Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Mixture characterization and real-time extrusion quality monitoring for construction-scale 3D printing (Contour Crafting)
(USC Thesis Other)
Mixture characterization and real-time extrusion quality monitoring for construction-scale 3D printing (Contour Crafting)
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Mixture Characterization and Real-Time Extrusion
Quality Monitoring for Construction-scale
3D Printing (Contour Crafting)
by
Ali Kazemian
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CIVIL AND ENVIRONMENTAL ENGINEERING)
December 2018
Copyright 2018 Ali Kazemian
ii
ACKNOWLEDGEMENTS
I would like to express my gratitude to my family for their love, support and encouragement
during my entire life. I dedicate this dissertation to my parents in appreciation of their endless
love and kindness.
I would also like to thank my advisor Dr. Behrokh Khoshnevis for providing me with the
opportunity to work on this exciting project, and for all his support and technical advice. A
special thank you to my dissertation committee member, Dr. Bora Gencturk, for his instructive
guidance and feedback. I also want to thank the other member of my dissertation committee, Dr.
Yong Chen, for his comments.
I am also grateful to my friends and lab mates Omid Davtalab, Hadis Nouri, Amir Mansouri, and
Xiang Gao for their support and encouragement during the past four years. And finally, last but
by no means least, a very special thank you to Dr. Xiao Yuan whose advice and technical
support was helpful in every step of this project.
iii
ABSTRACT
In this research a framework is proposed for performance-based laboratory testing of fresh
printing mixtures. An iterative laboratory testing procedure is described to evaluate the four
suggested workability aspects of a printing mixture, namely, print quality, robustness, shape
stability and printability timespan. Using different ingredients, four printing mixtures are
developed and used to demonstrate and discuss the implementation of the proposed procedure.
The second stage of this research is focused on the real-time quality monitoring of construction-
scale 3D printing. Four techniques are proposed and investigated as real-time extrusion quality
monitoring measures for Contour Crafting process: electrical resistivity measurements, extrusion
pressure measurements, extruder power consumption monitoring, and a vision-based technique.
The experimental results are used to discuss and compare the reliability and responsiveness of
each technique. Based on the obtained results, the vision-based technique is the most reliable and
responsive technique with respect to detecting changes in the water content of a printing mixture.
iv
Table of Contents
ACKNOWLEDGEMENTS ............................................................................................................ ii
ABSTRACT ................................................................................................................................... iii
1. INTRODUCTION ................................................................................................................... 1
1.1. Background ...................................................................................................................... 1
1.2. Research Significance ...................................................................................................... 7
2. LITERATURE REVIEW ...................................................................................................... 10
2.1. Characterization of Conventional Concrete and Mortar Mixtures ................................. 10
2.2. Construction-Scale 3D Printing ..................................................................................... 18
2.2.1. Past Research on Design and Fresh State Properties of Printing Mixtures ............ 19
2.2.2. Past Research on Hardened State Properties of Printing Mixtures ......................... 27
2.3. Quality Monitoring and Quality Control in Concrete Construction .............................. 34
2.3.1. Methods for On-site Testing of Conventional Concretes ....................................... 34
2.3.2. Methods for On-site Testing of Special Concretes ................................................. 36
2.3.3. Real-time Quality Monitoring in Concrete Construction ....................................... 37
2.4. Research Gaps and Problem Statement .......................................................................... 47
3. RESEARCH PLAN (METHODODLOGY) ......................................................................... 50
4. CHARACTERIZATION OF PRINTING MIXTURES ........................................................ 52
4.1. Construction of a Linear Extrusion Setup ...................................................................... 54
4.2. Developing Printing Mixtures ........................................................................................ 56
4.3. Laboratory Testing of Printing Mixtures ....................................................................... 60
4.3.1. Print Quality ............................................................................................................ 60
4.3.2. Shape Stability of Fresh Printing Mixture: Nozzle Design Considerations ........... 63
4.3.3. Shape Stability of Fresh Printing Mixture: Theoretical Analysis ........................... 66
4.3.4. Shape Stability of Fresh Printing Mixture: Experimental Measurement ................ 69
4.3.5. Robustness .............................................................................................................. 76
v
4.3.6. Printability Timespan .............................................................................................. 78
5. REAL-TIME QUALITY MONITORING OF CONTOUR CRAFTING ............................. 82
5.1. Using Electrical Resistivity of Mixture .......................................................................... 84
5.1.1. Optimization of Testing Parameters and System Calibration ................................. 92
5.1.2. Experimental Evaluation and Discussion ............................................................... 95
5.2. Using Extrusion Pressure ............................................................................................... 98
5.2.1. Sensor Installation and Data Acquisition ................................................................ 99
5.2.2. Results and Discussion ......................................................................................... 103
5.3. Using Power Consumption Level of Mixture Agitator Motor ..................................... 105
5.4. Using Computer Vision ................................................................................................ 109
5.4.1. Proposed Algorithm for Vision-based Extrusion Quality Monitoring ................. 112
5.4.2. Calibration and Experimental Evaluation ............................................................. 116
5.4.3. Vision-based Closed-loop Extruder: Concept Demonstration .............................. 121
5.5. Comparison of Investigated Quality Monitoring Techniques ...................................... 124
6. CONCLUSION AND FUTURE RESEARCH ................................................................... 127
6.1. Research Summary and Conclusions ........................................................................... 127
6.2. Recommendations for Future Work ............................................................................. 129
7. REFERENCES .................................................................................................................... 131
vi
List of Figures
Figure 1: Big Canopy system by Obayashi Corp. [10] ................................................................... 2
Figure 2: The cross-section of the Obayashi ABCS integrated construction automation system-
the self-contained ‘factory’ completes two floors before rising [10] ............................................. 2
Figure 3: (a) Smart dynamic casting (b) Mesh mold projects at ETH Zürich [11] ........................ 4
Figure 4: Contour Crafting Machine............................................................................................... 5
Figure 5: Layer-based construction using Contour Crafting .......................................................... 5
Figure 6: Enhance articulated robot (left) and the printed sulfur concrete sample (right) [18] ...... 6
Figure 7: (a) VSI=0- The mixture mass is homogeneous and no evidence of bleeding ............... 12
Figure 8: J-ring test [24] ............................................................................................................... 12
Figure 9: Flow table for measuring flow of hydraulic cement mortar [26] .................................. 13
Figure 10: Details of column mold [27] ....................................................................................... 14
Figure 11: Penetration test [28]..................................................................................................... 15
Figure 12: V-funnel mold [30] ...................................................................................................... 15
Figure 13: Constitutive relationships for fresh concrete plotted on a flow curve [32] ................. 17
Figure 14: Twenty years of 3D printing in architecture, an infographic by ELstudio [33] .......... 18
Figure 15: Sand gradation methods used by Weng et al. [40] ...................................................... 20
Figure 16: The printing result for mixture A (a) maintained shape until 42
nd
layer (b) sudden
deformation at 42
nd
layer (c) fell down at 43
rd
layer [40] ............................................................. 21
vii
Figure 17: Buildability of studied mixtures [41] .......................................................................... 22
Figure 18: Rheological behavior of the fresh mixture (a) viscosity (b) yield stress, and (c)
thixotropy [41] .............................................................................................................................. 23
Figure 19: Effect of superplasticiser dosage on workability with time [42]................................. 24
Figure 20: Yield stress evolution with time .................................................................................. 25
Figure 21: Filament of length of 2000 mm extruded from the small square nozzle [45] ............. 26
Figure 22: Images of structures built-up with twenty layers of filament in a single process [45] 27
Figure 23: (a) Diagram of cutting slabs and testing directions (b) Equivalent 100mm cube
compressive strengths of printed concretes compared with mold-cast specimens [36] ............... 29
Figure 24: Variation of tensile bond strength with printing gap and comparison with direct tensile
strength [36] .................................................................................................................................. 30
Figure 25: Inter-layer strength of the printed concrete [20] ......................................................... 31
Figure 26: (a) 1 day time gap (b) 10 minutes time gap ................................................................. 33
Figure 27: Three-point bending test setups used by Nerella et al. [48] ........................................ 33
Figure 28: (a) Slump test for measuring workability of fresh concrete on construction site (b)
Measuring early-age compressive strength of concrete specimen [49] ........................................ 35
Figure 29: Some of common quality control tests for self-consolidating concrete as a special
concrete [52] (a) Slump flow test, (b) J-ring test, (c) L-box, and (d) Column segregation test ... 37
Figure 30: Results of three tests with same composition [53] ...................................................... 39
viii
Figure 31: Electrical resistivity response of concretes: (a) minimum point P m and three
development periods on curves ρ–t; (b) transition point P m and three development periods on
curves ρ –t; (b) transition point Pt on curves ρ –t (log scale) [54]................................................ 41
Figure 32: Evolution of UPV of mortar/concrete specimens with (a) w/cm= 0.5 and (b) w/cm=
0.35 [55] ........................................................................................................................................ 43
Figure 33: VERIFI equipment [56]............................................................................................... 44
Figure 34: Variation in slump measurement comparing to acceptable precision in ASTM C143
(shown by gray shaded area) [56] ................................................................................................. 45
Figure 35: Wireless maturity sensor [58]...................................................................................... 46
Figure 36: Early age strength estimation using maturity method [59] ......................................... 47
Figure 37: Proposed framework for performance-based testing of printing mixture at fresh state
....................................................................................................................................................... 52
Figure 38: Schematics of the control system of the printing setup ............................................... 55
Figure 39: Android application developed for remote control of concrete printing setup ........... 56
Figure 40: Tearing in the printed layers due to excessive stiffness of the mixture ...................... 61
Figure 41: (a) Variations in width of printed layer using different mixtures at same printing speed
(dimension conformity) (b) Variations in width of a single layer (dimension consistency) ........ 62
Figure 42: Nozzle design used in this study ................................................................................. 63
Figure 43: Flow trajectory for the used nozzle ............................................................................. 65
Figure 44: Pressure cut plot (a) vertical section (b) horizontal bottom section of the nozzle ...... 66
ix
Figure 45: Cross section of printed layers, extrusion pressure, and defined parameters .............. 67
Figure 46: 3D model and plan of the house considered as a case study [73] ............................... 70
Figure 47: Printing a double layer specimen with zero time gap resulting in deformation of first
layer (Top) Same experiment using a 19-minute time gap (Bottom) ........................................... 71
Figure 48: 3D model and 3D printed components of cylinder stability test ................................. 73
Figure 49: Cylinder stability test................................................................................................... 74
Figure 50:Torque development of samples with admixture and Nano-clay at 600 rpm [76] ....... 75
Figure 51: Mortar penetrometer readings for the three mixtures .................................................. 79
Figure 52: Two-probe technique for electrical resistivity measurements [83] ............................. 85
Figure 53: Wenner technique for measuring resistivity [84] ........................................................ 87
Figure 54: Relationship between BI-2 and external bleeding [89] ............................................... 89
Figure 55: Schematics and wiring of the developed electrical resisitivity setup .......................... 91
Figure 56: Wenner probes installed on a 10x20cm cylinder ......................................................... 91
Figure 57: Main effects plot for variation in electrical resistivity values ..................................... 94
Figure 58: Influence of change in water content on electrical resistivity ..................................... 96
Figure 59: Prolonged electrical resitivity measurements for QCREF and QCREF+10 ............... 97
Figure 60: (a) The force-sensitive resistor [91] used for measuring extrusion pressure and (b) its
construction [92] ........................................................................................................................... 98
Figure 61: Calibration curve for the extrusion pressure setup ...................................................... 99
Figure 62: Embedding the pressure sensor in the nozzle for extrusion pressure measurements 100
x
Figure 63: Extrusion pressure data presentation and logging system ......................................... 100
Figure 64: (a) Pressure sensor data during extrusion process (b) Filtering to obtain peak pressure
values .......................................................................................................................................... 102
Figure 65: Pressure sensor data for QCREF-10 .......................................................................... 104
Figure 66: Description of the power consumption evolution [95] .............................................. 106
Figure 67: Influence of change in water content on agitator motor power consumption ........... 108
Figure 68: Construction activity recognition using computer vision techniques [97] ................ 110
Figure 69: Examples of extrusion nozzle assemblies described in the patent US 8944799 B2 . 111
Figure 70: An example of original frame and produced binary image for top view of layer
extrusion ...................................................................................................................................... 114
Figure 71: A frame with incorrectly detected layer, which is automatically ignored by the
algorithm ..................................................................................................................................... 115
Figure 72: Examples of detecting different extrusion conditions through developed vision system
..................................................................................................................................................... 116
Figure 73: The molded concrete layer and experiment setup for vision system calibration ...... 117
Figure 74: Variations in width of a layer made with QCREF+15 mixture ................................. 119
Figure 75: Change in extrusion rate as a result of change in printing mixture water content .... 120
Figure 76: Performance of the developed vision-based closed-loop extruder for two different
mixtures....................................................................................................................................... 122
xi
Figure 77: Visualization of multi-material construction-scale 3D printing using color mixtures
..................................................................................................................................................... 123
Figure 78. Use of computer vision for construction progress monitoring .................................. 124
xii
List of Tables
Table 1: Standard test methods for characterization of fresh properties of concrete and mortar . 10
Table 2: Visual Stability Index Values (reproduced from [23]) ................................................... 11
Table 3: Mixture proportions [40] ................................................................................................ 20
Table 4: Rheological properties of studied mixtures [40] ............................................................ 21
Table 5: Summary of layer number during which deformation and final collapse occured [40] . 21
Table 6: Printing mixtures studied by Zhang et al. [41] ............................................................... 22
Table 7: Mix proportions of trial mixes [42] ................................................................................ 24
Table 8: Mixture proportions used by Ma et al. (reproduced from [45]) ..................................... 26
Table 9: Mix proportions used by Sanjayan et al. as printing mixture [20] ................................. 31
Table 10: Mixture and process parameteres (Nerella et al. [48]) ................................................. 32
Table 11: Flexural strength of printed specimens with different time gaps and test at 28-day age
[48] ................................................................................................................................................ 34
Table 12: Mixture proportions designed by Li et al. [54]............................................................. 40
Table 13: Mixture proportions designed by Lee et al. [55] .......................................................... 42
Table 14: List of conventional characterization test methods and the measured properties ........ 47
Table 15: Sieve analysis of fine aggregate ................................................................................... 57
Table 16: Printing mixtures proportions (kg/m
3
) .......................................................................... 58
Table 17: Selection basis for printing mixtures ............................................................................ 59
xiii
Table 18: Compressive strength of printing mixture at 7-day and 28-day ages ........................... 60
Table 19: Layer settlement results (ImageJ)- no time gap ............................................................ 71
Table 20: Layer settlement results (ImageJ)- 19min time gap ..................................................... 72
Table 21: Cylinder stability test results for printing mixtures ...................................................... 74
Table 22. Average width of reference and altered mixtures ......................................................... 77
Table 23: Initial setting time, printability limit and blocking limit of mixtures (minutes) ........... 80
Table 24: Workability loss for the three printing mixtures (flow diameters in mm) .................... 81
Table 25: Mixtures used for testing of different quality monitoring techniques .......................... 84
Table 26: Comparison of chloride penetrability levels established for standards based on
electrical resistivity (AASHTO TP 95) and charged passed (ASTM C1202) (reproduced from
[83])............................................................................................................................................... 87
Table 27: Composition and properties of mortar mixtures [89] ................................................... 88
Table 28: Factors and their levels ................................................................................................. 92
Table 29: 3x3x2 factorial design for Wenner technique optimization ......................................... 93
Table 30: Experimental results for Wenner technique optimization (ohm-m) ............................. 94
Table 31: Experimental results for pressure sensor measurements ............................................ 103
Table 32: The experimental results for power consumption measurements (watt) .................... 107
Table 33: Vision system measurements for reference and altered mixtures (all values in mm) 118
Table 34: Sensitivity index of propsoed techniques at designed variation levels ....................... 125
1
1. INTRODUCTION
1.1. Background
After facing criticism for limited adoption of cutting edge technologies for several decades, the
construction industry is undergoing some profound changes now. Building Information Modelling
(BIM) is gradually becoming ubiquitous, in which the essential building design and project data are
generated and managed in digital format throughout the building life cycle [1, 2, 3, 4]. Sensing
automation and IoT technologies are also being used in buildings to improve energy efficiency and
occupants’ satisfaction [5, 6]. Use of drones in construction sites is also reported, where drones are
adopted for applications such as safety inspection [7] and 3D modelling of construction sites [8]. 3D
laser scanning is another emerging technology which provides the information on existing building
conditions with the accuracy needed for construction planning. It also enables comparison of the newly
constructed work against the as-designed model or drawings for quality assurance [9]. Another area
with great potential for innovation and automation is the actual construction process on the
construction site, which is almost entirely done manually. It should be mentioned, however, that there
have been earlier attempts to use robotic systems for in-situ construction. For instance, the Big Canopy
system (Figure 1) was developed in 1980s by Japanese engineers in an attempt to develop an in-situ
robotic construction system. The system consisted of a 13-ton rack and pinion gondola-type
construction lift for vertical material delivery and automated overhead cranes for horizontal delivery
and structural element orientation and positioning [10]. Even though this system resulted in some
improvement in productivity during few projects, it was unable to survive and replace traditional
construction methods. Based on Japan Construction Mechanization Association (JCMA), the failure of
Big Canopy and similar projects (such as Automated Building Construction System (ABCS)- Figure 2)
in 1980s and 1990s could be attributed to the inability to recover the research, development and
2
manufacturing costs, as well as the inability to significantly reduce on-site labor requirements [10].
Installation, maintenance, and repair of the complex robot systems on the construction site required
presence of experts which significantly added to the construction cost.
Figure 1: Big Canopy system by Obayashi Corp. [10]
Figure 2: The cross-section of the Obayashi ABCS integrated construction automation system- the self-
contained ‘factory’ completes two floors before rising [10]
3
There are several projects that can be viewed as a more recent major attempt toward robotic
construction during past several years. Mesh mold and smart dynamic casting which are developed at
ETH Zurich are among these projects. Figure 3 demonstrates smart dynamic casting which is inspired
by slip-forming. In this process a formwork, significantly smaller than the structures produced, is
attached to a 6-axis robotic arm. The robotic arm replaces the hydraulic jacks commonly used in slip-
forming, allowing for precise control of velocity and movement in space. This enhanced robotic
control is informed by the physical properties of the material, which are monitored by a feedback
system measuring the setting and hardening kinetics of the concrete and allowing for best formability
during the delicate transition from plastic to hardened state of the cementitious mixture. In this process
the complex dynamic relationship of the formwork’s geometry, its trajectory and the rheological
properties of the material are the parameters which determine the possible geometrical results [11]. In
mesh mold metal process, an industrial robot bends and welds metal wires to create a 3D mesh (Figure
3-b). Fresh concrete mix is infilled after the fabrication of the mesh. The mesh structure serves a two-
fold purpose. First, the mesh acts as a stay-in-place formwork for the fresh material, and second, it acts
as reinforcement for the hardened material [11, 12].
Another recently developed approach to robotic construction is based the novel idea of scaling up
additive manufacturing techniques. Additive manufacturing is defined as “a process of joining
materials to make objects from 3D model data, usually layer upon layer, as opposed to subtractive
manufacturing methodologies [13]”. In 2015, additive manufacturing was a $5.1 billion industry [14].
Major applications of AM include industrial businesses and machines (20%), aerospace (17%), motor
vehicles (14%), consumer products/electronics (13%), medical/dental (12%), and academic institutions
(11%) [14].
4
(a) (b)
Figure 3: (a) Smart dynamic casting (b) Mesh mold projects at ETH Zürich [11]
It should be mentioned that additive manufacturing technologies have previously been used for small-
scale concept modelling in architecture [15]. However, full-scale automated building construction is an
ongoing revolution in the construction industry. Contour Crafting (CC), D-Shape [16], and Digital
Construction Platform (DCP) [17] are well-known processes invented based on this idea.
Contour Crafting is the pioneering large-scale additive fabrication technology that uses computer
control to exploit the superior surface-forming capability of troweling to create smooth and accurate
planar and freeform surfaces. CC was developed by Dr. Behrokh Khoshnevis of USC and is the first
and one of the few layered fabrication technologies that is uniquely applicable to construction of large
structures such as residential buildings. The Contour Crafting prototype machine (shown in Figure 4)
has work envelope dimensions of 5m x 8m x 3m. Some elements built by CC machine are presented in
Figure 5.
5
Figure 4: Contour Crafting Machine
Figure 5: Layer-based construction using Contour Crafting
The CC technology can be described as a large-scale construction “platform” and various applications
could be realized using variations of this technology. Other than building construction, “infrastructure
development” and “planetary construction” are two other areas that have been already explored in this
regard. Different infrastructure elements may be automatically built with variations of the CC
technology. For example, in a previous research project at USC, sulfur concrete was used as the
material for planetary construction purposes [18]. In this process, sulfur was pre-melted and then
mixed with ingredients (elemental sulfur, sulfur modifier, coarse aggregate and fine aggregate) at
6
150
o
C, and were kept in the reservoir for one hour until the elemental sulfur were modified. The
enhanced articulated robot used in this project as well as a successfully extruded multi-layer sample
(made with sulfur concrete) are presented in Figure 6.
Figure 6: Enhance articulated robot (left) and the printed sulfur concrete sample (right) [18]
Compared to the traditional construction methods, numerous advantages could be offered by a well-
developed layer based automated construction process, including architectural design freedom, lower
construction cost, superior construction speed, higher job site safety, and higher degree of
customization. These benefits highlight the importance of research works in the field of robotic
construction.
A review of past studies, ongoing projects, and structural requirements for building construction
reveals that Portland cement mixtures are the most viable options as the material to be widely used in
automated building construction processes in near future [19, 11, 20, 21]. Other major reasons include
the well-understood unique fresh and hardened properties of cementitious materials, accessibility and
reasonable price, as well as availability of a large variety of admixtures to customize their
7
performance. On the other hand, there are many related challenges yet to overcome. Specifically, there
is no guideline for design and testing of printing mixtures. Interlayer adhesion between 3D printed
layers is a major concern from structural engineering perspective. Furthermore, extensive future
research is needed to develop fundamental understanding of rheology of printing materials in order to
regulate material flow as well as to avoid clogging and segregation [12]. Reinforcement of 3D printed
buildings is another area which is commonly discussed as a major challenge for automated building
construction systems.
In this study, the fresh state properties of cementitious printing mixtures have been investigated. In the
next section, the specific research objectives which have been followed in the present work are
described.
1.2. Research Significance
In this research two important challenges with respect to construction-scale 3D printing have been
investigated. Since currently there is no guideline or step-by-step procedure for characterization of
printing mixture, a framework for testing fresh properties of printing mixture is proposed herein. An
important characteristic of the proposed framework is that it uses a “performance-based” approach to
evaluate properties of printing mixtures. Accordingly, it is not dependent on the employed pumping
and extruding mechanism and focuses on properties of actual printed layers. As such, it could be used
by any construction-scale 3D printing system. Within the developed framework, several test methods
are described to quantitatively evaluate different aspects of workability of a printing mixture. In
specific, the shape stability of printing mixture is extensively discussed through a theoretical analysis
as well as experimental results. The proposed test methods for different aspects of printing mixture
workability are used to evaluate and differentiate performance of four designed mixtures.
8
With respect to real-time extrusion quality monitoring, four techniques, namely, electrical resistivity,
extrusion pressure, agitator motor power consumption, and computer vision are investigated in this
study. The implementation details, as well as the obtained experimental results on the reliability and
responsiveness of each technique form a solid basis for future research.
In summary, the present research work is designed based on the following two major objectives:
Objective 1: The first research objective is to present and examine a framework for
performance-based laboratory testing and evaluation of printing mixtures. It should be mentioned that
only fresh properties of a printing mixture are considered herein. Development of a comprehensive
framework for laboratory testing of printing mixture would be a starting point for systematic
investigation on this special concrete by researchers, and a basis for future specifications and
guidelines. Establishing universal acceptance criteria for printing mixture would be the next step, as it
is possible only after many relevant studies have been carried out, and a reasonable amount of data is
available on the performance of different printing mixtures used in actual construction projects.
Objective 2: The second major research objective is to develop and propose real-time in-
process quality monitoring measures for extrusion of printing mixtures. Considering the continuous
mixture preparation and consumption in robotic construction systems, traditional quality control
techniques could not be applied to construction scale 3D printing. In fact, real-time, in-process quality
monitoring techniques specially developed and customized for construction-scale 3D printing are
needed such that any issue is recognized instantaneously and could be modified in a timely fashion.
This important aspect of robotic construction has not been investigated in depth so far. In this research
work, several novel quality monitoring techniques are proposed. An experimental program is designed
to examine these techniques, and to compare the proposed methods in terms of sensitivity to variations
in printing material as well as measurement repeatability.
9
In the following chapter, the past studies on the design, fresh state properties, and hardened state
properties of printing mixtures and printed elements are reviewed in order to provide an understanding
of current literature on construction-scale 3D printing using cementitious materials.
10
2. LITERATURE REVIEW
To provide the current status of the knowledge and research with respect to material characterization,
on-site quality monitoring, and construction scale 3D printing, a review of conventional
characterization methods as well as some of the major studies with respect to construction-scale 3D
printing are presented in this chapter.
2.1. Characterization of Conventional Concrete and Mortar Mixtures
Many test methods have been developed and standardized over last decades in order to characterize the
fresh and hardened state properties of cementitious mixtures. In this section, the most common
standard test methods which are used to evaluate fresh properties of concrete and mortar mixtures are
described to provide an understanding of available methods. Table 1 presents the list of standard test
methods which are described in this section for evaluation of fresh properties of concrete and mortar.
Table 1: Standard test methods for characterization of fresh properties of concrete and mortar
Test Method Standard
Slump ASTM C 143/C143M - 15a
Slump flow ASTM C1611/C1611M - 14
J-ring ASTM C1621/C1621M - 17
Flow table ASTM C1437 - 15
Static segregation using column technique ASTM C1610/C1610M - 17
Static segregation resistance using penetration test ASTM C1712 - 17
V-funnel EN 12350
L-box EN 12350
Slump is the most common test method for quantitative evaluation of workability of fresh cementitious
mixtures in the lab or on the construction site. Based on ASTM, this test method is considered
applicable to plastic concrete with coarse aggregate up to 1.5 in. To carry out the slump test, a sample
of fresh mixture is placed and compacted by rodding in a cone-shaped mold. The mold is lifted, and the
11
mixture is allowed to subside. The vertical distance measured between the initial and final position of
the center of the top surface of the mixture is reported and used as the slump of the mixture at any
specific time [22].
Slump flow is used to monitor the consistency of fresh, unhardened concrete and mortar and its
unconfined flow potential. It is commonly used for self-consolidating or highly flowable cementitious
mixtures. To measure the slump flow, a sample of fresh mixture is placed in a mold either in the
upright or inverted position. The concrete is placed in one lift without any agitation. Then, the mold is
lifted, and mixture will spread. After spreading is completely stopped, two diameters of the mass are
measured in approximately orthogonal directions. Slump flow is the average of the two diameters [23].
ASTM C 1611 also provides an optional visual rating criteria that could be used to classify the ability
of a mixture to resist segregation (stability). Visual Stability Index (VSI) values with corresponding
criteria to qualitatively assess the stability of mixture are presented in Table 2.
Table 2: Visual Stability Index Values (reproduced from [23])
VSI value Criteria
0= Highly stable No evidence of segregation or bleeding
1= Stable
No evidence of segregation and slight bleeding observed as a sheen on the
concrete mass
2= Unstable A slight mortar halo ≤ 10mm and/or aggregate pile in the concrete mass
3= Highly unstable
Clearly segregating by evidence of a large mortar halo > 10mm and/or a large
aggregate pile in the center of the concrete mass
Figure 7 presents two cases with respect to slump flow and VSI values, demonstrating a highly stable
mixture as well as a highly unstable concrete mixture.
12
(a)
(b)
Figure 7: (a) VSI=0- The mixture mass is homogeneous and no evidence of bleeding
(b) VSI=3- Concentration of coarse aggregate at center of mixture mass and presence of a mortar halo
[23]
J-ring test method is used to determine passing ability of self-consolidating or highly flowable concrete
mixtures. To carry out the J-ring test, a sample of fresh mixture is placed in a slump mold that is
concentric with the J-Ring. The cone shaped mold is then raised, and the mixture is allowed to spread
through the J-Ring (Figure 8). After mixture flow is completely stopped, two diameters of the mass are
measured in orthogonal directions and the J-Ring flow is the average of the two diameters. Then the
test is repeated without the J-Ring to obtain the slump flow, as described before. The difference
between the slump flow and J-Ring flow is an indicator of the passing ability of the mixture [23].
Figure 8: J-ring test [24]
13
Flow table is another test method commonly used to determine flow of hydraulic cement mortars, and
of mortars containing cementitious materials other than hydraulic cements [25]. To measure the flow, a
flow mold is placed at the center of flow table (Figure 9). Then, a layer of mortar (about 25 mm thick)
is placed in the mold and compacted using a tamper. The mold is then filled with mortar and the
second layer is also compacted. Then the mortar surface is flattened by a sawing motion across the top
of the mold. Finally, the mold is lifted away from mortar and the table is dropped 25 times in 15
seconds and the diameter of the mortar mass is measured along the four lines scribed in the table top.
The flow is the resulting increase in average base diameter of the mortar mass, expressed as a
percentage of the original base diameter [25].
Figure 9: Flow table for measuring flow of hydraulic cement mortar [26]
Column technique is a standard test method which is used to determine the potential static segregation
of self-consolidating concrete. The details of the column mold which is used in this test is shown in
Figure 10. To carry out the column segregation test, a sample of fresh mixture is placed in the column
mold without any agitation. The mold is then separated into three sections. Portions of mixture from
14
the top and bottom section are washed on a No. 4 sieve such that coarse aggregate remains on the
sieve. The masses of coarse aggregate in the top and the bottom sections are measured and the percent
static segregation is calculated [27].
Figure 10: Details of column mold [27]
Penetration test (Figure 11) provides a rapid assessment of the static segregation resistance of self-
consolidating concrete during mixture development in the laboratory as well as prior to placement of
the mixture in the field. In penetration test, a sample of fresh mixture is placed in an inverted slump
mold without any agitation being applied to the sample. The hollow cylinder attached to a metal rod is
aligned in the center of the inverted slump mold. The hollow cylinder is then lowered onto the surface
of the mixture and released to freely penetrate into the mixture. The penetration depth is measured and
used to evaluate the static segregation resistance of the mixture.
15
Figure 11: Penetration test [28]
V-funnel test is primarily used to measure the filling ability of self-consolidating and highly flowable
mixtures and can also be used to evaluate segregation resistance [29]. To perform the V-funnel test, the
mold is filled with mixture without applying any vibration and it is left undisturbed for 1 minute.
Then, the gate at the bottom of the funnel is opened and the time for the mixture to completely exit the
funnel is measured and reported as the V-funnel flow time. A long flow time can be due to high paste
viscosity or blockage of flow by coarse aggregates. Non-uniform flow of mixture from the funnel
suggests a lack of segregation resistance [30].
Figure 12: V-funnel mold [30]
16
L-box test is used to measure the filling and passing ability of self-consolidating mixtures. To carry out
the L-box test, fresh mixture is initially placed in the vertical portion of the box. The gate is then
opened and mixture is allowed to flow through a row or reinforcement bars and into the horizontal
portion of the box. The times for concrete to reach points 200 mm (T20) and 400 mm (T40) down the
horizontal portion of the box are recorded. After the flow is completely stops, the heights of the
mixture at the end of the horizontal portion, H2, and in the vertical section, H1, are measured to
compute the blocking ratio, H2/H1 [30].
It should be mentioned that all of the aforementioned test methods are referred to as empirical
characterization methods which are commonly used for laboratory and (sometimes) on-site
characterization of fresh mortar and concrete. These empirical test methods are known to provide a
good indication of mixture field performance as they typically involve the simulation of a relevant field
placement condition and the measurement of a value, such as a time measurement, that serves as an
index of workability. There is another category of test methods which aim at measurement of
fundamental properties of mixtures, often referred to as fundamental rheology measurements. In
theory, rheological test methods measure fundamental parameters that are not specific to the used test
equipment. In reality, however, each rheometer available for cement, mortar, or concrete features
various artifacts and variations in geometry and surface friction such that absolute results can vary
considerably [30]. The characterization of concrete and mortar rheology is based on the concept that
concrete can be considered a fluid. Freshly-mixed concrete is essentially a concentrated suspension of
aggregate particles in cement paste. The cement paste is a concentrated suspension of cement grains in
water [31]. A fluid deforms continuously under a constant shear stress and experiences no recovery of
this deformation upon removal of the load. Therefore, in characterizing the fundamental flow
properties of a material, the relationship between shear stress, τ, and shear rate, 𝛾 ̇, is considered. This
17
relationship is commonly represented with a flow curve. The behavior of a fluid material may be
idealized with a constitutive relationship [30]. Six such relationships equations are shown in Figure 13.
Figure 13: Constitutive relationships for fresh concrete plotted on a flow curve [32]
The Bingham model is the most widely used constitutive relationship for concrete due to
its simplicity and its ability to represent concrete flow accurately [31]. The Bingham model requires
the determination of two parameters- yield stress, 𝜏 0 , and plastic viscosity, 𝜇 as shown in Equation 1
[30].
𝜏 = 𝜏 + 𝜇 𝛾 ̇
(Equation 1)
Yield stress represents the amount of stress to initiate or maintain flow while plastic viscosity describes
the resistance to flow once the yield stress has been exceeded. Increased plastic viscosity results in
greater resistance to material flow. The apparent viscosity is equal to the shear stress divided by the
shear rate at any given shear rate. As such, for a Bingham material, the apparent viscosity decreases
with increasing shear rate [30].
18
2.2. Construction-Scale 3D Printing
Figure 14 is developed by architecture and design firm EL Studio [33] and maps 20 years of 3D
printing in architecture and construction. All these projects were intended for construction of a building
or an architectural element. Considering the projects presented in the graph, it is observed that most of
these projects used either mortar or clay as the 3D printing material. While clay seems a viable option
for non-structural purposes, its mechanical properties are not acceptable for construction projects. On
the other hand, concrete is the most widely used construction material in the world. It is estimated that
the present global consumption of concrete exceeds 11 billion tons every year [34].
Figure 14: Twenty years of 3D printing in architecture, an infographic by ELstudio [33]
The three primary reasons for Portland cement based mixtures to be the most widely used engineering
material for conventional construction are excellent resistance to water, deformability in fresh state,
and the fact that it is usually the cheapest and most readily available material. Unlike wood and
ordinary steel, the ability of cementitious materials to withstand the action of water without serious
19
deterioration makes it an ideal material for building structures to control, store, and transport water.
Formability of cementitious mixtures enables engineers to form structural elements into a variety of
shapes and sizes. This is plastic behavior of fresh mixtures, which enables the material to flow into a
prefabricated formwork. After a cementitious mixture solidifies and hardens into a strong mass, the
formwork can be removed. Finally, the relatively low cost of cementitious materials has significantly
contributed to its wide application. The principal components for making mortar, namely aggregate,
water, and Portland cement are relatively inexpensive and are commonly available in most parts of the
world. It is to be mentioned that there are many other properties of mortar or concrete which could be
significant based on the specific application, such as its fire and termite resistance.
When comparing conventional construction to 3D printing of cementitious materials, “no use of
formwork” in the latter approach is a significant distinction. Formwork typically accounts for 40 to 60
percent of the total construction cost [35], therefore formwork elimination in automated construction
leads to considerable savings. On the other hand, because surface of printed elements is not covered by
molds water evaporation is accelerated, which potentially results in excessive shrinkage and cracking.
In the following sections the past research works on design and testing of cementitious mixtures for
construction-scale 3D printing is presented.
2.2.1. Past Research on Design and Fresh State Properties of Printing Mixtures
With respect to composition of printable mixture, use of different chemical admixtures and high
powder content to enhance the viscosity of the mixture for pumping, transportation in small diameter
hoses and extrusion is common in reported printing mixtures [36, 21, 37, 19, 38]. In addition, there is
not any published information available in the literature about use of coarse aggregate in the printed
mixtures, and the maximum size of aggregate is invariably smaller than 2 mm [37, 21, 39, 38].
20
In a study by Weng et al. (2018), aggregate grading and particle size distribution has been considered
as the main parameter for printing mixture design. Three gradation methods, namely, Fuller Thompson
gradation, uniform-gradations, and gap-gradations were used as described in Figure 15 [40]. Table 3
presents the mixture proportions used in this research. Rheological properties were also characterized
by static/dynamic yield stress and plastic viscosity in Bingham model. In addition, some printing tests
were carried out in order to evaluate the shape stability of mixtures.
Figure 15: Sand gradation methods used by Weng et al. [40]
Table 3: Mixture proportions [40]
OPC Sand W FA SF SP/(g/L)
1 0.5 0.3 1 0.1 1.3
* Ingredient content are expressed as weight proportion of cement content
Rheological test results (Table 4) indicate that mixture A designed by continuous gradation possesses
the highest static yield stress and the lowest plastic viscosity, which is very desirable to ensure low
pumping pressure and high buildability [40]; mixture E possesses a highest viscosity among all the
mixtures, which suggests that the pumpability of mixture E could be the worst.
21
Table 4: Rheological properties of studied mixtures [40]
Printing test results are summarized in Table 5, where layer number associated with noticeable
deformation and final collapse are presented and used as an indication of shape stability.
When using mixture A, the printed object maintained its shape until the 42
nd
layer, followed by large
deformations and finally collapsed at the 43
rd
layer (Figure 16). A similar phenomenon was reported by
these researchers for other mixtures, except that collapse happened at an earlier stage.
Table 5: Summary of layer number during which deformation and final collapse occured [40]
Layer number with noticeable deformation Layer number with final collapse
Mixture A 42
nd
43
rd
Mixture B 30
th
31
st
Mixture C 27
th
32
nd
Mixture D 31
st
36
th
Mixture E 32
nd
34
th
Mixture F 24
th
25
th
Figure 16: The printing result for mixture A (a) maintained shape until 42
nd
layer (b) sudden deformation
at 42
nd
layer (c) fell down at 43
rd
layer [40]
22
Finally, a large-scale printing experiment was carried out with mixture A and a 80 cm structure was
printed successfully without notable deformation. Density, compressive strength and flexural strength
of printed filaments were also characterized. Mechanical performance test results showed that mixture
A has the highest density and appropriate compressive strength, and a relative high flexural strength at
different curing ages [40].
Zhang et al. (2018) studied the buildability and rheological properties (viscosity, yield stress and
thixotropy) of the fresh 3D printing mixture [41]. Table 6 presents the five mixtures that were designed
and used in that study.
Table 6: Printing mixtures studied by Zhang et al. [41]
Material CM CS CC CCR CCS
Portland cement 900 882 882 882 864
Nano clay 0 0 18 18 18
Silica fume 0 18 0 0 18
Fine aggregate 900 900 900 900 900
W/B 0.35 0.35 0.35 0.35 0.35
Superplasticizer (%) 0.26 0.26 0.26 0.26 0.26
Thicker agent (%) 0.0125 0.0125 0.0125 0.0125 0.0125
Retarder agent (%) - - - 0.1 -
In order to evaluate the buildability of different mixtures, the height of extruded filament layers that
did not result in any visual deformation or collapse was used. The experimental results are presented in
Figure 17.
Figure 17: Buildability of studied mixtures [41]
23
The obtained results in this study indicated that the buildability of printing mixture was increased by
150% and 117%, respectively, by addition of a small quantity of NC or SF. Figure 18 shows the
rheological results of the fresh 3D printing mixture. It was reported that the viscosity with NC addition
(CC) exhibits almost the same value with the increase of time, the viscosity was increased by 8.03% as
compared with the fresh mixture without additive (CM) at 0 min. However, the increase in yield stress
and thixotropy, results in a great improvement for the fresh mixture with NC (CC). When SF was
incorporated into the fresh 3D printing mixture (CS), the viscosity, yield stress and thixotropy were
enhanced significantly, especially the yield stress [41].
Figure 18: Rheological behavior of the fresh mixture (a) viscosity (b) yield stress, and (c) thixotropy [41]
24
Another study [42] reports the experimental results regarding the mix design and fresh properties of a
high-performance fine-aggregate mixture for construction-scale 3D printing. The composition and
proportions for printing mixtures which were investigated in this study are presented in Table 7. The
mixtures also contained 1.2 kg/m
3
of 12 mm long polypropylene micro fibers.
Table 7: Mix proportions of trial mixes [42]
Material
Mixture proportions (kg/m
3
)
Mix 1 Mix 2 Mix 3 Mix 4 Mix 5
Sand 1612 1485 1362 1241 1123
Cement 376 446 513 579 643
Fly ash 107 127 147 165 184
Silica fume 54 64 73 83 92
Water 150 178 205 232 257
In order to determine the workability of printing mixtures, authors adopted a shear vane apparatus
(originally used for measuring the shear strength of soil) and shear strength was determined from the
maximum torque. Figure 19 presents the accepted shear strength range (0.3-0.9MPa) for printability of
a mixture, as well as the influence of polycarboxylate-based superplasticizer dosage on the workability
over time.
Figure 19: Effect of superplasticiser dosage on workability with time [42]
25
These researchers reported that Mix 4 (60:40 sand: binder ratio, comprising 70% cement, 20% fly ash
and 10% silica fume, plus 1.2 kg/m
3
micro propylene fibers) was the optimum mixture. This mixture
was then used to print a full scale curved bench. The printed component was 2 m long, 0.9 m
maximum width and 0.8 m high and comprised 128 layers of 6 mm thickness.
Perrot et al. [43] discussed the influence of building rate on the vertical stress which is applied to the
already deposited layers during 3D printing of cementitious mixture. In this study the plate deposition
method was used to evaluate the shape stability. A plate with a known weight was placed on top of
cylindrical sample at a certain time gap until failure was observed on the surface. Furthermore, the
evolution of the yield stress with time at rest was studied and compared to a linear model (proposed by
Roussel [44]) and an exponential model. The results (presented in Figure 20) show that the yield stress
evolution, of the tested cement paste, with time can be considered as linear during the first 40 min. It
follows that the Roussel linear model [15] can be used to describe the yield stress evolution for
construction process lasting less than this critical time. For longer processing, the non-linear model
better describes the yield stress increase.
Figure 20: Yield stress evolution with time
Finally, they proposed a theoretical framework to find the highest building rate for layer by layer
construction-scale 3D printing. This theoretical framework is based on the comparison of the vertical
26
stress acting on the first deposited layer with the critical stress related to plastic deformation that is
linked to the material yield stress.
In a recent study, an environmentally friendly mixture that is compatible with construction-scale 3D
printing is proposed where copper tailing is used to replace sand at six different ratios [45]. Various
fresh and hardened properties of mixtures, including extrudability and flowability has been
investigated in this study. Moreover, extrudability and buildability coefficients are suggested for design
of printing mixtures. Table 8 presents the proportions for the mixtures which were studied in this work.
Table 8: Mixture proportions used by Ma et al. (reproduced from [45])
Mix
No.
Natural
sand
Tailings
Replacement
(%)
Cement
Fly
Ash
Silica
fume
Water Superplasticizer
Polypropylene
fiber (kg/m
3
)
R0 1.2 0 0
0.7 0.2 0.1 0.27 0.0029 1.2
R10 1.08 0.12 10
R20 0.96 0.24 20
R30 0.84 0.36 30
R40 0.72 0.48 40
R50 0.60 0.60 50
The extrudability of fresh paste was evaluated by the continuity and stability of the extruded filament.
Each filament was designed 2000 mm long, extruded in eight return processes. Figure 21 presents the
printed filament extruded from the 8 x 8 mm
2
square opening with a total length of 2000 mm.
Figure 21: Filament of length of 2000 mm extruded from the small square nozzle [45]
27
In order to evaluate shape stability, the structures were designed to vertically stack twenty layers of
extruded filaments of length of 250 mm and width of 30 mm without collapse at a rest time of 10 min.
The authors mention that the 10 min delay was designed to facilitate the evaluation of shape stability
and to avoid the formation of cold joints and weaken interfaces. The ultimate stacking height of R0 –
R30 is 138, 140, 120, 117 mm, respectively. The width of the bottom layer of each case is 30, 31, 33
and 33 mm, respectively. The results for all mixtures are also illustrated in Figure 22. Based on the
obtained experimental results, the authors have determined the optimal mixture which contains 30%
copper tailings.
Figure 22: Images of structures built-up with twenty layers of filament in a single process [45]
2.2.2. Past Research on Hardened State Properties of Printing Mixtures
While the fresh properties of printing mixtures have been the main focus of research in the field of
construction-scale 3D printing, there are some concerns with respect to the hardened properties of
printing mixtures and several research works have investigated properties of printed elements.
28
Considering the construction-scale 3D printing process, the resulting structure is layered and likely to
be anisotropic as voids can form between deposited layers to weaken the structural capability. The
interlayer adhesion could influence the hardened properties of printed elements and structures.
Additionally, a low shrinkage seems essential as the printed elements are not covered by formwork and
this could accelerate water evaporation in the mixture and result in cracking [36].
Le et al. [36] studied the hardened properties of a high-performance mixture extruded through a 9mm
diameter nozzle to build layer-by-layer structural components in the 3D printing process. The mixtures
were reinforced by polypropylene fibers. The effects of the layering process on density, compressive
strength, flexural strength, tensile bond strength and drying shrinkage were investigated. Compressive
strength was measured in both mold-cast and printed specimens. For printed elements, 100 mm cube
specimens were extracted from one 350×350×120 mm slab and three 500×350×120 mm slabs. Nine
cubes were extracted from the 350×350×120 mm slab and loaded in one of three directions: direction I
for specimens 1–3; direction II for specimens 4–6; and direction III for specimens 7–9 (as shown in
Figure 23-a). Figure 23-b presents the obtained results, where SD represents standard 100 mm mold-
cast cubes, S3I-S3III represent 100 mm cubes extracted from the 350x350x120 mm slab, tested in
loading direction I, II and III, S5I-S5III represent 100 mm cubes extracted from three 500x350x120
mm slabs, tested in loading direction I, II and III, and B cy represents 58×63 mm cylinders cored from
the trial curvy bench. Testing of printed samples in various directions relative to the layers revealed a
strength from 75 to 102MPa. The average compressive strengths of the 100 mm cube specimens
extracted from a 350×350×120 mm slab was 102MPa in direction I (specimens 1, 2, 3) and the same in
direction II (specimens 4, 5, 6). In direction III it was 91MPa (specimens 7, 8, 9). Compared to the
standard mold-cast compressive strength, the printed concrete strength was similar in directions I and
II and 15% lower in direction III.
29
(a)
(b)
Figure 23: (a) Diagram of cutting slabs and testing directions (b) Equivalent 100mm cube compressive
strengths of printed concretes compared with mold-cast specimens [36]
Furthermore, 11 mortar specimens with varying time gap between new and old layers were cast to
study the effects on bond strength of printed element. Bond strength was measured using direct tensile
test carried out on cored cylinder specimens. The results are presented in Figure 24.
30
Figure 24: Variation of tensile bond strength with printing gap and comparison with direct tensile
strength [36]
The reported results indicate that increased time gap results in lower bond strength. For instance, the
authors reported that a 30-minute and a 7-day time gap results in 53% and 77% reduction in the tensile
bond strength. The authors used the minimum bond strength of 0.8MPa (recommended by [46]) as the
acceptance criteria. The main implication of the obtained results is that the concrete layers should be
printed upon each other with the shortest possible time gap, so that the structural properties of concrete
elements are not severely affected. However, the underlying mechanisms of the bond strength
(chemical and mechanical) were not discussed by the authors.
Sanjayan et al. [20] studied the parameters which influence the interlayer strength in concrete 3D
printing. To measure the interlayer strength of printed elements, 50x25x30 mm specimens were sawn
from the 250x25x30 mm printed samples and loaded in uniaxial tension. Compressive and flexural
strengths of extrusion-based 3D printed concrete samples were also measured in different directions. A
piston-type extruder was used in this printer, and the material was extruded by means of a piston
through a cylinder measuring 500 mm in diameter and 600 mm in length. A nozzle with a 25 mm x15
31
mm opening and having an angle of 45
o
to the build platform was attached to the end of the extruder.
Specimens were printed with 10, 20 and 30 min delay times. The mixture proportions used in this
study are presented in Table 9.
Table 9: Mix proportions used by Sanjayan et al. as printing mixture [20]
OPC Fine silica sand Course silica sand Water
1.0 0.375 1.125 0.38
* Numbers are mass ratios of the OPC mass
The effect of delay time (i.e. the printing time interval between layers) on the mechanical properties of
extrusion-based 3D printed concrete was investigated in this study. The experimental results for
interlayer bonding strength (measured by carrying out uniaxial tension tests) are presented in Figure
25. These researchers have the moisture level at the surface as an important factor. The moisture level
between the layers is a function of many parameters including printing process, evaporation rate and
bleeding rate of the mixes, as well as the level of moisture expelled to the surface during the extrusion
process.
Figure 25: Inter-layer strength of the printed concrete [20]
32
The compressive strengths of printed concrete layers were different in all three directions, exhibiting
an orthotropic strength behavior. The highest mean compressive strength was observed in the
longitudinal direction where the extrusion pressure is likely to increase compaction and strength. The
lowest mean compressive strength was observed in the lateral direction. The mean compressive
strength in perpendicular (vertical) direction was in-between the strength of the other directions. These
trends were true irrespective of the delay time [20].
Nerella et al. [47] investigated interlayer adhesion of 3D printed layers in micro and macro scales.
Table 10 presents the mixture information and some process parameters which were used in this study.
Table 10: Mixture and process parameteres (Nerella et al. [48])
For microstructural study of interlayer properties, specimens were prepared with time gaps of 1 minute,
10 minutes, and 1 day between two printed layers under investigation. Scanning electron microscope
(SEM) images of the layer interface with 10 minutes and 1 day time gap are shown in Figure 26. It can
be observed that for the sample printed with a 1-day time gap, cavities at interface of clearly separated
layers exist. For the other sample, on the other hand, no interface could be detected. Based on this, the
authors concluded the evident influence of the time gap on the quality of layer interface.
33
(a) (b)
Figure 26: (a) 1 day time gap (b) 10 minutes time gap
For macroscopic studies, these researchers used 3-point bending test to measure flexural strength of
samples printed with different time gaps. The sample orientations which were used for testing are
presented in Figure 27.
Figure 27: Three-point bending test setups used by Nerella et al. [48]
Table 11 presents the results for flexural strength measurements. It is evident that mixture C2 has
much better interface strength in comparison to mixture C1. The authors have recognized “differences
in material composition in conjunction with corresponding rheological properties” as the major source
of these effects. Mixture C2 had a finer binder containing 15% micro silica but superior rheological
properties due to replacement of 30% of cement by fly ash [48].
34
Table 11: Flexural strength of printed specimens with different time gaps and test at 28-day age [48]
2.3. Quality Monitoring and Quality Control in Concrete Construction
As mentioned before, the second main objective of the current research is real-time quality monitoring
of fresh mixture during Contour Crafting process. It should be noted that Contour Crafting process is
designed for on-site automated building construction. As such, quality control and quality monitoring
techniques which are suitable for on-site application are mainly considered herein, and rheological
measurements, as well as similar fundamental workability test methods which are mostly used for
laboratory studies are not the focus of discussion.
2.3.1. Methods for On-site Testing of Conventional Concretes
Based on the size and importance of a construction project, a large number of experiments might be
carried out in order to design and validate a mixture prior to its production in large volume for use in a
project. Different aspects such as fresh behavior (setting time, plastic shrinkage, workability, air
content, etc.), mechanical strength (compressive strength, flexural strength, tensile strength, creep,
modulus of elasticity, etc.), and durability of hardened concrete (chloride resistance, sulfate attack
resistance, permeability, carbonation, etc.) are usually investigated using various standard procedures,
35
some of which take months to be completed. There are various mixture design methods used by
engineers, such as ACI Recommended Practice 211.1 which is commonly used in north America. On
the other hand, after a mixture is designed, quality monitoring of fresh concrete on the construction site
is commonly simple and limited. For conventional concrete mixtures, quality control is often limited to
slump and early age strength measurement (Figure 28), and sometimes is complemented by unit weight
and air content measurements.
(a)
(b)
Figure 28: (a) Slump test for measuring workability of fresh concrete on construction site (b)
Measuring early-age compressive strength of concrete specimen [49]
36
2.3.2. Methods for On-site Testing of Special Concretes
In general, on-site quality control for special concretes is more comprehensive than conventional
concretes. For instance, procedures for quality control of Self Consolidating Concrete (SCC) are
discussed herein. SCC is selected for discussion because, similar to a printing mixture, these are mainly
fresh properties of SCC that differentiate it from other types of cementitious mixtures. American
Concrete Institute suggests that, when producing SCC, at least the slump flow and visual stability
index (VSI) tests should be performed each day by testing the first batch of SCC, and then consecutive
batches until two consecutively produced batches are within specification. Thereafter, slump flow and
VSI testing should be performed as per the project requirements. Testing should be performed as
outlined in ASTM C 1611 [50]. Furthermore, unit weight and air content should be assessed at the job
site for quality assurance [51]. Slump flow test, as well as J-ring test, L-box test, and column
segregation test are presented in Figure 29. Among these, slump flow test (along with VSI) is the only
test which is commonly used as on-site quality control measure for checking variations in produced
SCC mixtures. Moreover, the frequency of designated quality control experiments is variable and is
often determined based on the mixture type, environmental conditions, and the importance of a project.
In general, it seems that for a test to be accepted as an on-site quality control measure, it needs to be
fast, use only small volume of material, and reliable.
37
(a)
(b)
(c)
(d)
Figure 29: Some of common quality control tests for self-consolidating concrete as a special concrete [52]
(a) Slump flow test, (b) J-ring test, (c) L-box, and (d) Column segregation test
2.3.3. Real-time Quality Monitoring in Concrete Construction
In robotic construction systems the construction speed is considerably higher than a conventional
construction process. Layer-based robotic construction technologies such as Contour Crafting promise
custom-designed houses completed in a short time (e.g. printing building shell in one day). As such,
traditional quality control methods which are manual, need sample preparation, and are time-
consuming cannot be used by automated construction systems. Moreover, in conventional concrete
construction large volumes of concrete are poured (slab, beam, etc.) in a relatively short time, and as
long as structural requirements are satisfied, small variations in mixture proportions might not cause
any problem. On the other hand, in construction-scale 3D printing layers of 1-2inch dimensions are
38
being extruded, and small variations in printing mixture could possibly lead to process failure.
Deviations from specifications could result in extrusion of layers with undesirable properties, which
later can cause significant issues such as failure and collapse of freshly printed structure. As such,
automated construction processes need real-time, in-process, accurate quality monitoring techniques
specially developed and customized for construction-scale 3D printing such that any issue is
recognized instantaneously and could be modified in a timely fashion. However, considering the fact
that all industrial construction-scale 3D printers are at early ages of development, such quality
monitoring procedures have not been developed yet and no extensive research has been done with
respect to this important aspect of robotic construction.
While there is not any conventional quality control method for fresh concrete that can be directly
applied to robotic construction with cementitious materials, there are some commercial products and
research projects that aim at continuous measurement of properties of concrete. These studies and
products are presented here to provide relevant insight.
Reinhardt and Grosse [53] developed a testing device which utilizes the velocity of ultrasound (US)-
waves in order to continuously monitor the setting and hardening of cementitious materials. The
equipment is triggered automatically on intervals which are chosen by the operator in advance.
Depending on the binder, the intervals are 5–10 min. Once the container is filled with mortar and
compacted, the transducers are connected to the hardware such that the first measurement can be taken
several minutes after mixing of the mortar. All subsequent measurements are taken and stored
automatically.
In order to examine the reproducibility of the developed testing equipment, three mortar mixtures with
the same composition were produced and tested with the same equipment. The results are presented in
Figure 30, where the measured velocities vary only approximately 1%. This low variation proves the
39
high reproducibility of the proposed approach. The authors also tried to identify the initial and final
setting time using this approach and suggested that the point of first maximum in curvature of the
velocity versus age plot could be used as the initiation of setting. No conclusive result was obtained for
final setting.
Figure 30: Results of three tests with same composition [53]
Li et al. [54] investigated the possibility of using continuous electrical resistivity measurements for
monitoring the hydration process of fresh cementitious mixtures. The mixtures which were used in this
study are presented in Table 12.
40
Table 12: Mixture proportions designed by Li et al. [54]
These researchers have identified two points on the electrical resistivity development curves, namely,
the minimum point (Pm) and the transition point (Pt). The time tm at which Pm occurred, represented the
time when the curve dropped to a minimum and before increasing as a result of the onset of hydration.
The time tt at which Pt occurred, marked the time when the kinetics of hydration transition changed
from the setting to the hardening stage. Three examples of the electrical resistivity development with
time for Mix 1, the concrete containing a superplasticizer, Mix 3, and concrete with a higher water
cement ratio (Mix 10) are shown in Figure 31. Figure 31 (a) shows the resistivity changes with time up
to 400 minutes and the critical points, P m, are marked as solid dots on the curves. Figure 31 (b) shows
the resistivity changes up to 1440 minutes on a log-scale while the second critical points P t, are
identified on the curves.
41
(a) (b)
Figure 31: Electrical resistivity response of concretes: (a) minimum point P m and three development
periods on curves ρ–t; (b) transition point P m and three development periods on curves ρ –t; (b) transition
point Pt on curves ρ –t (log scale) [54]
Based on the experimental results, these researchers have developed two equations to estimate the
initial setting time and the final setting time of concrete mixtures. The following equation provides a
quantitative relationship between final setting time (t f) and tt [54]:
tf= 0.9202tt +0.2129
(Equation 2)
The coefficient of determination for this relationship has been reported as 0.9895 which indicates the
high correlation between the two parameters, given the experimental results of this study. The second
developed formula estimates the initial setting time [54]:
ti = 1.8807tm + 0.4429tt (Equation 3)
where ti=initial setting time (h); tm and tt=times of occurrence of the minimum point and transition
point (h), from the resistivity response. In summary, the experimental results and findings of this
research imply that electrical resistivity could be used to detect the initial and final setting time of a
cementitious mixture.
42
Lee et al. [55] developed an ultrasonic pulse velocity (UPV) measurement system to monitor concrete
and mortar mixtures (made with different w/cm ratios) during the first 24 hours. Table 13 presents the
details of mortar and concrete mixtures used in this study.
Table 13: Mixture proportions designed by Lee et al. [55]
Figure 32 shows the evolution of the UPV during the first 24 hours for designed mortars (hollow marks
in the figures) and concretes (solid marks). The UPV development curves for all mixtures was similar
to a curve that consists of three steps. At very early ages in the first step, the UPVs of all mixtures with
the same w/cm were almost the same regardless of the presence of coarse aggregates or FA. Lee et al.
explained this based on the hypothesis that ultrasonic waves propagate through the phase of viscous
suspension, which is initially common among all specimens [55]. The UPV profiles at later ages
(second and third steps) differed from each other: UPVs of concretes were larger than those of mortar
mixtures and those of OPC series were greater than those of fly ash incorporated mixtures. These
features were respectively explained by the higher stiffness of coarse aggregates and the retarding
effects of fly ash.
43
(a) (b)
Figure 32: Evolution of UPV of mortar/concrete specimens with (a) w/cm= 0.5 and (b) w/cm= 0.35 [55]
In addition, the UPVs corresponding to the initial and final setting, identified through the method of
penetration resistance, were examined for various mortar mixtures. As a result, certain ranges of UPV
with reasonable widths were suggested for the initial setting times: i.e., 800 –980 and 920– 1070 m/s
for the initial setting of OPC and FA mortar series, respectively. The UPVs corresponding to the initial
setting increased with decreasing w/cm in mortar specimens but did not show a monotonous tendency
in concrete specimens. The same was true of the UPVs corresponding to the final setting. In summary,
Lee et al. concluded that the methods and monitoring device used in this research were useful for the
in-situ monitoring of the setting of concrete, particularly in HPC [55].
Khayyat and Libre [56] developed an automated system that enables a real-time measurement and
recording of the properties of fresh concrete in a truck mixer (VERIFI). VERIFI automatically adds
water and/or water reducing admixtures within the allowable water-to-cementitious ratio (w/cm) limit
to manage and maintain the slump at the target value at the plant, during transit, and at the jobsite. All
water or admixture additions are recorded and can be monitored online during construction. This
automated slump adjustment reduces the variation in fresh properties of concrete and improves
44
uniformity of delivered concrete. The hardened properties are also expected to be improved since
lower amount of water is added through automated VERIFI slump retention. Figure 33 illustrates
different parts of VERIFI system. Sensors on the truck measure concrete slump, temperature, additions
of water and admixture, drum speed, number of revolutions, and time of activity (loading, arrive site,
begin pour, finish pour, etc.).
Figure 33: VERIFI equipment [56]
Each time the slump value is more than 0.5 inch lower than the target, VERIFI can automatically add
water or admixture to reach the slump target. VERIFI automatically calculates the amount needed to
reach the target slump. The ready-mix company sets up VERIFI instructions for each mix design or
group of mix designs. These instructions set how the truck is to manage a load of concrete to comply
with project specifications, including mixing requirements after batch and water additions, and when to
45
add water or admixture. All data is recorded in real time and transmitted via cellular connection to
VERIFI servers, where all data can be viewed from any internet connected device. In addition, VERIFI
mounts displays in the cab and on the outside of the truck, where drivers, inspectors, and contractors
can see current concrete properties, including the amount of water added to the load versus the
maximum water content. Slump is calculated using a proprietary algorithm that considers the drum
speed, hydraulic pressure to turn the drum (related to torque), load size, and mix design. The relation
between the slump values reported by VERIFI and manually measured slump is shown in Figure 34.
The regression analysis of data reveals that slump is overestimated in low slump concrete while it is
underestimated in concrete with high slump values. In zero slump concrete, the spread is 0.80 inch that
means VERIFI reported the slump 0.8 inch higher than the value measured by the technicians. On the
other hand, in flowable concrete with slump value of 10 inches, VERIFI reported the slump 0.4 inch
lower than the slump measured by the technicians.
Figure 34: Variation in slump measurement comparing to acceptable precision in ASTM C143 (shown by
gray shaded area) [56]
46
With respect to early-age strength estimation of concrete, wireless sensors have been developed which
enable contractors to get real-time information about the strength and temperature of poured concrete
as it matures [57].
Figure 35: Wireless maturity sensor [58]
These 2-by-2 inch battery-powered sensors (Figure 35) could be tied to the rebar before concrete is
placed. The sensors use Bluetooth technology to transmit data that can read with an Android or iOS
smartphone app. The apps draw from data stored in the cloud to analyze the performance of the
concrete [58]. In this system, the strength estimation of concrete over time is based on ASTM C1074
standard for maturity-based strength (Figure 36).
47
Figure 36: Early age strength estimation using maturity method [59]
2.4. Research Gaps and Problem Statement
A brief review of the most common test methods for characterization of concrete and mortar mixtures
in fresh state was provided section 2.1. Table 14 presents the list of properties which are measured and
evaluated by these empirical workability test methods.
Table 14: List of conventional characterization test methods and the measured properties
Test Method Measured property/properties
Slump Deformation under self-weight
Slump flow Flowability under self-weight
J-ring
Passing ability (through
reinforcement)
Flow table Flow ability
Static segregation using column technique Segregation
Static segregation resistance using penetration test Segregation
V-funnel
Flow and passing ability under
self-weight
L-box
Flow and passing ability under
self-weight
48
As it is shown in the table, some test methods including slump, slump flow, V-funnel and L-box aim at
measuring deformations and flowability of fresh mixtures. Deformation and flow under self-weight is a
desirable property for conventional mixtures and for self-consolidating mixtures and is actually
required for some applications. In construction-scale 3D printing, on the contrary, the freshly printed
layer is expected to show minimum deformations under its own weight and deformations and flow
should happen only as a result of the extrusion pressure. Other than flowability and deformation, some
test methods such as column technique and penetration test evaluate segregation resistance of concrete
mixture. Segregation resistance is a critical property for flowable mixtures such as self-consolidating
and self-leveling mixtures and is specifically important when coarse aggregate (3/8” or more) is
present in the mixture. For printing mixtures, segregation is not anticipated for two reasons. First, due
to requirements of construction-scale 3D printing, the viscosity of printing mixtures must be high
which significantly reduces the risk of segregation. Secondly, there has not been any published data
regarding use of coarse aggregates in printing mixtures so far, which eliminates the possibility of
severe segregation in the fresh printing mixtures. The third property which is evaluated by some test
methods such as L-box and J-ring is passing ability. Passing ability refers to the ability of the fresh
mixture to flow through the narrow openings between reinforcement bars and mainly depends on
aggregate characteristics and paste volume. Similarly, this is not applicable to printing mixture as it is
not designed to flow through steel reinforcement and other techniques are considered for reinforcement
of printed structures. Considering the aforementioned reasons, there is a need for development of new
test methods for printing mixtures, considering the requirements of construction-scale 3D printing
process. It should be noted that rheological measurements can provide insight and understanding of
properties of mixture which affect the performance. However, they cannot be directly used as an
indication of performance of mixtures in construction scale 3D printing. In addition, various pumping
49
and extrusion mechanisms which are used in industry and by researchers make it impossible to
describe a specific range for rheological parameters that could be considered acceptable universally.
Currently there is no relevant guideline or widely accepted procedure for characterization and
evaluation of printing mixtures. Similar to other special concretes, well-defined acceptance criteria are
needed for evaluation of short-term and long-term performance of printing mixtures. As discussed in
this chapter, few prior studies have focused on specific properties of printing mixtures such as shape
stability (also called shape retention and green strength) [60, 42, 61, 62, 63]. However, a
comprehensive list of performance requirements and test methods for a printing mixture has not yet
been developed. The present research aims at providing an understanding of the requirements for a
mixture to be used by construction-scale 3D printing process. Some characterization methods have
been proposed within a framework for step-by-step evaluation and design of a printing mixture.
50
3. RESEARCH PLAN (METHODODLOGY)
The current research work includes two major stages. In the first stage, a framework for performance-
based laboratory testing of fresh printing mortar is proposed. As mentioned before, a widely accepted
list of performance requirements and test methods for a printing mixture has not yet been developed. In
an attempt to address this knowledge gap, the proposed framework provides a step-by-step procedure
for testing of fresh printing mixture. This testing procedure is designed such that it is applicable to
different concrete 3D printing systems, as it focuses on properties of printed layers rather than
employed pumping or extrusion mechanism (performance-based testing). Within the framework,
properties of a fresh printing mixture such as print quality, shape stability, robustness, and printability
timespan are described. In addition, to enable quantitative evaluation of these properties for a specific
mixture, several test methods have been proposed and described. In order to demonstrate and elaborate
on the proposed framework, the results of a comprehensive experimental program which is carried out
using four different printing mixtures are presented and discussed.
The second stage of the present work focuses on the real-time extrusion quality monitoring of Contour
Crafting. This is highly critical as variations in materials and jobsite conditions are inevitable in any
construction project. In the case of Contour Crafting which aims at on-site automated building
construction, these variations could affect the quality of the printed structure (surface quality,
deformations, etc.), and in severe cases could result in structural failure and collapse of freshly printed
structure. It should be noted that, to the best of author’s knowledge, this important aspect of automated
building construction has not been investigated before, and the present work is the first extensive
research on this topic. As such, and in order to provide a basis for future endeavors, several novel
techniques are investigated herein as real-time quality monitoring measures for Contour Crafting.
Electrical resistivity, extrusion pressure, agitator motor power consumption, and computer vision are
51
the techniques which are studied. Initially, the sensory system development, calibration and
optimization of each technique is discussed. Then, in order to conduct a reliable evaluation and
comparison of proposed techniques, a series of experiments are designed and carried out, wherein
changes in water content (aggregate moisture content) is considered the major source of variations in
the printing mixture. Based on the experimental results from the first stage of the work, a reference
mixture is selected and 6 levels of variation in free water content is applied to this mixture, resulting in
a total of seven mixtures. The results of the experimental program are used to discuss and compare the
responsiveness (sensitivity to variations) and reliability of proposed quality monitoring techniques.
Finally, some recommendations are made regarding real-time quality monitoring of Contour Crafting
process.
52
4. CHARACTERIZATION OF PRINTING MIXTURES
Figure 37 presents the proposed performance-based framework for laboratory testing of printing
mixture in the fresh state. Based on the past experience as well as other studies [47, 42], it seems that
printing mixtures could be characterized by the high powder content, limited maximum aggregate size,
increased paste fraction, and use of Viscosity Modifying Agent (VMA). In fact, reported mixture
proportions of successful printing mixtures could be used as a starting point.
Figure 37: Proposed framework for performance-based testing of printing mixture at fresh state
After designing an initial mixture, fulfilling print quality requirements is the first step in the proposed
iterative mixture assessment and modification process. Three requirements are proposed with respect
53
to print quality, namely, surface quality, squared edges, and dimension conformity and consistency. All
these three requirements must be satisfied by a mixture with acceptable print quality.
Next, shape stability of a mixture must be examined and relevant adjustments should be made
accordingly. Shape stability is defined as the ability of extruded material to withstand self-weight and
also the load applied as a result of following layers being printed (the extrusion pressure and also
weight of deposited layers). In this regard, cylinder stability test is proposed herein for rapid evaluation
and comparison of influence of different materials (additives or admixtures) on shape stability of a
mixture. However, acceptance or rejection decision must be based on layer settlement experiment,
where concrete layers are printed on top of each other with the same extrusion mechanism as the full-
scale concrete printer. No visible deformations should occur when target interlayer time gap is used.
Next, robustness of developed mixtures is evaluated, where the influence of material variations on the
fresh properties of the printing mixture is investigated. Robustness of a printing mixture is defined as
the capacity of mixture to retain its printability when small variations in the properties or quantities of
the constituent materials occur. In this study, “change in water content” is considered as the major
source of variation and ±10liter/m
3
mixture is the water content variation level selected to evaluate
mixture robustness.
The next step in laboratory testing of a printing mixture refers to printability timespan of a mixture. To
this aim, printability limit (the longest time when a mixture can be printed with acceptable print
quality) and blockage limit (the longest time when a mixture can remain in the nozzle before the
blocking happens) are measured and reported for a specific mixture. Finally, when the laboratory
testing is aimed at developing a mixture for an actual construction project, the proposed framework
suggests a large-scale verification test. It should be carried out using the full-size printer in a similar
ambient temperature and humidity as the intended project. Using the same concrete batching, mixing,
54
and transporting equipment as the actual project is highly recommended. The main function of the
proposed verification test is to subject the designed mixture to actual jobsite-based assessment.
In following sections of this chapter, the experimental program which is carried out based on the
proposed framework is presented and the details of the developed concrete printer, as well as proposed
requirements and test methods for print quality, shape stability, robustness, and printability timespan of
a printing mixture are discussed.
4.1. Construction of a Linear Extrusion Setup
Investigation on the fresh properties of printing mixtures requires a large number of experiments to be
carried out, using a variety of materials and chemical admixtures. Carrying out all these experiments
using a full-scale building printer such as Contour Crafting gantry robot is costly, difficult, and time-
consuming. As such, a linear printing setup is developed
1
for this research work. The developed system
is capable of printing up to 10 layers of 1.2m length. The nozzle prints 38.1x25.4mm (1.5x1inch)
layers. The control system (Figure 38) is developed using a combination of Arduino Mega (based on
ATmega1280) and Arduino Uno (based on ATmega328) microcontrollers. Motion system consists of
two stepper motors and one DC motor. The system is able to print layers at different linear speeds and
deposition rates. A linear speed of 6 cm/s is used in this study to print cementitious layers.
1
This was done in collaboration with former PhD student Xiao Yuan
55
Figure 38: Schematics of the control system of the printing setup
In addition, Bluetooth communication with the printing setup enables the user to easily control the
system using a smartphone or tablet. An Android application (Figure 39) is developed to facilitate the
wireless control of the setup. This application enables the user to select the parameters (such as linear
speed and deposition rate) for concrete printing process, with a set of pre-defined commands such that
one or several layers of concrete can be printed conveniently.
56
Figure 39: Android application developed for remote control of concrete printing setup
4.2. Developing Printing Mixtures
After developing the linear printing machine, several printing mixtures are designed for extrusion tests.
Type I/II Portland cement is used for preparing all mixtures. A commercially available manufactured
sand with nominal maximum aggregate size of 2.38mm is used as fine aggregate. This sand is oven
dried before packaging, which ensures the consistent moisture content of particles. The sieve analysis
results are presented in Table 15.
A polycarboxylate based high-range water-reducing admixture (HRWRA) with a specific gravity of
1.1, and a pH value of 5.6 is used to achieve the required workability for the mixtures. Typical addition
rates for this HRWARA (commercially known as ADVA Cast 600) varies from 130 to 650 mL/100 kg
of cementitious materials. Furthermore, in order to increase the plastic viscosity and cohesiveness of
printing mixtures, a commercially available viscosity modifying admixture (VMA) for anti-washout
57
concrete is used. Typical dosage for this admixture is 260 to 1040 mL/100 kg of cementitious material.
Commercially labeled as “V-Mar3”, it works by increasing the viscosity of the concrete while still
allowing the concrete to flow without segregation. The specific gravity of V-Mar3 is 1.02 and it has a
pH value of 6.
Table 15: Sieve analysis of fine aggregate
Sieve
Percent passing
Number Size (mm)
4 4.75 100
8 2.36 83
16 1.18 62
30 0.6 38
50 0.3 19
100 0.15 w8
200 0.075 3.5
MASTERFIBER F100 fiber (manufactured by BASF) is used as a shrinkage and temperature
(secondary) reinforcement for one of printing mixtures. Addition of polypropylene fibers inhibits and
controls the formation of plastic and drying shrinkage cracking in cementitious materials [64]. This is
critical for printing mixture due to anticipated higher rate of water evaporation for plastic material.
This is because no formwork is used in automated construction and surface of printed structure will
remain uncovered. The length of used fiber is 6mm, the aspect ratio is 29, the specific gravity is 0.91,
and the tensile strength is 60,000 psi (415 MPa).
A highly-purified attapulgite clay, refined from bulk attapulgite, is used in one of the printing mixtures.
The particles are 1.75 µm in average length and 3 nm in average diameter, thereby considered to be
Nano-sized. The specific gravity of this clay is 2.29. The clay particles have been chemically exfoliated
to preserve their uniform shape and size while removing all impurities (such as quartz and swelling
clays). The aspect ratio (average length divided by average diameter = 583) of the clay particles is very
high [65]. Therefore, they may form a highly entangled gel even at a small volume concentration,
58
provided they are properly dispersed into individual particles. This Nano clay is commercially
available and several studies [65, 66, 67] have investigated the influence of addition of this clay on
properties of other types of concrete, including formwork pressure of self-consolidating concrete. A
significant difference between Nano-clay and supplementary cementitious materials (SCM) is the
dosage. While typical dosages of SCMs in concrete is 10-30% by mass of cement, clays have been
reported to be effective when small replacements dosages (less than 0.5% of cement mass) are
included. This is important from large-scale material production standpoint and issues related to
handling and mixing of materials.
Silica fume is also used as a supplementary cementitious material to replace Portland cement at 10%
(suggested addition dosage is 5-15% by cement mass). The specific gravity of silica fume is measured
2.2. Silica fume is proved to add cohesiveness and reduce bleeding of fresh concrete, and to improve
mechanical strength, long-term durability, and impermeability of hardened concrete [34, 68].
The mixture proportions for the different printing mixtures are presented in Table 16. An entrapped air
content of 20 liters per cubic meter of concrete is assumed. The total mixing time for all mixtures is 8
minutes to ensure homogeneity of the resulting mixture. To ensure uniform dispersion, chemical
admixtures, Nano-clay and silica fume are mixed with water (stirred for 30 seconds) before the mixing
procedure started.
Table 16: Printing mixtures proportions (kg/m
3
)
Mixture
ID
Proportions
Fine
aggregate
(SSD)
Portland
cement
Free
Water
Silica
fume
Fiber
Nano
clay
(%)*
Superplasticizer
(%)*
Viscosity
Modifying
Agent (%)*
PPM 1379 600 259 0 0
0 0.05 0.11
SFPM 1357 540 259 60 0 0 0.16 0
FRPM 1379 600 259 0 1.18 0 0.06 0.10
NCPM 1379 600 259 0 0
0.30 0.15 0
* The percentages are reported by cementitious materials mass
59
Plain printing mixture (PPM) only incorporates Portland cement as the cementitious material and is
selected as a reference and basic printing mixture. For silica fume incorporated printing mixture
(SFPM), Portland cement is replaced by silica fume at 10% of cementitious materials by weight, which
is commonly considered to be the optimum replacement level [69, 34]. Fiber-reinforced printing
mixture (FRPM) incorporates 1.18 kg/m
3
of polypropylene fibers, which is equivalent to 0.13% of total
concrete volume. The fiber dosage is selected based on the manufacturer recommendations for
effective shrinkage prevention. Finally, Nano-clay incorporated printing mixture (NCPM) is designed
and prepared to study the influence of attapulgite clay on the fresh properties of printing mixture.
Previous studies [66] reported that purified attapulgite clay significantly accelerates rate of recovery of
formed internal 3D network after being broken down under shear and maintained under stress. This
behavior can be highly related to the extrusion process wherein the mixture is initially under shear
stress while the material will be at rest after the layer is extruded. It could be anticipated that the
addition of the Nano-clay increases the rate of green strength recovery after concrete is printed. Table
17 summarizes the considerations and basis for selection of the four printing mixtures.
Table 17: Selection basis for printing mixtures
Mixture
ID
Selection basis
PPM
This mixture represents a basic mixture containing the conventional raw materials which are
commonly used in conventional concrete construction. From the practical standpoint,
production of this mixture in large scale is more convenient compared to the other three
mixtures as the ingredients are readily available.
SFPM
This mixture represents silica fume incorporated mixtures which are commonly used in the
projects were durability and long-term performance of concrete is important due to
environmental conditions.
FRPM
FRPM mixture represents an important family of cementitious mixtures which are fiber
reinforced mixtures. The justification for design of FRPM is that fiber inclusion could
address the shrinkage induced cracking in the printed elements.
NCPM
NCPM mixture is designed to investigate the performance of a natural viscosity modifying
agent. In previous studies on self-consolidating concrete, Nano-clay inclusion is shown to
enhance thixotropic behavior of cementitious mixtures. Accordingly, the hypothesis is that
its inclusion can enhance shape stability of a printing mixture as well.
60
Compressive strength of the four printing mixtures in hardened state is also presented in Table 18. The
mechanical strength test is carried out according to ASTM C109 [70], using 2in cubes at a loading rate
of 1200 N/s. Based on American Concrete Institute definition (ACI 363 committee), all these mixtures
are considered “high-strength concrete”, which is desirable due to concerns regarding structural
behavior of printed structures. A minimum compressive strength of 41 MPa is specified for high-
strength concrete by ACI.
Table 18: Compressive strength of printing mixture at 7-day and 28-day ages
Mixture
ID
7-day Compressive
Strength (std. dev.)
28-day Compressive
Strength (std. dev.)
MPa MPa
PPM 32.9 (0.7) 44.7 (1.3)
SFPM 35.2 (1.6) 48.5 (1.6)
FRPM 31.0 (1.9) 45.1 (1.1)
NCPM 31.8 (1.2) 45.9 (1.5)
4.3. Laboratory Testing of Printing Mixtures
In this step, four main performance requirements are defined for fresh printing mixtures and a series of
experiments are carried out using the developed printing mixtures to evaluate these properties.
4.3.1. Print Quality
“Print quality” refers to the characteristics of printed layers using a specific printing mixture. These
required properties are mainly related to the dimensions and also surface quality of the deposited
layers. Several criteria are defined to assess the print quality of a mixture, and acceptance or rejecting
that mixture based on that. A printing mixture is considered acceptable when the following conditions
are met:
61
C1: The printed layers should be free of surface defects, including any tearing due to excessive
stiffness and inadequate cohesion (Figure 40)
C2: The layer edges should be visible and squared (versus round edges)
C3: Dimension conformity and dimension consistency
Figure 40 presents two cases where the printing mixture is rejected by C1, since tearing is observed in
the layers due to inappropriate fresh properties of the printing mixture.
Figure 40: Tearing in the printed layers due to excessive stiffness of the mixture
With respect to the third acceptance criteria, “dimension conformity” guarantees that the dimensions of
the printed layers are within an acceptable range of the designed dimensions (Figure 41-a), while
“dimension consistency” refers to changes in width of a printed layer and acceptable variations (Figure
41-b). In this study, a width of 38.1mm (1.5inch) is targeted for each layer and after running a large
number of experiments, 10% error in the target width was selected as a reasonable bound for accepting
or rejecting printed layers. In other words, the width of printed layers using all the accepted printing
mixtures is in the range 38.1±3.81mm. Note that five measurements are done along each printed layer
to check the dimension conformity, and for each printing mixture the experiment is carried out four
times (four replicates of each printing mixture). It should also be noted that variations in extrusion rate
62
is another parameter which could be measured and considered as an alternative criterion for
consistency evaluation.
The four printing mixtures presented in Table 16 are obtained after a large number of experiments and
modifications (trial and error approach while considering conventional concrete characteristics), while
they all pass the print quality requirements. The reason for adopting a trial and error approach is that
currently there is no guideline or procedure for design and testing of printing mixtures. Finding such
approach is one of objectives of the current study.
(a) (b)
Figure 41: (a) Variations in width of printed layer using different mixtures at same printing speed
(dimension conformity) (b) Variations in width of a single layer (dimension consistency)
63
4.3.2. Shape Stability of Fresh Printing Mixture: Nozzle Design Considerations
One of the critical properties of printing material in fresh state is the ability to keep its geometry during
the printing process, which is referred to as shape stability herein. Shape stability is also important in
some other construction methods such as slip-forming [11]. Shape stability of fresh cementitious
mixtures is commonly attributed to the interparticle friction and cohesion of the mixture [71]. In
construction-scale 3D printing, deformations of a deposited layer could be attributed to three sources:
(1) Layer self-weight, (2) Weight of the following layers, and (3) The extrusion pressure. The
deposited layer should be able to withstand the load which is applied to it during the printing process.
In this section, some considerations with respect to effect of nozzle design on the layer deformations
are provided.
(A)
Figure 42: Nozzle design used in this study
64
To perform an analysis of flow and pressure variations for the used nozzle design, SOLIDWORKS
Flow Simulation package was used [72]. SOLIDWORKS Flow Simulation is a general parametric flow
simulation tool that uses the Finite Volume Method (FVM). To simulate the flow of printing mixture in
the nozzle, a non-Newtonian liquid was defined and the material for the solid 3D model (nozzle) was
defined as ABS plastic with surface roughness of 0.2mm.
To define the boundary conditions for the simulation problem, an extrusion pressure of 49 kPa was
defined in the inlet area of the nozzle. The details for experimental measurement of the extrusion
pressure will be presented in Chapter 5. In addition, a volume output flow of 58 cm
3
/sec flow volume
was defined for the nozzle outlet. This output flow is calculated based on the layer dimensions and the
printing speed which was used in all extrusion tests.
Figure 43 presents the material flow trajectory. It should be mentioned that 10% of the output flow
volume was assigned to the bottom opening of the nozzle outlet. The reason is that it is almost
impossible to have zero gap between the nozzle and the previous layer and there will be some material
flow under the nozzle.
65
Figure 43: Flow trajectory for the used nozzle
Figure 44 shows the pressure cut plots for the simulated nozzle flow. The reduction in pressure toward
the outlet of the nozzles can be observed in the pressure cut plots. Figure 44 (b) shows the pressure
distribution on a horizontal bottom section of the nozzle. Considering the bottom area of the nozzle
outlet, it is anticipated that an extrusion force of 47 N is applied to the previous layer. This extrusion
force could potentially improve the bonding strength of layers that are printed successively. The
experimental measurement of the bonding strength is out of the scope of the current research and needs
to be addressed in the future studies. On the other hand, it should be noted that the extrusion pressure
could increase the risk of deformation of bottom layers.
66
(a) (b)
Figure 44: Pressure cut plot (a) vertical section (b) horizontal bottom section of the nozzle
4.3.3. Shape Stability of Fresh Printing Mixture: Theoretical Analysis
Previous studies show that longer interlayer time gap in the printing process results in inferior
interlayer adhesion, which is not desirable in terms of structural properties of printed structure. This
indicates that after deposition of a layer, next layer should be printed as soon as the previous layer
acquires sufficient shape stability. The minimum interlayer time gap, however, is determined by the
shape stability of printed layers. Accordingly, shape stability of printing mixture could affect the total
construction time in construction-scale 3D printing as well, and increased shape stability could increase
the construction speed. Ideally, no delay time between successive layers is desired.
Considering the importance of shape stability of printing mixture, and to provide a better
understanding of this property of fresh mixture, a model in its generalized form is presented in this
section. The derivation steps are first shown for a specimen with three layers and later expanded to the
more general case of a printed wall with N layers. The cross section of the specimen with three layers,
as well as the parameters used in the model, are illustrated in Figure 45.
67
Figure 45: Cross section of printed layers, extrusion pressure, and defined parameters
The deformation of the first (top) layer could be calculated by:
∆
(𝑥 ) = (𝐹 (𝑡 ) + 𝜌 𝐴 𝑥 )
𝐴 𝐸 𝑑 𝑥 =
1
𝐴 𝐸 (𝐹 (𝑡 )𝑥 + 𝜌 𝐴 𝑥 2
) + 𝐶 (Equation 4)
where:
∆
(𝑥 ): 𝐴𝑥𝑖𝑎𝑙 𝑑𝑒𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛 𝑜𝑓 𝑙𝑎𝑦𝑒𝑟 1 𝑎𝑡 𝑝𝑜𝑖𝑛𝑡 𝑥
𝐹 (𝑡 ): 𝐸𝑥𝑡𝑟𝑢𝑠𝑖𝑜𝑛 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒
𝜌 : 𝐷𝑒𝑛𝑠𝑖𝑡𝑦 𝑜𝑓 𝑝𝑟𝑖𝑛𝑡𝑖𝑛𝑔 𝑚𝑎𝑡𝑒𝑟𝑖𝑎𝑙
𝐴 : 𝐶𝑟𝑜𝑠𝑠 𝑠𝑒𝑐𝑡𝑖𝑜𝑛 𝑎𝑟𝑒𝑎 𝑜𝑓 𝑙𝑎𝑦𝑒𝑟 1
𝑥 : 𝐻𝑒𝑖𝑔 ℎ𝑡 𝑜𝑓 𝑡 ℎ𝑒 𝑝𝑜𝑖𝑛𝑡 𝑤 ℎ𝑒𝑟𝑒 𝑑𝑒𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛 𝑖𝑠 𝑐𝑎𝑙𝑐𝑢𝑙𝑎𝑡𝑒𝑑 (𝑠𝑒𝑒 Figure 45)
𝐸 : 𝑀𝑜𝑑𝑢𝑙𝑢𝑠 𝑜𝑓 𝑒𝑙𝑎𝑠𝑡𝑖𝑐𝑖𝑡𝑦 𝑜𝑓 𝑡 ℎ𝑒 𝑙𝑎𝑦𝑒𝑟 (𝐸 𝑐𝑜𝑢𝑙𝑑 𝑏𝑒 𝑡𝑖𝑚𝑒 𝑑𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡 (𝐸 (𝑡 ))
𝐶 : 𝐼𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡
Similarly, the deformations in the second and third layer could be obtained by following equations:
68
∆
(𝑥 ) = (𝐹 (𝑡 ) + 𝑊 + 𝜌 𝐴 𝑥 )
𝐴 𝐸 𝑑 𝑥 =
1
𝐴 𝐸 ((𝐹 (𝑡 )+𝑊 ) 𝑥 + 𝜌 𝐴 𝑥 2
) + 𝐶 (Equation 5)
∆
(𝑥 ) = (𝐹 (𝑡 ) + 𝑊 + 𝑊 + 𝜌 𝐴 𝑥 )
𝐴 𝐸 𝑑 𝑥 =
1
𝐴 𝐸 ((𝐹 (𝑡 )+𝑊 + 𝑊 ) 𝑥 + 𝜌 𝐴 𝑥 2
) + 𝐶 (Equation 6)
where:
Wj: Total weight of layer j.
In order to calculate integration constants, boundary conditions are considered. We know that the
deformation of the third layer at x 3=L3 must be zero, since it is on the printing bed. Therefore:
∆
(𝑥 = 𝐿 ) = 0 → 𝐶 = −
1
𝐴 𝐸 (𝐹 (𝑡 ) + 𝑊 + 𝑊 )𝐿 + 𝑊 𝐿 2
(Equation 7)
Accordingly, C2 and C1 could be calculated using boundary conditions:
∆
(𝑥 = 𝐿 ) = ∆
(𝑥 = 0) → 𝐶 = −
𝐿 𝐴 𝐸 (𝐹 (𝑡 ) + 𝑊 + 𝑊 ) + 𝑊 1
2
−
𝐿 𝐴 𝐸 𝐹 (𝑡 ) + 𝑊 +
𝑊 2
(Equation 8)
∆
(𝑥 = 𝐿 ) = ∆
(𝑥 = 0) → 𝐶 = −
𝐿 𝐴 𝐸 (𝐹 (𝑡 ) + 𝑊 + 𝑊 ) + 𝑊 1
2
−
𝐿 𝐴 𝐸 𝐹 (𝑡 ) + 𝑊 +
𝑊 2
−
𝐿 𝐴 𝐸 𝐹 (𝑡 ) +
𝑊 2
(Equation 9)
From the obtained results, the deformations at layer j of a sample with N layers could be calculated by
the following equation:
∆
𝑥 =
1
𝐴 𝐸 𝐹 + 𝑊 𝑥 +
𝜌 𝐴 𝑥 2
+ 𝐶 (Equation 10)
where:
𝐶 = −𝐿 𝐴 𝐸 𝐹 (𝑡 ) +
𝑊 2
+ 𝑊 (Equation 11)
69
The initially calculated deformations indicate that the level of deformation that has happened in one or
multiple layers would result in a significant change in the values assumed for their modules of
elasticity. In this case, it is possible to replace the initial modulus of elasticity with the updated values
and use Equation 10 in a recursive procedure to calculate the new layer deformations and update the
modules of elasticity accordingly.
4.3.4. Shape Stability of Fresh Printing Mixture: Experimental Measurement
For experimental study, and to have a realistic perception of interlayer time gap during Contour
Crafting process, a one-story building is considered as a case study. The original house plan is
designed by FreeGreen architectural and design company [73]. This is a 1160 ft
2
house with two
bedrooms and one bathroom. The 3D model and plan of the house is presented in Figure 46.
Considering the construction process of such a house using a Contour Crafting gantry robot, the nozzle
traveling distance for each layer will be approximately 67 meters. Using the linear printing speed of 6
cm/sec (which was used in this study), this is equivalent to a 19-minute time gap between two
successive layers. It is to be noted that the existence of openings such as doors and windows do not
affect the printing time since the printing nozzle still has to travel the corresponding distance. The
obtained time gap (19 minutes) is used for laboratory investigation of shape stability of four printing
mixtures.
70
(a) 3D renderings of the house
(b) House plan
Figure 46: 3D model and plan of the house considered as a case study [73]
“Layer settlement test” is proposed as a test method to provide a quantitative measure of shape
stability of a printing mixture. In each test, two layers are printed on top of each other with two
different time gaps: 19 minutes (realistic construction scenario, based on the one-story building) and 0
(extreme scenario). The likely negative effect of longer time gap on the interlayer bond strength should
also be considered as another parameter, which is not within the scope of current work. The extreme
scenario is also considered in order to augment the differences in behavior of different mixtures in
fresh state and also to investigate the worst-case scenario. A camera is placed in front of printed layers,
and photos are taken both before and after the second layer is printed. A ruler is also placed next to the
layers as a scale. Then, ImageJ software is used to analyze the photos taken from concrete layers and to
71
measure layer settlements. ImageJ is a public domain, Java-based image processing program developed
at the National Institutes of Health [74]. It provides extensibility via Java plugins and recordable
macros. Figure 47 presents the result of printing a double layer specimen using fiber reinforced
printing mixture, with two different time gaps.
The results for the two scenarios are presented in Table 19 (no time gap) and Table 20 (19-minute time
gap). The result of each test (Test 1, 2, or 3) is the average of five readings for each printed specimen,
and the average of three experiments (three printed layers) is used as the final test result for a printing
mixture.
Figure 47: Printing a double layer specimen with zero time gap resulting in deformation of first layer
(Top) Same experiment using a 19-minute time gap (Bottom)
Table 19: Layer settlement results (ImageJ)- no time gap
Mixture
ID
Test 1 Test 2 Test 3
Average Reading
(Std. Dev.)
mm mm mm mm
PPM Collapse Collapse Collapse -
SFPM 2.2 1.8 1.5 1.8 (0.3)
FRPM 2.8 3.3 2.5 2.9 (0.3)
NCPM 2.0 1.1 1.6 1.6 (0.4)
72
Table 20: Layer settlement results (ImageJ)- 19min time gap
Mixture
ID
Test 1 Test 2 Test 3
Average Reading
(Std. Dev.)
mm mm mm mm
PPM 1.9 1.1 1.6 1.5 (0.3)
SFPM 0 0 0 0
FRPM 0 0 0 0
NCPM 0 0 0 0
Considering the obtained results, it is observed that all the printing mixtures except PPM possess high
shape stability (no visible deformation), when the second layer is printed with a 19-minute delay. On
the other hand, deformations happened in all cases when the second layer is printed right after the first
layer deposition. For PPM mixture with no time gap, considerable deformations (change in width and
height of the first layer) happens, and change in height is not an accurate indicator of deformations.
Therefore, the test result is simply reported as “collapse” instead of change in height value. The lowest
deformations are measured for SFPM and NCPM mixtures. Considering the standard deviation of the
obtained results, there is no significant difference between shape stability of these two mixtures.
Another finding is that acceptable print quality (as in the case of four printing mixtures) does not
guarantee high shape stability, as the four printing mixtures with acceptable printability show different
levels of shape stability herein. As such, it is suggested this property should be evaluated
independently for each printing mixture.
In order to make it possible to evaluate and compare the shape stability of different printing mixtures
without printing specimens, cylinder stability test is designed and carried out using different printing
mixtures. The test apparatus includes a frame, two semi-cylinder shells, a tamping rod, two guides and
a container for uniform application of load to fresh concrete specimen. These parts are designed using
73
SOLIDWORKS [75] CAD software and 3D printed using ABS plastic. The joints between two semi-
cylinders are sealed (using rubber) to prevent water leakage during compaction.
Figure 48: 3D model and 3D printed components of cylinder stability test
The following test procedure is carried out to measure the shape stability of a printing mixture using
cylinder stability test: The semi-cylinders are fixed in place and locked, and a concrete layer of 40mm
is placed. Using the tamping rod shown in Figure 46 (right), the layer is consolidated by rodding 15
times evenly distributed around the layer. The same procedure is repeated for second layer and
excessive concrete is removed from the top. Then the two semi-cylinders are unlocked and detached
from concrete cylinder. Any possible change in height as a result of self-weight is measured. Then a
load of 5.5 kg (equivalent to a 4.77 kPa stress) was applied and the resulting deformation in the fresh
concrete cylinder is measured (Figure 49). The weight is selected such that it causes visible
deformations in the concrete cylinder.
74
Figure 49: Cylinder stability test
The cylinder stability test is carried out after the 8min mixing procedure is finished for each printing
mixture. The obtained results are presented in Table 21.
Table 21: Cylinder stability test results for printing mixtures
Mixture ID
Test 1 Test 2 Test 3
Average
Reading
Standard
Deviation
mm mm mm mm mm
PPM 41 37 38 38.7 1.7
SFPM 15 15 14 14.7 0.5
FRPM 34 29 31 31.3 2.1
NCPM 12 15 11 12.7 1.7
The largest deformation in cylinder is observed for PPM mixture, followed by FRPM. NCPM mixture
exhibits the highest shape stability, while SFPM mixture results in a slightly larger deformation. The
high shape stability of NCPM, proved by the results of both cylinder stability and layer settlement
tests, can be explained by the results and observations reported in studies on the influence of Nano-clay
addition on the rheological properties of cementitious mixtures. Kawashima et al. [66] reported higher
build-up rate of internal structure after shear-induced breakdown for Nano-clay incorporated
cementitious mixtures, especially at early ages. The high specific surface area of Nano-clay in
combination with their fine size, which leads to higher surface forces between particles, is reported
75
influential to this observation [66]. In a 2018 study, Qian and De Schutter [76] have investigated this
thixotropy enhancing behavior of Nano-clay in the presence of polycarboxylate ether superplasticizer
(PCE) water reducing agent. A thixotropic index which shows the relationship between static and
dynamic yield stress has been defined and used in this study. Figure 50 shows the torque development
over time under 600 rpm. It is observed that increasing the amount of Nano-clay addition increases
both the initial and equilibrium shear stresses. Furthermore, the gap between initial and equilibrium
shear stresses is significantly improved, which implies the higher thixotropic behavior in the presence
of Nano-clay. In summary, these researchers reported that combination of Nano-clay and PCE based
superplasticizer enhances the thixotropic behavior of mixtures and increases the static yield stress, and
could be used to develop mixtures with low dynamic yield stress [76].
Figure 50:Torque development of samples with admixture and Nano-clay at 600 rpm [76]
Considering the process of construction-scale 3D printing, where the cementitious mixture undergoes
considerable shear stress and agitation before being deposited as a layer (mixing, pumping, and
76
extrusion), as well as the aforementioned results of other studies, it is concluded that changes in
internal structure and the thixotropic behavior play an important role in the shape stability of a printing
mixture. The faster so-called ‘‘structuration at rest” and high static yield stress of Nano-clay included
mixtures, even in the presence of PCE superplasticizer which was used in the present work, could be
considered as a major reason for enhanced shape stability of NCPM mixture which was experimentally
proven in this study.
Considering the similar performance of SFPM and NCPM, the experimental results indicate that
cylinder stability test ranks the shape stability of mixtures similar to layer settlement test. This implies
that, upon further testing, this simple test could potentially be suggested for evaluation and comparison
of shape stability of different printing mixtures. Further research and experimental data is required to
investigate any correlation between the values given by the two test methods.
4.3.5. Robustness
The fourth characteristic of a printing mixture which is proposed and considered in this study is the
mixture robustness. It should be noted that there are various chemical admixtures and materials
(variety of supplementary cementitious materials, for example) to be used for developing printable
mixtures. It is recommended that several mixtures with acceptable print quality and shape stability are
developed and the mixture with highest robustness against variations is selected. According to RILEM
TC 288MPS [77], the concrete robustness (as a general concept) is the characteristic of a mixture
representing its tolerance to variations in constituent characteristics and quantities, variations during
concrete mixing, transport, and placement, as well as environmental conditions [78]. Considering the
high importance of fresh properties of printing mixture, robustness of printing mixture can be defined
as the capacity of mixture to retain its printability when small variations in the properties or quantities
77
of the constituent materials occur. This is very important from practical standpoint, since the
cementitious mixture extrusion process needs to continue for one or few days uninterrupted and as
planned, in order to realize the promise of fast building construction made by construction-scale 3D
printing. Higher robustness of the printing mixture could minimize the problems caused by usual
variations in the material. Since there is no earlier research work on robustness of printing mixture,
EFNARC (European Federation of National Associations Representing for Concrete)
recommendations for SCC could serve as a good starting point: “A well designed and robust SCC can
typically accept a 5 to 10 liter/m
3
change in water content without falling outside the specified classes
of performance when fresh” [29]. Accordingly, in this study ±10liter/m
3
change in water content was
selected as the level of variation used for assessment.
Based on shape stability and print quality test results, NCPM and SFPM are selected for robustness
evaluation. As such, four layers are printed using each mixture (4 replicates of each mixture is
produced) and each layer is measured at 5 different points using a digital caliper (similar to print
quality test) in order to measure the average width of the layer. Change in the width of printed layers is
used as a measure of mixture sensitivity to variations. Table 22 presents the average width of the two
reference mixtures as well as deliberately altered mixtures.
Table 22. Average width of reference and altered mixtures
Mix ID
Average Layer Width
(mm)
SFPM 39.4 [0.9]
a
SFPM+10 47.8 [2.7]
SFPM-10 Not printable (tearing)
NCPM 40.1 [1.2]
NCPM+10 44.3 [1.8]
NCPM-10 Not printable (tearing)
a
Values in brackets are standard deviations of the test results (mm)
As a reminder, based on print quality experiments, the width of an acceptable printing mixture must be
in the range of 38.1±3.81mm. Therefore, the results show that none of the two mixtures (NCPM and
78
SFPM) were able to retain acceptable print quality when the water content changed by ±10litre/m
3
.
Both mixtures show similar behavior when the amount of free water is reduced by 10litre/m
3
, that is
the resulting mixture is not printable. On the other hand, when the free water is increased by this
amount the layer width increases. The average increase for NCPM is 10.7%, while for SFPM it is
21.3%. This implies relatively higher robustness of NCPM mixture. In other words, when considering
print quality, the NCPM mixture shows higher resistance to variations in water content compared to
SFPM.
4.3.6. Printability Timespan
The fourth characteristic of a printing mixture which is proposed and considered in this study is
printability timespan. It is defined as the acceptable time period during which a mixture could be
successfully printed, considering the workability loss which happens over time. This is highly
important in terms of construction management, and timing of material delivery to the extruder and
operation of a building printer such as Contour Crafting machine. In this study, two time limits are
introduced and used to define printability timespan of a printing mixture:
- The time when the quality of printed layer is affected as a result of workability loss, mainly
recognized by tearing due to excessive stiffness (printability limit)
- The time when the material inside the extruder cannot be forced out by the extrusion system
(blocking limit)
An important concept related to the printability timespan is the initial setting time of concrete. Per
ASTM C125 [79], time of initial setting is defined as “the elapsed time after initial contact of cement
and water, required for the mortar sieved from the concrete to reach a penetration resistance of 500psi
(3.5MPa)”. In this study, a penetrometer was used to measure the initial setting time of printing
79
mixtures according to ASTM C403 [80]. Penetration plunger has a 1/20in
2
tip area, and it is steadily
pushed into the mortar to a 1in depth, at periodic time intervals. The point of initial set is reached when
the penetration value is 500psi. Initial testing reveals that the setting time of the four developed
printing mixtures is similar (in the range of 325 to 340 minutes). This can be justified by the same
Portland cement used in all mixtures and use of no accelerator or retarder. In order to develop mixtures
with different behavior in terms of setting time and workability loss, three different dosages of CaCl 2
(1%, 2%, and 3% of Portland cement mass) were added to PPM mixture and the resulting mixtures
were labeled as PPM1%CaCl, PPM2%CaCl, and PPM3%CaCl, respectively. It is to be noted that
calcium chloride is commonly known as the most effective accelerator for cementitious mixtures. It
acts like a catalyst for C3S reactions, which also results in larger hydration heat liberation [34].
The measurements using mortar penetrometer for each mixture are presented in Figure 51, and the
green dashed line defines the limit corresponding to the initial setting time. For each mixture, the
measurements are carried out three times and the variations in the readings are shown on the following
graph using error bars.
Figure 51: Mortar penetrometer readings for the three mixtures
80
Based on obtained measurements, the initial setting time of PPM, PPM1%CaCl, PPM2%CaCl, and
PPM3%CaCl is 335, 237, 181, and 163 minutes, respectively. As anticipated, addition of calcium
chloride resulted in accelerated hydration reaction and shorter setting times. The results for printability
timespan parameters as well as initial setting time are presented in Table 23. In order to determine the
printability limit, a single layer is printed every 5 minutes (starting from 20 minutes after initial water-
cement contact) and the earliest time when printability criteria (C1-C3) are not satisfied, is recorded as
the printability limit. Similarly, the earliest time when the mixture inside the extruder could not be
guided out of nozzle, is recorded as the blocking limit.
The results indicate that nozzle blockage happens long before the initial setting time of each mixture.
Nozzle blockage can result in significant time loss and extra cost during construction. As such,
measuring blocking limit for each mixture is recommended during mixture design and laboratory
testing.
Table 23: Initial setting time, printability limit and blocking limit of mixtures (minutes)
Mixture ID Initial setting time Printability limit Blocking limit
PPM 335 55 85
PPM1%CaCl 237 40 75
PPM2%CaCl 181 40 60
PPM3%CaCl 163 45 55
Finally, workability loss of the four mixtures are studied. Workability of mixtures is determined using
a flow table (ASTM C1437-15 [81]), where initially a mold is filled with mortar and compacted. Then
the mold is lifted away from mortar and the table is immediately dropped 25 times in 15 second. The
resulting diameter of mortar pat (circle) is used as an indicator of flowability. Flow diameters for
different mixtures at different times from 8 minutes to 90 minutes (after initial cement-water contact)
are measured and presented in Table 24.
81
Table 24: Workability loss for the three printing mixtures (flow diameters in mm)
The results indicate that the influence of calcium chloride addition appears only after 60 minutes, and
becomes more distinct at longer times. This finding justifies the similar printability limits which were
measured for the three CaCl2 incorporated mixtures (Table 23). In addition, results imply that higher
calcium chloride dosages lead to faster workability loss over time. However, increasing the accelerator
dosage from 2% to 3% is shown to have insignificant influence on the workability loss.
Setting time and workability loss measurements could be employed when studying effects of different
chemical admixtures on fresh state behavior of printing mixture. However, it seems that neither setting
time measurements nor workability loss measurements could replace the direct measurement of
printability limit and blockage limit. Furthermore, these two limits are directly dependent on the
specific extrusion mechanism used by the robot, therefore finding a robot-independent relationship
seems unlikely.
8 20 40 60 70 80 90
PPM 219 211 200 179 169 161 155
PPM1%CaCl 221 210 202 163 152 148 142
PPM2%CaCl 216 213 197 154 146 137 131
PPM3%CaCl 218 215 195 147 145 134 128
Mixture ID
Time (minutes)
82
5. REAL-TIME QUALITY MONITORING OF CONTOUR
CRAFTING
In conventional concrete construction, large volumes of concrete are often poured into the prepared
formwork in a relatively short time. In this approach small variations in mixture proportions do not
necessarily cause any problem as long as structural and durability requirements are later satisfied by
the hardened concrete. On the other hand, in Contour Crafting layers of 1-2inch thickness are being
extruded continuously for 24-48 hours, and even small variations in the printing mixture could possibly
lead to the process failure when the problematic section of the mix is extruded out of the nozzle. For
instance, temporary changes in the rheology of printing mixture (caused by variations in the mixture
proportions) could lead to extrusion of a layer with insufficient shape stability. Later in the process,
after subsequent layers are printed, this could result in significant deformations of the faulty layer
which, in turn, could lead to collapse of freshly printed structure or element above the problematic
point. It should be noted that that two competing processes during printing determine structural failure
of a printed structure: the increasing strength and stiffness caused by thixotropic build-up of the
extruded mixture versus the gradually increasing load as more layers are deposited on each other.
Furthermore, conventional construction quality control methods are manual (often done by 1 or 2
operators), need sample preparation, and are time-consuming. All these characteristics are in conflict
with the main objectives of robotic building construction. As such, it is concluded that automated
construction processes need real-time, in-process, responsive, and reliable quality monitoring
techniques specially developed and customized for automated construction systems. Any issue needs to
be detected instantaneously so that the required modifications are made in a timely fashion. Such
quality monitoring procedures have not been developed yet and the topic has not been investigated by
researchers and experts.
83
In this chapter, several novel techniques are investigated as real-time quality monitoring measures for
the Contour Crafting process: electrical resistivity measurements, extrusion pressure measurements,
extruder power consumption monitoring, and a technique based on computer vision. It should be
mentioned that these techniques are investigated under experimental conditions and using the
laboratory-scale printer which is developed as part of this project (described in Chapter 5). Considering
the fact that this is the first research work dedicated to real time quality monitoring of Contour
Crafting, it was decided that assessment conditions be simplified and some parameters that could
increase measurement noise for each technique be eliminated to the best extent possible. In this way,
the true potential of each technique could be evaluated in the first step and the results could be used by
other researchers for further investigation, improvement, and final implementation in full-scale robotic
systems.
In the following sections the sensory system development, calibration, or test configuration is initially
discussed for each technique. Then, in order to enable a reliable evaluation and comparison of the
proposed techniques, a series of experiments are designed and carried out. In these experiments,
changes in the water content (often caused by variations in aggregate moisture content or measurement
imprecisions) are considered to be the major sources of variations in the printing mixture. Based on the
experimental results from the first stage of the work, a mixture (NCPM with minor change in
superplasticizer dosage) is selected as reference and 6 levels of variation in the free water content is
applied to this mixture, resulting in a total of seven mixtures (Table 25). These seven mixtures are used
in order to evaluate the proposed quality monitoring techniques. It should be mentioned that
±10liter/m
3
represents the variation level in water content that commonly happens and is expected
during construction and it is suggested by guidelines such as EFNARC [29] for robustness evaluation.
On the other hand, ±5liter/m
3
and ±15liter/m
3
are designed to evaluate each quality control technique
84
in lower and higher levels of variation in printing mixture water content, respectively, such that a clear
understanding of performance of proposed techniques at different variation levels could be obtained. In
other words, sensitivity level of each technique could be determined and compared.
Table 25: Mixtures used for testing of different quality monitoring techniques
Mixture ID
Proportions (kg/m
3
)
Fine aggregate
(SSD)
Portland
cement
Free
Water
Nano clay
(%)*
Superplasticizer
(%)*
QCREF
1379 600
259
0.3 0.11
QCREF+5 264
QCREF+10 269
QCREF+15 274
QCREF-5 254
QCREF-10 249
QCREF-15 244
* The percentages are reported by cementitious materials mass
Finally, the results of the experimental program are used to discuss and compare the responsiveness
(sensitivity to variations) and reliability of proposed quality monitoring techniques. Based on the
results, some recommendations are made regarding real-time quality monitoring of Contour Crafting
process.
5.1. Using Electrical Resistivity of Mixture
The electrical conductivity of a cementitious mixture, either fresh or hardened, is a function of various
parameters, including the volume fraction of the cement paste, pore-size distribution, ionic
composition of the liquid phase, presence of chemical admixtures, and pore network [82]. Following
the initial contact between cement and water, conductivity increases due to rapid dissolution of alkali
sulfates and other cement oxides, which results in increasing the ionic concentration of the pore
85
solution. Conductivity increases with the hydration of cement process to reach a peak value beyond
which conductivity starts to decrease due to further precipitation of the hydration products and
reduction of mobility of ions through the pore solution. The variation of conductivity as a function of
time can indeed reflect internal changes of the pore solution with time.
There are several methods to measure the electrical resistivity of concrete, namely, two-probe
technique and four-probe (Wenner) technique. In two-probe technique the concrete sample is placed
between two electrodes (usually two parallel metal plates) with moist sponge contacts at the interfaces
to ensure a proper electrical connection. An AC current is applied, and the drop in the potential
between the electrodes is measured. The schematics of the test configuration for the two-probe
technique is shown in Figure 52.
Figure 52: Two-probe technique for electrical resistivity measurements [83]
The four-probe method is the most widely used technique for field measurement of concrete resistivity,
primarily due to its relatively higher reliability [84]. This method originally was developed by Frank
Wenner [85] to measure earth resistivity. Figure 53 shows the schematic of the process, where four
86
electrodes are equally spaced, and a small alternating current is applied between the outer electrodes
while potential is measured between the inner electrodes. The resistivity is then calculated using the
following equation [86]:
𝜌 =
2 × 𝜋 × 𝑎 × 𝑉 𝐼
(Equation 12)
where ρ is resistivity (ohm.m), a is distance between inner electrodes (m), V is voltage (volts) and I is
current (amps). This equation is sometimes presented as 𝜌 = 𝑘 𝑥 𝑅 , where k is referred to as the
geometrical factor. Geometrical factor depends on the size and shape of the sample as well as the
distance between the probes on the testing device [83]. One immediate advantage of this technique
over the two-probe arrangement is that the four-probe method measures the resistivity of the concrete
area between the inner electrodes. This is a larger area than that of the probe tip itself. The influence of
aggregate therefore can be avoided when the spacing between the inner electrodes exceeds the size of
the largest aggregate particle. This yields a more reliable resistivity measurement. Surface contacts are
less critical for the four-probe method than for the two-probe technique. However, for hardened
concrete, adequate electrical contact should be established between electrodes and the concrete surface,
such as by use of a conductive gel. Additionally, Wenner technique measurements are based on the
assumption that the material to be tested is homogenous. Therefore, material inhomogeneities affect
electrical resistivity measurements. A high-resistivity aggregate surrounded by low-resistivity cement
paste is the source of such inhomogeneity. The effect of aggregate can be reduced by increasing the
spacing between the inner electrodes.
87
Figure 53: Wenner technique for measuring resistivity [84]
Electrical resistivity of hardened concrete has been widely used to assess the long-term performance of
hardened concrete [86, 82, 87, 88]. For instance, the relationship between chloride penetration
resistance and electrical resistivity is well established, and electrical resistivity measurements are used
to assess chloride ion penetrability of hardened concrete by several standards and guidelines.
Table 26 presents the chloride ion penetrability levels of hardened concrete based on resistivity
measurements based on AASHTO TP 95 [83].
Table 26: Comparison of chloride penetrability levels established for standards based on electrical
resistivity (AASHTO TP 95) and charged passed (ASTM C1202) (reproduced from [83])
AASHTO TP 95 (kΩ.cm) ASTM C1202 (Coulombs) Chloride ion penetrability
<12 >4000 High
12-21 2000-4000 Moderate
21-37 1000-2000 Low
37-254 100-1000 Very Low
>254 <100 Negligible
88
While the electrical resistivity of hardened concrete has been extensively studied and widely adopted
by guidelines and standards, only few studies have investigated electrical resistivity of fresh mortar and
concrete. Khayyat et al. [89] used variations in electrical conductivity to evaluate the stability and
homogeneity of fresh mortar mixtures. Sixteen mixtures with different water-binder ratios (0.45-0.80)
and different replacement levels of Class F fly ash were evaluated (Table 27).
Table 27: Composition and properties of mortar mixtures [89]
In this study, a bleeding index (BI-1) is defined, which is the cumulative area bound between
conductivity values of the top and second pairs of electrodes monitored between T initial and the peak
time, divided by the time required to reach the peak conductivity. The first peak in the conductivity
curve obtained for the top electrode pair is considered by these researchers to correlate to the maximum
bleeding. The other bleeding index used in this study is BI-2, which is defined as:
BI-2 =
−
(Equation 13)
89
An increase in BI-1 and BI-2 reflects an increase in bleeding. Figure 54 presents the relationship
between BI-2 and external bleeding for the tested mixtures. The BI-2 is shown to increase with
increasing external bleeding with a correlation factor of 0.72 [89].
Figure 54: Relationship between BI-2 and external bleeding [89]
Based on the results, it seems that BI-1 correlates well with the measured external bleeding and it
seems to best reflect the degree of water migration to the surface [89]. In conclusion, Khayyat et al.
[89] mention that the conductivity values provide a quantitative basis to define various indexes that
reflect water migration, segregation, and overall homogeneity of mortars.
In the present study electrical resistivity of fresh mixture is proposed and used as a quality monitoring
measure for printing material. The hypothesis is that monitoring the changes in the electrical resistivity
reveals the unacceptable changes in the water content of the printing material. As mentioned before,
the instantaneous detection of material variations could prevent extrusion of unacceptable layers before
the overall quality of printed structure is affected.
90
In this study, to simplify the testing conditions and to eliminate the noise by factors such as extrusion
flow, the electrical measurements were carried out on 10x20cm cylinder specimens of fresh printing
mixture. A signal generator [90] was used to produce different waveforms with desired frequencies and
amplitudes. The maximum output amplitude of used signal generator is ± 10V pp, and the range of
output frequency is 1Hz-500kHz. An ATMega2560 based micro-controller was used to build the
measurement device. DHT22 digital temperature and humidity sensor, with temperature measurement
precision of ±0.5° and humidity measurement precision of ±2%, was used to measure ambient
temperature and humidity during experiments. Furthermore, waterproof DS18B20 digital temperature
sensor was used to measure internal mixture temperature over time. Different components and main
structure of the four-probe electrical resistivity device are presented in Figure 55. In order to increase
the accuracy of voltage readings, high precision voltage reference LM4040 was used as reference
voltage for the micro-controller. Moreover, a data logging system was developed such that all
measured values were directly saved to an SD card for future analysis.
91
Figure 55: Schematics and wiring of the developed electrical resisitivity setup
3 mmx30 mm stainless steel pins were used as electrical resistivity probes, and probe spacing was
selected as 20mm. To ensure the accurate alignment and spacing of probes, a 3D printed holder was
used as shown in Figure 56. The probes were placed such that a 3mm length of each probe would be
inside fresh mixture.
Figure 56: Wenner probes installed on a 10x20cm cylinder
92
5.1.1. Optimization of Testing Parameters and System Calibration
There is limited information available regarding electrical resistivity of fresh concrete and mortar.
Even for hardened concrete, there is no consensus among researchers about the optimum waveform,
frequency, and voltage to be used in Wenner technique. As such, a series of experiments were initially
designed and carried out to determine the optimum test configuration. Considering the proposed
application for electrical resistivity measurements, the main objective for optimization of testing
parameters was to find the combination of parameters that increases the repeatability of electrical
resistivity measurements. The considered levels for the three influential factors are presented in Table
28. These factor levels were selected based on the literature [84, 83, 89]. The 3x3x2 factorial design
used for test optimization is presented in Table 29. The electrical resistivity readings, the average value
for each test configuration and the variations for the designed experiments are presented in Table 30.
Table 28: Factors and their levels
Factor Level 1 Level 2 Level 3
Frequency (Hz) 13 500 1000
V p-p (V) 2 5 8
Waveform Sinusoidal Square
93
Table 29: 3 x3 x2 factorial design for Wenner technique optimization
Experiment
Frequency
(Hz)
V p-p
(V)
Waveform
1 1 1 1
2 1 1 2
3 1 2 1
4 1 2 2
5 1 3 1
6 1 3 2
7 2 1 1
8 2 1 2
9 2 2 1
10 2 2 2
11 2 3 1
12 2 3 2
13 3 1 1
14 3 1 2
15 3 2 1
16 3 2 2
17 3 3 1
18 3 3 2
In addition, main effects plot for these results are presented in Figure 57. The main effects plots are
commonly used in statistics to examine differences between level means for one or more factors. There
is a main effect when different levels of a factor affect the response differently. Based on the main
effects plot, it seems that changes in peak to peak voltage have the most significant effect on the
electrical resistivity variations, while changes in waveform result in the least significant changes in the
resistivity measurement variations.
94
Table 30: Experimental results for Wenner technique optimization (ohm-m)
Experiment Sample 1 Sample 2 Sample 3 Avg.
Std.
Deviation
COV (%)
1 6.61 8.43 6.12 7.05 0.99 14.09
2 5.93 5.34 8.15 6.47 1.21 18.69
3 6.42 8.45 8.1 7.66 0.89 11.57
4 5.57 7.79 7.2 6.85 0.94 13.70
5 7.75 5.81 5.87 6.48 0.90 13.91
6 6.35 8.43 8.06 7.61 0.91 11.90
7 4.86 7.13 5.37 5.79 0.97 16.80
8 5.19 4.87 7.68 5.91 1.26 21.24
9 6.46 4.87 4.61 5.31 0.82 15.39
10 5.64 6.84 6.76 6.41 0.55 8.54
11 7.76 6.94 6.51 7.07 0.52 7.33
12 6.35 6.87 7.91 7.04 0.65 9.21
13 7.35 6.91 5.77 6.68 0.67 9.97
14 6.67 8.09 6.1 6.95 0.84 12.03
15 6.32 6.94 7.78 7.01 0.60 8.53
16 8.49 7.05 7.24 7.59 0.64 8.41
17 7.29 7.61 6.58 7.16 0.43 6.01
18 6.34 6.97 5.61 6.31 0.56 8.81
Figure 57: Main effects plot for variation in electrical resistivity values
95
Based on the experimental results, a sine wave with a frequency of 1 kHz and peak-to-peak voltage of
8V (i.e. configuration 17 which resulted in lowest COV) was selected for the main round of
experiments. Finally, a conductivity standard solution (1417 μS/cm) with a resistivity of 7.05 ohm.m
was used for calibration purposes and for accurate calculation of the geometrical factor k. The
electrical resistivity of the solution was measured as 6.87 ohm.m (instead of 7.05 ohm.m). The
geometrical factor was adjusted accordingly, and the theoretical factor (2xπxa) was replaced by
k=0.122.
5.1.2. Experimental Evaluation and Discussion
To evaluate the reliability and responsiveness of the electrical resistivity measurements for quality
monitoring purposes, mixtures in Table 25 were used. Distilled water was used to prepare all mixtures,
in order to prevent city water minerals from affecting electrical resistivity of mixtures. Each mixture
was prepared four times and each time, one sample (10x20 cm cylinder) was used for electrical
resistivity measurements. For each test, the electrical resistivity was measured 10 times during a 10-
second period, and the average value was used as the test result. The experiments were done within 8
minutes of initial water-cement contact, and the fresh mortar temperature was in the range of 19-23
o
C
during all experiments. Figure 58 presents the results of electrical resistivity measurements (for 4 test
results) at different variation levels.
96
Figure 58: Influence of change in water content on electrical resistivity
Based on the experimental results, large variations and considerable overlap in the electrical resistivity
measurements for different mixtures are observed. Coefficient of variation (COV) at different variation
levels is in the range of 4%-6%. For all variation levels, except for +15 liter/m
3
, no statistically
significant change in electrical resistivity happens as a result of change in the QCREF water content. In
other words, the variations in readings are larger than the changes in electrical resistivity as a result of
change in water content. As an exception, for QCREF+15, a 15 liter/m
3
increase in water content of the
reference mixture resulted in a 13% reduction in electrical resistivity. As a general pattern, the
reduction in electrical resistivity of printing mixture as a result of increase in water content can be
inferred from Figure 58. This observation could be justified by the decrease in ion concentration of
pore solution (liquid phase), which results in lower conductivity of fresh mixture.
In summary, it seems that the proposed technique based on electrical resistivity cannot be considered
as a reliable quality monitoring measure for detecting variations in the water content of mixtures. Low
electrical resistivity of fresh mixtures and large variations in resistivity measurements could be stated
97
as the major reasons for unsatisfactory performance of electrical resistivity as a quality monitoring
technique.
As a side discussion, it should be mentioned that electrical resistivity measurements for two mixtures
were continued for one day, and the results are presented in Figure 59. The obtained results show that
the difference and distinction between electrical resistivity of mixtures increases over time. This is an
implication for possibility of use of electrical resistivity, as a non-destructive test, to monitor early age
properties of printed mixture such as mechanical strength. Considering the fact that Contour Crafting
aims at fast building construction (1-2 days), early age strength monitoring is another important aspect
of quality monitoring of such process. Future research and extensive experimental data are needed to
investigate this hypothesis.
Figure 59: Prolonged electrical resitivity measurements for QCREF and QCREF+10
98
5.2. Using Extrusion Pressure
The second proposed technique for real time extrusion quality monitoring is based on extrusion
pressure measurements. The hypothesis is that, at constant printing speed, monitoring the changes in
the extrusion pressure could reveal the unacceptable changes in the rheology of printing material. The
instantaneous detection of material variations could prevent extrusion of unacceptable layers before the
overall quality of printed structure is affected.
To implement the proposed technique, a 0.05cm thick resistive pressure sensor with an active sensing
area of 3.81cm x 3.81cm (1.5inx1.5in) was used. The pressure sensor (Figure 60) exhibits a decrease in
resistance when there is an increase in the force applied to the active area. The sensor is made of two
layers separated by a flexible spacer. The more pressure is applied to the active sensing area, the more
of active element spots touch the semiconductor, which in turn, reduces the resistance (Figure 60).
(a) (b)
Figure 60: (a) The force-sensitive resistor [91] used for measuring extrusion pressure and (b) its
construction [92]
Using an Arduino and a 10kΩ pull-down resistor, a circuit was developed for measuring sensor values.
Instead of the resistance, the voltage measured by the Arduino analog pin was used as an indicator of
the applied pressure. Since the analog to digital converter (ADC) on the Arduino is a 10-bit ADC, it
99
is able to detect 1024 (2
10
) discrete analog levels. This means that it maps input voltages between 0
and 5V into integer values between 0 and 1023. Figure 61 indicates the experimentally measured
sensor values at different pressure levels, which shows some inherent variability in sensor values.
Figure 61: Calibration curve for the extrusion pressure setup
5.2.1. Sensor Installation and Data Acquisition
In order to embed the sensor in the extrusion system, a nozzle with a side opening of the same size as
the sensor, was designed and 3D printed (Figure 62). A piece of thin latex was attached to the inner
surface of the nozzle to cover the opening and to protect the sensor from direct exposure to the printing
mixture, while not interfering with extrusion pressure measurements. To fix the sensor in place and to
provide a rigid support for the sensor backside, a 3D printed cover (white piece in Figure 62) was
attached to the nozzle from outside using epoxy.
100
Figure 62: Embedding the pressure sensor in the nozzle for extrusion pressure measurements
For experimental study, each printing mixture was prepared four times and each time a layer was
extruded. During each experiment, the extrusion pressure was measured every 15 milliseconds and
saved to an SD memory card using the developed data logging system. The developed extrusion
pressure data presentation and logging system is shown in Figure 63.
Figure 63: Extrusion pressure data presentation and logging system
101
The results of pressure sensor values for extrusion of four layers is used as the basis of evaluation for
each mixture. Figure 64 (a) presents a sample data set collected during extrusion of one layer.
Considering the pattern observed in the collected data, filtering was needed to obtain information from
the data. As such, extrusion pressure cycles were first detected using a filter (Figure 64 (b)). Then, the
maximum measured sensor values (peak values) were found and used. The values obtained over ten
seconds, excluding the initial few seconds of extrusion, were used to calculate the average pressure
value and standard deviations for each extruded layer.
102
(a)
(b)
Figure 64: (a) Pressure sensor data during extrusion process (b) Filtering to obtain peak pressure values
103
5.2.2. Results and Discussion
Table 31 presents the experimental results for pressure sensor measurements for layers extruded using
different mixtures. The largest variations are observed for QCREF-5 mixture, with a coefficient of
variation of 1.54%. Considering the average sensor values and the standard variations, it is concluded
that no statistically significant change happens in extrusion pressure as a result of change in QCREF
water content. The unsuccessful detection of material variations through extrusion pressure
measurements can be attributed to two factors. First one is the low precision and inherent variability of
the pressure sensor which was used in this study. The second factor could be the insignificant changes
in the extrusion pressure as a function of change in material viscosity caused by the change in water
content. It should be mentioned that the amount of change in viscosity as a result of change in water
content is mixture-dependent and depends on many parameters such as cement content of the mixture
and the included chemical admixtures.
Table 31: Experimental results for pressure sensor measurements
Layer 1 Layer 2 Layer 3 Layer 4 Average Std. Deviation COV (%)
QCREF+15 892.1 [5.7] 905.1 [10.0] 922.3 [6.2] 903.3 [4.9] 905.7 10.8 1.19
QCREF+10 909.9 [10.6] 928.6 [4.8] 920.0 [6.5] 919.3 [5.4] 919.5 6.6 0.71
QCREF+5 917.2 [5.9] 902.0 [7.4] 915.5 [3.9] 923.6 [4.4] 914.6 7.9 0.86
QCREF 915.6 [4.9] 942.6 [2.6] 908.5 [6.8] 920.6 [7.5] 921.8 12.7 1.38
QCREF-5 888.2 [6.4] 924.2 [6.8] 895.1 [9.6] 895.0 [6.6] 900.6 13.9 1.54
QCREF-10 Tearing -
QCREF-15 Not extrudable -
For two mixtures with lowest water contents, QCREF-10 and QCREF-15, continuous extrusion was
not possible due to excessive stiffness of fresh mixtures. In the case of QCREF-15, zero deposition was
made by the extruder and, accordingly, sensor values were consistently in the range of 0-20. For
104
QCREF-10, material extrusion was intermittent. The collected data for one layer is presented in Figure
65. Compared to other mixtures, there are larger variations in the sensor values in this case. Most
importantly, the periods during which extrusion was interrupted could be easily recognized on the
graph. The range of values measured during extrusion interruption is 0-200. The reason for measuring
such values, compared to near-zero values observed in the case of QCREF-15, is the presence of small
portions of material which are often inside the nozzle but not forced out by pressurized mixture. Based
on the observations for QCREF-10 and QCREF-15, it can be concluded that the pressure sensor
measurements could be used to monitor the extrusion flow through the nozzle. Any interruption in the
extrusion flow, due to mechanical system malfunction or excessive material stiffness can be detected
by observation of significant reduction in the pressure sensor readings.
Figure 65: Pressure sensor data for QCREF-10
105
It should be emphasized that the conclusions made here with respect to the extrusion pressure as a
quality monitoring technique are limited to the use of force sensor that was selected for this study and
cannot be generalized. Additionally, as a side note, one important observation during the experimental
study of extrusion pressure was the high sensitivity of the sensor performance to humidity. In several
iterations, as a result of sensor damage, the pressure sensor was replaced and the sealing for sensor
installation was improved, and experiments were repeated until consistent sensor values were obtained.
5.3. Using Power Consumption Level of Mixture Agitator Motor
The third technique which is proposed and investigated as a real-time quality monitoring measure for
Contour Crafting is the continuous power consumption measurement of mixture agitator motor. The
hypothesis is that by monitoring the changes in the power consumption of the motor, unacceptable
changes in the rheology of printing material (caused by any reason) could be detected almost
instantaneously to prevent deposition of unacceptable layers. The proposed idea can also be described
as embedding an “inline” viscometer or torque meter in the wet material flow circuit.
While the proposed idea has not been investigated for an extrusion system before, several past research
projects have used power consumption of the concrete “mixer” for different purposes. For instance,
Chopin et al. [93] have used power consumption curve of a laboratory mixer in order to evaluate
concrete homogenization during the mixing procedure. Cazacliu and Roquet [94] provided a
description of different stages of mixing based on the power consumption of a twin shaft mixer, and
presented a model for concrete mixing kinetics. Figure 66 presents a graph developed by Teillet et al.
[95] which includes a notional curve and as well as several experimental measurements. It describes
different stages of mixing based on the power consumption rate of the mixer motor.
106
Figure 66: Description of the power consumption evolution [95]
To study the proposed technique for Contour Crafting extrusion process, several 3D printed blades
were installed on a shaft connected to a DC motor. The design of the agitator blades was inspired by
continuous concrete mixers which are commonly used in the construction industry while the
functionality of these blades is similar to a viscometer paddle. The power consumption of this motor
was used as an indication of the printing material viscosity.
It should be noted that the employed hopper-agitator system was loaded batch by batch. To reduce the
variations in the power consumption readings, the hopper opening was closed so that the material
stayed inside during the measurements. For all experiments, the same amount of fully mixed printing
mixture, 4kg, was used. Then the mixture agitator motor was turned on and after 30 seconds (6 minutes
and 30 seconds from the initial water-cement contact) the power consumption readings were carried
out during a 10-second period. The initial delay of 30 seconds, while the motor was operating, was
essential because of the large variations which were observed during that period. These initial
107
variations are anticipated to be reduced when there is a continuous feed of material to the extruder,
compared to the batch-by-batch loading process which was adopted in the present research.
For main experiments, each printing mixture was prepared four times and the aforementioned
procedure was carried out. The average of the four test results was used as the basis of evaluation. To
measure electrical power consumption, defined as the product of voltage times current, a constant
voltage of 15V was used for all experiments and the corresponding current was recorded. Table 32
presents the experimental results for power consumption measurements for different mixtures, as well
as when the hopper was empty.
Table 32: The experimental results for power consumption measurements (watt)
Test 1 Test 2 Test 3 Test 4 Average Std. Deviation COV (%)
Empty extruder 24.5 [1.1] 26.6 [1.0] 29.0 [1.5] 26.9 [1.2] 26.8 1.6 5.97
QCREF+15 72.6 [2.1] 71.3 [1.8] 75.4 [2.4] 71.8 [1.8] 72.8 1.6 2.20
QCREF+10 88.4 [2.8] 97.7 [2.5] 86.8 [3.1] 94.2 [3.1] 91.8 4.4 4.79
QCREF+5 78.2 [2.6] 88.2 [3.3] 81.8 [2.2] 86.4 [2.6] 83.6 3.9 4.67
QCREF 85.4 [2.5] 89.4 [2.8] 90.0 [2.5] 88.1 [2.1] 88.2 3.1 3.51
QCREF-5 81.7 [2.8] 91.4 [1.8] 83.8 [2.4] 81.3 [2.3] 84.6 4.1 4.85
QCREF-10 101.5 [3.4] 105.1 [2.4] 98.6 [3.6] 104.3 [3.7] 102.4 2.6 2.54
QCREF-15 119.4 [2.8] 112.7 [2.2] 117.0 [2.9] 118.5 [3.0] 116.9 2.6 2.22
A significant increase of 229% was observed in the motor power consumption as a result of
introducing QCREF mixture to the extruder. For mixtures with different water contents, the standard
deviation of each experiment is in the range of 1.6-4.4 watts, resulting in coefficients of variations of
2.2%-4.85%.
108
For visualization purposes, the influence of different variation levels of QCREF water content on the
motor power consumption is presented in Figure 67. The standard deviation related to each variation
level is also shown using vertical bars.
Figure 67: Influence of change in water content on agitator motor power consumption
Figure 67 shows that, as a general trend, increased water content reduces the current drawn by the
agitator motor, which means reduced power consumption. It is important to note that, however, the
measured data points do not form a strictly decreasing sequence. In specific, considering the variations,
changing the water content in the range of -5liter/m
3
to +10liter/m
3
does not result in a statistically
meaningful change in the power consumption. On the other hand, it seems that at ±15liter/m
3
variation
levels, the power consumption readings reflect the change in the properties of the printing material.
This is also true for -10liter/m
3
variation level. In general, the results reveal that reduction in the
QCREF water content, compared to increase in QCREF water content, results in larger changes in the
extruder power consumption. In other words, the proposed technique is more responsive when the
water content is lower than the target value, compared to the cases where additional water is added to
109
the mixture. This conclusion cannot be generalized to all printing mixtures, and in-depth rheological
studies are needed to define the performance of this technique at different ranges of plastic viscosity
and yield stress.
Another important observation during the experimentation was the high sensitivity of the readings to
the changes in the components of the setup, or test configurations. For instance, a considerable 0.7-amp
increase in current, equivalent to a 10.5 watts power increase, happened as a result of minor tension
adjustment on the coupling of the mixer motor to the motor shaft. Such a high level of sensitivity to
changes in the physical system calls into question the reliability of this technique for full-size
implementation. It could be troublesome during the operation of the Contour Crafting robot in a real-
life project which may continue for several days, as it can significantly increase the variations in the
power measurements and require frequent calibration of the system.
5.4. Using Computer Vision
Many research projects have investigated the use of computer vision techniques for tracking of
construction personnel, equipment, and materials on conventional construction sites. Progress
monitoring, safety analysis, and better communication between different architecture, engineering, and
construction teams [96] are some of the areas explored by researchers in this regard. A comprehensive
overview of application of computer vision techniques for continuous monitoring of unsafe conditions
and actions is provided in [97]. The overall procedure for action recognition techniques is presented in
Figure 68.
110
Figure 68: Construction activity recognition using computer vision techniques [97]
In this section, a real-time extrusion quality monitoring technique based on computer vision is
developed for Contour Crafting process. The details of the experimental setup, the data acquisition and
processing unit, and the proposed algorithm are discussed. Then the results of the experiments which
are carried out to evaluate the proposed algorithm are presented. In addition, as an early-stage concept
demonstration, a closed-loop extrusion system is developed based on the proposed algorithm and by
adopting proportional control. It should be noted that the idea of using computer vision for extrusion
monitoring has been patented by Dr. Behrokh Khoshnevis [98] (Figure 69).
111
Figure 69: Examples of extrusion nozzle assemblies described in the patent US 8944799 B2
For experimental study, all concrete layers were extruded on a white surface. The white printing bed
enhances the contrast between the layer and the background, and facilitates the segmentation process
of video frames. Additionally, low power consumption LEDs (17.3cm x 1.7cm) were used to enhance
the lighting of the extrusion area and to keep the illumination conditions constant for all experiments.
112
With respect to 2D data acquisition and processing system, a Logitech 720p HD camera was used. It
was firmly attached to the extruder using a 3D printed mount, such that it was facing the top surface of
the printed layer. The distance from camera lens to the top surface of the extruded layer was fixed at
40cm. All the extrusion videos were processed in real time by a Raspberry Pi 3 model B. Raspberry Pi
is a low cost, credit-card sized single-board computer equipped with Broadcom BCM2837 chip with
1.2GHz 64-bit quad-core processor [99]. OpenCV 3 with Python bindings are installed on the
Raspberry Pi and used to implement the developed algorithm to process the extrusion videos. OpenCV,
originally developed by Intel, is a library of programming functions mainly aimed at real-time
computer vision [100, 101].
5.4.1. Proposed Algorithm for Vision-based Extrusion Quality Monitoring
The developed algorithm measures the width of freshly extruded layer and compares it with the target
layer width, in order to detect over-extrusion or under-extrusion conditions. It should be mentioned that
over-extrusion or under-extrusion conditions could happen as a result of two different occurrences:
1- Inappropriate selection of “extrusion rate” and “nozzle travelling speed” parameters for a given
mixture, or
2- Changes in the properties of the printing mixture given a constant printing speed. For example,
at a given printing speed, reducing the viscosity of the printing mixture could increase the layer
width.
Furthermore, while over-extrusion can be identified by increased layer width, for under-extrusion three
different scenarios could be described. Reduced layer width (compared to the target width), tearing and
discontinuity of extruded layer, and zero deposition are three occurrences that could happen as a result
of imbalance between extrusion parameters and printing material properties leading to under-extrusion.
113
Regarding the video processing algorithm, a shape based approach is adopted, rather than color or
texture based approaches which usually do not provide reliable results for detection of slender objects
[102, 103]. Moreover, the proposed method is not required to process every frame of the video.
Therefore, to reduce the computational cost of the algorithm, only every tenth frame is used for
processing while the original recorded video speed is 30 fps. The frequency of grabbing frames can be
adjusted based on the nozzle travelling speed. For each extrusion test, the video capture process starts
at the same time the nozzle starts travelling. However, in order to prevent the initial transition state- the
first few seconds before steady state extrusion- from affecting the measurements, the initial 120
captured frames are disregarded.
To pre-process the data, initially the input frame is resized to a lower resolution, converted to
grayscale, and the region of interest (considering the position of the layer in the scene) is cropped and
used for further processing. The reason for this is to make the resulting image simplified and easier to
process, compared to the original full color frame which contains a larger number of pixels with B, G,
R, and alpha (degree of transparency) channels. Next, the image is blurred using a Gaussian filter.
Image blurring, also referred to as image smoothing, removes high frequency content (e.g. noise) from
the image. Subsequently, Otsu’s binarization is applied to segment the background (printing bed) from
the foreground (concrete layer) through inverted binary thresholding. In this approach, instead of
assigning a specific value as threshold by trial and error (global thresholding), it is automatically
calculated based on the scene lighting conditions [104, 101, 102]. After this step, a binary image is
produced in which the extruded layer is represented in white and the background is presented in black
(Figure 70).
114
Figure 70: An example of original frame and produced binary image for top view of layer extrusion
Then, all the image contours are extracted, resulting in occasional detection of instances of blobs on the
concrete layer or on the background area. To eliminate these random detected contours, also called
false positives, size and shape discrimination is performed based on the geometry of the selected layer
segment. To this end, parameters such as contour area and contour extent (the ratio of contour area to
bounding rectangle area [101]) are used. The applied size and shape discrimination techniques also
filter out the frames in which layer edges are detected incorrectly. Figure 71 demonstrates such a frame
that is automatically excluded by the developed algorithm due to the incorrectly detected contour.
115
Figure 71: A frame with incorrectly detected layer, which is automatically ignored by the algorithm
Upon finding and verifying a contour corresponding to a layer segment, the width of the layer is
measured at different y coordinates along the height of the rectangular contour. These measurements in
pixel are then converted to inches using “pixel per inch ratio” obtained through the calibration
procedure. The measurements related to the last 10 acceptable processed frames are used for
calculating average width of the layer. This is equivalent to approximately 20cm of the length of a
layer. Figure 72 presents three examples of different extrusion conditions which are detected by the
vision system.
116
Figure 72: Examples of detecting different extrusion conditions through developed vision system
5.4.2. Calibration and Experimental Evaluation
A 3.81cm x 2.54cm x 38.1cm (1.5in x 1in x 15in) concrete layer was cast and used for the vision system
calibration. In order to ensure the accuracy of layer dimensions, a mold was 3D printed using ABS
plastic and was filled with QCREF mixture. After one day, the plastic mold was removed and the
width of hardened layer was measured using the developed vision system (Figure 73). For all the
117
processed video frames, the layer width was measured as 63 pixels. Accordingly, a “pixel per inch
ratio” of 42 was selected. For verification purposes, a 5.08cm x 2.54cm x 38.1cm (2in x 1in x 15in)
concrete layer was also prepared and tested in a similar fashion. The width of the 2in wide layer was
continuously measured as 85 pixels. This result verifies the selected “pixel per inch” ratio and indicates
the high accuracy of the developed vision system under the experimental conditions.
Mixtures described in Table 25 were used for evaluation of the vision-based quality monitoring
system. Each mixture was prepared four times and a single layer was printed each time at a linear
printing speed of 6cm/s. Layers were deposited within 5 minutes after the completion of the mixing
procedure, in order to avoid any significant changes in the rheology of the mixture.
Figure 73: The molded concrete layer and experiment setup for vision system calibration
Table 33 presents the vision system measurements and variations for the layers extruded using the
reference and deliberately-altered mixtures. As anticipated, the width of the layers extruded with
118
QCREF are within the range of 3.81±0.381 cm, complying with print quality requirements (Chapter 5).
Increase in QCREF water content by 5, 10, and 15 liter/m
3
results in 8%, 17%, and 41.5% increase in
the layer width, respectively. Considering the standard deviation of each experiment, at all levels of
variation the layer width is changed as a result of change in water content. For mixtures with reduced
water content, the maximum reduction in water content while having a continuous layer deposition was
-5 liter/m
3
. In this case, a minor 5% reduction in layer width was observed as a result of insufficient
material being deposited to fill the nozzle width. For lower water contents, either discontinuity in
material deposition was observed, or in the case of QCREF-15, the extrusion system was not able to
deposit any material at all. The vision system identified both these cases as under-extrusion, as a result
of detecting no contour that could be associated with a concrete layer.
Table 33: Vision system measurements for reference and altered mixtures (all values in mm)
Mix ID
Layer
Average width
1 2 3 4
QCREF 39.7 [0.7]
a
40.3 [0.2] 40.5 [0.5] 37.9 [1.1] 39.6 [1.0]
QCREF+5 44.7 [0.4] 43.1 [1.1] 41.6 [1.0] 41.6 [0.3] 42.8 [1.3]
QCREF-5 37.8 [0.2] 37.5 [0.7] 38.3 [0.5] 36.3 [0.9] 37.5 [0.7]
QCREF+10 46.1 [0.8] 46.4 [0.5] 45.8 [0.7] 46.8 [0.8] 46.3 [0.4]
QCREF-10 Tearing -
b
QCREF+15 54.9 [1.4] 55.3 [2.1] 57.5 [1.2] 56.8 [1.7] 56.1 [1.1]
QCREF-15 Not extrudable -
b
a
Values in brackets are standard deviations of the measurements for 10 different frames of the video (in mm)
b
Detected as “under extrusion condition” by the vision system
Except for QCREF+15 mixture, the standard deviation of width measurements for 10 processed frames
(within one layer) is less than 1.2mm. However, for the mixture with the largest amount of additional
water (QCREF+15), the standard deviation of measurements for a single layer is significantly larger.
This is mainly due to the actual variations in the width of any layer made using these mixtures and the
119
irregularities in the geometry of such layer (Figure 74). In these cases, however, the width of the layers
is considerably larger than the target value of 3.81 cm (1.5 in) and even with the existing variations the
over-extrusion condition, caused by erroneous rheology of the printing mixture, can easily be detected
by the vision system.
In addition, it should be mentioned that, the width of each layer was also measured at five different
points using a digital caliper and the average value was compared to the vision system measurement. In
all cases, the vision system measurement was within ±1.7 mm of the average width measured by the
digital caliper. The observed difference in the measured values could not necessarily be attributed to
the vision system error, but also to the selection of the points which were chosen to be measured by the
caliper (specially for the cases such as the one shown in Figure 74).
Figure 74: Variations in width of a layer made with QCREF+15 mixture
Finally, the changes in the extrusion rate as a result of change in printing mixture water content should
be discussed. The increase in layer width caused by reduction in mixture viscosity (Table 33) clearly
120
proves the dependency of extrusion rate on the material rheology. Considering the printing speed of 6
cm/s, the extrusion rate at different water content levels is plotted in Figure 75.
Figure 75: Change in extrusion rate as a result of change in printing mixture water content
It is observed that a significant 41% increase in extrusion rate occurs as a result of reduced plastic
viscosity of material when ±15liter/m
3
of additional water was present in the mixture. To have an
estimate of the pressure caused by the mixture in the extrusion barrel, hydrostatic pressure is calculated
here. It should be mentioned that hydrostatic pressure is used as an indication of the formwork pressure
caused by self-consolidating concrete. With respect to the printing mixture, calculating the hydrostatic
pressure results in an upper hand estimate of the amount of pressure applied to the extrusion system,
which is responsible for variations in the extrusion rate. (Equation 14 shows the hydrostatic pressure
applied by the mixture when the extrusion barrel (used in this study) is full.
P= ρ x g x h = 2300 kg/m
3
x 9.81 m/s
2
x 0.30 m= 6769 Pa ≈ 6.8 kPa
(Equation 14)
121
where P is the hydrostatic pressure, ρ is the density of the printing material and h is the height of
column of printing mixture in the extrusion barrel.
Using the results from Section 6.2 (Table 31 and Figure 61), an average extrusion pressure of 51 kPa
can be assumed. As such, it can be concluded that the pressure from printing material, depending on its
viscosity, can be up to a considerable 13.3% of extrusion pressure. This is the likely major reason for
variations in the extrusion rate, and in turn, layer width when the viscosity of printing material
changes. In general, the ideal case would be independency of extrusion rate from printing material
rheology such that at a given printing speed, the geometry of layers be independent of material
properties. Further research on dependency of different extrusion mechanisms on the printing material
rheology and influential parameters is needed for a deeper understanding of this phenomenon.
5.4.3. Vision-based Closed-loop Extruder: Concept Demonstration
In this section, the proposed vision algorithm is used to develop a feedback-control extrusion system
for Contour Crafting. The developed system is able to automatically adjust the extrusion rate such that
a layer with target width is extruded without the need for prior calibration. For experimental
demonstration, the nozzle traveling speed was kept constant at 6cm/s and the extrusion rate was
dynamically controlled based on the feedback provided by the vision system.
“Proportional control” was used to adjust the extrusion rate based on the vision system feedback, while
initially the extrusion rate is set at maximum. In brief, the output of a proportional controller is a
function of the error signal times the proportional gain (K p.e(t)). After some initial experimentation and
system tuning, the proportional gain was selected as 25. In order to reduce the response time of the
extrusion system, the layer segment closest to the nozzle was selected for measurements. Furthermore,
in order to prevent frequent unnecessary changes in extrusion rate and increase stability of the system,
122
any change in the extrusion rate was made based on the last 3 processed frames, only if there was
insignificant scatter (a maximum difference of 0.1 in) among the most recent 3 measurements. The
performance of the developed closed-loop system for two different mixtures made with different
ingredients (one with Portland cement and the other with Calcium Sulfo Aluminate cement as the
binder), with no prior information on their extrusion parameters, is presented in Figure 76. It is
observed that in both cases the initial width of the layer at maximum extrusion rate (>1.8in) is changed
to the values close to the ±10% range of target value in less than 3 seconds.
Figure 76: Performance of the developed vision-based closed-loop extruder for two different mixtures
As an early stage work, this proves the feasibility of developing a smart vision-based extruder which is
able to automatically print layers of specific dimensions using any printable mixture without the need
for prior calibration. Such smart extrusion system could increase the robustness of automated
construction systems and prevent the quality of printed structure from being affected by regular
variations in the material properties. Furthermore, it facilitates the use of several different printing
123
mixtures, with different rheological parameters, during a project. This is commonly referred to as
“multi-material 3D printing” or “functionally graded materials”, which is visualized using iron oxide
pigments to produce layers of different color in Figure 77. In a similar fashion, per project
requirements, mixtures with different mechanical strength, unit weight, or insulation properties could
be used for different elements of a building (walls, roof, etc.). An automatic extrusion system would
enable transition from one material to another material (with different rheology) without adverse
effects on the quality of printed layers.
Figure 77: Visualization of multi-material construction-scale 3D printing using color mixtures
Possibilities offered by computer vision for Contour Crafting process are beyond extrusion monitoring.
An important aspect of robotic construction which needs extensive future research is construction
inspection and progress monitoring. As illustrated in Figure 78, computer vision techniques could be
used to monitor the automated construction progress and to provide useful as-built information to
detect any deviation from the initial design. It is anticipated that inspection of automated construction
processes in future will not solely rely on cognitive capabilities of a human inspector or superintendent
anymore and automated systems will be involved in the inspection process. For instance, any
unacceptable deformations in the freshly printed walls could be detected by the vision system so that
construction process could be temporarily stopped to prevent collapse of the structure. Surface quality
of printed layers could also be monitored using a vision system, to detect tearing or similar defects on
the surface of printed walls.
124
Figure 78. Use of computer vision for construction progress monitoring
5.5. Comparison of Investigated Quality Monitoring Techniques
In this section, an overview and comparison of the performance of the proposed quality monitoring
techniques is presented. In order to have a quantitative metric for responsiveness to material variations,
a dimensionless statistic called sensitivity index (also referred to as standardized response mean [105])
is adopted. This index is commonly used in signal detection theory to provide the separation between
the means of the signal (𝜇 ) and noise distributions (𝜇 ), compared to the standard deviation of the
signal (𝜎 ) or noise distribution (𝜎 ) [106]. The sensitivity index (SI) is calculated using Equation 13.
SI =
. ( )
(Equation 15)
Table 34 presents the sensitivity index values for different techniques at designed variation levels. The
cases which are identified with a cross, cannot be detected by the corresponding technique. The cases
which are identified with a checkmark, can be qualitatively detected by the corresponding technique.
125
For instance, in the case of QCREF-10, the extrusion pressure readings drop to values under 200 for
periods of few seconds, which makes it easy to recognize the extrusion interruptions.
Table 34: Sensitivity index of propsoed techniques at designed variation levels
Variation level
(liter/m
3
)
Proposed quality monitoring techniques
Electrical
resistivity
Extrusion
pressure
Power
consumption
Computer
vision
+5 x x x 2.8
-5 x x x -2.4
+10 x x x 8.8
-10 x 4.9
+15 -2.9 x 6.5 15.7
-15 x 10.0
Based on the results, it seems that the vision-based technique has the highest precision and
responsiveness to changes in material properties. For smallest variation level (±5 liter/m
3
), the mean of
change in the measured layer width, compared to the reference mixture, is more than twice the
variations in measurements. As anticipated, for larger variations, the sensitivity index increases. For
instance, for +10 liter/m
3
and +15 liter/m
3
change in water content, sensitivity indices of 8.8 and 15.7
are calculated, which means a significant change in response that could be easily detected. In general,
it seems that computer vision has high potential as a quality monitoring technique for construction-
scale 3D printing.
Power consumption is the only other proposed measure which was able to detect ±15 liter/m
3
variation
levels in material water content. Considerable sensitivity indices of 6.5 and 10.0 indicate reliable
detection of changes at these variation levels through power consumption measurements. A major
concern regarding implementation of this technique in a real-life project would be the high sensitivity
to the changes of mechanical components, as discussed in 6.3.
126
Electrical resistivity monitoring was the least responsive and reliable technique among the four
proposed techniques. In all cases, except for +15 liter/m
3
, this technique was not able to distinguish a
significant change in electrical resistivity as a result of changes in water content. The main reason for
inferior performance of this technique is the large variations in electrical resistivity measurements as
well as low electrical resistivity of the fresh printing mixtures, that prevent detection of changes in
material.
Finally, it should be emphasized that the focus of the current research project is on the variations in
material properties, and the comparison was made solely based on this aspect of quality monitoring.
However, there are various other applications that could be considered and investigated. For instance,
power consumption and extrusion pressure measurements could possibly be used for monitoring the
pumping system, detecting blockage in the wet material circuit, or detecting any related hardware
malfunction. Further research work is needed in the area of quality monitoring of Contour Crafting
process in order to investigate its different aspects.
127
6. CONCLUSION AND FUTURE RESEARCH
6.1. Research Summary and Conclusions
In this research two major topics were investigated. Initially, a framework was proposed for
performance-based laboratory testing of fresh printing mixtures. An iterative laboratory testing
procedure was described to assess and modify the four suggested workability aspects of a printing
mixture, namely, print quality, robustness, shape stability and printability timespan. This testing
procedure was designed such that it is applicable to different concrete 3D printing systems, as it
focuses on properties of printed layers rather than employed pumping or extrusion mechanism. To
realize the proposed framework, a laboratory-scale linear concrete printer was constructed. Four
different printing mixtures were also developed and the results for a combination of conventional and
new test methods were reported. Specifically, for evaluation of shape stability of printing mixture two
different test methods, namely, “layer settlement” and “cylinder stability” were proposed.
Experimental data revealed that inclusion of silica fume and Nano-clay (a highly-purified attapulgite
clay) enhances the shape stability of fresh printing mixtures, while minor improvement was observed
as a result of polypropylene fiber addition.
The second major topic which was investigated in this research was real-time quality monitoring of
construction-scale 3D printing. Several novel techniques were proposed and investigated as real-time
extrusion quality monitoring measures for Contour Crafting process: electrical resistivity
measurements, extrusion pressure measurements, extruder power consumption monitoring, and a
vision-based technique. An experimental program was designed and carried out in order to investigate
the performance of proposed techniques under controlled and simplified laboratory conditions.
Changes in the water contents (six variation levels of ±5 liter/m
3
, ±10 liter/m
3
, and ±15 liter/m
3
) were
128
considered as the main source of variations in the system. The experimental results indicated that the
vision-based technique is the most reliable and responsive technique with respect to detecting changes
in mixture water content. Power consumption was the second most responsive technique, which was
able to detect ±15 liter/m
3
variation levels in water content. Considerable sensitivity indices of 6.5 and
10.0 indicated reliable detection of changes at +15 liter/m
3
and -15 liter/m
3
variation levels,
respectively, through power consumption measurements. The third proposed technique, extrusion
pressure measurements, was only able to detect reduction in water content by -10liter/m
3
and -15
liter/m
3
, by temporary (but significant) drop in measured extrusion pressure values. This was due to the
tearing and discontinuity in the extruded layers, caused by stiffness of printing mixture at fresh state.
The fourth investigated technique, electrical resistivity monitoring was the least responsive and reliable
technique among the four proposed quality monitoring techniques. In all cases, except for +15 liter/m
3
,
this technique was not able to distinguish a significant change in the measured values as a result of
changes in water content. The main reason was the large variations in electrical resistivity
measurements as well as low electrical resistivity of the fresh printing mixtures, that prevented
detection of changes in material.
Finally, it is very important to note that the conclusions of this study are limited to the specific
materials, methods, and equipment (type of the sensors, etc.) used herein, and cannot be generalized.
Specifically, the proposed idea of using extrusion pressure for quality monitoring could potentially
have an acceptable performance given that a more accurate pressure sensor is developed or becomes
commercially available. Furthermore, the performance of proposed techniques was evaluated based on
sensitivity to variations in the water content of printing material. However, quality monitoring of
Contour Crafting is far beyond this, and extensive research work is needed to study other aspects as
briefly discussed in the previous chapter. For instance, it seems that computer vision has high potential
129
for construction progress monitoring and providing as-built information during the automated
construction process, or power consumption and extrusion pressure measurements could possibly be
used for monitoring the extrusion mechanical system, detecting extruder blockage, or detecting any
related hardware malfunction.
6.2. Recommendations for Future Work
Considering the short history of construction-scale 3D printing, there are various unexplored research
areas within this field. The main topics which need extensive research include:
- Structural design and performance of 3D printed buildings
- Process planning and toolpath optimization
- Process monitoring and inspection
- Reinforcement techniques
- Curing and strength development of printed structure
- Plastic and drying shrinkage of 3D printed elements and structures
- Durability and long-term performance of 3D printed buildings
It should be noted that there is no data on performance history of actual 3D printed buildings. As such,
many uncertainties with respect to structural performance and durability of these buildings exist.
Extensive future research and actual field data are needed in order to enable wide application of this
technology by the construction industry. With respect to the printing mixture, in specific, four
workability aspects were proposed and studied in this research work and test methods were suggested
for quantitative evaluation of each workability aspect. However, a more fundamental understanding of
influence of mixture properties on its performance in construction-scale 3D printing application is
needed. Thixotropy, dynamic yield stress, static yield stress, and plastic viscosity are the parameters
which are suggested for deeper investigation on fresh state properties of printing mixture. With respect
130
to the hardened properties of a 3D printed element or structure, there are some major concerns which
need to be addressed in future research. Shrinkage of 3D printed elements is anticipated to be
significant due to absence of formwork which usually prevent extensive water evaporation. The
shrinkage, if not controlled, could lead to cracking of printed elements. Influence of interlayer adhesion
on the mechanical properties of a 3D printed element, such as flexural strength, is another issue which
needs to be addressed in the future research. As discussed in the literature review section, interlayer
adhesion is directly affected by the time gap between layers, but underlaying bonding mechanism
(chemical or physical) is not known yet.
131
7. REFERENCES
[1] S. Kubba, Handbook of Green Building Design and Construction (Second Edition),
Butterworth-Heinemann, 2017.
[2] H. Wei, S. Zheng, L. Zhao and R. Huang, "BIM-based method calculation of auxiliary
materials required in housing construction," Automation in Construction, vol. 78, pp. 62-82, 2017.
[3] H. Penttilä, "Describing the changes in architectural information technology to understand
design complexity and free-form architectural expression," Electronic Journal of Information
Technology in Construction, vol. 11, pp. 395-408, 2006.
[4] O. Davtalab, "Benefits of real-time data driven BIM for FM departments in operations control
and maintenance," in ASCE International Workshop on Computing in Civil Engineering (IWCCE
2017), Seatle, 2017.
[5] S. Ahmadi-Karvigh, b. Becerik-Gerber and L. Soibelman, "A framework for allocating
personalized appliance-level disaggregated electricity consumption to daily activities," Energy and
Buildings, vol. 111, pp. 337-350, 2016.
[6] F. Jazizadeh, S. Ahmadi-Karvigh, B. Becerik-Gerber and L. Soibelman, "Spatiotemporal
lighting load disaggregation using light intensity signal," Energy and Buildings, vol. 69, p. 572–583,
2014.
[7] J. Irizarry, M. Gheisari and B. N. Walker, "Usability assessment of drone technology as safety
inspection tools," Journal of Information Technology in Construction, vol. 17, pp. 194-212, 2012.
[8] J. Irizarry and D. B. Costa, "Exploratory Study of Potential Applications of Unmanned Aerial
Systems for Construction Management Tasks," Journal of Management in Engineering , vol. 32, no. 3,
2016.
[9] "TruePoint," [Online]. Available: https://www.truepointscanning.com/construction. [Accessed
07 04 2018].
[10] M. Taylor, S. Wamuziri and I. Smith, "Automated Construction in Japan," in Proceedings of
the Institution of Civil Engineers, 2003.
132
[11] T. Wangler, E. Lloret, E. Reiter, N. Hack, F. Gramazio, M. Kohler and M. Bernhard, "Digital
concrete: opportunities and challenges," RILEM Technical Letters , pp. 67-75, 2016.
[12] "NSF Workshop on Additive Manufacturing (3D Printing) for Civil Infrastructure Design and
Construction," NSF, Arlington, Virginia, 2017.
[13] "ASTM F2792-12a (withdrawn)," American Society of Testing and Materials, 2012.
[14] T. Wohlers, "Wohlers report 2016," Wohlers Associates, Inc, 2016.
[15] R. A. Buswell, R. C. Soar, A. G. Gibb and A. Thorpe., "Freeform construction application
research," Advances in Engineering Structures, Mechanics & Construction, pp. 773-780, 2006.
[16] "D-shape," [Online]. Available: http://d-shape.com/. [Accessed 01 05 2016].
[17] "MIT Media Lab," [Online]. Available: https://www.media.mit.edu/projects/3d-printed-hemi-
ellipsoidal-dome/overview/. [Accessed 02 04 2018].
[18] B. Khoshnevis, X. Yuan, B. Zahiri, J. Zhang and B. Xia, "Construction by Contour Crafting
using sulfur concrete with planetary applications," Rapid Prototyping Journal, vol. 22 , no. 5, pp. 848-
856.
[19] S. Lim, T. Le, J. Webster, R. Buswell, A. Austin and A. Gibb, "Fabricating construction
components using layered manufacturing technology," in Global Innovation in Construction,
Loughborough University, Leicestershire, UK, 2009.
[20] J. G. Sanjayan, B. Nematollahi, M. Xia and T. Marchment, "Effect of surface moisture on inter-
layer strength of 3D printed concrete," Construction and Building Materials, vol. 172, p. 468–475,
2018.
[21] V. N. Nerella, V. Mechtcherine, M. Näther and V. Mechtcherine, "Studying printability of fresh
concrete for formwork free Concrete on-site 3D Printing technology (CONPrint3D)," in 25th
Conference on Rheology of Building Materials, Regensburg, Germany, 2016.
[22] "ASTM C 143/C143M - 15a: Standard Test Method for Slump of Hydraulic-Cement Concrete,"
American Society of Testing and Materials, 2015.
[23] "ASTM C1611/C1611M - 14: Standard Test Method for Slump Flow of Self-Consolidating
Concrete," American Society of Testing and Materials, 2014.
133
[24] "ASTM C1621/C1621M-17: Standard Test Method for Passing Ability of Self-Consolidating
Concrete by J-Ring," American Society of Testing and Materials, 2017.
[25] "ASTM C1437 − 15: Standard Test Method for Flow of Hydraulic Cement Mortar," American
Society of Testing and Materials, 2015.
[26] [Online]. Available: http://civilblog.org/2013/05/10/workability-of-concrete-using-flow-table-
test-apparatus-is1199-1959/. [Accessed 15 8 2018].
[27] "ASTM C1610/C1610M − 17: Standard Test Method for Static Segregation of Self-
Consolidating Concrete Using Column Technique," American Society of Testing and Materials, 2017.
[28] "ASTM C1712-17: Rapid Assessment of Static Segregation Resistance of SelfConsolidating
Concrete Using Penetration Test," American Society of Testing and Materials, 2017.
[29] "The European Guidelines for Self-Compacting Concrete: Specification, Production and Use,"
01 05 2005. [Online]. Available: http://www.efnarc.org/pdf/SCCGuidelinesMay2005.pdf.
[30] E. Koehler and D. Fowler, "ICAR Project 108: Aggregates in Self-Consolidating Concrete,"
International Center for Aggregates Research (ICAR), Austin, 2007.
[31] C. 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, ., vol.
104, no. 5, pp. 461-478, 1999.
[32] V. Hackley and C. Ferraris, "The Use of Nomenclature in Dispersion Science and Technology
(Special Report 960-3)," National Institute of Standards and Technology, Gaithersburg, 2001.
[33] "EL Studio," [Online]. Available: http://elstudioarch.com/. [Accessed 01 02 2016].
[34] P. K. Mehta and P. J. M. Monteiro, Concrete: Microstructure, Properties, and Materials,
McGraw-Hill Education, 2013.
[35] I. Lofgren, "Fibre-reinforced Concrete for Industrial Construction (PhD Thesis)," Chalmers
university of technology, 2005.
[36] T. Le, S. Austin, S. Lim, R. Buswell, R. Law, A. Gibb and T. Thorpe, "Hardened properties of
high-performance printing concrete," Cement and Concrete Research, vol. 42, no. 3, pp. 558-566,
2012.
134
[37] R. Buswell, A. Thorpe, R. Soar and A. Gibb, "Design, data and process issues for mega-scale
rapid manufacturing machines used for construction," Automation in Construction, vol. 17, pp. 923-
929, 2008.
[38] L. H. Anell, "Concrete 3D printer (MSc Thesis)," Lund University , Sweden , 2015.
[39] S. Lim, R. A. Buswell, T. T. Le, S. A. Austin, A. G. Gibb and T. Thorpe, "Developments in
construction-scale additive manufacturing processes," Automation in construction,, vol. 21, pp. 262-
268, 2012.
[40] Y. Weng, M. Li, M. J. Tan and S. Qian, "Design 3D printing cementitious materials via Fuller
Thompson theory and Marson-Percy model," Construction and Building Materials, vol. 163 , p. 600–
610, 2018.
[41] Y. Zhang, Y. Zhang, G. Liu, Y. Yang, M. Wu and B. Pang, "Fresh properties of a novel 3D
printing concrete ink," Construction and Building Materials, vol. 174, pp. 263-271, 2018.
[42] T. T. Le, S. A. Austin, S. Lim, R. A. Buswell, A. G. F. Gibb and T. Thorpe, "Mix design and
fresh properties for high-performance printing concrete," Materials and Structures, vol. 45, pp. 1221-
1232, 2012.
[43] A. Perrot, D. Rangeard and A. Pierre, "Structural built-up of cement-based materials used for
3D-printing extrusion techniques," Materials and Structures, vol. 49, pp. 1213-1220, 2016.
[44] N. Roussel, "A thixotropy model for fresh fluid concretes: theory, validation and applications,"
Cement and Concrete Research, vol. 36, pp. 1797-1806, 2006.
[45] G. Ma, Z. Li and L. Wang, "Printable properties of cementitious material containing copper
tailings for extrusion based 3D printing," Construction and Building Materials, vol. 162, pp. 613-627,
2018.
[46] S. Austin, P. Robins and C. Goodier, "The performance of hardened wet-process sprayed
mortars 52 (2000) 195–208," Magazine of Concrete Research, vol. 52, p. 195–208, 200.
[47] V. Nerella, M. Krause, M. Nather and V. Mechtcherine, "Studying printability of fresh concrete
for formwork free concrete on-site 3D printing technology (CONPrint3D)," in 25th Conference on
Rheology of Building Materials, Regensburg, 2016.
135
[48] V. N. Nerella, S. Hempel and V. Mechtcherine, "Micro and macroscopic investigations on the
interface between layers of 3D printed cementitious elements," in International Conference on
Advances in Construction Materials and Systems, Chennai, 2017.
[49] "Southern Testing," [Online]. Available: https://www.southerntesting.co.uk/services/laboratory-
testing/concrete-testing/. [Accessed 30 03 2018].
[50] "American Society for Testing Materials (ASTM)," [Online]. Available:
https://www.astm.org/Standards/C1611.htm. [Accessed 08 04 2018].
[51] "Self-Consolidating Concrete," American Concrete Institute (ACI), 2007.
[52] R. A.A. and Kazemian A., Self-consolidating concrete: Technology and Application (ISBN:
978-964-463-514-4), Amirkabir University press, 2013.
[53] H. Reinhardt and C. Grosse, "Continuous monitoring of setting and hardening of mortar and
concrete," Construction and Building Materials, vol. 18, p. 145–154, 2004.
[54] Z. Li, L. Xiao and X. Wei, "Determination of Concrete Setting Time Using Electrical
Resistivity Measurement," Journal of Materials in Civil Engineering, vol. 19, no. 5, pp. 423-427 ,
2007.
[55] H. Lee, K. Lee, Y. Kim, H. Yim and D. Bae, "Ultrasonic in-situ monitoring of setting process
of high-performance concrete," Cement and Concrete Research, vol. 34 , p. 631 – 640, 2004.
[56] K. H. Khayyat and N. A. Libre, "Automated Measurement and Control of Concrete Properties
in a Ready Mix Truck with VERIFI," Center for Transportation Infrastructure and Safety (Missouri
S&T), 2013.
[57] "Giatec," [Online]. Available: https://www.giatecscientific.com/smart-concrete-
monitoring/?utm_keyword=banner. [Accessed 02 04 2018].
[58] "Giatec," [Online]. Available: https://www.giatecscientific.com/education/want-to-check-on-
that-concrete-slab-you-just-poured-now-theres-a-smart-phone-app/. [Accessed 02 04 2018].
[59] [Online]. Available: https://www.giatecscientific.com/education/concrete-maturity-in-situ-
compressive-strength/. [Accessed 08 04 2018].
136
[60] A. Perrot, D. Rangeard and A. Pierre, "Structural built-up of cement-based materials used for
3D-printing extrusion techniques," Materials and Structures, vol. 49, no. 4, pp. 1213-1220, 2016.
[61] S. Shah, "Design and Application of Low Compaction Energy Concrete for Use in Slip-form
Concrete Paving," Infrastructure Technology Institute, Northwestern University, 2008.
[62] T. Voigt, J. Mbele, K. Wang and S. Shah, "Using fly ash, clay, and fibers for simultaneous
improvement of concrete green strength and consolidatability for slip-form pavement," Journal of
Materials in Civil Engineering, vol. 22, no. 2, pp. 196-206, 2010.
[63] A. Kazemian, X. Yuan, R. Meier, E. Cochran and B. Khoshnevis, "Construction-Scale 3D
Printing: Shape Stability of Fresh Printing Concrete," in 12th International Manufacturing Science and
Engineering Conference (MSEC 2017), Los Angeles, 2017.
[64] "BASF," BASF, [Online]. Available: https://www.master-builders-solutions.basf.us/en-
us/products/masterfiber/1649. [Accessed 01 04 2016].
[65] S. Kawashima, M. Chaouche, D. J. Corr and S. P. Shah, "Influence of purified attapulgite clays
on the adhesive properties of cement pastes as measured by the tack test," Cement & Concrete
Composites, vol. 48, p. 35–41, 2014.
[66] S. Kawashima, M. Chaouche, D. J. Corr and S. P. Shah, "Rate of thixotropic rebuilding of
cement pastes modified with highly purified attapulgite clays," Cement and Concrete Research, vol.
53, pp. 112-118, 2013.
[67] M. B. S. P. S. Jae Hong Kim, "Effect of mineral admixtures on formwork pressure of self-
consolidating concrete," Cement and Concrete Composites, vol. 32, no. 9, pp. 665-671, 2010.
[68] A. A. Ramezanianpour, A. Kazemian, M. Nikravan and M. A. Moghadam, "Influence of a low-
activity slag and silica fume on the fresh properties and durability of high performance self-
consolidating concrete," in International Conference on Sustainable Construction Materials &
Technologies (SCMT3), Kyoto, Japan, 2013.
[69] C. Poon, S. Kou and L. Lam, "Compressive strength, chloride diffusivity and pore structure of
high performance metakaolin and silica fume concrete," Construction and Building Materials, vol. 20 ,
p. 858–865, 2006.
137
[70] "American Society of Testing and Materials (ASTM)," [Online]. Available:
https://www.astm.org/Standards/C109.htm. [Accessed 08 2016].
[71] R. Wolfs, F. Bos and T. Salet, "Early age mechanical behaviour of 3D printed concrete:
Numerical modelling and experimental testing," Cement and Concrete Research, vol. 106, pp. 103-116,
2018.
[72] SOLIDWORKS, 2018. [Online]. Available: https://www.solidworks.com/product/solidworks-
flow-simulation.
[73] FreeGreen, [Online]. Available: https://www.houseplans.com/plan/1160-square-feet-2-
bedroom-1-bathroom-0-garage-modern-39050. [Accessed 15 06 2016].
[74] "ImageJ," [Online]. Available: https://imagej.nih.gov/ij/.
[75] "SOLIDWORKS," [Online]. Available: https://imagej.nih.gov/ij/.
[76] Y. Qian and G. De Schutter, "Enhancing thixotropy of fresh cement pastes with nanoclay in
presence of polycarboxylate ether superplasticizer (PCE)," Cement and Concrete Research, vol. 111,
pp. 15-22, 2018.
[77] K. Khayat and G. De Schutter, "Mechanical properties of self-compacting concrete," RILEM
Technical Committee TC 228- MPS, 2013.
[78] P. Ghoddousi and A. M. Salehi, "The robustness of self consolidating concrete due to changes
in mixing water," Periodica Polytechnica. Civil Engineering , vol. 61, no. 2, p. 216–225, 2017.
[79] "ASTM C125-15b: Standard Terminology Relating to Concrete and Concrete Aggregates,"
American Society of Testing and Materials (ASTM), 2015.
[80] "ASTM C403/C403M-08: Standard Test Method for Time of Setting of Concrete Mixtures by
Penetration Resistance," American Society of Testing and Materials (ASTM), 2008.
[81] "ASTM C1437-15: Standard Test Method for Flow of Hydraulic Cement Mortar," American
Society of Testing and Materials (ASTM), 2015.
[82] A. Ramezanianpour, A. Kazemian, M. Moghaddam, F. Moodi and A. Ramezanianpour,
"Studying effects of low-reactivity GGBFS on chloride resistance of conventional and high strength
concretes," Materials and Structures, vol. 49, no. 7, pp. 2597-2609, 2016.
138
[83] H. Layssi, P. Ghods, A. R. Alizadeh and M. Salehi, "Electrical Resistivity of Concrete,"
Concrete International, vol. 37, no. 5, p. 41.46, 2015.
[84] "Electrical Resistivity of Concrete- A Literature Review," Portland Cement Association (PCA),
2003.
[85] F. Wenner, "A method of measuring earth resistivity," U.S. Dept. of Commerce, Bureau of
Standards, 1916.
[86] K. R. Gowers and S. G. Millard, "Measurement of Concrete Resistivity for Assessment of
Corrosion Severity of Steel Using Wenner Technique," ACI Materials Journal, vol. 96, no. 5, pp. 536-
541, 1999.
[87] A. A. Ramezanianpour, A. Kazemian, M. Sarvari and B. Ahmadi, "Use of natural zeolite to
produce Self-consolidating concrete with low portland cement content and high durability," Journal of
Materials in Civil Engineering, vol. 25, no. 5, 2012.
[88] S. Feliu, C. Andrade, J. A. Gonzalez and C. Alonso, "A new method forin-situ measurement of
electrical resistivity of reinforced concrete," Materials and structures, vol. 29, no. 6, pp. 362-365, 1996.
[89] K. H. Khayat, T. V. Pavate, J. Assaad and C. Jolicoeur, "Analysis of variations in electrical
conductivity to assess stability of cement-based materials," Materials Journal, vol. 100, no. 4, pp. 302-
310, 2003.
[90] [Online]. Available: https://www.amazon.com/TOTMC%C2%AE-1HZ-500KHz-Function-
Generator-
Triangle/dp/B00NU8ECRQ/ref=pd_day0_23_6?_encoding=UTF8&pd_rd_i=B00NU8ECRQ&pd_rd_r
=3SCHGJ90E3SVEBRJNB7B&pd_rd_w=z5qiv&pd_rd_wg=HR3yd&psc=1&refRID=3SCHGJ90E3S
VEBRJNB7B. [Accessed 13 07 2017].
[91] "Adafruit," [Online]. Available: https://www.adafruit.com/product/1075. [Accessed 09 01
2018].
[92] [Online]. Available: https://learn.adafruit.com/force-sensitive-resistor-fsr?view=all. [Accessed
04 05 2018].
[93] D. Chopin, B. Cazackliu, F. De Larrard and R. Schell, "Monitoring of concrete homogenisation
with the power consumption curve," Materials and Structures, vol. 40, pp. 897-907, 2007.
139
[94] B. Cazacliu and N. Roquet, "Concrete mixing kinetics by means of power measurement,"
Cement and Concrete Research, vol. 39, no. 3, pp. 182-194, 2009.
[95] R. Teillet, S. Bruneaud and Y. Charonnat, "Suivi et contrôle de la fabrication des mélanges-une
nouvelle jeunesse pour le wattmètre différentiel," Bulletin de Liaison des Laboratoires des Ponts et
Chaussées, vol. 174 , 1991.
[96] J. Gauri M., H. Fathi and I. Brilakis, "Automated computation of the fundamental matrix for
vision based construction site applications," Advanced Engineering Informatics , vol. 25, no. 4, pp.
725-735, 2011.
[97] S. J., H. S., L. S. and K. H., "Computer vision techniques for construction safety and health
monitoring," Advanced Engineering Informatics, vol. 29, no. 2, pp. 239-251, 2015 .
[98] B. Khoshnevis, "Techniques for sensing material flow rate in automated extrusion". United
States Patent US8568121B2, 2013.
[99] "Raspberri Pi," [Online]. Available: https://www.raspberrypi.org/. [Accessed 05 03 2018].
[100] K. Pulli, A. Baksheev, K. Kornyakov and V. Eruhimov, "Realtime Computer Vision with
OpenCV," ACMQueue, vol. 10, no. 4, 2012.
[101] "OpenCV," [Online]. Available: https://opencv.org/. [Accessed 15 03 2018].
[102] H. Hamledari, B. McCabe and S. Davari, "Automated computer vision-based detection of
components of under-construction indoor partitions," Automation in Construction, vol. 74, pp. 78-94,
2017.
[103] D. Philip and D. DeMenthon, "Object recognition in high clutter images using line features," in
ICCV 2005. Tenth IEEE International Conference on. Vol. 2. IEEE, 2005., 2005.
[104] "pyimagesearch," [Online]. Available: https://www.pyimagesearch.com/2015/11/02/watershed-
opencv/. [Accessed 09 05 2018].
[105] N. A. Libre, R. Khoshnazar and M. Shekarchi, "Repeatability, responsiveness and relative cost
analysis of SCC workability test methods," Materials and structures, vol. 45, no. 7, pp. 1087-1100,
2012.
140
[106] "Wikipedia," [Online]. Available: https://en.wikipedia.org/wiki/Sensitivity_index. [Accessed
02 06 2018].
[107] B. Khoshnevis, M. P. Bodiford, K. H. Burks, E. Ethridge, D. Tucker, W. Kim, H. Toutanji and
M. R. Fiske, "Lunar contour crafting—a novel technique for ISRU-based habitat development," in 43rd
AIAA Aerospace Sciences Meeting and Exhibit, Reno, Nevada, 2005.
[108] "Development of a BIM-based automated construction system," Procedia Engineering, vol. 85,
pp. 123-131, 2014.
[109] P. Feng, X. Meng, J. Chen and L. Ye, "Mechanical properties of structures 3D printed with
cementitious powders," Construction and Building Materials, vol. 93, pp. 486-497, 2015.
[110] "Contour Crafting Corporation," [Online]. Available: www.contourcrafting.com. [Accessed 02
04 2018].
[111] "EE Publishers," [Online]. Available: http://www.ee.co.za/article/concrete-towers-lift-wind-
turbines-new-heights.html. [Accessed 02 04 2018].
[112] "Towards A Material Ecology," 2017. [Online]. Available:
https://static.tti.tamu.edu/conferences/tamu-engineering/nsf-3dp-workshop/day1/invited-talks-
3/oxman-leland.pdf. [Accessed 02 04 2018].
[113] W. McGee and M. P. d. Leon, Robotic Fabrication in Architecture, Art and Design 2014,
Springer, Cham, 2014.
[114] "Incremental3D," [Online]. Available: https://www.incremental3d.eu/. [Accessed 04 04 2018].
[115] [Online]. Available: http://www.3ders.org/articles/20171220-scg-3d-prints-3m-tall-triple-s-
concrete-building-with-new-surface-structure-shelter-
concept.html?lipi=urn%3Ali%3Apage%3Ad_flagship3_feed%3Bbh3pDcYLRkWXjR0ZvQ5laA%3D
%3D. [Accessed 04 04 2018].
[116] B. Cazacliu, "In-mixer measurements for describing mixture evolution during concrete
mixing," Chemical Engineering Research and Design, vol. 86, p. 1423–1433, 2008.
Abstract (if available)
Abstract
In this research a framework is proposed for performance-based laboratory testing of fresh printing mixtures. An iterative laboratory testing procedure is described to evaluate the four suggested workability aspects of a printing mixture, namely, print quality, robustness, shape stability and printability timespan. Using different ingredients, four printing mixtures are developed and used to demonstrate and discuss the implementation of the proposed procedure. ❧ The second stage of this research is focused on the real-time quality monitoring of construction-scale 3D printing. Four techniques are proposed and investigated as real-time extrusion quality monitoring measures for Contour Crafting process: electrical resistivity measurements, extrusion pressure measurements, extruder power consumption monitoring, and a vision-based technique. The experimental results are used to discuss and compare the reliability and responsiveness of each technique. Based on the obtained results, the vision-based technique is the most reliable and responsive technique with respect to detecting changes in the water content of a printing mixture.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
A data driven software platform for process automation, planning and inspection of Contour Crafting large-scale robotic 3D printing system
PDF
Contour crafting construction with sulfur concrete
PDF
Thermal and deformation analysis of multiphase sulfur concrete extrusion for planetary construction
PDF
Evaluation of interlayer bonding strength in contour crafted structures: experimental and statistical approach
PDF
Selective Separation Shaping (SSS): large scale cementitious fabrication potentials
PDF
3D printing of polymeric parts using Selective Separation Shaping (SSS)
PDF
A data-driven approach to compressed video quality assessment using just noticeable difference
PDF
Fabrication of ultrasound transducer and 3D-prinitng ultrasonic device
PDF
Multi-scale biomimetic structure fabrication based on immersed surface accumulation
PDF
Novel and efficient schemes for security and privacy issues in smart grids
PDF
Energy control and material deposition methods for fast fabrication with high surface quality in additive manufacturing using photo-polymerization
PDF
Automatic test generation system for software
PDF
Managing functional coupling sequences to reduce complexity and increase modularity in conceptual design
PDF
Towards a cross-layer framework for wearout monitoring and mitigation
PDF
An analytical and experimental study of evolving 3D deformation fields using vision-based approaches
PDF
Vision-based studies for structural health monitoring and condition assesment
PDF
Functional real-time MRI of the upper airway
PDF
Software quality understanding by analysis of abundant data (SQUAAD): towards better understanding of life cycle software qualities
PDF
Studies into vibration-signature-based methods for system identification, damage detection and health monitoring of civil infrastructures
PDF
Real-time channel sounder designs for millimeter-wave and ultra-wideband communications
Asset Metadata
Creator
Kazemian, Ali
(author)
Core Title
Mixture characterization and real-time extrusion quality monitoring for construction-scale 3D printing (Contour Crafting)
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Civil Engineering
Publication Date
10/08/2018
Defense Date
08/27/2018
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
cementitious materials,contour crafting,OAI-PMH Harvest,quality monitoring,robotic construction
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Khoshnevis, Behrokh (
committee chair
), Chen, Yong (
committee member
), Gencturk, Bora (
committee member
)
Creator Email
akazemia@usc.edu,ali.kazemian@contourcrafting.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-75509
Unique identifier
UC11670356
Identifier
etd-KazemianAl-6793.pdf (filename),usctheses-c89-75509 (legacy record id)
Legacy Identifier
etd-KazemianAl-6793.pdf
Dmrecord
75509
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Kazemian, Ali
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
cementitious materials
contour crafting
quality monitoring
robotic construction