University of Southern California Dissertations and Theses

Selective Separation Shaping (SSS): large scale cementitious fabrication potentials

(USC Thesis Other)

Selective Separation Shaping (SSS): large scale cementitious fabrication potentials

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Selective Separation Shaping (SSS) – Large Scale
Cementitious Fabrication Potentials

By
Xiang Gao
____________________________________________________________________________

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)
May 2019

ii

Acknowledgements
I would like to thank my advisor, Professor Behrokh Khoshnevis, Dean’s Professor of Industrial
& Systems Engineering and a Fellow of the National Academy of Inventors, for providing me
with the opportunity to work on great projects and to study at one of the world’s top research
institutions, the University of Southern California. His innovative thinking in solving complex
problems and detail-oriented attitude gave me a path to follow as I developed my own problem-
solving skills and tackled new challenges.
My most sincere thanks are also owed to my committee members, Professor Yong Chen,
Professor Qiang Huang, Professor Geoffrey Shiflett, and Professor Qiming Wang, who gave me
invaluable suggestions during my research and dissertation. Their comments were essential to
take my work to a higher level.
I would also like to offer thanks to all the colleagues who have worked with me and helped me at
USC. I would especially like to thank Dr. Ali Kazemian, Dr. Hadis Nouri, Dr. Xiao Yuan, and
Dr. Jing Zhang for their support of my work. I would also like to thank Brittany Barbara, my
master’s assistant, and Yunpeng Zhang, the lab manager of M.C. Gill Composites Center, who
helped me with strength testing. Other lab members I enjoyed having the opportunity to work
with include Dr. Dongping Deng, Dr. Huachao Mao, and Dr. Xuan Song. Finally, I would like to
express my appreciation to my parents, my grandparents, my wife, and all my family members.
It would have been impossible for me to finish this without their constant support.

iii

Abstract
Additive manufacturing (AM) or 3D printing has been applied for rapid prototyping and direct
manufacturing for years. AM has shown it capability in fabricating complex shapes without extra
tooling and fixture [1, 2]. However, application of AM in fabrication of large scale products has
been a challenge due to low fabrication speed. This work introduces a new high speed fabrication
process which is especially suitable for fabrication of large scale objects.
Selective Separating Shaping is a new additive manufacturing technique which is capable of
processing polymeric, metallic, ceramic and composite materials such as cementitious
construction materials. In earlier research the capabilities of SSS in making metallic and ceramic
parts have been demonstrated. The focus of the research reported in this dissertation is on
expansion of capabilities of SSS for efficient creation of large scale cementitious composite
parts.
A prototype machine has been used to create specimens made of regular Portland cement (lime
based), Sorel cement (magnesia based) and gypsum based composite, as shown in Chapter 6.
The fabrication result and process characteristics based on several experiments are presented.
The factors that impact the surface quality of samples are also discussed in Chapter 6. The
interaction between the two powder materials (i.e., Separation powder and Base material) and the
separation powder deposition process are simulated and optimized to ensure an adequate powder
flow. The mechanical behavior of cementitious samples built by SSS and the factors affecting it

iv

are also presented in Chapter 7. In Chapter 8 statistical predictive models are also developed and
discussed to provide information about the samples curing characteristics under different
temperatures. Finally, the potential of the current SSS on the fabrication of large scale objects
and complicated three dimensional parts are presented and discussed in Chapter 9.

v

Contents
Acknowledgements ..................................................................................................................... ii
Abstract ...................................................................................................................................... iii
Contents....................................................................................................................................... v
List of Tables .............................................................................................................................. ix
List of Figures ............................................................................................................................. x
1. Introduction ................................................................................................................................. 1
2. Literature Review ....................................................................................................................... 4
2.1. Major Additive Manufacturing technologies with potential for large scale fabrication ...... 4
2.1.1. Large Scale Fused Deposition Technologies ................................................................ 4
2.1.2. Large Scale Binder Jetting Technologies ...................................................................... 5
2.1.3. Big Area Additive Manufacturing ................................................................................. 7
2.2. Major Additive Manufacturing technologies for cementitious materials ............................ 8
2.2.1. Contour Crafting ............................................................................................................ 8
2.2.2. D-Shape ......................................................................................................................... 9

vi

2.2.3. Smart Dynamic Casting ............................................................................................... 10
2.2.4. Stone Spray .................................................................................................................. 11
2.3. Critique of the Past Approaches ......................................................................................... 13
2.4. SSS for Large-Scale Fabrication ........................................................................................ 14
3. Explanation of the SSS Fabrication Process and Hypothesis ................................................... 18
3.1. Mechanical System and Slicing Strategy ........................................................................... 18
3.2. SSS Process Steps and Preliminary Results on Cement Material ...................................... 19
3.3. Research Challenges and Hypotheses ................................................................................ 23
4. Preliminary Results ................................................................................................................... 26
4.1. Portland Cement ................................................................................................................. 26
4.2. Gypsum .............................................................................................................................. 29
4.3. Sorel Cement ...................................................................................................................... 30
4.4. Magnesium-based material ................................................................................................ 31
5. Surface Quality Analysis .......................................................................................................... 34
5.1 Surface’s Quality Interacted with Separation Powder ........................................................ 34
5.1.1 Impact of Separation Powder Choice ........................................................................... 34

vii

5.1.2 Separation Powder Deposition ..................................................................................... 36
5.1.3 Deposition Simulation .................................................................................................. 41
5.2 Quality of Top and Bottom Surfaces .................................................................................. 48
5.2.1 Material and Roughness Measurement Process ........................................................... 49
5.2.2 Bottom Surface Quality ................................................................................................ 51
5.2.3 Top Surface Quality ..................................................................................................... 58
5.3 Demonstration Samples with Optimized Process ............................................................... 64
6. Analysis of Mechanical Properties of Parts Made by SSS ....................................................... 66
6.1. Specimens .......................................................................................................................... 66
6.2 Test Procedure ..................................................................................................................... 68
6.3 Analysis of Results .............................................................................................................. 69
6.4 Strength Analysis of a Honeycomb Structure ..................................................................... 71
7. Statistical Predictive Modeling of Strength of SSS-Produced Parts ......................................... 75
7.1 Experimental Material and Process ..................................................................................... 76
7.2 Experiment Results ............................................................................................................. 79
7.3 Statistical Predictive Model Development .......................................................................... 81

viii

7.3.1 Models under Different Temperatures ......................................................................... 81
7.3.2 Predictive model including the temperature ................................................................. 85
7.4 Determination of Best Separation Time .............................................................................. 87
8. Potential of SSS on Large-scale and Complicated Shape Fabrication ..................................... 89
8.1 Scale Up SSS for Large-Scale Fabrication ......................................................................... 89
8.1.1 Printing Speed of SSS under Different Layer Thicknesses .......................................... 90
8.1.2 Comparison of SSS and Commercialized Mega-scale 3D-printing Technologies ...... 94
8.2 Fabrication with Multi-layer Thickness .............................................................................. 96
9. Conclusion and Future Research ............................................................................................ 100
9.1. Research Summary ........................................................................................................... 100
9.2. Recommendations for Future Work ................................................................................. 101
9.2.1 Fabrication with more materials ................................................................................. 101
9.2.2. Large scale device development ................................................................................ 102
9.2.3. Mutli-degree of Freedom Movement Device Application ........................................ 102
Bibliography ............................................................................................................................... 104

ix

List of Tables
Table 1. Main building parameters and results comparison on three different materials ............. 33
Table 2. The 23 full factorial design table .................................................................................... 39
Table 3. The bottom surface roughness of samples fabricated under dry/wet sand layers with
different water speeds ................................................................................................................... 56
Table 4. Parameter settings for strength measurement. ................................................................ 69
Table 5. Mean flexural strength index (the blue bar represents the strength of different samples;
the orange bar represents the strength of sample 2). ..................................................................... 70
Table 6.The flexural strength of samples maintained in 20 ℃ water. .......................................... 80
Table 7. The flexural strength of samples maintained in 50 ℃ water. ......................................... 80
Table 8. The fitted models for the linear relationship between flexural strength and temperature
and age (t indicates time, and T indicates temperature). ............................................................... 86
Table 9. The maximum printing speed of nozzles with different opening sizes. ......................... 93
Table 10. The fabrication parameters of mega-scale 3D printing technologies. .......................... 95

x

List of Figures
Figure 1. Transformative impact of Contour Crafting (Source: EL Studio, Amsterdam) .............. 1
Figure 2. Left: the Fortus 900 machine; Right: a multi-segmented part made with the machine
[10] .................................................................................................................................................. 5
Figure 3. Left: The S-Max machine; Right: Sample parts made by the machine [11] ................... 6
Figure 4. Left: The VXC 800 machine; Right: produced parts [12] ............................................... 6
Figure 5. Left: The BAAM setup; Right: A part made by BAAM [13] [14] ................................. 7
Figure 6. Left: A concrete wall built by Contour Crafting; Right: Building construction by
Contour Crafting [15] ..................................................................................................................... 8
Figure 7. Left: the prototype machine of D-Shape; Right: the product built by magnesium sand
[16] ................................................................................................................................................ 10
Figure 8. Left: the process of dynamic casting; Right: products of dynamic casting [17] ........... 11
Figure 9. Visualization of a sculpture on the beach by the designer [17] ..................................... 12
Figure 10. Left picture shows the building process, when a head is injecting sand and binder
from two pipes; Right picture shows a finished product. ............................................................. 12
Figure 11. Mental samples built by SSS ....................................................................................... 16
Figure 12. Deposition of thin wall of separating powder on a layer with 12mm thickness ......... 17
Figure 13. SSS beta machine CAD model .................................................................................... 18

xi

Figure 14. Shape building algorithm ............................................................................................ 20
Figure 15. Left: CAD model of the part; Middle: printing process of SSS; Right: separated part22
Figure 16. Selective Separation Shaping prototype ...................................................................... 26
Figure 17. An early stage Portland cement part made by SSS with smooth side surfaces but poor
edges and rough top surface .......................................................................................................... 28
Figure 18. Gypsum parts made by SSS ........................................................................................ 29
Figure 19. Sorel cement parts made by SSS ................................................................................. 30
Figure 20. Magnesium-based part made by SSS .......................................................................... 32
Figure 21. Soda lime’s micro and macro image ........................................................................... 35
Figure 22. A 24mm tall Portland cement part built with SSS ...................................................... 36
Figure 23. Left: Cross section of layer without sufficient soda lime as S powder; Right: Cross
section of layer with sufficient S powder ..................................................................................... 37
Figure 24. Plot of arch generation inside the syringe [13] ............................................................ 37
Figure 25. Half norm plot for dispersion of experiment results ................................................... 40
Figure 26. Half norm plot for variance of experiment results ...................................................... 40
Figure 27. Left: actual deposition device; Right: model built in Maya ........................................ 43
Figure 28. Left: SolidWorks model on deposition device; Right: simulation result of syringe’s
displacement under vibration ........................................................................................................ 44

xii

Figure 29. Left: displacement of syringe for first natural frequency; Right: displacement of
syringe for third natural frequency ............................................................................................... 45
Figure 30. Left: an indication of angle of repose, where 𝜶𝑹 is angle of repose [34]. Right: an
actual pile of Soda Lime ............................................................................................................... 46
Figure 31. Comparison of angle of repose between Nucleus model and real Soda Lime pile ..... 47
Figure 32. Left: dynamic simulation result of powder deposition in syringe with no vibration;
Right: powder deposition under vibration of 1000 Hz ................................................................. 47
Figure 33. The dial gauge applied to measure the roughness of samples ..................................... 50
Figure 34. Different parameters used for the measurement of roughness [35] ............................ 50
Figure 35. One layer of sand (2mm) after the water is added....................................................... 53
Figure 36. Surface qualities of sand layers at different thicknesses ............................................. 54
Figure 37. Water dispersion on the bottom surface of cement ..................................................... 55
Figure 38. These images show the bottom surface of samples fabricated with dry/wet sand layers
with different water speeds. .......................................................................................................... 57
Figure 39. A spray system with four nozzles to spread water ...................................................... 59
Figure 40.The testing system for top surface quality analysis and the image of top surface of a
sample fabricated with water sprayed directly on the cement ...................................................... 60
Figure 41. These images show the top surface of samples before water was added .................... 61
Figure 42. These images show the top surfaces of samples after water was added ..................... 62

xiii

Figure 43. Images of the top surface of samples fabricated under different water spraying speeds
....................................................................................................................................................... 63
Figure 44. Demonstration of samples with improved surface qualities ........................................ 64
Figure 45. Measurement of Arithmetical mean deviation of sides of samples ............................. 65
Figure 46. Loading on flexural specimens. ................................................................................... 68
Figure 47. Left: flexural strength measuring device. Right: a broken part after measurement. ... 69
Figure 48. (a) The honeycomb structure’s compression displacement simulation result. (b) The
block structure’s compression displacement simulation result. .................................................... 73
Figure 49. A 23.50mm high honeycomb structure developed by SSS using type II/IV Portland
cement. .......................................................................................................................................... 73
Figure 50. Samples ready for the flexural strength test. ............................................................... 76
Figure 51. Samples inside a maintenance box ............................................................................. 77
Figure 52. Maintenance tank with temperature control capability. .............................................. 77
Figure 53. The load frame without a sample is shown on the left, and a sample during the
flexural strength test process is on the right. ................................................................................. 78
Figure 54. Load change during the flexural strength test. ............................................................ 79
Figure 55. The figure on the left shows the regression line using logarithm time under 20 ℃, and
the figure on the right shows the regression line for the time under 20 ℃................................... 83

xiv

Figure 56. The graph on the left shows the regression line using logarithm time under 50 ℃, and
the graph on the right shows the regression line for the time under 50 ℃. .................................. 84
Figure 57. The upper printing head has a slot of 25.5mm high, and the lower printing head has a
slot of 12.5mm high. ..................................................................................................................... 90
Figure 58. The gap formed by the nozzle with a 12.5mm slot a different moving speeds. .......... 92
Figure 59. The gaps formed by the nozzle with a 25.5mm slot under different moving speeds. . 92
Figure 60. The image of a sample built with the layer thickness of 22mm. ................................. 93
Figure 61. Slicing strategy under a 12mm fixed layer thickness. ................................................. 97
Figure 62. The sample built with different layer thicknesses. ...................................................... 98
Figure 63. The left image shows a printing head on the end of a 6 axes movement device; the
right image shows a printing head installed on the end of a KUKA robotic manipulator. [60] . 103

1

1. Introduction
Extensive research has been carried out to utilize AM technologies to produce functional parts
using a wide range of materials [3, 4]. Building large scale structures still remains an issue due to
the big barrier of fabrication speed and applicability of available approaches. The first large –
scale AM method, called Contour Crafting, has been invented by Dr. Behrokh Khoshnevis at
USC over two decades ago and application feasibility of the technology for terrestrial and
planetary construction have been demonstrated [5-8]. Contour Crafting is an extrusion-based
approach which as Figure 1 shows has inspired numerous related activities around the world.
Figure 1. Transformative impact of Contour Crafting (Source: EL Studio, Amsterdam)
Invented by Dr. Khoshnevis, SSS (Selective Separation Shaping) is another pioneering large-
scale AM method as it is the first powder -based (as opposed to being extrusion-based) process
which enables rapid fabrication of 3D geometries at a very low cost and high speed. The

2

capabilities of SSS have been demonstrated in previous efforts in printing metallic and high-
performance ceramic parts [9]. This research intends to study the capabilities of SSS in
constructing cementitious composite parts. Applications of SSS can expand to the rapid
fabrication of public and urban structures, large-scale prototyping, and pavement engineering.
The latter has applications in road and platform construction using 3D printed interlocking tiles.
In 2016 the SSS technology won a NASA international competition Grand Prize for planetary
autonomous construction of landing pads and roads using in-situ materials [10].
There is a critical need for a low-cost additive manufacturing process such as SSS which can
utilize available local material such as crushed rocks and sand to accelerate the Additive
Manufacturing of free form large objects at a high speed for a variety of applications.
In this dissertation, the background of Selective Separation Shaping is first presented where the
design of the device, the slicing strategy, and the fabrication process are discussed in detail. In
the preliminary result section, fabricated samples using Portland cement, Sorel cement, gypsum,
and magnesium-based material are shown. Additionally, surface quality-related topics, which
includes the selection of separation powder (S-powder), S-powder deposition characteristics and
effects of hydration by water spraying are discussed in Chapter 6. In Chapter 7, Portland
cement’s mechanical behaviors are analyzed and advantages of SSS in the fabrication of high
strength-to-weight ratio structures are presented. In Chapter 8, statistical predictive models are
discussed to provide information for the strength change during the curing process of samples.
Models with only time and with time and temperature are developed using linear regression and

3

logistic regression methods. These model help to provide guidance on the time of separation of
the fabricated part from its surrounding material and to improve the efficiency of the whole
fabrication process. In Chapter 9, the application of an adaptive slicing strategy which makes the
fabrication of complex 3D shapes possible is discussed. The capability of SSS to potentially
construct building scale structures is shown as well, where the fabrication speed of the current
SSS is studied and compared with other large scale 3D printing technologies, like Contour
Crafting.

4

2. Literature Review
In this section the main additive manufacturing processes to build large scale parts, and
processes to develop cementitious parts are discussed.
2.1. Major Additive Manufacturing technologies with potential for large scale
fabrication
2.1.1. Large Scale Fused Deposition Technologies
The Fused Deposition Modeling is a widely applied technology in additive manufacturing field.
With solid material melted and deposited through a heated nozzle, the device can spread material
following certain pattern in a platform, where manufactured material becomes solid again. After
layer by layer’s deposition, a part can be fabricated.
Fortus 900 is the biggest FDM printer produced by Stratasys Co., with the capability to
manufacture products within a 36 by 24 by 36 inches space. The advantages of this process are
its good product quality, ability to make thin bodies (as small as 0.007 inches), and relatively fast
building speed. On the other hand this technology cannot build larger structures and has a limited
number choice of materials to use.

5

Figure 2. Left: the Fortus 900 machine; Right: a multi-segmented part made with the
machine [10]
2.1.2. Large Scale Binder Jetting Technologies
Binder jetting is an additive manufacturing method in which the base material power, like starch,
gypsum or plaster, is spread in a thin layer and then a binder material is deposited on selected
areas of each powder layer to consolidate the layer profile. After layer by layer deposition only
the section of the material which is exposed to the binder will be solidified to form the final part.
In the last step the unsolidified powder is removed and the desired part is extracted.
S-Max is a binder jetting based 3D printer produced by ExOne whose related focus is on
applying binder jetting technology on sandcasting foundries. The machine’s building volume is
70.9 by 30.4 by 27.6 inches, which is relatively large.
VXC 800, built by Voxeljet is another large-scale binder jetting based 3D printer and is used for
large scale sand mold manufacturing. Its workspace can reach to 850 by 500 by 1500 mm (33.46
by 19.69 by 59.06 inches).

6

The S-Max machine has a manufacturing speed of 2.12 to 3.3 cubic feet per hour and the
smallest layer thickness is 0.011 inches. Despite the advantages that this machine possesses it has
some disadvantages, like the high price and limitation of material choices. VXC 800 has a
fabrication speed of 35 mm per hour and a layer thickness of 0.3 mm.

Figure 3. Left: The S-Max machine; Right: Sample parts made by the machine [11]

Figure 4. Left: The VXC 800 machine; Right: produced parts [12]

7

2.1.3. Big Area Additive Manufacturing
Developed under the partnership between Lockheed Martin and Oak Ridge National Lab [13],
and improved under the cooperation between Cincinnati and ORNL, Big Area Additive
Manufacturing is considered to have the capability of producing large size products.

Figure 5. Left: The BAAM setup; Right: A part made by BAAM [13] [14]
With high-performance engineered thermoplastic melted in the nozzle and extruded, the device
is believed to be able to build large-scale structures within the robotic arm’s work envelope. In
recent work, a gantry structure based machine is developed with the capability to develop base
for a single person vehicle.
The advantages of BAAM include its big product size, relatively fast speed, and relatively good
part strength qualities. However, it also has problems like rough part surfaces, and limited
material choices.

8

2.2. Major Additive Manufacturing technologies for cementitious materials
2.2.1. Contour Crafting
Contour Crafting (CC) is introduced by Dr. Behrokh Khoshnevis as the first process with
potential to build mega scale constructions. CC is based on extrusion of concrete materials. In
this process, a path plan file is first transferred to the controller of the machine and then the
machine extrudes a semi-fluid phase concrete material through a specifically designed nozzle
following the planned path. The next layer is then deposited on the first layer after preliminary
hardening of the first layer. As shown on the left side of Figure 6 very large concrete structures
may be fabricated by the CC technology.
With relatively fast building speed, structure strength and flexible building capability, this
technology possesses huge potential to be applied to construction of buildings, infrastructures
and planetary structures.

Figure 6. Left: A concrete wall built by Contour Crafting; Right: Building construction by
Contour Crafting [15]

9

2.2.2. D-Shape
The D-Shape technology, invented by Enrico Dini, is aimed at large art sculpture fabrication
with environmental friendly materials. The process is based on binder-jetting and solidification
of magnesium oxide in presence of magnesium hexahydrate. In the first step a 5 to 10 mm layer
of sand combined with magnesium oxide is spread on the build surface. In the second step a
solution including magnesium hexahydrate is deposited on the sand through several single-
orifice printing heads (nozzles) which are installed 20 mm apart in a row. The nozzles have a
nominal resolution of 1 mm. The print head is programmed to move reciprocate to cover the 20
mm distance between the nozzles. A new layer of sand is then spread manually after completion
of printing for the layer. Steps 1 and 2 are repeated until the whole part is developed. After
completion of printing the unsolidified sand is recollected for future use and the consolidated
part is removed.
The advantage of this process is that because of the support structure that the sand volume
provides relatively complex structures with overhang may be possible to fabricate. The process
also uses eco-friendly material, but extrusion based processes like CC can also use similar
material the cost of which would be unjustifiable for building construction.

10

Figure 7. Left: the prototype machine of D-Shape; Right: the product built by magnesium
sand [16]
The main disadvantage of this process is the relatively rough surface quality that it produces.
Also, the speed of the process is very low.
2.2.3. Smart Dynamic Casting
Smart Dynamic Casting, also called Merging Existing Casting, was developed by Ena Lloret
[17]. In this process formable concrete is deposited into a vertical mold segment that can raise
and rotate at the same time. The mold movement pattern and speed are such that the desired
rotational shape can be ensured. Each material addition happens after the previous segment
receives enough hardness to stand on its own when the mold is moved up and away from it.

11

Figure 8. Left: the process of dynamic casting; Right: products of dynamic casting [17]
This process is mainly used for construction of freeform columns. The fastest movement speed
of the robotic arm is set to 2.4 cm/min [17]. Some shortcomings of the process include shape
limitation and low fabrication speed. The main advantage is good surface quality as layer
separation lines are not visible.
2.2.4. Stone Spray
Anna Kulik, Inder Shergill and Petr Novikov [18] from IAAC in Barcelona have developed a
granular adhesive based additive manufacturing technology, called Stone Spray, aiming to build
structures with environmentally friendly materials. Utilizing natural sand and binder, it is
claimed that the technology is eco-friendly and has the potential to build large scale structures,
like the sculpture shown in Figure 9.

12

With a similar principle to Direct Laser Melting
[19], Stone Spray applies two pipes to deposit
sand/dirt and a binder separately to form a solid
structure on the desired surface. With structure
formation following certain pattern every layer,
this technology obtains the capability to build three dimensions structures. So far large scale
structures like those shown in Figure 10 have been produced.

Figure 10. Left picture shows the building process, when a head is injecting sand and binder
from two pipes; Right picture shows a finished product.
Stone Spray is considered to be a promising technology to build large scale structures, since it
could be economical and environmentally friendly. There are some challenges faced by this
technology: first, the surface quality is relatively rough, which will limit its application, second,
Figure 9. Visualization of a sculpture
on the beach by the designer [17]

13

building speed of this technology is slow, which will be a critical for large scale construction,
and third, the strength of the structures produced this way may not be sufficient for serious
applications.
There are also other large scale Cementous material based processes, but they each fall in one of
the above process categories.
2.3. Critique of the Past Approaches
Speed limitation is the main challenge faced by most additive manufacturing processes, which
becomes more critical when large scale fabrication is the goal. The aforementioned technologies
are all additive manufacturing processes which are based on layerwise fabrication. With the
exception of Contour Crafting (which uses trowels at its nozzle orifices) the layer thickness in
these processes is typically small. With limited layer thickness it is hard to significantly increase
manufacturing speed. Furthermore, for powder based processes a lot of time is also is consumed
between depositions of layers, when powder material is added to form a new layer.
When large scale products are considered, parts’ surface quality in most of the above processes
will become a problem, since most large scale manufacturing technology (except Contour
Crafting) are just scale up versions of their small scale implementation and hence they
exaggerate errors that are inherent in their base approach. Accordingly, most of the technologies
mentioned above cannot achieve good surface quality and typically need post processing to
condition the surface. Moreover, fabrication speed and surface quality are opposing factors,

14

hence in the aforementioned processes higher fabrication speeds are typically attained at the
expense of sacrificing surface quality.
Selective Separation Shaping (SSS) which has been pursued in this research has a number of
unique advantages which allow it to build large-scale complex parts with smooth surfaces at a
high speed. The capabilities of SSS in fabrication of small scale metallic and polymeric parts
have been demonstrated by other researchers. Investigation and development of large-scale
implementation of SSS is the subject of the proposed research.
2.4. SSS for Large-Scale Fabrication
In the initial efforts, another AM technology called Selective Inhibition Sintering (SIS) process,
was conceived and developed to fabricate parts based on powder sintering. In SIS a liquid
material, called the inhibitor, is deposited into the perimeter of the part on each layer of the
powder bed. In the subsequent step, the base powder is uniformly spread on top of the build tank
and the deposition process is repeated. Based on the selection of the base material, the sintering
profile varies from bulk sintering to layer by layer heating for metal and polymeric parts,
respectively. Experimental results have shown that the selection of the inhibitor is critical in the
final quality of the fabricated part. The chemical reaction between the inhibitor and base material
needs to be extensively studied before applying in the SIS machine [20]. As an alternative
method to selective separation of part using fluid agent, the new 3D printing technique called
SSS [21] was developed. Compared to SIS the SSS technology is very robust because instead of
relying on the chemical impact of the inhibitor powder on the separation of part from its

15

surrounding powder SSS creates a mechanical separation boundary around each part layer profile
by inserting a powder agent named Separation Power. In SSS the solid separation agent works as
a physical barrier among base powder particles. Separation of the regions of a base power by
insertion of a different powder which may have a higher melting point or a different chemical
reaction than the base powder expands the flexibility of SSS for applicability to a large choice of
materials and various scales of fabrication.
The manufacturing principle of SSS is as following: 1) a thick layer of the base powder (B
powder) is uniformly spread on the build tank, 2) a thin nozzle is moved to the printing location
and is inserted into the border of the part layer to deliver the B powder, 3) the nozzle is vibrated
by a piezo element and the separating agent (S powder) is deposited inside the B powder on the
boundary of the part, 4) the nozzle is raised and moved back to the home position, and the
process continues for the next layer: Finally, the powder volume is removed from the platform
and is placed in a sintering furnace to sinter and is held there until it reaches the desired density
and mechanical strength. Some small scale metallic parts made with SSS are shown in Figure 11.

16

Figure 11. Mental samples built by SSS
In the meso-scale sintering-based SSS process, which has been recently reported in the literature,
a thin wall of high melting point separator powder material (S-powder) is deposited within the
base material powder (B-powder) to form a barrier on the boundary of each layer. This barrier
creates a separation between the part and surrounding material, which allows for the separation
of the part from the surrounding powder after sintering is complete. Finally, the part is removed
from the platform and inserted into sintering furnace. After sintering the part is easily separated
from the surrounding as the S-powder remains in powder state because of its relatively higher
sintering temperature.

17

In the extended SSS process for large-
scale part fabrication the thickness of
each progressive powder layer is
significantly higher than is in the case
of meso-scale
fabrication. Thick layers are achieved
by an S-powder insertion nozzle that
is thin tube with a longitudinal slot
with height equal to the desired layer
thickness. In the preliminary experiments a 1.30mm diameter metallic nozzle tube with a 12 mm
slut is used as shown in Figure 12.

Figure 12. Deposition of thin wall of separating
powder on a layer with 12mm thickness

18

3. Explanation of the SSS Fabrication Process and Hypothesis
Selective Separation Shaping for Large Scale is an expansion for Selective Separation Sintering,
which is a unique additive manufacturing process to fabricate metal products. With the new
Selective Separation Shaping for Large Scale technology, the manufacturing speed is supposed
to be improved significantly, and good surface quality for product is supposed to be obtained.
3.1. Mechanical System and Slicing Strategy
A beta machine is developed for the purpose of this research. The machine includes an XYZ
movement system and allows the rotation movement for the nozzle, shown in Figure 13.

Figure 13. SSS beta machine CAD model

19

Unlike common additive manufacturing processes, a different slicing method is required for the
SSS process. We use an open-source software called ReplicatorG to generate a primary slicing
data, which is presented as a G-code collection. . For SSS the internal filler is not needed, and
only contour shape of layer profiles are desired. A program is written to modify the primary
slicing document in C++. In this program the contour information is extracted for each layer, and
some points are added for when the nozzle needs to make sharp turns.
3.2. SSS Process Steps and Preliminary Results on Cement Material
At the beginning, S-powder is selected from candidate powders, as it is a critical element in this
process. After selection of the base powder the S-powder is selected from candidate powders as
it is a critical element in this process which impacts the final part quality. In the preliminary
experiments extra fine Soda Lime from Potters Co., is selected to be the suitable S-powder for its
small particle size and smooth surface dimension. The bead particle size range is 45 to 90
microns which is a good match to our requirement as too large or too light particle size will
result in nozzles clogging. The part printing process is shown in Figure 14.

20

Initialization of the
device
Movement of the powder
supply hopper and powder
depositing
separation powder
deposition
Move the nozzle to
the initialization
position
Lower the nozzle
into the powder
Separation of nozzle
and hopper
Layer
finished
Raise
the
nozzle
Raise
the
nozzle
No
Product
finished
End
Yes
Yes
No

Figure 14. Shape building algorithm
In the first step the machine gantry bridge on which the nozzle is installed gets hooked to the
base material hopper and moves it forward and back to spread a layer of the base powder on top
of the build platform. The base powder needs to get compacted under uniform pressure. Next the
bridge unhooks from the hopper to move the nozzle along the part layer boundary while the
nozzle is inserted in the base powder and delivers the S-powder. During S powder delivery the
nozzle is vibrated by the piezo element at the frequency of 2000 Hz. After comp0letion of

21

delivery of the S-powder the nozzle moves up and out of the base powder. The powder volume
in the build tank is lowered by a layer thickness. The gantry bridge hooks to the base power
hopper to deliver the next layer of power to the build tank. The process is repeated until all part
layers are completed.
After the printing process and completion of all layers water spraying on top of the base powder
surface is performed. The water spray nozzle produces very fine water particle which do not
disturb the top surface of the base power.
A water spraying schedule is followed to achieve a good top surface quality and enough water
penetration. In the first step a relatively small amount of water is sprayed onto the powder
surface. After several minutes the powder block becomes strong enough to hold its own shape.
After the early stage solidification of wet powder takes place (for Portland cement, usually 45
minutes) the power tank piston is raised up and the powder block is extracted. In the second step
more water is added to the solidified powder block to complete the hydration process. For the
elected base power 40 grams of water was added for each 100 grams of cement powder, as
specified by the cement vendor. In the final step a frame is also added around the powder block
to prevent the wet powder from possible collapse.
In the third step, certain amount of water is added to the powder block once a day until the
ultimate solidification is achieved. For Portland cement, usually it takes 7 days for final strength
to be achieved. A plastic cover is used in the process of curing to prevent water evaporation.

22

At the end of the curing process the part is extracted by separating it from the surrounding. A
brush is used to remove the extra sand away from the extracted part. After this, the final product
is obtained (Figure 15).

Figure 15. Left: CAD model of the part; Middle: printing process of SSS; Right: separated
part
It should be noted that compaction of the powder volume after injection of the separation power
would result in improved part quality. The compaction force is applied to the top of the powder
and because of the sand content of the mix the reduction in height of the part is often negligible.
To assure a smooth surface for the top layer of the part a thin layer of fine sand is spread on the
top surface of the base powder mix prior to spraying the surface with water. The thin sand layer
protects the bas powder surface from getting disturbed by the pressure of sprayed water.
Moreover, this layer of sand prevents rapid water evaporation, which is a main reason for inner
and surface crack generation.

23

3.3. Research Challenges and Hypotheses
This r focuses on the application of SSS in the fabrication of cementitious samples and the
extension of SSS to large-scale products fabrication. This new technology faces several
challenges. Firstly, a proper choice of separation powder and improved control of the deposition
process is required to achieve good surface quality for fabricated parts. Secondly, a system
should be designed for the process of adding water to ensure an appropriate proportion and
homogenous mixture of water and cement; in addition, water addition should have little impact
on the surface quality. Thirdly, the strength of the fabricated samples needs to be improved
because no complete blending and kneading process between the base powder and the curing
agent liquid exists in the SSS process.
Three hypotheses are proposed for solving the challenges mentioned above:
Hypothesis 1: It is hypothesized that a certain kind of S-powder and a stable deposition of
the S-powder will ensure good surface quality of the side surfaces (Arithmetical mean deviation
is smaller than 0.1mm), and that a proper combination of vibration and air flow will provide a
stable S-powder deposition flow.
Evaluation: The increased deposition flow under various conditions of vibration and air
flow rates can be verified by experiments. The surface quality can be used as a measure of

24

whether the selected S-powder is appropriate and adequate deposition has been achieved. This
research is presented in Chapter 6.1.
Hypothesis 2: It is hypothesized that spreading fine sand over the top surface prior to
spraying water will ensure a good quality surface. In addition, it is posited that spreading a sand
layer will provide good bottom surface quality. The roughness (Arithmetical mean deviation)
will be used to indicate the surface qualities of samples and is hypothesized that it can be
reduced to under 0.2mm.
Evaluation: The roughness (Arithmetical mean deviation) of the top and bottom surfaces
of the parts fabricated by SSS can be measured and used for verification of the hypothesis, as
discussed in Chapter 6.2.
Hypothesis 3: The early strength (2 weeks) of samples could be increased to at least 70%
of the quality of concrete structures made by conventional methods using the same powder
material mix. The strength will be improved after fabrication by exposing the specimen to a
water environment under a controlled temperature.
Evaluation: Flexural strength tests on produced samples will be conducted to determine
samples’ mechanical behavior resulting from certain maintenance process, which will be
designed to prove this hypothesis.

25

All these hypotheses will be discussed in the following chapters. Here, the hypothesis 1
and hypothesis 2 are regarding the surface quality of samples and will be explained and prove
mainly in Chapter 5. Hypothesis 3 is relevant to the strength of samples fabricated by SSS and
will be discussed in Chapter6 and Chapter 7.

26

4. Preliminary Results
Parts are built with Portland cement, Sorel cement,
gypsum and magnesium-based material using a
XYZ three axes printing device shown in Figure
16. Critical results of preliminary experiments are
shown and analyzed in the following sections.
4.1. Portland Cement
Curing of Portland cement is complex and involves
many physical and chemical processes.
According to F.W. Locher [22] the main
reactions that take place during the curing
process of Portland cement are shown as follows:
3CaO ∙ SiO
2
+ nH
2
O = xCaO ∙ SiO
2
∙ yH
2
O + (3 − x)Ca(𝑂𝐻 )
2

2CaO ∙ SiO
2
+ mH
2
O = xCaO ∙ SiO
2
∙ yH
2
O + (2 − x)Ca(𝑂𝐻 )
2

3𝐶𝑎𝑂 ∙ 𝐴𝑙
2
𝑂 3
+ 27H
2
O = 4𝐶𝑎𝑂 ∙ 𝐴𝑙
2
𝑂 3
∙ 19H
2
O + 2𝐶𝑎𝑂 ∙ 𝐴𝑙
2
𝑂 3
∙ 8H
2
O
Figure 16. Selective Separation Shaping prototype

27

4𝐶𝑎𝑂 ∙ 𝐴𝑙
2
𝑂 3
∙ 𝐹𝑒
2
𝑂 3
+ 4Ca(𝑂𝐻 )
2
+ 22H
2
O = 2[4𝐶𝑎𝑂 ∙ (𝐴𝑙
2
𝑂 3 ∙ 𝐹𝑒
2
𝑂 3
) ∙ 13H
2
O]
Equation 1
This curing process is usually divided into three solidification processes:
a) Saturated solution is generated and Ca(𝑂𝐻 )
2
crystals is precipitated out.
b) More Ca(𝑂𝐻 )
2
, Ettringite, C-S-H components are generated and connected with each
other to form strong microstructure.
c) More hydrates are generated gradually, and product’s strength grows slowly.
It is essential to have a good understanding of the curing process to obtain products with high
surface quality and strength. Extensive efforts have been made to find proper parameters and
procedures to enhance the Portland cement solidification process, where solidification situation
becomes different from the traditional method where water and cement are stirring for a long
time.
The Portland cement type that is chosen in this set of experiments is type II/V Portland cement
made by CalPortland Company, which satisfies ASTM C 150 standard [26].

28

Using the above choice of Portland cement many parts have been produced rather rapidly by SSS
which have good dimensional accuracy and smooth surfaces. The single layer 52 X 52 mm parts
shown in Figure 17 are built within 13 minutes, and the 130 by 130 mm double layer part is
printed within 30 minutes.

Figure 17. An early stage Portland cement part made by SSS with smooth side surfaces but
poor edges and rough top surface
The thickness of the petal shape shown above is 10 mm. The side surface shown is very smooth
without any post polishing. The reasons for the good surface quality are expected to be:
As the scale of fabricated specimens is currently much less than the scale of commercial
structures like buildings and large concrete sculptures, the quality of specimen surfaces is very
sensitive to the size of aggregates (sand) in the cement mix. Accordingly the commercial cement
mix used has been sieved using a No.18 mesh sieve, therefore, the sand particles in the cement
are very fine.

29

Commercial concrete is often cast in lumber of steel molds which are rigid and impermeable,
while in SSS the printed parts are surrounded by the very fine separation power which is
permeable and flexible with respect to minuscule displacements. This difference in the
surrounding media causes a significant reduction in micro-cracks on the surface of the parts
made with SSS.
4.2. Gypsum
The gypsum we used is Plaster of Paris from DAP Co. and is chosen for its good quality and
diverse applications. The result of gypsum printing by SSS is shown in Figure 18.

Figure 18. Gypsum parts made by SSS
The single layer gypsum petal shape is 10 mm high and has been printed within 13 minutes.
After two hours for solidification the part is separated from the surrounding material. The surface
of gypsum parts experienced some cracks, which is likely to be caused by fast evaporation and
the heat generated by the chemical reactions during the hydration process. Many experiments

30

were carried out before parameters like water particle size and water flow rate were decided.
Further analysis is still needed, since water penetration rate is very low inside gypsum, and
shrinkage of gypsum is not uniform in different directions.
4.3. Sorel Cement
Sorel cement based parts have also been fabricated by SSS. Sorel cement, also called magnesia
cement, is a widely applied form of cement, as it possesses special properties like faster
solidification and hardening speed, and larger early strength. Fabrication experiments are
conducted and some results are shown in Figure 19. The parts were built with a layer thickness
of 10 mm.

Figure 19. Sorel cement parts made by SSS
Following are the reasons for the relatively poor surface quality of Sorel cement parts:

31

First, the Sorel cement utilized here is not regular size powder, but a combination of powder
particles with a large size range. The existence of large size particles affects the formation of the
separation wall and damages the surface quality.
Second, Sorel cement is more sensitive to the amount of water sprayed on the surface, and the
water absorption rate is less in Sorel cement relative to in Portland cement. Water should be
added with well controlled flow rate, otherwise bubbles would be generated which destroy the
inner and surface structure of the part resulting in porous surfaces. More experiments are planned
to be conducted to find the optimum rate by which a solidifying agent needs to be added to the
base material.
4.4. Magnesium-based material
Magnesium Oxychloride is a combination of magnesium chloride (MgCl) and magnesium oxide
(MgO), which can achieve relatively high strength and has good appearance, therefore it has
been chosen as a potential candidate material for the SSS process. The formula for chemical
reaction between MgCl and MgO solution is shown below [23]:
2Mg𝐶𝑙
2
+ 6Mg𝑂 + 𝑛 𝐻 2
𝑂 = 3Mg(OH)
2
· MgCl
2
· nH
2
0 + 5Mg(OH)
2
· MgCI
2
· nH
2
0 Equation 2

32

First, Magnesium Oxide (with purity of
98% obtained from SIGMA-ALDRICH) is
mixed with sand (200 to 400 micron from
ACTIVE PRODUCTS, INC.) with weight ratio
of 1 to 2. Next, certain amount of Magnesium
Chloride (99.9+% Hexahydrate, MgCl
2
∙ 6𝐻 2
𝑂
from SIGMA-ALDRICH) whose weight is
calculated by molar ratio 7.5. The solution is next
sprayed onto base changing it to a magnesium chloride. After two days of reaction between
MgCl and MgO, the part is solidified and the desired part is separated from the rest of the base
material. In the picture of the part shown in Figure 20 the part surface quality is relatively rough
due to interference between sand inside magnesium oxide powder and the outer separation and
covering sand. Even though the specimens made have relatively high strength, this choice of
material is not considered a preferred material for the SSS application, due to its high price and
chemical instability when exposed to other powders.

Figure 20. Magnesium-based part
made by SSS

33

Table 1. Main building parameters and results comparison on three different materials
Material Type Solidification
Time
W/C Ratio
(water to cement
weight)
Surface
Quality
Portland
Cement
7 days 0.4 Good
Sorel Cement 2 days 0.4 Rough
Gypsum 2 hours 0.55 Mediate
Magnesium-
based Material
2 days
0.5
(solution to powder)
Rough
The results for the different materials are summarized in Table 1. It can be seen in the table that
Portland cement samples possess the best surface quality, while gypsum has the shortest
solidification time. More experiments are needed to improve the results for Sorel Cement and
Magnesium-based materials.

34

5. Surface Quality Analysis
The surface quality of Portland cement parts built by SSS has been studied in more detail. The
results of the experiments indicate that SSS part surface quality may be better than what can be
achieved with other cement based 3D printing processes. In SSS part surfaces are of two kinds:
surfaces which are created by the insertion of the S-powder in the part boundary, and bottom and
top horizontal surfaces which are not created by S-powder insertion.
5.1 Surface’s Quality Interacted with Separation Powder
5.1.1 Impact of Separation Powder Choice
The choice of S-powder will impact part surface quality considerably. A smooth and compact
separation wall made of the right choice of S-powder can create smooth non-horizontal surfaces.
For experiments conducted in this study Soda Lime Potters Co. (size from 40 to 90 microns) is
selected as S-powder because of the following favorable properties:
Firstly, with low water penetration rate, this material entraps most of the water in the base
powder preventing it from easily escaping across the part boundary.

35

Secondly, with 40 to 60 micron particle size, as shown in Figure 21, and high stiffness Soda
Lime can easily generate a robust capillary bridge between particles when exposed to liquid,
which leads to a stronger separation wall structure post water addition. Moreover, fine S-powder
size prevents the diffusion of cement slurry through the part's
boundary. Such diffusion can badly affect part surface quality
as base powder segments outside the layer boundary may also
be fused to the piece.
Thirdly a separation wall made with Soda Lime retains
most of the delivered water (or any fluid used as the
bonding agent) within the volume of the part thereby
preventing rapid evaporation (which causes cracking), and allowing for a complete curing
process which results in stronger parts.
A cement part surface without any post-processing is shown in Figure 22. It can be seen that the
part surface is very smooth and, as measurements have shown, geometric dimensions of the part
shown have been within a few microns of target dimensions.
Figure 21. Soda lime’s micro and
macro image

36

Figure 22. A 24mm tall Portland cement part built with SSS
5.1.2 Separation Powder Deposition
Even though Soda Lime have these excellent properties, insufficient S-powder deposition will
limit its function. If powder deposition is not sufficient, a solid separation wall cannot be
generated or can only be produced with low quality, then Soda Lime’s ability to keep part’s
shape and dimensions close to specification cannot take effect. In Figure 23 two pictures are
shown for comparison between surfaces with and without sufficient S-powder deposition.

37

Figure 23. Left: Cross section of layer without sufficient soda lime as S powder; Right: Cross
section of layer with sufficient S powder
From the picture above, it can be easily seen that surface quality with sufficient S-powder is
much better than surfaces that resulted from insufficient deposition of S-powder.
To ensure sufficient deposition, it is critical to keep the deposition flow stable and prevent the S-
powder from clogging inside the powder container, which
is a common challenge in powder flow situations [13].
The reasons for the clogging in the deposition process
found from experiments are: 1) Contamination happens
inside the S-powder, as there is no seal from powder
container and syringe to outer environment; 2) Under
effects of gravity powder grains can form an arch during
Figure 24. Plot of arch
generation inside the syringe [13]

38

powder deposition, which would block powder from flowing down, as shown in the Figure 24.
To solve the first problem, a sieving stage is added every time before the S-powder is filled into
the syringe to prevent contamination. To solve the second problem both aggressive vibration and
air flow to be directed to the bottom of the hopper to blow up the formed arches may be used.
Considering that the hopper applied in SSS is in small size, the vibration method is implemented
in the first stage. After several experiments with piezo vibration an amplitude of 5 microns and a
frequency of 2KHz were successfully applied which effectively prevented arch formation.
For increased deposition rate, airflow was applied to the upper side of the hopper containing the
S-powder. The result of experiments showed that this method is capable of increasing low
deposition rate caused by high density powder particles. To obtain stable and adequate S-powder
deposition, a miniature pump with proper air pressure was installed on the S-powder container to
pressurize the container and increase the outflow rate of the S-powder from the nozzle.
To understand the effects of the main factors on the deposition a full factorial design experiment
was developed and conducted. Three factors are chosen to be analyzed according to previous
experimental results and lab equipment. As presented in Table 2, “A” represents vibration of
cellphone vibrator, which has low frequency but high amplitude, “B” represents vibration of
piezo vibration, has high frequency but low amplitude, “C” represents airflow, which is provided
by a diaphragm pump. In the experiments, only two status of factors are considered: “+”

39

indicates the element is present, i.e., airflow is added, “-” indicates the element is not present.
Three replication is finished according to the design table, and the weight of soda lime deposited
within 300 seconds is recorded and marked as “Y” in Table 2.
The design table and result is shown in Table 2:
Table 2. The 𝟐 𝟑 full factorial design table
A B C Y1/g Y2/g Y3/g
- - - 0 0 0
- - + 0 0 0
- + - 2.755 1.611 0.795
- + + 9.167 19.569 14.206
+ - - 3.265 3.05 0.625
+ - + 1.985 6.87 6.667
+ + - 3.11 0.969 0.953

40

+ + + 6.014 15.805 8.949
From the experiments’ results, half norm plots about displacement and dispersion are generated
to show the significance of parameters.

Figure 25. Half norm plot for dispersion of experiment results

Figure 26. Half norm plot for variance of experiment results

41

From the two plots we can see that main factors B and C are significant in affecting the
dispersion of deposition, while the main factor C and interaction between B and C are significant
to change the deposition quality. Furthermore, regression models based on the significant factors
are calculated. According to the 2 step strategy, B and C are chosen to be positive, while A is
selected as negative, to achieve maximum deposition amount in low energy cost and system
complexity. In conclusion, a combination solution of both piezo vibration and air flow is
introduced as an economical and efficient solution to ensure a stable and sufficient powder flow.
The regression equations are shown below:
𝑦 = 4.432 + 2.560 ∗ 𝐵 + 3.004 ∗ 𝐶 Equaiton 3
𝜎 = 0.1924 + 0.7466 ∗ 𝐶 + 0.2580 ∗ 𝐵 ∗ 𝐶 Equation 4
5.1.3 Deposition Simulation
S-powder deposition is a critical process for the success of SSS, because the quality of powder
deposition flow impacts part surface quality and resolution. With changing of base material and
S-powder material, achieving stable powder flow for different materials is a major challenge.

42

In previous work performed on development of SSS for printing ceramic and metallic materials
[25, 27] a lot of experiments have been conducted to optimize the settings for deposition of
tungsten powder to achieve stable powder flow. Although experiments provide the results closest
to the real fabrication process, they usually take a lot of time to be conducted, and those results
are not transferable between different deposition powders. To make the whole process more
efficient and switch between materials more smoothly, a numerical simulation of powder
deposition is considered as a good way to provide a guide for setting adjustment of specific S-
powder deposition.
The S-powder flow is a special case of the general granular flow [28], which mainly concerns
solid particles and the gaps between them. When considering the quality of granular flow the
interaction between powder particles plays a dominant role in flow properties, and simulation of
granular flow is commonly considered a challenging problem because the interaction between
particles is very complicated and most general hypotheses and conclusions in fluid dynamics
cannot be applied to the granular flow phenomenon.
Based on Nucleus and n-particle technology [29], the commercial software Maya is considered
to be a capable platform to build a practical and accurate simulation to show the effect of
different parameters on particle deposition. In this proposal, the clogging problem, which is
described in section 5.1.3, is considered the most critical and central problem. A Nucleus and n-
particle-based simulation, where a ball-spring model is applied, is deemed to be proper for
deposition flow quality simulation because the issues of robust longtime simulation and micro-

43

fluid dynamic influence are not considered to be primary in this simulation, and hence are not
considered.
In the simulation a model of the syringe-needle system is built in Maya according to the actual
syringe-needle configuration. Then, a vibration simulation conducted in Solidworks to model the
vibration information of the syringe-needle system. Next, a model of soda lime flow is developed
in Maya and the parameters of the powder are entered in the simulation model. In the last stage
of simulation the deposition animation under different scenarios is built.
In detail, a model is built under Maya 2017. As shown in Figure 27 the model includes a
container (combination of a syringe and a needle) and S-powder (a mixture of particles of 50 and
200 microns). The syringe is modeled to have a diameter of 16mm, the needle is modeled to
have a diameter of 1200 microns.

Figure 27. Left: actual deposition device; Right: model built in Maya

44

For the vibration parameters applied here, they are calculated by a SolidWorks vibration
simulation. As shown in Figure 28, a syringe with the same dimension as that in Maya is made,
and the vibration, which is simplified to a uniform force, is added from a piezo disc placed under
the needle (shown as red arrows in Figure 28).

Figure 28. Left: SolidWorks model on deposition device; Right: simulation result of syringe’s
displacement under vibration
From the displacement simulation of a syringe, shown in the right side of Figure 28, it is
reasonable to decouple the syringe’s vibration into two kinds of vibrations: vertical vibration and
horizontal vibration. Moreover, we can see that the amplitude of vertical vibration is two times
of the horizontal vibration amplitude.
First and third natural frequencies of the syringe, which is considered the best representation of
syringe’s dynamic properties under vibration, are also obtained, with the aim of getting more
information about the syringe’s reaction to vibration wave. In the results shown in Figure 29 the

45

first natural frequency is found as 300.72 Hz, where the primary displacement is bending of the
syringe. The third frequency is found as 1700.3 Hz, where the primary deformation is the
shortening and expanding of the syringe. To avoid bending displacement of the syringe, which
may affect SSS’s fabrication accuracy, the frequency of piezo vibration is always chosen to be
larger than 600Hz.

Figure 29. Left: displacement of syringe for first natural frequency; Right: displacement of
syringe for third natural frequency
After obtaining dynamic information of syringe reacting to vibration, efforts are made to find the
proper parameters for soda lime in the simulation. In general, the powder has four critical
properties: particle size distribution, bulk density, dilatancy, and shear force [30-33]. The
measurements for these characteristics of powder usually require expensive devices. Instead, a
natural and low-cost method to measure powder’s main feature is applied here, which utilizes the
concept known as the angle of repose. [30-33]

46

Figure 30. Left: an indication of angle of repose, where 𝜶 𝑹 is angle of repose [34]. Right: an actual
pile of Soda Lime
The angle of repose is an indicator to show a combination effect of main characteristics of
particle flow. The Angle of repose is shown in the left picture of Figure 30, and we can see that it
is the angle between the tangential plane of a pile and ground, where the pile is generated free
from deposition of powder through a hopper. In most cases, two powders can be considered to
have similar shear force and viscosity property if they have the same angle of repose. In the
Nucleus-based simulation applied in this research, parameters of the powder, which include
friction, viscosity, rotation, incompressibility, stickiness and surface tension, are adjusted to
make sure the angle of repose of model and real Soda Lime pile are the same, as shown in Figure
31.

47

Figure 31. Comparison of angle of repose between Nucleus model and real Soda Lime pile

Figure 32. Left: dynamic simulation result of powder deposition in syringe with no vibration;
Right: powder deposition under vibration of 1000 Hz
After the parameters of soda lime particle in the simulation are fixed, different vibration settings
are applied and the animation results are recorded. From the left image in Figure 32 we can see
that powder deposition will be clogged without vibration, and only some leakage of powder
show up in the animation. With a wave of 1000Hz in both vertical and horizontal directions, as
shown in the right image of Figure 32, a stable powder flow is generated and maintained. In
conclusion, the simulation can show the conditions of clogging of powder and various behaviors
of powder under different vibrations conditions.

48

As the results of vibration simulation show, the deposition is enhanced until the vibration
frequency of the piezo disk reaches 1500 Hz and then it worsen as the frequency of vibration
increases. The reason behind this phenomenon is that the vibration breaks the arches that cause
clogging by increasing particle dynamic, while the high-frequency vibration disturbs the
deposition and causes flow stoppage. This conclusion is consistent with previously reported
experiments [27].
5.2 Quality of Top and Bottom Surfaces
The top and bottom surface of parts are considered of great importance for many applications,
such as consumer products (e.g., furniture) and public infrastructure (e.g., sculptures,
landmarks). In the current SSS process the top and bottom surfaces are not defined by deposition
powder but by base machine platform plate, and by the last powder layer spread on top of the
build tank. Controlling the quality of the top surface is somewhat more difficult than the bottom
surface, because a direct influence from the water spray is not avoidable for the top surface. In
this chapter, the parameters that are considered to have significant influence on the quality of the
top and bottom surfaces are analyzed and methods to improve the surface qualities are discussed.
Spraying water on every layer of the part as it is being printed would disturb the top of all layers.
Accordingly, the approach used is to spread each base powder layer and deposit the S-powder on
the layer profile lines and continue the process until all layers are treated. Water is then sprayed
on top of the last layer. The challenge is to spray the water in such a way that all layers

49

homogenously receive a sufficient amount of water for cement hydration without disturbing the
top surface.
A layer of sand is also added at the bottom of the build tank to make the separation between the
fabricated green part and the build tank piston easier, as explained in Chapter 3. This layer of
sand introduces new factors that affect the surface quality of the part bottom. Since the causes of
the roughness of the top and bottom surfaces are different, these two surface quality cases will be
discussed separately.
5.2.1 Material and Roughness Measurement Process
Type II/IV Portland cement from the CalPortland Company was chosen for analysis since it is a
common building material that is widely used in the construction industry. The research on this
material is also considered applicable to other cementitious materials since this cement has
typical characteristics in common with other cement types.
The surface quality is analyzed by measuring the roughness of samples using a dial gauge as
shown in Figure 33. The two parameters [35] of arithmetical mean deviation and maximum
(peak) height have been chosen to indicate the surface roughness since the overall surface quality
and the largest surface displacement are the two most important characteristics for commercial
products.

50

Figure 33. The dial gauge applied to measure the roughness of samples
Arithmetical mean deviation of the assessed surface profile is used as the indicator of general
surface roughness. During the measurement, a line is drawn on the surface of the samples and
then spots are marked along the line at a constant distance. The equation to calculate the
Arithmetical mean deviation is shown in Equation 5.

Figure 34. Different parameters used for the measurement of roughness [35]

51

𝑅 𝑎 =
1
𝑛 ∑ |𝑦 𝑖 |
𝑛 𝑖 =1
Equation 5
where |𝑦 𝑖 | is the height of a spot on the surface of a sample relative to the standard surface.
Another parameter utilized is the maximum peak height, which shows the largest
displacement on the surface. Equation 6 shows how to calculate the maximum peak height of a
sample.
𝑅 𝑝 = max|𝑦 𝑖 | Equation 6
where |𝑦 𝑖 | has the same definition as equation 5.
5.2.2 Bottom Surface Quality
A smooth bottom surface will allow the samples fabricated by SSS to be used in additional
applications (e.g., sculptures) while saving time needed for post-processing. In the current SSS
process a layer of sand (with grain sizes ranging from 200 microns to 400 microns) is paved on
top of the piston inside the build tank. The first layer of the base material is then spread over this
sand layer. The interaction of the sand layer and building material after water is added will be the
focus of this section since it is believed to be vital in determining the final bottom surface
quality.

52

During the interaction between the sand layer and the building material, two variables are
considered to potentially influence the bottom surface quality:
a) The surface of the bottom sand layer will become uneven when water penetrates the sand
layer and flows away at different rates.
b) An uneven expansion of cement during curing due to non-uniformwater dispersion will
cause the displacement of the cement surface.
These two problems are studied separately below.
a. Analysis of the Displacement of the Sand Layer after Water Flows Away
During the fabrication process water will be added to the whole block after all layers are
deposited and processed. Since the sand layer is at the bottom, a portion of the water that has
penetrated the cement material will accumulate in the sand layer at the bottom. After a few
minutes (approximately ten minutes for the prototype machine) the sand layer will become
saturated with water and the excess water will flow away from the sand. The different dispersion
rates of water streams flowing toward the edgiest of the build tank piston will cause an uneven
surface in the sand layer and negatively impact the bottom surface of the sample. To study this
further, experiments were conducted as described below.

53

To more clearly observe the surface change of the sand layer
when water flows away, only the sand layer is studied in this
experiment. In the first step, one layer of sand (thickness 2mm,
4mm, and 8mm) is paved in the 200mm x 200mm building tank
as shown in Figure 35. In the second step, the liquid is sprayed
directly on the sand until the sand layer is saturated.
Thirty minutes after the sand is sprayed it is considered that the
dispersal of water from the sand layer is completed and the
surface of the sand layer has become stable. At this time pictures were taken to determine if a
non-uniform dispersion rate of water in different sand areas occurred. The area in the middle is
noticed to have a relatively lower dispersion rate compared to the area close to the edge, and this
was believed to influence the surface roughness of the sand layer.
As a result of this experiment it was found that different dispersion rates of water do not
significantly influence the final surface quality. As shown in Figure 36, the surfaces of the sand
layer do not show obvious displacement after water is added and flows away from the sand,
regardless of the thickness of the sand layer. In conclusion, the non-uniform dispersion rates of
water inside the sand layer do not significantly affect the surface quality of the sand layer.
Figure 35. One layer of sand (2mm)
after the water is added

54

Figure 36. Surface qualities of sand layers at different thicknesses
b. Uneven Dispersion of Water Inside Cement
Since the cement volume expands after water is added, the uneven dispersion of water inside
cement will lead to non-uniform expansion of different surface areas within the samples. To study
the water dispersion conditions inside the cement powder, a camera is placed under the bottom of
a 160mm x 160mm transparent tank to record the water penetration on the bottom of the cement
powder; a 12mm thick layer of cement powder is paved inside the tank (Figure 39). Water is
sprayed onto the cement powder at the speed of 120g/min with the system shown in Figure 39.
The results from the different times are shown in Figure 37. From left to right, the images were
taken when water is added for 80, 90, and 140 seconds, respectively

55

Figure 37. Water dispersion on the bottom surface of cement
From these results we can see that water dispersion in the cement powder is uneven. After 40
seconds water reaches the bottom surface of the cement. In about 160 seconds the whole block
becomes saturated with water and the whole bottom surface of cement becomes darker. It was
found that the spread of water inside the cement powder will remain non-uniform for about 120
seconds, which is considered long enough for the displacement on the bottom surface to happen.
It is believed that the uneven weight distribution of cement and the non-uniform expansion of
cement caused by the uneven dispersion of water is the reason for the increased roughness on the
bottom surface of the samples.
c. Current Solution
Since the non-uniform dispersion of water inside cement is difficult to avoid, methods to
increase the stiffness of the sand layer to reduce the negative effects from the cement are
promising. At present, the solution utilized is to mix one part water with eight parts sand before

56

paving the sand as the bottom layer. After mixing with water, sand grains can form capillary
bridges between each other to create a stronger structure to hold the cement sample. [36]
A set of comparison experiments were conducted to study the influence of the wet sand layer and
water spraying speed on the roughness of the bottom surface. Samples of the size 200mm x
200mm x 12mm are fabricated with the SSS machine for these experiments.
In the first experiment a 2mm thick dry sand layer is paved on the building platform and the
sample is fabricated. Water is then added to the cement at a rate of 60g/min. In the second
experiment the same sand layer thickness is applied while the speed of the water is decreased to
30g/min. In the third experiment, a 2mm thick wet sand layer is paved on the building tank and a
sample is fabricated on the wet sand layer. Water is sprayed onto the cement at the speed of
30g/min. The characteristics of the samples fabricated under these different conditions is shown
in Figure 38 and the roughness parameters measured in these experiments are recorded in Table
3.
Table 3. The bottom surface roughness of samples fabricated under dry/wet sand layers with
different water speeds

Spray speed Roughness (mm)
Maximum peak height (mm) Wet sand
1
60g/min 0.262
2.22 No
2
30g/min 0.246
1.70 No
3
30g/min 0.192
0.79 Yes

57

Figure 38. These images show the bottom surface of samples fabricated with dry/wet sand layers
with different water speeds.
Based on Equation 5 and 6, the arithmetical mean deviation and the maximum peak height of the
samples are recorded in Table 3. From the table, the arithmetical mean deviation of the cement
sample built on dry sand is 0.246mm and 0.262mm at the two spraying speeds, and the
maximum peak height of the samples is 1.70mm and 2.22mm, respectively. The surface quality
of samples fabricated under slower water speeds is considered better in both measured roughness
parameters. It is believed that the reason for this is that under high spraying speeds water will not
have enough time to spread in the lateral direction uniformly inside the cement and non-uniform
expansion caused by the dispersion of water will worsen.

58

Figure 38 shows the bottom surface qualities of three samples fabricated in the three sets of
parameters as presented in Table 3. From the images and calculated roughness parameters, the
samples fabricated on the wet sand under the condition of 30g/min spraying speed have the best
surface quality. In conclusion, the mixture of water and sand as the separation bottom layer has a
positive influence on the bottom surface quality of cement samples, and a slow water spraying
speed benefits the bottom surface quality.
5.2.3 Top Surface Quality
The surface quality of the top of samples is considered to be of equal or higher importance since
the top surface will be would be viewable in most situations. The main challenge faced when
trying to increase the upper surface quality of samples is that the direct impact of water on the
cement powder is strong. During the water spraying process, water mist is sprayed directly on
the cement powder bed and in this process large water droplets may form which damage the
cement surface due to impact under gravity. Cracks may also occur due to the non-uniform
dispersion of water inside the cement.
5.2.3.1 Water spraying system
A mist spraying system, shown in Figure 39, has been developed to generate water mist,
which has been helpful in reducing the high-pressure impact of large water droplets on powder
surfaces. With the spray system small water droplets are spread onto the powder bed, which
ensures a relatively even water distribution and a small impact on the base powder.

59

Figure 39. A spray system with four nozzles to spread water
5.2.3.2. Analysis of Sand Cover
As shown in Figure 40, an experimental system has been developed to analyze the influence of
water spraying on the surface of the cement. Firstly, experiments were conducted in which water
was sprayed directly onto the cement inside a transparent tank. During the experiment a 20mm
thick layer of cement was paved inside the 160mm x 160mm tank and water was sprayed at the
speed of 120g/min. The cement block was removed from the tank once it completely solidified.
In Figure 40 the picture on the left shows some cracks and displacement, indicating that the mist
spray system cannot produce a smooth top surface.

60

Figure 40.The testing system for top surface quality analysis and the image of top surface of a
sample fabricated with water sprayed directly on the cement
To make the water distribution even and reduce the impact of the spray pressure on the surface
deformation, coarse sand with particle sizes ranging from 200 microns to 400 microns was used
to cover both bottom and upper surfaces. On the upper surface, water droplets reach the coarse
sand before penetrating into the cement. Water can spread quickly inside coarse sand, promoting
even water distribution. Furthermore, the sand reduces the direct impact of water spray on the
base powder (cement) surface.
The proper thickness of the sand layer should also be determined for the SSS process; a good
quality surface in combination with the least material consumed is ideal. To check how thickness
influences the surface quality of samples, a set of four experiments were conducted using 2mm,
4mm, 6mm, and 8mm thick sand layers. These were paved onto type II/IV Portland cement

61

inside molds as shown in Figure 41, and water was sprayed onto the molds from the height of
340mm at the speed of 120g/min.

Figure 41. These images show the top surface of samples before water was added
In Figure 41 and Figure 42 we can see that samples fabricated with a protective sand layer show
no obvious defects after solidification. Comparing images of samples with sand protection

62

(Figure 42) and samples without sand protection (Figure 40), we can see that a 2mm thick sand
layer can prevent the negative impact of direct water spraying and ensure better surface quality.
Furthermore, as shown in Figure 42, there is no significant difference between the surface
qualities of samples fabricated under different thicknesses of sand layers. In conclusion, a 2mm
thick sand layer should be utilized to ensure a good quality top surface.

Figure 42. These images show the top surfaces of samples after water was added
5.2.3.3 The Influence of Spraying Speed
Preliminary experiment results showed that a fast water spray rate will increase the roughness of
the top surfaces of samples even if a sand layer is used on top of the cement powder. In this
section the impact of the water spraying speed will be analyzed.
To better understand the influence of water spraying speed on the cement, three sets of
experiments were conducted. In these experiments a 12mm thick cement layer was paved on the
200mm x 200mm building tank of the SSS machine, and water was sprayed directly on the

63

cement using the equipment shown in Figure 40 at a distance of 340mm. During the
experiments, the sand layer was not used because it was easier to observe changes to the cement
surface without the sand layer present.

Figure 43. Images of the top surface of samples fabricated under different water spraying speeds
From left to right, the images in Figure 43 show the water spraying speed applied at 120g/min,
60g/min, and 30g/min. It can be observed that the number of cracks on the surface significantly
decreases with a decrease in spraying speed. At the spraying speed of 30g/min, small cracks
show up on the surface of the sample. With the protection of the sand layer, it is believed that a
spraying speed higher than 30g/min may still provide a good top surface. In conclusion, to
ensure a smooth and stable top surface, a spraying speed of 30g/min was chosen as the best one
for the SSS process.

64

5.3 Demonstration Samples with Optimized Process

Figure 44. Demonstration of samples with improved surface qualities
The double layer petal shape shown in Figure 44 is 22mm thick with an outer diameter of
125mm. The left image demonstrates the bottom surface quality of the sample (Arithmetical
mean deviation is below 0.2mm), and the image on the right illustrates the upper and side surface
qualities (Arithmetical mean deviation is below 0.1mm), which is rather impressive in the world
of 3D printing; this is comparable with conventional cement molding methods.

65

Figure 45. Measurement of Arithmetical mean deviation of sides of samples
In conclusion, with the proper plan for using building materials and careful control of the
fabrication process, it is possible to achieve smooth surface qualities in parts fabricated by SSS.

66

6. Analysis of Mechanical Properties of Parts Made by SSS
In this section, several parameters that influence the mechanical properties of cement parts made
by SSS are discussed, and a proper process to economically improve cement strength is
introduced.
6.1. Specimens
Portland cement (Type II/IV from Quikrete Corp. [37] complied with ASTM C-150) was chosen
for this study of how to improve cement strength because this type of cement is the most
common [38] and economic building material for building scale products and structures. The
samples created as part of this research underwent flexural testing to determine the relative
strength of each specimen. This test is preferred to compressive and tensile strength tests because
of its simplicity and direct indication of the material’s strength property.
Different parameters are studied to improve the mechanical property of SSS-produced samples,
where the water–cement ratio (by weight) is the most crucial parameter for cement’s ultimate
strength. The water–cement ratio impacts a cement’s saturation level, porosity level, and
microstructure [39], which collectively determine a cementitious product’s mechanical behavior.
For conventional processes the water–cement ratio is usually 0.35 to 0.5. In the SSS process a
specific water–cement ratio is expected because, unlike conventional cement preparation, no
stirring or mixing is included in SSS. This may lead to an insufficient amount of water reaching

67

certain cement segments, thereby negatively impacting the curing process. Additives, such as
silica fume and superplasticizer, are also considered as potentially helpful components that can
be added to increase concrete strength [40].
Maintenance during the curing process is another factor that impacts cement strength. Hydration
temperature and moisture are the two most important maintenance factors during the curing
process. An environment with proper temperature and moisture keeps cement’s hydration speed
at a controlled level and reduces crack generation [41].
To analyze the effects of different parameters, eight sets of experiments have been conducted.
Table 3 shows the parameter values for the seven different settings, where every experiment with
the same setting is repeated three times and flexural strength is calculated as the average of the
three results. Set 1, set 2, and set 3 tested samples with different water–cement ratios. The
material used in set 4 was a mixture of cement and sand to understand the effect of sand on the
specimen’s strength. In set 5, silica fume was added to cement in which silica fume’s weight was
5% of the whole mass. In set 6, a superplasticizer admixture was mixed with water in the
percentage of 14%, and water was added into cement with water ratio of 0.4 by mass. In set 7,
cement was stirred with water and paste, and injected into a mold. In set 1 to 7, samples were
submerged in pure water at 25 ℃ after primary curing (within 24 hours). In set 8, samples were
not submerged in water to serve as a basis for comparison.

68

6.2 Test Procedure
The flexural strength of the specimens was
measured after 14 days of curing according to
standard ASTM C348 - 14. During the
measurement process, 20mm x 20mm x
80mm samples were first fabricated by the
SSS machine, and force was applied on the
center of the parts as shown in Figure 46. The
parts were fixed on a testing bed and the load
was applied by the NK-500 analog force gauge, as shown in Figure 47. The first crack load was
recorded as F, and the specimens’ flexural strengths were calculated following the equation
shown below:
𝑅 𝑓 =
1.5 𝐹 ∗𝐿 𝑏 3
Equation 7
where L is the distance between supporters, and b (20mm) indicates width and height of samples.
Figure 46. Loading on flexural specimens.

69

Figure 47. Left: flexural strength measuring device. Right: a broken part after measurement.
6.3 Analysis of Results
Table 4. Parameter settings for strength measurement.
Sample
Number
W/C
Ratio
Cement
Agent
Percent
cement Stirring
Days
Submerged

Additive
Force
Applied (N)
Flexural
Strength (MPa)
1 0.35 Portland 100 No 14
N/A
191.25
2.15

2 0.4 Portland 100 No 14 N/A 217.5 2.45
3 0.45 Portland 100 No 14 N/A 177.24 1.99
4 0.4 Portland 60 No 14 N/A 211.5 2.38

70

5 0.4 Portland 100 No 14
Silica
Fume 228.48 2.57
6 0.4 Portland 100 No 14
Plasticiz
er 195.54
2.20
7 0.4 Portland 100 Yes 14 N/A 250.13 2.81
8 0.4 Portland 100 No 1 N/A 114.76 1.29
Table 5. Mean flexural strength index (the blue bar represents the strength of different samples;
the red bar represents the strength of sample 2).

71

According to Table 4 and Table 5, 0.4 is the best water –cement ratio for the SSS process, and the
addition of silica fume and superplasticizer do not significantly increase a sample’s final
strength. As shown in the data for set 4, the mixture with sand is an excellent choice if cost
reduction has a high priority as this mixture does not reduce the specimen’s strength. Although
set 7 has a higher flexural strength compared to set 4, SSS is still considered a proper fabrication
process for cement because samples built with SSS are only 15% weaker than typical industrial
concrete building methods. Further SSS research will help to reduce this strength deficiency. As
shown in set 8, samples without a maintenance process during curing are much weaker than
samples cured under a maintenance environment, which leads to the conclusion that the
submersion maintenance process is critical to the strength of samples.
6.4 Strength Analysis of a Honeycomb Structure
Honeycomb structures are proven to have superior mechanical behavior with the least weight
[42]. A high strength-to-weight ratio of materials development is always an attractive field in
material science since this characteristic has broad appeal in areas such as aerospace, defense,
and construction. For years, people have sought to increase a material’s strength-to-weight ratio
by reducing a material’s density or by increasing a material’s strength. Instead of approaches
based on material composition, more recent research [43-45] has focused on a microstructure-
optimization-based study to reduce a material’s density, as well as on part geometry using
topology optimization. As discussed by Jens Bauer in [42], honeycomb structures (a stretching
dominated structure) are believed to be a promising design for materials to replace traditional

72

structures, allowing the material to have a high strength-to-weight ratio. Honeycomb structures,
as shown in Figure 48(a), have a relatively complicated shape in the horizontal direction, but stay
uniform in the vertical direction; therefore, the SSS technology would be a good match for
fabrication of such a structure.
Before experiments were conducted to fabricate a honeycomb structure, a SolidWorks-based
simulation was performed to show the advantages of honeycomb structures over other shapes. In
this research a part is considered a failure if its strain goes beyond 0.1%. The density of a
honeycomb structure and a block structure with the same compression strength have been
compared to show the advantages of the former shape. A honeycomb model with a surface area
of 2418.36mm
2
in the horizontal plane was created and is illustrated in Figure 48(a). In the first
step, a uniform force was applied to the top of the honeycomb part and was found to generate
0.078mm strain (0.1% of a part’s height) in the vertical direction.
Under the same force, a square block structure with an area of 4900mm
2
in the horizontal plane
is found to generate the same strain as the honeycomb part. As the simulation shows, the
honeycomb structure’s compression yield strength could be the same as that of a square block
structure, but with 49.34% less horizontal surface area. In conclusion, a honeycomb structure can
reach the same compression yield strength as a square structure with only half the weight of the
square block shape.

73

Figure 48. (a) The honeycomb structure’s compression displacement simulation result. (b)
The block structure’s compression displacement simulation result.

Figure 49. A 23.50mm high honeycomb structure developed by SSS using type II/IV Portland
cement.
A honeycomb shape built with SSS is shown in Figure 49. From the picture, we can see that SSS
has the capability to manufacture complex 2.5D forms with high resolution and surface quality.
Additionally, the SSS process is shown to be more efficient than conventional methods in the
b a

74

construction of such complicated 2.5D structures, especially with regards to cementitious
materials.

75

7. Statistical Predictive Modeling of Strength of SSS-Produced Parts
Information about how strength changes during the curing process is essential for increasing the
efficiency and reliability of the whole SSS process. There are two important time nodes in the
curing process that decide how efficient the fabrication will be. The first time node is when the
fabricated part can obtain preliminary strength so that the desired part can be safely separated
from the green part; the second is when the fabricated part reaches the strength necessary to
satisfy industrial requirements and can be taken out of the maintenance environment. The
knowledge of when the first node will be reached is needed so that the separation time can be
chosen to allow for improved safety and efficiency in the process.
Most of the previous predictive models of strength are based on cement material that is
developed by fully blending water and cement [46, 47, 48]. As discussed earlier, in SSS blending
and kneading is not used in mixing water and cement. In this chapter, a predictive model on the
change of strength of the parts fabricated by SSS is developed. SSS functions through the use of
water sprayed onto cement powder; this chapter focuses on developing a predictive model on the
change of strength in parts fabricated in such a manner. Flexural strength has been chosen as the
indicator of the sample strength because it is one of the most commonly applied indicators of
mechanical behavior of cement material products and is considered a more important sign for
safety during the separation process.

76

7.1 Experimental Material and Process

The material used is Portland type II/IV cement,
the same as in section 6.2. According to the
standard ASTM - C348, parts measuring 20mm
x 20mm x 80mm are fabricated by a SSS
prototype machine. All parts are exposed to air
in the first 12 hours after fabrication and then
submerged in water for maintenance under
certain temperatures.
Several sets of samples were fabricated
successfully, with no obvious crack or
boundaries between layers; they are shown in
Figure 50. These sets are then submerged in water for different lengths of time and at multiple
temperatures to study the relationship between the flexural strength of samples, time submersed,
and curing temperature.
As discussed above, samples are submerged in water in the maintenance box for curing.
Different water temperatures are used to study the strength change of samples under these
Figure 50. Samples ready for the flexural strength test.

77

variable conditions. As shown in Figure 51, a maintenance box is
applied for the curing process under room temperature.
A maintenance tank was built to provide a consistent environment for
the curing of samples; this is shown in Figure 52. This maintenance
system includes a water-heating tank, a PID temperature
controller, and a solid-state relay.

Figure 52. Maintenance tank with temperature control capability.
After the planned length of time, samples are taken out of the maintenance devices and tested for
flexural strength. The flexural strength tests are conducted on the load frame machine INSTRON
5567, which is shown in Figure 8.4.
Figure 51. Samples inside a
maintenance box

78

Figure 53. The load frame without a sample is shown on the left, and a sample during the flexural
strength test process is on the right.
Based on Figure 53, it can be observed that the samples have lightened in color compared to
Figure 50. Samples are placed in the air for six hours before the tests to reduce their interior
humidity since the humidity of samples is an important element that influences sample strength.
During the measurement, the load on the parts will be increased at a consistent speed until the
resistance force from the samples decreases by more than 40%, which means the structures of the
tested part are broken. A typical plot that shows the load change during the test is shown in
Figure 54. The maximum flexural load (N), 192N as shown in Figure 54, is chosen to calculate
the flexural strength of the samples in further analysis.

79

Figure 54. Load change during the flexural strength test.
7.2 Experiment Results
In total, seven sets of experiments were conducted. Four sets of samples were fabricated and
submerged in 20 ℃ water for three, four, five, and seven days, and three sets of samples were
fabricated and submerged in 50 ℃ water for three, four, and six days; the latter temperature is
considered as providing the strongest structure for samples [49].
During the flexural strength tests, the largest load to break the fabricated parts is recorded in
Newtons. Within each set of experiments, the average strength of all the samples is calculated.
As discussed in Chapter 7, Equation 7.1 is applied to calculate the flexural strength of samples.
The results for the samples submerged in the 20 ℃ water is shown in Table 6, and the results for
samples maintained in the 50 ℃ water is shown in Table 7.

80

Table 6. The flexural strength of samples maintained in 20 ℃ water.

7 days 5 days 4 days 3 days
Maximum load (N) 142.1618 135.6159 122.733 126.1307
Flexural strength (MPa)
1.48515 1.356323 1.320431 1.279829
Table 7. The flexural strength of samples maintained in 50 ℃ water.

6 days 4 days 3 days
Maximum load (N) 292.2635 249.8186 132.7036
Flexural strength (MPa) 2.96555625 2.60979 1.386323
Based on the data in Tables 6 and 7, the samples submerged at a higher temperature are stronger
within the first week. This result is consistent with other research on the influence of temperature
[50]. Compared with results from Chapter 7, where after two weeks the flexural strength of
samples built by conventional method is 2.81 MPa, the samples built with SSS can obtain the
same level of flexural strength (2.96MPa). In conclusion, samples built by SSS can obtain
improved flexural strength by using a proper maintenance method, and there is great potential in
utilizing this method in real industrial applications.

81

7.3 Statistical Predictive Model Development
In this part of the chapter statistical predictive models are developed and analyzed for the
relationship between strength and sample age. In the first section, different predictive models
will be separately fitted for the relationship between flexural strength and the curing time of
samples under different temperatures. Furthermore, the predictive model including the
temperature is desired since strength prediction under different maintenance temperatures would
be a useful. In the second section, predictive models that include temperatures and ages are
developed and studied.
7.3.1 Models under Different Temperatures
In previous research, models to predict the compressive strength change of samples with age
were developed by Metwally [51]. In his research, it is claimed that compression strength and the
logarithmic timeline are linearly related.
As described in references [51], a model is applied as shown in Equation 7:
𝑓 𝑡 = 𝐴𝑙𝑛 (𝑡 ) + 𝐵 Equation 7
where 𝑓 𝑡 is compressive strength at age t, A and B are parameters.

82

With the consideration that the flexural strength is mainly focused on and that the curing time
(one or two weeks) is short, a linear relationship (shown in Equation 8) between the strength and
age is also assumed reasonable.
𝑓 𝑡 = 𝐴𝑡 + 𝐵 Equation 8
The coefficient of determination 𝑅 2
[50], which is commonly used to show how well the
fitted equation can explain the experiment results, is chosen as the quantity to estimate the
performance of the predictive model.
7.3.1.1 Predictive Model for Samples Cured at 20 ℃
With an ordinary least squares linear regression model applied, a line is fitted to show the
relationship between the strength and age of parts by using Equation 7. The calculated regression
model is shown below:
𝑓 𝑡 = 0.2395𝑙𝑛 (𝑡 ) + 0.9988
The 𝑅 2
is of this model is calculated to be 0.9315.
If the model 𝑓 𝑡 = 𝐴𝑡 + 𝐵 is used, the fitted model will be:
𝑓 𝑡 = 0.0515t + 1.1158

83

The 𝑅 2
is of this model is 0.9806. The two fitted lines and the original data are shown in Figure
55.

Figure 55. The figure on the left shows the regression line using logarithm time under 20 ℃, and
the figure on the right shows the regression line for the time under 20 ℃.
From the 𝑅 2
score we can conclude that the performance of the model 𝑓 𝑡 = 𝐴𝑡 + 𝐵 is better,
because a higher 𝑅 2
score indicates a better fitness between the equation and the real data [52].
8.3.1.2 Predictive Model for Samples Cured at 50 ℃
As with the predictive model development for samples submerged in 20 ℃ water, two models
are also fitted for the samples cured at 50℃.
For the line between the logarithmic timeline and strength, the fitted model is
𝑓 𝑡 = 2.1864𝑙𝑛 (𝑡 ) − 0.7963

84

The 𝑅 2
score is calculated as 0.84476.
For the line between the time and strength, the fitted model is:
𝑓 𝑡 = 0.4766t + 0.2552
The 𝑅 2
score is calculated as 0.77241.

Figure 56. The graph on the left shows the regression line using logarithm time under 50 ℃, and
the graph on the right shows the regression line for the time under 50 ℃.
The regression model 𝑓 𝑡 = 𝐴𝑙𝑛 (𝑡 ) + 𝐵 shows a smaller error for the samples cured at 50 ℃
compared to those cured at 20 ℃. This result is consistent with the research conducted by
Metwally [51]. This result is considered reasonable, since the curing process accelerates at
higher temperatures and becomes closer to the long-term relationship between the strength and
age, as shown by previous research [51].

85

In conclusion, a linear relation between flexural strength and age is believed to be more
reasonable for samples maintained at 20 ℃. For samples maintained at 50 ℃, the solidification
process of cement material, where the CSH increases significantly, is significantly accelerated.
In this way, a regression model between the logarithm of time and strength is considered more
accurate.
7.3.2 Predictive model including the temperature
From the measurement results, it is observed that the curing temperature has a significant impact
on the strength of samples. The strength development will be accelerated, and the final strength
will be higher, if samples are maintained at a higher temperature. To help predict the flexural
strength change at different temperatures, predictive models that include temperature have been
developed.
Based on section 8.3.1, it has been observed that the strength of samples has a linear relation to
either time or the logarithmic timeline. It is assumed that the relationship between temperature
and strength change is also linear and, with this belief, three models have been studied. The
results are shown in Table 8.

86

Table 8. The fitted models for the linear relationship between flexural strength and temperature
and age (t indicates time, and T indicates temperature).
Model 𝑅 2
Fitted model
𝑓 𝑡 = 𝐴𝑡 + 𝐵𝑇 + 𝐶 0.710 𝑓 𝑡 = 0.1994𝑡 + 0.0348𝑇 − 0.2820
𝑓 𝑡 = 𝐴𝑡 + 𝐵𝑙𝑜𝑔 (𝑇 ) + 𝐶 0.710 𝑓 𝑡 = 0.1994t + 1.1385𝑙𝑜𝑔 (𝑇 ) − 2.9972
𝑓 𝑡 = 𝐴𝑙𝑜𝑔 (𝑡 ) + 𝐵𝑙𝑜𝑔 (𝑇 ) + 𝐶 0.738 𝑓 𝑡 = 0.9926𝑙𝑜𝑔 (t) + 1.1394𝑙𝑜𝑔 (𝑇 ) − 3.5518
From the relatively low 𝑅 2
score, we can see that all three fitted regression models have limited
performance. The reasons for this are considered to be as follows:
a. The impact of temperature is very influential on the slope rate. The linear parameter of
time (parameter A in the model) will change significantly with the change of
temperature.
b. As discussed previously, the linear relationship between time and strength cannot be
assured under different temperatures.
With the hope of solving these problems, a linear model based on logistic regression has been
developed and applied. The model is described below:
𝑓 𝑡 = (𝐴 (
1
1+𝑒 −(𝑇 −35)
) 𝑡 + 𝐵 (1 −
1
1+𝑒 −(𝑇 −35)
) log(𝑡 )) + 𝐶 ∗ log (T) + 𝐷 Equation 9

87

In the new model, the relationship between time and strength are divided into two parts: one part
describes the relationship between strength and time, and the second part describes the
relationship between strength and the logarithmic timeline. As discussed in section 8.3.1, the
logic behind this model is that the linear relationship between strength and time is more
significant at low temperatures, while the linear relationship between strength and the logarithm
of time is more reasonable at high temperatures. The new function, which includes these two
relationships at the same time, is believed to perform better.
After the regression calculation, the fitted equation is:
𝑓 𝑡 = ( 0.6575(
1
1 + 𝑒 −(𝑇 −35)
) log (𝑡 ) + 0.2099 (1 −
1
1 + 𝑒 −(𝑇 −35)
) 𝑡 ) + 1.2289log(T) − 3.3912
With a high 𝑅 2
score (0.858from the result), the model is considered reliable enough to show
the relationship between strength and time at different temperatures. As a result, the statistical
predictive model has been successfully developed for the relationship between the flexural
strength of samples and the age at different temperatures.
7.4 Determination of Best Separation Time
As mentioned earlier in this chapter, the information of two time nodes are needed for improving
the efficiency of SSS. For the first time node, we need to know the best time to separate the
desired part from its surrounding. From the data of previous experiments, typically a part that

88

can tolerate 50N load (0.75MPa as flexural strength) can be considered safe to be separated from
its surrounding material. From the fitted model, it is calculated that the sample can reach this
strength 9 hours, if it is cured under 50 ℃ and in 53 hours if it is maintained under 20 ℃.
In some applications, like road pavement, a flexural strength of 2.50 MPa is considered sufficient
for the lower boundary of requirement. By applying the second model, it is calculated that 11
days of maintenance at 20 ℃ and 5 days of maintenance at 50 ℃ is required for samples to be
utilized in these situations. Generally speaking, two weeks of maintenance of fabricated parts
will give the part the strength that will satisfy the requirements for commercial utilization.
It is believed that with new data included, the current predictive model will be more accurate, if
the strength of older samples are tested. At present, the regression model developed is considered
satisfactory for optimization of the SSS process.

89

8. Potential of SSS on Large-scale and Complicated Shape
Fabrication
In the previous chapters the capabilities and advantages of SSS in the fabrication of samples with
different materials and of certain sizes were presented. SSS is also believed to have great
potential for large part fabrication in the fields of construction and public art, and it can be
utilized to build 3D structures because it is a layer-based process. In this chapter the speed of
SSS will be studied and compared to other commercial 3D printers to show its potential in the
fabrication of large-scale products, and an extension of current SSS capabilities for building true
3D shapes is introduced.
8.1 Scale Up SSS for Large-Scale Fabrication
The main challenge for extending small-scale fabrication technologies to large-scale fabrication
is the slow building speed. In this section, the building speed of SSS at different layer
thicknesses is discussed and compared to Contour Crafting and D-shape to show its potential for
large-scale fabrication.
As discussed above, the current printing head gives SSS the capability to print high-quality parts
with a layer thickness of 12.5mm. Theoretically, the layer thickness of SSS can be increased with
no limit. To show the capability of SSS with increased layer thickness a printing head with a
25.5mm slot has been developed, as shown in Figure 57.

90

Figure 57. The upper printing head has a slot of 25.5mm high, and the lower printing head has a
slot of 12.5mm high.
As shown in Figure 57, the diameter (OD) of the upper and bottom needle are both 1.2mm. For
the upper needle, a slot of 25.5mm long and 0.7mm wide was cut. For the lower needle, the slot
is 12.5mm long and 0.7mm wide.
8.1.1 Printing Speed of SSS under Different Layer Thicknesses
To ensure a smooth deposition flow, vibration and air pump settings are adjusted to achieve the
maximum separation powder deportation rate (𝑅 𝑚𝑎𝑥
) for different layer thicknesses. Several

91

experiments were also conducted to analyze the maximum printing speed (Equation 10) of
different nozzles.
𝑉 𝑚𝑎𝑥
=
𝑅 𝑚𝑎𝑥 𝜌 ∗𝐿 ∗𝐷 Equation 10
The parameters are explained below:
𝑉 𝑚𝑎𝑥
: Maximum printing head moving speed (mm/s)
𝑅 𝑚𝑎𝑥
: Maximum deposition rate (g/s)
𝜌 : Deposition of separation powder (g/𝑚𝑚
3
)
L: Layer thickness (mm)
D: Diameter of nozzle (mm)
The first experiment tested the maximum moving speed of the head with the 12.5mm slot.
During the experiment a 12mm layer of cement is paved in the building tank and then the nozzle
is inserted into the cement to start depositing the separation powder. Different moving speeds
were tested and it was found that the maximum moving speed of the head with the 12.5mm slot
is 18mm/s. At the moving speed of 18mm/s the nozzle can form a 1.2mm tall separation gap
filled with sufficient amount of separation powder, as shown on the right of Figure 58.

92

Figure 58. The gap formed by the nozzle with a 12.5mm slot a different moving speeds.
For the nozzle with the 25.5mm slot several moving speeds were also tested, and the maximum
moving speed was found to be 12mm/s. Results under various moving speeds are shown in
Figure 59.

Figure 59. The gaps formed by the nozzle with a 25.5mm slot under different moving speeds.
Based on Table 9, fabrication speed is not necessarily reduced by half as the layer thickness is
doubled. This is because the maximum deposition rate can be increased by adjusting the
vibration and air pump settings. The printing speed will be even faster if we consider large-scale

93

printing, since a larger printing head will be applied for fabrication and a higher deposition rate
can be achieved accordingly. In conclusion, the printing speed limitation can be avoided for
large-scale fabrication by SSS
Table 9. The maximum printing speed of nozzles with different opening sizes.
Index Opening height Layer thickness Maximum speed
1 12.5mm 12mm 18mm/s
2 25.5mm 24mm 12mm/s

Figure 60. The image of a sample built with the layer thickness of 22mm.
A 22mm thick sample, shown in Figure 60, is fabricated with the 25.5mm head to show the
surface quality of samples built with greater layer thickness. The surface of the sample is

94

smooth, which means the deposition of separation powder is sufficient, as discussed in Chapter
6. Comparing the sample shown in Figure 60 and samples shown in Chapter 5, scaling up the
layer thickness of the SSS process will not negatively affect the quality of the samples.
Furthermore, it can be concluded that the SSS process can be scaled up for large-scale
fabrication because the possible limitations of fabrication speed and surface qualities can be
easily overcome as discussed above.
8.1.2 Comparison of SSS and Commercialized Mega-scale 3D-printing Technologies
To show the capability of SSS in commercial large-scale fabrication the building speed
(Equation 11) of the current SSS machine and the commercial large-scale additive manufacturing
machines are compared.
𝑇 𝑚𝑖𝑛 = ∑ (
𝑝 𝑖 𝑉 𝑚𝑎𝑥 + 𝑤 𝑖 )
𝑛 𝑖 =1
Equation 11
Where:
𝑇 𝑚𝑖𝑛 : Total time needed for fabrication (s)
n: The total number of layers
𝑝 𝑖 : The length of the trajectory that printing heads need to traverse in layer i (mm)
𝑉 𝑚𝑎𝑥
: Maximum printing head movement speed (mm/s)

95

𝑤 𝑖 : Pavement time (s) in layer i for powder-based printing
Contour Crafting [53] and D-shape [54] are the two commercialized fabrication technologies
with the capability for large-scale fabrication. In the Contour Crafting process the waiting time
(𝑤 𝑖 ) between layers is negligible because no pavement is needed. For SSS and D-shape there is
waiting time, but it is much shorter than the separation powder deposition time (
𝑝 𝑖 𝑉 𝑚𝑎𝑥 in Equation
11), especially when the building pattern is significantly larger for large-scale fabrication. In this
study, only the separation powder deposition time is considered. All fabrication parameters of
the three additive technologies are compared in Table 10.
Table 10. The fabrication parameters of mega-scale 3D printing technologies.
Name Nozzle movement speed Layer thickness
Contour Crafting [53] 150mm/s 25-130mm
D-Shape [54] 100mm/s 5mm
SSS 12-18mm/s No limitation (e.g., 12mm, 24mm)
Based on Table 10 the data indicates that the fabrication speed of SSS is at the same level of D-
shape. The nozzle moving speed of SSS is 20% of D-shape, while the layer thickness of SSS is at
least 500% of D-shape. For now, the printing speed of SSS is relatively slower compared to

96

Contour Crafting, while the printing speed of the SSS process can be significantly increased
when a larger nozzle, with a wider slot and hence higher flow rate of the separation powder, is
used for larger scale building tasks.
In conclusion, the current SSS technology has the capability to complete large-scale fabrication
jobs within a reasonable timeframe compared to current commercialized technologies; SSS has a
huge potential if a larger nozzle and vector printing is used.
8.2 Fabrication with Multi-layer Thickness
As discussed above, it has been shown that SSS can be utilized to fabricate parts with smooth
surfaces within a limited time. The extension of the current SSS machine to fabricate complex
shapes is another important topic. The challenge is that the detailed structure of complex shapes
may be lost due to the large layer thickness of SSS. Figure 61 shows that the information of the
area within a grid will be lost if a fixed layer thickness is used with the vertical needle.

97

Figure 61. Slicing strategy under a 12mm fixed layer thickness.
Two solutions are considered feasible for this challenge. The first solution is the use of a
machine with multi-degree of freedom movement capability. For example, the printing head can
be installed at the end of a 6-axis robotic manipulator. The second solution will be using an
adaptive slicing strategy to choose the proper layer thicknesses for different geometric
information.
Complicated three-dimensional shapes can be fabricated using a combination of small and large
layer thicknesses. Slicing procedures for adaptive layer thickness have been discussed in

98

previous studies in the Additive Manufacturing field, and they are considered suitable for the
SSS process. Many slicing strategies are based on Dolenc’s cusp height [55].
With the measurement of resolution based on cusp height, Kulkarni [56], Yan [57, 58] and Mao
[59] have developed adaptive slicing strategies to optimize the surface smoothness of samples by
using different layer thicknesses. Their strategies can be utilized for slicing process of SSS.
Adaptive slicing strategies can be used to generate layer patterns for SSS when different layer
thicknesses are required for complicated samples.
To show the capability of SSS to fabricate with different layer thicknesses, the sample shown in
Figure 62 is built by the 12.5mm nozzle. The lower part of the sample is fabricated with a layer
thickness of 10mm, and the upper part is built with a layer thickness of 0.25mm.

Figure 62. The sample built with different layer thicknesses.

99

From the results, we can see that the thick layer of the sample is smooth and that the upper
pyramidal shape shows an accurate geometric dimension. It can therefore be claimed that SSS
technology has the capability to fabricate true three-dimensional parts using an adaptive slicing
strategies.

100

9. Conclusion and Future Research
9.1. Research Summary
This research started with the introduction of a modified SSS process to implement large
scale products fabrication. Specimens of Portland cement (lime based), Sorel cement (magnesia
based) and gypsum based composite were fabricated with fair surface qualities by SSS. Base on
theses preliminary experiments, Portland cement is chosen as the focus of this dissertation and
three topics were discussed.
The first topic focused on the surface qualities of specimens fabricated. After several
experiments, soda lime was chosen as the separation powder for SSS because it had outstanding
chemical and physical characteristics as the separation powder. The effort was then followed by
a 2
3
full factorial design of experiments, where a functional combination of vibration and air flow
was achieved to ensure a sufficient soda lime deposition. A Nucleus technology based simulation
was implemented successfully to simulate the influence of vibration and air pressure during the
powder deposition process. Furthermore, the water spray speed and extra support sand layers (for
the bottom and upper surface) are studied to ensure a good surface qualities of specimens. By
using the standard arithmetical mean deviation, a water spray speed of 30 g/min was chosen to
achieve the lowest roughness. A 2mm layer of sand was also paved on the top of the whole base
material before water spraying to increase the upper surface quality of samples fabricated.

101

The second topic was on the mechanical properties of samples fabricated. A water cement
ratio of 0.4 and a water maintenance process under 50℃ were chosen after experiments, to give a
fair flexural strength (2.97 MPa for 6 days) of specimens comparing to samples built with
conventional method. A statistical predictive model was developed successfully to provide
guidance on the separation time decision of desired part from green part after the curing process
under different temperatures.
Under the third topic, the building speeds of SSS with 12.5mm and 25.5mm opening
printing head were compared to Contour Crafting and D-shape to show the capability of SSS on
large scale product fabrication. A sample was also fabricated by applying different layer
thicknesses (10mm and 0.5mm). The research was successful in proving the potential of SSS on
large scale product and complex three dimensional shapes fabrication.
9.2. Recommendations for Future Work
AM, especially AM for large scale fabrication, is a new topic and needs a lot of explorations. In
this section, several meaningful topics are suggested for further research.
9.2.1 Fabrication with more materials
In this dissertation, samples fabricated with Sorel cement (magnesia based) and gypsum are
presented in the preliminary results chapter but not discussed in detail. These materials also have
wide applications in real industry, many challenges exist during the fabrication process. Sorel
cement is very sensitive to the S-powder material choice, and gypsum has a relatively high

102

shrinkage ratio. The further studies on the methods to improve the surface qualities and strength
of samples built with these materials are needed.
9.2.2. Large scale device development
Although the capability of SSS to fabricate large scale products are proved by showing its fast
building speed, a device with a larger size is needed to show the advantages of SSS on the
fabrication of mega scale constructions. For now, the printing head used has a diameter of
1.2mm and the opening of 25.5mm. In the next research, a printing head with a 10mm or large
diameter can be built, and a gantry 3 axes system with meters wide and high can be developed.
With the large device, the surface quality and strength of samples with meters large sizes can be
studied and optimized.
9.2.3. Mutli-degree of Freedom Movement Device Application
As shown in Chapter 9, the fabrication of complex 3D shapes was implemented by using the
adaptive slicing strategy. A more efficient solution may be adding the printing head on the end of
devices with multi-degree of freedom movement capability, like a Stewart platform or a robotic
manipulator.

103

Figure 63. The left image shows a printing head on the end of a 6 axes movement device; the right
image shows a printing head installed on the end of a KUKA robotic manipulator. [60]
In the future research, these flexible platforms may be utilized for the SSS. With better
movement flexibility, SSS will show its advantages in fabricating large scale constructions or
products in less time. A new slicing strategy will be needed accordingly, where a combination of
subtractive and additive manufacturing path planning is considered with great potential.

104

Bibliography
[1] Gibson, I., Rosen, D. W., and Stucker, B., 2010. Additive manufacturing technologies: rapid
prototyping to direct digital manufacturing. Springer.
[2] Campbell, T., Williams, C., Ivanova, O., and Garrett, B., 2011. Could 3d printing change the
world? technologies, potential, and implications of additive manufacturing. Washington: Atlantic
Council of the United States, 5, p. 2012.
[3] Leong, K. F., C. M. Cheah, and C. K. Chua. Solid freeform fabrication of three-dimensional
scaffolds for engineering replacement tissues and organs. Biomaterials 24.13 (2003): 2363-2378.
[4] Chia, Helena N., and Benjamin M. Wu. Recent advances in 3D printing of biomaterials.
Journal of biological engineering 9.1 (2015): 1.
[5] B. Khoshnevis, Automated construction by contour crafting - Related robotics and
information technologies, Autom. Constr., vol. 13, pp. 5–19, 2004.
[6] B. Khoshnevis, M. Thangavelu, X. Yuan, and J. Zhang, Advances in Contour Crafting
technology for extraterrestrial settlement infrastructure buildup, AIAA SPACE 2013.

105

[7] B. Khoshnevis, X. Yuan, B. Zahiri, and B. Xia, Construction by Contour Crafting using
sulfur concrete with planetary applications. Conference and Exposition, 2013.
[8] Khoshnevis, Behrokh, Bahram Asiabanpour, Mehdi Mojdeh, and Kurt Palmer, SIS-a new
SFF method based on powder sintering. Rapid Prototyping Journal9, no. 1 (2003): 30-36.
[9] Zhang, J. and B. Khoshnevis, Selective Separation Sintering (SSS) – A new additive
manufacturing approach for metals and ceramics, Solid Free Form Fabrication Symposium,
Austin, Texas 2015.
[10] New 3-D printing process could lead to construction on Mars and the moon
[Online].Available: https://news.usc.edu/97707/new-3-d-printing-process-could-lead-to-
construction-on-mars-and-the-moon/
[10] Stratasys F900 overview [Online]. Available: http://www.stratasys.com/3d-
printers/production-series/fortus-900mc
[11] S-max Tech Specs [Online]. Available: https://www.exone.com/Systems/Production-
Printers/-S-Max
[12] Voxeljet vx4000 [Online]. Available: http://www.voxeljet.de/en/systems/vx4000/

106

[13] Christopher Holshouser, Clint Newell, Sid Palas, Chad duty out of bounds additive
manufacturing, Advanced Materials & Processes , March 2013.
[14] BAAM overview [Online]. Available: http://www.e-ci.com/baam/
[15] Introducing Contour Crafting technology [Online]. Available:
http://www.contourcrafting.org/
[16] D shape [Online]. Available: http://d-shape.com/
[17] Ena Lloret , Amir R. Shahab, Complex concrete structures merging existing casting
techniques with digital fabrication, Computer-Aided Design 60 (2015) 40–49.
[18] Stone Spray Robot by Anna Kulik, Inder Shergill and Petr Novikov [Online]. Available:
https://www.dezeen.com/2012/08/22/stone-spray-robot-by-anna-kulik-inder-shergill-and-petr-
novikov/
[19] B. Dutta* S. Palaniswamy, J.Choi, et al. Additive manufacturing by Direct Metal
Deposition. Advanced Materials& Processes, May 2011.
[20] Nouri, H., and B. Khoshnevis. Study on inhibition mechanism of polymer parts in Selective
Inhibition Sintering process.

107

[21] Khoshnevis, B. and J. Zhang, Selective Separations Sintering (SSS) – An additive
manufacturing approach for fabrication of ceramic and metallic parts with applications in
planetary construction, American Institute of Aeronautics and Astronautics (AIAA) Conference,
Pasadena, CA Sep 2015.
[22] F.W. Locher, Setting of cement part I: Reaction and development of structure, Zement-
Kalk-Gips 29(10), 435-442, 1976.
[23] Edwin S. Newman, A study of the system magnesium Oxide-Magnesium Chloride-Water
and the heat of formation of Magnesium Oxychloride”, Journal of Research of the National
Bureau of Standards Vol. 54, No. 6, June 1955.
[24] Broomfield, John P. Corrosion of steel in concrete: understanding, investigation and repair.
CRC Press, 2002.
[25] Kruth, Jean-Pierre, et al, Binding mechanisms in selective laser sintering and selective laser
melting. Rapid Prototyping Journal 11.1 (2005): 26-36.
[26] American Society for Testing and Materials (ASTM), ASTM International, [Online].
Available: https://www.astm.org/. [Accessed 2016]
[27] Zhang, Jing. Phd Dissertation, University of Southern California

108

[28] Brennen, Christopher Earls (2005), Fundamentals of multiphase flow, Cambridge
University Press. ISBN 13, Chapter 13.
[29] Create nParticles [Online]. Available: https://knowledge.autodesk.com/support/maya/learn-
explore/caas/CloudHelp/cloudhelp/2018/ENU/Maya-SimulationEffects/files/GUID-87BE424C-
46D1-4245-9546-86C44797B104-htm.html
[30] nParticle Workflows [Online]. Available:
https://knowledge.autodesk.com/support/maya/learn-
explore/caas/CloudHelp/cloudhelp/2018/ENU/Maya-SimulationEffects/files/GUID-AEBE18A1-
C2C8-4830-B600-8C8690979A35-htm.html
[31] K Rietema, The dynamics of fine powders, Springer 1991.
[32] Nucleus introduction [Online]. Available: https://autodeskresearch.com/projects/nucleus
[33] nDynamic collisions [Online]. Available:
https://knowledge.autodesk.com/support/maya/learn-
explore/caas/CloudHelp/cloudhelp/2018/ENU/Maya-SimulationEffects/files/GUID-
BABBD1CE-C82C-4EDB-BBF6-4C0EBC854DAA-htm.html.

109

[34] Richard Holdich, Fundamentals of particle technology. Midland Information Technology
and Publishing, Chapter 10
[35] Degarmo, E. Paul; Black, J.; Kohser, Ronald A. (2003), Materials and processes in
manufacturing (9th ed.), Wiley, p. 223, ISBN 0-471-65653-4.
[36] Maryam Pakpour (2012), How to construct the perfect sandcastle, Scientific Report.
[37] CalPortland Company, N.p., n.d. Web. 15 July 2016.
[38] Courland, Robert (2011). Concrete planet: the strange and fascinating story of the world's
most common man-made material, Amherst, N.Y.: Prometheus Books. ISBN 978-1616144814.
[39] Wenke Yang, The issues and discussion of modern concrete science, Springer Press.
[40] Michael Thomas, Optimizing the use of fly ash in concrete, Portland Cement Association.
[41] Michael Thomas, Optimizing the use of fly ash in concrete, Portland Cement Association.
[42] Jens Bauera, High-strength cellular ceramic composites with 3D microarchitecture, PNAS |
February 18, 2014, vol. 111, no. 7, 2453–2458.

110

[43] Weiner S, Wagner HD (1998) The material bone: Structure-mechanical function relations,
Annu Rev Mater Sci 28:271–298.
[44] Spatz H-C, Köhler L, Speck T (1998), Biomechanics and functional anatomy of hollow-
stemmed sphenopsids. I. Equisetum giganteum (Equisetaceae) Am J Bot 85(3):305–314.
[45] Deshpande VS, Ashby MF, Fleck NA (2001), Foam topology bending versus stretching
dominated architectures, Acta Mater 49(6):1035–1040.
[46] M.S. Shetty, Concrete technology theory and practice, S.Chand & Company Ltd., New
Delhi, 2006 (Chapter 7).
[47] ECP 203 Egyptian code for designing and executing the building works, 2nd ed., HBRC,
Egypt, 2009.
[48] C. Yeh, Generalization of strength versus water–cementitious ratio relationship to age,
Cement and Concrete Research 36 (2006) 1865–1873.
[49] Ming Xia, Jay G. Sanjayan, Methods of enhancing strength of geopolymer produced from
powder-based 3D printing process, Material Letters 227 (2018) 281 – 283.
[50] Jay Devore, Probability and statistics for engineering and sciences (8th edition).

111

[51] Metwally abd allah Abd elaty, Compressive strength prediction of Portland
cement concrete with age using a new model, Housing and Building National Research Center,
(2014) 10, 145–155.
[52] S.-H. Han, J.-K. Kim, Effect of temperature and age on the relationship between dynamic
and static elastic modulus of concrete, Cement and concrete research 34 (7) (2004) 1219–1227.
[53] D. Hwang and B. Khoshnevis, Concrete wall fabrication by Contour Crafting, Dooil Hwang
and Behrokh Khoshnevis, 2004.
[54] G. Cesaretti, E. Dini, X. De Kestelier, V. Colla, and L. Pambaguian, Acta Astronautica,
Building components for an outpost on the Lunar soil by means of a novel 3D printing
technology, Acta Astronaut, vol. 93, pp. 430–450, 2014.
[55] Dolenc A, Mäkelä I., Slicing procedures for layered manufacturing techniques,
Comput Aided Des 1994;26(2):119–26.
[56] Kulkarni P, Dutta D. An accurate slicing procedure for layered manufacturing,
Comput Aided Des 1996;28(9):683–97.
[57] Yang P, Qian X. Adaptive slicing of moving least squares surfaces: toward direct
manufacturing of point set surfaces, J Comput Inf Sci Eng 2008;8(3):031003–11.

112

[58] Yang P, Li K, Qian X. Topologically enhanced slicing of MLS surfaces, J Comput
Inf Sci Eng 2011;11(3):031003–9.
[59] H Mao, TH Kwok, Y Chen, CCL Wang. Adaptive slicing based on efficient profile analysis,
Computer-Aided Design 107, 89-101.
[60] Kinematic Calibration of Six-Axis Robots [Online]. Available:
https://www.researchgate.net/figure/KUKA-KR-15-2-DH-Parameters_fig3_270887659

Abstract (if available)

Abstract Additive manufacturing (AM) or 3D printing has been applied for rapid prototyping and direct manufacturing for years. AM has shown it capability in fabricating complex shapes without extra tooling and fixture [1, 2]. However, application of AM in fabrication of large scale products has been a challenge due to low fabrication speed. This work introduces a new high speed fabrication process which is especially suitable for fabrication of large scale objects. ❧ Selective Separation Shaping is a new additive manufacturing technique which is capable of processing polymeric, metallic, ceramic and composite materials such as cementitious construction materials. In earlier research the capabilities of SSS in making metallic and ceramic parts have been demonstrated. The focus of the research reported in this dissertation is on expansion of capabilities of SSS for efficient creation of large scale cementitious composite parts. ❧ A prototype machine has been used to create specimens made of regular Portland cement (lime based), Sorel cement (magnesia based) and gypsum based composite, as shown in Chapter 6. The fabrication result and process characteristics based on several experiments are presented. The factors that impact the surface quality of samples are also discussed in Chapter 6. The interaction between the two powder materials (i.e., Separation powder and Base material) and the separation powder deposition process are simulated and optimized to ensure an adequate powder flow. The mechanical behavior of cementitious samples built by SSS and the factors affecting it are also presented in Chapter 7. In Chapter 8 statistical predictive models are also developed and discussed to provide information about the samples curing characteristics under different temperatures. Finally, the potential of the current SSS on the fabrication of large scale objects and complicated three dimensional parts are presented and discussed in Chapter 9.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Multi-scale biomimetic structure fabrication based on immersed surface accumulation

PDF

Selective separation shaping: an additive manufacturing method for metals and ceramics

PDF

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

PDF

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

PDF

Metallic part fabrication wiht selective inhibition sintering (SIS) based on microscopic mechanical inhibition

PDF

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

PDF

Contour crafting construction with sulfur concrete

PDF

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

PDF

Analyses of strength of layered structures fabricated by Contour Crafting

PDF

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

PDF

Scalable polymerization additive manufacturing: principle and optimization

PDF

Hybrid vat photopolymerization: methods and systems

PDF

Experimental and numerical techniques to characterize structural properties of fresh concrete

PDF

Fabrication-aware machine learning for accuracy control in additive manufacturing

PDF

Deformation control for mask image projection based stereolithography process

PDF

Contour Crafting process planning and optimization

PDF

Machine learning-driven deformation prediction and compensation for additive manufacturing

PDF

Some scale-up methodologies for advanced manufacturing

PDF

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

Asset Metadata

Creator Gao, Xiang (author)

Core Title Selective Separation Shaping (SSS): large scale cementitious fabrication potentials

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Industrial and Systems Engineering

Publication Date 02/25/2019

Defense Date 01/24/2019

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

Tag additive manufacturing,cementitious material,large scale,OAI-PMH Harvest,statistical model,strength analysis,surface quality

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Khoshnevis, Behrokh (committee chair), Chen, Yong (committee member), Shiflett, Geoffrey (committee member)

Creator Email ethanxiangg@gmail.com,xiangg@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-126763

Unique identifier UC11675671

Identifier etd-GaoXiang-7111.pdf (filename),usctheses-c89-126763 (legacy record id)

Legacy Identifier etd-GaoXiang-7111.pdf

Dmrecord 126763

Document Type Dissertation

Format application/pdf (imt)

Rights Gao, Xiang

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