University of Southern California Dissertations and Theses

Analyses of strength of layered structures fabricated by Contour Crafting

(USC Thesis Other)

Analyses of strength of layered structures fabricated by Contour Crafting

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ANALYSES OF STRENGTH OF LAYERED STRUCTURES FABRICATED BY
CONTOUR CRAFTING

by
Amir Mansouri

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
(Industrial and Systems Engineering)

August 2016

ii

Acknowledgments
I would like to express my special appreciation and thanks to my advisor Professor Behrokh
Khoshnevis, you have been a tremendous mentor for me. I would like to thank you for
encouraging my research and for allowing me to grow as a research scientist, tinkerer and
innovator. Your advice on both research as well as on my career have been priceless. This thesis
would not have been possible without your profound knowledge, wisdom, and encouragement.
I would like to thank Professor Yong Chen for all his support during this Ph.D. program. I
am also very grateful to Professor Edward Goo for his evaluation of my Ph.D. study.
I am also thankful to my beloved parents, Mohamadreza Mansouri and Pari Araghi for their
unconditional love and support throughout my life. I would have not been able get to this point if
I had not had you in my life. My appreciation also goes to my brother Alireza and sister Atefeh
for their love and support.
Last but not the least, I would like to thank my greatest friends; Elahe Ahmady, Mahmood
Shirooyeh, Payman Torabi, and Matthew Petros whose encouragement and emotional support
made this work possible far away from home.
Thank you All.

iii

Table of Contents
1 Chapter One: Introduction .................................................................................................. 1
1.1 Introduction to Additive Manufacturing (AM) ......................................................................... 1
1.2 Additive Manufacturing and Construction Industry ................................................................. 2
1.3 Statement of the Problem ........................................................................................................ 3
1.4 Research Purpose ..................................................................................................................... 4
1.5 Research Contribution .............................................................................................................. 5
1.6 Organization of the Dissertation .............................................................................................. 6
2 Chapter Two: Background of Study ..................................................................................... 7
2.1 Introduction to Contour Crafting .............................................................................................. 7
2.2 Concrete ................................................................................................................................... 8
2.2.1 Introduction to Concrete ................................................................................................... 8
2.2.2 Fresh Concrete Properties ................................................................................................. 9
2.2.3 Stability of Concrete ....................................................................................................... 10
2.2.4 Hardened Properties of Concrete ................................................................................... 11
2.3 Bonding Strength of Concrete ................................................................................................ 14
2.3.1 Masonry Structures ........................................................................................................ 14
2.3.2 Bonding Strength between Fresh Concrete and Old Existing Concrete Surface .............. 23
2.4 Critique of Past Approaches ................................................................................................... 25
2.5 Summary ................................................................................................................................ 26
3 Chapter Three: Research Methodology ............................................................................ 27
3.1 Introduction ............................................................................................................................ 27
3.2 Stage One: Analytical study .................................................................................................... 27
iv

3.2.1 Geometry of the Model .................................................................................................. 28
3.2.2 Boundary and Loading Condition ................................................................................... 28
3.2.3 Material .......................................................................................................................... 29
3.2.4 Elasto-Plasticity .............................................................................................................. 29
3.2.5 Simulation of Interaction ................................................................................................ 30
3.3 Stage Two: Experimental Study .............................................................................................. 32
3.3.1 Double-Cantilever Beam (DCB) Test ............................................................................... 32
3.3.2 Specimen Preparation .................................................................................................... 34
3.3.3 Material .......................................................................................................................... 35
3.3.4 Experimental Design ....................................................................................................... 36
3.4 Stage Three: Primary Analytical Analyses ............................................................................... 37
3.4.1 Finite Element Analysis ................................................................................................... 39
3.5 Summary ................................................................................................................................ 40
4 Chapter Four: Results ........................................................................................................ 42
4.1 Analysis of Stage One Results ................................................................................................. 42
4.1.1 Testing the Layer Adhesion Model .................................................................................. 42
4.1.2 Full Scale Model .............................................................................................................. 43
4.1.3 Validation ....................................................................................................................... 49
4.1.4 Summary ......................................................................................................................... 57
4.2 Analyses of Stage Two Results ................................................................................................ 57
4.2.1 Initial Experiments .......................................................................................................... 58
4.2.2 Primary Experiments ...................................................................................................... 66
4.2.3 Final Experiments ........................................................................................................... 80
4.2.4 Summary ......................................................................................................................... 85
v

4.3 Stage Three Results ................................................................................................................ 86
4.3.1 Simulation Comparison of Walls with Different Layer Bonding Strength ....................... 86
4.3.2 Determining the Minimum Bonding Strength ................................................................ 90
4.3.3 Effect of Geometry Complexity on a Structure’s Strength .............................................. 92
4.3.4 Section Summary .......................................................................................................... 114
4.4 Chapter Summary ................................................................................................................. 115
5 Concluding Remarks ........................................................................................................ 116
5.1 Conclusion ............................................................................................................................ 116
5.2 Future Work ......................................................................................................................... 120
6 References ....................................................................................................................... 121
7 Appendices ...................................................................................................................... 128
7.1 Appendix I: Curve fitting for the compliance calibration methods ...................................... 128
7.2 Appendix II: Figures of the simulations related to curvature analysis. ................................. 130

vi

List of Tables
Table 3.1. Typical Characteristics for M30 grade concrete .........................................................................29
Table 3.2. Characteristics of concrete used in the experiments ...................................................................36
Table 3.3. Boundary and geometry information ..........................................................................................40
Table 4.1. Deflection of the wall in different heights ..................................................................................44
Table 4.2. Deflection of the layered and rigid wall at different heights. .....................................................47
Table 4.3. Delamination magnitude of the first 15 layers. ...........................................................................48
Table 4.4. Delamination magnitude of the first 15 layers. ...........................................................................55
Table 4.5. Summary of initial experiments sample .....................................................................................59
Table 4.6. Summary of round 1 experiments ...............................................................................................60
Table 4.7. Experimental Design ...................................................................................................................66
Table 4.8. Coding of the primary experiment samples ................................................................................66
Table 4.9. The results of plain interface samples .........................................................................................71
Table 4.10. The growth of the initial crack for all plain interface samples .................................................74
Table 4.11. Summary of the results for PR-01, PR-02 and PR-03 samples ................................................75
Table 4.12. Summary of the results for PS-01, PS-02 and PS-03 samples ..................................................77
Table 4.13. The design of the final experiments ..........................................................................................80
Table 4.14. Summary of the results of the final experiments ......................................................................83
Table 4.15. Wall deflection for different heights .........................................................................................89
Table 4.16. Delamination for plain interface and rod reinforced walls .......................................................89
vii

Table 4.17. Predictive simulation design with results .................................................................................91
Table 4.18. Rectilinear wall deflections ......................................................................................................95
Table 4.19. Factorial levels for the simulations ...........................................................................................98
Table 4.20. Structural parameters used in different simulations .................................................................99
Table 4.21. ANOVA table of the factorial simulations .............................................................................101
Table 4.22. Structural parameters used in different simulations for the concave wall ..............................102
Table 4.23. Curved versus rectilinear deflection comparison of 10 cm thick wall ...................................103
Table 4.24. Curved versus rectilinear deflection comparison of 20 cm thick wall ...................................105
Table 4.25. Simulation plan for the curvature effectiveness study ............................................................106
Table 4.26. Simulation results plan for the curvature effectiveness study ................................................107
Table 4.27. The result of the 30 simulations ..............................................................................................108
Table 4.28. The simulation results of the corrugated wall study ...............................................................112

viii

List of Figures
Figure 1.1 The process of AM .......................................................................................................................1
Figure 1.2. The Contour Crafting Process .....................................................................................................3
Figure 2.1 Construction of Building Using Contour Crafting .......................................................................7
Figure 2.2. Bingham Model. τ
0
is the minimum shear stress needed to make the concrete flow ................10
Figure 2.3. Masonry with irregular stones on the left and representation of the regular masonry ..............15
Figure 2.4. Failure of Shear Walls ...............................................................................................................16
Figure 2.5 Common failure mode of unreinforced concrete are a) Sliding b) Diagonal failure ..................17
Figure 2.6. Diagonal crack in masonry walls with different aspect ratio ....................................................19
Figure 2.7. Typical testing of the interface. a) Tension b) Shear .................................................................20
Figure 2.8. Two of the developed experimental approaches to test the behavior of ...................................21
Figure 2.9. Putra interlocking blocks ...........................................................................................................21
Figure 2.10. Interfacial Zone .......................................................................................................................24
Figure 2.11. The wall effect is highlighted in the interface of large aggregates and cement particle. ........25
Figure 3.1. Research Methodology ..............................................................................................................27
Figure 3.2. The 3D model of the shear wall with the boundary and loading condition .............................29
Figure 3.3. The cohesive law .......................................................................................................................31
Figure 3.4. Fracture at the bond zone ..........................................................................................................31
Figure 3.5. DCB Test ...................................................................................................................................33
Figure 3.6 Dimension of the samples ..........................................................................................................34
ix

Figure 3.7. Mold used to fabricate samples .................................................................................................34
Figure 3.8. The DCB sample loaded in the Instrum 100kgf tensile machine ..............................................35
Figure 3.9. The flowchart of the DCB experiments .....................................................................................37
Figure 3.10 Roadmap of Stage Three of the study ......................................................................................38
Figure 4.1. Stress Simulation of two layers. a) unscaled b) scaled to show the deflection ........................42
Figure 4.2. Displacement distribution a) weak bonding B) strong bonding ................................................43
Figure 4.3. Strain distribution (a), stress distribution (b) .............................................................................44
Figure 4.4. The selected top nodes of one layer ..........................................................................................45
Figure 4.5. Stress distribution (a), strain distribution (b) .............................................................................46
Figure 4.6. The comparison of deflection in layered and none layered walls .............................................47
Figure 4.7. The comparison of deflection in layered and none layered walls .............................................49
Figure 4.8. The construction of the wall ......................................................................................................50
Figure 4.9. The distribution of the glue .......................................................................................................50
Figure 4.10. Final Wall ................................................................................................................................51
Figure 4.11. Loading fixture ........................................................................................................................51
Figure 4.12 The simulation of the layered concrete wall under a distributed load ......................................52
Figure 4.13. Delamination for the first 15 layers of the simulated wall ......................................................52
Figure 4.14. Delamination of the first 15 layers ..........................................................................................53
Figure 4.15. Delamination and deflection of the wall .................................................................................53
Figure 4.16. Delamination and deflection of the wall .................................................................................54
x

Figure 4.17. The model of each layers .........................................................................................................55
Figure 4.18. Simulation results of the wall. .................................................................................................56
Figure 4.19. Deflection comparison for the first 13 layers ..........................................................................56
Figure 4.20. 4 Samples of round 1. With two types of brackets and two types of glues .............................59
Figure 4.21. IH-01 Sample (with aluminum bracket and Crystalbond) ......................................................61
Figure 4.22. Delaminated layers of sample IB-02 .......................................................................................62
Figure 4.23. The top layer of delaminated samples that showed a crack. ...................................................63
Figure 4.24. IH-02 sample delamination result ...........................................................................................63
Figure 4.25. Initial crack propagation ..........................................................................................................64
Figure 4.26. Polished and threaded rods used for reinforcement ................................................................67
Figure 4.27. Distribution of the rods between layers ...................................................................................67
Figure 4.28. Threaded rods installed after the first layer is casted (PS-01 sample) ....................................68
Figure 4.29. Fabricated samples with plain interface ..................................................................................69
Figure 4.30. Fabricated samples with rod reinforcing mechanism ..............................................................69
Figure 4.31. Fabricated samples using threaded rods. .................................................................................70
Figure 4.32. The delaminated PP-02 sample ...............................................................................................72
Figure 4.33. The delaminated PP-03 sample ...............................................................................................72
Figure 4.34. The growth of initial crack ......................................................................................................73
Figure 4.35. Initial crack growth at PP-03 sample .......................................................................................73
Figure 4.36. The cross section of the top layers in all plain interface samples ...........................................74
xi

Figure 4.37. DCB result of PR-03 sample ...................................................................................................75
Figure 4.38. Initial crack growth in PR-03 sample ......................................................................................76
Figure 4.39. PS-01 sample after the DCB test .............................................................................................77
Figure 4.40. Initial crack growth in PS-01 sample ......................................................................................78
Figure 4.41. Comparison of threaded and polished rod reinforced samples ...............................................79
Figure 4.42. The position of the reinforcing rod in the samples ..................................................................81
Figure 4.43. Samples of the final experiment ..............................................................................................82
Figure 4.44. FS-01 sample after the DCB test .............................................................................................83
Figure 4.45. FS-02 sample after the DCB test .............................................................................................84
Figure 4.46. FS-03 sample after the DCB test .............................................................................................85
Figure 4.47. Suggested testing to calculate the bonding strength ................................................................86
Figure 4.48. Deflection comparison of the simulated walls ........................................................................88
Figure 4.49. Delamination comparison graph for the first 15 layers ...........................................................90
Figure 4.50. Stress and strain distribution comparison for the optimized model ........................................92
Figure 4.51. Interior of an adobe house .......................................................................................................93
Figure 4.52. Rectilinear wall simulations with the thickness of 10cm, 15cm, and 20cm ............................94
Figure 4.53. Rectilinear walls deflection comparison .................................................................................95
Figure 4.54. The simulation result of 10 cm thick rectilinear wall ..............................................................96
Figure 4.55. The simulation result of 15cm thick rectilinear wall ...............................................................96
Figure 4.56. The simulation result of 20cm thick rectilinear wall ...............................................................97
xii

Figure 4.57. Curvature parameters ...............................................................................................................98
Figure 4.58. Simulation 1 of the curvature study .........................................................................................99
Figure 4.59. Simulation 2 of the curvature study .......................................................................................100
Figure 4.60. Simulation 3 of the curvature study .......................................................................................100
Figure 4.61. Simulation 4 of the curvature study .......................................................................................100
Figure 4.62. Simulation 5 of the curvature study .......................................................................................101
Figure 4.63.Main effect plot ......................................................................................................................102
Figure 4.64. Curved versus rectilinear deflection comparison of 10 cm thick wall ..................................104
Figure 4.65. Curved versus rectilinear deflection comparison of 20 cm thick wall ..................................104
Figure 4.66. Deflection comparison of a thin curved wall and a thick rectilinear wall. ............................106
Figure 4.67. The deflection difference surface shown in a 3D plot ...........................................................109
Figure 4.68. The deflection difference surface shown in a 2D contour plot .............................................109
Figure 4.69. Corrugated wall created by Contour Crafting .......................................................................110
Figure 4.70. The cross section of the simulated hollow wall .....................................................................111
Figure 4.71. The simulation result of the walls with two different patterns ..............................................112
Figure 4.72. The deflection of the simulated walls in various heights. .....................................................113
Figure 4.73. Curved corrugated model ......................................................................................................114
Figure 4.74. The simulation of curved corrugated structure ......................................................................114
Figure 4.75. Deflection comparison of the walls with 20 thickness ..........................................................115

xiii

Abbreviations
ASTM American Society for Testing and Materials
CC Contour Crafting
CZM Cohesive cone modeling
EL Elevation
FE Finite Element
FEA Finite Element Analysis
FEM Finite Element Method
IZ Interfacial Zone
RGS Recycled Glass Sand
RA Recycled Aggregates

xiv

Notations
d Cohesion
p Applied load
q Deviatoric or Mises equivalent stress Friction
β Friction Angle
ɛ Strain
σ Stress
σ0 Yield limit for uniaxial stress-strain curve
σ1 Principal stress
σ2 Principal stress
σ3 Principal stress
σc Uniaxial compression yield stress
E Modulus of elasticity (Young’s modulus)
ν Poisson’s ratio
V Volume

xv

Abstract
The construction industry had been deprived from automation until the emergence of
additive manufacturing approaches such as Contour Crafting (CC) ) opened up new opportunities.
CC is an automatic fabrication technology on the basis of additive manufacturing principles. The
process stacks relatively thick layers of cementitious material (e.g., 25.4 mm) on top of each other
until the full scale structure is achieved. There are several problems in conventional construction
approaches, including inefficient use of labor, time, materials, and capital. All of these efficiency
problems can be addressed by CC. Besides the construction of building structures, CC is capable
of performing sub tasks such as electrical, plumping, wiring, and painting.
The operational behavior of layered structures under vertical and in-plane loads are not
known to date. To expand the application of the structures fabricated by automated approaches,
the strength and failure mechanisms of these structures must be completely understood. The
adhesion strength between the layers is one of the structural concerns in CC that should be
studied and enhanced if needed to achieve the desired structural properties. In this research, a
methodology to analyze the structural properties of specimens made by CC was developed. The
methodology includes both experimental and numerical analyses that objectively study the
bonding strength of concrete layers to each other and predict the behavior of Contour Crafting
structures. The interface of layers is simulated based on the Cohesive Zone Modeling (CZM)
approach. The simulation is followed and calibrated by experimental procedures to quantify the
bonding strength of the layers. A final simulation model to investigate the strength of the layers is
developed based on experimental results. The study also includes the effect of incorporating
geometrical complexity on the behavior of CC structures. The model investigates and compares
curved, rectilinear, hollow, and curved hollow structures. Results and findings of this study show
the effectiveness of incorporating curvature as compared with rectilinear walls. The results also
show the significance of corrugated structures.
1

1 Chapter One: Introduction
1.1 Introduction to Additive Manufacturing (AM)
The major manufacturing approaches include 1) manufacturing by material removal such as
lathing, 2) manufacturing by forming the material such as forging, and 3) manufacturing by
adding material such as casting. Additive manufacturing (AM), which fits in the third approach,
is a Computer Aided Manufacturing (CAM) process where the part is being built in a layer by
layer fashion from a 3D CAD model. Figure 1.1 shows the process of the AM.

Figure 1.1 The process of AM

AM has some advantages and disadvantages over conventional manufacturing approaches.
The advantages are:
• The ability of manufacturing complex parts
• Complexity does not add to the cost of fabrication
• Low setup time
• Having no or minimum waste
The disadvantages are:
• Fair surface quality
• Not suitable for mass production
• Fair material properties

2

The common AM technologies are Stereo Lithography (SLA), Laser Powder Forming
(LPF), Selective Laser Sintering (SLS), Fused Deposition Modeling (FDM), and Three
Dimensional Printing (3DP). The mentioned technologies have some advantages and
disadvantages, making them suitable for different applications. In general, the ability of
manufacturing highly complex and occasionally functional parts using AM has made them
relatively preferable over conventional manufacturing methods.
1.2 Additive Manufacturing and Construction Industry
Additive manufacturing is the only manufacturing approach that has enabled the
construction industry with automation. Contour Crafting is the first practical technology that
utilizes AM principles to build structures layer by layer. CC was invented at the University of
Southern California by Professor Behrokh Khoshnevis [1]. It solves the problems of the
conventional construction approaches such as being time consuming, having low labor efficiency,
and high costs.
Contour Crafting can be used in the fabrication of complete structures. CC utilizes computer
control to create complex forms and uses troweling to create smooth surfaces. Figure 1.2 shows
the CC process. Contour Crafting uses two trowels, one on the side and one on the top, creating
extremely smooth surfaces.
3

Figure 1.2. The Contour Crafting Process

1.3 Statement of the Problem
The properties of structures fabricated by CC are the defining factors in evaluating the CC
process. The bonding between the layers, the widths to height ratio of the layer, and the geometry
of the structures play important roles in the overall properties of the structures built by Contour
Crafting. The strength of the structures fabricated by Contour Crafting should be comparable with
the structures made by conventional construction approaches. The fabricated buildings should be
capable of bearing both vertical and lateral loads. Not all aspects of properties of structures made
by CC are known to date, as analyzing the structures under proper loading and boundary
conditions has not yet been done in a systematic way. More importantly, the surface sections
between the layers are shown to be the weakest section of the fabricated samples by CC.
Accordingly, the challenge of this study is to investigate the structures under given loads as well
as the bonding strength of the CC built structures using Finite Element Analysis (FEA), based on
the following given information:
a) The characteristics of construction material
4

b) Design specifications of the structure
c) Vertical loading and wind loading conditions
Moreover, the comparison of the behavior of the layered structures under different loading
conditions with structures fabricated by conventional approaches has not been done before. Such
a study answers questions such as how much better or worse the layered structures behave under
a specific loading condition in comparison with conventional structures. Therefore, the problem
can be restated as follows:
• How do CC structures behave under vertical and in-plane loads, and what happens to the
bonding between layers when walls a re
subjected to in-plane loads?
• Given the ability of CC in producing curved and corrugated structures, how much added
strength can be achieved because of this feature?

1.4 Research Purpose
Following are the objectives of this research:
1. The proposed research utilizes available tools and techniques for evaluating the strength of
structures fabricated by the Contour Crafting (CC) technology which fabricates structures
layer by layer through extrusion of cementitious materials. The first research objective is to
develop a methodology to investigate the strength of structures fabricated by CC under both
vertical loads (caused by the weight of a structure itself) and the in-plane loads, such as
wind.
2. The second objective is to develop a methodology to predict the bonding strengths between
layers for various approaches that may be used to enhance the inter-layer bonding strength.
5

3. The geometry of conventional walls has always been limited to simple rectilinear shapes
often without complexity and curves. Even this type of structure could not be constructed
precisely according to engineering specification because construction is done manually by
labor. Contour Crafting enables rapid fabrication of structures with complex geometries;
curved and corrugated for example. Implementing such geometries can allow the structures
to carry more loads while having smaller profiles compared to the conventional structures.
This can potentially change the way structures are designed and fabricated. Hence, the third
objective is to utilize the developed methodologies to analyze and compare the conventional
structures with layered structures with geometrical complexity under the same loading
conditions. The research purpose here is determination of the extent to which layered
structures can potentially improve the overall properties of buildings as compared with
conventional structures.
1.5 Research Contribution
Based on the review of the available literature, this work appears to be an extensive study of
the bonding strength of the layers of concrete both numerically and experimentally. This work
offers a methodology for investigating the bonding strength as well as the overall structural
strength of cementitious structures fabricated by Contour Crafting. The experimental part of this
thesis concerns the bonding strength of concrete layers for additive manufacturing. Studying the
effect of proposed reinforcement methods for enhancing bonding strength between fresh concrete
layers (such as adding rods) is also new. The experiments revealed that layers would be
delaminated under applied load. Delamination could be prevented by incorporating reinforcing
elements leading to enhancing the bonding strength of concrete layers.
Moreover, this work has utilized experimental and simulation results in studying the
influence of innovative geometrical shapes, such as curvature, hollow, and corrugated structures
that are only feasible using Contour Crafting, in increasing overall structural behavior. The
6

simulation studies have proven the significant impacts of both incorporating curvature, and
making the structures hollow on the flexural strength of walls. It was demonstrated that if the
thickness of curved walls is reduced to one half of equivalent size of rectilinear walls, the curved
wall still has higher strength.
1.6 Organization of the Dissertation
Chapter 1. This chapter includes an introduction to the problem of bonding strength of the
layers of cementitious material fabricated by CC.
Chapter 2. This chapter reviews all the fundamentals of concrete, bonding strength of
concrete, and previous studies focused on the bonding strength of concrete. The critiques of the
past approaches and the research gap are covered in this chapter.
Chapter 3. This chapter presents the methodology utilized for carrying out research in
studying the structural behavior of structures fabricated by Contour Crafting. The methodology is
categorized into three stages. In Stage One, the preliminary simulation model for studying the
strength of layered structures is developed. In Stage Two, experimental approaches were
conducted to tune the simulation model to the real world samples. In Stage Three, the results of
Stage Two were utilized in the simulation model.
Chapter 4. The analysis of the results of all three stages of the study are presented in this
chapter. For Stage One, the preliminary results were gathered and studied. In the experimental
Stage Two, statistical approaches were used to quantify the bonding strength and select the best
performer of the experiment to be used in Stage Three. For Stage Three, statistical approaches of
the design of the experiment were utilized to analyze the strengths of the structures fabricated by
CC.
Chapter 5. This chapter discusses the accomplished results and the area of contribution as
well as the suggestions for the future research direction.
7

2 Chapter Two: Background of Study
2.1 Introduction to Contour Crafting
The fabrication of structures using automation has been very slow. This is mainly due the
reason that the current fabrication technologies are not suitable for building structures, the ratio of
production quantity/type is small and there are always material and managerial issues. The
conventional fabrication approaches are facing problems such as low labor efficiency, high rate of
accident, low quality and lack of control [2].
CC can be used for both fabrications of a single house or mass production of a colony of
houses. It can easily eliminate the complications and limitations of the conventional fabrication
methods and open doors to new features, shapes and structure that conventional methods are
unable of doing. CC consists of a gantry system equipped by the extrusion system. Not only the
construction of the building can be done automatically, but also most of the subtasks such as
reinforcement, plumping, tiling, electrical wiring and painting [1] as shown in Figure 4.1.

Figure 2.1 Construction of Building Using Contour Crafting together with automated tiling
and plumbing
Contour Crafting has caused a paradigm shift by bringing automation in the construction
fields. Hence it aims to reduce and save labor, energy and material by taking advantage of
automation. Since the construction of the buildings becomes automated, less labor are involved in
the process which significantly reduces injuries to labors. Automation will significantly reduce
the lead time in the construction which in turn reduces the energy consumption. Moreover, the
8

tool paths and construction approaches are optimized to construct the buildings in the shortest
possible time. Finally, the saving of the material is achieved by taking advantages of computer
simulations before the process begins so the amount of material that is needed is clearly
estimated. Plus, the automated delivery of material to the CC machine eliminates the formwork
and wastes within the process.
This research aims to investigate the strength structures fabricated by CC. A more complete
description of Contour can found in the following references number [3], [4], [5] and [6].
2.2 Concrete
This section provides a brief overview of previous research related to the strength of fresh
and hardened concrete.
2.2.1 Introduction to Concrete
Concrete is one of the most common materials in construction which has been existed for
many centuries. In fact, concrete is a composite material consisting of cement, aggregate and
water. A variety of additives are also added to concrete to achieve varying desired characteristics.
Portland cement, patented in 1824, is the main ingredient of concrete. Cement plays the role
of binder in concrete mix that holds all the aggregates together. Cement contains calcium oxides,
silicon and aluminum. When cement comes in contact with water, the hydration process takes
place in which the cement converts to a binder that can hold everything in the mix together.
Hydration is in fact a chemical reaction that is initiated with water and generates heat in the
process. The chemical reactions are shown below:
Standard notation: Ca
3
SiO
5
+ H
2
O → (CaO) (SiO
2
) (H
2
O) + Ca (OH)
2

Cement chemist notation: C
3
S + H → C-S-H + CH

9

The water to cement ratio has a direct influence on the final properties of concrete. Lower
water to cement ratios makes the final concrete much stronger while higher ratios makes it
weaker. Fine and coarse aggregates are the other ingredients of concrete. Aggregates constitute
the bulk of concrete and are made from sand, crushed stones and recycled aggregates from
demolition of construction. Cement glues the aggregates together and contribute to the durability
of concrete. Finally, chemical admixtures are added to concrete to give it certain properties that
the target applications require. These admixtures are retarders, accelerators, plasticizers,
pigments, etc. These admixtures consists of less than %5 of weight of concrete and are added to
the concrete during mixing [7].
2.2.2 Fresh Concrete Properties
Fresh concrete is usually treated as yield stress fluid in which the fluids do not flow by
themselves and requires applying a minimum amount of shear stress. Equations 2.1 and 2.2
approximate the behavior of fresh concrete as a yield stress fluid.
γ˙ = 0 → τ < τ
0
Equation 2.1
γ˙ ≠ 0 → τ = τ
0
+ f (γ˙) Equation 2.2
τ
0
is the minimum yield stress needed to make the concrete flow . γ˙is the shear rate which
can be related to flow rate. The model in fact expresses the shear stress as a function of shear rate
[7]. On the other hand, other researchers treat concrete as Bingham fluid. Equitation 2.3 models
concrete as a Bingham fluid. Bingham fluids require a certain amount of shear stress to start to
flow as shown in Figure 2.2. It can be seen once the shear stress applied to the fluid reaches the
τ
0
, the fluid starts to move. Toothpaste and honey are common examples of Bingham fluids [8],
[9].
10

Figure 2.2. Bingham Model. τ
0
is the minimum shear stress needed to make the concrete flow

The rheological properties of cement paste can be measured using conventional tools such as
Couette Viscometer [10] or parallel plates [11] while heavy duty machines are developed to
measure the rheology of concrete in which large aggregates exist [12]. Time plays an important
role among the factors affecting the rheology of fresh concrete. In fact, the viscosity of concrete
reduces over time. This is mainly due to the thixotropic behavior of cementitious materials. When
concrete stops moving, internal structures are formed on top of each other which cause viscosity
increase. When concrete moves, during mixing for example, its viscosity decreases leading to the
drop of the minimum shear stress that is needed to make the flow happen [13], [14]. This change
of behavior should be reversible for a short period of time and is called thixotropic. All types of
concrete have this thixotropic behavior in which the time span differs. After the thixotropic time
span is passed, the hydration process begins which is not reversible anymore and the concrete
hardens permanently [15].
2.2.3 Stability of Concrete
Concrete is a multi-phase composite in which different materials are involved with
viscosities ranging from 1000 kg/m
3
for water to 3200 kg/m
3
for the cement paste. This difference
between the viscosities of ingredients combined with the effect of gravity results in some
11

phenomena such as bleeding and segregation that severely affect the homogeneity of the fresh
concrete.
Bleeding: The difference between the density of water and cement grains leads to
accumulation of water on the surface of the paste that is called bleeding. This phenomena in
which water displaces upward and come to the surface is described as by consolidation process
rather than individual cement grains settlement in a dilute system. Bleeding can be controlled and
slowed down by modifying the viscosity of the fluid that tends to travel upward. Hence, the
viscosity modifying agents are added to the mixture to slow down the bleeding effect. In fact,
these agents thicken the interstitial water [16], [17].
Segregation: In most times the aggregate’s density is higher than that of the suspending
fluid (cement paste and water). In this situation, the static sedimentation happens in which
aggregates tend to settle in a concrete mix. This phenomenon is called segregation. Segregation
can be called a multi-phase phenomenon where at least two phases are involved: the first one is
the cement paste and the second one is the =aggregates. It is shown that the thixotropic behavior
of the cement paste can help make the concrete mix more homogenous and reduce segregation
[18].
2.2.4 Hardened Properties of Concrete
The studies in hardened properties of concrete revolve around the compressive strength,
tensile strength, shrinkage, creep and crack resistance of concrete samples. The detail study of the
mentioned properties of different concretes is out of the scope of this research and they mostly
relate to casting and formwork applications. The effect of different parameters such as fresh
properties and hardened properties, aggregate type, reinforcements and chemical additives on the
hardened properties of concrete are discussed in this chapter.
12

2.2.4.1 Effect of Rheology
The correlation between the mix design and rheology of the fresh concrete on the
mechanical properties of hardened concrete has been investigated extensively. It has been shown
that there are some correlation between the rheology of fresh concrete and the final mechanical
properties of cured and hardened concrete [19]. In addition to mechanical properties, concrete
shows crack when shrinkage happens during hardening. Moreover, when the permeability of
concrete increased, the durability of concrete reduced. The rheology and water to concrete ratio
directly affect the permeability and crack resistance properties of hardened concrete [20], [21].
2.2.4.2 Effect of Aggregates Type
Two types of aggregates are used widely in preparing concrete mix: natural and recycled.
Each of these aggregates have their own benefits that are described in this section. Natural
aggregates such as limestone possesses a relatively good strength, they do not tend to have
reaction with silica and they decrease the drying shrinkage in concrete. Studies show that
compressive strength, flexural strength, module of elasticity as well as drying shrinkage of
concrete increase when the proportion of fine limestone goes up in the mix [22]. Gabbro, basalt,
quartzite, and sandstone are other aggregates that are used in addition to limestone as natural
aggregates. Gabbro has the highest compressive and flexural strength while sand stone has the
lowest value. Other researchers have shown that the compressive and flexural properties of basalt,
quartzite and limestone are relatively similar to each other [23].
Recycled Aggregates (RA) is another type of aggregates that are being studied widely by
researchers. Most researches aims to compare the properties of hardened concrete made with
natural against RA. It has been shown than concrete samples made with RA yield to a lighter
concrete than natural aggregates [24]. The recycled aggregates do not only come from crushed
demolition remnants, colored waste recycled glass are another source of aggregates that are used
by researchers as RA in concrete. Researchers utilized recycled glass sand (RGS) as the
13

aggregates in developing the concrete mix. It has been shown that the compressive and flexural
strength of concrete with RGS can be matched with concrete with other types of RA [24].
2.2.4.3 Effect of Reinforcement
Reinforcing bar (rebar) is commonly used to increase the tensile and flexural behavior of
concrete structures. Rebars that manufactured by hot rolling are commonly used in the concrete
and masonry structures to keep the concrete in the compressive state. In fact, rebars compensate
for the low tension properties of concrete. Rebars are shown to increase the tension strength of
concrete up to many times while having a limited effect on the compressive behavior of concrete.
The use of rebars is coupled with some challenges such as corrosion of the steel rebars and
adhesion of the rebars to the concrete which is a topic of interest among researchers.
Besides rebars, fiber reinforced concretes have received quite a lot of attention recently. The
main reason for adding fibers is to increase tensile behavior of concrete. Polypropylene, polyester
and glass are the major fibers that are being used in fiber reinforced concrete widely. In one
study, the mechanical properties such as compressive, tensile and flexural strength of self-
compacting concrete samples in which polypropylene was added was measured and compared to
unreinforced ones. While compressive strength of concrete was not changed, the tensile and
flexural strength were increased 14.4% and 10.7% respectively [25].
Recycled Aggregates (RA) is another type of aggregates that are being studied widely by
researchers. Most researches aims to compare the properties of hardened concrete made with
natural against RA. It has been shown than concrete samples made with RA yield to a lighter
concrete than natural aggregates [26]. The recycled aggregates do not only come from crushed
demolition remnants, colored waste recycled glass are another source of aggregates that are used
by researchers as RA in concrete. Researchers utilized recycled glass sand (RGS) as the
aggregates in developing the concrete mix. It has been shown that the compressive and flexural
strength of concrete with RGS can be matched with concrete with other types of RA [26].
14

In another study the effect of mixed reinforcement including hooked steel and a non-metallic
fiber on flexural was studied. The different mixture of steel-polypropylene, steel-polyester and
steel-glass are used as the reinforcement. In analyzing the flexural strength it has been shown that
the pre-peak and post-peak regions of the load deflection curved was boosted. It has also been
shown that micro-cracking has been slowed down due to the increase in energy absorption while
fibers are added [26].
2.2.4.4 Effect of Chemical Additives
Chemical additives mainly influence the fresh properties of concrete. The effect of the
additives on properties and durability of hardened concrete was investigated by some researchers.
Corrosion inhibitor and super-plasticizers are used by researchers to investigate the durability of
hardened concrete. It is shown that the final properties of concrete do not change significantly
when these admixtures are added to the concrete [27] , [28].
2.3 Bonding Strength of Concrete
The literature on the bonding strength of concrete which is relevant to Contour Crafting has
been investigated in two sections. The first one is the strength and bonding strength of masonry
structures and the second one is the bonding strength between old layers of existing concrete and
fresh concrete on top it.
2.3.1 Masonry Structures
This chapter reviews the literature on the structural behavior, failure modes and finite
element approaches in investigating masonry walls.
2.3.1.1 Introduction to Masonry Structures
Masonry walls are made by stacking units on top of each other that are connected to each
other by joins. Brick, block, and adobe irregular stones are all among the units in masonry
structure. The joints between the layers can be achieved in two ways. Adhesion using clay, chuck
15

and grout and mechanical interlock using groves and pultrusions [29]. Masonry structures in
general form are heterogeneous structures unless the units and joins are homogenously place on
top of each other. Figure 2.3 shows a typical irregular masonry structure together with the
representation of the regular masonry.

Figure 2.3. Masonry with irregular stones on the left and representation of the regular
masonry[29].
2.3.1.2 Structural Behavior of Masonry Structures
The structural behavior of masonry structures is similar to Contour Crafting structures. In
masonry, the units, such as bricks blocks, are stacked dry on top of each other while are
connected through cementitious material such as grout or mortar. In Contour Crafting wet
concrete extrudate plays the role of both units and joints, therefore, interlayer adhesion is
expected to be better than stacking dry blocks on wet mortar. There are two types of masonry
walls. The first type is only subjected to vertical loads such as gravity. The second type is
subjected to both vertical and in-plane loads such as wind. The latter wall is also called shear
wall. The structural behavior of buildings are influenced by the arrangement of shear walls and
strength of the floors. Unreinforced masonry structures are designed for gravity loads. Their
capacity for in-plane loads such as wind and earthquake are limited hence masonry structures
should be enhanced by partial or full reinforcement [30].
16

a) Failure of Unreinforced Masonry Walls
In general, the failure of shear walls, not specifically masonry structures, happens in two
modes: the first mode is flexure failure and the second one is tensile shear failure at the diagonal
direction. Figure 2.4 shows both failure modes of typical shear walls.

Figure 2.4. Failure of Shear Walls [30].
Unreinforced masonry shear walls show two major failure modes under in-plane loading.
These failure modes are mostly caused by their heterogeneity properties so they are slightly
different from the failure modes of typical shear walls shown in Figure 4.5 [31]. These failure
modes are:
1. Shear sliding: This type of failure occurs when the shear stress on the layers exceeds the bonds
strength between the layers.
2. Diagonal failure: If the diagonal tensile stress on the wall exceeds the splitting tensile strength,
diagonal failure occurs which is common in all shear wall.

17

Figure 2.5 Common failure mode of unreinforced concrete are a) Sliding b) Diagonal
failure [31].
The shear capacity of masonry shear walls is a function of bond strength (τ
o
) and axial
compressive strength (σ
v
) of the mortar based on Coulomb criterion.
τ =τ
o
+µ σ
v
Equation 2.3
τ
o
= Bond strength between the mortar and unit
µ = Friction coefficient at the interface
b) Failure of Reinforced Masonry Walls
Unreinforced masonry walls are not suitable for highly seismic regions because their low
shear capacity. Hence, reinforced masonry (RM) is used to increase the shear capacity of the
walls. RMs are similar to reinforced concrete (RC). So the analysis and prediction of shear and
load capacity of RM follows the RC approaches. The resistance of RM shear walls to in-plane
loads is influenced by the wall geometry, axial stress level, masonry properties, and the
distribution pattern of the reinforcement [32].
Failure of the RM occurs due to several modes: tensile cracking and yielding of vertical bars
in zones with high flexural, tensile shear cracking near the wall center, propagation of shear
cracks to form rupture planes, or yielding of steel and crushing of masonry in compression at the
18

loaded corners. The amount and spacing of the reinforcement, aspect ratio of the wall and the
compressive strength of the masonry are also among the factors affecting the overall capacity of
the RM shear walls [33].
c) Factors Influencing the Masonry Walls Strength
Percentage and dispersion of the reinforcement are important factors in the overall strength
of shear RM shear walls. The initial stiffness of the walls is not affected by the amount of
horizontal reinforcement bars [34]. Other researchers found out that the load capacity of the RM
is not affected by the quantity of the reinforcements but the ductility is influenced. In addition, the
shear capacity of walls goes up by the horizontal reinforcement as well as ultimate capacity [34],
[35].
Aspect ratio of the masonry walls (i.e., the ratio of height to length) is also crucial in
estimating the shear capacity of the walls. Some researchers investigated the effect of the aspect
ratio and figured out that any aspect ratio in the range of .75-2.5 influences the capacity
significantly while not improving the shear deformation [33]. Figure 2.6 shows the effect of
aspect ratio on failure mode. In fact, squat walls having the aspect ratio (H/L) of 0.6 have the best
flexural behavior. For tall walls a 45° crack occurs in the lower parts of wall leading to flexural
failure. In the aspect ratios close to 1, a 45° cracks meets the base of the wall converting part of
the stress to the compression zone.

19

Figure 2.6. Diagonal crack in masonry walls with different aspect ratio [33].

The material properties of the mortar and grout have been the topic of research among many
researchers. It has been found that the material properties of the grout and masonry has a slight
influence on the ultimate capacity of the masonry walls while better tensile strength of the
masonry units, such as bricks, increases the ultimate compressive capacity of the walls [36].
2.3.1.3 The Interface between Masonry Units
The properties of interface in two types of masonry is discussed in this section. Mortar is the
first type of interface used to hold the units together. The second type of interface is ungrouted
masonry units that are stacked on top of each other and mechanically interlocked using groves
and pultrusions.
a) Masonry Structure with Mortar
There are three important factors affecting the bond strength between the units and grout.
The first one is the properties of masonry units such as material, strength, perforation and size.
The second one is the properties of mortar including the mix ratio and water content. And the last
one is how well the workmanship has been conducted [29]. The weakest link in masonry
structures is the bond between units and joints. Grout as a joint shows a nonlinear behavior. When
20

grout comes in contact with units of masonry it enables the assemblage with two types of failure:
tensile and shear. So the interface of masonry unit and grout should be tested in both tension and
shear force. Figure 2.7 shows the schematic of two types of test that are commonly used to
investigate the shear strength and tensile capacity of the interface.

Figure 2.7. Typical testing of the interface. a) Tension b) Shear
Many experimental procedures have been developed by other researchers to study the brick
and grout interaction behavior. Figure 2.8.a shows one of the developed approaches to investigate
the shear behavior of an assemblage of 210 bricks and grout. Here a monotonous growing load
was applied to the assemblage until failure occurred. Figure 2.8.b shows an actual shear test
mechanism developed to test the interaction between two bricks having grout in the middle.
21

Figure 2.8. Two of the developed experimental approaches to test the behavior of brick to
brick interaction [29].
b) Masonry Structure without Mortar/Grout
Mortarless masonry structures in which mechanical interlock has been used to hold the units
together have been investigated by researchers. Dry friction based contact has been the basis of
the investigation in which it has been shown that behavior of the interlocking systems are
influenced significantly by the behavior of the dry mortarless units in elastic and inelastic zones
[39]. In another research an interlocking type of block called “Putra block” were used without any
grout. It includes three sections: stretcher block, corner and half block as shown in Figure 2.9.

Figure 2.9. Putra interlocking blocks [37].
22

The stretcher block is the main unit of masonry, the cornet blocks are the interlocking parts
and the half block unit completes the courses of the wall. Blocks interlock to the next course
through pultrusions. Bonding and slipping mechanism for the interface has been considered in
their study. A finite element model has been developed that was claimed to have agreement with
experimental results. No comparison of shear capacity with masonry structures with mortar has
been done in their study.
2.3.1.4 Finite Element Analysis of Masonry Structures
Investigating masonry structures using finite element approaches has been a topic of interest
among researchers. There are two types of methods in using finite element in studying the
structures: macro and micro modeling.
a) Macro Modeling:
In analyzing the masonry using macro modeling, the assemblage of the units and joints is
considered as an ideal homogenous body and the structure is studied as a one-homogeneous
material model. This type of simulation neglects the influence of mortar joins and is used in low
stress loading condition in which the overall behavior of the masonry (deformation of the
assemblage) is required. Material properties are selected based on the average masonry
properties. In other studies, the effect of non-linearity and progressive failure was implemented in
the finite element simulation but still the structure is treated as one-phase material [38]. To
incorporate the non-linearity of the masonry structure materials in the finite element analysis, a
simulation model should be defined merely for the material. There are two types of material
modeling methods that are commonly used by researchers: damaged plasticity and smeared crack
concrete method.
The smeared crack concrete method explains that the cracks occur when the stress in
concrete reaches the failure surface in the tension region or in a combined tension-compression
23

region [39]. This model is suitable for the conditions in which the concrete is under monotonic
straining while the material shows either tensile cracking or compressive crushing. In fact, the
focus of method is on the growth of the micro cracks in small regions [40]. Concrete under
compression shows an elastic behavior in which the strain is reversible. Upon applying more
stress, the strains become non-reversible and concrete softens. However, in tension, the strength
is only 7%-10% of the compression so the strain occurs rather quickly which makes the
observation relatively difficult.
The concrete damage plasticity method focuses on low confining pressures in which
concrete behaves in a brittle fashion. In this way, concrete cracks in tension and crushes in
compression. The damage which is due to micro cracks in brittle materials is defined by
dissipated fracture energy [41].
b) Micro Modeling:
In this modeling approach the masonry is treated as a two-phase material in which the
elements of the interface are taken into account. Micro modeling provides a more reliable study
of masonry structure and failure mechanism. Researchers have incorporated the non-linearity of
concrete using multi-surface plasticity models described in the last section 2.3.1.4.a to predict the
behavior of the masonry shear walls under in-plane loading condition. Researchers have
developed a micro model to predict the behavior of the masonry shear walls where they
succeeded to predict the peak load. However, they failed to predict the post peak behavior
including the prediction of the brittle failure seen in the diagonally loaded unreinforced
masonry panels [42], [43].
2.3.2 Bonding Strength between Fresh Concrete and Old Existing Concrete Surface
There are two major applications in which an overlay of concrete is added to an existing
surface: repairing damaged concrete and increasing the capacity of existing structures. This may
24

cause two major concerns with respect to the properties of the final structure. The first concern is
the mismatch in the properties of the two layers of concrete and the second one is the bonding
strength between the layers. To better investigate the bonding strength for such applications, the
mechanical and physical behavior of the interface is studied [44]. The interfacial zone (IZ) has
different characteristics compared to other parts of the mixture. IZ is the main cause of weak
bonding due to “wall effect”. When the mixture of the fresh concrete comes in contact with the
hardened layer of old concrete the aggregates cannot penetrate inside the old concrete to form a
strong bond, hence there will be gaps between the new fresh concrete and old existing concrete.
Figure 2.10 shows this gap in green which is caused due to the wall effect in the IZ.

Figure 2.10. Interfacial Zone
Besides the interface of the fresh layer and old concrete layers, IZ is also seen in the
interface of the large aggregates and cement paste. Although cement paste is very fine, it is a
particulate material. When it comes in contact with large aggregates of the similar gap is created
that decrease the strength of concrete. This is the reason that concrete breaks from the interface of
the aggregates and cement paste. Figure 2.11shows the wall effect between large aggregate sand
cement past.
25

Figure 2.11. The wall effect is highlighted in the interface of large aggregates and cement
particle.
2.4 Critique of Past Approaches
The previous CC research projects have mainly focused on the study of feasibility of the CC
process, planning the operation of the CC process, and developing different components of the
CC system such as flow control mechanism or the robotic systems. The few studies related to
structural properties of the artifacts produced by CC had the following shortcomings:
• The work only simulated the CC extrudates right after printing while the concrete is still
fresh. The structural properties of the hardened structures being subjected to different loading
conditions were not studied.
• The structures were modeled as unibody hence the effect of the interface between layers was
completely ignored, i.e., the bonding strengths of the layers was not studied.
• The studies merely simulated CC fabricated structures under vertical loads such as fabrication
loads. No study of the in-plane loads such as wind and earthquake has been conducted.
• Comparison of performance of CC built structures with the structures fabricated by
conventional methods such as casting has not been performed.
26

• No study of the gained strength resulting from curved features, which are easily
accomplished by CC but are expensive and difficult in conventional methods, has been made.
2.5 Summary
In this section, different studies regarding concrete, the fresh properties and hardened
properties of concrete have been reviewed. Separate studies regarding masonry structures, their
properties, strength and their bonding strength were also surveyed. Finally, the literature on the
bonding strength of concrete to concrete in repairing structure was reviewed.

27

3 Chapter Three: Research Methodology
3.1 Introduction
Significant efforts have been put into finding the appropriate model, running the right type
of simulation, calibrating the simulation with experimental work, simulating the behavior to study
the effect of layer by layer fabrication, and, finally, studying how free form fabrication of the
structures can influence the behavior of the walls fabricated by Contour Crafting. The key
activities fall into three stages that are show in Figure 3.1.

Figure 3.1. Research Methodology
The above categories are all related to one another. In each step of the research, the
concurrent theoretical studies, experimental, and analytical work was conducted. The outcome
was analyzed and a new set of experiments were designed.
3.2 Stage One: Analytical study
This section discusses the preliminary work done on numerical analysis and engineering of
Contour structures to compare the strengths of an unreinforced layered shear wall made by CC
with an unreinforced poured shear wall. Also reported is the analysis carried out on the layered
structure. Finite element analysis using ABAQUS has been utilized in this study
28

3.2.1 Geometry of the Model
Based on the US census 2010, the average square meter of feet floor area in the United
States is 200 square meters that includes all type of houses such as concrete and wood. The
longest wall of a single family house is typically those of the living rooms which are 426.72 cm
on average. The height of the typical single family houses in the United States are 304.8 cm per
floor and thickness is 15.26 cm [47]. Hence, the shear wall that was studied in this work has the
dimension of 426.72 cm (lengths) x 304.8 cm (height) x 15.24cm (width).
Contour Crafting is capable of fabricating different layer heights. The analysis in this work
will be studied based on 2.54cm layers on top of each other until the full scale structure is built.
However, the model can be generalized for any layer size. Therefore, the model for the layered
simulation consists of walls made of 120 extruded layers.
3.2.2 Boundary and Loading Condition
Shear walls under the load of wind was simulated in the preliminary modeling. In the typical
residential-scale wind speed, 64 Kilometer per hour is reported to be the fastest that blows in the
United States [45] which approximately around 2000 Pascals according to conversion tables.
Considering a typical factor of safety of 5 for concrete structure, the wind pressure on the wall at
this wind speed would be 10000 Pa.
The bottom of the wall is fixed in the ground where all the degrees of freedom are restricted.
Likewise, the sides of the wall are also fixed as they are connected to other walls where the
degrees of freedom are restricted. However, the top part of the wall is left unrestricted to better
study and compare the overall deflection of the wall. Figure 3.2 shows the wall models developed
in this study. The weight of the roof is not considered in this study to focus on the wall in
preliminary simulations.
29

Figure 3.2. The 3D model of the shear wall with the boundary and loading condition
3.2.3 Material
The typical characteristics of concrete have been selected according to the M30 grade
concrete specifications [46]. Table 6.1 summarizes the typical properties of concrete utilized for
the simulation.

Table 3.1. Typical Characteristics for M30 grade concrete
Parameter Value
Young's Modulus 27 Giga Pascals
Poisson Ratio 0.18
Density 2400 kg/m
3

Compressive Strengths 25 Mega Pascals
Tensile Strengths 5 Mega Pascals

3.2.4 Elasto-Plasticity
Finite element analysis can be used to analyze and compare the structures, study interaction
of the layers, and simulate the different steps of the Contour Crafting process. To incorporate the
elasto-plasticity of concrete, damage plasticity approach that has been explained in previous
30

chapters was utilized. The parameters of damage plasticity approach that have been selected
based on the previous works [46].
3.2.5 Simulation of Interaction
The realistic analysis and comparison of the CC structures with the conventionally fabricated
structures should be carried out through micro analysis of the structures where the interaction
between the layers are taken into account. During the Contour Crafting process, the adhesion
occurs between the layers when the fresh layers of concrete are placed on top of the previous
ones. The behavior of the bonding of the layers in CC can be well described on the basis of
cohesive zone modeling (CZM) and fracture mechanics (FM) principles.
The Cohesive Zone Modeling (CZM) approach has been utilized in practice to simulate the
adhesion between the layers. CZM is a common tool for the simulation of delamination and
deboning of composites and adhesive joints. CZM uses the principle of fracture mechanics (FM)
to evaluate the integrity of the bond in joints and layers of composites. In FM two parameters are
compared to each other. One parameter is the function of the load and the body of the crack (the
strain release energy “G
c
”) and the other one is the resistance to crack (G) [47]. The condition
where G
c
= G starts the propagation of the crack. CZM simulates the physical crack zone where
traction and separation behavior tears and delaminates the layers. In order to better explain the
theory, cohesive law which is used in FM to correlates the interfacial strength (σ) with the
displacement (δ) is shown in Figure 6.2. Where the load reaches the critical value T = σ
crititcal
,
δ
critical
is the critical separation/displacement and the crack start to grow till δ
failure
where failure
takes place. G
crtitical
is the critical strain release energy and K
i
is the initial stiffness penalty [48].
31

Figure 3.3. The cohesive law
The fracture at the bond zone is described in Figure 6.3 that shows the different stages of
crack formation based on the cohesive zone modeling. Stage One is the linear elastic stage where
the interface resists to the load. In Stage Two, the crack starts to stretch where T = σ
crititcal
. At
Stage Three cohesive law becomes non-linear and the crack continues to grow until the failure
occurs at stage 4 [49].

Figure 3.4. Fracture at the bond zone [49].

32

The fracture stiffness (K) plays an important role in predicting the behavior of the interface
of the layers in the contour crafting structures. The higher the value of the stiffness, the higher the
resistance of the layers to delamination.
The stiffness of the initial response of the cohesive element can be defined as:
Equation 3.1
Where Ti and δ
critical
are the critical separation and displacement, respectively. Two
stiffnesses for the shear mode (K
ss
and K
tt
) and one stiffness for the normal mode (K
nn
) are
required to properly evaluate the bonding of the layers with cohesive zone modeling. The values
of K
ss
, K
tt,
and K
nn
are assumed to be to same according to the previous studies [49]. The fracture
stiffness has been formulated according to the material properties and geometry of the samples
[53]:
K ≥ 50. E/t Equation 3.2
3.3 Stage Two: Experimental Study
This stage discusses the experimental approaches to calibrate the simulation model and
extract the cohesion parameters that are used in the simulation model. CZM uses the cohesive
strength of layers in tension as well the as the strain energy release energy of the layers (G
n
). The
critical value of the energy, fracture toughness, is needed for the FEM models (cohesion
parameter). Double Cantilever Beam (DCB) is the experimental approached that is commonly
utilized to derive the cohesion parameters. [50]. The method utilized to derive the cohesion
parameter is based on the work of R. Campilho et al. [51], [52].
3.3.1 Double-Cantilever Beam (DCB) Test
The DCB test is a standard procedure to determine the fracture toughness of inter-laminar
interface [53]. DCB test is developed to calculate the fracture stiffness that can be used in the
finite element model while other available tastings such as slant shear test, short beam shear test
33

or pull off test do not provide outputs that can be incorporated in the simulations. ASTM
international 2001b [52] provides recommendations for the procedure of this test utilized in this
study. In the DCB test, there is an initial crack length that is attempted to propagate by applying
loads to two laminates. Figure 3.5 shows the DCB test.

Figure 3.5. DCB Test [50]

The Compliance Calibration Method (CCM) was utilized in this work to derive the critical
strain release energy (G
C
)
,
the results of which were compared to each other. Equation 3.3 shows
the modified Compliance Calibration Method (CCM).

Equation 3.3
where G
c
is the critical energy release rate, P is the applied force, b is the specimen width,
and, finally, C = δ/P is the compliance (δ is the specimen displacement). A cubic polynomial
curve was used to represent compliance (C) as a function of crack length (a).
34

3.3.2 Specimen Preparation
The dimension of the samples was selected in accordance to the ASTM D5528-01 2001. The
limitation of the available tension/compression machine was also a factor in selecting the right
dimensions. Figure 3.6 shows the dimensions of the sample.
The samples were fabricated using a one-piece wooden mold. The surface of the wood was
laminated to avoid absorption of the water from the concrete mix. Figure 3.7 shows the wooden
mold that was used to fabricate the samples.

Figure 3.6 Dimension of the samples

Figure 3.7. Mold used to fabricate samples
35

An Instrum 210 series tensile machine was utilized to run the DCB tests. The load cell was
100kgf. The strain rate was selected at 0.5 mm/min in accordance to ASTM standards. Figure 3.8
illustrates the samples during the DCB test on the tensile machine.

Figure 3.8. The DCB sample loaded in the Instrum 100kgf tensile machine
3.3.3 Material
The concrete used in the experiments developed was engineered for extrusion in the Contour
Crafting technology. The concrete showed minimal sign of segregation and bleeding. The
concrete utilized in the experiments had a moderate viscosity. The material contained rapid
setting ingredients that are essential for stacking layers on top of each other. The concrete mix
was sieved on the filter size of 4.75 millimeters to ensure the maximum size of the aggregates are
less than 4.75 mm. The work ability of the used concrete was also tested. The flow diameter was
230 mm after 25 revolutions. The Yong’s modulus, compressive strength, and tensile strength of
the material were measured at 27GPa, 25 MPa, and 5 MPa respectively. The characteristics of the
concrete used in the experiments are listed in Table 3.2

36

Table 3.2. Characteristics of concrete used in the experiments
Parameter Value
Young's Modulus 27 GPa
Density 2400 kg/m
3

Compressive Strength 25 MPa
Tensile Strength 5 MPa

3.3.4 Experimental Design
The Stage Two experiments were designed and analyzed in three steps. Step One included
some preliminary experiments to study the feasibility of DCB with concrete layers. The aim of
this step was to figure out the best experimental setup, tools, and measurement techniques that
can get the most out of each DCB sample.
Step Two of the experiments were called primary experiments. It included design of a series
of experiments to study the bonding strength of concrete to concrete. The experiments also
studied the influence of adding reinforcing rods between concrete layers on the bonding strength
of the layers.
After analyzing the results of step two, further experiments were designed for step 3 of the
experiments, targeted to further investigate and quantify the best performer of Step Two
experiments. After quantifying the influence of reinforcing, the cohesive parameters of the plain
interface samples and reinforced interface samples were computed in order to be used in the next
stage of the study. Figure 3.9 portrays the flowchart of the experiments.
37

Figure 3.9. The flowchart of the DCB experiments
3.4 Stage Three: Primary Analytical Analyses
This stage is based on the work of Stage One and Stage Two. The results of the experimental
work in Stage Two were plugged into the analytical analysis of Stage One. The methodology was
based on the methodology presented in Stage One. Stage Three begins with re-simulating the wall
section that was studied in Stage One using the experimental data that was gathered from Stage
Two. The simulations were then compared with Stage One. The next step of the study was to
determine the minimum bonding strength that was required to make the layered structures
comparable with none layered structures. This happened using a sequence of simulations that
were aimed to minimize the delamination between concrete layers.
The final step of Stage Three included the investigating the effect of geometry complexity
on the flexural strength of the walls. Three types of geometries were the objective of the study. It
started with simulating curved walls and comparing them with the rectilinear structures of the
38

same dimensions. Then, a sensitivity analysis was conducted to find out the significance of wall
thickness and curvature radius on the strength of the wall. After that, a thorough study of
curvature radius and comparison of them with rectilinear structures including of 30 simulations
were conducted to find the effective regions where curvature influences the strength of the walls.
The final step of Stage Three included the finite element simulations of corrugated hollow walls
and curved corrugated hollow walls. Figure 3.10 shows the summary of the results and analyses
of Stage Three of this study.

Figure 3.10 Roadmap of Stage Three of the study

39

3.4.1 Finite Element Analysis
In FEM of each step of the simulations, a wall section is simulated as a computer model. The
model includes small elements that represent the shape of the wall section. The equations are then
applied to each of the elements. The model can analyze the deformation of the wall as well as the
delamination of the layers on top of each other. After the performance of the different wall
sections were analyzed, the necessary modifications could be made to enhance the performance
of the wall sections. This wall section was modeled as a layered structure. Each layer was
designed separately in the FEA software and stacked on top of each other. Then the boundary and
loading conditions were applied.
a) Material Properties
Material properties such as Young’s modulus, Poisson's ratio, and density are required for
the FEM model. Such properties are selected based on typical M30 grade concrete properties. The
developed model is not dependent on any specific material. Typical material properties tastings
can be conducted for any material, and the results can be plugged into the model in the future.
Elasto-plasticity of concrete was incorporated into the model using the concrete damage plasticity
approach.
b) Boundary Conditions on the Wall
Similar to section 3.2.2, there were two types of load that were applied to the wall. The first
one was gravity and the second was wind. This is the result of a 128 kilometer/hour wind that is
reported in the literature. Similar to section 3.2.3, the bottom and the sides of the walls were
restricted so they had zero degree of freedom. The top portion of the wall was left unrestricted to
better study the deformation and delamination. Table 3.3 summarizes the boundary condition and
geometry information.

40

Table 3.3. Boundary and geometry information
Item Value
1 Dimension 426.72 cm (lengths) x 304.8 cm (height) x 15.24cm (width).
2 Number of Layers 120
3 Load magnitude 10000 Pascal (N/mm
2
)
4 Material E = 27GPa, Density = 2400 kg/m
3
, Compressive Strengths =
25Mpa, Tensile Strengths = 5 Mpa

3.5 Summary
In order to investigate the strength of structures fabricated by Contour Crafting, three stages
of events were designed and performed. The beginning of the analysis was Stage One. It covered
the proper simulation approach to study structures. CZM was selected as the best approach to
study the effect of load on the bonding strength of the layers. Then, a Finite Element Model for
studying 120 layers of concrete on top of each other, representing a section of a wall, was created.
The model used theoretical cohesive parameters to study and compare a layered wall with a none
layered wall of the same size.
Stage Two of the experiments covered experimental work to study and analyze the bonding
strength of concrete layers. As explained, cohesive zone modeling based on fracture mechanics
was the approach of studying the layer interaction. DCB test was picked as the best approach to
extract the cohesive parameters such as fracture toughness and energy release rate where the
layers start to delaminate.
Stage Three was the final step in this work. It utilized and combined the results of the Stages
One and Two to better study the effect of layering in construction. Moreover, the effect of
geometrical incorporating curvature that can potentially be achieved by automatic construction
approaches such as Contour Crafting was studied on the strength and layer delamination of the
41

wall. A similar Finite Element Model was developed to study the effect of curvature on
mentioned characteristics.

42

4 Chapter Four: Results
This section presents the results of each of the three stages of the research and discusses
them in detail.
4.1 Analysis of Stage One Results
4.1.1 Testing the Layer Adhesion Model
Two layers of concrete were modeled to evaluate the effectiveness of the bonding model
between two layers. Figure 4.1 shows the simulation of two layers of concrete on top of each
other.

Figure 4.1. Stress Simulation of two layers. a) unscaled b) scaled to show the deflection
The maximum stress of 0.1 MPa applies to the left and right sides of layer 2. Layer 1 is
under significantly lower stress because the bottom of layer 1 is fixed.
In order to prove that the model of bonding between two layers works, two more
experiments were carried out. In these experiments a relatively low and a relatively high value of
the fracture stiffness between the layers were selected. The results of the analyses were compared
to the original value of the stiffness to correspond to a very weak bond and a strong bond between
layers. It can be shown in Figure 4.2.a that a weak bond led to the separation of the layers. The
displacement of the second layer on the first layer is 6 mm and takes place approximately in the
middle of the layer and there is not a significant stress on the first layer which is fixed to the
43

bottom. However, in the case of the strong bonding strengths (Figure 4.2.b) the displacement of
layer 2 on top of layer 1 is 1.7 x 10
-5
millimeter that may be considered as being negligible.

Figure 4.2. Displacement distribution a) weak bonding B) strong bonding

4.1.2 Full Scale Model
The full scale shear wall is modeled in this section under two cases:
1. Case I: The model in this case assumed a wall which is constructed by conventional methods such
as casting. Therefore, the wall was a single rigid body without having any layers.
2. Case II: The model in this case assumed a wall which is built by Contour Crafting in 120 layers.
Hence, there are 119 inter-layer interactions.
The geometry and boundary and loading conditions applied to the wall was explained in
previous sections.
4.1.2.1 Case I:
The full-scale casted wall was simulated using Abaqus. Because of the large size of the wall,
a relatively rough mesh size (hexagonal 3D mesh with dimensions 5.08cm x 5.08cm x 5.08cm)
was used for the analysis. The model has 21,231 nodes and 15,555 elements. Table 4.1 shows the
wall deflection at different heights that was measured at 30 cm intervals. Figure 4.3 shows the
stress and strain distribution.
44

Table 4.1. Deflection of the wall in different heights
Wall Height (m) Deflection (mm)
0.30 0.0434
0.61 0.1392
0.91 0.2574
1.22 0.3788
1.52 0.4930
1.83 0.5957
2.13 0.6862
2.44 0.7670
2.74 0.8430
3.05 0.9353

(a)

(b)
Figure 4.3. Strain distribution (a), stress distribution (b)
45

The measurements showed that the maximum deflection took place at the top center of the
wall (red area in Figure 4.3.a) with the magnitude of 1.87 mm. The maximum stress of 2.7 MPa
happened at the top right and left of the wall. While the maximum stress that the wall receives is
lower than the critical value, the wall showed a relatively large deflection. In fact, not having
reinforcement caused the wall to be deflected so much. When in-plane loads such as wind are
applied to the wall, reinforcement should be used to control the deflection. The relatively low
flexural strengths of the concrete that was used in the model can also be the reason for the
occurred deflection. Therefore, improving the flexural properties can decrease the deflection.
4.1.2.2 Case II
A full-scale layered model has been developed consisting of 120 layers (with 119 inter-layer
joints). The dimension of each layer was 426.72 cm (lengths) x 15.24 cm (widths) x 2.54cm
(thickness), as mentioned previously. Node to surface type of interaction was selected for the
interaction between one layer and the bottom one. Therefore, the model searches for the
separation of the nodes from the surface. Figure 4.4 shows the top nodes of one layer. Again, a
relatively coarse mesh was utilized in the simulation to decrease the computation time. The size
of each element was 7.62cm x 9.652cm x 2.54cm. In total, there were 31,680 nodes and 10,320
elements in the model.

Figure 4.4. The selected top nodes of one layer
46

The stress and strain distributions of the layered model are shown in Figure 4.5. The
deflection of the wall was measured every 30cm. For comparison purposes, the results of both
deflections for Case I and Case II are shown in Table 3.1. Figure 4.6 also shows the plot of the
deflection data for both layered and casted walls.

(a) Stress curve

(b) Strain curve
Figure 4.5. Stress distribution (a), strain distribution (b)

47

Table 4.2. Deflection of the layered and rigid wall at different heights.
Wall Height(cm) Layered(mm) Poured(mm)
0 0 0
0.3048 1.4196 0.08
0.6096 1.4432 0.2784
0.9144 1.4432 0.5148
1.2192 1.4437 0.7576
1.524 1.4437 0.9860
1.8288 1.4437 1.1914
2.1336 1.4432 1.3724
2.4384 1.4497 1.5340
2.7432 1.4515 1.6860
3.048 1.4827 1.8706

Figure 4.6. The comparison of deflection in layered and none layered walls
The measurements show that the bottom 15 layers experience the highest amount of
delamination. Table 4.3 shows delamination magnitudes of the first 15 layers. The shear
stress on the first bottom layers was more than the shear capacity of the interface. This is
because the layers next to the anchored bottom are under more shear stress than top
48

layers. So the strain release energy on the interface passed the critical value. The
delamination of the layer after layer 15
th
became negligible (< 1 micron). The
delamination of the first 20 layers caused the major part of the wall deflection. Figure 4.6
shows the deflection of the wall at different heights. It can be seen that the maximum
deflection of the layered model is lower than the homogenous casted wall model. In fact,
in the casted model the deflection increases gradually as the height increases while most
of the deflection in the layered wall happened in the bottom layers where the
delamination occurred. Hence, it can be concluded that the layer structures tend to
increase the flexural strengths of the wall compared to a none layered (homogenous) wall
of the same size. Note that the maximum deflection of the layered wall was about %20
less than the none layered wall.
Table 4.3. Delamination magnitude of the first 15 layers.
Layer Number Delamination(mm)
2 0.434
3 0.308
4 0.215
5 0.149
6 0.103
7 0.071
8 0.049
9 0.034
10 0.023
11 0.016
12 0.011
13 0.007
14 0.003
15 0.001

49

4.1.3 Validation
The purpose of the following experiment was to verify and compare the pattern of the finite
element simulation model with physical experiments.
4.1.3.1 Experimental Procedure
44 layers of 6 mm x 6 mm x 60 mm basswood Figure 4.7 were stacked on top of each other
to create a 600 mm (length) x 400 mm (width) x 6 mm wall. Such dimension kept the aspect
ratio of the simulated concrete wall. Wooden spacers with the thickness of 1.5 mm were placed
between the layers to control ensure that the thickness of the adhesive were uniform for all the
layers.

Figure 4.7. The comparison of deflection in layered and none layered walls
Silicone sealing adhesive was chosen to add between the layers as its flexibility and strength
enable the delamination of the layers to be observed and compared with the simulation. Figure
4.8 shows the process of fabricating. The entire gap between the layers was not covered with glue
to prevent a strong bonding. Four locations between the layers were glued. The distribution of the
glue between the layers is shown in Figure 4.9. After the glue was added, the wall was taped on
50

the middle and two studs were added from side to press the layers to each other and ensure a
proper bonding.

Figure 4.8. The construction of the wall

Figure 4.9. The distribution of the glue
51

Figure 4.10. Final Wall
Loading fixtures that are used in testing structures in the School of Architecture was used to
test the wall. Figure 4.11 shows the wall that was placed on the loading fixture. To create a
distributed load, a relatively large plastic bag was filled with mortar. The size of the plastic bag
allowed moving the mortar to evenly distribute the load on the wall. One 20 Kg bag of mortar
was used and distributed on the wall to create a distributed load. It was extremely 20 kg more
weight was added later to the wall to create a visible delamination.

Figure 4.11. Loading fixture
52

4.1.3.2 Results
The deflection and delamination results from the finite element simulation is shown in
Figure 4.12. The simulation showed that bottom 15 layers went through a major delamination
while the top layers did not delaminate significantly. Figure 4.12 plots the delamination of the
interlayer shift of the first 5 layers.

Figure 4.12 The simulation of the layered concrete wall under a distributed load

Figure 4.13. Delamination for the first 15 layers of the simulated wall

The experiments showed that the second layer that was next to the anchored first layer
showed the highest amount of delamination that was 1.87mm. The delamination became so
difficult to measure after layer nine. Figure 4.14 plots the delamination of the first layers for the
physical wall. The pattern of the delamination decrease for the first layer of the wall matches the
pattern that the model had generated. In general, the bottom layers are more vulnerable to the
53

delamination according to both experiments and simulations. Figure 4.15 shows the deflection
and delamination behavior of the physical wall that were tested. The maximum deflection of the
physical wall was estimated to be around 6 mm.

Figure 4.14. Delamination of the first 15 layers

Figure 4.15. Delamination and deflection of the wall
54

Figure 4.16. Delamination and deflection of the wall
The physical wall that was tested under loading was simulated in the developed finite
element model in 44 layers. To make modeling more feasible and to avoid unnecessary
assemblies in the finite element environment, the adhesive was modeled as part of the layers
themselves as shown in Figure 4.17. The drawback of such approach was the fact that adhesives
faced each layer from one side instead of two sides that happens in reality.
55

Figure 4.17. The model of each layers

The material properties of a typical basswood are shown in Table 4.4. The shear and tensile
capacity of the silicone adhesive that was used in the experiment were derived from existing
database.
Table 4.4. Delamination magnitude of the first 15 layers.
Property Value
Young's Modulus 10.7 GPa
Tensile Strength 59 MPa
Poisson's Ratio 0.12
Density 0.6 Kg/m3

The pattern of the deflection complied with the pattern of the conducted physical
experiment as shown in the Figure 4.18. Simulations shows that the maximum deflection of the
wall was 5.16 mm that was about 2-3mm lower than what was recorded from the experiments.
Figure 4.19 shows the delamination magnitude of the the first 13 layers that the simulations
recorded. While the overall pattern matches the experimental data, the value of delamination
that the simulation reported for the first 3-5 layers was lower than the values that were reported
for the experiments. The method that the adhesive was modeled in the simulation increased the
capacity of the adhesive so it was expected to get lower delamination magnitude for the first
56

few layers in the simulations. Also, the shear and tensile capacities of the adhesive that were
used in the simulations were derived from typical libraries and does not correspond to the
adhesive that were used in the experiments. However, the overall pattern of the simulation
model matches the experimental data.

Figure 4.18. Simulation results of the wall. The deformation has been scaled up 14 times for
a better observation of the deformation

Figure 4.19. Deflection comparison for the first 13 layers
57

4.1.4 Summary
In summary, an initial finite element simulation model was developed to investigate the
structural properties of structures fabricated by Contour Crafting. The simulation model takes the
effect of the layers into consideration. Cohesive Zone Modeling (CZM) was utilized to
investigate the delamination of the layers. After successfully modeling two layers, a section of
wall consisting of 120 layers were modeled. A second model of a same dimension wall but
layered was developed to be compared with the layered structure. The model showed that the
layered wall faced a highest amount of delamination in the first 15 layers. After the first 15
layers, the delamination rate dropped significantly. It was also shown that the deflection of the
layered wall was lower than the deflection of the layered wall.
4.2 Analyses of Stage Two Results
In this section, the results of the experimental work to investigate and quantify the bonding
strength of the layers of concrete are presented and discussed. The experiments of this section
were conducted in 3 steps. At first, a series of initial experiments was conducted to study the
feasibility of the DCB test with concrete layers. Then, the primary experiments were designed
based on the results of the initial experiments, and, finally, experiments were conducted to
finalize and conclude the results of Stage Two.
In order to categorize the experiment and samples of each section and refer to them back and
forth, a coding system was created, and each sample of the experiment was given a unique code.
This coding system is used throughout this study. The code of the samples belonging to the initial
experiments starts with letter “I”, the samples of the primary experiments start with letter “P”,
and the code of the samples of the final experiments start with letter “F”. Other letters and
58

numbers are then added to the samples based on the type and stage of the experiments, which are
explained in future chapters.
4.2.1 Initial Experiments
The purpose of the initial experiments was to study the feasibility of doing DCB on concrete
samples. Three factors needed to be identified for the test. The first factor was the type of bracket
to be glued to the concrete layers to pull the layers away from each other. The second factor was
the type of glue to bond the brackets to the concrete layers, and the final factor was the behavior
of layers of concrete in the DCB test.
4.2.1.1 Experimental Procedure
Two types of brackets were used in the experiment. The first one was an aluminum 90º
which was used to connect aluminum extrusions to each other. The dimensions of each side of
such brackets were 25.4 mm long and 25.4 wide. The second type was the steel butterfly hinge.
The butterfly hinge came in a length of 30.34 mm (1 foot) so they were cut into pieces of 25.4
mm which is the width of the concrete layers. The length of the hinges was 38 mm. A hole was
drilled on one surface of the hinge for placing the pins of the load cell. Two types of glue were
used to bond the hinges to the concrete samples. The first type was Loctite concrete to metal
epoxy. The second type was Crystalbond glass glue. Crystalbond is typical mounting glue that
should be molten before applying and cures very fast.
4.2.1.2 Samples
Two samples were made using the mold described in section 3.3.2 with steel hinges, one
glued with epoxy and the other with Crystalbond, and two samples were made using the
aluminum brackets, one glued with epoxy and the other with Crystalbond. So, in total, four
experiments were designed for this step to test the feasibility (Figure 4.20). Table 4.5
summarizes the fabricated samples together with their codes. It should be noted that the second
letter of each code is descriptive of the type of experiment. For example, IB-01 means the sample
59

belongs to the initial experiments where aluminum brackets (“B”) were used in the samples.
Likewise, IH-01 means the first experiment of the initial experiment where steel hinges (“H”)
were used to pull the layers away from each other.
Table 4.5. Summary of initial experiments sample
Sample Description Sample Label
1 Aluminum bracket glued with Crystalbond IB-01
2 Aluminum bracket glued with Loctite epoxy glue IB-02
3 Steel butterfly hinge with Crystalbond IH-01
4 Steel butterfly hinge with Loctite epoxy glue IH-02

Figure 4.20. 4 Samples of round 1. With two types of brackets and two types of glues
4.2.1.3 Results
The results of this section are analyzed from two aspects. The first aspect was to study the
feasibility of the DCB experiment with concrete layers that included testing two different
60

mechanisms to pull the concrete layers from each other and two different bonding agents for
gluing the pulling mechanisms to the concrete layers. The second aspect analyzed the results of
the DCB test with concrete layers.
a. Feasibility Results
The samples were aged for 21 days. The glue was applied to the concrete layers 7 days after
the samples were created. The strain rate was set at 0.5 mm/min according to the standard. Table
4.6 reviews the results of the experiments. Crystalbond failed for both of IB-01 and IH-01
samples. They came off of the sample disabling the testing machine to pull the layers apart. This
happened in both aluminum bracket and steel butterfly hinge. The sample is shown in Figure
4.21. It can be seen that the concrete samples were intact.
Table 4.6. Summary of round 1 experiments
Code Sample Description Results
1 IB-01 Aluminum bracket glued with Crystalbond Glue failed
2 IB-02 Aluminum bracket glued with Loctite epoxy glue Glue worked
Fixed bracket created issues
3 IH-01 Steel butterfly hinge with Crystalbond Glue failed
4 IH-02 Steel butterfly hinge with Loctite epoxy glue Glue worked
Hinge worked

61

Figure 4.21. IH-01 Sample (with aluminum bracket and Crystalbond)
The Loctite epoxy glue worked well. The glue did not come off during the course of the
experiments for both aluminum bracket and steel hinge samples (IH-02 and IB-02). Although the
aluminum bracket worked as well as the steel butterfly hinge, mounting the sample on the testing
machine created some complications. Because the angel of the bracket is fixed at 90º, everything
should have been perfectly aligned to run the experiments, which was not the case. So, samples
were installed in the testing machine under some pre-stressed tension not recommended by the
standard. The steel hinge with Loctite epoxy glue was selected as the best alternative to conduct
the main experiment with.
b. Analysis of DCB Results for IB-02
Figure 4.22 shows the delaminated layers of sample IB-02 after the DCB test. This was the
sample that was glued using Loctite epoxy glue and aluminum bracket. It can be seen that the
layers came apart completely.
62

Figure 4.22. Delaminated layers of sample IB-02
The top layer showed a cracked at about 51 mm of the edge of part and around 24.6 mm
away from the initial crack length (Figure 4.23). As per the DCB test, there was an initial crack
length of 25.4 mm between the layers. The pulling load applied energy to the parts and tried to
delaminate the layers but the crack resisted against delamination until the energy reached the
propagation energy and so it started to grow.
According to Figure 4.23, there is a second crack seen in the top layer. This crack happened
after the initial crack growth ended. The separation of the layers due to the crack growth created a
load and momentum on the top layer that led to the creation of the secondary crack and failure of
the samples. This is similar to a cantilever beam under load. Such failure of short concrete
cantilever beams is also reported in the study of concrete structures [54]. The load and
momentum applied to a short cantilever beam could lead to more growth of the crack so the
layers delaminated completely. The brittle nature of concrete together with having a low tensile
strength could be the main reason for the secondary crack. The DCB test on polymer composites
63

did not show any sign of a secondary crack while the crack grew continually all the way [55]. The
maximum normal stress that the sample could carry before full delamination was 1.49 MPa.

Figure 4.23. The top layer of delaminated samples that showed a crack.
c. DCB Results Analysis for IH-02 Sample
The delamination results of the sample with steel hinge was similar to the sample with
aluminum bracket. Figure 4.24 shows the delamination results of the IH-02. It can be seen that
similar to the IB-01 sample, the layers came apart completely.

Figure 4.24. IH-02 sample delamination result
64

It can be seen from Figure 4.25 that the top layer showed a second crack. The red line shows
the length border of the initial crack length. Similar to the IB02 samples, three points were
selected to measure the depth of the crack propagation in this sample. Two of them are along the
edges of the sample and one point is in the middle. The average of the crack growth in this
sample was 26 mm.

Figure 4.25. Initial crack propagation
65

The second crack occurred for the same reason it happened in the IB-02 sample. After the
propagation of the initial crack, the portion was under a load and momentum similar to a short
cantilever beam that created a force and a momentum on the fixed end which led to both more
delamination of the layers and creation of the crack near the fix end. The crack only appeared in
the top layer similar to the IB-02 sample. Such a crack is not seen in the DCB result of the
polymer composite laminates because of greater ductility of the polymer matrix.
The maximum traction stress at the failure was at 1.58 MPa. This is comparable with the IB-
02 sample and the slight difference could be because the concrete mixture is heterogeneous and
the existence of different particle sizes might create such differences.
4.2.1.4 Summary and Planning
The feasibility of DCB with concrete layers to investigate the bonding strength of the
concrete layers was studied in the initial experiments. Two types of adhesive were combined with
two types of brackets to pull the layers of concrete away from each other to test the bonding
strength of the layers. It was shown that the concrete epoxy with butterfly hinge showed the best
results in pulling the samples away from each other.
The DCB test results of the two samples with bracket aluminum and steel hinge were studied
as well. The initial crack that was intentionally created between the two layers was propagated
with the loading. The average of crack growth until the failure of the model (emergence of
secondary crack) was about 24.6 mm and 26 mm for aluminum bracket and steel hinge
receptively.
For the next stage of the experiments, two other interfacing mechanisms were selected to
incorporate to the parts and to study their effect on the delamination. So, in total, three types of
interfaces were designed. The first one was plain interface, the second one was rod reinforced
interface, and the third was threaded rod reinforced interface.
66

4.2.2 Primary Experiments
This section studies the bonding strength of the concrete layers further. It also investigates
the possibility of adding reinforcements to the layers to increase the bonding strength.
4.2.2.1 Experimental Procedure
In order to investigate the effect of different reinforcing mechanisms on bonding strength of
the concrete layers, statistical approaches were selected. Table 4.7 shows the experimental design.
Similar to the initial experiments, the coding starts with the letter “P” to show the sample belongs
to the primary experiments. Three types of interface were studied in this stage: plain interface,
polished rod reinforced interface, and threaded rod reinforce interface. Each type of interface was
considered at one level in a one-way layout design, and each level was replicated three times. So,
there were 9 experiments in round one. Table 4.8 shows the coding system for the 9 fabricated
samples.
Table 4.7. Experimental Design
Level Factor
1 Plain Interface
2 Polished Rod Reinforced
3 Threaded Rod Reinforced

Table 4.8. Coding of the primary experiment samples

Plain
Interface
Polished Rod
Reinforced
Threaded Rod
Reinforced
Sample 1 PP-01 PR-01 PS-01
Sample 2 PP-02 PR-02 PS-02
Sample 3 PP-03 PR-03 PS-03

4.2.2.2 Samples
Two interface reinforcing elements, including threaded rod and polished rod, were studied in
this section, as shown in Figure 4.26. The rods had 6mm diameter and were cut into pieces of 38
mm in length. Therefore, 14 mm of each rod was placed inside each layer. In each sample, three
67

rods were placed every 25.4 mm. The first rod was sited 10 mm away from where the initial crack
ends. Figure 4.27 shows the distribution of both the threaded and the polished rods between the
layers.

Figure 4.26. Polished and threaded rods used for reinforcement

Figure 4.27. Distribution of the rods between layers
In order to ensure an exact 25.4 mm for the length of the initial crack, a thin plastic film was
placed before casting the second layer. This plastic film that was coated with lubricant prevented
the first layer from bonding to the second layer. The plastic film had 25.4 mm length which was
68

the length of the initial crack and 25.4 mm width that was the width of the concrete layers. Figure
4.28 indicates the thin film and reinforcing rods before casting the second layer.

Figure 4.28. Threaded rods installed after the first layer is casted (PS-01 sample)
All of the samples were fabricated in one day using the same batch of material. Similar to
the initial experiments, the steel hinge was glued by the concrete-metal epoxy 7 days after the
samples were fabricated. The initial crack length (a
0
) was 25.4 mm. The second layer of concrete
was deposited 7 minutes after the first layer. Besides the plain interface samples, three pieces of
rod were used in each sample as described in Figure 4.27. The samples were aged 21 days to
reach the nominal properties. Figure 4.29, Figure 4.30, and Figure 4.31 show all 9 of the
fabricated samples for the primary experiments.

69

Figure 4.29. Fabricated samples with plain interface

Figure 4.30. Fabricated samples with rod reinforcing mechanism

70

Figure 4.31. Fabricated samples using threaded rods.

71

4.2.2.3 Results
In this section the results of the primary experiment were analyzed. The results of each of the
bonding reinforcement mechanisms were studied separately and compared to each other. The
DCB experiments for all nine samples were conducted in one day. The strain rate of the tension
machine was set at 0.5

mm/min.

1. Plain Interface
Table 4.9 summarizes the results of the plain interface samples. The hinge of the PP-01 was
not properly bonded so it came off during the course of the experiment. The layers of PP-02 and
PP-03 were delaminated. Figure 4.32 and Figure 4.33 show the delamination of both PP-02 and
PP-03 samples.
In the PP-02 sample, the layers of were separated. The sample showed a crack at around 54
mm of the edge of the sample. It should be noted that 25.4 mm of this growth belonged to the
initial crack test of the DCB test. When the strain energy reached the propagation energy, the
initial crack started to grow. Similar to the section 4.2.1 experiments, the initial crack showed an
early growth under the tension load. The growth of the initial crack is assumed to be at the point
that the second crack occurred, as shown in Figure 4.34. Similar to the initial experiments, three
points were selected, two on the two edges and one on the middle, to measure the growth. The
average depth of the initial crack growth was about 22 mm.
Table 4.9. The results of plain interface samples
Sample Description
PP-01 Glue failed
PP-02 Delaminated
PP-03 Delaminated

72

Figure 4.32. The delaminated PP-02 sample

Figure 4.33. The delaminated PP-03 sample

73

Figure 4.34. The growth of initial crack

The PP-03 sample showed a similar behavior under similar loading, as expected. The initial
crack showed an early growth. Then the growth led to a second crack. The average growth of the
initial crack in the PP-03 sample was measured at 25.6 mm as shown in Figure 4.35. This value is
4 mm larger than the PP-02 sample.

Figure 4.35. Initial crack growth at PP-03 sample

74

Table 4.10 shows the comparison of the initial crack growth for four tested samples (two
belong to the initial experiments and two belong to the primary experiments). The average of the
crack growths for all the samples was 24.55 mm. After the growth, the delaminated portion of the
layers behaved similar to short cantilever beams and failed at the fixed end. This failure was
consistent in all of the samples, as shown in Figure 4.36.
Table 4.10. The growth of the initial crack for all plain interface samples
Sample Crack Growth
PP-02 22 mm
PP-03 25.6 mm
IB-02 24.6 mm
IH-02 26mm

Figure 4.36. The cross section of the top layers in all plain interface samples
75

The traction stress at failure was 1.72 MPa, which was comparable to the other plain
interface samples. The peak traction stress was recorded at 1.83 MPa.
a. Polished Rod Reinforcement
The results of the DCB of the samples that were reinforced using polished rods are presented
and analyzed in this section. Table 4.11 lists the summary of the results. Similar to the PP-01
sample, the epoxy glue of the two rod reinforced samples failed and did not lead to conclusive
results. The PR-03 sample did make it through the test. The results can be seen in Figure 4.37.
Table 4.11. Summary of the results for PR-01, PR-02 and PR-03 samples
Sample Description
PR-01 Glue failed
PR-02 Glue failed
PR-03 Delaminated

Figure 4.37. DCB result of PR-03 sample

76

The delamination behavior of the rod reinforced samples was different from the plain
interface samples. According to Figure 4.38, there were two major differences between the
behaviors. The first difference was the fact that the growth of the initial cracked stopped when it
reached the reinforcing rod. As described in 4.2.2.2, the first rod was placed 10 mm away from
the initial crack. The propagation of the initial crack seemed to have been prevented by the
existence of the rod. So, the propagation of the initial crack was about 10 mm. After the initial
propagation, the delaminated portion of the layer acted as a short cantilever beam similar to the
previous samples. The applied load and the momentum on the delaminated portion could not pull
the reinforcing rod out of the concrete, so the short beam failed at the fixed end.
The other difference between the rod reinforced and plain interface samples was the fact that
the top layer did not delaminate all the way. As it was described in the plain interface samples,
the top layer came off of the bottom layer completely. However, the existence of the rod in the
sample prevented the growth of the crack even further and the sample cracked where the first
reinforcing rod was placed as can be seen in Figure 4.38. Therefore, the bonding strength of the
layers was improved.

Figure 4.38. Initial crack growth in PR-03 sample

77

The traction that propagates the crack was 2.61 MPa, showing the sample’s resistance to
crack propagation was about 40% more than the plain interface sample. The difference was
believed to be because of the existence of the reinforcing rods. In fact, the reinforcing rods
increased the resistance to the crack propagation and led to carrying more load upon failure.
b. Threaded Rod Reinforcement
The summary of the DCB results of the threaded rod reinforced samples is listed in Table
4.12. Two of the samples did not perform well under the DCB test. The contact of the epoxy glue
and the surface of the concrete seemed to cause issues for the samples. The PS-01 sample after
the DCB is shown in Figure 4.39.
Table 4.12. Summary of the results for PS-01, PS-02 and PS-03 samples
Sample Description
PS-01 Delaminated
PS-02 Test failed
PS-03 Test failed

Figure 4.39. PS-01 sample after the DCB test
The delamination behavior of the PS-01 sample was similar to the PR-03. The propagation
of the initial crack stopped when it reached the threaded rod. The first rod was placed about 10
78

mm away from the initial crack. It can be seen in Figure 4.40 that the crack propagated until it
reached the rod. After that, the delaminated portion of the top layer behaved like a short
cantilever beam and cracked at the fixed end. Similar to the PR-03 sample, the load and
momentum on the delaminated section could not pull the rod out of the bottom layer. The threads
of the rod caused the concrete to stick more to the rod compared to the case of polished rod.
The peak of the traction happened at 2.96 MPa, which was 35% more than samples with
plain interface. This could be because of the existence of the threaded rod in the concrete layers.
This amount was higher than the PR-01 (the polished rod reinforced sample). The extra amount
of load that the layers could carry might be because of the threads of the rod that stuck to the
concrete better than the polished rod, as seen in Figure 4.41.

Figure 4.40. Initial crack growth in PS-01 sample

79

Figure 4.41. Comparison of threaded and polished rod reinforced samples

4.2.2.4 Summary and Planning
In the primary experiments, the possibility of incorporating reinforcements for increasing the
bonding strength was studied. One way lay-out design was conducted where each level included
one of the reinforcing mechanisms. Level one was samples with plain interface and no
reinforcement, level two was samples where a polished rod was used as reinforcement, and level
three was samples where a threaded rod was used as reinforcement. The plain interface samples
were completely delaminated. A second crack was seen on the top layer of the concrete. The
average length of crack growth in those samples were 24.55 mm. The results of the plain interface
samples were in compliance with the initial experiments.
The reinforced samples showed a different behavior when the load was applied. In the
polished rod sample, the crack started to propagate and it stopped when it reached the polished
80

rod. So the growth of the initial crack was believed to be about 10 mm, which was where the
polished rod was placed into the sample. The second and third rod did not do much in reality. The
threaded rod sample behaved similar to the polished rod with one difference. The concrete made a
better bond to the threads of the rod which helped to hold the layers together more than the
polished rod.
According to the results of primary experiments, the presence of reinforcing rods stopped the
crack propagation. In order to investigate this phenomenon deeper, an experimental procedure
was designed where the rods are placed in various distances from the initial crack, which is
explained in the next section.
4.2.3 Final Experiments
The purpose of the experiments of this section was to investigate the possibility of adding
reinforcing rods to prevent the initial crack from propagating.
4.2.3.1 Experimental Procedure
For this stage, four experiments were designed. In each of the samples, one rod was used
with a specific distance from the initial crack. The first sample had a rod 15 mm away from the
crack, the rod of the second sample was placed 30 mm away from the crack, the third sample had
a rod 45 mm away from the crack, and the last sample had a rod 60 mm away from the crack.
Table 4.13 and Figure 4.42 summarize the experiments.
Table 4.13. The design of the final experiments
Rod Distance from the crack Code
1 15 mm FS-01
2 30 mm FS-02
3 45 mm FS-03
4 60 mm FS-04

81

FS-01

FS-02

FS-03

FS-4
Figure 4.42. The position of the reinforcing rod in the samples
4.2.3.2 Samples
The samples of the final experiments were fabricated using the same wood mold that was
used for previous experiments. The same batch of material was utilized for consistency. After the
first layer was casted, the plastic film was used to create the initial crack length. Then, the
reinforcing rod was placed in the predefined distances according to Figure 4.42. The second layer
was added to the sample seven minutes after the first layers similarly as the previous samples.
The hinges were glued to the samples seven days after the samples were created. All of the
samples were aged 21 days before the DCB test. Figure 4.43 shows the fabricated samples.

82

Figure 4.43. Samples of the final experiment
4.2.3.3 Results
The experiments were all conducted in one day. The strain rate was chosen at 0.5 mm/min.
Table 4.14 summarizes the result of the experiments. The FS-01 sample, with the rod installed at
10 mm from the initial crack, is shown in Figure 4.44. It can be seen that the crack reached the
rod and then the sample showed the second crack similar to the previous samples.
The FS-02 sample was the one that had a reinforcing bar at 30 mm away from the initial
crack. Figure 4.45 describes the sample after the DCB test. The result of the DCB test on FS-02
was similar to the FS-01; the reinforcing rod stopped the crack from propagation, and then the
material failed and cracked under the applied load and momentum resulting from load.

83

Table 4.14. Summary of the results of the final experiments
Sample Result
1 FS-01 Crack stopped at reinforcing rod
2 FS-02 Crack stopped at reinforcing rod
3 FS-03 Crack did not reach the rod
4 FS-04 Crack did not reach the rod

Figure 4.44. FS-01 sample after the DCB test

84

Figure 4.45. FS-02 sample after the DCB test

The result of the FS-03 sample after the DCB test is shown in Figure 4.46. The result of FS-
03 is different from the other two samples. The crack did not reach the reinforcing rod that was
placed at 45 mm away from the initial crack when the failure of the sample happened. This
matches the results of the initial and primary experiments for plain interface samples where the
initial propagation of the crack ranges from 22-30 mm. Therefore, it was rational to expect that
the failure of the sample under the load and momentum happens before the crack reaches the
reinforcing bar that was placed 30 mm away from the initial crack. The average growth of the
initial crack for the FS-03 sample for three points that was 23.6 mm, which matches the
experiments of the plain interface.
85

Figure 4.46. FS-03 sample after the DCB test
4.2.4 Summary
The results of the FS-01, FS-02, and FS-03 samples indicates that the propagation of the
crack could be stopped by placing the reinforcing rods inside the sample. Unlike the plain
interface samples that were fully delaminated, the reinforcing rods of the FS-01 and FS-02
samples stopped the crack propagation, thus preventing the layers from being delaminated. This
can be used in the design of the reinforcing rods during the Contour Crafting process to prevent
delamination of the layers.
Moreover, the bonding strength that were generated using DCB test belongs to the worst
situation where bonding is the weakest. The bonding strength of other parts of the layers should
be calculated according to the suggested testing at Figure 4.47.
86

Figure 4.47. Suggested testing to calculate the bonding strength

4.3 Stage Three Results
This section talks about the results and analyses of the simulation model that was developed
using the experimental data. The simulation model that was developed at the beginning of the
study required layer adhesion parameters. Such parameters could only be derived from
experiments that were conducted in Stage Two. The results of this section are explained in two
steps. At first the findings of the experiments were used in the developed finite element model of
Stage One. Two models were developed using plain interface and reinforced interface data. The
results and data were analyzed and compared.
In the next step, the developed simulation tool was used to predict the minimum bonding
strength required between the layers. And. finally, the possibility of adding curvature and
corrugation to the walls and investigating the effect of such shapes on the strength of the layered
structures was studied by developing a complex finite element model.
4.3.1 Simulation Comparison of Walls with Different Layer Bonding Strength
A full-scale layered model developed in Stage One was used for this part of the study. It
consisted of 120 layers (with 119 inter-layer joints). The dimension of each layer was 426.72 cm
(length) x 15.24 cm (width) x 2.54 cm (thickness). For the interaction model, the CZM settings
87

were used. Node to surface type of interaction was selected for the interaction between the top
layer and the bottom one. The size of each element was 7.62cm x 9.652cm x 2.54cm. In total,
there were 31,680 nodes and 10,320 elements in the model, similarly as the initial simulation. The
simulation was executed once with the cohesive parameters of the threaded rod reinforced and
once with the cohesive parameters of plain interface samples from the Stage Two results. The
model does not contain the existence of the rods to prevent complexity and to increase
computation time.
Table 4.15 compares the wall deflection for four developed simulations. The second and
third columns hold the deflection data of layered wall simulation with plain and reinforced
interface. The fourth column corresponds to the deflection data of the wall simulation of Stage
One where the theoretical cohesive parameters were utilized. The deflection data are also plotted
in the Figure 4.50. The simulation of the plain interface wall showed the highest amount of
deflection among all four simulations. This was because of the relatively weak bonding strength
between the concrete layers that were studied in Stage Two of this study. The maximum
deflection of 2.76 mm took place at a height of 3 meters according to the simulation. The
reinforced interface simulation showed a huge improvement compared to the theoretical
simulation and plain interface sample.
88

Figure 4.48. Deflection comparison of the simulated walls
The delamination values of the first 15 layers of the plain interface, reinforced interface, and
theoretical data walls are shown in

Table 4.16 and Figure 4.49. The delamination of first layer for the plain interface simulation
was about 0.5 mm more than the theoretical data while the reinforced interface simulation was
reduced to 0.189mm. The delamination became negligible after 10 where the value dropped to
less than 0.009 mm for the reinforced interface simulation. The delamination value became
negligible after 15 layers for the plain interface samples. The theoretical values are placed
between the plain interface and reinforced interface samples.
The bonding strength of the layers should be enhanced in order to the lower the delamination
magnitude of the first layer below 0.01 mm. Using threaded rods, the critical energy to propagate
89

the delamination was increased from 0.4 N/mm to 1.1 N/mm, which decreased the delamination
from 0.996 mm to 0.189 mm.
Table 4.15. Wall deflection for different heights
Wall
Height(m)
Plain
Interface
Wall(mm)
Reinforced
Wall(mm)
Wall with
Theoretical CZM
Parameters
0 0 0 0
0.3048 1.5942 0.6346 1.4196
0.6096 1.8251 0.6354 1.4432
0.9144 1.9693 0.6395 1.4432
1.2192 2.1020 0.6395 1.4437
1.524 2.3922 0.6395 1.4437
1.8288 2.3922 0.6416 1.4437
2.1336 2.3914 0.6416 1.4432
2.4384 2.4022 0.6447 1.4497
2.7432 2.5401 0.6547 1.4515
3.048 2.7657 0.6568 1.4827

Table 4.16. Delamination for plain interface and rod reinforced walls
Layer Number
Plain
interface
wall (mm)
Rod
Reinforced
wall (mm)
Theoretical
Interface
Wall(mm)
2 0.996 0.189 0.434
3 0.856 0.123 0.308
4 0.465 0.083 0.215
5 0.325 0.071 0.149
6 0.231 0.054 0.103
7 0.150 0.027 0.071
8 0.112 0.02 0.049
9 0.070 0.017 0.034
10 0.050 0.009 0.023
11 0.035 0.006 0.016
12 0.027 0.006 0.011
13 0.019 0.007 0.007
14 0.006 0.003 0.003
15 0.003 0.001 0.001
90

Figure 4.49. Delamination comparison graph for the first 15 layers
4.3.2 Determining the Minimum Bonding Strength
In order, to find the minimum bonding strength to make the delamination negligible,
numerous finite element models with various bonding strengths were conducted.
Similar to the previous simulations, a full-scale layered model was used in the simulation.
The model contained 120 layers of 426.72 cm (lengths) x 15.24 cm (widths) x 2.54 cm
(thickness). The first stimulation was done using the cohesive parameters and critical strain
release energy of the reinforced interface. According to the previous simulations, it was apparent
that the first layer went through the maximum delamination magnitude. Hence, the value of the
fracture stiffness was reduced for each experiment and the magnitude of delamination was
measured only for the first layer. After each simulation, the magnitude of the delamination was
91

measured. If the amount of delamination did not satisfy the requirement, the strain release energy
was increased by 20%.
Table 4.17 shows the results of the designed simulations. The critical propagation energy for
the reinforced samples tested in Stage Two was 1.1 N/mm, which led to delamination of 0.189
mm in the first layer. Upon exploring higher values, 3.3 N/mm was determined as the minimum
bonding strength required for layer by layer fabrication.
It can be concluded that the adhesion of the layers in the layer by layer fabrication should
have the minimum critical strain release energy to be comparable to the none layered structures.
Figure 4.50 shows the strain and stress distribution of the layered structure with layers that had
the minimum required adhesion. That strain distribution has shifted from the bottom to the center
and top of the wall and looks similar to the none layered structures.
Table 4.17. Predictive simulation design with results
Simulation Bonding Strength (MPa)
Delamination
(mm)
1
2.6
0.189
2
3.1
0.133098592
3
3.8
0.080665813
4
4.5
0.052380398
5
5.5
0.031271879
6
6.5
0.015793879
7
7.9
0.008491333
8 9.3 0.007514454

92

Figure 4.50. Stress and strain distribution comparison for the optimized layered model and
none layered model

4.3.3 Effect of Geometry Complexity on a Structure’s Strength
After investigating the bonding strength of the concrete layers and determining the minimum
strength required to have comparable results with conventional structures, it was time to develop
a statistical approach to investigate the extent to which geometry complexity, which may be
effortlessly achieved by the Contour Crafting technology, can enhance the strength of structures.
93

4.3.3.1 Curved Walls
Creating curvature in structures is an old construction approach to enhance the strength.
Engineers of centuries past placed curved adobe plates on top of each other, creating structures
that are still standing even in earthquake prone regions (Figure 4.51). The secret to that
exceptional strength was the curvature incorporated in them. Such approaches are not being
utilized so frequently any more due to complexity of implementation and additional costs that are
imposed on the projects. However, these types of assemblies can be implemented effortlessly
using automatic construction approaches. This section studies how curvature will affect the
flexural strength of the walls under wind loading.

Figure 4.51. Interior of an adobe house [59]

a. Rectilinear Walls Flexural Study
The wall section that was used in the previous simulations was used to study the effect of
curvature in this part of the work. It includes 120 layers of 25 mm thickness, having height of 3
meters. The thickness of the wall was chosen as variable in the study. In order to have a basis for
the comparison, the layered rectilinear walls with various thicknesses were simulated to have a
basis of the comparison at first. The values of the wall thickness that the rectilinear wall
94

simulations had were 10, 15 and 20 centimeters, which were selected based on the average
dimensions of typical residential walls explained in section 3.1. Figure 4.52 shows the modeled
walls before the simulation and Figure 4.54 through Figure 4.56 portray the results. The
deflection results of each of the walls for different heights are also listed in Table 4.18 and shown
in Figure 4.53. The wall with 10 cm thickness had the highest magnitude of deflection (19.86
mm) and the wall with 20 cm thickness had the lowest amount of deflection (0.84 mm). However,
the maximum deflection of the 15 cm thick wall was 2.34 mm, which was close to the 20 cm
thick wall.

Figure 4.52. Rectilinear wall simulations with the thickness of 10cm, 15cm, and 20cm

95

Table 4.18. Rectilinear wall deflections
Wall Height(m)
10 cm thick Wall
Deflection
15 cm thick Wall
Deflection
20 cm thick Wall
Deflection
0 0 0 0
0.25 1.36 0.192 0.06
0.5 2.94 0.39 0.13
0.75 4.58 0.56 0.21
1 6.2 0.78 0.26
1.25 7.79 0.97 0.33
1.5 9.35 1.17 0.4
1.75 10.7 1.36 0.46
2 12.19 1.56 0.534
2.25 13.5 1.76 0.6
2.5 16.54 1.95 0.67
2.75 18.2 2.15 0.73
3 19.85 2.34 0.84

Figure 4.53. Rectilinear walls deflection comparison
96

Figure 4.54. The simulation result of 10 cm thick rectilinear wall

Figure 4.55. The simulation result of 15cm thick rectilinear wall
97

Figure 4.56. The simulation result of 20cm thick rectilinear wall

b. Curved Walls Flexural Study
Before incorporating curvature and studying its effect on the deflection and comparing it
with the rectilinear wall, a sensitivity simulations-analysis was designed to study the significance
of the influencing factors on the deflection of the wall. Curvature radius, curvature direction
(concave or convex), and the thickness of the structures were identified as the influencing factors
on the deflection of the structures.
The curvature in the walls is defined by the radius of the circle that corresponds to or
matches the arch. The arch length is a portion of the perimeter of the circle that defines the arc.
Figure 4.57 shows the curvature parameters. The arch length was not a variable, nor was the arch
angle defining the arch length, which was fixed. So the radius was selected as a variable. The
other variable was convexity or concavity of the wall. We were interested to see which surfaces
would enhance the strength of the wall further, and, finally, the thickness of the wall was selected
as a variable.
98

Figure 4.57. Curvature parameters
A two level factorial experiment with no replicates of the center point were selected to do
the sensitivity analysis on the suggested factors. Curvature direction was qualitative data and
could not be used for center point simulations, so the full factorial experiment was designed to be
conducted in two separate parts, one with concave curvature and one with convex. So the
simulations were divided into two separate factorial experiments each of them having two factors.
Table 4.19 shows the factorial levels for convex and concave simulations.
Table 4.19. Factorial levels for the simulations
Factorial levels for concave structure simulation
Factors Low Center Point High
Curvature Radius 8 meter 10 meter 12 meter
Wall Thickness 10 centimeter 15 centimeter 20 centimeter

Factorial levels for concave structure simulation
Factors Low Center Point High
Curvature Radius 8 meter 10 meter 12 meter
Wall Thickness 10 centimeter 15 centimeter 20 centimeter
99

The low level of curvature radius was selected at 8 meters, center point at 16 meters, and
high level at 32 meters. The levels of wall thickness were selected at 10, 15, and 20 centimeters
respectively. Figure 4.58 through Figure 4.62 illustrates the samples of the convex simulations,
and the details of the curvature and wall thickness parameter of each simulation is included in
Table 4.20.
Table 4.20. Structural parameters used in different simulations
Simulation Order Radius Wall Thickness
Maximum Deflection
1 8 meter 20 centimeter 0.275 mm
2 32 meter 10 centimeter 5.1 mm
3 8 meter 10 centimeter 0.87 mm
4 32 meter 20 centimeter 0.51 mm
5 20 meter 15 centimeter 1.1 mm

Figure 4.58. Simulation 1 of the curvature study
100

Figure 4.59. Simulation 2 of the curvature study

Figure 4.60. Simulation 3 of the curvature study

Figure 4.61. Simulation 4 of the curvature study

101

Figure 4.62. Simulation 5 of the curvature study
Analysis of variances (ANOVA) table of the conducted sensitivity analysis is listed in Table
4.21. The ANOVA tables show that the curvature radius and wall thickness were significant and
influencing the deflection. The plot of the main effects are illustrated in Figure 4.63. The plot
suggested the lower value of radius is preferred over higher values and higher values of layer
thickness led to lower deflection, as was expected. Therefore, as suggested parameters based on
the sensitivity analysis, the radius should be selected a low value and the wall thickness should be
selected at a high value.
Table 4.21. ANOVA table of the factorial simulations
Source DF Adj SS Adj MS F-Value P-Value
Model 3 11.9824 3.9941 1 0.609
Linear 2 11.7051 5.8526 1.47 0.504
Radius 1 4.9841 4.9841 1.25 0.075
Wall Thickness 1 6.7211 6.7211 1.68 0.034
Curvature 1 0.2773 0.2773 0.07 0.836
Error 1 3.99 3.99

Total 4 15.9724

102

Figure 4.63.Main effect plot
After the sensitivity analysis of the concave wall, the second part of the simulation design
explained in Table 4.19 was conducted. The results are recorded in Table 4.22. The results of the
concave loading of the wall was very similar to the convex samples so the rest of the finite
element models were conducted on the convex wall.

Table 4.22. Structural parameters used in different simulations for the concave wall
Simulation Order Radius Wall Thickness
Maximum Deflection
1 8 meter 20 centimeter 0.292 mm
2 32 meter 10 centimeter 5.38 mm
3 8 meter 10 centimeter 0.86 mm
4 32 meter 20 centimeter 0.63 mm
5 20 meter 15 centimeter 1.23 mm

c. Curved Wall versus Rectilinear
The last part of study of the curved walls included the comparison of such structures with
rectilinear structures of the same dimensions. The deflection data of the rectilinear wall with 10
103

cm thickness are compared to the curved walls with 10 cm thickness. The two walls of the
comparison had 8 meter and 32 meter radiuses that were explained during the sensitivity analysis.
Figure 4.41 demonstrates such deflection comparison. The deflection values are also included in
Table 4.23. Such data for the 20 cm wide walls are recorded in Table 4.24 and Figure 4.65.

Table 4.23. Curved versus rectilinear deflection comparison of 10 cm thick wall
Height (meter)
32m Radius Deflection
(mm)
8m Radius
Deflection (mm)
Rectilinear Deflection
(mm)
0 0 0 0
0.25 0.425 0.07 1.36
0.5 0.85 0.14 2.94
0.75 1.2 0.22 4.58
1 1.7 0.29 6.2
1.25 2.16 0.366 7.79
1.5 2.55 0.44 9.35
1.75 2.97 0.51 10.7
2 3.4 0.58 12.19
2.25 3.82 0.65 13.5
2.5 4.25 0.73 16.54
2.75 4.67 0.806 18.2
3 5.1 0.87 1.85
104

Figure 4.64. Curved versus rectilinear deflection comparison of 10 cm thick wall

Figure 4.65. Curved versus rectilinear deflection comparison of 20 cm thick wall

105

Table 4.24. Curved versus rectilinear deflection comparison of 20 cm thick wall
Height ( meter)
8m Radius Deflection
(mm)
32m Radius
deflection(mm)
Rectilinear deflection
(mm)
0 0 0 0
0.25 0.02 0.042 0.06
0.5 0.045 0.085 0.13
0.75 0.068 0.12 0.21
1 0.09 0.17 0.26
1.25 0.11 0.21 0.33
1.5 0.137 0.25 0.4
1.75 0.16 0.29 0.46
2 0.183 0.34 0.534
2.25 0.206 0.38 0.6
2.5 0.229 0.42 0.67
2.75 0.25 0.46 0.73
3 0.27 0.51 0.84

According to the deflection data in Figure 4.41 and Table 4.23, curvature in fact has a
significant effect on the flexural strength of the walls. The rectilinear wall of the study with 10 cm
thickness showed 19.8 mm deflection under load. The relatively small curvature of 32 meter
reduced the deflection to 5.1 mm while the radius of 8 meters to the curvature brought the
maximum deflection down to 0.87 mm which was approximately 4 times reduction for a small
curvature and more than 15 times reduction for a more curved wall.
The deflection data of the 20 cm wall that were reported in Figure 4.65 and Table 4.24
revealed the effect of curvature in the walls. The relatively thick rectilinear wall of 20 cm showed
the small deflection of 0.84 mm under load. Featuring the 32 m radius curvature reduced the
deflection to 0.51 mm (about 1.5 times) and the curvature of 8 m radius reduced the deflection to
0.18 mm (about 4.5 times).
Comparing the results of the 10 cm thick wall and 20 cm thick wall, it could be inferred that
the wall with 10 cm thickness and 8 m radius curvature showed more flexural strength as did the
wall that had twice the thickness of the curved wall. Therefore, the curvature would potentially
increase the flexural strength of structures which would lead to a huge savings in material and
106

construction lead time, while also being stronger. Such curvature incorporation can only by
implemented using automated construction approaches. Figure 4.66. compares the deflection of
the rectilinear wall of with 20 cm thickness with the curved wall of the 8 m radius with 10 cm
thickness. Other researchers recorded a similar effect of curvature on the flexural strength of
various types of structures such as trusses or precast beams [60], [61].

Figure 4.66. Deflection comparison of a thin curved wall and a thick rectilinear wall.
On the other hand, by comparing the results of the rectilinear versus curved wall with 20 cm
thickness, it could be concluded that the effect of curvature is more effective on thinner walls
while it diminishes when the thickness increases. To study and locate the “sweet spot” where the
curvature has the peak effect, a new series of simulations were planned, as shown in Table 4.25.
Table 4.25. Simulation plan for the curvature effectiveness study

Radius

8 12 16 20 24 Rectilinear
Thickness 5 Run 1 Run 6 Run 11 Run 16 Run 21 Run 26
10 Run 2 Run 7 Run 12 Run 17 Run 22 Run 27
15 Run 3 Run 8 Run 13 Run 18 Run 23 Run 28

20 Run 4 Run 9 Run 14 Run 19 Run 24 Run 29
25 Run 5 Run 10 Run 15 Run 20 Run 25 Run 30

107

Five different curvatures were combined with five different wall thicknesses, and the
deflection of each combination was deducted from the deflection of the rectilinear wall of the
same thickness. There were 25 combinations of wall thickness to wall curvature plus 5
experiments with the rectilinear walls. Therefore, 30 finite element simulations were conducted in
order to study the influence of curvature.
The simulation results are listed in Table 4.26. Each cell corresponds to the deflection of that
combination. The simulation figures are included in Appendix II. Table 4.26 reveals significant
information about changes of the wall deflection when the thickness is fixed and radius is a
variable and vise versa. For instance, it can be seen from the 5 cm thickness row that the
deflection of the wall increases when the radius goes up. The slope of the increase of the
deflection of the 5 cm thick wall is a lot more than the 15 cm wall, for example. The data were re-
sorted in Table 4.27 to indicate the regions where curvature affected the deflection. So the new
order of the table has 4 columns: thickness, radius, deflection, and deflection difference (delta
deflection). The data in Table 4.27 are plotted in a 3D graph and as 2D contours in Figure 4.67
and Figure 4.68.
Table 4.26. Simulation results plan for the curvature effectiveness study

Radius

8m 12m 16m 20m 24m Rectilinear
Thickness 5 cm 1.93 mm 4.5 mm 8.4 mm 12.1 mm 17.8 mm 28.1 mm
10 cm 0.87 mm 1.69 mm 2.5 mm 3.25 mm 3.9 mm 19.85 mm
15 cm 0.49 mm 0.77 mm 0.94 mm 1.1 mm 1.21 mm 2.34 mm

20 cm 0.27 mm 0.32 mm 0.43 mm 0.46 mm 0.48 mm 0.84 mm
25 cm 0.16 mm 0.21 mm 0.22 mm 0.23 mm 0.23 mm 0.4 mm

108

Table 4.27. The result of the 30 simulations
Radius (m) Thickness (cm) Deflection (mm) Delta Deflection(mm)
8 5 1.93 26.17
12 5 4.5 23.6
16 5 8.4 19.7
20 5 12.1 16
24 5 17.8 10.3
12 10 1.69 18.16
8 10 0.87 18.98
16 10 2.5 17.35
20 10 3.25 16.6
24 10 3.9 15.95
16 15 0.94 1.4
8 15 0.49 1.85
12 15 0.77 1.57
20 15 1.1 1.24
24 15 1.21 1.13
20 20 0.46 0.38
8 20 0.27 0.57
12 20 0.32 0.52
16 20 0.43 0.41
24 20 0.48 0.36
24 25 0.23 0.17
8 25 0.164 0.236
12 25 0.21 0.19
16 25 0.22 0.18
20 25 0.23 0.17

109

Figure 4.67. The deflection difference surface shown in a 3D plot

Figure 4.68. The deflection difference surface shown in a 2D contour plot
110

Figure 4.67 shows the surface formed by delta deflection between the curved wall
simulations and rectilinear wall simulations. It is evident that the value of the delta deflection
rises when the curvature radius increases and wall thickness decreases. This phenomenon can be
seen more clearly in the plotted contour of Figure 4.68. The dark blue regions represent the area
where the delta deflection is negligible, and red regions are those of the highest delta deflection. It
can be seen that when the thickness of the wall goes above 15 cm, the effect of curvature
diminishes, and when it goes below 15 cm and the radius is around 8 m, the curvature makes a
great difference. The contour of Figure 4.68 can be used as a guideline for designing structures
for Contour Crafting.
4.3.3.2 Innovative Wall Shapes
Contour Crafting has proved to be able to create geometrically complex structures such as
the corrugated wall [2] that is shown in Figure 4.69. The aim of this section is to study the
flexural behavior of such walls in comparison with rectilinear solid walls of the same dimensions.

Figure 4.69. Corrugated wall created by Contour Crafting
111

a) Corrugated Rectilinear Wall Study
The 20 cm thick solid rectilinear wall was selected for the comparison. Such a wall has
enough thickness to allow for incorporation of corrugation patterns. Two innovative patterns of
parallel hatch and zigzag, were selected to be simulated and compared with solid walls. Figure
4.70 demonstrates the cross section of the two types of corrugated walls that were studied. The
angles of the bars were 45° for both zigzag and hatched simulations, while the thickness of the
bars were 5 cm. Figure 4.71 and Table 4.28 indicate the simulation results of the study. The
maximum deflection of the hatched and zigzag corrugated walls were recorded at 5.4 mm and
1.74 mm. It is evident that corrugation did not improve the flexural strength and made the
strength worse than the solid rectilinear wall as the maximum deflection of the rectilinear wall
was 0.87 mm. This was believed to be due to the relative thinness of the walls of the corrugated
structure. The deformation of the cross section was also not symmetric due to the asymmetric
geometry of the corrugated simulated wall. The deflection of various heights of the three
simulated walls are plotted in Figure 4.72. Although the simulated rectilinear corrugated
structures did not have comparable load carrying capacity as compared to the solid structures,
they provide other benefits such as sound and energy insulation and could be used for internal
walls.

Figure 4.70. The cross section of the simulated hollow wall
112

Table 4.28. The simulation results of the corrugated wall study
Simulation Max Deflection
Hatched Wall 5.4 mm
Corrugated Wall 1.74 mm
Solid Wall 0.87 mm

Figure 4.71. The simulation result of the walls with two different patterns

113

Figure 4.72. The deflection of the simulated walls in various heights.
b. Corrugated Curved Walls
Incorporating curvature in the rectilinear walls led to a significant increase in the flexural
strength of the contour crafting structures in section 4.3.2. Section 4.3.3.a studied the influence of
hollowing the walls and creating corrugated structures. It was shown that corrugating the walls
decreases the flexural strength in comparison with solid structures. This section studies the effect
of curvature implementation in the corrugated structures. The curvature of the structure was
selected based on the result of section 4.3.3.c. It was concluded that the curvature of 8 m
increased the flexural strength of the structures. So, 8 m curvature was incorporated in the zigzag
corrugated structures of chapter 4.3.3.a, which had a better flexural strength compared to the
hatched corrugated one. Figure 4.73 shows the model of the layered zigzag corrugated wall, and
Figure 4.74 illustrates the simulation results. The simulations revealed that the maximum
deflection of the wall reduced from 1.74 mm to 0.36 mm, verifying the effect of curvature that
was studied in the previous sections. The maximum deflection of the solid rectilinear wall was
114

recorded at 0.87 mm, while the deflection of the curved solid wall was at 0.27 mm. The
deflection results of the 20 cm thick walls of all types are plotted in Figure 4.75.
4.3.4 Section Summary
The layered, curved zigzag corrugated structures presented the most desirable flexural
strength and seemed to be the most appropriate type of structures to be incorporated in
construction. Such structures not only increase the strength of structures by more than 100%, they
decrease the weight and material cost of structures and provide energy and sound insulation.
Implementation of such structures are only possible using automated construction approaches
such as Contour Crafting.

Figure 4.73. Curved corrugated model

Figure 4.74. The simulation of curved corrugated structure
115

Figure 4.75. Deflection comparison of the walls with 20 thickness
4.4 Chapter Summary
This chapter presented and analyzed the results of the three stages of this thesis. The study
started with a preliminary simulation that studied the bonding strength of the layers and flexural
behavior of the structures. That was followed by an experimental stage that deeply studied the
bonding strength of the concrete layers. And, finally, the last stage of the study incorporated the
experimental data into the simulation model while it studied and compared the strength of the
solid rectilinear structures, solid curved structures, and corrugated hollow structures.

116

5 Concluding Remarks
5.1 Conclusion
The purpose of this research is to investigate the structural properties of structures built by
automated construction approaches such as Contour Crafting. The major objectives of the work
were:
• To make an appropriate model that takes the layer interaction into consideration
• To establish a methodology that quantifies the bonding strength of the layers of concrete
• To study and compare the layered structures with none layered structures
• To investigate the effect of geometry complexity such as curvature or corrugated structures
on the strength of walls
• To create a guideline for future users of Contour Crafting
The study was conducted in three steps that were mutually connected. The results of each
stage were analyzed and used in next stage of the study. The summary of each stage of the study
is listed below:
• Stage One: This stage began with an effort to find the most proper model to investigate the
bonding strength of the concrete structures. The events and findings of this stage were:
• A proper simulation model for studying bonding strength was developed. Cohesive zone
modeling (CZM) was selected as an appropriate approach to study the concrete layer
adhesion. CZM is a common approach for studying the delamination of adhesive joints
and composite laminates.
• A full scale simulation model for a layered concrete wall was presented. It was shown
that the layers demonstrated a great deal of delamination toward the bottom of the wall.
• The strength and behavior of a full scale wall was compared with a solid wall of the same
dimensions. It was shown that the majority of the deflection of the layered wall came
117

from the delamination of the bottom layers. The rate of increase of the deflection was
steep at the beginning, then it did not change toward the height of the layered wall while
the deflection of the none layered wall was increased gradually toward the end of the
wall.
• It was concluded that an experimental approach is required to derive the parameters of
the CZM model that quantifies the bonding strength of the layers.

• Stage Two: Stage Two was an experimental approach aimed at quantifying the bonding
strength of the concrete structures. The major activities and findings of this stage included:
• A Double Cantilever Beam (DCB) test was chosen as the most appropriate test for
extracting the parameters of the Cohesive zone modeling for the finite element
simulation.
• A DCB test was successfully conducted on the concrete structures.
• The DCB test showed that the depth of the growth of the initial crack between the layers
is between 10-30 millimeters. Such layers had a plain interface and were not reinforced.
• The possibility of enhancing the bonding strength between the layers of concrete was
studied.
• Two types of reinforcements were utilized and studied. A small piece of threaded rod and
polished rods were placed between the layers of concrete for the DCB test.
• The DCB test on the reinforced interface was conducted together on samples with plain
interface.
• The experiments showed that the reinforcement rods stopped the growth of the initial
crack as both reinforced samples showed less delamination as did the plain interface.
• A final round of DCB experiments was conducted on the threaded rod reinforced
samples to verify if such rods could prevent the crack growth.
118

• The tests showed that the initial crack propagation stopped when it reached the rods that
were positioned 15 mm and 30 mm away from the crack. The crack did not reach the rod
for the samples whose rods were placed 45 mm and 60 mm away from the crack.
Therefore, it was concluded that the reinforcing rods could prevent the crack growth and
delamination of the layers at the end.
• It was also suggested that the reinforcing rods should be placed at a maximum at 20 mm
away from the crack to be considered effective in helping the bonding strength.
• The critical strain release energy of the plain interface sample and the reinforced samples
were calculated. This critical strain release energy was 0.5 N/mm and 1.1 N/mm for plain
and reinforced interfaces respectively.
• Stage Three:
• The critical strain release energy of the plain and reinforced interface samples was used
to re-run the FEM model that was developed in Stage One.
• The simulations showed that he maximum deflection of the walls for the plain and
reinforced interface were 2.6 mm and 0.78 mm respectively.
• The delamination of the first layer for plain and reinforced interface samples were
recorded at 0.996 and 0.189 mm.
• A sequence of simulations was designed to determine the minimum bonding strength
required to make the delamination negligible. It was found that the minimum bonding
strength required for such a result was 3.3 N/mm.
• The effects of geometry twists, such as curvature or hollow corrugated structure, on the
flexural strength of the walls were studied.
• The deflection of the solid rectilinear structure was quantified as the basis of the
comparison.
119

• A sensitivity analysis was conducted to report the significance of wall thickness and
curvature radius in enhancing the flexural strength. It was found that both factors were
significant, and the lower curvature radius was preferred.
• The simulations revealed that the flexural strength of a 10 cm thick curved wall is
comparable to a 20 cm thick rectilinear wall. This leads to constructing lightweight
structures while carrying more load capacity.
• The results of the sensitivity analysis showed that the effect of curvature diminishes when
the thickness of the wall increases. This means that the flexural strength of a 20 cm thick
curved wall is close to that of the 20 cm rectilinear wall.
• A two-way layout simulation series was designed to investigate and determine the effect
of range of wall thickness and wall curvature that enhances strength. The design included
30 individual simulations. The maximum deflection of the walls was compared with the
maximum of that of the rectilinear wall of the same dimensions. The difference between
the deflections were plotted into a 3d graph and a 2d contour. The graphs showed when
the thickness goes above 15 cm, the curvature becomes ineffective, and the deflection of
the wall would not make any difference. The results of this section could be used as a
guideline for future builders using automated construction approaches in order to choose
the best geometry for the structures they are designing.
• Creating corrugated walls was successfully implemented before with Contour Crafting
technology. The behavior of corrugated rectilinear walls was simulated and compared
with that of solid rectilinear walls of the same dimension.
• Parallel hatched corrugated and zigzag corrugated walls were simulated and compared
with a rectilinear 20 cm thick wall. It was shown that the solid structures exhibited a
significantly lower deflection than the hollow structures. The zigzag corrugated structure
also had better performance than the hatched corrugated wall.
120

• The 8 m curvature that was selected as the most effective radius among the previous
simulations was implemented in the zigzag corrugated wall.
• It was shown that the performance of the corrugated curved wall was better than the
rectilinear wall of the same dimensions. Such a wall brings significant benefits to
structures, such as being lightweight and energy and sound insulation, while also having
more strength. This type of the geometry was selected as the best performer of all the
conducted simulations.
In conclusion, the proposed research was successful in establishing a methodology of
quantifying the bonding strength of concrete layers, creating a simulation model in predicting the
behavior of layered structures, and revealing the positive effect of implementing non-
conventional geometries such as curvature and corrugation on future structures that would be
fabricated using automated construction approaches. Such structures would be lightweight, more
durable, and more thermal energy and sound insulating.
5.2 Future Work
The following suggestions are made for the direction of future works in this field:
• While the research was successful in simulating a wall section, the simulation can be utilized
in simulating specific types of structures instead of wall sections.
• This work established a methodology in testing the bonding strength of concrete to concrete
as well as the possibility of reinforcing the interface. Many other reinforcing mechanisms and
their effect can be studied further such as adding adhesive or fiber.
• This work studied two types of innovative shapes and found a significant merit in using such
patterns. This study can be extended in identifying the more geometry factor, optimizing
those factors, and coming up with more innovative shapes that can change the way the
current structures are built.
121

6 References
[1] B. Khoshnevis, “Contour Crafting - State of Development,” in Solid Freeform Fabrication
Proceedings, 1999, pp. 743–750.
[2] B. Khoshnevis, “Automated construction by contour crafting—related robotics and information
technologies,” Autom. Constr., vol. 13, no. 1, pp. 5–19, 2004.
[3] T. Di Carlo, “Experimental and Numerical Techniques To Characterize Structural Properties of
Fresh Concrete Relevant To Contour Crafting,” 2012.
[4] J. Zhang, “CONTOUR CRAFTING PROCESS PLANNING AND OPTIMIZATION,” 2009.
[5] J. Zhang and B. Khoshnevis, “Contour Crafting Process Plan Optimization Part I : Single-Nozzle
Case,” J. Ind. Syst. Eng., vol. 4, no. 1, pp. 33–46, 2010.
[6] J. Zhang and B. Khoshnevis, “Optimal machine operation planning for construction by Contour
Crafting,” Autom. Constr., vol. 29, pp. 50–67, 2013.
[7] K. Kovler and N. Roussel, “Properties of fresh and hardened concrete,” Cem. Concr. Res., vol. 41,
no. 7, pp. 775–792, 2011.
[8] P. F. G. Banfill, “Rheology of Fresh Cement and Concrete,” Rheol. Rev., vol. 2006, pp. 61–130,
2006.
[9] G. . Tattersall, Workability and quality control of concrete. 1991.
[10] R. Shaughnessy and P. E. Clark, “The rheological behavior of fresh cement pastes,” Cem. Concr.
Res., vol. 18, no. 3, pp. 327–341, 1988.
[11] M. Nehdi and M. a. Rahman, “Estimating rheological properties of cement pastes using various
rheological models for different test geometry, gap and surface friction,” Cem. Concr. Res., vol.
34, no. 11, pp. 1993–2007, 2004.
122

[12] C. Hu, F. de Larrard, T. Sedran, C. Boulay, F. Bosc, and F. Deflorenne, “Validation of
BTRHEOM, the new rheometer for soft-to-fluid concrete,” Mater. Struct., vol. 29, no. 10, pp.
620–631, 1996.
[13] N. Roussel, C. Stefani, and R. Leroy, “From mini-cone test to Abrams cone test: Measurement of
cement-based materials yield stress using slump tests,” Cem. Concr. Res., vol. 35, no. 5, pp. 817–
822, 2005.
[14] Y. Otsubo, S. Miyai, and K. Umeya, “Time-dependent flow of cement pastes,” Cem. Concr. Res.,
vol. 10, no. 5, pp. 631–638, Sep. 1980.
[15] L. Nachbaur, J. C. Mutin, a. Nonat, and L. Choplin, “Dynamic mode rheology of cement and
tricalcium silicate pastes from mixing to setting,” Cem. Concr. Res., vol. 31, no. 2, pp. 183–192,
2001.
[16] C. A. Clear and D. G. Bonner, “Settlement of fresh concrete—an effective stress model,” Mag.
Concr. Res., vol. 40, no. 142, pp. 3–12, Mar. 1988.
[17] K. H. Khayat, “Viscosity-enhancing admixtures for cement-based materials — An overview,”
Cem. Concr. Compos., vol. 20, no. 2–3, pp. 171–188, Jan. 1998.
[18] N. Roussel, “A Theoretical Frame to Study Stability of Fresh Concrete,” Mater. Struct., vol. 39,
no. 1, pp. 81–91, Aug. 2007.
[19] S. E. Chidiac, O. Maadani, A. G. Razaqpur, and N. P. Mailvaganam, “Correlation of Rheological
Properties to Durability and Strength of Hardened Concrete,” J. Mater. Civ. Eng., vol. 15, no. 4,
pp. 391–399, Aug. 2003.
[20] D. P. Bentz, “A review of early-age properties of cement-based materials,” Cem. Concr. Res., vol.
38, no. 2, pp. 196–204, Feb. 2008.
[21] K. Kovler and A. Bentur, “Cracking sensitivity of normal- and high-strength concretes,” ACI
123

Mater. J., vol. 106, no. 6, pp. p537–542, 2009.
[22] C. Aquino, M. Inoue, H. Miura, M. Mizuta, and T. Okamoto, “The effects of limestone aggregate
on concrete properties,” Constr. Build. Mater., vol. 24, no. 12, pp. 2363–2368, Dec. 2010.
[23] A. Kılıç, C. D. Atiş, A. Teymen, O. Karahan, F. Özcan, C. Bilim, and M. Özdemir, “The influence
of aggregate type on the strength and abrasion resistance of high strength concrete,” Cem. Concr.
Compos., vol. 30, no. 4, pp. 290–296, Apr. 2008.
[24] İ. B. Topçu and M. Sarıdemir, “Prediction of properties of waste AAC aggregate concrete using
artificial neural network,” Comput. Mater. Sci., vol. 41, no. 1, pp. 117–125, Nov. 2007.
[25] H. Mazaheripour, S. Ghanbarpour, S. H. Mirmoradi, and I. Hosseinpour, “The effect of
polypropylene fibers on the properties of fresh and hardened lightweight self-compacting
concrete,” Constr. Build. Mater., vol. 25, no. 1, pp. 351–358, Jan. 2011.
[26] A. Sivakumar and M. Santhanam, “Mechanical properties of high strength concrete reinforced
with metallic and non-metallic fibres,” Cem. Concr. Compos., vol. 29, no. 8, pp. 603–608, Sep.
2007.
[27] A. L. Gamino, J. Sousa, O. L. Manzoli, and T. Bittencourt, “A comparative study between
smeared and embedded crack models for finite element analysis of reinforced concrete beams,”
Fract. Mech. Concr. Concr. Struct. Adv. Fract. Mech. Concr. Seoul, Korea, 2010.
[28] T. A. Söylev, C. McNally, and M. G. Richardson, “The effect of a new generation surface-applied
organic inhibitor on concrete properties,” Cem. Concr. Compos., vol. 29, no. 5, pp. 357–364, May
2007.
[29] M. Angelillo, Mechanics of Masonry Structures, vol. 551. 2014.
[30] S. Nazir and M. Dhanasekar, “A non-linear interface element model for thin layer high adhesive
mortared masonry,” Comput. Struct., vol. 144, pp. 23–39, Nov. 2014.
124

[31] Z. Yan, “Nonlinear dynamic response of unreinforced masonry under in-plane lateral loads,”
Queensland University of Technology, 1995.
[32] P. B. Shing, M. Schuller, and V. S. Hoskere, “In-Plane Resistance of Reinforced Masonry Shear
Walls,” J. Struct. Eng., vol. 116, no. 3, pp. 619–640, Mar. 1990.
[33] S. G. Fattal, The effect of critical parameters on the behavior of partially-grouted masonry shear
walls under lateral loads. Engineering Research Database, 1993.
[34] J. B. Mander, M. J. N. Priestley, and R. Park, “Observed stress-strain behavior of confined
concrete,” J. Struct. Eng., vol. 114, no. 8, pp. 1827–1849, 1988.
[35] G. Benli and Z. Houqin, “The seismic resistance of behaviour of partially reinforced brick walls,”
in 9th International Brick/Block Masonry Conference.
[36] N. F. Naom, “Prediction of masonry ultimate compressive strength,” MPhil Thesis. University of
Essex, 1992.
[37] W. A. Thanoon, A. H. Alwathaf, J. Noorzaei, M. S. Jaafar, and M. R. Abdulkadir, “Nonlinear
finite element analysis of grouted and ungrouted hollow interlocking mortarless block masonry
system,” Eng. Struct., vol. 30, no. 6, pp. 1560–1572, 2008.
[38] A. D. Tzamtzis and P. G. Asteris, “Finite element analysis of masonry structures: Part I: Review of
previous work,” in 9th North American masonry conference, 2003, pp. 101–111.
[39] S. V.Chaudhari and M. A. Chakrabarti, “Modeling of Concrete for Nonlinear Analysis using
Finite Element Code ABAQUS,” Int. J. Comput. Appl., vol. 44, no. 7, pp. 14–18, Apr. 2012.
[40] A. D. Tzamtzis and P. G. Asteris, “Finite Element Analysis of Masonry Structures Part II--
Proposed 3-D Nonlinear Microscopic Model,” in 9th North American Masonry Conference, 2003.
[41] J. Lee and G. L. Fenves, “Plastic-damage model for cyclic loading of concrete structures,” J. Eng.
125

Mech., vol. 124, no. 8, pp. 892–900, 1998.
[42] T. Janaraj and M. Dhanasekar, “Finite element analysis of the in-plane shear behaviour of masonry
panels confined with reinforced grouted cores,” Constr. Build. Mater., vol. 65, pp. 495–506, 2014.
[43] P. B. Lourenço and J. G. Rots, “Multisurface Interface Model for Analysis of Masonry
Structures,” J. Eng. Mech., vol. 123, no. 7, pp. 660–668, Jul. 1997.
[44] E. N. B. . Júlio, F. A. . Branco, and V. D. Silva, “Concrete-to-concrete bond strength. Influence of
the roughness of the substrate surface,” Constr. Build. Mater., vol. 18, no. 9, pp. 675–681, Nov.
2004.
[45] Census.gov, “Median and Average Square Feet of Floor Area in New Single-Family Houses
Completed by Location.”
[46] S. G. Fattal, The Effect of Critical Parameters on the Behavior of Partially-grouted Masonary
Shear Walls Under Lateral Loads. US Department of Commerce, National Institute of Standards
and Technology, 1993.
[47] K. , C. G. Dávila, and C. A. Rose, “Guidelines and parameter selection for the simulation of
progressive delamination,” 2008.
[48] G. Giuliese, A. Pirondi, and F. Moroni, “A cohesive zone model for three-dimensional fatigue
debonding/delamination,” Procedia Mater. Sci., vol. 3, pp. 1473–1478, 2014.
[49] M. Alfano, F. Furgiuele, A. Leonardi, C. Maletta, and G. H. Paulino, “Cohesive zone modeling of
mode I fracture in adhesive bonded joints,” in Key Engineering Materials, 2007, vol. 348, pp. 13–
16.
[50] T. Carlberger and U. Stigh, “Influence of Layer Thickness on Cohesive Properties of an Epoxy-
Based Adhesive—An Experimental Study,” J. Adhes., vol. 86, no. April 2015, pp. 816–835, 2010.
126

[51] R. D. S. G. Campilho, D. C. Moura, D. J. S. Gonçalves, J. F. M. G. da Silva, M. D. Banea, and L.
F. M. da Silva, “Fracture toughness determination of adhesive and co-cured joints in natural fibre
composites,” Compos. Part B Eng., vol. 50, pp. 120–126, 2013.
[52] ASTM, “D5528-01 2001. Standard Test Method for Mode I Interlaminar Fracture Toughness of
Unidirectional Fiber-Reinforced Polymer Matrix Composites,” Am. Soc. Test. Mater., pp. 1–13,
2014.
[53] F. Roudolff and Y. Ousset, “Comparison between two approaches for the simulation of
delamination growth in a DCB specimen,” Aerosp. Sci. Technol., vol. 6, no. 2, pp. 123–130, 2002.
[54] N. Alver, M. Efe, and O. Burak, “The effect of short cantilever beam formation on the structural
behavior of precast post-tensioned connections,” Constr. Build. Mater., vol. 35, pp. 232–239,
2012.
[55] Z. Mikulik, B. G. Prusty, R. S. Thomson, and D. W. Kelly, “Application of fracture mechanics-
based methodologies for failure predictions in composite structures,” Conf. Damage Compos.
Mater. Stuttgart, 2006.
[56] J. A. B. P. Neto, R. D. S. G. Campilho, and L. F. M. Da Silva, “Parametric study of adhesive joints
with composites,” Int. J. Adhes. Adhes., vol. 37, no. September, pp. 96–101, 2012.
[57] G. Ji, Z. Ouyang, G. Li, S. Ibekwe, and S. S. Pang, “Effects of adhesive thickness on global and
local mode-I interfacial fracture of bonded joints,” Int. J. Solids Struct., vol. 47, no. 18–19, pp.
2445–2458, 2010.
[58] R. D. S. G. Campilho, D. C. Moura, M. D. Banea, and L. F. M. da Silva, “Adherend thickness
effect on the tensile fracture toughness of a structural adhesive using an optical data acquisition
method,” Int. J. Adhes. Adhes., vol. 53, pp. 1–8, 2014.
[59] N. Khalili, Ceramic House & Earth Architecture. California: Cal-Earth Press, 2000.
127

[60] W. Xu, L. H. Han, and Z. Tao, “Flexural behaviour of curved concrete filled steel tubular trusses,”
J. Constr. Steel Res., vol. 93, pp. 119–134, 2014.
[61] N. A. Brake, H. Allahdadi, and F. Adam, “Flexural strength and fracture size effects of pervious
concrete,” Constr. Build. Mater., vol. 113, pp. 536–543, 2016.

128

7 Appendices
7.1 Appendix I: Curve fitting for the compliance calibration methods
PS01:

Polynomial Regression Analysis: C versus a

The regression equation is
C = 17.54 - 8.058 a + 1.325 a^2 - 0.06251 a^3

S = 2.14255 R-Sq = 86.6% R-Sq(adj) = 85.6%

Analysis of Variance

Source DF SS MS F P
Regression 3 1158.08 386.026 84.09 0.000
Error 39 179.03 4.591
Total 42 1337.11

Sequential Analysis of Variance

Source DF SS F P
Linear 1 711.788 46.67 0.000
Quadratic 1 276.899 31.79 0.000
Cubic 1 169.391 36.90 0.000

Fitted Line: C versus a

Figure 1. Fitted line of C versus a

1 4 1 2 1 0 8 6 4 2 0
20
1 5
1 0
5
0
S 2.1 4255
R-S q 86.6%
R-S q(adj) 85.6%
a
C
Fitted L ine Plot
C = 17.54 - 8.058 a
+ 1.325 a^2 - 0.06251 a^3
129

PP02

Welcome to Minitab, press F1 for help.
Executing from file: C:\Program Files (x86)\Minitab\Minitab
17\English\Macros\Startup.mac

This Software was purchased for academic use only.
Commercial use of the Software is prohibited.

Polynomial Regression Analysis: Sigma versus Crack growth

The regression equation is
Sigma = 6.068 + 0.3769 Crack growth - 0.04895 Crack growth^2 + 0.001389
Crack growth^3

S = 1.02843 R-Sq = 99.4% R-Sq(adj) = 99.4%

Analysis of Variance

Source DF SS MS F P
Regression 3 11391.2 3797.08 3590.04 0.000
Error 63 66.6 1.06
Total 66 11457.9

Sequential Analysis of Variance

Source DF SS F P
Linear 1 7338.78 115.81 0.000
Quadratic 1 3572.43 418.25 0.000
Cubic 1 480.02 453.84 0.000

Fitted Line: Sigma versus C

Figure 2. Fitted line of C versus a

50 40 30 20 1 0 0
60
50
40
30
20
1 0
0
S 1 .02843
R-S q 99.4%
R-S q(adj) 99.4%
a
C
Fitted L ine Plot
C = 6.068 + 0.3769 a
- 0.04895 a^2 + 0.001389 a^3
130

7.2 Appendix II: Figures of the simulations related to curvature analysis.

Run 1:

Run 14:

Run 2:

Run 15:

Run3:

Run 16:

Run 4

Run 17:

131

Run 5

Run 18

Run 6

Run 19:

Run 7:

Run 20:

Run 8:

Run 21:

132

Run 9:

Run 22:

Run 10

Run 23:

Run 11

Run 24:

Run 12

Run 25:

133

Run 13:

Abstract (if available)

Abstract The construction industry had been deprived from automation until the emergence of additive manufacturing approaches such as Contour Crafting (CC)) opened up new opportunities. CC is an automatic fabrication technology on the basis of additive manufacturing principles. The process stacks relatively thick layers of cementitious material (e.g., 25.4 mm) on top of each other until the full scale structure is achieved. There are several problems in conventional construction approaches, including inefficient use of labor, time, materials, and capital. All of these efficiency problems can be addressed by CC. Besides the construction of building structures, CC is capable of performing sub tasks such as electrical, plumping, wiring, and painting. ❧ The operational behavior of layered structures under vertical and in-plane loads are not known to date. To expand the application of the structures fabricated by automated approaches, the strength and failure mechanisms of these structures must be completely understood. The adhesion strength between the layers is one of the structural concerns in CC that should be studied and enhanced if needed to achieve the desired structural properties. In this research, a methodology to analyze the structural properties of specimens made by CC was developed. The methodology includes both experimental and numerical analyses that objectively study the bonding strength of concrete layers to each other and predict the behavior of Contour Crafting structures. The interface of layers is simulated based on the Cohesive Zone Modeling (CZM) approach. The simulation is followed and calibrated by experimental procedures to quantify the bonding strength of the layers. A final simulation model to investigate the strength of the layers is developed based on experimental results. The study also includes the effect of incorporating geometrical complexity on the behavior of CC structures. The model investigates and compares curved, rectilinear, hollow, and curved hollow structures. Results and findings of this study show the effectiveness of incorporating curvature as compared with rectilinear walls. The results also show the significance of corrugated structures.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Contour Crafting process planning and optimization

PDF

Contour crafting construction with sulfur concrete

PDF

Experimental and numerical techniques to characterize structural properties of fresh concrete

PDF

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

PDF

The extension of selective inhibition sintering (SIS) to high temperature alloys

PDF

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

PDF

Enhancing the surface quality and dimensional accuracy of SIS-metal parts with application to high temperature alloys

PDF

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

PDF

Multi-scale biomimetic structure fabrication based on immersed surface accumulation

PDF

Deformation control for mask image projection based stereolithography process

PDF

Scalable polymerization additive manufacturing: principle and optimization

PDF

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

PDF

Statistical modeling and machine learning for shape accuracy control in additive manufacturing

PDF

Numerical study of shock-wave/turbulent boundary layer interactions on flexible and rigid panels with wall-modeled large-eddy simulations

PDF

Fabrication-aware machine learning for accuracy control in additive manufacturing

PDF

Hybrid vat photopolymerization processes for viscous photocurable and non-photocurable materials

PDF

Machine learning-driven deformation prediction and compensation for additive manufacturing

PDF

The effect of lattice structure and porosity on thermal conductivity of additively-manufactured porous materials

PDF

Next generation seismic resistant structural elements using high-performance materials

Asset Metadata

Creator Mansouri, Amir (author), Mansourighasri, Amirhossein (author)

Core Title Analyses of strength of layered structures fabricated by Contour Crafting

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Industrial and Systems Engineering

Publication Date 08/04/2016

Defense Date 08/01/2016

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

Tag additive manufacturing,cementitious materials,Concrete,corrugated walls,curved walls,damage mechanics,finite element model,fracture toughness,OAI-PMH Harvest

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Khoshnevis, Behrokh (committee chair)

Creator Email a.mansouri.q@gmail.com,amirhosm@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-296250

Unique identifier UC11281100

Identifier etd-Mansourigh-4736.pdf (filename),usctheses-c40-296250 (legacy record id)

Legacy Identifier etd-Mansourigh-4736.pdf

Dmrecord 296250

Document Type Dissertation

Format application/pdf (imt)

Rights Mansourighasri, Amirhossein; Mansouri, Amir

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