University of Southern California Dissertations and Theses

Evaluation of interlayer bonding strength in contour crafted structures: experimental and statistical approach

(USC Thesis Other)

Evaluation of interlayer bonding strength in contour crafted structures: experimental and statistical approach

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

EVALUATION OF INTERLAYER BONDING STRENGTH IN CONTOUR CRAFTED
STRUCTURES:
EXPERIMENTAL AND STATISTICAL APPROACH

by
Babak Zareiyan
_________________________________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(CIVIL ENGINEERING)

December 2017

Copyright 2017 Babak Zareiyan
ii

Dedication

To my Family for their endless love, encouragement, and sacrifices..

iii

Acknowledgments
I would like to express my gratitude and sincere thanks to my adviser Professor Behrokh
Khoshnevis, for his tireless devotion to Contour Crafting and providing me the opportunity to
work on CC.
I would like to acknowledge my members of Dissertation Committee, Dr. Gregg
Brandow, Dr. Vincent Lee, and Dr. Anders Carlson for all their help and encouragement with
this project. This research would not be possible without their contributions and support.
Last, but most important, I am also thankful to my family for inspiration and emotional
support throughout the PhD program. I dedicate this dissertation to them whose encouragement
helped me to reach my goals.

iv

Table of Contents
Dedication ....................................................................................................................................... ii
Acknowledgments.......................................................................................................................... iii
Table of Contents ........................................................................................................................... iv
List of Tables ............................................................................................................................... viii
List of Figures ................................................................................................................................. x
Abstract ........................................................................................................................................ xiii
1. Introduction ............................................................................................................................. 1
1.1. Contour Crafting .............................................................................................................. 1
1.2. Statement of the problem ................................................................................................. 3
1.2.1. Opportunities and limitations of the integration of Contour Crafting in
construction. ............................................................................................................................ 3
1.2.2. Research significance................................................................................................ 6
1.3. Literature review .............................................................................................................. 7
1.3.1. Contour Crafting ....................................................................................................... 7
1.3.2. Bond interface adhesion of concrete layers .............................................................. 8
1.3.3. Mechanical test methods for interface strength in concrete structure ....................... 9
1.3.4. Stability of masonry wall ........................................................................................ 13
2. Effects of aggregate size, extrusion rate, and layer thickness on interlayer adhesion and
strength of structures in Contour Crafting .................................................................................... 16
v

2.1. Effect of aggregate size on failure behavior of concrete under uniaxial compression
test….. ....................................................................................................................................... 16
2.1.1. Concrete Mixture Development .............................................................................. 18
2.1.2. Experiment .............................................................................................................. 20
2.1.3. Results and discussion ............................................................................................ 21
2.2. Effect of aggregate size on early strength development of concrete under uniaxial
compression test and split- cylinder test ................................................................................... 24
2.2.1. Concrete Mixture Development .............................................................................. 25
2.2.2. Experiment .............................................................................................................. 25
2.2.3. Results and discussion ............................................................................................ 26
2.3. Effects of extrusion rate and layer thickness on interlayer adhesion and strength of
structure in Contour Crafting .................................................................................................... 27
2.3.1. Specimen preparation.............................................................................................. 29
2.3.2. Mechanical Test ...................................................................................................... 31
2.3.3. Results and discussion ............................................................................................ 35
2.4. Summary ........................................................................................................................ 37
3. Effects of Interlocking on interlayer adhesion and strength of structures in Contour
Crafting.. ....................................................................................................................................... 40
3.1. Concrete mixture ............................................................................................................ 41
3.2. Specimen preparation ..................................................................................................... 42
3.3. Mechanical test ............................................................................................................... 45
3.4. Results and discussion .................................................................................................... 47
vi

3.5. Summary ........................................................................................................................ 53
4. Effects of mixture ingredients on interlayer adhesion in Contour Crafting .......................... 54
4.1. Introduction .................................................................................................................... 54
4.2. Design of experiment ..................................................................................................... 56
4.2.1. Factorial Design ...................................................................................................... 58
4.2.2. Material properties .................................................................................................. 62
4.2.3. Flow-ability test ...................................................................................................... 68
4.2.4. Specimen Preparation (Interlayer adhesion) ........................................................... 72
4.2.5. Mechanical Test (Interlayer adhesion) .................................................................... 74
4.3. Experimental results ...................................................................................................... 76
4.3.1. Flow-ability ............................................................................................................. 76
4.3.2. Strength-28 days ..................................................................................................... 81
4.4. Statistical model ............................................................................................................. 87
4.4.1. Regression model (Flow-ability) ............................................................................ 88
4.4.2. Diagnostic plots (Flow-ability) ............................................................................... 90
4.4.3. Regression model (Strength-28 days) ..................................................................... 93
4.4.4. Diagnostic plots (Strength-28 days)........................................................................ 95
4.5. Result and discussion ..................................................................................................... 98
4.5.1. Limitations ............................................................................................................ 100
4.5.2. Response optimization, the desirability approach ................................................ 102
4.6. Validation of optimization results ................................................................................ 103
vii

4.7. Summary ...................................................................................................................... 107
5. Conclusion ........................................................................................................................... 109
5.1. Research contribution ................................................................................................... 109
5.2. Research methodology ................................................................................................. 110
5.3. Result summary and applications in practice ............................................................... 111
5.4. Suggested future research............................................................................................. 113
5.4.1. Physical properties of the extruded part ................................................................ 113
5.4.2. Analysis of crack formation and crack propagation in Contour Crafted wall ...... 115
6. Bibliography ........................................................................................................................ 117

viii

List of Tables
Table 1: Sieve analysis for different maximum aggregate size (Cumulative Percentage Passing)
....................................................................................................................................................... 18
Table 2: Major Compounds of the Calcium Sulfoaluminate [42] ................................................ 19
Table 3: The weight and volume of base mixture’s ingredient (maximum aggregate size=1/2") 19
Table 4: Composition of the four types of concrete mixture ........................................................ 20
Table 5: Compression test results for different maximum aggregate size .................................... 21
Table 6: Initial Young modulus of different maximum aggregate size ........................................ 23
Table 7: Primarily test result of 3/16” maximum size aggregate; [47], [43] ................................ 25
Table 8: Compression and split cylinder test results at early age for 3/32” and 3/16” maximum
aggregate size ................................................................................................................................ 26
Table 9: the fabrication rates for different layer thicknesses ........................................................ 30
Table 10: Layering fabrication schedule (1”, 2”, ad 4”) ............................................................... 30
Table 11: Compressive and tensile strength for layers of fresh concrete (1”, 2”, and 4”) ........... 34
Table 12: The weight and volume of concrete mixture’s ingredient ............................................ 42
Table 13: Fabrication rate for 2" height layer ............................................................................... 43
Table 14: Fabrication sequence .................................................................................................... 44
Table 15: Compressive and tensile strength for layers of fresh concrete (0”, 0.25”, 0.5”, and
0.75” interlock) ............................................................................................................................. 47
Table 16: Factor’s operational range affecting interlayer adhesion .............................................. 60
Table 17: Values of different levels of factors .............................................................................. 61
Table 18:Combination of different levels for 36 concrete mixtures ............................................. 62
Table 19: Physical properties of Silica fume ................................................................................ 63
ix

Table 20: Composition of ingredient of MasterLife SF 100......................................................... 64
Table 21: Sieve analysis of the concrete sand .............................................................................. 65
Table 22: Physical properties of the aggregate ............................................................................. 66
Table 23: Physical properties of Superplasticizer ......................................................................... 67
Table 24: Physical properties of Fiber .......................................................................................... 68
Table 25: Fabrication rate for 2" height layer ............................................................................... 72
Table 26: The sequence of fabrication for 2 layers ....................................................................... 73
Table 27: Summary of Flow-ability tests for three different trials ............................................... 77
Table 28: Effect of the mixture on flow-ability of the concrete ................................................... 79
Table 29: Summary of Strength-28 days’ tests for three different trials ...................................... 82
Table 30: effect of the mixture on strength-28 days of the concrete ............................................ 83
Table 31: Estimated coefficient and significance of factors for modeling flow-ability ............... 89
Table 32: Residuals analysis (Flow-ability) ................................................................................. 91
Table 33: Estimated coefficient and significance of factors for modeling strength-28 days ........ 94
Table 34: Residual Analysis (Strength-28 days) .......................................................................... 96
Table 35: Target responses for composite desirability ............................................................... 102
Table 36: Predictions of multiple optimal values and responses ................................................ 103
Table 37: Property of validation mixture .................................................................................... 105
Table 38: Experimental results for the validation of the model .................................................. 106
Table 39: Fitted solution for response optimization ................................................................... 107
Table 40: Design of experiment (DOE) for surface roughness and part accuracy measurement 114

x

List of Figures
Figure 1: Single residence building construction with Contour Crafting [1] ................................. 1
Figure 2: Additive manufacturing process ...................................................................................... 2
Figure 3: Adjustable side trowel to shape different geometry [8] .................................................. 2
Figure 4 : Contour Crafting Technology and Research Specimen; [7] .......................................... 3
Figure 5: Research components ...................................................................................................... 5
Figure 6: Interfacial Zone between old and new concrete [16] ...................................................... 9
Figure 7: Interface bond strength test methods ............................................................................. 10
Figure 8: Plane of failure in specimen with pull-off test method ................................................. 13
Figure 9: Modeling strategies for masonry structures .................................................................. 14
Figure 10: Sieve analysis of the different maximum aggregate size (Cumulative Percentage
Passing) ......................................................................................................................................... 17
Figure 11: Compression test result for 14 days (Blue line) and 28 days (red line) for different
maximum aggregate size............................................................................................................... 22
Figure 12: Surface fracture comparison of different maximum aggregate size ............................ 24
Figure 13: Compression (left) and Split- cylinder (right) test results at early age of concrete
mixture for two different maximum aggregate sizes .................................................................... 27
Figure 14: Cold joint formation in specimen ................................................................................ 28
Figure 15: Effect of time rest and weight of the upper layer on the interlayer adhesion .............. 29
Figure 16: the sequence of fabrication for 3 different specimens ................................................. 31
Figure 17: Direction of applied force on the specimens ............................................................... 32
Figure 18: Strength development of three types of specimens ..................................................... 36
Figure 19: Failure under applied force.......................................................................................... 37
xi

Figure 20: Preliminary study on the size of the interlocking ........................................................ 43
Figure 21: Layering Process and fabrication schedule (2” Interlocking) ..................................... 45
Figure 22: Direction of applied force on the specimens ............................................................... 46
Figure 23: Comparison of bond interface for four types of interlock ........................................... 48
Figure 24: 28 days’ strength development of four types of interlock ........................................... 49
Figure 25: Typical mode of failure in four types of notch ............................................................ 50
Figure 26: Modes of failure for specimen under shear stress [15] ............................................... 51
Figure 27: Failure under applied force.......................................................................................... 52
Figure 28: Design of experiment of the concrete mixture for the optimization of the interface
strength and flow-ability ............................................................................................................... 57
Figure 29: Hierarchy of the process and factors for Design of Experiment ................................. 59
Figure 30: Flow-ability (Work-ability) test methods .................................................................... 70
Figure 31: Flow table apparatus[37] ............................................................................................. 71
Figure 32: Layering Process and fabrication schedule ................................................................. 73
Figure 33: Surface preparation; smoothed with finishing trowel ................................................. 74
Figure 34: Molding and Extrusion Apparatus .............................................................................. 74
Figure 35: Direction of applied force on the specimens (Splitting prism) ................................... 75
Figure 36: Extrudability at different flow-ability level................................................................. 78
Figure 37: Boxplot of flow-ability for three different trial and average ....................................... 78
Figure 38: Pairwise comparison for 95% family-wise confidence level (Flow-ability) ............... 79
Figure 39: Main effect plot for Flow-ability (Fitted mean) .......................................................... 80
Figure 40: Boxplot of strength-28 days for three different trial and average ............................... 83
Figure 41: Pairwise comparison for 95% family-wise confidence level (Strength- 28 days) ...... 84
xii

Figure 42: Main effect plot for Strength- 28 days (Fitted mean) .................................................. 85
Figure 43: Effect of fiber at interface in splitting test ................................................................... 87
Figure 44: Two-way interaction plot (Flow-ability) ..................................................................... 90
Figure 45: Residual plots (Flow-ability) ....................................................................................... 91
Figure 46: Normal Q-Q plot (Flow-ability) .................................................................................. 92
Figure 47: Scale-Location and Residual vs. Leverage plot (Flow-ability) ................................... 93
Figure 48: Two-way interaction plot (Strength-28days) .............................................................. 95
Figure 49: Residual plots (Strength-28 days) ............................................................................... 96
Figure 50: Normal Q-Q plot (Strength-28 days)........................................................................... 97
Figure 51: Scale-Location and Residual vs. Leverage plot (Strength-28 days) ........................... 98
Figure 52: Micro scale analysis of a broken sample under splitting test ...................................... 99
Figure 53: Average flow-ability vs. Average strength-28 days .................................................. 101
Figure 54: Optimization plot ....................................................................................................... 103
Figure 55: Predicted vs. average observed values ...................................................................... 104
Figure 56: Strain- Stress graph of the experimental results for the validation of the model ...... 106

xiii

Abstract
Contour Crafting (CC) is a method that builds concrete structures in a layer-by-layer fabrication.
During construction, layers must be bonded together to make a homogenous structure, as there is
no vibration or external force during layer deposition. In this research, interlayer bonding
strength in contour crafted structure is evaluated by analyzing experimental data and applying
statistical techniques. The interface bond strength is studied based on controlling the properties
of the material in mixture and the impact of mechanical factors (e.g. fabrication parameters,
interlocking). An initial concrete mixture in which the compressive and tensile strength is
significantly improved by modifying the size of aggregate and cement to aggregate ratio is
developed. This mixture is compatible with existing extrusion system. First, the impact of the
fabrication parameters, including extrusion rate, layer thickness, and layer width on the interface
bond strength are analyzed. Secondly, different interlock configurations and sizes are tested to
find the optimal adhesion between layers. Finally, a design of experiment approach is pursued to
investigate the effect of different mixture’s components on the integrity of the extruded part.
Some important factors that affect the mixture’s design are recognized including binder, flow-
ability, and tensile stress. Therefore, full factorial design is used to study these independent
variables and measure their effect on extrudability and strength at interface. Identifying an
appropriate model requires design of experiments, regression modeling, optimization, and
validation methods. The proposed models can balance different variables affecting extrudability
and strength at interface. Moreover, understanding of significant factors in the mixture can
facilitate the material optimization for CC fabrication.

1

1. Introduction
1.1. Contour Crafting
Contour crafting (CC) uses computer controlled layering of material to fabricate more affordable
and sustainable structures by integrating material delivery, workmanship, and installation in one
system. Figure 1 shows the construction of a single-story residence with Contour Crafting by
extruding layers of a special concrete mixture from bottom to top to shape the structure.

Figure 1: Single residence building construction with Contour Crafting [1]
CC is based on deposition of successive layers of material to shape the object [2]. Figure 2 shows
the process which starts with modeling the object that can be of almost any geometry. At second
level, the 3D model was processed and sliced in layers. Finally, the object was fabricated through
laying down material layer by layer.
Rapid mega-scale manufacturing [3], utilizing Building Information Modeling (BIM) in different
phases of construction [4], passive design strategies [5], and sensing automation technology [6]
are possible responses to lower down the cost of construction, improve energy efficiency, and
increase occupants’ satisfaction.
2

Figure 2: Additive manufacturing process
Rapid fabrication and robotic application of internal component installations are advantages of
Contour Crafting. CC uses polymer, ceramic, cement, and a variety of other materials to build
large scale, smoothly finished objects. However, the implementation of Contour Crafting for
building construction has focused on concrete structures due to low cost, availability, and
structural performance.
Layers are extruded subsequently from a nozzle which indicates the point of deposition. Figure 3
shows one type of extrusion unit which carries uncured paste. The trowel control mechanism is
the main parts of the machine. The angle and orientation of the side trowel are adjustable to
shape a complex geometry [7].

Figure 3: Adjustable side trowel to shape different geometry [8]
3

Computer aided manufacturing (CAM) and design (CAD) are integrated in to the building design
and construction industry by connecting design software to the Contour Crafting robot as a rapid
fabrication technique. Figure 4 shows a Contour Crafting machine for concrete processing.

Figure 4 : Contour Crafting Technology and Research Specimen; [7]
Coupling design and construction will optimize the whole process while considering
environmental concerns and economic issues [8]. This promising sustainable construction
method reduces resource consumption, can reuse material, and is economical. Therefore, the
introduction and application of Contour Crafting in construction will positively affect the design
and construction process especially for low income housing, and/or emergency shelters.
1.2. Statement of the problem
1.2.1. Opportunities and limitations of the integration of Contour Crafting in construction
The problem of the bonding in layered manufacturing process is important in various
applications, including extrusion deposition. The mechanical properties of material and strength
of the bond between layers define the structural performance of the system. Even though, bonds
having strength more than the strength of the material is achievable, structural performance of
interface must be investigated. Accordingly, it is proposed to develop an experimental and
4

statistical method to measure the bond strength of layers of Contour Crafted structures. The main
objectives of the proposed research are:
1. Development of an experimental technique to analyze the impact of the following
mechanical factors on the interface bond strength:
1.1. Fabrication parameters, including extrusion rate, layer thickness and layer width [9]
1.2. Alternative interlocking configurations [2]
1.3. Comparison of methods for evaluating the bond strength between layers (compressive,
tensile, and shear stress)
1.4. Analysis of crack formation and crack propagation at interface area
2. Creation and evaluation of concrete mixtures for the CC process to achieve maximum inter-
layer adhesion based on:
2.1. Significant factors affecting inter-layer adhesion (cement binder, fiber, water to cement
ratio, chemical admixture, etc.)
2.2. Effect of different component on extrudability of the mixture [10]
2.3. Compatibility of the mixture to the Contour Crafting process
2.4. Optimization of different factors for strength and extrudability
5

Figure 5: Research components
The prime objective of this study is to understand the structural integrity of a Contour Crafted
wall and characterize the failure behavior at the interface of cementitious composite. Figure 5
shows the research component and pathway toward the proposed research. The problem will be
tackled in three different phases. In the first phase, which focuses on mechanical properties
through compression and tension test, fabrication procedure and interlocking between layers are
tested to achieve the simplest structural configuration. This configuration will be based on
maximum strength at the interface. In the second phase, significant factors in the concrete
mixture that strengthen interlayer adhesion will be identified and optimized, and a suitable
mixture for concrete will be introduced. Once the second phase is completed we will have a
working concrete mixture that yields optimized interlayer strength. This mixture will lay the
Preliminary
Tests on the
configuration of
Interlocking
DESIGN OF EXPERIMENT
(FACTORS IN MIXTURE THAT AFFECT BOND STRENGTH)
DESIGN OF EXPERIMENT
(FACTORS IN MIXTURE THAT AFFECT BOND STRENGTH)
COMPRESSION AND TENSION TEST COMPRESSION AND TENSION TEST
PROPOSED
MIXTURE AND
FABRICATION
PROCEDURE
Fabrication
Procedure
Interlocked
Layers
Workability
Compressive
Strength
Concrete
Mixture
Compatibility
with Extrusion
System
INITIAL CONCRETE MIXTURE INITIAL CONCRETE MIXTURE
Sieve Analysis
Aggregate to
Cement ratio
Aggregate
Early Strength
development
Grading (Particle
size distribution)
Tensile Strength
Workability
Binder
Cementitious Material:
- Portland Cement (II)
- Silica Fume
Water to Cement ratio
Chemical Admixture
- Superplasticizer
Fiber
Interface
Strength
Flowability
OUTPUT OUTPUT INPUT INPUT
Confirm
Compatibility
with Extrusion
System
Model
Optimization
Mathematical
Model
Validation
6

groundwork for tailoring the optimized concrete mixture and analyzing the properties of the
extruded part (e.g. roughness and accuracy).
1.2.2. Research significance
There is a great interest in enhancing the structural integrity of the 3D printed (Contour Crafted
in this research) part due to increasing interest in additive manufacturing and rapid prototyping in
construction. Despite several attempts in research and experimentation to enhance the
homogeneity of the CC structure, a lower-than-desired level of homogeneity persists. This shows
through the failure of the structure due to formation of cracks at the interface and debonding of
layers. A possible solution to this problem can be top surface preparation of the substrate,
application of different binders, and better managing of the curing process.
While bond tests have been developed for old to new concrete application, there is no
comprehensive study for measuring the development of the bond strength between concrete
layers at an early age. This research intends to develop structural analysis tools for fresh concrete
structures with Contour Crafting. The objective of the research is to use experimental and
statistical techniques to describe the elements of fresh concrete related to CC. Experimental
methods are elaborated for simple structures, and the results will be advanced to understand the
performance of more complex structures. Further application and analytical methods such as
finite element analysis can also be applied to understand more complex structures.
The bond strengths measured by different test methods are compared in this report, and proper
mixes is developed and validated. The applicability of different mixtures to the Contour Crafting
robot must be examined because of:
• The sensitivity of the CC robot (delivery and extrusion system) to the mixture.
7

• The structure is subject to loading during fabrication before the concrete has developed
adequate strength.
The result of this research, which is based on material selection and the fabrication regime, will
be used to introduce a methodology to enhance the bond interface of cementitious material
subject to CC fabrication, a detailed understanding of the application of CC for homogenous and
sustainable structures, and a framework for understanding the critical aspect of building design
and construction that relates to the integration of additive manufacturing in construction. The test
results will be in the interest of designers and engineers involved in developing additive
manufacturing.
1.3. Literature review
1.3.1. Contour Crafting
In 1999, Richard J. Russell II [11] completed his PhD dissertation on analyzing polystyrene melt
flow using Contour Crafting through experimental approach. In 2002, Hongkyu Kwon [12]
continued the research on CC by using experimental approach for uncured ceramic material. He
investigated the effect of side trowel on the capability and quality of the Contour Crafting
process to fabricate 2.5D and 3D parts, and he concluded that the surface quality of the extruded
part with trowels on two sides was better than a single side trowel on the exterior angle. Kwon
simulated the pattern of flow in the CC nozzle during fabrication processes with CFD software to
study the effect of the pressure on the extrusion and geometry.
In 2005, Dooli Hwang [13] used experimental approach to study the application of Contour
Crafting on a full scale concrete wall. His research showed that designing the setting time
depends on time of deposition cycle, material delivery, CC machine preparation, and fabrication
8

rate because decreasing the setting time by using rapid set cement risks mortar hardening inside
the CC machine. He also added Bentonite (A12O3-5SiO2-7H2O), plastic clay, to the mix to
increase the paste plasticity and decrease the water seepage.
In 2012, Tony Di Carlo [14] applied experimental and numerical techniques to analyze the
structural properties of fresh concrete subject to Contour Crafting. He developed a special mortar
mixture which can be safely used for layered fabrication. His proposed cementitious mixture was
suitable for freeform-layered fabrication and was tested for a full-scale demonstration. Di Carlo
also studied the structural properties of fresh concrete for safe layering by developing analytical
and numerical tools.
1.3.2. Bond interface adhesion of concrete layers
The strength at the interface of two layers of fresh concrete is severely understudied since there
was not any practical application of the phenomenon before Contour Crafting. Comparison can
be made to studies about adhesion of new and old concrete, and the formation of cold joint
during the construction of concrete structure.
The connection in the boundary of two layers of extruded material is defined as “adhesion”.
Mechanical interaction, chemical bonding, and thermodynamic linkage are three major adhesion
characteristics [15]. Although material characteristics and environmental conditions can affect
the bond, interlayer adhesion models can be introduced as the interaction of the layers in both the
micro and macro scale. In the micro scale, chemical forces and in macro scale, layer
configurations (e.g. interlocking and surface roughness) are responsible for bonding between
layers.
9

Figure 6: Interfacial Zone between old and new concrete [16]
Interface strength in concrete structures is the required stress to separate the bond between two
layers of concrete. Mechanical and chemical compatibility of the material and its properties (e.g.
plastic or elastic deformation) influence the strength at the interface of two components [17].
However, opinions change from one condition to another among researchers. Figure 6 shows the
three-layer interface between old and new concrete material as presented by Xie and Xiong [16].
The penetration layer, which is the first layer, is the result of the formation of Calcium Silica
Hydrate (C-S-H) and a small amount of Ca(OH)2, and the second layer is mainly Ca(OH)2 which
is the weakest layer. The third layer has almost the same structure as new concrete.
The bond between old and new concrete usually shows a weak link, and the characteristics of
different effects of that have been investigated widely in the field of repair (e.g. highway
structures), and enhancing the structural performance of an existing structure.
1.3.3. Mechanical test methods for interface strength in concrete structure
Shear, tension, and torsion tests are common test methods for interface strength. Several tests
have been developed to measure the bond strength of new and old concrete, and test results may
vary substantially based on loading rate, specimen size, experiment set-up, etc. [18]. However,
the applicability of these tests to Contour Crafting and the difference between the results of
10

methods have not been studied yet. Momayez [19] compared four test methods to measure the
bond strength between old and new concrete. He found out that the bond strength from some
tests is eight times larger than others, and pull-off and splitting test provides the most
conservative results.

Figure 7: Interface bond strength test methods
Figure 7 shows the existing test methods for measuring the bond strength between layers of old
and new concrete. The most common test method for the bond strength is the pull-off test
(Figure 7A). Delatte et al. [20] indicated that the result of a pull-off test is highly dependent on
eccentricity of applied loads, and result can be largely scattered. Moreover, the tensile stress
strength of the layer should be more than the bond strength otherwise the specimen will break at
layers and interlayer adhesion will not be quantified.
11

The splitting test of concrete (Figure 7B) was first introduced by Akazawa in Japan [21], and
now has been adopted as a standard test ASTM C496 [22] as a way of measuring indirect
tension. Ramey and Strickland [23] developed a splitting test to measure the bond strength
between old concrete and repair material. They used the same equation as that of a homogenous
cylinder to calculate the tensile strength of the cylinders made of two layers of old and new
concrete, and obtained the consistent results. Figure 7B shows that in the splitting test, the
longitudinal compression load is applied on the cross section of the prism sample and results in
the tension at the interface, which split the sample into two parts.
The wedge splitting test, which uses a 4”x4”x4” cube with a notch on the top and interface in the
middle, was introduced by Tschegg [17]. Groove and the starter notches were designed during
the fabrication process. Figure 7C shows the principle of wedge splitting concept. Force from
testing machine was applied on the center line of the wedge load, starter notch, and support area
which are all in a vertical line. Horizontal force causes splitting of the specimen, which can be
calculated based on a wedge of approximately 5° to 10° and the cube is split by loading the
wedge at the interface of two layers. This test measures the interface strength by fracture
mechanics parameters (tensile interface strength and fracture energy).
The slant shear method (Figure 7D) measures the bond strength for shear and compression
combined. Slant shear was proposed by Kreigh for the bond strength of resinous repair material
and old concrete [24]. Further modification and researches on Kreigh’s test were performed
using a prism with a cross-section that was one third of the length used by Wall and Shrive [25],
and also using the test for cementitious material instead of resinous by Abu-Tair [26]. The
specimen was loaded in a standard compression test machine with unrealistic loading conditions,
which affected the experiment to report results were generally higher than that of other tests.
12

The torsion bond strength test was introduced by Silfwerband (Figure 7E) and can be applied to
measure the shear strength at interfaces [27]. Figure 7F shows the shear strength measurement by
applying forces parallel to interface. There are different modified shear strength tests but the
majority of them have difficulties in load alignment and the results are scattered. For example, in
Figure 7F-3, the ratio of the interface to the length of the specimen has to be designed in such a
way that cracks propagate at interface, or in Figure 7F-2 the result shows the strength at two
interfaces. In the case of the concrete repair material, it is not practical [28], but can resemble the
Contour Crafted wall.
Mechanical adhesion in tension determined by transverse anchorage in pores and voids differs
from mechanical adhesion in shear[15]. The choice of bond strength test method is crucial, and
different test methods cause distinct interface stress conditions in specimens, therefore, the
appropriate test is the one for which the nature of loading is the most like that of stress conditions
of the actual structure. Preliminary experiments were performed in this research based on the
pull-off test method. For pull-off tests the force must be applied perpendicular to the interface,
and misalignment causes large scattering in results. For the majority of trials, there was a mixed
mode of failure, and failure occurred in both layers and interfaces instead of just interfaces.
Figure 8 illustrate failure mode in specimens tested with the pull-off test method. Though this
can be interpreted to show that there is higher strength at interfaces, such an interpretation within
the context of this research makes the comparison between results impractical. Therefore,
splitting prism test was used to quantify the tensile strength at interfaces.
13

Figure 8: Plane of failure in specimen with pull-off test method
1.3.4. Stability of masonry wall
Adobe buildings have the highest risk of suffering earthquake damage among all building types.
However, the availability of materials, thermal insulation characteristics, and minimal
construction skill required make adobe structures one of the most popular choices globally since
historic times.
Similarity between layers of brick (or block) units and mortar with layered manufacturing makes
the analysis of these structure somehow useful for understanding the CC structures. Three
significant factors on performance of adobe structure are:
1. The shape of the adobe structure is important. Symmetrical form like circular adobe
houses has shown great performance without any additional ductile reinforcement [29].
2. Height to thickness ratio has direct relationship with seismic stability of the structure [30]
3. The continuity in element and minor restrains can improve the performance of adobe
structure.
When working with a purely adobe structural system, the influential factors are adobe units,
mortar, and the interaction in between. In such a structural system, modeling the adobe walls can
be done using two finite element methods of macro- and micro-modeling as shown in Figure 9.
14

The choice of using which one of these methods depends mostly on the desired accuracy and
simplicity for the model.

Figure 9: Modeling strategies for masonry structures
(a) masonry sample; (b) detailed micro-modeling; (c) simplified micro-modeling; (d) macro
modeling [31].
In macro-modeling, the infill is modeled by a single or multiple compressive diagonal strut. The
advantage of this model is that it accounts for the global behavior of a structural system without
modeling all the components and all possible failure modes. In this approach, the focus is towards
the physical understanding of the system’s behavior, where a small number of elements are
representative of overall behavior. Polliakov [32] was the first to suggest this strategy for infill
wall modeling. The method was then improved by Holmes through his suggestion of using an
equivalent pin-jointed diagonal strut with the same material and thickness of the infill [33]. The
behavior of the infill was then studied by many researchers based on changing the number of struts
[34], material properties, and inelastic behavior of the strut [35].
In micro-modeling, approach of the structure is divided into different types of elements: brick,
mortar, boundary of brick and mortar, boundary of adobe panel and frame, and the frame elements.
15

Finite element method (FEM) or discrete element method (DEM) are numerical methods used to
model the surrounding frame and the infill components. There are two different methods of micro-
modeling with different degrees of precision: detailed micro-modeling and simplified micro-
modeling. In detailed micro-modeling, solid continuum elements are used to model bricks, units,
and mortar, while the brick-mortar interfaces are modeled by discontinuous interface elements
(planes of failure and slippage) [31]. Using this configuration, the structural behavior and possible
failure mechanisms can be evaluated in detail. To thoroughly understand the failure mechanism of
adobe construction, one may find that micro-modeling is the best method, since it models bricks
and mortar separately.

16

2. Effects of aggregate size, extrusion rate, and layer thickness on interlayer
adhesion and strength of structures in Contour Crafting
This chapter outlines a research methodology aimed at the development of a framework for
understanding the critical aspects of interlayer adhesion in Contour Crafting. A series of
experimental case studies regarding Contour Crafting and concrete are introduced. The main
objective of the proposed research is the development of an experimental technique to analyze
the impact of the fabrication parameters, including extrusion rate, layer thickness, and layer
width on the interface bond strength, crack formation, and crack propagation at interface areas
[9].
1

2.1. Effect of aggregate size on failure behavior of concrete under uniaxial
compression test
A concrete mix consists of various aggregate sizes, cementitious material, and water. The
concrete strength is a factor of the amount of water to amount of cementitious material,
aggregate to cement ratio, grading, shape, and size of the aggregates, and the bond between
binder and aggregate [36]. Fine and coarse aggregates displace a large portion of the mixture’s
volume and the maximum size of the aggregate, the gradation, and the fineness affect the
strength of the mix.
In this report, the effects of the size of a rigid surface aggregate and aggregate to cement ratio on
the mechanical performance of the concrete composite were experimentally studied and a proper
mixture which was compatible with extrusion system used was introduced. Desired workability
was calculated using a flow table, ASTM C 230[37]. Superplasticizer (ADVA Cast600 from

1
This chapter is published in Automation in Construction journal.
17

GRACE at the range of 3 to 6 fl oz./100 lbs. of cementitious material) and Viscosity Modifying
Admixtures (MasterMatrix-VMA450 from BASF at the range of 0.5 to 4 fl oz./ 100 lbs. of
cementitious material) were added to achieve constant workability and shape stability.
The main difference in mixtures is the maximum size of the aggregate, which are 3/32”, 3/16”,
1/4”, and 1/2”. Based on ASTM C33, Standard Specification for Concrete Aggregate, the
distribution of different aggregate sizes decreases the void between the aggregate particles and
provides a homogenous mix [38]. Figure 10 shows a smooth gradation curve as the result of
different particle sizes (base case is the mixture with maximum aggregate size of 1/2”).

Figure 10: Sieve analysis of the different maximum aggregate size (Cumulative Percentage
Passing)
Table 1 shows the sieve analysis for different aggregate compositions. To develop a mix,
knowing the composition of standard concrete, the distribution of particle sizes was modified to
make the mix suitable for the extrusion system while providing the desirable strength.

18

Table 1: Sieve analysis for different maximum aggregate size (Cumulative Percentage Passing)

Cumulative percentage passing
Maximum Aggregate Size
Sieve Size 1/2" 1/4" 3/16" 3/32"
No.200 4 10 14 24
No.100 10 18 21 35
No.50 25 28 38 49
No.30 36 42 53 73
No.16 49 58 69 89
No.8 61 73 85 100
No.4 74 88 100 100
0 . 2 5 ” 81 100 100 100
0 . 3 7 5 ” 98 100 100 100
0 . 5 ” 100 100 100 100
2.1.1. Concrete Mixture Development
Control mortar mixes having 2000 psi compressive strength at 28 days were prepared (2000 psi
was selected based on previous research on CC [14]). The mortar was mixed based on ASTM C
305, Mechanical Mixing of Hydraulic Cement Pastes and Mortars of Plastic Consistency [39].
The binder was composed of 60% Calcium Sulfoaluminate (CSA) and 40% type I Ordinary
Portland Cement (OPC) by mass. CSA cement works as a catalyst in hardening process of
concrete; therefore, the mixture of OPC and CSA shows higher early strength development [40].
Table 2 shows the major compounds of CSA from Buzzi Unicem USA Inc. Calcium oxide,
aluminum, sulfur, silica oxide, and iron comprise the majority of CSA. The main advantages of
using CSA for Contour Crafting are:
- Rapid strength gain: Rapid strength gain is critical in Contour Crafting since the process
is based on the deposition of layers of concrete on top of each other in short amounts of time.
- Lower shrinkage: First, the concrete strength increases more rapidly than the concrete
shrinkage stress, and second, CSA requires about 50% more water than OPC for hydration.
Therefore, most of the water in the mixture is used for hydration and hence less water remains to
cause excessive shrinkage [41].
19

Table 2: Major Compounds of the Calcium Sulfoaluminate [42]
Major Compounds
CaO 40% to 50%
Al2O3 15% to 25%
SO3 15% to 25%
SiO2 5% to 10%
Fe2O3 1% to 5%
Natural sand river of 1/2" maximum size and tap water were used to prepare the mortar. Table 3
illustrates the weight and volume of the ingredients in the base mixture.
Table 3: The weight and volume of base mixture’s ingredient (maximum aggregate size=1/2")
Batch Weight (lb.) Specific Gravity Absolute Volume (ft
3
)
CSA 6.00

Cement (Type I) 4.00

Total Cementitious Material 10.00 3.25 0.05
Aggregate 27.50 2.61 0.17
Water (W/C=0.50) 5.00 1.00 0.08
Total 42.50

0.30
Density of water 62.27 lbs./ft
3
at 73.4°F
Superplasticizer and Viscosity Modifying Admixtures were added to achieve desired workability and shape
stability.
In this experiment, the main factors are the maximum size of the aggregate (3/32”, 3/16”, 1/4",
and 1/2") and aggregate to cement ratio while the aggregate type remains the same through all
trials. Table 4 shows the weight of different materials in four batches. In different batches,
portions of the cement material increase when the mixture is prepared based on smaller size
aggregates (Larger amount of cement in the mixture with smaller size aggregate is attributed to
particle distribution). Numerous trial mixtures were created to find the suitable range for
aggregate to cement ratio and its compatibility with the existing Contour Crafting nozzle until a
satisfactory efficiency of the extruded part was achieved.

20

Table 4: Composition of the four types of concrete mixture
Mixture ID
Maximum
aggregate size (In.)
Cement
(lb.)
Aggregate
(lb.)
Water/Cement
ratio
Total aggregate and
binder (lb.)
M1 3/32 17.5 20 0.5 37.5
M2 3/16 15 22.5 0.5 37.5
M3 1/4 12.5 25 0.5 37.5
M4 (base case) 1/2 10 27.5 0.5 37.5
Superplasticizer and Viscosity Modifying Admixtures were added to achieve desired workability and shape
stability.
2.1.2. Experiment
Six specimens of each concrete mixture were prepared. All specimens were cast in 4×8-inch
cylinder molds and were kept in room temperature for 24 hours. The molds were stripped off
afterwards, and the samples were kept in water up to the testing time. Prior to uniaxial
compression testing, the specimens were capped and loaded at a rate of 25 to 40 psi/sec based on
ASTM C 39[43] using a 100,000 lbs. capacity hydraulic compression machine.
All specimens were subject to axial compression stress until failure. The results, which include
24 compression tests (12 for 14 days and 12 for 28 days), and average values of recorded peak
loads of the three specimens are summarized in Table 5. Due to preliminary defects in the
sample preparation, 8% of the total number of specimens was rejected during the test.
Table 5 indicates that an increase in compression strength results from lower aggregate sizes
with higher cement to aggregate ratio. The maximum difference between compression strengths
of base case and smallest aggregate size is about 118% after 14 days and 104% after 28 days.

21

Table 5: Compression test results for different maximum aggregate size
Aggregate
Maximum
Size
Test Age
(14 Days) (28 Days)
1 2 3 Average (psi) COV 1 2 3 Average (psi) COV
3/32 ” 3552 3651 3783 3662 3.2% 4023 3855 3954 3944 2.1%
3/16 ” 2930 3166 3102 3066 4.0% 3788 3921 3685 3798 3.1%
1/4" 2706 2945 2773 2808 4.4% 2754 2827 2912 2831 2.8%
1/2" 1751 1500 1786 1679 9.3% 1971 1828 2012 1937 5.0%
2.1.3. Results and discussion
According to different researchers the effect of aggregate size on mechanical properties of
concrete has not been fully measured [44], but understanding the effect of the aggregate size has
become more important in Contour Crafting by considering its influence on the extrusion system,
the surface quality of the extruded part, and the strength of the concrete in CC wall.
Developing higher strength concrete is important because of its effect on bond interface.
Beushausen; M. G. Alexander [15] investigated the influence of the new concrete compressive
strength on shear bond strength between old and new concrete, and they found a constant
correlation between shear bond strength and compressive strength. Neville [36] also mentioned
the increase in tensile strength was the result of increase in compressive strength, but at slower
rate than shear. Thus, developing higher strength concrete concluded in better interlayer
adhesion.
Different strength results have been observed among researchers in terms of maximum aggregate
size in concrete. Hilleberg [45] investigated the fracture energy in mortar and concluded that
fracture energy in concrete was more than that of mortar. Moreover, the fracture energy in
concrete had a direct relationship with aggregate size. However, large scattering of results
increased the uncertainty.
22

The average compression strength as a function of the aggregate size and aggregate to cement
ratio obtained from the experimental results is shown in Figure 11. The results could be
categorized into two groups: one for mixes with maximum aggregate sizes of 3/32” and 3/16”
(which are compatible with the existing CC machine), and the other for maximum aggregate
sizes from 1/4" to 1/2". Figure 11 also shows that the coefficient of correlation is almost constant
for two regression lines at different ages. Interestingly, there are significant differences in terms
of compressive strength. Overall, compression test results show that smaller size aggregates with
higher cement to aggregate ratio yield higher compressive strength. The observed increase in
compression strength can be due to a decrease in aggregate volume relative to the total
composite volume, and less so due to the difference between the elastic moduli of the two phases
[9].

Figure 11: Compression test result for 14 days (Blue line) and 28 days (red line) for different
maximum aggregate size
The effect of the maximum aggregate size on moduli of elasticity is shown in Table 6. The initial
Young’s modulus was calculated as a tangent of the linear part of the stress-strain graph. The
23

experimental results show an increase of 85 percent in 28 days from maximum aggregate sizes of
1/2" to 3/32”. Elices and Rocco [44] also showed in their experiments that the modulus of
elasticity decreases as the aggregate size increases.
Table 6: Initial Young modulus of different maximum aggregate size

Test Age
Aggregate Maximum Size (14 Days) (28 days)
3/32” 191700 214300
3/16” 163200 194700
1/4" 132500 144400
1/2" 104300 115800

Fracture surface is another phenomenon impacted by aggregate type. In test specimens, failure
under uniaxial compression happens by the debonding of the cementitious mix from aggregate
particles. These micro-cracks form at the interface of aggregate and binder, and the slopes of the
stress-strain curves indicate their formation under uniaxial compression tests. Therefore, crack
propagates around the aggregate rather than through it. Figure 12 shows the average crack
propagation in specimens during the test. Increase in fracture toughness was observed with an
increase in aggregate size. In case of 1/2", large number of cracks that propagate parallel to
loading axis were spotted. In 1/4", cracks were expanding through the specimens, and 3/16”
involved large number of microcracks in contact with platen. In 3/32”, gradual softening of the
specimens resulted in smoothest fracture surface and smaller (width, depth, and length) cracks.
24

Figure 12: Surface fracture comparison of different maximum aggregate size
2.2. Effect of aggregate size on early strength development of concrete under
uniaxial compression test and split- cylinder test
The structural properties of fresh concrete subject to uniaxial compression loads and split
cylinder tests minutes after mixing is investigated in this part of the report. In Contour Crafting,
it is essential for the concrete to develop strength in a minimal amount of time such that the first
layer can sustain the weight of the layers on the top since successive layers will be applied
minutes after the previous layer’s deposition. Thus, using high early strength concrete is an
approach to achieve adequate strength in a few minutes.
There are some uncertainties on the optimum required time in terms of early strength
development because of the lack of research in this area. The current Contour Crafting robot
specifies a requirement for concrete to reach a minimum compressive strength of 100 psi in 30
minutes; Therefore, the test was performed in three different periods up to 70 minutes after
mixing to understand the strength development.
25

2.2.1. Concrete Mixture Development
Concrete mixtures were based on the experiment presented in the previous section. 3/32” and
3/16” maximum aggregate sizes were selected because of their higher strength and compatibility
with the extrusion system used. Table 4 shows the weight of different materials in two batches.
2.2.2. Experiment
36 samples (three samples for each trial) were constructed. Half of the samples were tested under
uniaxial compression tests, and the other half were tested under split cylinder tests.
Table 7 shows the preliminary results for 3/16” maximum aggregate size. Gillmore Needles [46]
tests were performed on mixtures to measure the initial setting time. Wet density was measured
immediately after mixing using a standard container and with the same consolidation regimen
that has been implemented during fabrication of the test specimen.
Table 7: Primarily test result of 3/16” maximum size aggregate; [47], [43]
Maximum Aggregate Size 3/16"
Setting Time (ASTM C191) 45 mins
Wet Density (Wet Mix) 140 lbs./cu ft.
2 hrs. Compressive strength (ASTM C 39) 400 psi
24 hrs. Compressive strength (ASTM C 39) 900 psi
7 Day Compressive strength (ASTM C 39) 2400 psi
14 Day Compressive strength (ASTM C 39) 3066 psi
28 Day Compressive strength (ASTM C 39) 3800 psi

The model is loaded incrementally, with the load applied at a rate of 5 to 10 psi/sec using
100,000 lbs. capacity hydraulic compression machine. Each test result is an average of the three
specimens. The results can be seen in Table 8. The first unconfined compression tests were
performed 14 minutes after mixing. The mold was modified to allow the extraction of the barely
set cylinder without damaging it. Additional data points at 29 and 42 minutes after mixing were
measured, note that strength development over time is not usually linear. The table indicates
26

higher early compression and tension strength at early ages in smaller aggregate sizes with lower
aggregate to cement ratio.
Table 8: Compression and split cylinder test results at early age for 3/32” and 3/16” maximum
aggregate size
Maximum
Aggregate Size
Specimen
Compression (psi) Split Cylinder (psi)
16 Mins 29 Mins 42 Mins 24 Mins 34 Mins 47 Mins
3 /3 2 ”
1 157 415 497 33 111 123
2 184 396 456 28 96 108
3 178 413 472 32 96 111
Average 173 408 475 31 101 114
COV 8.2% 2.6% 4.4% 8.5% 8.6% 7.0%
16 Mins 44 Mins 69 Min 18 Mins 40 Mins 62 Mins
3 /1 6 ”
1 47 178 308 21 39 66
2 52 209 287 18 43 58
3 45 204 290 21 38 65
Average 48 197 295 20 40 63
COV 7.5% 8.4% 3.9% 8.7% 6.6% 6.9%
2.2.3. Results and discussion
Since in the past there has not been a fabrication process like Contour Crafting, there is little
research regarding load bearing development of concrete in early age. Understanding fresh
concrete maturity is the central piece for investigating and calibrating interlayer adhesion, and
identifying the possible fabrication rate of the structure. Figure 13 shows the evolution of
concrete mixtures over time at three different mortar ages. In the figure, blue points depict the
result of unconfined compression tests and split cylinder tests for concrete mixes with maximum
aggregate size of 3/32”, and red points are for mixes with maximum aggregate size of 3/16”.
Split cylinder tests show that the tensile strength of the specimens increases very gradually,
while the compression strength increases more rapidly. Figure 13 shows that compressive
strength increases linearly with a coefficient of correlation of 0.91 for ages up to 70 minutes for
aggregate sizes of 3/32”. The results of this experiment also indicate that the strength of the
27

3/16” maximum size aggregate develops linearly until 70 minutes after mixing for both
compression (R
2
=0.99) tests and split cylinder tests (R
2
=0.99).

Figure 13: Compression (left) and Split- cylinder (right) test results at early age of concrete
mixture for two different maximum aggregate sizes
Experimental results show a higher tensile and compressive strength in mixtures with smaller
sized aggregate and higher cement to aggregate ratio. Elices and Rocco [44] investigated the
effect of aggregate size on tensile stress of the concrete and they found the same trend for tensile
strength, but their results for compression tests were different.
For maximum aggregate size of 3/32”, fail strain and Young’s modulus as a function of time
indicates linear increase with very low slope and then exponential evolution in material response
up to 7 days after mixing with maximum of 214,300 in 28 days.
2.3. Effects of extrusion rate and layer thickness on interlayer adhesion and
strength of structure in Contour Crafting
In Contour Crafting, the porosity and permeability increase at interfaces, and the bond between
old and new concrete layers usually presents a weak link in the structure; thus, the quality of
interlayer adhesion must be evaluated through mechanical and chemical tests. Several tests are
28

available to measure the bond strength between old and new concrete and it must be picked
carefully for CC since the test method has great influence on the measured bond strength [19].
The time lapse between layer depositions in a Contour Crafted structure has a great influence on
its structural properties. A poorly controlled time lapse may result in cold joints at the interface
of layers and reduce the strength of the structure (Figure 14). To avoid a plane of discontinuity
and eliminate long delays, the optimum lapse time between successive layer depositions should
be determined and new layers must be added after the previous layer is sufficiently hardened and
before it is over-cured. Therefore, the fabrication speed must be designed to allow layers to bond
together while they simultaneously gain enough strength to sustain their own weight as well as
the weight of any layers above them.

Figure 14: Cold joint formation in specimen
The thickness of the layer and amount of concrete that can be placed in one layer depend on the
properties of the concrete mixture, the design of the nozzle, and speed of the fabrication process.
An optimized extrusion system places concrete in a constant rate and does not affect the interface
of the layers. Figure 15 shows factors that have been studied in this section. This section
compares the strength of Contour Crafted structure for different fabrication regimes to quantify
the weakness at the interface region due to timing sequences. The concrete mixture was designed
based on a maximum aggregate size of 3/16” (Table 4 and Table 7).
29

Figure 15: Effect of time rest and weight of the upper layer on the interlayer adhesion
2.3.1. Specimen preparation
Sample blocks with dimensions of 4” × 4” ×4” (composed of one 4” block, two 2” layers, and
four 1” layers) were fabricated to investigate three parameters affecting the strength at interfaces
of fresh layers of concrete. Parameters were uniaxial compression tests, uniaxial compression
tests parallel to the interface, and split prism tests.
Fabrication time was selected through several preliminary experiments based on material
strength development and setting time (Table 7). Hwang’s study on fast setting concrete
concluded that initial setting time of the mixture should be more than one complete layer
deposition cycle [13].
The time interval between subsequent layers must be designed in a way such that the lower layer
supports the upper layer, which cannot be too short, and also be able to adhere to each other,
which cannot be too long. [14]. These initial experiments proved the compatibility of the
material with the extrusion system, since decreasing the setting time can risk mortar hardening
inside the CC machine. This issue has a direct relationship with nozzle idling time, which can
also cause mortar solidification and clog the extrusion system.
Deposition rate of material in inches of concrete per hour defines the CC construction speed and
how fast a concrete wall can be fabricated. The target fabrication rate is to erect a ten-foot tall
wall in 12 hours. This means stacking rates of 10 inches per hour. Based on an initial setting time
of 15-20 minutes, Table 9 shows the calculations for fabrication of different layer thicknesses.
30

Table 9: the fabrication rates for different layer thicknesses
Total Height of the Specimen H = 4 in
Wall height 10 ft. = 120 in
Total fabrication time 12 hr. = 720 min
Fabrication rate of 720 min/120 in = 6 min/in
Height of a single layer h 1 = 1 in
Interval between layers (Int.) t 1 = 6 min
Height of a single layer h 2 = 2 in
Interval between layers t 2 = 12 min

The sizes of all samples are 4” ×4” ×4”, but they were fabricated in 3 different layering
processes (1-inch, 2-inch, and 4-inch). Table 10 and Figure 16 show the sequence of fabrication
for 3 different categories and the extrusion system for 1 inch, 2 inches, and 4-inch layer
specimens. A height of 4 inches was used as the base case, 2-inch height layers were casted in
12-minute intervals, and 1-inch layers were deposited every 6 minutes. The mixture is poured
inside the first layer immediately after mixing. For instance, for 2-inch layer samples concrete
was placed in at time = 00:5:30. After 12 minutes at time = 00:17:30 the second layer was added,
and after 24 minutes at time = 00:29:30 the mold was removed. Each layer gains strength to
support its own weight and after 12 minutes to support the weight of the layers above.
Table 10: Layering fabrication schedule (1”, 2”, ad 4”)
1 Inch Height
Layer No. Mixing Start Mixing Stop Layer Placing Start Layer Placing End
1 0:00:00 0:04:00 0:04:00 0:05:00
2 0:06:00 0:10:00 0:10:00 0:11:00
3 0:12:00 0:16:00 0:16:00 0:17:00
4 0:18:00 0:22:00 0:22:00 0:23:00
2 Inch Height
Layer No. Mixing Start Mixing Stop Layer Placing Start Layer Placing End
1 0:00:00 0:04:00 0:04:00 0:05:30
2 0:12:00 0:16:00 0:16:00 0:17:30
4 Inch Height
1 0:00:00 0:04:00 0:04:00 0:06:30

31

To avoid the bonding between concrete and the inner surface of the mold oil was applied to the
surface of the mold before pouring in the concrete. Any extra particles were removed from the
top surface of the layers and interfaces of the layer were flattened by a finishing trowel to
provide a consistent surface roughness among all the specimens. The specimens were placed
inside water after 24 hours and were kept there until the test time.

Figure 16: the sequence of fabrication for 3 different specimens
2.3.2. Mechanical Test
Compression and split tensile tests were performed to quantify the homogeneity of the specimens
after 75 minutes and after 3, 7, and 28 days. This section includes the results of testing 108
samples. Figure 17 shows the size of the specimen, height of the layers, and direction of the
applied force. Forces were applied in 3 different ways: uniaxial compression, uniaxial
compression in direction of the layer interfaces (interlayer adhesion), and splitting prism test. For
the splitting prism test a square cross section prism (4” ×4” ×4” cube) was placed under
32

longitudinal compressive loading, and tension stress splits the sample into two parts along upper
and lower axes of loading.

Figure 17: Direction of applied force on the specimens
(1. Uniaxial Compression test; 2. Uniaxial compression test (interlayer adhesion); 3. Splitting
prism)
All 108 specimens were loaded statically at the rate of 20 to 30 psi/sec using 100,000 lbs.
capacity hydraulic compression machine. Table 11 gives the test results for each sample,
including the age of the sample, the average strength, and the coefficient of variation (COV).
COV’s vary from 2.4% to 6.5% for uniaxial compression tests, 2.1% to 9% for compression tests
(interlayer adhesion), and 6.4% to 9.8% for splitting prism. The ranges of these coefficients of
variations are reasonable.
33

The strengths for compression tests were calculated by dividing the maximum load by the area
(16 in
2
). The splitting prism strength was calculated by Ϭ= 2P/ πA, where Ϭ = splitting prism
strength (psi), p = applied force (lbf), and A = area of the interface (16 in
2
). For samples that
were casted in one stage there was no interface plane, therefore, those samples were treated as a
base case using the same formula.

34

Table 11: Compressive and tensile strength for layers of fresh concrete (1”, 2”, and 4”)
Compression Test (psi)
Layer Specimen (75 Min) (3 Days) (7 Days) (28 Days)
1-inch
1 316 1836 2398 4784
2 300 1749 2306 4573
3 341 1746 2298 4611
Average 319 1777 2334 4656
COV 6.5% 2.9% 2.4% 2.4%
2-inch
1 303 1392 2201 4183
2 291 1469 2088 4409
3 312 1495 2239 4503
Average 302 1452 2176 4365
COV 3.5% 3.7% 3.6% 3.8%
4-inch
(base case)
1 286 1094 1789 3611
2 291 1115 1692 3847
3 305 1184 1832 3876
Average 294 1131 1771 3778
COV 3.3% 4.2% 4.0% 3.8%
Compression Test (Layer Adhesion) (psi)
Layer Specimen (75 Min) (3 Days) (7 Days) (28 Days)
1-inch
1 199 698 1085 2393
2 178 766 1011 2564
3 187 759 1084 2627
Average 188 741 1060 2528
COV 5.6% 5.0% 4.0% 4.8%
2-inch
1 229 852 1298 3105
2 239 823 1147 2987
3 234 716 1323 2872
Average 234 797 1256 2988
COV 2.1% 9.0% 7.6% 3.9%
4-inch
(base case)
1 257 978 1557 3518
2 238 912 1718 3321
3 282 1008 1522 3376
Average 259 966 1599 3405
COV 8.5% 5.1% 6.5% 3.0%
Splitting Test (Layer Adhesion) (psi)
Layer Specimen (75 Min) (3 Days) (7 Days) (28 Days)
1-inch
1 34 159 191 433
2 39 136 192 377
3 35 152 226 426
Average 36 149 203 412
COV 7.3% 7.9% 9.8% 7.4%
2-inch
1 47 178 256 477
2 41 155 243 485
3 41 180 212 409
Average 43 171 237 457
COV 8.1% 8.1% 9.5% 9.1%
4-inch
(base case)
1 60 211 281 556
2 57 188 249 480
3 51 186 256 491
Average 56 195 262 509
COV 8.2% 7.1% 6.4% 8.1%
35

2.3.3. Results and discussion
The results of this work indicate the influence of the size of the layer and time interval on the
bonding interface [9]. The result of interlayer adhesion up to 28 days are shown in Figure 18.
Specimens that were monolithically cast in one stage (4-inch specimen) have the highest strength
under compressive strength (interlayer adhesion) and splitting prism tests. The decrease in bond
strength is due to improper bonding between layers. In contrast, specimens made of 4 layers of 1-
inch show higher strength under uniaxial compression tests.
When the heights of the layers are smaller compared to the base case (monolithic sample), the
bond strength decreases less for compression tests (interlayer adhesion) than for splitting prism
tests. In other words, splitting prism tests demonstrate more conservative results, which are 84%
less than that of compressive strength tests along layer interfaces.
36

Figure 18: Strength development of three types of specimens
(1. Uniaxial Compression test; 2. Uniaxial compression test (interlayer adhesion); 3. Splitting
prism)
37

The integrity of structures fabricated from layers of concrete under mechanically applied loads
was also investigated through analysis of crack propagation in specimens. Bond failure is
characterized as failure along the interface of two layers without leaving material attached to the
substrate (bottom layer) or overlay (top layer). Failure cracks propagate at the bond interface for
all samples during splitting prism tests (Figure 19-a), while only 43% of them failed at interface
(Figure 19-b) in uniaxial compression tests (layer adhesion). In contrast, fracture never occurs at
bonding interfaces under uniaxial compression tests (Figure 19-c). It was also observed that
crack growth rates in 1-inch layer specimens are higher than those of 2-inch layers.

Figure 19: Failure under applied force
a) Splitting Prism (100% of the samples), b) Uniaxial compression test in direction of the
interface (43% of the specimens), c) Uniaxial compression test
2.4. Summary
In this work, a series of experiments for four different mixtures based on the effect of aggregate
sizes and aggregate to cement ratio was performed. In addition, the effect of aggregate size on
the maturity model of a mix is explored, and a time frame for safe fabrication was suggested.
Furthermore, the integrity of a Contour Crafted wall and crack propagations at layer interfaces
was investigated under mechanically applied loads using compression and tension over a range
38

of fabrication procedures and layer assemblies. The following conclusions are made from this
study:
1- Tests at age 14 and 28 days showed that the mixture with smaller maximum aggregate
size (aggregate to cement ratio: 1.15) yielded a higher compressive strength. The 3/32”
mix yielded a 104 percent higher compressive strength than the 1/2” mix aggregate at the
age of 28 days.
2- Early age strength tests showed that mixtures with smaller maximum aggregate size
yielded higher compressive and split cylinder test strength. 3/32” (aggregate to cement
ratio: 1.15) mixtures yield a 141 percent higher compressive strength after 42 minutes
than that of 3/16” (aggregate to cement ratio: 1.5) mixtures.
3- In terms of early strength, the linear behavior of 3/16” concrete mixture which was
compatible with the Contour Crafting nozzle helped in predicting the strength of a
deposited layer at an early age more accurately.
4- Shorter setting times in mixture with 3/32” maximum aggregate size may produce cold
joints at layer interface, less homogenous structures, or even paste hardening inside the
extrusion system.
5- For 3/16” maximum aggregate size, the bond strength between layers increased with
increase in height of the layers and more time lapse between layers. In contrast, the
compressive strength of the specimen was higher when the height of the layers was
smaller and time lapse between layers was shorter.
6- The bond strength for 1-inch layer fabrication (6 minutes time lapse between layer
depositions) decreased from 19 to 36 percent with an average of 25% compare to
monolithic samples in splitting prism tests.
39

7- The failure of the specimens can be cohesive or adhesive. As shown in Figure 19, in
adhesive mode (splitting prism test), the failure happens along the interface of two layers
rather than the partial failure in cohesive mode (uniaxial compression test- layer
adhesion).
Although this study provides insight into effect of aggregate size, aggregate to cement ratio, and
extrusion rate, further experiments shall be performed on proposed mixture to calculate the
significance of each factor individually (e.g. effect of aggregate size in constant aggregate to
cement ratio and/or constant extrusion rate for different layer thickness).

40

3. Effects of Interlocking on interlayer adhesion and strength of structures in
Contour Crafting
Bonding between layers in 3D printing is critical in many applications especially in Contour
Crafting. A series of preliminary experiments on layered concrete structures show the
vulnerability of the structures due to low strength at bond interfaces. In some cases, however,
bonds with even a higher strength than the bulk (the monolithic segments) of the material are
achievable by using commercially available chemical bonding agents at the interface.
3D printed structure with interlocking of subsequent layers represent a new approach to support
integrity of the structure. Interlocking layers are often used in the 3D printing of mechanical
structures, jewelry, and parts with complex geometries.
Topological interlocking is a design principle for structures that elements are hold together
purely by geometrical constraints without binder or connector [48]. This type of assemblies
provide an alternative to monolithic structures and can be effective in addressing some of the
critical engineering problems [49–53]. It can also be applied to improve the performance of
composite materials (e.g. sandwich panels) [54].
Topologically interlocked material are categorized as class of granular crystal which made of an
assembly of polyhedral elements [55]. This fabrication technique allows some limited movement
of the blocks and can be used to build mortarless structures [56]. Topologically interlocked
structures are damage tolerant since the structural defect and cracks contained in the individual
units [57]. Therefore, the structure has higher resistance to fracture propagation [58].
Research by Schaare et al. [51] has shown potential of high energy absorption and self-reversible
mechanical behavior in topologically interlocked cubes. Mechanical and functional properties
41

including resistance to crack propagation, tolerance to local failure, and energy absorption are
advantages of topological interlocking [48,59–61].
The prime objective of this study is to investigate the impact of interlocking on the strength of
structure in Contour Crafting. This approach has yielded a working concrete mixture with a
range of interlocking configurations. Inspired by the traditional wood joinery techniques, a
simple tongue and groove shape lock was selected as a means of interlocking two concrete
layers. The configuration of interlocking was varied to find the best performance.
The outcome of the research is a detailed understanding of the application of interlocking on
layers of Contour Crafted structure for homogenous and sustainable fabrication, and a
framework for understanding the critical aspect of building design and construction that relates
to the integration of additive manufacturing [2].
2

3.1. Concrete mixture
Concrete mixture having a 3800 PSI compressive strength at 28 days is prepared. The binder is
composed of 60% Calcium Sulfoaluminate (CSA) and 40% type I ordinary Portland cement
(OPC) by mass. Natural Sand River of 3/16” (4.75 mm) maximum size and tap water are used to
prepare the mortar. Table 12 illustrates the weight and volume of the ingredients in the base
mixture.

2
This chapter is published in Automation in Construction journal.
42

Table 12: The weight and volume of concrete mixture’s ingredient
Batch Weight (lb.) Specific Gravity Absolute Volume (ft3)
CSA 9.00

Cement (Type I) 6.00

Total Cementitious Material 15.00 3.25 0.07
Aggregate 22.50 2.61 0.14
Water (W/C=0.50) 7.50 1.00 0.12
Total 45.00 0.33
Density of water 62.27 lbs./ft3 at 73.4F
Chemical Admixtures (Superplasticizer and Viscosity Modifying Admixtures) were added to achieve
desired workability and shape stability.
3.2. Specimen preparation
Before designing the size of the tongues, a series of preliminary experiments were conducted to
investigate the practicality of the approach and proper width to depth ratio for interlock features.
Constant factors in initial experiments were:
- Workability of the mixture which was designed to be compatible with the existing CC
machine.
- Width of the groove
Figure 20 shows that concrete filled the interface between layers up to 3/4" depth for the groove
and further increase resulted in non-uniform distribution. Therefore, 1/4”, 1/2”, and 3/4” sizes for
tongues were selected and compared. In addition, a layer with flat top surface (without the
tongue and groove joint) was studied as base case.
43

Figure 20: Preliminary study on the size of the interlocking
The target fabrication rate was to erect a ten-foot height wall in 12 hours (i.e., stacking rates of
10 inches per hour). Table 13 shows the fabrication rate of the layers. Height of a single layer
was determined using results of previous research by authors regarding interlayer adhesion. It
was expected that each layer gained strength to support its own weight, and after 12 minutes to
support the weight of layers above.
Table 13: Fabrication rate for 2" height layer
Total Height of the Specimen H = 4 in
Wall height 10 ft. = 120 in
Total fabrication time 12 hr. = 720 min
Fabrication rate of
720 min/120 in = 6
min/in
Height of a single layer h= 2 in
Interval between layers t= 12 min

44

The size of all specimens was 4” ×4” ×4”. Therefore, the mold system was designed based on
casting two layers of 2-inch height in 12-minute intervals. Table 14 and Figure 21 show the
sequence of fabrication and the extrusion system. The process simulated Contour Crafting, with
the exception that CC is extrusion based and fully automated. After four minutes of mixing,
concrete was placed in first layer. At time = 00:5:30 after placing was done, any extra particles
were removed from the top surface of the layer and it was flattened by a finishing trowel to
provide a consistent surface roughness for all the specimens. Mixing for second layer started
after 12 minutes. At 15 minutes, mold was detached from the first layer and was prepared for
second layer right before placing the concrete (No.3 to 6 on Figure 21). At time = 00:16:00, the
second layer was added and after 24 minutes at time = 00:29:30 the entire mold was removed.
The specimens were placed in water after 24 hours and were kept in water until the time of
testing.
Table 14: Fabrication sequence
Layer No. Mixing Start Mixing Stop Layer Placing Start Layer Placing End
1 0:00:00 0:04:00 0:04:00 0:05:30
2 0:12:00 0:16:00 0:16:00 0:17:30

The mold system was for 2-inch layer specimens to fabricate unconfined and self-supporting 4-
inch specimens. Each layer gains strength to support its own weight, and after 12 minutes to
support the weight of layers above. The mixture was poured on top of the first layer immediately
after mixing, and remaining layers were filled with the same freshly mixed concrete after twelve-
minute time lapses.
45

Figure 21: Layering Process and fabrication schedule (2” Interlocking)
3.3. Mechanical test
To ensure that fabricated structures can support the expected loads the bonding between layers
should resist mechanical loads. Interlayer adhesion must be investigated to quantify the bonding
quality of the structures. Samples based on same concrete mixtures are designed and placed
under compression tests in the direction of interface and split tensile to quantify bond strengths
after 1, 3, 7, and 28 days.
This section includes the result of testing 96 samples in uniaxial compression and splitting prism.
Figure 22 shows the size of the specimen, depth of the tongue, and direction of the applied force.
The sizes of all samples were 4”x4”x4”. Forces are applied in uniaxial compression in the
direction of the layer interface (interlayer adhesion). For splitting prism tests a square cross
section prism (4”x4”x4” cube) is placed under longitudinal compressive loading until tension
stress split the sample into two parts along upper and lower axes of loading.
46

Figure 22: Direction of applied force on the specimens
(A. Uniaxial compression test (interlayer adhesion), B. Splitting prism)
All 96 specimens are tested under uniform load. Table 15 shows the test results for each sample,
including the age of the sample, the average strength, and the coefficient of variation (COV)
after exclusion of outliers. COV’s vary from 3.2% to 8.4% for compression tests (interlayer
adhesion), and 4.2% to 9.7% for splitting prism tests, which all are in reasonable ranges.
The strengths for compression tests are calculated by dividing the maximum load by the area (16
in
2
). The strength for splitting prism tests are calculated by Ϭ= 2P/ πA, where Ϭ = splitting prism
strength (psi), p = applied force (lbf), and A = area of the interface (16 in
2
). For samples that
don’t have tongue and groove there is no interlocking, thus, these samples are treated as base
case using the same formula.

47

Table 15: Compressive and tensile strength for layers of fresh concrete (0”, 0.25”, 0.5”, and
0.75” interlock)

Compression Test
(Layer Adhesion)

Splitting Test
(Layer Adhesion)
Interlocking Specimen
1
Day
3
Days
7
Days
28
Days
Specimen
1
Day
3
Days
7
Days
28
Days
0.75"
1 715 718 1468 2805 1 146 158 249 432
2 670 849 1336 3022 2 143 165 279 437
3 664 803 1297 3047 3 162 185 246 488
Average 683 790 1367 2958 Average 150 169 258 452
COV 4.1% 8.4% 6.6% 4.5% COV 6.9% 8.2% 7.1% 6.8%
0.50"
1 803 982 1412 3276 1 177 214 317 516
2 789 912 1567 3360 2 173 186 266 609
3 754 881 1512 3498 3 201 224 315 598
Average 782 925 1497 3378 Average 184 208 299 574
COV 3.2% 5.6% 5.2% 3.3% COV 8.2% 9.5% 9.7% 8.9%
0.25"
1 685 948 1452 3212 1 147 188 234 552
2 723 859 1372 3026 2 151 175 281 589
3 743 815 1265 3266 3 164 203 262 522
Average 717 874 1363 3168 Average 154 189 259 554
COV 4.1% 7.8% 6.9% 4.0% COV 5.7% 7.4% 9.1% 6.1%
0.00"
1 640 734 1194 3098 1 146 162 246 473
2 691 823 1220 2889 2 128 163 251 462
3 703 834 1354 2977 3 153 188 214 436
Average 678 797 1256 2988 Average 142 171 237 457
COV 4.9% 6.9% 6.8% 3.5% COV 9.0% 8.6% 8.5% 4.2%
3.4. Results and discussion
Beushausen; M. G. Alexander [15] compared the shear bond interface of new and old concrete
for three different surface roughness’s and notches. Their results showed that notched interfaces
have higher bond strength, up to 45% more than sandblast surfaces and 220% more than smooth
interfaces at 28 days.
The results of strength test up to 28 days are shown in Figure 23 and Figure 24. Interlocking
features at the interface of the layers resulted in an increase in interlayer adhesion for samples of
all different ages; however, there is an optimum in the length of tongue for interlock. The
influence of interlocking appears to peak at 0.5 in, and additional depth to 0.75” does not
increase the strength of specimen at interface regions. Moreover, the increased strength caused
48

by interlocking is shown to be higher in splitting prism tests than in compression tests (interlayer
adhesion).

Figure 23: Comparison of bond interface for four types of interlock
(Uniaxial compression test (interlayer adhesion); Splitting prism)
49

Figure 24: 28 days’ strength development of four types of interlock
(Uniaxial compression test (interlayer adhesion); Splitting prism)
Higher strengths in specimens with notches can be investigated through the comparison of
failure modes. Notch interfaces provide a mechanical factor that increases the bond strength and
results in different pattern in failure at interfaces. Tschegg [17] defined two extreme categories
for crack propagation: 1) cracks forming in one side of the inhomogeneous cement based
material, and 2) cracks forming on both sides of the material.
50

Figure 25 shows the average modes of failure for four different samples under splitting prism
tests. Cracks can propagate in an interface of two layers in two different ways. In 0.25” and 0.5”
length notches, stress transferred to a large extent through the top layer while in specimens with
0.75” notches and specimens without notches the plane of failure are the same. Considering the
crack propagation in four different types of interlocking in 0.75” notches and base case
categories, the fracture process zone is formed on one side of the specimen (at interface), while
in 0.25” and 0.5” notches a combination of cases occurs. Therefore, more energy is needed to
separate the layers. The role of overlays in the interlocking of layer interfaces indicate that the
material properties play an important role in overall strength of the interface, and higher strength
materials can result in stronger interlayer adhesion.

Figure 25: Typical mode of failure in four types of notch
51

Delatte et al. [20] studied the bond strength between old and new concrete based on concrete
maturity at early ages and concluded that the new concrete layer is the dominant zone, and
tensile and shear strength at interfaces increase as early-age strength of new concrete increases.
Beushausena and Alexander [15] investigated the failure mode of old and new concrete under
shear stress in specimens with notches. They examined the fracture properties to determine the
behavior of layers at interfaces. As shown in Figure 26, they observed transformations of stress
to be mostly inside the overlay.

Figure 26: Modes of failure for specimen under shear stress [15]
Crack formation and propagation play an important role in the structural performance of Contour
Crafted walls, and the location of bond failure will aid in the investigation of the stress zones at
interfaces. In specimens with interlocks fracture occurs at bonding (adhesive failure) and layers
(cohesive), depending on the size of the notch.
52

Figure 27: Failure under applied force
a) Splitting Prism (100% of the samples), b) Uniaxial compression test in direction of the
interface (28% of the specimens)
According to Pigeon and Saucier [62], characteristics of the interfaces of new and old concrete
are quite similar to the bond between cementitious materials and aggregate in concrete mixtures,
and can contribute to the understanding of the formation and propagation of cracks in layers of
concrete. Crack formation and propagation play an important role in the structural performance
of Contour Crafted walls, and the location of bond failure will aid in the investigation of the
stress zones at interfaces. In specimens with interlocks fracture occurs at bonding (adhesive
failure) and layers (cohesive), depending on the size of the notch.
Under splitting prism tests, cracking progresses at layer interfaces, whereas under compression
failure they may be observed in the layers themselves. Micro-cracks develop in compression
tests, while shear cracks are generated during splitting prism tests and influence the shear
capacity of the specimen. Figure 27a shows failure at interfaces during splitting prism tests, and
Figure 27b shows that 28% of the specimens failed by failure in layers when uniaxial
compression loads were applied (interlayer adhesion).
53

3.5. Summary
The bond strengths of 96 samples (4”x4”x4” cube) constructed with four different interlocking
sizes (base case of 0.25”, 0.5”, and 0.75”) were investigated [2]. Further application and
analytical methods such as finite element analysis can also be applied to understand more
complex structures [63]. The four types of interfaces are quite different in terms of strength at
interface regions. The test results revealed the following:
1- Regardless of the test method used, interlocking layers of 0.25” and 0.5” (tongue’s depth)
increases the bond at interfaces of the contour crafted layers.
2- The bond strength for specimens with 0.5” (tongue’s depth) interlocking layers increases
from 16% to 19% with an average of 17% in comparison with the base case under
compression along layer interfaces (interlayer adhesion). The increase in bonding
strength was, on average, 26% for splitting tests. This result shows that bonding strengths
are sensitive to interlocking.
3- The results also indicate that increasing the length of tongue at interlock to 0.75”
diminishes the effect of the interlock, and the results are identical to the base case.
4- Cracks at the interface of layers of a Contour Crafted wall cause loss of tensile strength of
the wall. Thickness of the wall can be an important factor in keeping the wall stable and
carries vertical loads.
5- Despite the relatively simple construction method, Contour Crafted structures usually
show complex behavior when undergoing different load conditions. This necessitates a
suitable design approach to properly account for different loads.

54

4. Effects of mixture ingredients on interlayer adhesion in Contour Crafting
4.1. Introduction
CC is an additive manufacturing method that constructs in a layer by layer fabrication process.
An interface appears whenever a layer of concrete is applied to an existing one. The performance
of a Contour Crafted wall is directly related to the strength of these interfaces. According to Lim
[64], because of stress concentrations at interface in a layered concrete structure the chance of
failure at interfaces is higher. In most of existing literature the interface between old and new
concrete is considered. However, there is little information about the strength and fracture
process at the interface between two extruded layers of fresh concrete when they are not subject
to any post processing (e.g. vibration).
The design of the concrete mixture proportion is complicated due to different requirements for
viscosity, workability, shape stability, and bonding strength. For instance, a decrease in water to
cement ratio can increase the shape stability of the extruded layer, but increase viscosity and thus
lead to blockage of the flow inside nozzle. On the other hand, low viscosity concrete can
segregate during placement, cause a non-uniform aggregate distribution, develop micro-
structures, and result in low bonding strength between layers. Another example is inter-particle
friction between coarse aggregate, sands, and fine aggregate as well as between two adjacent
layers which increase bonding at interface, at the same time, such a friction is an obstacle to the
flow of fresh concrete.
55

The contradicting workability requirements (low viscosity and shape stability) for CC necessitate
a homogenous concrete mixture to provide an acceptable trade-off [10]
3
. The combined use of
superplasticizers and Viscosity Modifying Admixtures (VMA) can result the required
workability of the concrete. Therefore, the need for proper admixtures shall be observed and
quantified.
A full factorial design test was performed to identify the effects of key mixture parameters on
interlayer adhesion of a Contour Crafted wall. Parameters of interest to be tested include
interface strength. 36 mixtures were prepared to investigate the relationship between different
factors and to derive a final statistical model. This section presents the statistical model for effect
of key mixture parameters and their interactions on strength. Statistical model shows the balance
between mixture’s parameters and helps to develop a prediction model. This model will be used
to narrow down the upper and lower bounds of concrete mixture’s components for optimization
of bond interfaces in Contour Crafting construction.
The main factors of interest in this section are bonding agents (cementitious and supplementary
cementitious materials), fibers for reinforcement, water to cement ratio, and chemical admixtures
(superplasticizer). To identify the value of different factors for the experiment, a range of
operating conditions was defined. This range was the result of initial experiments with
measurable and acceptable strengths for bonding interfaces. The objective of this chapter is to
optimize conditions for maximum achievable interlayer adhesion within the CC nozzle
constraints (e.g. extrudability, aggregate size).

3
Effects of mixture ingredients on extrudability of concrete in Contour Crafting is published in Rapid Prototyping
Journal.
56

4.2. Design of experiment
Properties and characteristics of concrete mixtures depend on the type of binder (cementitious
material), aggregate, and the amount of water. Uniformity of aggregate distribution is affected by
mixture of cementitious material and water which fills the voids between aggregates. Among all
different mixture’s processing conditions (factors), mechanical properties of the mixture were
investigated through experiments, and statistical tools were applied to enhance structural
integrity and shape stability.
Figure 28 shows the research process. The objective of this section is to quantify the effects of
different mixture parameters, identify their significance, and observe their interactions by
implementing a statistical experimental design that allows for manipulation of several variables
in order to improve the bonding strength of layers while ensuring the desired flow-ability.
Interlayer adhesion strength depends on the bonding agent (interface adhesion), aggregate size
(friction and aggregate interconnection), the time interval between layer depositions, roughness
of the top surface of substrate, and the shape and surface of the topmost surface. Considering
constant material compaction, substrate age, and roughness of the surface, this section examines
different parameters for concrete mixtures (e.g. bonding agent, water to cement ratio, etc.). The
experiment will be based on identifying maximum strength at interface, desired workability, and
the nature of relationships between them.
Considering that flowability is an important factor for extrudability and strength of structures
fabricated by CC, initial attempts to optimize all the responses concluded with several runs that
yielded unacceptable results before range of operating conditions was identified. After a region
for the acceptable operation condition was identified, the impact of different factors on the final
57

flow-ability and interlayer adhesion were studied. An empirical model of strength versus other
factors was introduced based on responses from experimental studies. As can be seen in Figure
28, each model is validated by experimentation.

Figure 28: Design of experiment of the concrete mixture for the optimization of the interface
strength and flow-ability
Research about strength at interface of old and new concrete have been conducted by others who
studied the related factors individually. Momayez [19] investigated the effect on bond strength of
the old and new concrete when silica fume is added to the mixture. His results showed that 7%
silica fume is the optimum amount to add, which increases bond strength by up to 22%. After
comparing four different test methods, it was concluded that silica fume was more effective
under shear than tension. Beushausen, and Alexander [15] examined the bond strength of new
and old concrete. Their results show that higher bond strength was achievable with lowering
viscosity in concrete.
Khayat [65] studied the influence of viscosity enhancing admixture and high-range water
reducers, water to cement ratio, content of cementitious materials, and the volume of coarse
aggregate on self-consolidating concrete. He modeled the deformability of concrete in non-
restrained areas (slump flow, rheological parameters), deformability in restrained areas (V-funnel
and filling capacity), and compressive strength of self-consolidating concrete. There are other
statistical researches regarding concrete mixtures and different conclusions have been reported in
literature regarding the relationship between bond strength and material properties [15].
Initial
Experiment
Operating
Region
Condition
Design of
Experiment
Empirical
Model
Model
Optimization
Validation
Experiment
Strength
Model
Workability
Model
58

The goals of this part of research are to investigate the effect of different concrete mixtures on
the bond strength between layers of a Contour Crafted wall. Bonding agents are divided into two
main categories: cement based, and modified cementitious material. 108 specimens were
fabricated and subjected to splitting prism tests. In this section, mixture parameters impacting
bonding interfaces are studied via Design of Experiment. Experiments were designed statistically
to gather necessary data to describe the relationship between the factors and structural
performance. Finally, the results are analyzed to develop an empirical model for further research.
4.2.1. Factorial Design
To analyze different combinations of concrete mixes, statistical tools are applied using full
factorial design. The force required to overcome interface adhesion is the quantifiable response
in relation to the value of different factors. At the same time, the impacts of different factors on
extrudability are calculated through analysis of flow-ability. At the end of the experiment a
functional model is introduced to describe the mix conditions for obtaining maximum interlayer
adhesion and required flow-ability, and the model is validated through a series of experiments.
A proper concrete mixture for Contour Crafting must be able to flow such that it can be pumped
easily without blockage or segregation. Since there is no vibration or post processing after
placement of the layer, the mixture should have a relatively high workability, but it should have
sufficient resistance against segregation. Segregation and bleeding can result in non-uniform
distribution of aggregates in the mix during and after extrusion. These tradeoffs in flow-ability of
the mixture and segregation are studied in addition to investigating the effects of water reducer
chemical admixture (e.g. superplasticizer) to optimize the proper proportions of the mix.
59

Concrete should have proper shape stability after extrusion and gain sufficient strength fast
enough to support the weight of the upper layers and this requires a customized concrete mixture.
These initial tests provide the required information to identify the range of factors.
To find a suitable homogenous mixture, in this part of research different variables are
investigated and a statistical model is introduced based on the design of experiments. The
resulting empirical model is a guideline that can be used to predict the influence of different
mixture parameters on properties of the concrete. Four key factors that can impact interlayer
adhesion of two layers and flow-ability are selected to derive analytical models. The four
variables included water to cement ratio, choice of chemical admixture, choice of cementitious
material and bonding agent, and the amount of fiber contents. The strength at interface after 28
days are measured along with flow-ability after mixing. Figure 29 shows the list of factors.
Water to cement ratio and the amount of Superplasticizer are two factors that have direct
influence on flow-ability of the mixture. The type of the cement and percentage of silica fumes
relate to the bonding agent, and fiber reinforced concrete enhances the tensile strength of the
concrete.

Figure 29: Hierarchy of the process and factors for Design of Experiment
Tensile
Strength
Workability
Binder
Cementitious Material:
- Portland Cement Type II
- Silica Fume
Cementitious Material:
- Portland Cement Type II
- Silica Fume
Water to Cement ratio
Chemical Admixture
- Superplasticizer
Water to Cement ratio
Chemical Admixture
- Superplasticizer
Fiber Fiber
Confirm
Compatibility
with Extrusion
System
Interface
Strength
Flowability
OUTPUT OUTPUT INPUT INPUT
60

The sub factors that have impact on bonding strength are too complicated to allow the
development of an exact empirical model. For instance, the air content of the concrete mixture
was not calculated as a component of the mix, although it affects the calculation of the yield. Air
content was not considered as being important for small batches, and can be added as a factor
after the selection of the final mix.
The operational range for four factors was obtained through preliminary tests and previous
research regarding bonding between old and new concrete. One factor at a time was varied
during the preliminary experiments. The main experimental interests are mixtures that are
extrudable or can be extruded with only slight changes in flow-ability, as well as experimental
parts that have reasonable strength at interface.
Table 16 shows the factors and their level (min, max, and center point). The sand/cement ratio of
the mix was 2.4, and W/C was 0.45 and 0.55. The total cementitious material was kept constant
by replacing 5% and 10 % of weight of type II Portland cement with silica fume. Center points
were added to the two different factors, since the expectation is that the strength and flow
responses do not follow a linear pattern.
Table 16: Factor’s operational range affecting interlayer adhesion
Variables
Silica Fume SF0 SF1 SF2 3
Fiber F1 _ F2 2
Superplasticizer SP0 SP1 SP2 3
W/C ratio W1 _ W2 2
Total Concrete Mixtures 36
Time _ _ 28-Days 1
Sample 1 2 3 3
Test Split 1
Total Samples 108

61

The concrete strength is influenced by binder, ratio of water to cementitious material, the ratio of
aggregate to binder, and the bond between binder and aggregate. Table 17 shows the levels of the
factors. The initial levels for the factors were selected after running preliminary experiments
based on the required concrete properties. Setting for maximum, minimum, and center points
were concluded in 36 runs.
Table 17: Values of different levels of factors
Factor Level Unit
Sand S 2400 Gr _
Silica Fume SF0 0 Gr -1
SF1 50 Gr 0
SF2 100 Gr 1
Superplasticizer SP0 0 CC -1
SP1 5 CC 0
SP2 10 CC 1
Fiber F1 0 Gr -1
F2 5 Gr 1
W/C W1 0.45 W/C -1
W2 0.55 W/C 1
Total cementitious material is 1000 gr.

Table 18 shows that the experiment focuses on full factorial designs, involving 36
combinations
4
. Responses are the influence of different levels for each of four main mixture
variables on interlayer adhesion (after 28 days) and flow-ability and they were measured for each
run of the experiment and a statistical model will be derived over a range of mixture proportions.

4
Silica Fume (3 levels) *W/C ratio (2 levels) *Superplasticizer (3 levels) *Fiber (2 Levels) = 36 combinations
62

Table 18:Combination of different levels for 36 concrete mixtures
Sand Cement W/C Silica Fume Superplasticizer Fiber W SF SP F
M1 S C W1 SF0 SP0 F1 -1 -1 -1 -1
M2 S C W2 SF0 SP0 F1 1 -1 -1 -1
M3 S C W1 SF0 SP1 F1 -1 -1 0 -1
M4 S C W2 SF0 SP1 F1 1 -1 0 -1
M5 S C W1 SF2 SP0 F1 -1 1 -1 -1
M6 S C W2 SF2 SP0 F1 1 1 -1 -1
M7 S C W1 SF2 SP1 F1 -1 1 0 -1
M8 S C W2 SF2 SP1 F1 1 1 0 -1
M9 S C W1 SF2 SP0 F2 -1 1 -1 1
M10 S C W2 SF2 SP0 F2 1 1 -1 1
M11 S C W1 SF2 SP1 F2 -1 1 0 1
M12 S C W2 SF2 SP1 F2 1 1 0 1
M13 S C W1 SF1 SP0 F1 -1 0 -1 -1
M14 S C W2 SF1 SP0 F1 1 0 -1 -1
M15 S C W1 SF1 SP1 F1 -1 0 0 -1
M16 S C W2 SF1 SP1 F1 1 0 0 -1
M17 S C W1 SF1 SP0 F2 -1 0 -1 1
M18 S C W2 SF1 SP0 F2 1 0 -1 1
M19 S C W1 SF1 SP1 F2 -1 0 0 1
M20 S C W2 SF1 SP1 F2 1 0 0 1
M21 S C W1 SF0 SP0 F2 -1 -1 -1 1
M22 S C W2 SF0 SP0 F2 1 -1 -1 1
M23 S C W1 SF0 SP1 F2 -1 -1 0 1
M24 S C W2 SF0 SP1 F2 1 -1 0 1
M25 S C W1 SF0 SP2 F1 -1 -1 1 -1
M26 S C W2 SF0 SP2 F1 1 -1 1 -1
M27 S C W1 SF0 SP2 F2 -1 -1 1 1
M28 S C W2 SF0 SP2 F2 1 -1 1 1
M29 S C W1 SF2 SP2 F1 -1 1 1 -1
M30 S C W2 SF2 SP2 F1 1 1 1 -1
M31 S C W1 SF2 SP2 F2 -1 1 1 1
M32 S C W2 SF2 SP2 F2 1 1 1 1
M33 S C W1 SF1 SP2 F1 -1 0 1 -1
M34 S C W2 SF1 SP2 F1 1 0 1 -1
M35 S C W1 SF1 SP2 F2 -1 0 1 1
M36 S C W2 SF1 SP2 F2 1 0 1 1
Two Factors at Two Levels and Two Factors at 3 Levels (36 Runs)
4.2.2. Material properties
The mortar was prepared based on ASTM C 305, Mechanical Mixing of Hydraulic Cement
Pastes and Mortars of Plastic Consistency [39]. Mixing started with blending sand for 30
63

seconds, then adding 70% of the mixing water and all the superplasticizer. Cementitious material
(and fiber if applicable) was added after 30 seconds of mixing. Flowing one minute of mixing,
the remaining water was added. Mixing was continued for two more minutes, two minutes of
rest, and additional two more-minute mixing at the end.
Binder
Binder is a vital factor on interlayer strength since it affects the microstructure of the interface. A
number of researchers reported the significance of the binder in bond strength of the new and old
concrete [66]. To enhance the rheological properties and strength of the concrete mixture, series
of preliminary experiments were performed to investigate different binders and admixtures. The
binder contains Portland cement type II (OPC II) and silica fumes. The chemical composition of
the silica fume is given in Table 19.
Table 19: Physical properties of Silica fume
PH Density Solubility in Water ASTM
Neutral 2.1-2.3 g/cm3 Miscible C 1240

Silica fume usually makes up 10 percent by mass of the cement and increases the compressive
strength of the concrete. The dry compacted silica fumes from BASF (Master life SF 100)
contained more than 85% SiO2. The composition of the ingredient MasterLife SF 100 is shown
in Table 20. MasterLife SF 100 is a dense silica fume admixture that enhances the strength,
durability, and performance of the concrete. As a pozzolan this admixture increases the
cohesiveness and modulus of elasticity and decreases bleeding and permeability.

64

Table 20: Composition of ingredient of MasterLife SF 100
Silicon Dioxide, amorphous > 85
Carbon <10
Magnesium Oxide <5
Iron Oxide <5
Calcium Oxide <5
Silicon dioxide, crystalline (Quartz) <0.5

In concrete’s hardened state, calcium silicate hydrate (C-S-H) is the bonding agent that holds the
cementitious mixture together. During chemical reaction with the cementitious material in
concrete, silica fumes increase the amount of C-H-S by reacting with the calcium hydroxide
which is produced by hydration of cement [67]. Cong et al. [68] reported that partial replacement
of cement with silica fume increased compressive strength due to increase in strength of the
cement paste. However, the change in binder-aggregate interface due to presence of silica fume
had little effect.
Tayeh et al. [69] reported that adhesion strength of the new and old concrete significantly
increased when silica fume added to OPC. They showed the improvement of the microstructure
as result of the adding silica fume. Additional C-H-S fills the pores at interface of old and new
concrete and results in a stronger bonding [66]. Shin [70] studied the interfacial properties
between new and old concrete and reported that adding 7% silica fume increased the
compressive strength of the new concrete and shear bond strength at interface.
Aggregate
Non-contaminated continuously virgin aggregate materials, reclaimed, or recycled materials
were used in the experiment. Sand was used for all mixes compatible with ASTM C144 [71].
The sieve analysis of the aggregate and gradation is summarized in Table 21. The effect of
65

maximum aggregate size on the mechanical properties of concrete mixture was investigated and
reported in previous chapters. It was concluded that concrete with smaller size aggregate (4.75
mm) showed higher compression and tensile strength than corresponding concrete with larger
size aggregate (12.5 mm).
Table 21: Sieve analysis of the concrete sand
5

Sieve Size Plaster Sand ASTM C144
4.75 mm (#4) 100 100
2.36 mm (#8) 98 95-100
1.18 mm (#16) 86 70-100
600 µm (#30) 62 40 – 75
300 µm (#50) 28 10 – 35
150 µm (#100) 8 2 – 15
75 µm (#200) 3 0 - 5

The grading, shape, size, surface, texture, minerology, and strength of the aggregate influence
the strength of the concrete. Physical properties of the aggregate are shown in Table 7. The
aggregate was plaster river sand with a fineness modulus of 2.2, bulk specific gravity of 2.65,
and absorption (dry) of 1.1 percent. The unit weight (saturated surface dry) was 102 lbs./ft
3
.
Aggregate moisture content was calculated per ASTM C 70 prior to mixing.
The influences of aggregate on strength of concrete have been investigated by different
researchers. Zhou [72] reported that higher modulus of elasticity for concrete is achievable with
stiffer aggregate. Aitcin [73] concluded that compressive strength and elastic behavior of the
high strength concrete mixture are affected by mineralogical properties of the aggregate. Brown
[74] observed that aggregate size and shape is an important factor in rheology of the concrete
mixture.

5
Email from Hanson Aggregates
66

Table 22: Physical properties of the aggregate
6

Fineness Modulus 2.2
Specific Gravity, Bulk S.S.D. 2.65
Absorption, % 1.1
Sand Equivalent, CT 217 80
Sodium Sulfate Soundness, CT 214 0.011
Organic Impurities Satisfactory
Durability Index, CT 229 81
Unit Weight (ASTM C29) 102 lbs./ft3

Chemical Admixture (Superplasticizer)
Superplasticizer modifies the workability of concrete while reducing the amount of water needed
for a specific slump, and increases the compressive strength of the concrete. It also changes the
flow-ability of the mix and makes the pumping process easier. However, very high dosage of
superplasticizer lessens the cohesiveness of the mixture and would necessitate additional
chemical admixtures to overcome this problem. Aghabaglou [75] observed that influence of the
superplasticizer on compressive strength of the concrete mixture was asserted at early ages of
concrete. On the other hand, high dosages of superplasticizer can reduce the compressive and
flexural strength of the concrete [76].
The water reducer used in the experiment was ADVA® CAST 600 (GRACE) which is a
polycarboxylate based chemical that extends the slump life of the concrete mix while increasing
its workability and its main application is in low water to cement ratio mixtures [77]. The
admixture has a specific gravity ranging from 1.065 to 1.085 g/cm
3
, and was used at rates
ranging from 200 to 390 mL to 100 kg of cementitious materials with addition rates of 130 to

6
Email from Hanson Aggregates
67

650 mL/100 kg. The water reducer was mixed with the water before adding dry materials. Table
23 shows the physical properties of superplasticizer.
Table 23: Physical properties of Superplasticizer
PH Density Solubility in Water ASTM
5.1-6.1 1.065-1.085 g/cm3 Miscible C 494

Fiber
The weakness in tensile strength of the concrete can be overcome to some extent by adding fiber
to the mixture. MasterFiber M 100 (BASF) is a polypropylene, monofilament, and ultra-thin
fiber designed to control plastic shrinkage of the concrete mixture. It helps transform cracks from
macro to micro in concrete [78]. M 100 is translucent polypropylene with density of 0.91 g/cm
3
.
Table 24 shows the physical properties of the fiber.
Ahmed [79] observed an increase in flexural, tensile, and shear strength of the concrete as a
result of adding fiber to the mixture, while the compression strength did not change. Patel et al
[80] reported that compressive strength of the concrete increased by adding 1.5% polypropylene
fibers. Madhavi et al [81] suggested that addition of fibers in mixture improves the failure mode
of concrete and control the development of micro crack during the curing. It also controls post-
cracks, and increases residual flexural strength in concrete. Gencel et al [82] found that fibers
bridge the micro cracks and hold back the propagation of cracks.

68

Table 24: Physical properties of Fiber
Density 0.91 g/cm3
Tensile Strength 70 psi (480 MPa)
Modulus of Elasticity 1,230 psi (8.48 GPa)
Lengths 0.5 in. (13 mm)
Equivalent Diameter 0.00047 in. (12 microns)
ASTM C 1116/C 1116M
Absorption Nil
Number of fibers per pound 225 million
Dosage 0.50 lb./yd³ (0.3 kg/m³)

Adding fibers reduces the workability of the concrete. Patel et al [80] reported that higher
polypropylene fiber decreased the workability of the concrete due to increase in entrapped air
voids and air content. Mohamed [83] conducted investigation on effect of polypropylene fibers
and results showed significant increase in water absorption of the specimen. Fiber also decrease
the plastic shrinkage [84]. Slight increases in fine aggregate contents and use of superplasticizer
might be needed to achieve required workability after adding fiber to the mix.
Water
Water to cementitious material ratio is an important factor on flow-ability. Mixtures which have
a low yield value, but contains high amounts of water may result in the segregation of the
aggregates after extrusion. Potable water available in the laboratory, which satisfies drinking
standards was used for mixture’s preparation, and curing the specimen. In this study, a full
factorial design is used for modeling of the flow-ability of the mixture and interlayer adhesion of
concrete for water to cement ratio of 45% and 55%.
4.2.3. Flow-ability test
Concrete used for Contour Crafting shall meet special combinations of extrusion and uniform
placement which is strongly related to ease of extrusion and placement without segregation of
69

fresh concrete that cannot be achieved by using conventional concrete mixing. Workability and
flow-ability cover pump-ability, ease of placement, consistency, stability, and finish-ability of
the extruded part.
There are different standard and nonstandard tests to measure the flow of concrete. Majority of
these test, can only be used to compare mixtures when using the same test. Figure 30 shows an
overview of tests that are used to describe the rheology of the fresh concrete. A flow test should
be able to determine the yield stress and plastic viscosity to measure the flow of concrete.
Gravity and vibration are two common methods to exceed the yield stress. The Slump test
(Figure 30A) is the most common test to measure the workability of fresh concrete. The test is
applicable to plastic and cohesive concrete based on ASTM C143 [85]. Slump can be distributed
evenly (true slump) or on one side of the cone (shear slump). Shear slump is a sign of
segregation in mixture and it is measured based on the difference of slides on two sides of the
cone.
Figure 30B shows the V funnel test to measure the flow-ability and evaluate the segregation
resistance of Self Compacting Concrete (SCC). High viscosity pastes are associated with higher
flow time thorough the inverted cone. The flow table test (Figure 30C) measure the consistency
of SCC with slump more than 175 mm [86]. The pattern of the concrete which spread on the
table is an indication of the cohesiveness in mixture. A non-circular distribution is a sign of
segregation. Figure 30D shows the compaction factor apparatus which is used gravity and
vibration to measure the workability of the mixtures with low workability. Stress applied based
on gravity in these tests and there is no external force or vibration except compaction factor test
which is based on the difference in weight of the partial and fully compacted concrete.
70

Figure 30: Flow-ability (Work-ability) test methods
In this experiment, to measure the flow-ability, concrete considered as a homogenous fluid in its
fresh state and the objective is to predict the flow-ability of concrete given the composition of
different materials. To study the effect of different mixtures’ parameters on flow-ability of the
concrete a total of 36 mixtures were made and several measurements were conducted to relate
properties of the mixture with flow-ability.
71

Figure 31: Flow table apparatus[37]
To characterize the flow-ability response of the concrete, it is fundamental that the test and
experimental procedure resemble the CC criteria. The apparatus set up to measure the flow-
ability is based on ASTM C230, flow table for use in tests of hydraulic cement which like CC is
based on gravity and vibration to overcome the yield stress of concrete. Figure 31 shows the flow
table apparatus. The conical mold’s height is 2” with diameters of 2.75” for the top and 4.00” for
the bottom openings [37]. The top circular rigid surface of the flow table apparatus was modified
to measure up to 14-inch diameters.
The tests were performed three times for each set of factors combination based on ASTM C1437,
flow of hydraulic cement mortar [87]. The mortar was placed about 1” in thickness inside conical
mold and was tamped 20 times with the tamper. After that the mold was filled with concrete and
was tamped as specified in the first layer. The mortar was cut in a plane surface using the straight
edge. After 1 minute, the mold was removed and table was dropped 25 times in 15 seconds. The
diameter of the mortar was measured after it was spread by 1/8” accuracy along four lines and
the average was recorded. A non-circular distribution is an indication for tendency for
72

segregation. Results of this experiment is introduced as flow-ability to describe the extrudability
and placement of the concrete through Contour Crafting.
4.2.4. Specimen Preparation (Interlayer adhesion)
In this experiment, two identical layers interact through contact, which allows for the testing
tensile strength at the interface. The base part of the experiment is a 2”x4”x4”sample block. The
target fabrication rate is 10 feet height in 12 hours, which requires 10 inches per hour. The
following calculations show the relevant data associated with this fabrication rate.
Table 25: Fabrication rate for 2" height layer
Total Height of the Specimen H = 4 in
Wall height 10 ft. = 120 in
Total fabrication time 12 hr. = 720 min
Fabrication rate of 720 min/120 in = 6 min/in
Height of a single layer h= 2 in
Interval between layers t= 12 min

All mixtures were prepared in 6 quart batches based on ASTM C305 [39]. Mixing was lasted
until mixtures became uniform. The same concrete was used for the two layers of each sample.
Variations in different samples were determined by the statistical design of experiment. The
mixes were based on Portland Cement Type II, with bonding agents containing 0%, 5%, and
10% silica fume, and maximum crushed aggregate size of 4.75 mm (3/16 in.). 36 mixes were
designed based on the same interface condition (consistent surface roughness). For each mixture,
the consistency was measured based on ASTM C230 [37] as well as the extrudability of the
concrete through the nozzle.
73

Figure 32: Layering Process and fabrication schedule
Table 26: The sequence of fabrication for 2 layers
Layer No. Mixing Start Mixing Stop Layer Placing Start Layer Placing End
1 0:00:00 0:06:00 0:06:00 0:07:30
2 0:12:00 0:18:00 0:18:00 0:19:30

Figure 32 and Table 26 shows the sequence of fabrication for the specimens. The process
simulates Contour Crafting, with the exception that CC is fully automated. Concrete was placed
in a mold at time = 00:6:00. The contact surface of the substrate with the subsequent layer was
smoothed with a finishing trowel to keep surface roughness consistent in all specimens (Figure
33). After 12 minutes at time = 00:19:30 the second layer was added, and after 24 minutes at
time = 00:31:30 the entire mold was removed. To avoid bonding between concrete and the inner
surfaces of mold, oil was applied to the inner surfaces of the mold before pouring concrete.
Finally, the samples were placed in water and then tested 1 day and 28 days after fabrication.
74

Figure 33: Surface preparation; smoothed with finishing trowel
Figure 34 demonstrates the extrusion system for 2-inch layer specimens to fabricate unconfined
4-inch specimens. A layer of 2-inch concrete was cast as substrate, and a second layer was
deposited on it. The mixture is poured on top of the first layer immediately after mixing, and the
remaining layers were placed with the same freshly mixed concrete after six minutes’ time
lapses.

Figure 34: Molding and Extrusion Apparatus
4.2.5. Mechanical Test (Interlayer adhesion)
To be sure that structure assembled with the contour crafting process lasts, interlayer adhesion
must be investigated to quantify homogeneity of the structure. Split tensile tests were performed
to quantify those different variables after 1 and 28 days. The consistent applied rate of loading is
a representative of the lateral load and its effect on a CC structure and the reaction of the
75

specimen and its failure under the uniaxial compression load at interface allows to measure the
strength at this region particularly.
The size of the base part for the experiment was a sample block with dimension of 2”x4”x4’. The
size of the specimen is enough which boundary condition did not contribute to the stress at
interface. This section includes the results of 108 samples subjected to splitting prism tests based
on ASTM C496 [22]. Figure 35 shows the size of the specimen and direction of the applied
force. This assembly limited the orientation of crack propagation to the plane of interface.
Compressive forces were applied uniformly over the length of the layers and the specimen failed
due to development of tensile stress at interface. The bond strength was measured and the failure
stress was called interlayer adhesion strength.

Figure 35: Direction of applied force on the specimens (Splitting prism)
All 108 specimens were loaded statically at the age of 28 days, and all of them failed at interface.
Failure resulted from development of shear strength over the specimen’s interface. Because the
goal of this study is to determine the load at which adhesion at interlayer fails, any deviation
from the orientation of interface was eliminated by controlling the angle of the applied stress.
The specimen failed when principal stress exceeded the bonding at interface and interlayer
76

adhesion strength was calculated by reading the maximum stated stress and Ϭ= 2P/ πA, where Ϭ
= maximum stress (psi), p = applied force (lbf), and A = area of the interface (8 in
2
), and the
average value was based three specimens. Interlayer adhesion results are introduced as Strength-
28 days.
4.3. Experimental results
Full factorial design is a statistical technique used for modeling the results. To identify the model
from results, experiments were designed statistically, and regression model technique were
applied. Input variables (W/C ratio, Superplasticizer, Silica Fume, and Fiber) are independent
variables which defines the response (Flow-ability, and Interlayer adhesion).
4.3.1. Flow-ability
To test the flow-ability of the mixture, 3 runs for each combination of factors were made in
random order. Table 27 shows the test results for 108 samples, including the average flow-ability
and the coefficients of variation (COV) which vary from 2.8% to 9.8%.

77

Table 27: Summary of Flow-ability tests for three different trials
W SF SP F A B C AVERAGE COV EXTRUDABILITY
M1 -1 -1 -1 -1 7.375 6.625 6.625 6.875 6.30% Maybe (Dry)
M2 1 -1 -1 -1 8.125 9.25 9.625 9 8.67% OK
M3 -1 -1 0 -1 9.125 8.25 8.125 8.5 6.41% OK
M4 1 -1 0 -1 11.5 9.875 11.25 10.875 8.05% Watery
M5 -1 1 -1 -1 6 6.625 5.75 6.125 7.36% Maybe (Dry)
M6 1 1 -1 -1 8.25 7 8 7.75 8.53% OK
M7 -1 1 0 -1 7.25 7.125 8.125 7.5 7.26% OK
M8 1 1 0 -1 8.875 10 10 9.625 6.75% Maybe (Watery)
M9 -1 1 -1 1 4.875 4.25 4.375 4.5 7.35% NO (Dry)
M10 1 1 -1 1 4.675 5.575 5.5 5.25 9.51% NO (Dry)
M11 -1 1 0 1 5.375 6 5.5 5.625 5.88% NO (Dry)
M12 1 1 0 1 6.5 7.5 7.75 7.25 9.12% OK
M13 -1 0 -1 -1 6.25 6 6.875 6.375 7.07% Maybe (Dry)
M14 1 0 -1 -1 8.5 8.875 7.375 8.25 9.46% OK
M15 -1 0 0 -1 7.625 7.5 8.5 7.875 6.92% OK
M16 1 0 0 -1 9.125 10.5 10.75 10.125 8.64% Maybe (Watery)
M17 -1 0 -1 1 4.625 5.125 4.5 4.75 6.96% NO (Dry)
M18 1 0 -1 1 6 5.25 6 5.75 7.53% NO (Dry)
M19 -1 0 0 1 5.75 6.5 5.75 6 7.22% Maybe (Dry)
M20 1 0 0 1 8.125 6.875 7.875 7.625 8.67% OK
M21 -1 -1 -1 1 5.5 5 4.875 5.125 6.45% NO (Dry)
M22 1 -1 -1 1 5.875 6.95 6.675 6.5 8.59% Maybe (Dry)
M23 -1 -1 0 1 7 7 5.875 6.625 9.80% Maybe (Dry)
M24 1 -1 0 1 8.875 7.75 8.875 8.5 7.64% OK
M25 -1 -1 1 -1 9.75 9.5 10.75 10 6.61% Maybe (Watery)
M26 1 -1 1 -1 13.5 13.625 11.875 13 7.51% Watery
M27 -1 -1 1 1 8 8.875 7.875 8.25 6.60% OK
M28 1 -1 1 1 11.125 9.5 10.875 10.5 8.33% Maybe (Watery)
M29 -1 1 1 -1 9.75 8.875 8.75 9.125 5.97% OK
M30 1 1 1 -1 10.875 12.25 12.5 11.875 7.37% Watery
M31 -1 1 1 1 7.75 7 7 7.25 5.97% OK
M32 1 1 1 1 9 9.25 9.5 9.25 2.70% OK
M33 -1 0 1 -1 9.25 10.125 9.125 9.5 5.74% OK
M34 1 0 1 -1 11 12.875 12.5 12.125 8.18%
M35 -1 0 1 1 8.25 7 7.625 7.625 8.20% OK
M36 1 0 1 1 10.375 9 10.25 9.875 7.70% Maybe (Watery)

Flow-ability is an index for extrudability of the mixture through the Contour Crafting nozzle
used [10]. Figure 36 shows the variation in quality of the extruded part. Four different colors
indicate different levels as follow:
- Red: Not extrudable
- Yellow: Extrudable but requires more flow-ability
78

- Green: Extrudable
- Light blue: high flow ability
- Dark blue: Vey flow-able concrete mixture

Figure 36: Extrudability at different flow-ability level
Figure 37 shows boxplot for flow-ability of three different trial for each mixture. The median,
the lower quartile as 25 percentiles, and upper quartile as 75 percentiles are plotted.

Figure 37: Boxplot of flow-ability for three different trial and average
Null hypothesis was that there is no difference in mean flow-ability across the 36 mixtures,
therefore, a preliminary calculation was performed in R to test the validity of the hypothesis.
79

Probability (<2e-16) at 0.001 level on Table 28 shows mixture has significant effect on flow-
ability. It was concluded that there was a difference in the mean for different mixtures and
rejected the null hypothesis.
Table 28: Effect of the mixture on flow-ability of the concrete
Df Sum-Sq. Mean-Sq. F-value Pr(>F)
Mixture 35 478.1 13.659 34.62 <2e-16 ***
Residuals 72 28.4 0.394

Significant codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Pairwise comparison was performed to find out where the most significant difference between
mixtures was. Figure 38 shows the result of the TukeyHSD in R statistics. Zero interval between
mixtures suggested no difference. Looking at difference in flow-ability between mixtures, for
instance M36 and M1 (M36-M1), the mean difference was 3 (M36 was higher). Lower
confidence interval and upper level at 95% were 0.94708927, 5.05291073 respectively (The
length of the intervals was the same.) and P adj. (0.0000739) showed that the difference (3) was
significant.

Figure 38: Pairwise comparison for 95% family-wise confidence level (Flow-ability)
80

The effect of different levels of main factors can be observed in Figure 39. To evaluate the effect
of factors individually, the average responses were compared based on changing the level of one
factor at a time while the level for other factors are consistent. The summary of these
comparisons is:
- Changing water to cement ratio from 45% to 55 % increase the flow-ability of the
mixtures from 17% (M9-M10) to 31% (M1-M2).
- Silica fume were added at three different levels. Replacing 5% of Portland cement with
silica fume decreased the flow-ability of the mixtures from 5% (M25-M33) to 12%
(M22-M18). Adding more silica fume to the mixture and increasing the percentage to
10% changed the flow –ability of the mixture from -9% (M25-M29) to -19% (M22-M10)
compare to mixture with Portland cement only. Flow-ability and silica fume percentage
did not follow a linear relationship while we compare the change in level 0 to level +1 to
previous changes. The maximum change was 9% decrease from M18 to M10 and the
minimum is 2% decrease for M34 to M30.

Figure 39: Main effect plot for Flow-ability (Fitted mean)
• The behavior of concrete in different levels of water to cement ratio seems to be identical
with adding superplasticizer in terms of flow-ability. Superplasticizer increase the flow-
ability of the concrete, especially addition of superplasticizer overcome the extrudability
81

issues in mixtures with fiber which we dealt with lower workability. The effect of this
factor was measured at three different levels. Adding 10cc superplasticizer increased the
flow-ability from 44% (M2-M26) up to 76% (M10-M32). The first 5cc (level -1 to 0)
increased the flow-ability between 21% (M2 to M4) to 38% (M10 to M2). Level +1
compare to 0 was 18% more flow-able (M3-M25) with maximum of 30% for M20 to
M36.
• Adding fiber to the mixture decreased the flow-ability of the mixture form 18% (M25-
M27) to 32% (M6-M10). In general, workability of concrete mixture decreased by adding
polypropylene fibers. It was also observed that fiber reduced the bleeding in concrete
mixtures. Madhavi et al [81] suggested that increase in the entrapped air voids and air
content are results of adding fiber to mixture which reduces the workability.
4.3.2. Strength-28 days
To test the strength at interface, 3 runs for each combination of factors were made in random
order. Table 29 shows the test results for 108 samples, including the average bonding strength
and the coefficients of variation (COV) which vary from 3.52% to 9.89%.

82

Table 29: Summary of Strength-28 days’ tests for three different trials
W SF SP F A B C AVERAGE COV
M1 -1 -1 -1 -1 484 430 442 452 6.27%
M2 1 -1 -1 -1 567 551 487 535 7.91%
M3 -1 -1 0 -1 485 488 572 515 9.59%
M4 1 -1 0 -1 502 571 529 534 6.51%
M5 -1 1 -1 -1 519 446 505 490 7.91%
M6 1 1 -1 -1 548 607 540 565 6.48%
M7 -1 1 0 -1 592 520 547 553 6.58%
M8 1 1 0 -1 576 671 652 633 7.94%
M9 -1 1 -1 1 547 487 472 502 7.91%
M10 1 1 -1 1 529 558 602 563 6.53%
M11 -1 1 0 1 599 582 514 565 7.96%
M12 1 1 0 1 698 619 597 638 8.32%
M13 -1 0 -1 -1 504 466 443 471 6.54%
M14 1 0 -1 -1 582 500 565 549 7.88%
M15 -1 0 0 -1 518 501 583 534 8.10%
M16 1 0 0 -1 579 509 535 541 6.54%
M17 -1 0 -1 1 512 497 440 483 7.86%
M18 1 0 -1 1 586 530 522 546 6.39%
M19 -1 0 0 1 513 584 541 546 6.55%
M20 1 0 0 1 658 565 640 621 7.94%
M21 -1 -1 -1 1 512 428 449 463 9.44%
M22 1 -1 -1 1 500 569 527 532 6.54%
M23 -1 -1 0 1 558 480 543 527 7.85%
M24 1 -1 0 1 589 558 674 607 9.89%
M25 -1 -1 1 -1 554 598 525 559 6.58%
M26 1 -1 1 -1 462 518 544 508 8.25%
M27 -1 -1 1 1 570 652 584 602 7.29%
M28 1 -1 1 1 684 601 632 639 6.56%
M29 -1 1 1 -1 648 667 572 629 7.99%
M30 1 1 1 -1 590 602 632 608 3.56%
M31 -1 1 1 1 603 634 686 641 6.54%
M32 1 1 1 1 748 661 769 726 7.89%
M33 -1 0 1 -1 601 592 637 610 3.90%
M34 1 0 1 -1 607 561 533 567 6.59%
M35 -1 0 1 1 659 566 641 622 7.93%
M36 1 0 1 1 724 691 676 697 3.52%

Figure 40 shows boxplot for interlayer strength- 28 days of three different trial of each mixture.
The median, the lower quartile as 25 percentiles, and upper quartile as 75 percentiles are plotted.
83

Figure 40: Boxplot of strength-28 days for three different trial and average
Null hypothesis was that there was no difference in means strength across the 36 mixtures,
therefore, a preliminary calculation was performed in R to test the validity of the hypothesis.
Table 30 shows mixture had significant effect on bonding strength with Probability (<9.83 e-13) at
0.001 level. Author concluded there was a difference in the mean for different mixtures and
rejected the null hypothesis.
Table 30: effect of the mixture on strength-28 days of the concrete
Df Sum Sq. Mean-Sq. F-value Pr(>F)
Mixture 35 429542 12273 7.209 9.83 e-13 ***
Residuals 72 122568 1702

Significant codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘’ 1
Pairwise comparison was performed to find out where the most significant difference between
mixtures was lied. Figure 41 shows the result of the TukeyHSD in R statistics. Those were
contained zero in interval suggested no difference. For instance, looking at difference between
M24 and M1 (M24-M1), the mean difference was 155 (M24 was higher). Lower confidence
84

interval and upper level at 95% were 20.1438449, 289.856155 respectively while the length of
the intervals was the same and P adj. (0.0076116) showed the difference of 155 was significant.

Figure 41: Pairwise comparison for 95% family-wise confidence level (Strength- 28 days)
The effect of different levels of factors on strength- 28 days are shown in Figure 42. To evaluate
the effect of factors individually, the average responses were compared based on changing the
level of a factor at a time while the level for other factors are consistent. The summary of these
comparisons is:
• Changing Water to cement ratio from 45% to 55 % increased the strength at the interface
up to 18% (M1-M2) and decreased it as low as 9% (M25-M26).
• Silica fume were added at three different levels. Silica fume strengthen the interlayer
adhesion by consuming calcium hydroxide and making the interface denser, and more
uniform. Tayeh et al. observed the formation of additional C-S-H at interface of old and
new concrete as result of the presence of silica fume in binder [69].
85

• Replacing 5% of Portland cement with silica fume increased the interlayer adhesion from
1% (M4-M16) to 12% (M26-M34). Adding more silica fume to the mixture and
increasing the percentage to 10% changed the strength at interface from 5% (M24-M12)
up to 20% (M26-M30) compare to mixture with Portland cement only. Strength and silica
fume percentage did not follow a linear relationship while we compare the change in
level 0 to level +1 to previous changes. The maximum change was 17% increase for M16
to M8 and the minimum was 3% for M14 to M6.

Figure 42: Main effect plot for Strength- 28 days (Fitted mean)
• The effect of superplasticizer was measured at three different levels. Adding 10cc
superplasticizer changed the strength from -5% (M2-M26) up to 30% (M21-M27). The
first 5cc (level -1 to 0) changed the strength between -1% (M14-M16) to 14% (M1-M3).
Change in level +1 as compared to 0 is from 14% stronger at interface (M23-M27) to 5%
weaker for M4 to M26.
• Stability of the CC structure depends on the extent of development of cracks at interfaces
of layers. Therefore, the appetite of various material to form more micro cracks during
crack propagation at interface was compared. Brittle samples were those with steep drop
at load after reaching the maximum. Implementing approaches that keep the elasticity of
the structure (e.g. fiber-reinforced concrete) can prevent significant damage. Fiber is
86

mostly used to control crack propagation in concrete structure not strengthen the
structure. Tschegg [17] investigated the bonding between old and new concrete by
comparing maximum load and fracture energy (GF). His results showed that, although
different specimen had different result when maximum load compared, they show
dramatic difference when studied in terms of fracture energy. This approach can be
applied to study the fracture energy in different samples by analyzing the strain-stress
diagram, especially on those that have fiber.
The addition of fiber changed the bonding strength from 26% (M26-M28) to -1% (M14-
M18). It appeared that the addition of fiber can increase or decrease the bonding strength
between layers. The decline in strength at interlayer adhesion can be the result of lower
cohesiveness at interface due to lower workability of the concrete or non-uniform
distribution of fibers. The ductility of mixture with fiber and its strength at interface
depends on the ability of the fibers to bridge cracks at layer. Figure 43 shows the increase
in strength at interlayer due to bridging effect of fiber at two layers in mixtures with
higher workability. In general, a ductile and gradual failure at interface was observed in
presence of fibers in mixtures.

87

Figure 43: Effect of fiber at interface in splitting test
4.4. Statistical model
R-statistics was used for computing and analyzing the data and regression technique was applied
to find the best matching polynomial function. The proposed models demonstrate the
significance of various parameters and their interactions on responses. The polynomial
regression model which describe the four input factors is given in Equation 1 [88]. As shown in
the Equation 1, interaction among all four factors were also included in the regression. In
Equation 1, β1234 is the effect of the interaction of all four factors, βijk is three factor interaction
effect between i, j, and k, βij defines the effect of two factors interaction, βi is the linear effect of
factor i, and β0 represent the intercept while Y is the predicted response.
𝑌 = 𝛽 0
+ ∑ 𝛽 𝑖 𝑘 𝑖 =1
𝑥 𝑖 + ∑ ∑ 𝛽 𝑖𝑗
𝑥 𝑖 𝑥 𝑗 + ∑ ∑ ∑ 𝛽 𝑖𝑗𝑙 𝑥 𝑖 𝑥 𝑗 𝑥 𝑙 + 𝛽 1234
𝑥 1
𝑥 2
𝑥 3
𝑥 4

𝑙 𝑗 𝑖
𝑗 +
𝑖 𝑒
𝑖 < 𝑗 < 𝑙 (1)
88

The proposed model, the strength and flow-ability are functions of the content of all variables in
the mixture, including silica fume, water, superplasticizer, and fiber. This empirical model can be
utilized to calculate the effect of factors, the significance of their interaction, and to predict the
properties of future mixtures. Therefore, correlation coefficient (R
2
) and probability (t value) of
the derived model was estimated for flow-ability and strength at 28 days.
4.4.1. Regression model (Flow-ability)
Analysis of variance (ANOVA) procedure was used to analyze the results of the experiments.
Summary of the main effects and their interactions are given in Table 31. ANOVA table
obtained from R statistics shows estimate, standard error, t-value, and the probability values
associated with individual factors and all interactions. R-squared quantifies how close the
experimental results are to the fitted regression model. Adjusted R-square uses the degree of
freedom in computation and adjusted for the number of predictors in the model. Degree of
freedom (DF) of the model is associated with the number of parameters minus intercept (15). F-
statistic is the mean square of the model divided by the mean square of the error. Pr (>|t|) is the
null hypothesis probability, which all regression parameters, except intercept, are zero.

89

Table 31: Estimated coefficient and significance of factors for modeling flow-ability
Estimate Std. Error t value Pr(>|t|)
(Intercept) 8.076389 0.054328 148.658 < 2e-16 ***
W 0.986111 0.054328 18.151 < 2e-16 ***
SF -0.526042 0.066539 -7.906 5.65E-12 ***
SP 1.755208 0.066539 26.379 < 2e-16 ***
F -1.0625 0.054328 -19.557 < 2e-16 ***
W: SF -0.088542 0.066539 -1.331 0.186582

W: SP 0.255208 0.066539 3.835 2.29E-04 ***
SF: SP -0.023438 0.081493 -0.288 0.774298

W: F -0.166667 0.054328 -3.068 2.83E-03 **
SF: F -0.005208 0.066539 -0.078 0.937779

SP: F -0.015625 0.066539 -0.235 0.814865

W: SF: SP 0.039063 0.081493 0.479 0.632836

W: SF: F -0.005208 0.066539 -0.078 0.937779

W: SP: F 0.026042 0.066539 0.391 0.696423

SF: SP: F -0.023437 0.081493 -0.288 0.774298

W: SF: SP: F 0.007812 0.081493 0.096 0.923834

Significant codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Residual standard error: 0.5646 on 92 degrees of freedom
Multiple R-squared: 0.9421, Adjusted R-squared: 0.9327
F-statistic: 99.79 on 15 and 92 DF, p-value: < 2.2e-16

Figure 44 shows the two-way interaction effects. Polynomial regression analysis was used to
determine the relationship among factors while significant parameters are those with probability
less than 0.05. In Table 31, all significant factors with Pr <0.05 including coefficient of the linear
model found by least square are marked gray. The summary of the ANOVA table is:
• Main factors water, silica fume, superplasticizer, and fiber are statistically significant.
• Figure 44 shows two-way interaction water: superplasticizer and water: fiber are
statistically significant, other interactions are not.
90

Figure 44: Two-way interaction plot (Flow-ability)
To find out the proper model, the coefficients were entered the model and then the least
significant factors were removed until all remaining coefficient had the probability of less than
0.05. The fitted regression model is shown in equation 2 below [10]:
Mf= Flow= 8.076+0.986 W-0.526 SF+1.755 SP-1.063 F+0.255 W*SP-0.167 W*F (2)
4.4.2. Diagnostic plots (Flow-ability)
There are few assumptions with fitting a linear regression model:
• The response variable (Y) can be expressed as a linear function of the independent
variables (X).
• The errors are independent.
• Variation of observations around the regression line is constant and no obvious pattern is
observed (homoscedasticity).
• Errors are normally distributed for any given value of X.
91

Figure 45 shows residual vs fitted plot to conduct a residual analysis. This is the most important
and frequently used plot to look at. Fitted values (𝑦 ̂) are Flow-ability (response) predicted by the
model and residual (e) is the deviation from these fitted values. Figure 45 indicates a distribution
of dots scattered randomly around 0 which positive values on Y axis shows the prediction was
low, negative values interpret as higher prediction, and 0 means the correct estimation based on
suggested model.

Figure 45: Residual plots (Flow-ability)
Residual plot indicate that linear regression model describes the relationship between factors and
flow. In general, there is not any clear pattern in graph. However, as can be seen in Table 32,
there is slight variation (increasing in residuals’ absolute value) as the predicted flow moves
from smaller value to larger values.
Table 32: Residuals analysis (Flow-ability)
Min 1Q Median 3Q Max
-1.29167 -0.36632 0.03125 0.42274 0.78125

Figure 46 shows normal Q-Q plot. Q-Q plot sort observed standardized residuals and theoretical
residuals and plot them against each other in ascending order. The graph is a scatter plot of two
92

sets of quantiles on X (theoretical quantile) and Y (standardizes residual) axis, and it is a
technique for assessing whether data sets are normally distributed or not. If they were normally
distributed they approximately lie on a straight line. In R statistics, the line goes through the first
and third quartiles. Even though, larger values departure from straight line and they are not as
extreme as would be expected, there is no significant deviation from straight line, so both
quantiles are normally distributed.

Figure 46: Normal Q-Q plot (Flow-ability)
Figure 47 shows scale-location and residual vs. leverage plots which respectively present non-
linearity pattern in residuals and points with greatest influence on regression model. The scale-
location graph is a scatter plot of the square root of the absolute value of the residuals (Y axis)
against the fitted value (X axis). In a good model, residuals are equally distributed around a
horizontal line. There is no distinct pattern in Figure 47 which indicates that the relationship
between response variable and independent variables were explained by linear model.
93

Figure 47: Scale-Location and Residual vs. Leverage plot (Flow-ability)
Residual vs leverage present potential influence of the observed responses on predicted response
(leverage) on X axis against the absolute value of the residuals on Y axis. In this graph, patterns
are irrelevant unlike other diagnosis plots and points outside of cook’s distance are influential to
the egression. In graph Cook’s distance is not observed which indicates there is no influential
case in the model.
4.4.3. Regression model (Strength-28 days)
To determine the relationship among factors and their interactions polynomial regression
analysis (Equation. 1) was used. The polynomial regression model contains intercept, four main
effect terms, six two-factor interaction terms, four three-factor interactions, and a four-factor
interaction term (16 parameters). Equation 3 shows the regression model obtained from R
statistics after eliminating least significant terms (Pr>0.05).
M s-28 = Strength at 28 days= 565.917+23.472 W+26.667 SF+52.375 SP+18.528 F-
14.458 W*SP+ 10.861 W*F+17.458 SP*F +14.792 W*SP*F (3)
94

The analysis of variance (ANOVA) procedure was used to analyze the results of the
experiments. Table 33 gives the summary of main effects and their interactions. It also shows the
estimated effect and coefficient for all factors and their interaction. Coefficient of correlation
(R
2
) measured the accuracy of the fitted model.
Table 33: Estimated coefficient and significance of factors for modeling strength-28 days
Estimate Std. Error t value Pr(>|t|)
(Intercept) 565.917 3.632 155.834 < 2e-16 ***
W 23.472 3.632 6.463 4.82E-09 ***
SF 26.667 4.448 5.996 3.93E-08 ***
SP 52.375 4.448 11.776 < 2e-16 ***
F 18.528 3.632 5.102 0.0000018 ***
W: SF 4.833 4.448 1.087 0.280007

W: SP -14.458 4.448 -3.251 0.001609 **
SF: SP 9.875 5.447 1.813 0.07312 .
W: F 10.861 3.632 2.991 0.00357 **
SF: F -4.583 4.448 -1.03 0.30548

SP: F 17.458 4.448 3.925 0.000167 ***
W: SF: SP 5.875 5.447 1.079 0.283624

W: SF: F -2.083 4.448 -0.468 0.640602

W: SP: F 14.792 4.448 3.326 0.001268 **
SF: SP: F -2.875 5.447 -0.528 0.59892

W: SF: SP: F 1.125 5.447 0.207 0.836838

Significant codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 37.74 on 92 degrees of freedom
Multiple R-squared: 0.7627, Adjusted R-squared: 0.724
F-statistic: 19.71 on 15 and 92 DF, p-value: < 2.2e-16

Estimate, standard error, t-value, and probability associated with individual factors and all
interactions are shown in the ANOVA table. All significant factors with Pr <0.05 were marked
gray in Table 33. The summary of the table is:
• Adjusted R-squared is 0.724
95

• Main factors; water, silica Fume, superplasticizer, and fiber are statistically significant.
• Two-way interaction; Figure 48 shows Water: Superplasticizer, Water: Fiber, and
Superplasticizer: Fiber are statistically significant. Other two-way interactions are not.

Figure 48: Two-way interaction plot (Strength-28days)
• Three-way interaction of Water: Superplasticizer: Fiber is statistically significant. Other
three-way interactions are not.
• Four-way interaction: Water, Silica fume, superplasticizer, and fiber is not significant.
4.4.4. Diagnostic plots (Strength-28 days)
Figure 49 shows residual vs fitted plot. Fitted values are strength-28 days (response) predicted by
the model on the horizontal axis and residual is the deviation from these fitted values on the
vertical axis (Equation.4). The residual vs. fitted plot is used to check the linearity and
homoscedasticity assumptions. This is the most important and frequent plot to look at to check
the validity of the model.

96

Residual= Observed value – Predicted value
𝑒 = 𝑦 − 𝑦 ̂ (4)
If the variance of the error term (residuals) is constant (∑ 𝑒 = 0) and the mean is zero (𝑒 ̅ = 0),
the predicted linear regression model is valid. Figure 49 shows the residuals are scattered
randomly like cloud of points around zero and there is not any noticeable pattern in plot
(homoscedasticity), so the assumption that the mean of residuals is zero and the errors have
constant variance sounds reasonable. This pattern indicates that the predicted linear regression
model is credible.

Figure 49: Residual plots (Strength-28 days)
Table 34 shows the symmetrical distribution of errors which means the error is unpredicted and
explanatory information are not in the error.
Table 34: Residual Analysis (Strength-28 days)
Min 1Q Median 3Q Max
-66.778 -28.528 -4.806 32.014 80.056

Figure 50 shows normal Q-Q plot to analyze if the set of observations for predicted model of
strength-28 days are normally distributed. Y axis is ordered observed standardized residuals and
97

X axis is ordered theoretical residuals (Expected residuals if the residuals are normally
distributed). In Q-Q plot the theoretically expected value for each data point based on normal
distribution is calculated. If set of observation is normally distributed a normal quantile-quantile
(Q-Q) plot of the observations will result in an approximately straight line. Even though smaller
and larger values departure from straight line and largest values are not as large as would be
expected and smallest values are not as extreme as would be predicted, no significant deviation
from straight line was observed. Therefore, both quantiles are normally distributed.

Figure 50: Normal Q-Q plot (Strength-28 days)
Figure 51 shows scale-location plot (square rooted standardized residual vs. predicted value)
which present whether there is non-linearity pattern in residuals and check the assumption of
homoscedasticity. In a good model, like residual plot there should not be obvious pattern in the
plot and residuals are equally distributed around a horizontal line. There is no distinct trend in
Figure 51 and the graph is a scatter plot where residuals are independently distributed. This
indicates that the relationship between response variable and independent variables were
98

explained by linear model. Figure 51 also shows residual vs. leverage plot which present the
influence of each observatuion on regression coefficient. Any observaion larger than Cook’s
distance represent a highly influential point and should be studied. In Figure 51 cook’s distance
is not observed which indicates there is no influential case in the model.

Figure 51: Scale-Location and Residual vs. Leverage plot (Strength-28 days)
4.5. Result and discussion
Mechanical interlocking of the material from overlay to substrate allow the flow to fill open
pores and cavities on the top surface of the substrate. In interlayer adhesion’s test, all specimens
failed in the plane of the interface with very little material attached to the layers. As can be seen
in Figure 52, broken samples observed under a microscope shows thin layers of substrate that are
caused by mechanical interlocking between layers. The average thickness of these thin layers is
about 0.42 mm (almost 1/64 in).

99

Figure 52: Micro scale analysis of a broken sample under splitting test
The proposed models can be used to build blocks for experimenting different materials.
Moreover, comparison of different responses and test methods can be studied (e.g. compressive
strength vs. interlayer adhesion) to identify possible trends and enhance the quality of the
extruded part. Previous research on the influence of the new concrete’s compressive strength on
interface shear bond strength between old and new concrete found a constant ratio between shear
strength bond and compressive strength. Based on this result, Beushausen et al. [15] concluded
that at the interface, there is transition zone similar to the transition zone between aggregate and
cement. Their result indicated that the interface by itself is not the weakest part of the bond
between layers, but the material strength plays an important role too. In this experiment, the bond
tensile strength of two layers of the same material freshly deposited on top of each other were
subjects of investigation. To analyze the relationship between compressive strength of the
concrete and interlayer adhesion when deposited in two separated layers, the results from
compression test should be compared with interlayer adhesion tests.
100

An appropriate model should be able to predict the fundamental characteristic of the strength at
interlayer and flow-ability rather than specific test. The accuracy of the model is limited to the
level of the selected factors and large deviations from the set of materials will lower the quality
of the model. The model assumed that the concrete within the specimen is homogenous.
However, variability in concrete parameters, testing condition, and test set-up and procedure
cannot be incorporated into the model when there is not sufficient data.
4.5.1. Limitations
The accuracy of the predicted response will decrease with changes in materials. Chemical
properties, grain-size distribution, water, and admixture requirements will be affected because of
any modification of cementitious material. For instance, a mixture can have Metakaolin instead
of the whole or part of a portion of silica fume, which likely will not be predictable with the
proposed model. However, the models can be used to optimize and simulate for mixtures of
different materials, or the same approach can be applied to mixtures of other materials.
Even though there might be experimental data that defines the strength of interface, the random
nature of concrete mixtures makes the characterization of the modeling of the interface in
Contour Crafted wall a complex task. One of the application of the proposed models is the
establishment of relationship between flow-ability and 28-day strength. Figure 53 shows the
average flow-ability on X axis and average 28-days strength of three specimens on Y axis.
Maximum strength recorded within 6.5 up to 11 for flow-ability but no specific equation was
derived from the presented data.
101

Figure 53: Average flow-ability vs. Average strength-28 days
Khayat [65] modeled self-consolidating concrete and his result was acceptable in the range of the
experiment, but not extending to levels of the beyond the initial levels, which limited the
accuracy of the model. Mohamed’s investigation on effect of polypropylene fiber showed
significant increase in ductility and slight increase in compressive strength when 0.5% fiber (by
volume) were added. However, further increase of fiber volumes did not increase the bond
strength [83]. Alsadey carried out tests for different dosage of superplasticizer to determine the
optimum amount for compressive strength. He reported that superplasticizer increase the
compressive strength of the concert, on the other hand over dosage has reverse effect and
reduced the strength significantly [89]. The effect of silica fume appeared to be consistent and
increased the compressive and flexural strength of the concrete even in high slump mixtures
[76].
A sophisticated model should characterize the interface and flow-ability under different possible
combinations. The proposed model can respond to a limited number of mixtures in a specified
102

range. Considering the applicability of the mixture for the Contour Crafting machine, the desired
range of flow-ability is between 7.5 to 9 inches.
4.5.2. Response optimization, the desirability approach
The desirability approach is a widely used method to optimize the multiple response process
based on finding operation conditions for most desirable responses [90]. Analyzing mathematical
models for Mf and Ms-28 indicates some trade off among responses since extrudability of the
mixtures shall be satisfied to be applicable in CC.
Maximum strength was optimized after narrowing down the range of flow-ability to acceptable
extruded parts. Experimental results showed that the extruded parts are acceptable in the range of
7 to 9.5 for flow-ability while the best quality was achieved in the range of 7.5 to 8.5. The
maximum strength at this range is 641 psi with an average of 582 psi. Table 35 shows that 8 as
center point of this range was selected as the target for flow-ability and the target response for
strength was 650.
Table 35: Target responses for composite desirability
Response Goal Lower Target Upper Weight Importance
Strength-28 days Target 550 650 750 1 1
Flow-ability Target 7 8 9 1 1

Composite desirability is calculated by combining the individual desirability values. Table 36
shows four optimal values for predicted strength and flow-ability in the defined range. Solution
one indicates the maximum composite desirability and fitted values were predicted based on Mf
and Ms-28.

103

Table 36: Predictions of multiple optimal values and responses
Solution W SF SP F Fit (Strength-28 days) Fit (Flow-ability) Composite Desirability
1 1 0 0 1 618.78 7.83 0.70
2 -1 0 1 1 619.61 7.69 0.68
3 -1 -1 1 1 592.94 8.22 0.56
4 1 -1 0 1 592.11 8.36 0.55

Figure 54 shows individual desirability for Strength-28 days and flow-ability (d), the composite
desirability (D) and optimal values which were calculated in Minitab. This combination of levels
for factors provide the maximum desirable values for responses. Comparing physical failure and
responses of different samples would validate the mathematical model.

Figure 54: Optimization plot
4.6. Validation of optimization results
To validate the analytical model, the experimental values versus the predicted value were
compared. Figure 55 shows the predicted value on X axis versus the average three observe
104

experiment on Y axis. The predicted results are close to the observed values which is a validation
of the empirical model. There is a very small error in predicted value especially when compared
to average. These errors are larger when fitted model compared to 3 different measured results of
experiments.

Figure 55: Predicted vs. average observed values
To validate the predicted model, mixture based on optimization results was selected for further
investigation. The predicted interlayer adhesion at 28 days based on the proposed empirical
model (Ms-28) is 618.78 psi. and the observed results were six additional specimens.
Table 37 shows the property of the concrete mixtures which was based on optimization results
(0.55 W/C ratio), the mid level of silica fume (5% of the binder when total cementitious material
is 1000 gr), the mid level of superplasticizer (5 CC), the high level of fiber (5gr), and 2400-gram
sand.

105

Table 37: Property of validation mixture
Factor Level Unit
Sand S 2400 Gr _
Cement C 950 Gr _
Silica Fume SF1 50 Gr 0
Superplasticizer SP2 5 CC 0
Fiber F2 5 Gr 1
W/C W2 0.55 W/C 1
Total cementitious material is 1000 gr.

For each sample, mixtures based on the optimum operating point of analytical models were
prepared. The strain- stress graphs are shown in Figure 56 which present the failure of specimens
used in this validation. The response of specimen subjected to monolithically increasing loading
in tension is characterized by initiation, opening, and propagation of crack at interface. Here, the
stress normal to the orientation of interface is tensile which opens the crack and deteriorates the
specimen at plane of interface. Linear-elastic response of the specimen demonstrated the
initiation of micro cracks at interface and pick tensile strength occurs when the samples lost load
capacity. Specimens fail under unrecoverable deformation upon loading beyond yield limit while
continuous cracks were observed at interface and they remained open (plasticity). Experimental
graphs do not show significant variability in recorded data. The observed variability can be result
of variation in fabrication and testing conditions such as, temperature and humidity.
106

Figure 56: Strain- Stress graph of the experimental results for the validation of the model
The maximum stress, average, and coefficient of variation are shown in Table 38. The average
maximum strength at interface of 6 samples was 607.5 with coefficient of variation of 1.88%.
Experimental data describes the response of specimen subjected to uniaxial compression stress at
interface. The size of the specimen is large enough (considering the maximum aggregate size)
which we can assume the concrete mixture is homogenously distributed over the interface.
Table 38: Experimental results for the validation of the model
V1 V2 V3 V4 V5 V6 Average COV
625 618 607 598 591 606 607.5 1.88%
107

Table 39 shows the confidence interval for optimized responses. All six actual results except V5
fall within 95 percent prediction interval of 618.78 psi which can be interpreted as validity of the
model. Moreover, the average of experimental results (607.5 psi) are close to the predicted
response (618.78 psi).
Table 39: Fitted solution for response optimization
Response Fit SE Fit 0.95 CI
Strength-28 days 618.78 13.90 (591.23, 646.33)
Flow-ability 7.83 0.17 (7.499, 8.161)

The experimenal results validated the empirical model which indicate that the sample size used
in design of experiment was large enough to estimate statistically significant effects. Moreover,
the model explains the significance of different factors and their interactions sufficiently and the
fitted regession equation showed a good fit to the model, specially for optimal parameters.
4.7. Summary
Contour Crafting is a new additive manufacturing method that constructs in a layer by layer
fabrication process. In this chapter, the full factorial experiment and optimization of concrete
mixture proportions related to flow-ability and strength at interface of two layers were performed
by using design of experiment. The goal of this research was to introduce an accurate and
practical model through simplifying of the model which can predict the strength at interface and
flow-ability within a specific range of concrete components properties.
A full factorial design was applied to derive the statistical model for predicting the behavior of
the mixture in a specified range of factors. Experiments, key factors and their interactions
identification, empirical model development, and validation were performed to optimize the
required properties of the mixture. It was concluded that to obtain the maximum strength at
108

interface, high level of water (+1), the high level of silica fume (+1), the high level of
superplasticizer (+1), and the high level of fiber (+1) are needed. However, this mixture was not
optimized for extrudability. After applying an optimization method for flow-ability, the
following conclusions were obtained:
• Optimized mixture is at: W/C (+1), silica fume (0), superplasticizer (0), and fiber (+1).
• Interlayer adhesion and its development in time are greatly dependent on the flow-ability
of the mixture.
• Adding silica fume while keeping the total cementitious material constant increases the
strength at the interface.
• The recommended approach is to use the existing model to estimate and optimize the
design, and to compare the predicted results with the results of other experiments to
quantify the deviation.
• This study provides a tool for improvement of concrete mixture for Contour Crafting.
The outcome can be used to increase accuracy of the extruded part and interlayer
adhesion. The next step will be to develop a new model based on modified material.

109

5. Conclusion
5.1. Research contribution
Results from the present study as well as other Contour Crafting research indicate a very
promising future for CC process. CC will be a competitive process because of its ability to apply
in variety of locations, cost, and speed. The contributions of the research explained in this
dissertation, to develop an innovative approach to enhance the structural integrity of CC
structure, are listed below:
• Effect of the aggregate size on early strength development of the concrete mixture was
studied.
• The impact of the fabrication parameters, including extrusion rate, layer thickness, and
layer width on the interface bond strength was analyzed.
• The applicability of different mixtures to the Contour Crafting process was examined
because of the sensitivity of the delivery and extrusion system to the mixture.
• The effect of interlocking on bond strength between layers was investigated.
• Strength of the bond for different size of interlock is studied and optimum size was
achieved.
• Full factorial design was used to identify and analyze the impact of binder, water to
cement ratio, superplasticizer, and fiber on the extrudability and interface bond strength.
• Significant factors and their interactions were estimated and an empirical model was
proposed.
• The condition for maximum desirability was identified and the optimized model was
validated.
110

• A specially formulated concrete mixture which can be extruded, layered without
framework, and achieve required strength at layer interfaces within 10 inch/hour vertical
fabrication rate was demonstrated.
5.2. Research methodology
In recent years, there has been a tremendous effort in 3D printing of concrete to develop models
to predict the response of the concrete specimen in different conditions (e.g. loading, flow-
ability). However, issues and characteristics of the CC built structure requires the development
of a specific model. The case of mixture development for CC requires solution for:
• Extrudability of the concrete mixture
• Shape stability of the extruded part
• Strength of fresh concrete under compression loads from the layer’s above
• Strength of the CC structure under compression, tension, and shear stress
• Evolution of failure at interface of two layers
The experimental investigations incorporated the characteristic of aggregates (size and aggregate
to cement ratio) and developed a mixture within constrains of the CC nozzle. The preliminary
result showed dependency to the aggregate size, interlocking, and size of specimen.
Statistical models were applied to experiments involving varied combination of concrete
components for more accurate prediction. To enhance the efficiency of the proposed solution,
this work has a suggested strength model coupled with flow-ability to study the consistency in
parameters associated with the CC nozzle. Therefore, the quality of the extruded part was
updated in relationship to the state of the material. Observation of the solutions showed that the
111

coupling between statistical models were significant and the condition which resulted in the
maximum strength at interface satisfied the required flow-ability for CC nozzle.
Results indicate a very promising approach for enhancing the integrity of CC structures while
controlling the quality of the extruded material and its shape. However, the variability in
concrete parameters, testing condition, and test set-up and procedure cannot be incorporated into
the model when there is not sufficient data available. The accuracy of the model is limited to the
level of the selected factors and large deviations between portions of materials in different
experiments lower the reliability of the model’s output.
5.3. Result summary and applications in practice
This research has demonstrated a special concrete mixture which develops sufficient early
strength to carry the load of layers above and achieved the required strength at layer interfaces.
Moreover, the procedure for engineering of the material for CC, given different constraints like
extrudability, early strength development, and structural integrity was outlined.
First, compressive and tensile strength was significantly improved through modifying the
aggregate size and cement to aggregate ratio. The mixture was compatible with an existing
extrusion system and it was tested to quantify the strength at interface for different extrusion
rates and layer thickness. The results showed that an improvement of up to 16% at the layer
interface is achievable for different fabrication conditions. After developing a concrete mixture,
different interlock configurations were tested. The results showed that bonding strengths is
sensitive to interlocking and it can be increased using interlock by an average of 26% as shown
by splitting test.
112

Design of experiment was used to quantify the significance of different factors in concrete
mixtures. After the experiments were run, the collected data was analyzed using R statistics. R
provided ANOVA which was studied for significant factors and their interactions. At the end, the
condition for maximum desirability was identified and the model was validated. The model
predicted the fundamental effect of different factors within the appropriate range on the stiffness
at plane of interface of two layers of concrete of CC wall in a variety of situations.
The goal of this study was to identify the mechanism to achieve the ultimate strength at the
interface of two layers of fresh concrete deposited on top of each other by Contour Crafting.
However, in practice, it is important to understand how to approach the development of a
suitable mixture considering other constraints like cost and availability. A suitable mixture can
be developed by following the recommendations below:
- Vertical build rate is a function of initial strength development which is a factor of water
to cement ratio, aggregate grading, and chemical admixture. So, an initial concrete
mixture in which the size of aggregates is modified to be compatible with the available
extrusion system should be considered.
- Material rate and initial strength should be measured to calibrate the material maturity
model for different conditions.
- Coupling the flow-ability and structural analysis is very important due to the sensitivity
of the nozzle to the over flowing or clogging. Therefore, the tradeoff between different
dependent factors (extrudability, shape stability, and strength) should be considered to
maximize the quality of the extrusion.
113

- The optimized mixture which is compatible with current nozzle provides a guide line for
industry. Parameters can be manipulated by means of additives to adopt required strength
at structure or different curing conditions.
- It is recommended that to apply the empirical model introduced in this research as
foundation for development of a model to characterize the interlayer adhesion in CC
structure and extrudability based on concrete components.
5.4. Suggested future research
Even though there might be experimental data that can define the strength of interface in Contour
Crafted wall, the random nature of concrete mixtures makes the characterization of modeling
complex. A systematic approach was perused in this research to optimize the concrete mixture
for higher strength at interface. However, this thesis has focused on a limited number of concrete
mixture’s component. Based on required structure properties many other materials combination,
like grain-size distribution, binder, and VMA may be studied. For instance, a mixture can have
metakaolin instead of the whole or part of a portion of silica fume, which likely will not be
predictable with the proposed model. Moreover, other optimization methods, such as Taguchi,
can be applied to the proposed material model.
5.4.1. Physical properties of the extruded part
After developing the mixture with optimized strength at interface of layers and compatible with
current CC nozzle, physical properties of extruded part should be studied. Factors that affect
accuracy of the part and roughness of the surface are similar in many ways, so the physical
properties of the extruded part can be investigated in one extruded layer and multiple extruded
layers. Sharpness of the edge, and smoothness of the sides (especially top of the extruded part)
should be measured to quantify the quality of the extruded part.
114

Flow of the material from overlay to substrate and filling these open pores and cavities between
two layers mechanically attaches layers together. On the micro scale, open pores and cavities,
and on the macro scale, surface texture exists. Laser scanner may be used to measure the
roughness at top surface.
All three surfaces of each extruded specimen (two sides and top surface) should be studied
individually. Roughness of the top surface should be evaluated and compared between different
samples and with sides of other areas of the same sample to find any possible relationship.
Factors and operational range can be selected for design of experiment based on the procedure
demonstrated in this research (Effects of mixture components on interlayer adhesion in Contour
Crafting).
Table 40: Design of experiment (DOE) for surface roughness and part accuracy measurement
Surface Roughness & Extrusion Quality (One Layer)
Variables
Flow-ability
W/C ratio
Superplasticizer
Viscosity Modifying Admixture
Extrusion Rate -
Nozzle Speed -
Measurement
Layer Deformation
Width
Height (Slump)
Surface Roughness -
Fresh Unit Weight -
Surface Roughness & Extrusion Quality (Multi-Layer)
Variables
Rapid Set -
Time interval between layers -
Measurement
Layer Deformation
Width
Height (Slump)
Interlayer Adhesion -

Table 40 shows factors and responses for DOE to evaluate the quality of the extruded part. In the
first step, one layer should be extruded and factors to consider are flow-ability of the mixture,
rate of extrusion, and horizontal speed of the nozzle. The original mixture can be based on the
115

optimized mixture introduced in this research. It should be noted that high degree of water
reducer and high water to cement ratio can lead to bleeding and segregation in mixture and
adding viscosity modifying admixture (VMA) can increase the homogeneity of the mixture.
Other researchers combined the use of high-range water reducer and viscosity-enhancing agent
(VEA) to control the viscosity of self-compacting concrete [65]. Nozzle is an important factor,
since the height of the side trowel limits the maximum height of layer in CC [7]. In the second
step, rapid set and time intervals between layers should be investigated for multi-layer
fabrication.
5.4.2. Analysis of crack formation and crack propagation in Contour Crafted wall
Sliding, mixed mode, and pure tension are three cracking failure modes. Compressive stress on
interface causes the sliding mode when it meets the shear strength limit, pure tension occurs
when the interface is under tensile stress without any shear stress, and mixed models are the
result of combination of these two stresses. The assumption is that the failure at interface take
place under a plane deformation (deformation is restricted to the interface plane) and after
normal stress (σ) and tangential stress (τ) pass the cracking failure envelope that limits the crack
formations. Mohr’s failure hypothesis describes the condition as:
F (σ, τ) =0
According to Tschegg [17], measurement of maximum load alone is not a sufficient indication of
the interface strength for old and new concrete, and predictions of the crack propagation and the
investigation of the fracture energy are required. The experimental data presented in the
preceding sections provided the response of CC structure subjected to compressive and tensile
strength. The localized damage was the result of formation and propagation of crack to the
116

orientation of the stress. Under this condition the ultimate capacity of the structure depends on
the response of the interface of layers under compressive and tensile stress.
A proposed approach is based on investigation of several fracture tests and their integration in to
finite element analysis. The final method should be applied in results of splitting prism tests from
this report. The method must be able to explain formation of new cracks because of external
loads, growth of existing cracks, and their propagation at bond interface.
The analytical model should be based on theoretical models that are supported by mechanical
test. Fundamental rules for crack formation and propagation come from fracture mechanics, and
the final model should be used to study more complicated structures.

117

6. Bibliography
[1] http://www.contourcrafting.org, Contour Crafting Robotic Construction System, (2016).
[2] B. Zareiyan, B. Khoshnevis, Effects of interlocking on interlayer adhesion and strength of
structures in 3D printing of concrete, Automation in Construction. 83 (2017) 212–221.
doi:10.1016/j.autcon.2017.08.019.
[3] R. Buswell, R. Soar, A. Gibb, A. Thorpe, Freeform construction: Mega-scale rapid
manufacturing for construction, Automation in Construction. 16 (2007) 224–231.
doi:10.1016/j.autcon.2006.05.002.
[4] O. Davtalab, Benefits of Real-Time Data Driven BIM for FM Departments in Operations
Control and Maintenance, in: Computing in Civil Engineering 2017, American Society of
Civil Engineers, Reston, VA, 2017: pp. 202–210. doi:10.1061/9780784480823.025.
[5] B. Zareiyan, Performance of roof materials high SRI, low SRI, and green roof in
California climate zone 8, Los Angeles, California, Uniersity of Southern California,
2011.
[6] S. Ahmadi-Karvigh, B. Becerik-Gerber, L. Soibelman, A framework for allocating
personalized appliance-level disaggregated electricity consumption to daily activities,
Energy and Buildings. 111 (2016) 337–350. doi:10.1016/j.enbuild.2015.11.029.
[7] B. Khoshnevis, D. Hwang, K.T. Yao, Z. Yeh, Mega-scale fabrication by Contour Crafting,
International Journal of Industrial and Systems Engineering. 1 (2006) 301.
doi:10.1504/IJISE.2006.009791.
[8] B. Khoshnevis, Automated Construction By Contour Crafting – Related Robotics and
Information Technologies, Automation in Construction. 13 (2004) 5–19.
doi:10.1016/j.autcon.2003.08.012.
[9] B. Zareiyan, B. Khoshnevis, Interlayer adhesion and strength of structures in Contour
Crafting - Effects of aggregate size, extrusion rate, and layer thickness, Automation in
Construction. 81 (2017) 112–121. doi:10.1016/j.autcon.2017.06.013.
[10] B. Zareiyan, B. Khoshnevis, Effects of mixture ingredients on extrudability of concrete in
Contour Crafting, Rapid Prototyping Journal. 24 (2018). doi:10.1108/RPJ-01-2017-0006.
[11] R.J. Russell II, An Experimental Approach to Analyzing Polystyrene Melt Flow When
Troweling Using the Contour Crafting Process, University of Southern California, 1999.
[12] H. Kwon, Experimentation and analysis of contour crafting (CC) process using uncured
ceramic materials, University of Southern California, 2002.
[13] D. Hwang, Experimental study of full scale concrete wall construction using contour
crafting, University of Southern California, 2005.
118

[14] T. Di Carlo, Experimental and numerical techniques to characterize structural properties
of fresh concrete relevant to contour crafting, University of Southern California, 2012.
[15] H. Beushausen, M.G. Alexander, Bond strength development between concretes of
different ages, Magazine of Concrete Research. 60 (2008) 65–74.
doi:10.1680/macr.2007.00108.
[16] X. Hui-cai, L. Geng-ying, X. Guang-jing, Microstructure model of the interfacial zone
between fresh and old concrete, Journal of Wuhan University of Technology-Mater. Sci.
Ed. 17 (2002) 64–68. doi:10.1007/BF02838421.
[17] E.K. Tschegg, S.E. Stanzl, Adhesive power measurements of bonds between old and new
concrete, Journal of Materials Science. 26 (1991) 5189–5194. doi:10.1007/BF01143212.
[18] ASTM Standard C1583, Tensile Strength of Concrete Surfaces and the Bond Strength or
Tensile Strength of Concrete Repair and Overlay Materials by Direct Tension (Pull-off
Method), ASTM International, West Conshohocken, PA. (2013).
doi:10.1520/C1583_C1583M-13.
[19] A. Momayez, M.R. Ehsani, A.A. Ramezanianpour, H. Rajaie, Comparison of methods for
evaluating bond strength between concrete substrate and repair materials, Cement and
Concrete Research. 35 (2005) 748–757. doi:10.1016/j.cemconres.2004.05.027.
[20] N.J. Delatte, Jr., D.M. Wade, D.W. Fowler, Laboratory and field testing of concrete bond
development for expedited bonded concrete overlays, ACI Materials Journal. 97 (2000)
272–280. doi:10.14359/4622.
[21] S. Akazawa, Splitting tensile test of cylindrical specimens, Journal of the Japanese Civil
Engineering Institute. 6 (1943) 12–19.
[22] ASTM Standard C496, Splitting Tensile Strength of Cylindrical Concrete Specimens,
ASTM International, West Conshohocken, PA. (2011). doi:10.1520/C0496_C0496M-11.
[23] G.E. Ramey, A.M. Strickland, An experimental evaluation of rapid-setting patching
materials used in the repair of concrete bridges and pavements, 1984.
[24] J.D. Kreigh, Arizona slant shear test: a method to determine epoxy bond strength, ACI
Journal. 73 (1976) 372–373.
[25] J.S. Wall, N.G. Shrive, Factors affecting bond between new and old concrete, ACI
Materials Journal. 85 (1988). doi:10.14359/2329.
[26] A.I. Abu-Tair, S.R. Rigden, E. Burley, Testing the bond between repair materials and
concrete substrate, ACI Materials Journal. 93 (1996). doi:10.14359/9861.
[27] J. Silfwerbrand, Shear bond strength in repaired concrete structures, Materials and
Structures. 36 (2003) 419–424. doi:10.1007/BF02481068.
119

[28] P.-W. Chen, X. Fu, D.D.L. Chung, Improving the bonding between old and new concrete
by adding carbon fibers to the new concrete, Cement and Concrete Research. 25 (1995)
491–496. doi:10.1016/0008-8846(95)00037-D.
[29] M. Sassu, Vernacular Housing Construction, University of Pisa, Italy, Edizione Tipografia
Editrice Pisana. (2002).
[30] E.L. Tolles, E.E. Kimbro, F.A. Webster, W.S. Ginell, Seismic stabilization of historic
adobe structures, 2000.
[31] P.B. Lourenco, Computational strategies for masonry structures, TU Delft, Delft
University of Technology, 1996.
[32] S.V. Polliakov, Masonry in framed buildings: an investigation into the strength and
stiffness of masonry infilling, National Lending Library for Science and Technology.
(1963).
[33] M. Holmes, Steel frames with brickwork and concrete infilling., Proceedings of the
Institution of Civil Engineers. 19 (1961) 473–478. doi:10.1680/iicep.1961.11305.
[34] B. Stafford-Smith, Behavior of square infilled frames, ASCE Journal of the Structural
Division. 92 (1966) 381–403.
[35] A.M. Reinhorn, A. Madan, R.E. Valles, Y. Reichmann, J.B. Mander, Modeling of
masonry infill panels for structural analysis, Journal of Structural Engineering. 123 (1997)
1295–1302. doi:10.1061/(ASCE)0733-9445(1997)123:10(1295).
[36] A.M. Neville, Properties of concrete, 5th ed., Pearson, England, 2011.
[37] ASTM Standard C230, Flow Table for Use in Tests of Hydraulic Cement, ASTM
International, West Conshohocken, PA. (2013) 1–6. doi:10.1520/C0230_C0230M-14.
[38] ASTM Standard C33, Concrete Aggregates, ASTM International, West Conshohocken,
PA. (2011). doi:10.1520/C0033_C0033M-13.
[39] ASTM Standard C305, Mechanical Mixing of Hydraulic Cement Pastes and Mortars of
Plastic Consistency, ASTM International, West Conshohocken, PA. (2014) 1–3.
doi:10.1520/C0305-14.
[40] A. Telesca, M. Marroccoli, M.L. Pace, M. Tomasulo, G.L. Valenti, T.R. Naik, Expansive
and non-expansive calcium sulfoaluminate-based cements, in: Claisse.info, Third
International Conference on Sustainable Construction Materials and Technologies, 2013.
[41] The Concrete Countertop Institute, CSA Cements: Rapid Strength With A Low Carbon
Footprint, (2014). http://www.concretecountertopinstitute.com/csa-cements-rapid-
strength-with-a-low-carbon-footprint/ (accessed January 6, 2016).
[42] Heartland Cement Sales Company, Material Safety Data Sheet (MSDS) for CSA cement,
120

(2013).
[43] ASTM Standard C39, Compressive Strength of Cylindrical Concrete Specimens, ASTM
International, West Conshohocken, PA. (2015) 1–7. doi:10.1520/C0039_C0039M-15A.
[44] M. Elices, C.G. Rocco, Effect of aggregate size on the fracture and mechanical properties
of a simple concrete, Engineering Fracture Mechanics. 75 (2008) 3839–3851.
doi:10.1016/j.engfracmech.2008.02.011.
[45] A. Hillerborg, Results of three comparative test series for determining the fracture energy
GF of concrete, Materials and Structures. 18 (1985) 407–413. doi:10.1007/BF02472416.
[46] ASTM Standard C266, Time of Setting of Hydraulic-Cement Paste by Gillmore Needles,
ASTM International, West Conshohocken, PA. (2013). doi:10.1520/C0266-13.
[47] ASTM Standard C191, Time of Setting of Hydraulic Cement by Vicat Needle, ASTM
International, West Conshohocken, PA. (2014) 1–8. doi:10.1520/C0191-13.
[48] Y. Estrin, A.V. Dyskin, E. Pasternak, Topological interlocking as a material design
concept, Materials Science and Engineering: C. 31 (2011) 1189–1194.
doi:10.1016/j.msec.2010.11.011.
[49] Y. Feng, T. Siegmund, E. Habtour, J. Riddick, Impact mechanics of topologically
interlocked material assemblies, International Journal of Impact Engineering. 75 (2015)
140–149. doi:10.1016/j.ijimpeng.2014.08.003.
[50] S. Khandelwal, T. Siegmund, R.J. Cipra, J.S. Bolton, Adaptive mechanical properties of
topologically interlocking material systems, Smart Materials and Structures. 24 (2015)
45037. doi:10.1088/0964-1726/24/4/045037.
[51] S. Schaare, W. Riehemann, Y. Estrin, Damping properties of an assembly of topologically
interlocked cubes, Materials Science and Engineering: A. 521–522 (2009) 380–383.
doi:10.1016/j.msea.2008.10.069.
[52] A.J. Kanel-Belov, A. V. Dyskin, Y. Estrin, E. Pasternak, I.A. Ivanov-Pogodaev,
Interlocking of convex polyhedra: towards a geometric theory of fragmented solids,
Moscow Mathematical Journal. 10 (2008) 337–342.
[53] Y. Estrin, Negative stiffness of a layer with topologically interlocked elements, Scripta
Materialia. 50 (2004) 291–294. doi:10.1016/j.scriptamat.2003.09.053.
[54] A. Molotnikov, R. Gerbrand, O. Bouaziz, Y. Estrin, Sandwich Panels with a Core
Segmented into Topologically Interlocked Elements, Advanced Engineering Materials. 15
(2013) 728–731. doi:10.1002/adem.201300002.
[55] S. Khandelwal, T. Siegmund, R.J. Cipra, J.S. Bolton, Scaling of the Elastic Behavior of
Two-Dimensional Topologically Interlocked Materials Under Transverse Loading,
Journal of Applied Mechanics. 81 (2013) 31011. doi:10.1115/1.4024907.
121

[56] A. V. Dyskin, E. Pasternak, Y. Estrin, Mortarless structures based on topological
interlocking, Frontiers of Structural and Civil Engineering. 6 (2012) 188–197.
doi:10.1007/s11709-012-0156-8.
[57] A. Mather, R. Cipra, T. Siegmund, Structural integrity during remanufacture of a
topologically interlocked material, International Journal of Structural Integrity. 3 (2012)
61–78. doi:10.1108/17579861211210009.
[58] A. Molotnikov, R. Gerbrand, Y. Qi, G.P. Simon, Y. Estrin, Design of responsive materials
using topologically interlocked elements, Smart Materials and Structures. 24 (2015)
25034. doi:10.1088/0964-1726/24/2/025034.
[59] A. V. Dyskin, Y. Estrin, A.J. Kanel-Belov, E. Pasternak, Topological interlocking of
platonic solids: A way to new materials and structures, Philosophical Magazine Letters. 83
(2003) 197–203. doi:10.1080/0950083031000065226.
[60] A. V. Dyskin, Y. Estrin, A.J. Kanel-Belov, E. Pasternak, Toughening by Fragmentation—
How Topology Helps, Advanced Engineering Materials. 3 (2001) 885. doi:10.1002/1527-
2648(200111)3:11<885::AID-ADEM885>3.0.CO;2-P.
[61] S. Khandelwal, T. Siegmund, R.J. Cipra, J.S. Bolton, Transverse loading of cellular
topologically interlocked materials, International Journal of Solids and Structures. 49
(2012) 2394–2403. doi:10.1016/j.ijsolstr.2012.04.035.
[62] F. Saucier, M. Pigeon, Durability of new-to-old concrete bondings, ACI Special
Publication. 128 (1991) 689–706. doi:10.14359/2066.
[63] A. Mansourighasri, B. Zareiyan, Experimental and numerical analysis of structures
fabricated by contour crafting, in: IIE Annual Conference & Expo, Institute of Industrial
Engineers, Construction Division, Nashville, TN, 2015: pp. 1–5. http://www.iienet2.org/.
[64] Y. Lim, M. Kim, S. Shin, V. Li, Numerical simulation for quasi-brittle interface fracture
in cementitious bi-material system, in: In Proceedings of Fourth International Conference
on Fracture Mechanics of Concrete Structures, Cachan, France:Balkema, 2001: pp. 73–80.
[65] K.H. Khayat, A. Ghezal, M.S. Hadriche, Factorial design model for proportioning self-
consolidating concrete, Materials and Structures. 32 (1999) 679–686.
doi:10.1007/BF02481706.
[66] G. Li, H. Xie, G. Xiong, Transition zone studies of new-to-old concrete with different
binders, Cement and Concrete Composites. 23 (2001) 381–387. doi:10.1016/S0958-
9465(01)00002-6.
[67] BASF, MasterLife ®SF 100, (2015). www.master-builders-solutions.basf.us.
[68] X. Cong, S. Gong, D. Darwin, S.L. Mccabe, Role of silica fume in compressive strength
of cement paste, mortar, and concrete, The University of Kansas Structural Engineering &
Materials Laboratory, 1990. doi:10.14359/2570.
122

[69] B. Tayeh, B.H. Abu bakar, M.A. Megat Johari, A.M. Zeyad, The role of silica fume in the
adhesion of concrete restoration systems, Advanced Materials Research. 626 (2013) 265–
269. doi:10.4028/www.scientific.net/AMR.626.265.
[70] H.-C. Shin, Z. Wan, Interfacial Properties between New and Old Concretes, in: Second
International Conference on Sustainable Construction Materials and Technologies,
Ancona, Italy, 2010.
[71] ASTM Standard C144, Aggregate for Masonry Mortar, ASTM International, West
Conshohocken, PA. (2011). doi:10.1520/C0144-11.
[72] F.P. Zhou, F.D. Lydon, B.I.G. Barr, Effect of coarse aggregate on elastic modulus and
compressive strength of high performance concrete, Cement and Concrete Research. 25
(1995) 177–186. doi:10.1016/0008-8846(94)00125-I.
[73] P.C. Aitcin, P.K. Mehta, Effect of Coarse Aggregate Characteristics on Mechanical
Properties of High-Strength Concrete, Materials Journal. 87 (1990) 103–107.
doi:10.14359/1882.
[74] R.D. Browne, P.B. Bamforth, Tests to Establish Concrete Pumpability, ACI Journal
Proceedings. 74 (1977) 193–203. doi:10.14359/11001.
[75] A. Mardani-Aghabaglou, M. Tuyan, G. Yılmaz, Ö. Arıöz, K. Ramyar, Effect of different
types of superplasticizer on fresh, rheological and strength properties of self-consolidating
concrete, Construction and Building Materials. 47 (2013) 1020–1025.
doi:10.1016/j.conbuildmat.2013.05.105.
[76] R. Gagn, A. Boisvert, M. Pigeon, Effect of Superplasticizer Dosage on Mechanical
Properties, Permeability, and Freeze-Thaw Durability of High-Strength Concretes With
and Without Silica Fume, Materials Journal. 93 (1996) 111–120. doi:10.14359/1407.
[77] Grace, ADVA CAST 600, (2015). www.graceconstruction.com.
[78] BASF, MasterFiber® M 100, (2015).
[79] S. Ahmed, I. Bukhari, J. Siddiqui, A study on properties of polypropylene fiber reinforced
concrete, in: 31st Conference on Our World in Concrete & Structures, Singapore, 2006:
pp. 16–17.
[80] P. Patel, A. Desai, K. Desai, Evaluation of engineering properties for polypropylene fiber
reinforced concrete, International Journal of Advanced Engineering Technology. 3 (2012)
42–45.
[81] T.C. Madhavi, L.S. Raju, D. Mathur, Polypropylene Fiber Reinforced Concrete-A
Review, International Journal of Emerging Technology and Advanced Engineering. 4
(2014) 114–119.
[82] O. Gencel, C. Ozel, W. Brostow, Mechanical properties of self-compacting concrete
123

reinforced with polypropylene fibres, Materials Research. (2011).
doi:10.1179/143307511X13018917925900.
[83] R.A.S. Mohamed, Effect of polypropylene fibers on the mechanical properties of normal
concrete, Journal of Engineering Sciences, Assiut University. 34 (2006) 1049–1059.
[84] J. Xiao, H. Falkner, On residual strength of high-performance concrete with and without
polypropylene fibres at elevated temperatures, Fire Safety Journal. 41 (2006) 115–121.
doi:10.1016/j.firesaf.2005.11.004.
[85] ASTM Standard C143, Slump of Hydraulic-Cement Concrete, ASTM International, West
Conshohocken, PA. (2015). doi:10.1520/C0143_C0143M-15A.
[86] ASTM Standard C1611, Slump Flow of Self-Consolidating Concrete, ASTM
International, West Conshohocken, PA. (2014). doi:10.1520/C1611_C1611M-14.
[87] ASTM Standard C1437, Flow of Hydraulic Cement Mortar, ASTM International, West
Conshohocken, PA. (2015). doi:10.1520/C1437-15.
[88] C.F.J. Wu, M.S. Hamada, Experiments: Planning, Analysis, and Optimization, 2nd ed.,
John Wiley & Sons, New York, USA, 2009.
[89] S. Alsadey, Effect of Superplasticizer on Fresh and Hardened Properties of Concrete,
Journal of Agricultural Science and Engineering. 1 (2015) 70–74.
[90] NIST/SEMATECH e-Handbook of Statistical Methods, (2012).
http://www.itl.nist.gov/div898/handbook/.

Abstract (if available)

Abstract Contour Crafting (CC) is a method that builds concrete structures in a layer-by-layer fabrication. During construction, layers must be bonded together to make a homogenous structure, as there is no vibration or external force during layer deposition. In this research, interlayer bonding strength in contour crafted structure is evaluated by analyzing experimental data and applying statistical techniques. The interface bond strength is studied based on controlling the properties of the material in mixture and the impact of mechanical factors (e.g. fabrication parameters, interlocking). An initial concrete mixture in which the compressive and tensile strength is significantly improved by modifying the size of aggregate and cement to aggregate ratio is developed. This mixture is compatible with existing extrusion system. First, the impact of the fabrication parameters, including extrusion rate, layer thickness, and layer width on the interface bond strength are analyzed. Secondly, different interlock configurations and sizes are tested to find the optimal adhesion between layers. Finally, a design of experiment approach is pursued to investigate the effect of different mixture’s components on the integrity of the extruded part. Some important factors that affect the mixture’s design are recognized including binder, flow-ability, and tensile stress. Therefore, full factorial design is used to study these independent variables and measure their effect on extrudability and strength at interface. Identifying an appropriate model requires design of experiments, regression modeling, optimization, and validation methods. The proposed models can balance different variables affecting extrudability and strength at interface. Moreover, understanding of significant factors in the mixture can facilitate the material optimization for CC fabrication.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Experimental and numerical techniques to characterize structural properties of fresh concrete

PDF

Contour crafting construction with sulfur concrete

PDF

Contour Crafting process planning and optimization

PDF

Analyses of strength of layered structures fabricated by Contour Crafting

PDF

Thermal and deformation analysis of multiphase sulfur concrete extrusion for planetary construction

PDF

Mixture characterization and real-time extrusion quality monitoring for construction-scale 3D printing (Contour Crafting)

PDF

A data driven software platform for process automation, planning and inspection of Contour Crafting large-scale robotic 3D printing system

PDF

3D printing of polymeric parts using Selective Separation Shaping (SSS)

PDF

Slurry based stereolithography: a solid freeform fabrication method of ceramics and composites

PDF

Energy control and material deposition methods for fast fabrication with high surface quality in additive manufacturing using photo-polymerization

$Metasurfaces in 3D applications: multiscale stereolithography and inverse design of diffractive optical elements for structured light$

PDF

Metasurfaces in 3D applications: multiscale stereolithography and inverse design of diffractive optical elements for structured light

Asset Metadata

Creator Zareiyan, Babak (author)

Core Title Evaluation of interlayer bonding strength in contour crafted structures: experimental and statistical approach

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Civil Engineering

Publication Date 10/25/2017

Defense Date 09/15/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag 3D concrete printing,additive manufacturing,automation in construction,contour crafting (CC),design of experiment,interlayer adhesion,OAI-PMH Harvest

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Khoshnevis, Behrokh (committee chair), Brandow, Gregg (committee member), Carlson, Anders (committee member), Lee, Vincent (committee member)

Creator Email babak.zareiyan@gmail.com,zareiyan@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-448343

Unique identifier UC11266336

Identifier etd-ZareiyanBa-5790.pdf (filename),usctheses-c40-448343 (legacy record id)

Legacy Identifier etd-ZareiyanBa-5790.pdf

Dmrecord 448343

Document Type Dissertation

Rights Zareiyan, Babak

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