development, Solid Freeform
Fabrication Proceedings (1999) 743–750.
[68] A. Kazemian, X. Yuan, E. Cochran, B. Khoshnevis, Cementitious materials for
construction-scale 3D printing: Laboratory testing of fresh printing mixture,
Construction and Building Materials 145 (2017) 639–647.
doi:10.1016/j.conbuildmat.2017.04.015.
[69] B. Zareiyan, B. Khoshnevis, Interlayer adhesion and strength of structures in
Contour Crafting – Effects of aggregate size, extrusion rate, and layer thickness,
Automation in Construction 81 (2017) 112–121.
doi:10.1016/j.autcon.2017.06.013.
[70] S. Lim, R. Buswell, T. Le, S. Austin, A. Gibb, T. Thorpe, Developments in
construction-scale additive manufacturing processes, Automation in Construction
21 (1) (2012) 262–268. doi:10.1016/j.autcon.2011.06.010.
[71] T. T. Le, S. A. Austin, S. Lim, R. A. Buswell, R. Law, A. G. F. Gibb, T. Thorpe,
Hardened properties of high-performance printing concrete, Cement and
Concrete Research 42 (3) (2012) 558–566.
doi:10.1016/j.cemconres.2011.12.003.
[72] D. Lowke, E. Dini, A. Perrot, D. Weger, C. Gehlen, B. Dillenburger, Particle-bed
3D printing in concrete construction – Possibilities and challenges, Cement and
Concrete Research 112 (May) (2018) 50–65.
doi:10.1016/j.cemconres.2018.05.018.
[73] K. Henke, S. Treml, Wood based bulk material in 3D printing processes for
applications in construction., European Journal of Wood and Wood Products
71 (1) (2013) 139–141.
[74] Beton ohne schalung formen (2013).
URL http:
//www.ethlife.ethz.ch/archive_articles/130325_betonmischung_per/
[75] EP3042008A1 – Method of fabricating a 3-dimensional structure, mesh formwork
element for fabricating a 3-dimensional structure, and method of fabricating the
same.
[76] EMPA/ETHZ, MESH MOULD – DFAB HOUSE.
URL http://dfabhouse.ch/mesh{_}mould/
[77] Totalkustom, 3D printed castle – photo gallery (2015).
URL http://www.totalkustom.com/photo.html
[78] 3ders.org, 10 completely 3D printed houses appears in Shanghai, built under a
day (2014).
URL www.3ders.org/articles/20140401-10-completely-3d-printed-
houses-appears-in-shanghai-built-in-a-day.html
[79] Fraunhofer-IRB, Machbarkeitsuntersuchungen zu kontinuierlichen und
schalungsfreien Bauverfahren durch 3D-Formung von Frischbeton (2014).
[80] V. N. Nerella, V. Mechtcherine, Virtual sliding pipe rheometer for estimating
pumpability of concrete, Construction and Building Materials 170 (2018) 366 –
377. doi:https://doi.org/10.1016/j.conbuildmat.2018.03.003.
URL
http://www.sciencedirect.com/science/article/pii/S0950061818304823
[81] H. Ogura, V. Nerella, V. Mechtcherine, Developing and Testing of
Strain-Hardening Cement-Based Composites (SHCC) in the Context of
3D-Printing, Materials 11 (8) (2018) 1375. doi:10.3390/ma11081375.
URL http://www.mdpi.com/324900
[82] TU Eindhoven, TU Eindhoven starts using kingsize 3D concrete printer (2015).
URL https://www.tue.nl/en/university/news-and-press/news/22-10-
2015-tu-eindhoven-starts-using-kingsize-3d-concrete-printer/
[83] F. Bos, R. Wolfs, Z. Ahmed, T. Salet, Additive manufacturing of concrete in
construction: potentials and challenges of 3D concrete printing, Virtual and
Physical Prototyping (2016). doi:10.1080/17452759.2016.1209867.
[84] T. A. M. Salet, Z. Y. Ahmed, F. P. Bos, H. L. M. Laagland, Design of a 3D printed
concrete bridge by testing, Virtual and Physical Prototyping (2018)
1–15doi:10.1080/17452759.2018.1476064.
[85] Clement Gosselin, XtreeE: PROJECTS – 3D-Printed Concrete Wall.
URL http://www.xtreee.eu/projects-3d-printed-concrete-wall/
[86] RILEM, Nicolas Roussel, 276-DFC : Digital fabrication with cement-based
materials (2016).
URL https://www.rilem.net/groupe/276-dfc-digital-fabrication-with-
cement-based-materials-351
[87] Apis-cor, Apis Cor – construction technology (2017).
URL http://apis-cor.com/en/faq/texnologiya-stroitelstva/
[88] B. Panda, S. C. Paul, N. A. N. Mohamed, Y. W. D. Tay, M. J. Tan, Measurement of
tensile bond strength of 3D printed geopolymer mortar, Measurement: Journal of
the International Measurement Confederation 113 (September) (2018) 108–116.
doi:10.1016/j.measurement.2017.08.051.
[89] A. Pierre, D. Weger, A. Perrot, D. Lowke, Penetration of cement pastes into sand
packings during 3D printing: analytical and experimental study, Materials and
Structures 51 (1) (2018) 22. doi:10.1617/s11527-018-1148-5.
[90] R. J. Flatt, T. Wangler, Editorial for special issue on digital concrete, Cement and
Concrete Research 112 (2018) 1–4. doi:10.1016/j.cemconres.2018.07.007.
[91] N. Roussel, Rheological requirements for printable concretes, Cement and
Concrete Research 112 (April) (2018) 1–10.
doi:10.1016/j.cemconres.2018.04.005.
[92] D. Asprone, C. Menna, F. P. Bos, T. A. Salet, J. Mata-Falcón, W. Kaufmann,
Rethinking reinforcement for digital fabrication with concrete, Cement and
Concrete Research 112 (May) (2018) 111–121.
doi:10.1016/j.cemconres.2018.05.020.
[93] S. Lim, T. Le, J. Webster, R. Buswell, S. Austin, A. Gibb, T. Thorpe, Fabricating
construction components using layered manufacturing technology, in: Global
Innovation in Construction Conference, 2009, pp. 512–520.
[94] I. Agustí-Juan, A. Hollberg, G. Habert, Integration of environmental criteria in
early stages of digital fabrication, in: Proceedings of the 35th International
Conference on Education and Research in Computer Aided Architectural Design
in Europe: ShoCK! Sharing of Computable Knowledge!, Vol. 2, Education and
 Research in Computer Aided Architectural Design in Europe (eCAADe), 2017, pp.
185–192.
[95] D. Asprone, F. Auricchio, C. Menna, V. Mercuri, 3D printing of reinforced
concrete elements: Technology and design approach, Construction and Building
Materials (2018). doi:10.1016/j.conbuildmat.2018.01.018.
[96] G. Vantyghem, M. Steeman, V. Boel, W. D. E. Corte, Multi-physics topology
optimization for 3D-printed structures, in: Proceedings of the IASS Symposium
2018 Creativity in Structural Design, International association for shell and
spatial structures, MIT, Boston, 2018.
[97] Kira, WinSun China builds world’s first 3D printed villa and tallest 3D printed
apartment building (2015).
URL
http://www.3ders.org/articles/20150118-winsun-builds-world-first-3d-
printed-villa-and-tallest-3d-printed-building-in-china.html
[98] S. Lim, R. A. Buswell, P. J. Valentine, D. Piker, S. A. Austin, X. De Kestelier,
Modelling curved-layered printing paths for fabricating large-scale construction
components, Additive Manufacturing (2016).
doi:10.1016/j.addma.2016.06.004.
[99] G. W. Ma, L. Wang, Y. Ju, State-of-the-art of 3D printing technology of
cementitious material–An emerging technique for construction, Science China
Technological Sciences 149 (2017) 1–21. doi:10.1007/s11431-016-9077-7.
[100] N. Labonnote, A. Rønnquist, B. Manum, P. Rüther, Additive construction:
State-of-the-art, challenges and opportunities, Automation in Construction 72
(2016) 347–366. doi:10.1016/j.autcon.2016.08.026.
[101] T. Wangler, E. Lloret, L. Reiter, N. Hack, F. Gramazio, M. Kohler, M. Bernhard,
B. Dillenburger, J. Buchli, N. Roussel, R. Flatt, Digital Concrete: Opportunities
and Challenges, RILEM Technical Letters 1 (0) (2016) 67–75.
doi:10.21809/rilemtechlett.2016.16.
[102] J. Zhang, B. Khoshnevis, Optimal machine operation planning for construction by
Contour Crafting, Automation in Construction 29 (2013) 50–67.
doi:10.1016/j.autcon.2012.08.006.
[103] B. Berman, 3-D printing: The new industrial revolution, Business Horizons 55 (2)
(2012) 155–162. doi:10.1016/j.bushor.2011.11.003.
[104] Building a lunar base with 3D printing / Space Engineering & Technology / Our
Activities / ESA (2013).
URL http:
//www.esa.int/Our{_}Activities/Space{_}Engineering{_}Technology/
Building{_}a{_}lunar{_}base{_}with{_}3D{_}printing
[105] G. Cesaretti, E. Dini, X. De Kestelier, V. Colla, L. Pambaguian, Building
components for an outpost on the Lunar soil by means of a novel 3D printing
technology, Acta Astronautica 93 (2014) 430–450.
doi:10.1016/j.actaastro.2013.07.034.
[106] L. Sass, R. Oxman, Materializing design: the implications of rapid prototyping in
digital design, Design Studies 27 (3) (2006) 325–355.
doi:https://doi.org/10.1016/j.destud.2005.11.009.
[107] Lisa Ricciotti, Post in Aix-en-Provence (2017).
URL http://www.xtreee.eu/post-in-aix-en-provence/
[108] B. Khoshnevis, S. Bukkapatnam, H. Kwon, J. Saito, Experimental investigation of
contour crafting using ceramics materials, Rapid Prototyping Journal 7 (1) (2001)
32–42. doi:10.1108/13552540110365144.
[109] Lidija Grozdanic, Huge 3D Printer Can Print an Entire Two-Story House in Under
a Day – Inhabitat – Green Design, Innovation, Architecture, Green Building
(2014).
URL https://inhabitat.com/large-3d-printer-can-print-an-entire-two-
story-house-in-under-a-day/
[110] R. Duballet, Space truss masonry walls with robotic mortar extrusion, in:
C. Mueller, S. Adriaenssens (Eds.), Annual IASS Symposia: Creativity in structural
design, International association for shell and spatial structures, Boston, 2018.
[111] Federal Statistical Office, Baugenehmigungen im Hochbau 2014 (2015).
URL www.destatis.de
[112] V. N. Nerella, M. Krause, M. Näther, V. Mechtcherine, Studying printability of
fresh concrete for formwork free Concrete on-site 3D Printing technology
technology (CONPrint3D), in: W. Kusterle, O. Teubert, M. Greim (Eds.),
Rheologische Messungen an Baustoffen, no. March, Tredition GmbH, Hamburg,
Regensburg, 2016, pp. 236–246.
[113] F. Mahaut, S. Mokéddem, X. Chateau, N. Roussel, G. Ovarlez, Effect of coarse
particle volume fraction on the yield stress and thixotropy of cementious
materials, Cement and Concrete Research 38 (11) (2008) 1276–1285.
doi:10.1016/j.cemconres.2008.06.001.
[114] F. Mahaut, X. Chateau, P. Coussot, G. Ovarlez, Yield stress and elastic modulus of
suspensions of noncolloidal particles in yield stress fluids, Journal of Rheology
52 (1) (2008) 287–313. arXiv:0810.3487, doi:10.1122/1.2798234.
[115] H. Hafid, G. Ovarlez, F. Toussaint, P. Jezequel, N. Roussel, Effect of particle
morphological parameters on sand grains packing properties and rheology of
model mortars, Cement and Concrete Research 80 (2016) 44–51.
doi:10.1016/j.cemconres.2015.11.002.
[116] J. G. Sanjayan, B. Nematollahi, M. Xia, T. Marchment, Effect of surface moisture
on inter-layer strength of 3D printed concrete, Construction and Building
Materials (2018) 468–475doi:10.1016/j.conbuildmat.2018.03.232.
[117] Y. Qian, G. De Schutter, Enhancing thixotropy of fresh cement pastes with
nanoclay in presence of polycarboxylate ether superplasticizer (PCE), Cement and
Concrete Research 111 (September) (2018) 15–22.
doi:10.1016/j.cemconres.2018.06.013.
URL https://doi.org/10.1016/j.cemconres.2018.06.013
[118] Y. Qian, K. Lesage, K. El Cheikh, G. De Schutter, Effect of polycarboxylate ether
superplasticizer (PCE) on dynamic yield stress, thixotropy and flocculation state
of fresh cement pastes in consideration of the Critical Micelle Concentration
(CMC), Cement and Concrete Research 107 (September) (2018) 75–84.
doi:10.1016/j.cemconres.2018.02.019.
[119] M. Hambach, D. Volkmer, Properties of 3D-printed fiber-reinforced Portland
cement paste, Cement and Concrete Composites 79 (2017) 62–70.
doi:10.1016/j.cemconcomp.2017.02.001.
[120] B. Khoshnevis, M. Bodiford, K. Burks, E. Ethridge, D. Tucker, W. Kim, H. Toutanji,
M. Fiske, Lunar contour crafting–a novel technique for ISRU-based habitat
development, Automation in construction 13(1) (January) (2004) 5–19.
[121] W. R. L. Da-Silva, T. J. Andersen, A. Kudsk, K. F. Jørgensen, 3D Concrete Printing
of post-tensioned elements, in: C. Mueller, S. Adriaenssens (Eds.), Annual IASS
Symposia: Creativity in structural design, International association for shell and
spatial structures, Cambridge, Massachusetts, 2018.
[122] L. Reiter, M. Palacios, T. Wangler, R. J. Flatt, Putting Concrete to Sleep and
Waking It Up With Chemical Admixtures for Smart Dynamic Casting, in: V. M.
Malhotra, P. R. Gupta, T. C. Holland (Eds.), SP-302: Eleventh International
Conference on Superplasticizers and Other Chemical Admixtures in Concrete,
Ottawa, 2015, pp. 1–513.
[123] A. Vandenberg, H. Bessaies-Bey, K. Wille, N. Roussel, Enhancing Printable
Concrete Thixotropy by High Shear Mixing, in: T. Wangler, R. J. Flatt (Eds.), First
RILEM International Conference on Concrete and Digital Fabrication – Digital
Concrete 2018, Springer International Publishing, Zürich, 2019, pp. 94–101.
[124] M. Näther, V. Nerella, M. Krause, G. Kunze, V. Mechtcherine, S. Rainer,
Beton-3D-Druck Machbarkeitsuntersuchungen zu kontinuierlichen und
schalungsfreien Bauverfahren durch 3D-Formung von Frischbeton; Projektbericht
(AZ: SWD-10.08.18.7-14.07)., Tech. rep., TU Dresden, Dresden (2016).
[125] J. Buchli, M. Giftthaler, N. Kumar, M. Lussi, T. Sandy, K. Dörfler, N. Hack, Digital
in situ fabrication – Challenges and opportunities for robotic in situ fabrication in
architecture, construction, and beyond, Cement and Concrete Research
112 (January) (2018) 66–75. doi:10.1016/j.cemconres.2018.05.013.
[126] C. M. Eastman, R. Sacks, Relative Productivity in the AEC Industries in the
United States for On-Site and Off-Site Activities, Journal of Construction
Engineering and Management 134 (7) (2008) 517–526.
doi:10.1061/(ASCE)0733-9364(2008)134:7(517).
[127] P. M. Goodrum, D. Zhai, M. F. Yasin, Relationship between Changes in Material
Technology and Construction Productivity, Journal of Construction Engineering
and Management 135 (4) (2009) 278–287.
doi:10.1061/(ASCE)0733-9364(2009)135:4(278).
[128] H. Beushausen, M. G. Alexander, Localised strain and stress in bonded concrete
overlays subjected to differential shrinkage, Materials and Structures 40 (2) (2007)
189–199. doi:10.1617/s11527-006-9130-z.
URL https://doi.org/10.1617/s11527-006-9130-z
[129] S. Lim, R. Buswell, T. Le, R. Wackrow, S. Austin, A. Gibb, T. Thorpe, Development
of a Viable Concrete Printing Process, in: Proceedings of the 28th International
Symposium on Automation and Robotics in Construction (ISARC2011), Seoul,
South Korea, 2011, pp. 665–670. doi:10.22260/ISARC2011/0124.
[130] F. P. Bos, Z. Y. Ahmed, E. R. Jutinov, T. A. Salet, Experimental exploration of metal
cable as reinforcement in 3D printed concrete, Materials 10 (11) (2017).
doi:10.3390/ma10111314.
[131] Mike Murphy, A San Francisco startup is 3D-printing entire houses in just one
day (2017).
URL https://qz.com/924909/apis-cor-can-3d-print-and-entire-house-
in-just-one-day/
[132] V. Mechtcherine, V. N. Nerella, Integration der Bewehrung beim 3D-Druck mit
Beton, Beton- und Stahlbetonbau 113 (7) (2018) 496–504.
doi:10.1002/best.201800003.
[133] N. Valencia, World’s first 3D printed bridge opens in Spain (2017).
URL https://www.archdaily.com/804596/worlds-first-3d-printed-
bridge-opens-in-spain
[134] C. Menna, D. Asprone, T. Pastore, V. Mercuri, Implementation of a
stress-constrained topology optimization technique for 3D printed concrete
structures, in: C. Mueller, S. Adriaenssens (Eds.), Annual IASS Symposia:
 Creativity in structural design, International association for shell and spatial
structures, Cambridge, Massachusetts, 2018.
[135] V. M. Julia Kaufhold, Johannes Kohl, Venkatesh Naidu Nerella, Christof Schröfl,
Christoph Wenderdel, Paul Blankenstein, Wood-based support material for
extrusion based digital construction, Rapid Prototyping Journal (under review), in
Press (2018).
[136] M. Näther, V. Nerella, M. Krause, G. Kunze, V. Mechtcherine, S. Rainer,
Beton-3D-Druck Machbarkeitsuntersuchungen zu kontinuierlichen und
schalungsfreien Bauverfahren durch 3D-Formung von Frischbeton; Projektbericht
(AZ: SWD-10.08.18.7-14.07)., Tech. rep., TU Dresden, Dresden (2016).
[137] M. Krause, J. Otto, A. Bulgakov, D. Sayfeddine, Strategic optimization of 3D
concrete printing using the method of CONPrint3D, in: Proceedings of the 35th
ISARC, IAARC, Berlin, 2018, pp. 9–15.
[138] G. H. Tattersall, P. F. G. Banfill, The rheology of fresh concrete, Pitman, 1983.
[139] P. F. G. Banfill, Rheology of Fresh Cement and Concrete, Vol. 2006, Taylor &
Francis, Abingdon, UK, 1991. doi:10.4324/9780203473290.
[140] N. Roussel, Understanding the rheology of concrete, Elsevier, 2011.
[141] C. F. Ferraris, Measurement of the rheological properties of high performance
concrete: state of the art report, Journal of research of the national institute of
standards and technology 104 (5) (1999) 461.
[142] L. N. Thrane, Form Filling with Self-Compacting Concrete Form Filling with
Self-Compacting Concrete [Ph.D. Dissertation], Ph.D. thesis, Technical University
of Denmark (2007).
[143] V. Mechtcherine, A. Gram, K. Krenzer, J.-H. Schwabe, S. Shyshko, N. Roussel,
Simulation of fresh concrete flow using Discrete Element Method (DEM): theory
and applications, Materials and Structures 47 (4) (2014) 615–630.
doi:10.1617/s11527-013-0084-7.
[144] G. Heirman, L. Vandewalle, D. Van Gemert, Ó. Wallevik, Integration approach of
the Couette inverse problem of powder type self-compacting concrete in a
wide-gap concentric cylinder rheometer, Journal of Non-Newtonian Fluid
Mechanics 150 (2-3) (2008) 93–103. doi:10.1016/j.jnnfm.2007.10.003.
[145] S. Shyshko, Numerical simulation of the rheological behavior of fresh concrete,
Phd, Technische Universität Dresden (2013).
[146] N. Roussel, A. Gram, Simulation of Fresh Concrete Flow, Vol. 15 of RILEM
State-of-the-Art Reports, Springer Netherlands, Dordrecht, 2014.
doi:10.1007/978-94-017-8884-7.
[147] J. E. Wallevik, Relationship between the Bingham parameters and slump, Cement
and Concrete Research 36 (7) (2006) 1214–1221.
doi:10.1016/j.cemconres.2006.03.001.
[148] A. Perrot, A. Pierre, S. Vitaloni, V. Picandet, Prediction of lateral form pressure
exerted by concrete at low casting rates, Materials and Structures 48 (7) (2015)
2315–2322. doi:10.1617/s11527-014-0313-8.
[149] C. F. Ferraris, K. H. Obla, R. Hill, The influence of mineral admixtures on the
rheology of cement paste and concrete, Cement and Concrete Research 31 (2)
(2001) 245–255. doi:10.1016/S0008-8846(00)00454-3.
[150] E. P. Koehler, S. Amziane, V. K. Bui, Y. S. Deshpande, R. P. Ferron, J. Hu, N. L.
Moncef, R. G. Peleggi, J. Tanesi, D. Beaupre, S. E. Chidiac, E. Al., ACI 238.2T-14
TechNote Concrete Thixotropy – By ACI Committee 238, Tech. Rep. June,
American Concrete Institute (2014).
[151] Zhuojun Quanji, Thixotropic behavior of cement-based materials: effect of clay
and cement types, Ph.D. thesis, IOWA State University (2010).
URL http:
//lib.dr.iastate.edu/cgi/viewcontent.cgi?article=2686{&}context=etd
[152] N. Roussel, F. Cussigh, Distinct-layer casting of SCC: The mechanical
consequences of thixotropy, Cement and Concrete Research 38 (5) (2008)
624–632. doi:10.1016/j.cemconres.2007.09.023.
[153] A. F. Omran, K. H. Khayat, Y. M. Elaguab, Effect of SCC Mixture Composition on
Thixotropy and Formwork Pressure, Journal of Materials in Civil Engineering
24 (7) (2012) 876–888. doi:10.1061/(ASCE)MT.1943-5533.0000463.
[154] Y. Qian, S. Kawashima, Flow onset of fresh mortars in rheometers: Contribution of
paste deflocculation and sand particle migration, Cement and Concrete Research
90 (October) (2016) 97–103. doi:10.1016/j.cemconres.2016.09.006.
[155] K. H. Khayat, A. F. Omran, S. Naji, P. Billberg, A. Yahia, Field-oriented test
methods to evaluate structural build-up at rest of flowable mortar and concrete,
Materials and Structures 45 (10) (2012) 1547–1564.
doi:10.1617/s11527-012-9856-8.
[156] N. Roussel, A thixotropy model for fresh fluid concretes: Theory, validation and
applications, Cement and Concrete Research 36 (10) (2006) 1797–1806.
doi:10.1016/j.cemconres.2006.05.025.
[157] B. Patzák, Z. Bittnar, Modeling of fresh concrete flow, Computers & Structures
87 (15-16) (2009) 962–969. doi:10.1016/j.compstruc.2008.04.015.
[158] S. Amziane, C. F. Ferraris, Cementitious Paste Setting Using Rheological and
Pressure Measurements, ACI Materials Journal 104 (2) (2008) 137–145.
URL ACIMaterialsJournal
[159] F. Mahmoodzadeh, S. Chidiac, Rheological models for predicting plastic viscosity
and yield stress of fresh concrete, Cement and Concrete Research 49 (2013) 1–9.
doi:10.1016/j.cemconres.2013.03.004.
[160] F. de Larrard, C. F. Ferraris, T. Sedran, Fresh concrete: A Herschel-Bulkley
material, Materials and Structures 31 (7) (1998) 494–498.
doi:10.1007/BF02480474.
[161] K. Vance, A. Arora, G. Sant, N. Neithalath, Rheological evaluations of interground
and blended cement–limestone suspensions, Construction and Building Materials
79 (2015) 65–72. doi:10.1016/j.conbuildmat.2014.12.054.
[162] M. H. Mohammed, R. Pusch, S. Knutsson, G. Hellström, Rheological Properties of
Cement-Based Grouts Determined by Different Techniques, Engineering 06 (05)
(2014) 217–229. doi:10.4236/eng.2014.65026.
[163] A. F. Omran, K. H. Khayat, Choice of thixotropic index to evaluate formwork
pressure characteristics of self-consolidating concrete, Cement and Concrete
Research 63 (2014) 89–97. doi:10.1016/j.cemconres.2014.05.005.
[164] A. F. Omran, S. Naji, K. H. Khayat, Portable vane test to assess structural buildup
at rest of self-consolidating concrete, ACI Materials Journal 108 (6) (2011)
628–637.
[165] Q. Yuan, D. Zhou, K. H. Khayat, D. Feys, C. Shi, On the measurement of evolution
of structural build-up of cement paste with time by static yield stress test vs.
small amplitude oscillatory shear test, Cement and Concrete Research
99 (September 2016) (2017) 183–189. doi:10.1016/j.cemconres.2017.05.014.
[166] W. Mbasha, I. Masalova, R. Haldenwang, A. Malkin, The yield stress of cement
pastes as obtained by different rheological approaches, Applied Rheology 25 (5)
(2015) 53517. doi:10.3933/ApplRheol-25-53517.
[167] J. E. Wallevik, Thixotropic investigation on cement paste: Experimental and
numerical approach, Journal of Non-Newtonian Fluid Mechanics 132 (1-3) (2005)
86–99. doi:10.1016/j.jnnfm.2005.10.007.
[168] P. H. Billberg, The Structural Behaviour of SCC at rest, in: 36 th Conference on
Our World in Concrete & Structures, no. August, CI-Premier PTE LTD, Singapore,
2011, pp. 14–16.
[169] A. Perrot, T. Lecompte, H. Khelifi, C. Brumaud, J. Hot, N. Roussel, Yield stress and
bleeding of fresh cement pastes, Cement and Concrete Research 42 (7) (2012)
937–944. doi:10.1016/j.cemconres.2012.03.015.
[170] Q. Yuan, X. Lu, K. H. Khayat, D. Feys, C. Shi, Small amplitude oscillatory shear
technique to evaluate structural build-up of cement paste, Materials and
Structures 50 (2) (2017) 112. doi:10.1617/s11527-016-0978-2.
[171] A. M. Mostafa, A. Yahia, New approach to assess build-up of cement-based
suspensions, Cement and Concrete Research 85 (March) (2016) 174–182.
doi:10.1016/j.cemconres.2016.03.005.
[172] N. Roussel, G. Ovarlez, S. Garrault, C. Brumaud, The origins of thixotropy of fresh
cement pastes, Cement and Concrete Research 42 (1) (2012) 148–157.
doi:10.1016/j.cemconres.2011.09.004.
[173] D. C.-H. Cheng, F. Evans, Phenomenological characterization of the rheological
behaviour of inelastic reversible thixotropic and antithixotropic fluids, British
Journal of Applied Physics 16 (11) (1965) 1599–1617.
doi:10.1088/0508-3443/16/11/301.
[174] A. Papo, The thixotropic behavior of white Portland cement pastes, Cement and
Concrete Research 18 (4) (1988) 595–603. doi:10.1016/0008-8846(88)90052-X.
[175] P. Coussot, Q. D. Nguyen, H. T. Huynh, D. Bonn, Viscosity bifurcation in
thixotropic, yielding fluids, Journal of Rheology 46 (3) (2002) 573–589.
doi:10.1122/1.1459447.
[176] P. O. H. Wallevik, Introduction to Rheology of Concrete, Innovation Center
Iceland, Reykjavik, 2009.
[177] T. Ngo, E. Kadri, R. Bennacer, F. Cussigh, Use of tribometer to estimate interface
friction and concrete boundary layer composition during the fluid concrete
pumping, Construction and Building Materials 24 (7) (2010) 1253–1261.
doi:10.1016/j.conbuildmat.2009.12.010.
[178] S. D. Jo, C. K. Park, J. H. Jeong, S. H. Lee, S. H. Kwon, A Computational Approach
to Estimating a Lubricating Layer in Concrete Pumping, Computers, Materials
and Continua 27 (3) (2012) 189–210. doi:10.3970/cmc.2011.027.189.
[179] D. Feys, G. De Schutter, R. Verhoeven, Parameters influencing pressure during
pumping of self-compacting concrete, Materials and Structures 46 (4) (2013)
533–555. doi:10.1617/s11527-012-9912-4.
[180] V. N. Nerella, V. Mechtcherine, K. Kasten, Experimental study on pumpability of
concrete using sliding pipe rheometer (SLIPER), in: Rheology and processing of
Construction Materials – 7th RILEM International Conference on
Self-Compacting Concrete and 1st RILEM International Conference on Rheology
and Processing of Construction Materials, no. September, RILEM Publications
S.A.R.L., Paris, 2013, p. 406.
[181] M. Choi, N. Roussel, Y. Kim, J. Kim, Lubrication layer properties during concrete
pumping, Cement and Concrete Research 45 (null) (2013) 69–78.
doi:10.1016/j.cemconres.2012.11.001.
[182] D. Feys, K. H. Khayat, R. Khatib, How do concrete rheology, tribology, flow rate
and pipe radius influence pumping pressure?, Cement and Concrete Composites
66 (2016) 38–46. doi:10.1016/j.cemconcomp.2015.11.002.
[183] N. R. Guptill, D. J. Akers, R. A. Kelsey, J. S. Pierce, C. Bognacki, J. C. King, P. E.
Reinhart, J. L. Cope, W. C. Krell, R. J. Rhoads, M. Garner, G. R. Mass, K. L. Saucier,
D. J. Green, P. McDowell, P. R. Stodola, T. C. Holland, D. T. Parekh, W. X. Sypher,
T. A. Johnson, R. J. Phares, R. E. Tobin, S. A. Kalat, K. Wolf, Placing Concrete by
Pumping Methods-Report by ACI Committee 304, Tech. rep., American Concrete
Institute (1996).
[184] J. L. Poiseuille, Recherches expérimentales sur le mouvement des liquides dans les
tubes de très-petits diamètres, Imprimerie Royale, 1844.
[185] G. De Schutter, D. Feys, Pumping of Fresh Concrete: Insights and Challenges,
RILEM Technical Letters 1 (2016) 76. doi:10.21809/rilemtechlett.2016.15.
[186] S. W. Churchill, Viscous flows: the practical use of theory, Butterworth Publishers,
Stoneham, MA, 1988.
URL https://cds.cern.ch/record/1968108
[187] M. S. Choi, J. K. Young, K. K. Jin, Prediction of Concrete Pumping Using Various
Rheological Models, International Journal of Concrete Structures and Materials
8 (4) (2014) 269–278. doi:10.1007/s40069-014-0084-1.
[188] S. H. Kwon, C. K. Park, J. H. Jeong, S. D. Jo, S. H. Lee, Prediction of Concrete
Pumping: Part II—Analytical Prediction and Experimental Verification, ACI
Materials Journal 110 (6) (2013) 657–668. doi:10.14359/51686333.
[189] E. Secrieru, S. Fataei, C. Schröfl, V. Mechtcherine, Study on concrete pumpability
combining different laboratory tools and linkage to rheology, Construction and
Building Materials 144 (2017) 451–461.
doi:10.1016/j.conbuildmat.2017.03.199.
[190] V. N. Nerella, E. Secrieru, V. Mechtcherine, Experimental study on influence of
lubricating layer on pumping pressure using Sliding Pipe Rheometer (SLIPER),
in: 19h International Conference on Building Materials – ibausil, Weimar, 2015,
pp. 1–1247–1254.
[191] D. Feys, K. H. Khayat, A. Perez-Schell, R. Khatib, Development of a tribometer to
characterize lubrication layer properties of self-consolidating concrete, Cement
and Concrete Composites 54 (2014) 40–52.
doi:10.1016/j.cemconcomp.2014.05.008.
[192] S. H. Kwon, C. K. Park, J. H. Jeong, S. D. Jo, S. H. Lee, Prediction of Concrete
Pumping: Part I–Development of New Tribometer for Analysis of Lubricating
Layer, ACI Materials Journal 110 (6) (2013). doi:10.14359/51686332.
[193] V. N. Nerella, Experimental and CFD study on pumpability of concrete using
Sliding Pipe Rheometer and ANSYS Fluent [Master Thesis] (2012).
[194] B. Khoshnevis, X. Yuan, B. Zahiri, J. Zhang, B. Xia, Construction by Contour
Crafting using sulfur concrete with planetary applications, Rapid Prototyping
Journal 22 (5) (2016) 848–856. doi:10.1108/RPJ-11-2015-0165.
URL http://www.emeraldinsight.com/doi/10.1108/RPJ-11-2015-0165
[195] N. Roussel, M. R. Geiker, F. Dufour, L. N. Thrane, P. Szabo, Computational
modeling of concrete flow: General overview, Cement and Concrete Research
37 (9) (2007) 1298–1307. doi:10.1016/j.cemconres.2007.06.007.
[196] N. Roussel, A. Gram, M. Cremonesi, L. Ferrara, K. Krenzer, V. Mechtcherine,
S. Shyshko, J. Skocec, J. Spangenberg, O. Svec, L. N. Thrane, K. Vasilic, Numerical
simulations of concrete flow: A benchmark comparison, Cement and Concrete
Research 79 (2016) 265–271. doi:10.1016/j.cemconres.2015.09.022.
[197] S. Remond, P. Pizette, A DEM hard-core soft-shell model for the simulation of
concrete flow, Cement and Concrete Research 58 (2014) 169–178.
doi:10.1016/j.cemconres.2014.01.022.
[198] J. Anderson, Governing Equations of Fluid Dynamics, in: J. F. Wendt (Ed.),
Computational Fluid Dynamics, 3rd Edition, Springer-Verlag, Berlin, Heidelberg,
2009, Ch. 2, pp. 15–51. doi:10.1007/978-3-540-85056-4_2.
[199] K. Vasilic, B. Meng, H. Kühne, N. Roussel, Flow of fresh concrete through steel
bars: A porous medium analogy, Cement and Concrete Research 41 (5) (2011)
496–503. doi:10.1016/j.cemconres.2011.01.013.
[200] S. Tichko, J. Van De Maele, N. Vanmassenhove, G. De Schutter, J. Vierendeels,
R. Verhoeven, P. Troch, Numerical simulation of formwork pressure while
pumping self-compacting concrete bottom-up, Engineering Structures 70 (2014)
218–233. doi:10.1016/j.engstruct.2014.04.008.
[201] A. Gram, J. Silfwerbrand, Numerical simulation of fresh SCC flow: applications,
Materials and Structures 44 (4) (2011) 805–813.
doi:10.1617/s11527-010-9666-9.
URL http://www.springerlink.com/index/10.1617/s11527-010-9666-9
[202] O. Svec, J. Skocek, H. Stang, J. F. Olesen, P. N. Poulsen, Flow simulation of fiber
reinforced self compacting concrete using Lattice Boltzmann method, in: 8th
International Congress on the Chemistry of Cement, Madrid, 2011.
[203] K. Vasilic, A Numerical Model for Self-Compacting Concrete Flow through
Reinforced Sections : a Porous Medium Analogy, Ph.D. thesis, TU Dresden (2015).
[204] M. R. Geiker, M. Brandl, L. N. Thrane, D. H. Bager, O. Wallevik, The effect of
measuring procedure on the apparent rheological properties of self-compacting
concrete, Cement and Concrete Research 32 (11) (2002) 1791–1795.
doi:10.1016/S0008-8846(02)00869-4.
[205] J. E. Wallevik, O. H. Wallevik, Analysis of shear rate inside a concrete truck mixer,
Cement and Concrete Research 95 (2017) 9–17.
doi:10.1016/j.cemconres.2017.02.007.
[206] Y. Tan, H. Zhang, D. Yang, S. Jiang, J. Song, Y. Sheng, Numerical simulation of
concrete pumping process and investigation of wear mechanism of the piping
wall, Tribology International 46 (1) (2011) 137–144.
doi:10.1016/j.triboint.2011.06.005.
[207] L. Chen, G. Liu, W. Cheng, G. Pan, Pipe flow of pumping wet shotcrete based on
lubrication layer, SpringerPlus 5 (1) (2016) 945.
doi:10.1186/s40064-016-2633-3.
[208] D. Feys, G. De Schutter, K. H. Khayat, R. Verhoeven, Changes in rheology of
self-consolidating concrete induced by pumping, Materials and Structures 49 (11)
(2016) 4657–4677. doi:10.1617/s11527-016-0815-7.
[209] CONCRETE 101_A guide to understanding the qualities of concrete and how they
affect pumping, Tech. rep., ACPA – American Concrete Pumping Association
(2008).
[210] CIP 21 – Loss of Air Content in Pumped Concrete, Tech. rep., NRMCA – National
readymix concrete Association, Silver Spring, MD 20910 (2005).
[211] H. Stang, C. Pedersen, HPFRCC - Extruded Pipes., in: Materials for the new
Millennium. Fourth Materials Engineering Conference, American Society of Civil
Engineers, Washington, DC, 1996, pp. 261–270.
[212] C. Aldea, S. Marikunte, S. P. Shah, Extruded Fiber Reinforced Cement Pressure
Pipe, Advanced Cement Based Materials 8 (2) (1998) 47–55.
doi:10.1016/S1065-7355(98)00006-6.
[213] A. Peled, S. P. Shah, Processing effects in cementitious composites: Extrusion and
casting, Journal of materials in civil engineering 15 (April 2003) (2003) 192–199.
doi:10.1061/(ASCE)0899-1561(2003)15.
[214] A. Perrot, C. Lanos, P. Estellé, Y. Melinge, Ram extrusion force for a frictional
plastic material: model prediction and application to cement paste, Rheologica
Acta 45 (4) (2006) 457–467. doi:10.1007/s00397-005-0074-y.
[215] A. Perrot, C. Lanos, Y. Melinge, P. Estellé, Mortar physical properties evolution in
extrusion flow, Rheologica Acta 46 (8) (2007) 1065–1073.
doi:10.1007/s00397-007-0195-6.
[216] Archimedean Screw Pump.
URL https://www.ksb.com/centrifugal-pump-lexicon/archimedean-screw-
pump/191708/
[217] R. Moineau, A new Capsulism, Ph.D. thesis, The University of Paris (1930).
[218] J. Gamboa, J. Iglesias, P. Gonzalez, Understanding the Performance of a
Progressive Cavity Pump with a Metallic Stator, in: Proceedings Of The
Twenty-Fifth International Pump Users Symposium, Houston, 2003, pp. 1–19.
[219] P. A. S. Pessoa, E. E. Paladino, J. A. de Lima, A Simplified Model for the Flow in a
Progressive Cavity Pump, 20th International Congress of Mechanical
Engineering (2003) (2009).
[220] A. G. Wild, Progressing cavity pumps proper selection and application, in:
Proceedings of the ninth international pump users symposium, The Laboratory,
1992, p. 111. doi:https://doi.org/10.21423/R1ZQ3B.
[221] V. W. F. de Azevedo, J. A. de Lima, E. E. Paladino, A Computational model for
multiphase flow inside a metallic progressing vacity pump, in: 14th Brazilian
Congress of Thermal Sciences and Engineering, Rio de Janeiro, 2012, pp. 1–10.
[222] G. Vetter, W. Wirth, H. Körner, S. Pregler, Multiphase pumping with twin-screw
pumps – understand and model hydrodynamics and hydroabrasive wear, in:
Proceedings of the 17th international pump users symposium, Houston, 2000.
[223] E. E. Paladino, J. A. Lima, P. A. Pessoa, R. F. Almeida, A computational model for
the flow within rigid stator progressing cavity pumps, Journal of Petroleum
Science and Engineering 78 (1) (2011) 178–192.
doi:10.1016/j.petrol.2011.05.008.
[224] G. Vetter, W. Wirth, Understand Progressing Cavity Pumps Characteristics And
Avoid Abrasive Wear, in: Proceedings of the 12th international pump users
symposium, Houston., 1995.
[225] A. Perrot, Y. Mélinge, D. Rangeard, F. Micaelli, P. Estellé, C. Lanos, Use of ram
extruder as a combined rheo-tribometer to study the behaviour of high yield
stress fluids at low strain rate, Rheologica Acta 51 (8) (2012) 743–754.
doi:10.1007/s00397-012-0638-6.
[226] J. Benbow, J. Bridgewater, Paste flow and extrusion, Oxford University Press(UK),
1993, (1993) 153.
[227] R. Basterfield, C. Lawrence, M. Adams, On the interpretation of orifice extrusion
data for viscoplastic materials, Chemical Engineering Science 60 (10) (2005)
2599–2607. doi:10.1016/j.ces.2004.12.019.
[228] P. J. Martin, D. I. Wilson, K. Challis, Extrusion of Paste Through
Non-Axisymmetric Systems, in: 6th World Congress of Chemical Engineering,
no. 2, 2001.
[229] D. J. Horrobin, R. M. Nedderman, C. D. Nechifor, D. O. Dorohoi, C. Ciobanu,
Rahaman; M. (2003). Ceramic Processing and Sintering. Boca, Romanian Reports
of Physics 54 (3-4) (2009) 349–359. arXiv:0808.2157, doi:10.1007/12.
[230] T. Voigt, Z. Sun, S. P. Shah, Comparison of ultrasonic wave reflection method and
maturity method in evaluating early-age compressive strength of mortar, Cement
and Concrete Composites 28 (4) (2006) 307–316.
doi:10.1016/j.cemconcomp.2006.02.003.
[231] ASTM C403: Test Method for Time of setting of Concrete Mixtures by Penetration
Resistance (2008).
[232] Y. Chen, L. J. Struble, G. H. Paulino, Extrudability of cement-based materials,
American Ceramic Society Bulletin 85 (6) (2006) 9101–9105.
[233] Z. Toutou, N. Roussel, C. Lanos, The squeezing test: A tool to identify firm
cement-based material’s rheological behaviour and evaluate their extrusion
ability, Cement and Concrete Research 35 (10) (2005) 1891–1899.
doi:10.1016/j.cemconres.2004.09.007.
[234] A. Perrot, D. Rangeard, Y. Mélinge, Prediction of the ram extrusion force of
cement-based materials, Applied Rheology 24 (5) (2014).
doi:10.3933/APPLRHEOL-24-53320.
[235] A. Perrot, D. Rangeard, M. Yannick, P. Estell, A. Perrot, D. Rangeard, M. Yannick,
P. Estell, C. L. Ex, Extrusion Criterion for Firm Cement-based Materials, Applied
Rheology 19(5) (2009) 53042. doi:10.3933/ApplRheol-19-53042.
[236] Z. Malaeb, H. Hachem, A. Tourbah, T. Maalouf, N. El Zarwi, F. Hamzeh, 3D
Concrete Printing: Machine and Mix Design, International Journal of Civil
Engineering and Technology 6 (April) (2015) 14–22.
[237] T. D. Carlo, A. Carlson, B. Khoshnevis, T. Di Carlo, Experimental and Numerical
Techniques to characterize Structural Properties of Fresh concrete, Ph.D. thesis,
University of Southern California (2013).
[238] Y. Weng, M. Li, M. J. Tan, S. Qian, Design 3D printing cementitious materials via
Fuller Thompson theory and Marson-Percy model, Construction and Building
Materials 163 (2018) 600–610.
doi:https://doi.org/10.1016/j.conbuildmat.2017.12.112.
[239] N. Roussel, C. Lanos, Plastic fluid flow parameters identification using a simple
squeezing test, Applied Rheology 13 (3) (2003) 132–139.
[240] C. Schröfl, V. N. Nerella, V. Mechtcherine, Capillary water intake by 3d-printed
concrete visualised and quantified by neutron radiography, in: T. Wangler, R. J.
Flatt (Eds.), First RILEM International Conference on Concrete and Digital
Fabrication – Digital Concrete 2018, Springer International Publishing, Cham,
2019, pp. 217–224.
[241] D. S. Santos, P. M. D. Santos, D. Dias-da costa, Effect of surface preparation and
bonding agent on the concrete-to-concrete interface strength, Construction and
Building Materials 37 (2012) 102–110.
doi:10.1016/j.conbuildmat.2012.07.028.
[242] D. Dias-da costa, J. Alfaiate, E. N. B. S. Júlio, FE modeling of the interfacial
behaviour of composite concrete members, Construction and Building Materials
26 (1) (2012) 233–243. doi:10.1016/j.conbuildmat.2011.06.015.
[243] CEN, EN 1992-1-1: Eurocode 2: design of concrete structures – Part 1-1: General
rules and rules for buildings (1992).
[244] CEM-FIP: Model Code 90 – design manual (1993).
[245] E. N. Júlio, F. A. Branco, V. D. Silva, Concrete-to-concrete bond strength. Influence
of the roughness of the substrate surface, Construction and Building Materials
18 (9) (2004) 675–681. doi:10.1016/j.conbuildmat.2004.04.023.
[246] T. Lecompte, A. Perrot, Non-linear modeling of yield stress increase due to SCC
structural build-up at rest, Cement and Concrete Research 92 (2017) 92–97.
doi:10.1016/j.cemconres.2016.11.020.
[247] E. N. Júlio, F. A. Branco, V. D. Silva, Concrete-to-concrete bond strength. Influence
of the roughness of the substrate surface, Construction and Building Materials
18 (9) (2004) 675–681. doi:10.1016/j.conbuildmat.2004.04.023.
[248] K. R. Hindo, In-place bond testing and surface preparation of concrete, Concrete
International (April) (1990) 127–129.
[249] F. Saucier, M. Pigeon, Durability of new-to-old concrete bonding, in: Proceedings
of the ACI International Conference Evaluation and Rehabilitation of Concrete
Structures and Innovations in Design, Vol. 1, 1991, pp. 689–707.
[250] P. H. Emmons, Concrete Repair and Maintenance Illustrated,
Butterworth-Heinemann Limited, MA, 2003.
[251] S. Austin, P. Robins, Y. Pan, Tensile bond testing of concrete repairs, Materials and
Structures 28 (1995) 249–259.
[252] E. N. B. S. Júlio, F. A. B. Branco, V. D. Silva, J. F. Lourenço, Influence of added
concrete compressive strength on adhesion to an existing concrete substrate,
Building and Environment 41 (12) (2006) 1934–1939.
doi:10.1016/j.buildenv.2005.06.023.
[253] B. Panda, S. Chandra Paul, M. Jen Tan, Anisotropic mechanical performance of 3D
printed fiber reinforced sustainable construction material, Materials Letters
(2017). doi:10.1016/j.matlet.2017.07.123.
[254] T. Marchment, J. Sanjayan, Method of Enhancing Interlayer Bond Strength in 3D
Concrete Printing, in: T. Wangler, R. J. Flatt (Eds.), First RILEM International
Conference on Concrete and Digital Fabrication – Digital Concrete 2018, Springer
International Publishing, Cham, 2019, pp. 148–156.
[255] EN 197-1 – 2011-11: Cement – Part 1: Composition, specifications and conformity
criteria for common cements.
[256] ASTM, ASTM C618 : Standard Specification for Coal Fly Ash and Raw or Calcined
Natural Pozzolan for Use in Concrete.
[257] Instructions Manual HAAKE MARS rheometer (2005).
[258] M. Haist, Zur Rheologie und den physikalischen Wechselwirkungen bei
Zementsuspensionen, Ph.D. thesis, Universität Fridericiana zu Karlsruhe (2010).
[259] J. E. Wallevik, Rheology of particle supsensions – Fresh concrete, mortar and
cement paste with various types of lignosulfonates, Ph.D. thesis, The Norwegian
University of Science and Technology (NTNU) (2003).
[260] G. Heirman, Modelling and Quantification of the Effect of Mineral Additions on
the Rheology of Fresh Powder Type Self-Compacting Concrete, Ph.D. thesis,
Katholieke Universiteit Leuven (2011).
[261] CEN, EN 1015-3:2007-05 – Methods of test for mortar for masonry – Part 3:
Determination of consistence of fresh mortar (by flow table); German version EN
1015-3:1999+A1:2004+A2:2006 (2007).
[262] T. S. Rushing, G. Al-Chaar, B. A. Eick, J. Burroughs, J. Shannon, L. Barna, M. Case,
Investigation of concrete mixtures for additive construction, Rapid Prototyping
Journal 23 (1) (2017) 74–80. doi:10.1108/RPJ-09-2015-0124.
[263] A. Leemann, F. Winnefeld, The effect of viscosity modifying agents on mortar and
concrete, Cement and Concrete Composites 29 (5) (2007) 341–349.
doi:10.1016/j.cemconcomp.2007.01.004.
[264] D. S. A. Bravo, T. Cerulli, C. Maltese, C. Pistolesi, Effects of increasing dosages of
an alkali-free accelerator on the physical chemical properties of a hydrating
cement paste., ACI Special Publication 217 (2003) 211–225.
URL https://www.concrete.org/publications/
internationalconcreteabstractsportal/m/details/id/12915
[265] M. A. B. Beigh, Thixotropy characterization of fresh cement-based materials
through rotational rheometry, Master project, TU Dresden (2017).
[266] R. Schwarze, CFD-Modellierung: Grundlagen und Anwendungen bei
Strömungsprozessen, Springer Berlin Heidelberg, Berlin, Heidelberg, 2013.
[267] N. S. Martys, A classical kinetic theory approach to lattice Boltzmann simulation,
International Journal of Modern Physics C 12 (8) (2001) 1169–1177.
doi:10.1142/S0129183101002474.
[268] A. Gram, Modelling Bingham Suspensional Flow, PhD Thesis, Ph.D. thesis, KTH
Royal Institute of Technology (2015).
[269] W. R. Schowalter, G. Christensen, Toward a rationalization of the slump test for
fresh concrete: Comparisons of calculations and experiments, Journal of Rheology
42 (4) (1998) 865–870. doi:10.1122/1.550905.
[270] D. Feys, R. Verhoeven, G. D. Schutter, Pipe flow velocity profiles of complex
suspensions, like concrete, 8th National Congress on Theoretical and Applied
Mechanics (NCTAM 2009) (2009) 66–73.
[271] I. M. Krieger, T. J. Dougherty, A Mechanism for Non-Newtonian Flow in
Suspensions of Rigid Spheres, Transactions of the Society of Rheology 3 (1) (1959)
137–152. doi:10.1122/1.548848.
[272] R. J. Farris, Prediction of the Viscosity of Multimodal Suspensions from Unimodal
Viscosity Data, Journal of Rheology 12 (2) (1968) 281. doi:10.1122/1.549109.
[273] R. J. Flatt, P. Bowen, Yodel: A yield stress model for suspensions, Journal of the
American Ceramic Society 89 (4) (2006) 1244–1256.
doi:10.1111/j.1551-2916.2005.00888.x.
[274] X. Chateau, G. Ovarlez, K. L. Trung, Homogenization approach to the behavior of
suspensions of noncolloidal particles in yield stress fluids, Journal of Rheology
52 (2) (2008) 489–506. doi:10.1122/1.2838254.
[275] C. Hu, F. de Larrard, The rheology of fresh high-performance concrete, Cement
and Concrete Research 26 (2) (1996) 283–294.
doi:10.1016/0008-8846(95)00213-8.
[276] K. Krenzer, Development of a state-dependent DEM-Contact-Model to simulate
mixing processes of fresh concrete [PhD Dissertation], Ph.D. thesis, Technische
Universität Dresden (2016).
[277] D. Feys, G. De Schutter, R. Verhoeven, K. H. Khayat, Similarities and Differences
of Pumping Conventional and Self-Compacting Concrete, in: Design, Production
and Placement of Self-Consolidating Concrete, Springer Netherlands, Dordrecht,
2010, pp. 153–162. doi:10.1007/978-90-481-9664-7_13.
[278] H. Khelifi, A. Perrot, T. Lecompte, D. Rangeard, G. Ausias, Prediction of extrusion
load and liquid phase filtration during ram extrusion of high solid volume
fraction pastes, Powder Technology 249 (2013) 258–268.
doi:10.1016/j.powtec.2013.08.023.
[279] A. Iqbal, Development of test method to quantitatively characterize extrudability
of fresh cementitious materials in the context of additive construction., Master
project, TU Dresden (2017).
[280] D.-I. für-Normung-e. V., DIN 18202:2013-04 – Tolerances in building
construction – Buildings (2013).
Complementary research 9.2. Interface tensile strength

[281] Baukosteninformationszentrum Deutscher Architektenkammern GmbH,

Baukosten Gebäude Neubau 2017 – Statistische Kostenkennwerte, 2015.
[282] Hauptverband der Deutschen Bauindustrie, BGL Baugeräteliste 2015
Technisch-wirtschaftliche Baumaschinendaten, Bauverlag BV GmbH, Gütersloh,
2015.
[283] Cooper Hewitt, Why Design Now?: Contour Crafting – YouTube (2010).
URL https://www.youtube.com/watch?v=-yv-IWdSdns
[284] D. Hwang, B. Khoshnevis, Concrete Wall Fabrication by Contour Crafting, in:
ISAR 2004 21st International Symposium on Automation and Robotics in
Construction, International Association for Automation and Robotics in
Construction (IAARC), 2004. doi:10.22260/ISARC2004/0057.
[285] J. Stark, K. Bollmann, Late ettringite formation in concrete, Part 1, ZKG
International 53 (4) (2000) 1–25.
[286] J. Grafe, Diplomarbeit Bauen mit Beton, Ph.D. thesis, TU Dresden (2017).
[287] G. Rehm, "Über die Grundlagen des Verbundes zwischen Stahl und Beton",
Volume 138 (1961).
[288] V. Markin, Untersuchungen zum Verbund zwischen Schichten 3D-gedruckter
Betonelemente., Master thesis., TU Dresden (2017).
[289] A. K. Gajula, On layer-to-layer interface strengths of concrete in the context of
digital construction, Master thesis., TU Dresden (2018).

197
Appendix A

Additional figures and tables

23 45 60 23 45 60 23 45 60
90 120 150 90 120 150 90 120 150
0.3 0.3 0.3

0.2 0.2 0.2

0.1 0.1 0.1

0.0 0.0 0.0

0 10 20 30 0 5 10 15 0 2 4 6 8 10

a) b) c)
Figure A.1: Shear rate variation over td for varying shear rates with constant total applied
strain: a) γ̇ap of 0.08 s−1 for 30 s, b) γ̇ap of 0.15 s−1 for 16 s and c) γ̇ap of 0.24 s−1 for 10 s.
10
23

10-1 45

60
10-2
90

10-3 120

150
10-4
0 10 20 30 40

Figure A.2: Behaviour of highly thixotropic SCM mixture – shear rate variation with test
duration.

Table A.1: Applied shear rate, time duration and applied shear strain values in tests on SCM
mixture.
tage [min] 23 45 60 90 120 150
γ̇ap [s−1 ] 0.15 0.15 0.15 0.15 0.15 0.15
td [s] 30 30 25 30 35 40
γap [−] 4.50 4.50 3.75 4.50 5.25 6.00

199
Additional figures and tables

Table A.2: Hägermann flow table (HFT) test results for binder pastes investigated structural
build-up. B.S. and A. S. are the average diameters before and after the strokes are applied,
respectively; N.S. means no strokes were applicable.
tage Ref SP-0.50 SP-0.30 SP-0.00 ACC-2.50 SCM
N.S. N.S. B.S. A.S B.S. A.S B.S. A.S B.S. A.S
min mm
23 303 247 194 237 113 163 195 241 176 180
45 299 242 188 231 113 163 188 238 168 173
60 302 243 194 233 111 160 192 238 180 186
90 301 250 185 233 109 154 203 248 183 187
120 301 247 181 224 104 135 205 252 191 201
150 296 238 171 215 101 115 205 236 186 193

Figure A.3: Floor layout and dimensions of the target house. (Courtesy of M. Krause).

Figure A.4: HFT spread diameters of mixtures C1 and C2 after strokes.

200
Curriculum vitae

Curriculum vitae

Bio data
Name: Venkatesh Naidu Nerella M. Sc.
Address: Dorfhainer Str. 38
01189 Dresden
Born: 1988
Place of birth: Medrametla, India
Education
10/2010 – 09/2012 Masters in Advanced Computational Civil Structural
Studies at TU Dresden, Germany
Thesis: “Experimental and Numerical Study on
Pumpability of Concrete using Sliding Pipe Rheometer
and ANSYS Fluent”
Grade: 1.0 (100%)
07/2005 – 03/2009 Bachelor of Technology in Mechanical Engineering from
JNT University, Andhra Pradesh, India
Grade: Distinction with Gold Medal for best performing
student

Professional experience
11/2012 – present Research associate
Institute of Construction Materials, TU Dresden
Development of advanced construction materials
05/2012 – 09/2012 Research intern
Institut für Angewandte Bauforschung, Weimar
CFD simulation of non-Newtonian fluids
Associations
Active member RILEM technical committee on “Digital fabrication with
cement-based materials (TC-DFC)”
Active member German Research Foundation (DFG)’s Scientific Network
“Adaptive reinforcement of geometrically complex
cement-bound structures”

Dresden, 01. November 2018

Venkatesh Naidu Nerella

201
Publications

Selected publications
Peer-reviewed Journal Publications:
1 V.N. Nerella, M.A.B. Beigh, S. Fataei, V. Mechtcherine, Strain-based approach for measuring
structural build-up of cement pastes in the context of digital construction, Cement and
Concrete Research, 115 (2018), 530–544. doi:10.1016/j.cemconres.2018.08.003.
2 V.N. Nerella, V. Mechtcherine, Virtual Sliding Pipe Rheometer for estimating
pumpability of concrete, Construction and Building Materials, 170 (2018), 366–377.
doi:10.1016/j.conbuildmat.2018.03.003.
3 V.N. Nerella, M. Näther, A. Iqbal, M. Butler, V. Mechtcherine, Inline quantification of
extrudability of cementitious materials for digital construction, Cement and Concrete
Composites, 95 (2018), 260–270. doi:10.1016/j.cemconcomp.2018.09.015.
4 V.N. Nerella, M. Krause, V. Mechtcherine, Direct printing test for buildability of
3D-printable concrete considering economic viability, Automation In Construction, 109
(2020), 102986. doi:10.1016/j.autcon.2019.102986.
5 V.N. Nerella, S. Hempel, V. Mechtcherine, Effects of layer-interface properties on mechani-
cal performance of concrete elements produced by extrusion-based 3D-printing, Construc-
tion and Building Materials 205 (2019), 596–601. doi:10.1016/j.conbuildmat.2019.01.235.
6 V. Mechtcherine, J. Grafe, V.N. Nerella, E. Spaniol, M. Hertel, U. Füssel, 3D-printed
steel reinforcement for digital concrete construction – Manufacture, mechanical prop-
erties and bond behaviour, Construction and Building Materials, 179 (2018), 125–137.
doi:10.1016/j.conbuildmat.2018.05.202.
7 H. Ogura, V.N. Nerella, V. Mechtcherine, Developing and Testing of Strain-Hardening
Cement-Based Composites (SHCC) in the Context of 3D-Printing, Materials, 11 (2018),
1375. doi: 10.1016/j.cemconres.2018.08.003.
8 G. De Schutter, K. Lesage, V. Mechtcherine, V.N. Nerella, G. Habert, I. Agusti-Juan, Vision
of 3D printing with concrete — Technical, economic and environmental potentials, Cement
and Concrete Research, 112 (2018), 25–36. doi:10.1016/j.cemconres.2018.06.001.
9 V. Mechtcherine, V.N. Nerella, K. Kasten, Testing pumpability of concrete using
Sliding Pipe Rheometer, Construction and Building Materials, 53 (2014), 312–323.
doi:10.1016/j.conbuildmat.2013.11.037.
10 V. Mechtcherine, V.N. Nerella, F. Will, M. Näther, J. Otto, M. Krause, Large-scale digital
concrete construction – CONPrint3D concept for on-site, monolithic 3D-printing, Automa-
tion In Construction, 107 (2020), 102933. 10.1016/j.autcon.2019.102933.
11 B. Nematollahi, P. Vijay, J. Sanjayan, M. Xia, V. N. Nerella, and V. Mechtcherine, (2018),
Effect of Polypropylene Fibre Addition on Properties of Geopolymers Made by 3D Printing
for Digital Construction, Materials, 11 (2018), 2352. doi: 10.3390/ma11122352.
12 J. Kaufhold, J. Kohl, V.N. Nerella, C. Schröfl, C. Wenderdel, P. Blankeinstein, V. Mechtcher-
ine, Wood-based support material for extrusion-based digital construction, Rapid Prototyp-
ing Journal, 25 (2019), 690–698. doi:10.1108/RPJ-04-2018-0109.
13 V. Mechtcherine, V.N. Nerella, 3-D-Druck mit Beton: Sachstand, Entwicklungstendenzen,
Herausforderungen, Bautechnik. 95 (2018) 275–287. doi:10.1002/bate.201800001.
14 V. Mechtcherine, V.N. Nerella, Integration der Bewehrung beim 3D-Druck mit Beton,
Beton- und Stahlbetonbau. 113 (2018) 496–504. doi:10.1002/best.201800003.
15 V. Mechtcherine, V.N. Nerella, Beton-3D-Druck durch selektive Ablage: Anforderun-
gen an Frischbeton und Materialprüfung, Beton- und Stahlbetonbau. 114 (2018). doi:
10.1002/best.201800073.
16 R. Schach, M. Krause, M. Näther, V. N. Nerella, CONPrint3D, Bauingenieur. 9 (2017)
355–363.

202
Publications

Journal Publications without peer-review:

1 V. N. Nerella, M. Krause, M. Näther, V. Mechtcherine, 3D printing technology for

on-site construction, Concrete Plant International. (2016) 36–41.
2 V. N. Nerella, E. Secrieru, V. Mechtcherine, Rheological behaviour of fresh concrete
in continuous pumping and circuit-breakdown cases. Part 1: Principals and experi-
mental program, Concrete Plant International. 5 (2015) 222–226.

Conference proceedings:

1 V. N. Nerella, H. Ogura, V. Mechtcherine, Incorporating reinforcement into digital

concrete construction, Annual IASS Symposia: Creativity in Structural Design, In-
ternational association for shell and spatial structures, Cambridge, Massachusetts,
2018.
2 C. Schröfl, V. N. Nerella, V. Mechtcherine, Capillary Water Intake by 3D-Printed
Concrete Visualised and Quantified by Neutron Radiography, Digital Con-
crete 2018, Springer International Publishing, Zürich, 2019: pp. 217–224.
doi:10.1007/978-3-319-99519-9_ 20.
3 V. Mechtcherine, V. N. Nerella, H. Ogura, J. Grafe, E. Spaniol, M. Hertel, U. Füssel,
Alternative Reinforcements for Digital Concrete Construction, Digital Concrete
2018, Zürich, 2019: pp. 167–175. doi:10.1007/978-3-319-99519-9_15.
4 V. N. Nerella, S. Hempel, V. Mechtcherine, Micro- and macroscopic investigations
on the interface between layers of 3d-printed cementitious elements, Int. Conf. Adv.
Constr. Mater. Syst., Chennai, 2017: pp. 557–565
5 V. N. Nerella, M. Krause, M. Näther, V. Mechtcherine, Studying printability of fresh
concrete for formwork free Concrete on-site 3D Printing technology technology
(CONPrint3D), Rheol. Messungen an Baustoffen, Tredition GmbH, Hamburg, Re-
gensburg, 2016: pp. 236–246.
6 V. N. Nerella, E. Secrieru, V. Mechtcherine, Experimental study on influence of
lubricating layer on pumping pressure using Sliding Pipe Rheometer (SLIPER), 19h
International Conference on Building Materials – Ibausil, Weimar, Weimar, 2015:
pp. 1-1247–1254.
7 V. N. Nerella, J. Lipowsky, S. Shyshko, V. Mechtcherine, Numerical modeling
of concrete pumpability tests with Sliding Pipe Rheometer using CFD method,
IAB-Weimar-gGmbH; (Ed.), IAB Science Days / IAB Wissenschaftstage, Weimar,
2014: pp. 46–47.
8 V. N. Nerella, F. Galvani, L. Ferrara, Normal and tangential interaction between
discrete particles immersed in viscoelastic fluids: Experimental investigation as
basis for Discrete Element Modelling of fresh concrete, 8th Int. RILEM Symp.
Self-Compacting Concr. - SCC 2016, RILEM Publications SARL, Washington, 2016:
pp. 171–181.
9 V. N. Nerella, V. Mechtcherine, K. Kasten, Experimental study on pumpability of
concrete using sliding pipe rheometer (SLIPER), Rheol. Process. Constr. Mater. –
7th RILEM Int. Conf. Self-Compacting Concr. 1st RILEM Int. Conf. Rheol. Process.
Constr. Mater., RILEM Publications SARL, Paris, 2013.

203
Publications

Conferences (without a full paper)

1 V. N. Nerella, H. Ogura,V. Mechtcherine, Development and Characterization of

3D-printable Strain-Hardening Cement-based Composites (PSHCC), ACI Fall Con-
vention, Las Vegas, USA, 2018.
2 V. N. Nerella, V. Mechtcherine, Digital construction of building elements: chal-
lenges and opportunities, Norsk Betongdag, Trondheim, Norway, 2017.
3 V. N. Nerella, V. Mechtcherine, Concepts for assuring printability of concrete using
rheology as a tool for additive construction, in: Rheol. Messungen an Baustoffen,
Regensburg, Germany, 2017.
4 V. N. Nerella, J. Lipowsky, S. Shyshko, V. Mechtcherine, Numerical modeling
of concrete pumpability tests with Sliding Pipe Rheometer using CFD, in: IAB
Wissenschaftstage, Weimar, Germany, 2014.
5 V. N. Nerella, L. Ferrara, V. Mechtcherine, Experimental study on contact behaviour
of rigid particles suspended in visco-elastic matrix as basis for distinct element
modelling of fresh FRCs, in: IV International Conference on Particle-Based Meth-
ods. Fundamentals and Applications, Barcelona, Spain, 2014.

204
Schriftenreihe des Institut für Baustoffe
der TU Dresden

Herausgeber: Prof. Dr.-Ing. Viktor Mechtcherine

Anschrift Institut für Baustoffe

Fakultät Bauingenieurwesen
Technische Universität Dresden
01062 Dresden
Telefon: +49 351 463 36311
Telefax: +49 351 463 37268
E-Mail: [email protected]

Heft 2009/1 Marko Butler

Dauerhaftigkeit von Verbundwerkstoffen aus zementgebundenen Ma-
trices und AR-Glass-Multifilamentgarnen
ISBN 978-3-86780-128-7

Heft 2011/1 Petr Jun

Behaviour of strain-hardening cement-based composites (SHCC) un-
der monotonic and cyclic tensile loading
ISBN 978-3-76780-229-1

Heft 2011/2 Nick Bretschneider

Inverse Analyse zur Ermittlung der bruchmechanischen Eigen-
schaften entfestigender und verfestigender zementgebundener Werk-
stoffe
ISBN 978-3-00-034863-1

Heft 2012/1 Matthias Lieboldt

Transport flüssiger und gasförmiger Medien in Betonschichten mit
textiler Bewehrung
ISBN 978-3-86780-299-4

Heft 2012/2 Frank Altmann

A durability concept for strain-hardening cement-based composites
ISBN 978-3-86780-296-3

Heft 2013/1 Sergiy Shyshko

Numerical simulation of the rheological behavior of fresh concrete
ISBN 978-3-86780-371-7

205
Publications

Heft 2014/1 Björn Höhlig

Anwendung der Radiowellentechnologie in der Beton-Technik
ISBN 978-3-86780-394-6

Heft 2014/2 Rabea Barhum

Mechanism of the interaction between continuous and short fibres in
textile-reinforced concrete (TRC)
ISBN 978-3-86780-381-6

Heft 2016/1 Thomas Thiel

Entwicklung von Cellulosefaser-Leichtbeton (CFLC) und Unter-
suchung des bruchmechanischen Verhaltens
ISBN 978-3-86780-498-1
Heft 2016/2 Christian Wagner
Dauerhaftigkeitsrelevante Eigenschaften von dehnungsverfestigen-
den zementgebundenen Reparaturschichten auf gerissenen Betonun-
tergründen
ISBN 978-3-86780-505-6

Heft 2017/1 Ksenija Vasilic

A numerical model for self-compacting concrete flow through rein-
forced sections: A porous medium analogy
ISBN 978-3-86780-521-6

Heft 2017/2 Łukasz Dudziak

Mitigating autogenous shrinkage of ultra-high performance concrete
by means of internal curing using superabsorbent polymers
ISBN 978-3-86780-519-3

Heft 2017/3 Knut Krenzer

Entwicklung eines Zustandsabhängigen DEM-Stoffmodells zur Nach-
bildung von Mischprozessen für Frischbeton
ISBN 978-3-86780-522-3

Heft 2018/1 Iurie Curosu

Influence of fiber type and matrix composition on the tensile behavior
of strain-hardening cement-based composites (SHCC) under impact
loading
ISBN 978-3-86780-555-1

Heft 2018/2 Egor Secrieru

Pumping behaviour of modern concretes Characterisation and predic-
tion
ISBN 978-3-86780-562-9

Heft 2018/3 Erik Mündecke

Tensile behaviour of steel-reinforced elements made of
strain-hardening cement-based composites

206
Publications

ISBN 978-3-86780-564-3

Weitere Dissertationen am Institut für Baustoffe, die nicht in der Schriftenreihe erschienen sind:

2011 Sören Eppers

Assessing the autogenous shrinkage cracking properties of concrete by
means of the restrained ring test

2014 Oliver Millon

Analyse und Beschreibung des dynamischen Zugtragverhaltens von
ultrahochfestem Beton

207