part design and topology optimization for rapid sand …
TRANSCRIPT
The Pennsylvania State University
The Graduate School
Engineering Design, Technology, and Professional Programs
PART DESIGN AND TOPOLOGY OPTIMIZATION FOR RAPID SAND AND
INVESTMENT CASTING
A Thesis
in
Engineering Design
by
Jiayi Wang
2017 Jiayi Wang
Submitted in Partial Fulfillment
of the Requirements
for the Degree of
Master of Science
December 2017
The thesis of Jiayi Wang was reviewed and approved* by the following:
Guha P. Manogharan
Assistant Professor of Mechanical and Nuclear Engineering
Thesis Co-Advisor
Timothy W. Simpson
Paul Morrow Professor of Engineering Design and Manufacturing
Thesis Co-Advisor
Sven G. Bilén
Professor of Engineering Design, Electrical Engineering, and Aerospace
Engineering Head of School of Engineering Design, Technology, and Professional Programs
*Signatures are on file in the Graduate School
iii
ABSTRACT
The integration of additive manufacturing into traditional metal casting provides a wide
range of rapid casting solutions. One important motive for rapid casting is additive
manufacturing’s ability to create highly complex objects without any fixture or tooling
requirements. Such advantages provide great design freedom for the geometry of cast metal parts.
The objective in this thesis is to explore the part design opportunistic and restrictions of two rapid
casting processes: (1) sand casting with 3D Sand Printing-fabricated molds and (2) investment
casting using material extrusion–fabricated wax-like patterns. Knowledge-based design
guidelines are developed for both of these rapid casting processes through novel integration of
topology optimization with design for casting and design for AM principles. For each process, a
case study is conducted in which a mechanical metal benchmark is topologically optimized and
redesigned following the proposed design rules. The redesigned parts are successfully cast and
demonstrate improvements in mechanical performance and weight reduction.
Keywords: additive manufacturing, 3D Sand Printing, material extrusion, topology
optimization, rapid casting, sand casting, investment casting, cast part design
iv
TABLE OF CONTENTS
LIST OF FIGURES ................................................................................................................. vi
LIST OF TABLES ................................................................................................................... ix
ABBREVIATIONS ................................................................................................................. x
ACKNOWLEDGMENTS ....................................................................................................... xii
Chapter 1 INTRODUCTION AND LITERATURE REVIEW .............................................. 1
1.1. Introduction to Additive Manufacturing ................................................................... 1 1.2. Introduction to Topology Optimization ................................................................... 6 1.3. Introduction to Traditional Metal Casting ................................................................ 10 1.4. Part Design for Traditional Metal Casting ............................................................... 13 1.5. Rapid Casting ........................................................................................................... 20 1.6. Thesis Overview ....................................................................................................... 22
Chapter 2 REDESIGN OF TRADITIONAL METAL PARTS FOR 3D SAND
PRINTING USING TOPOLOGY OPTIMIZATION ...................................................... 24
2.1. Introduction .............................................................................................................. 24 2.2. Literature Review ..................................................................................................... 28 2.3. Methodology ............................................................................................................ 31
2.3.1. Step 1: Prepare the 3D CAD model for a metal part ..................................... 33 2.3.2. Step 2: Perform TO for the initial CAD model ............................................. 33 2.3.3. Step 3: Refinement and revision of the optimization output ......................... 34 2.3.5. Step 5: Perform FEA on the revised design .................................................. 42 2.3.6. Step 6: Analyze FEA results ......................................................................... 42 2.3.7. Step 7: Finalize the optimized geometry ....................................................... 42
2.4. Case Study ................................................................................................................ 43 2.4.1. TO and FEA validation ................................................................................. 43 2.4.2. Rigging design and casting ........................................................................... 46 2.4.3. Mechanical testing ......................................................................................... 48
2.5. Discussion ................................................................................................................ 49 2.6. Future Work ............................................................................................................. 50
Chapter 3 REDESIGN AND TOPOLOGY OPTIMIZATION OF TRADITIONAL
METAL PARTS FOR INVESTMENT CASTING WITH 3D-PRINTED WAX
PATTERNS ..................................................................................................................... 51
3.1. Introduction .............................................................................................................. 51 3.2. Literature Review ..................................................................................................... 55 3.3. Methodology ............................................................................................................ 57
3.3.1. Step 1: TO of the initial CAD ....................................................................... 58 3.3.2. Step 2: Redesign of the optimization output ................................................. 59
v
3.3.3. Step 3: FEA validation .................................................................................. 67 3.3.4. Step 4: Patterning of the optimized design .................................................... 68
3.4. Case Study ................................................................................................................ 68 3.4.1. TO and FEA validation ................................................................................. 68 3.4.2. Gating design and casting.............................................................................. 71
3.5. Discussion ................................................................................................................ 75 3.6. Future Work ............................................................................................................. 76
Chapter 4 CONCLUSIONS AND FUTURE WORK ............................................................. 78
4.1. Conclusions .............................................................................................................. 78 4.2. Future Work ............................................................................................................. 80
REFERENCES ........................................................................................................................ 82
vi
LIST OF FIGURES
Figure 1-1: Industries served and approximate revenues for additive manufacturing, 2016
[6]. .................................................................................................................................... 2
Figure 1-2: Material extrusion AM process principle and schematic [10]. ............................ 3
Figure 1-3: Binder jetting AM process principle and schematic [13]. .................................... 4
Figure 1-4: Sizing, shape, and topology optimization [22]. .................................................... 6
Figure 1-5: The influence of penalization factor p on topology optimization results [27]. .... 9
Figure 1-6: The general procedure of topology optimization based on SIMP algorithm
[31]. .................................................................................................................................. 9
Figure 1-7: Sand mold assembly [34]. .................................................................................... 11
Figure 1-8: Traditional sand casting process [35]. .................................................................. 12
Figure 1-9: Schematic of the traditional investment casting process [41]. ............................. 13
Figure 1-10: Corner and edge on cast part: (a) bad design; (b) improved design [45]. .......... 15
Figure 1-11: Section change on cast part: (a) bad design; (b) and (c) improved designs
[46]. .................................................................................................................................. 15
Figure 1-12: Intersection on cast part: (a) bad design; (b) and (c) improved designs [45]. .... 16
Figure 1-13: Pattern designed (a) without draft and (b) with draft [45]. ................................ 18
Figure 1-14: (a) Pattern with undercut; (b) and (c) improved designs [45]. ........................... 19
Figure 1-15: Rapid sand casting solutions: approaches and corresponding AM processes
[49]. .................................................................................................................................. 21
Figure 1-16: Rapid investment casting solutions: approaches and corresponding AM
processes [49]. .................................................................................................................. 21
Figure 2-1: The 3DSP process via binder jetting [56] ............................................................ 26
Figure 2-2: Two design stages of a 3DSP-based casting process: (a) cast part design; (b)
mold design. ..................................................................................................................... 28
Figure 2-3: Design flowchart for sand casting with 3D-printed molds. ................................. 32
Figure 2-4: Topology optimization setup and result of a triangular bracket. .......................... 34
vii
Figure 2-5: Refinement and revision using Inspire: (a) a coarse truss on a TO output
geometry; (b) refined truss structure 1; (c) refined truss structure 2. ............................... 34
Figure 2-6: A metal pipe nipple not able to cast traditionally due to undercuts. .................... 36
Figure 2-7: Redesign of a housing for 3DSP-based casting: (a) current cast part with
draft; (b) redesign. ............................................................................................................ 37
Figure 2-8: Redesign of a U-bracket for 3DSP-based casting: (a) current cast part with
no pilot holes on the side surfaces; (b) redesign with pilot holes. ................................... 38
Figure 2-9: Redesign of a hinge bracket for 3DSP-based casting: (a) current cast part
with a heavy section; (b) redesign with uniform wall thickness. ..................................... 39
Figure 2-10: Redesign of a pipe elbow for 3DSP-based casting: (a) current cast part with
a core; (b) redesign with a core. ....................................................................................... 40
Figure 2-11: Redesign of a wheel frame for 3DSP-based casting: (a) current cast part
with a thick solid hub; (b) redesign with a center hole. ................................................... 41
Figure 2-12: Alcoa Aircraft Bracket Challenge [87]: (a) design envelope and (b)
boundary conditions. ........................................................................................................ 43
Figure 2-13: (a) Alcoa challenge original loading conditions [87]; (b) Assumed loading
direction in this case study. .............................................................................................. 44
Figure 2-14: (a) Bearing bracket design envelop; (b) TO output geometry; (c) optimized
design. .............................................................................................................................. 45
Figure 2-15: Check for geometry based on redesign rules. .................................................... 46
Figure 2-16: FEA of (a) the target design and (b) the optimized design. ............................... 46
Figure 2-17: The (a) rigging design and (b) solidification simulation for the optimized
bracket. ............................................................................................................................. 47
Figure 2-18: (a) 3D-printed sand mold; (b) the final cast part. ............................................... 48
Figure 2-19: (a) Fixtures designed for the mechanical testing; (b) setup of the tensile
testing machine. ............................................................................................................... 48
Figure 2-20: The response of the bracket in the mechanical testing. ...................................... 49
Figure 2-21: The failure location of the bracket in (a) FEA simulation and (b) the
mechanical testing. ........................................................................................................... 49
Figure 3-1: Schematic of the traditional investment casting process [41]. ............................. 53
Figure 3-2: Framework of pattern design for a RIC process. ................................................. 54
viii
Figure 3-3: Design flowchart for 3D-printed IC patterns. ...................................................... 58
Figure 3-4: Topology optimization of a cantilever beam........................................................ 59
Figure 3-5: Generation of a smooth solid truss from TO output using Inspire
PolyNURBS. .................................................................................................................... 60
Figure 3-6: Influence of extrusion temperature on the surface finish of a 3D-printed wax
pattern: (a) under proper temperature, (b) under insufficient temperature. ...................... 62
Figure 3-7: Testing of dimensional accuracy by 3D-printing a cube. ..................................... 63
Figure 3-8: A benchmark for testing 3D printing features such as minimum feature size,
minimum wall thickness, surface finish, bridging and overhangs: (a) 3D model, (b)
printed part. ...................................................................................................................... 64
Figure 3-9: Quality of prints (a) with proper layer height and (b) when the layer height is
too low.............................................................................................................................. 64
Figure 3-10: Testing of support for wax-like filament: (a) an arch structure printed with
support, (b) bottom surface after the support is removed. ............................................... 65
Figure 3-11: Defects of a cast part resulting from slurry infiltration. ..................................... 66
Figure 3-12: Influence of build direction to a 3D-printed truss. ............................................. 66
Figure 3-13: Curling of wax layers. ........................................................................................ 67
Figure 3-14: (a) The target design envelope and (b) load and boundary conditions of the
bearing bracket. ................................................................................................................ 69
Figure 3-15: (a) TO output geometries and (b) the final redesigns......................................... 70
Figure 3-16: Check for geometries based on redesign rules. .................................................. 71
Figure 3-17: Comparison of safety factor of the target design and redesigns: (a) 1.0, (b)
1.3, (c) 1.4 and (d) 1.2. ..................................................................................................... 71
Figure 3-18: Solidification simulation for (a) target with 1-mm fillet, (b) redesign 1, (c)
redesign 2 and (d) redesign 3. .......................................................................................... 72
Figure 3-19: The pattern design for bracket geometries: (a) target with 1 mm fillet, (b)
redesign 1, (c) redesign 2 and (d) redesign 3. .................................................................. 73
Figure 3-20: (a) 3D-printed wax-like patterns; (b) final cast carbon steel brackets after
post-processing. ................................................................................................................ 75
ix
LIST OF TABLES
Table 1-1: List of common commercial and educational TO software [32]. .......................... 10
Table 1-2: Minimum wall thickness of different metals for sand casting [44]. ...................... 14
Table 1-3: Patternmaker’s shrinkage of different cast metals [44]. ........................................ 16
Table 1-4: Machining allowances of different cast metals in sand casting [44]. .................... 17
Table 1-5: Machining allowances for cast parts in investment casting [47]. .......................... 18
Table 1-6: Recommended draft angles for patterns used in sand casting [44]. ...................... 19
Table 1-7: Geometric requirements for holes on patterns in investment casting [48]. ........... 20
Table 2-1: Important part design rules for traditional sand casting [45]. ................................ 29
Table 2-2: Improvements of the use of 3DSP on traditional sand casting rules. .................... 41
Table 2-3: Comparison of the performance of the target and the optimized designs. ............ 46
Table 2-4: Comparison of the UTS and displacement between the FEA and mechanical
testing results.................................................................................................................... 49
Table 3-1: Parameters for the material extrusion of wax-like patterns. .................................. 74
Table 3-2: Comparison of the performance of the target design and the optimized
designs. ............................................................................................................................. 75
x
ABBREVIATIONS
3DP 3D Printing
3DSP 3D Sand Printing
ABS Acrylonitrile Butadiene Styrene
AM Additive Manufacturing
ATOM Abaqus Topology Optimization Module
CAD Computer-Aided Design
CAE Computer-Aided Engineering
CNC Computer Numerically Controlled
DfAM Design for Additive Manufacturing
DMLS Direct Metal Laser Sintering
FDM Fused Deposition Modelling
FEA Finite Element Analysis
FEM Finite Element Method
IC Investment Casting
ISE Isotropic-Solid or Empty
LOM Laminated Object Manufacturing
MBB Messerschmitt-Bölkow-Blohm
MM II Model Maker II
MMA Method of Moving Asymptotes
xi
NURBS Non-Uniform Rational Basis Spline
OC Optimality Criteria
PLA Polylactic Acid
RC Rapid Casting
RIC Rapid Investment Casting
RP Rapid Prototyping
RTV Room Temperature Vulcanized
SIMP Solid Isotropic Material with Penalization
SLA Stereolithography Apparatus
SLS Selective Laser Sintering
STEP Standard for the Exchange of Product model data
STL Stereolithography
TO Topology Optimization
UNI MCC The University of Northern Iowa Metal Casting Center
UTS Ultimate Tensile Strength
xii
ACKNOWLEDGMENTS
First, I want to express my sincere gratitude to my advisors Dr. Guha Manogharan and
Dr. Timothy Simpson. I am grateful that I could be a student of Dr. Manogharan, a nice and
responsible advisor. Without him, I could not have been able to work on this project and pursue
my M.S. degree at Penn State. Many thanks to Dr. Timothy Simpson for his guidance, expertise,
understanding, and patience to me during my study.
I would also like to thank Dr. Sven Bilén for his kind help on the thesis writing, and for
his continued support on studying, working and using facilities of the Engineering Design
Program.
I want to thank Dr. Robert Voigt, Daniel Supko, and Travis Richner from FAME
(Factory for Advanced Manufacturing Education) lab. Their professional guidance helped me to
have a better understanding of casting and manufacturing processes.
In addition, I want to recognize Ellis from Engineering Services and Bill Genet from the
Learning Factory for providing expertise in machining and testing.
My thanks also go to all my fellow lab mates in SHAPE (Systems for Hybrid-Additive
Process Engineering) lab, for their valuable suggestions and encouragement. It has been a great
experience working with and learning from them.
I want to say thank you to my family, especially my wife Hui Wang. It was they who
gave me the strength to finish my degree. I could not imagine how I could be here without their
understanding and support.
This thesis work is funded by America Makes project: Additive Manufacturing for Metal
Casting (AM4MC). Finally, I would like to thank the University of Northern Iowa, Youngstown
State University, and Tech Cast, Inc. for their generous help in assistance with casting design and
operation.
1
Chapter 1
INTRODUCTION AND LITERATURE REVIEW
1.1. Introduction to Additive Manufacturing
Additive manufacturing (AM), commonly known as three-dimensional printing (3DP), is
the process of joining materials to make objects from 3D model data, usually layer by layer, as
opposed to traditional subtractive (e.g., cutting and shearing) and forming (e.g., stamping,
bending, and molding) manufacturing methodologies [1]. AM used to be called rapid prototyping
(RP). This term is used in a variety of industries to describe a process for rapidly creating
prototypes for final products that will eventually be realized using conventional manufacturing
methods [2]. With the improvement of AM in material diversity, accuracy, speed, and cost, the
application of AM has expanded from rapid prototyping to rapid tooling and rapid manufacturing,
in which AM-fabricated objects are used for tooling or as end-use products [3]. Due to the speed
and efficiency with which AM can produce prototypes and parts, this technology is believed to
have the greatest impact on products that require high customization, have complex designs, and
are made in small quantities [4].
The AM industry is growing fast. According to Wohlers Report 2016, the AM industry
has had a compound annual growth rate (CAGR) of 26.2% over the past 3 decades [5]. The AM
market has grown to a total of $5,165 billion in 2015. AM has received tremendous attention and
is being used in a wide range of industries. Leading sectors, based on revenue for all AM
products and services in 2016, include aerospace (18.2%), automotive (14.8%), consumer
products/electronics (12.8%) and medical/dental (11.0%) (see Figure 1-1).
2
Figure 1-1: Industries served and approximate revenues for additive manufacturing, 2016 [6].
To date, numerous AM processes have been developed differing in the materials they use
and the methods of patterning and fusing layers they employ [4]. The identified 7 major
categories of AM processes and their definitions are as follows [7]:
• Material extrusion: selective dispensing of material through a nozzle or orifice;
• Material jetting: selective deposition of droplets of build material via ink-jet
printhead nozzles;
• Binder jetting: selective deposition of a liquid bonding agent through ink-jet
printhead nozzles to join powder materials;
• Sheet lamination: a process involving bonding sheets of material together;
• Vat photopolymerization: selective curing of liquid photopolymer (light-sensitive
polymer) in a vat via light-activated polymerization;
• Powder bed fusion: selective fusing of powder bed regions via thermal energy; and
• Directed energy deposition: simultaneous fusion and deposition of material.
In this thesis, two types of AM processes are the focus: (1) binder jetting and (2) material
extrusion given their ability to additively manufacture sand and wax-like parts, respectively.
Figure 1-2 shows a schematic of the material extrusion process. The core of a material
extrusion AM system is the heated extruder into which the filament is pushed and melted. The
3
melted material is then extruded through a small nozzle as a road or bead, which then solidifies
after laying on the base. The extruder is moved through the build environment on a gantry via
stepper motors. The planar, x-y motion of the printhead combined with the z-motion of the build
stage allows 3D structures to be created in a layer-by-layer manner [8]. Material extrusion is
among the most widely used rapid prototyping processes with growing application in finished-
part manufacturing. It can print 3D structures with a variety of materials, including wax,
thermoplastics, ceramics, and metals [9]. Material extrusion systems also have advantages such as
satisfactory printing accuracy, compact equipment size, and low maintenance cost.
Figure 1-2: Material extrusion AM process principle and schematic [10].
The binder jetting AM technology was first developed at MIT in the late 1980’s. A binder
jetting system includes a powder bed as the build chamber, a leveling roller, an ink-jet printhead,
and either a powder hopper or feed bin as the powder supplier (see Figure 1-3). During the
printing process, the build volume is repeatedly lowered to the desired layer height followed by a
thin layer of powder spread over the build area with the roller. The printhead passes over each
new layer of power and selectively deposits liquid binder similar to a conventional 2D ink-jet
printer. The powder layers are bound together where the binder is deposited, eventually forming
the printed object [11]. When the part is completely printed, unbound powder is removed and
4
reused. The powder used in binder jetting can be nearly anything, while metals, ceramics, sand,
and plastics are among the most commonly used.
Binder jetting is a rather cheap and fast process. A layer can be formed very quickly,
often in a matter of seconds, by using arrays of printheads. Binder jetting does not need support
structures as the powder surrounding the printed part naturally acts as a support for any
subsequent overhanging geometries. In addition, binder jetting is able to 3D-print full-color
objects by using binders with different colors [12].
Figure 1-3: Binder jetting AM process principle and schematic [13].
As a novel method of manufacturing 3D objects, AM has many advantages over
traditional processes, such as: easy fabrication of parts on-demand for customization and
personalization; no need for special tooling; efficient material use; reduced time and cost of
manufacturing for individualized parts and small-quantity productions; creation of novel
components and complex structures; reduced environmental impact; and simplified supply chain
[14]. Although AM is unable to replace conventional manufacturing methods in the foreseeable
future, especially for the high-volume production of parts with low complexity and high
5
accuracy, AM will still revolutionize the manufacturing industry through its integration with a
wide range of conventional manufacturing technologies [15]. For example, many Rapid Casting
(RC) solutions have been developed by combining AM techniques with traditional metal casting
processes [16].
AM is not only changing what we can make and how we fabricate parts, but also
reshaping the way we design things. One of the most promising aspects of AM is the possibility
to create highly complex geometries without increasing manufacturing cost. Namely, designers
now are able to realize new and different design ideas previously impossible or too expensive to
manufacture. To achieve such a goal, design engineers must learn first how to break out of the
conceptual barriers created by conventional fabrication techniques and design effectively for AM
processes [17]. However, one of the technical challenges currently faced by design engineers is
their limited knowledge of Design for Additive Manufacturing (DfAM) principles. Insufficient
understanding and application of DfAM are said to be limiting the overall penetration of AM in
industries [18], preventing designers from fully benefitting from AM [19], and preventing AM
from reaching its full potential in general [20]. Once the DfAM rules are well understood, they
should be used to develop systematic design guidelines for design engineers in different
industries.
Nowadays, there is a significant tendency for using AM for structural, load-bearing
structures, by taking advantage of the inherent design freedom of AM processes [21]. Thanks to
the digital workflow of AM, more advanced design approaches can be integrated into the CAD
design stage to improve the performance of final products. A good example is the increasing
application of topology optimization, a powerful technology to generate lightweight and
functional structures. Topology optimization relates well to AM because the optimized structures
often have complex geometries that are extremely difficult and expensive to make using
traditional manufacturing processes.
6
1.2. Introduction to Topology Optimization
In the field of engineering, a structure could be optimized for many purposes, such as
mechanical performance, thermal properties, and fluid flow. Optimization of structures can be
divided into 3 categories: (1) size optimization, (2) shape optimization, and (3) topology
optimization [22]. In a sizing or shape optimization problem, the goal is to find the optimal size
or shape of features of a structure. On the other hand, topology optimization is aimed at
determining the optimal material distribution of a structure, including the size, shape and location
of features, and the way they are connected (see Figure 1-4). Sizing or shape optimization only
deals with one specific geometric parameter, such as the size of trusses and the shape of holes. In
contrast, topology optimization is obviously more challenging but yet more powerful because it
allows unlimited design freedom within the design space; therefore, the true global optimum can
be achieved for a given set of boundary conditions and loads. For this reason, topology
optimization is the focus of this thesis to reduce the weight of structures and optimized their
mechanical performance under external loadings.
Figure 1-4: Sizing, shape, and topology optimization [22].
7
By definition, topology optimization (TO) is a mathematical approach to optimize the
distribution of material within a design space, for a given set of loads, boundary conditions, and
constraints with the goal of maximizing the performance of the system [23]. A typical TO
problem involves four important concepts: (1) design space, (2) objective, (3) constraints, and (4)
design variables. Design space describes the allowable region within which the design exists. An
objective defines a physical quantity that needs to be maximized or minimized, such as
compliance, stiffness and Von Mises stress. Constraints include the limits that the optimization
solution must satisfy, such as maximum volume fraction and displacement. Design variables are
numerical inputs that are allowed to change during the optimization process.
TO is mostly solved based on finite element method (FEM), by which a design space is
discretized into finite elements [24]. As a result, solving the material distribution problem is
converted to the so-called “0–1” problem: elements should either contain material (𝜌𝑒 = 1) or be
voids (𝜌𝑒 = 0), where 𝜌𝑒 is the material concentration or “pseudo-density”. For cases dealing with
isotropic materials, TO tries to generate a topology consisting of a finite set of either solid or
empty elements, namely the ISE (Isotropic-Solid or Empty) topology [25]. An obvious advantage
of this method is its high compatibility with finite element analysis (FEA), which can be easily
used to evaluate the optimization results. A typical TO task for minimizing the compliance or
maximizing the stiffness of a structure with a volume constraint can be written as:
min 𝑓 = 𝐹T𝑢
(1.1) Subject to: 𝐾𝑢 = 𝐹
∑ 𝜌𝑒𝑣𝑒
𝑒∈1
≤ 𝑉
where 𝐹 is an external load; 𝐾 is the stiffness matrix and 𝑢 is the displacements resulting from 𝐹;
𝜌𝑒 and 𝑣𝑒 are the pseudo-density and the volume, respectively, of element e; and 𝑉 is the
maximum allowable volume of the design.
8
A number of optimization techniques have been developed for the optimal ISE topology
design, such as density-based, level set, topological derivative, phase field, evolutionary, and
several others [26]. Among them, a density-based method named Solid Isotropic Material with
Penalization (SIMP) [27], has become one of the most widely used TO algorithms. Density-based
methods allow the material concentration 𝜌𝑒 take any value between 0 and 1. This relaxation
converts the discrete “0–1” problem to a continuous problem so that using gradient-based
optimization methods to solve it becomes possible. The core of SIMP algorithm can be expressed
as:
𝐸 = 𝜌𝑒𝑝
𝐸0, 0 ≤ 𝜌𝑒 ≤ 1, p ≥ 1 (1.2)
where 𝐸0 is the theoretical stiffness of element e, 𝐸 is the element stiffness used for analysis, the
pseudo-density 𝜌𝑒 is the only design variable, and p is the so-called penalization factor with a
typical value of 3 [27]. The function of 𝑝 is to reduce elements with intermediate densities by
“penalizing” their stiffness with respect to the element volume, because these elements do not
physically make sense for manufacturing (see Figure 1-5). For an element with isotropic material
density over it, the stiffness matrix 𝐾, which depends on the stiffness 𝐸, can be expressed as:
𝐾 = 𝜌𝑒𝑝
𝐾0, 0 ≤ 𝜌𝑒 ≤ 1, p ≥ 1 (1.3)
A typical SIMP-based TO procedure is shown in Figure 1-6. The optimization process
usually starts with a design domain with elements of a uniform distributed density and a volume
fraction. The initial element density and volume fraction are defined by users. The first step in the
optimization loop is an FEA in which the equilibrium equation 𝑓 = 𝐹𝑇𝑢 in Eq. (1.1) is solved.
Next, a sensitivity analysis is conducted by calculating derivatives of the objective function 𝑓
with respect to the design variable 𝜌𝑒. To ensure numerical stability, filtering techniques are then
applied to remedy the checkerboard problem [28]. In each optimization loop, the design variable
𝜌𝑒 will be updated by mathematical optimization approaches until convergence of the design is
9
reached. Many optimization approaches can be used for density updating, such as Optimality
Criteria (OC) [29] and the Method of Moving Asymptotes (MMA) [30].
Figure 1-5: The influence of penalization factor 𝑝 on topology optimization results [27].
Figure 1-6: The general procedure of topology optimization based on SIMP algorithm [31].
SIMP is a simple yet effective TO algorithm; therefore, it has been widely used in both
research efforts and commercial software tools. Table 1-1 lists some common TO software
available for either commercial or educational use.
10
Table 1-1: List of common commercial and educational TO software [32].
1.3. Introduction to Traditional Metal Casting
Metal casting is a manufacturing process with a history that dates back thousands of years,
in which molten metal is poured into a mold with a hollow cavity of the desired geometry, and then
allowed to solidify. The special factories where metal casting is performed are called foundries.
The term “casting” can also refer to a metal cast part produced via a casting process.
Sand casting is a type of metal casting process using sand as the mold material. Over
70% of all metal castings are produced via sand casting [33]. Greensand molds are the least
expensive and most widely used and are made of a mixture of sand, water, and a clay or binder
[33]. The primary components of a sand mold are shown in Figure 1-7. The mold is divided by a
parting surface into two halves: (1) the cope (upper half) and (2) the drag (bottom half). Both
cope and drag are packed inside a box, called a flask. The gating system, including pouring cup,
sprue, runner(s), and gate(s), forms a path for the molten metal to fill the mold cavity. Cores are
separate pieces that are used to provide geometric features of a casting that cannot be added
directly onto the pattern. Cores are typically raw sand mixed with binder and are molded in core
boxes. The structure of a mold to locate and support a core is called a core print. Risers, or
11
feeders, are reservoirs designed for storing hot molten metal to prevent porosity in casting due to
shrinkage.
Figure 1-7: Sand mold assembly [34].
In general, there are six steps involved in the sand casting process (see Figure 1-8):
(1) Make a pattern, normally of dimensionally stable materials like wood, and pack it
with sand to create a mold.
(2) Incorporate the pattern and sand in a rigging system, including gating and riser(s).
(3) Remove the pattern and form the sand mold cavity.
(4) Pour molten metal into the mold cavity.
(5) Allow the metal to solidify and cool in the mold.
(6) Break the sand mold and remove the casting.
12
Figure 1-8: Traditional sand casting process [35].
As the most widely used casting process, sand casting can deal with most metals and
alloys with flexibility in shape, size, and quantity. The cost for pattern and material in sand
casting is relatively low. However, it also suffers from several drawbacks such as limited
complexity of design, relatively coarse surface finish, and relatively poor dimensional accuracy
compared with other molding processes [36].
Investment casting (IC) is one of the oldest metal casting techniques and dates back to
4000–6000 B.C. [37]. IC is also called lost-wax casting, which employs wax patterns as
sacrificial patterns to cast metal parts. Paraffins and microcrystalline waxes are the most widely
adopted waxes in IC, and they are often used in combination because their properties tend to be
complementary [38]. In order to improve their properties, waxes are often blended with additive
materials such as plastics, resins, fillers, antioxidants, and dyes [39].
Traditional IC follows the process shown in Figure 1-9. Specifically, wax patterns are
first made via injection molding and attached to a wax sprue to create a pattern cluster. The
cluster is then repeatedly invested with refractory ceramic coatings until a hard and strong shell is
13
formed. Each coating consists of a fine ceramic slurry layer covered with a coarse layer of
ceramic “stucco” particles. Common refractory ceramics used for both slurry and stucco include
silica, zircon, alumina, and various aluminum silicates [38]. Upon the shell building, the
expendable wax patterns are removed at about 140 ℃ and 2000 kPa in a steam autoclave or flash
furnace [40]. Following this, the ceramic shell is fired to around 1000 ℃, and the molten metal is
poured into it. After the molten metal has solidified and cooled, the shell is broken away and
individual metal casting is removed for post-processing [40].
Figure 1-9: Schematic of the traditional investment casting process [41].
IC produces high-quality and geometrically complex near net shaped metal parts with
tight tolerances, but it has several disadvantages. First, the design and production of a metal die
for the injection molding of wax patterns often requires very high tooling cost and long lead time.
This limits the economic benefits of IC only to mass production. Second, once the die tooling is
completed, any changes to the wax pattern design could be inconvenient and costly [42]. Third, it
is difficult for IC to cast objects requiring cores.
1.4. Part Design for Traditional Metal Casting
All traditional metal casting processes share a set of important rules for the design of cast
parts and the molds used to make them. These design rules include:
14
• Minimum wall thickness: If the metal is flowing for a long distance in a mold, a
minimum wall thickness must be maintained to avoid incomplete casting. Thin and
long sections might be blocked by quickly solidified metal and result in unfilled areas
inside a mold. Generally, the minimum wall thickness is influenced by multiple
factors, such as flow rate, temperature and fluidity of molten metal, flow distance，
and complexity of mold. Table 1-2 lists the recommended minimum wall thickness
for some common materials in the sand casting process. Investment casting allows a
smaller minimum wall thickness because of the high temperature of the shell molds.
Wall thickness as low as 0.5 mm is possible within a small area. Normal minimum
wall thickness lies between 1 mm and 2 mm [43].
Table 1-2: Minimum wall thickness of different metals for sand casting [44].
• Corners and edges: Sharp corners and edges with acute angles should be rounded to
avoid the occurrence of “hot spots”, isolated pools of liquid trapped inside solidified
metal (see Figure 1-10). Hot spots are the most common defects due to uneven
cooling of molten metal during solidification. Sharp corners and edges encourage
high thermal gradients and promote a higher solidification rate of metal near the
surface, leaving the inside regions to become thermally isolated, which finally form
shrinkage porosity or cavities in the casting. Rounded edges and fillets can also
15
prevent stress concentration and tears when the metal is cooling down.
Figure 1-10: Corner and edge on cast part: (a) bad design; (b) improved design [45].
• Section changes: Abrupt change in section is another cause of hot spots because thin
sections solidify faster than heavy sections. Uniform sections are always preferred. In
general, different sections should have a relative thickness less than 2:1. If a heavy
section is unavoidable, then a fillet or taper are preferred instead of sharp steps (see
Figure 1-11). It is worth mentioning that multiple uniform sections, when intersecting
at one point, can also create a region with heavy cross-section, resulting in the
uneven cooling problem mentioned earlier. One way to minimize this is to core the
intersection by a hole, similar to a hub hole in a wheel with spokes (see Figure 1-12).
Figure 1-11: Section change on cast part: (a) bad design; (b) and (c) improved designs [46].
16
Figure 1-12: Intersection on cast part: (a) bad design; (b) and (c) improved designs [45].
• Patternmaker’s shrinkage: Patternmaker’s shrinkage refers to the contraction occurs
when the metal is cooling to room temperature after complete solidification in a
mold. It results in the change of casting dimensions from those of liquid in the mold
to those dictated by the metal’s rate of contraction. Thus, as the solid casting shrinks
away from the mold walls, a pattern designer must predict the dimensional change of
the casting based on the pouring material and add proper allowance on the pattern to
compensate such shrinkage. Table 1-3 lists the patternmaker’s shrinkage of some
common cast metals.
Table 1-3: Patternmaker’s shrinkage of different cast metals [44].
17
• Machining allowance: Sand casting products are generally poor in surface finish and
dimensional accuracy; therefore, they often need post machining for finish or
accuracy purposes. Then machining allowances should be added to the pattern
dimensions. The amount of machining allowance is affected by multiple factors, such
as the size and shape of the casting, the casting material, and the degree of accuracy
or finish required. The machining allowances recommended for different metals in
sand casting are given in Table 1-4. Investment cast parts often have superior surface
finishes and dimensional accuracy compared with sand castings, thereby requiring
less machining allowances. The machining allowance for an IC process only depends
on the overall dimension of a cast part (see Table 1-5).
Table 1-4: Machining allowances of different cast metals in sand casting [44].
18
Table 1-5: Machining allowances for cast parts in investment casting [47].
Due to different pattern fabrication and mold making methods, different casting processes
might have some special design rules. For sand casting, there are two important geometric
constraints for cast parts related to the withdrawal of pattern from the sand mold:
• Draft: To allow a pattern to slip out from the molding sand smoothly, a certain
amount of draft is needed during pattern design. Removal of a pattern with little or no
draft might damage the sand mold (see Figure 1-13). Proper drafts are selected based
on various factors such as surface position, method of molding, drawing of the
pattern, pattern material, surface smoothness, and degree of precision. The
recommended draft angles of pattern surface for sand casting are summarized in
Table 1-6.
Figure 1-13: Pattern designed (a) without draft and (b) with draft [45].
19
Table 1-6: Recommended draft angles for patterns used in sand casting [44].
• Undercut: In sand casting, undercuts are features on a casting that prevent the pattern
from being withdrawn from the sand mold. If a casting has undercuts, then cores or
extra mold pieces are used to form the desired mold cavity. However, this could
largely increase the time and cost for designing and fabricating them. Thus, undercuts
on sand cast parts are not preferred in general (see Figure 1-14).
Figure 1-14: (a) Pattern with undercut; (b) and (c) improved designs [45].
Draft and undercuts are not constraints for investment casting because the wax patterns
are drained out of the shell mold by heating. However, the investment casting process has special
requirements on the geometry of holes:
20
• Holes: To ensure the integrity and strength of ceramic shells, the maximum ratio of
hole depth (T) to the hole diameter (D) for IC patterns must be controlled. Depending
on the hole type and size, the maximum allowable depths for holes are listed in Table
1-7. In addition, all hole edges should be rounded to increase the core strength and
prevent shell fracturing during the metal pouring.
Table 1-7: Geometric requirements for holes on patterns in investment casting [48].
1.5. Rapid Casting
The application of AM in traditional metal casting processes to create patterns, cores,
cavities, and other toolings is often referred to Rapid Casting (RC). To date, almost all major AM
processes have been used in traditional metal casting and provided a wide range of RC solutions.
Figure 1-15 and Figure 1-16 summarize some of the AM systems used in rapid sand casting and
rapid investment casting (RIC), respectively. Generally, three AM integration approaches exist:
(1) direct AM of patterns, (2) indirect AM of patterns, and (3) direct AM of molds [49]. For the
direct pattern AM approach, AM-fabricated objects are directly used as patterns for mold making;
whereas, for the indirect approach, AM fabricated objects are used as direct tooling or as master
patterns to make indirect tooling for the casting patterns. The direct mold AM approach
introduces a “pattern-less” process for which molds are directly 3D-printed without the need for
patterns.
21
Figure 1-15: Rapid sand casting solutions: approaches and corresponding AM processes [49].
Figure 1-16: Rapid investment casting solutions: approaches and corresponding AM processes
[49].
The intensive interest and wide application of RC were motivated by traditional metal
casting’s drawbacks such as long lead-time, high tooling cost, and restricted part complexity.
Traditionally, in order to produce cast prototypes, a model and eventually cores have to be
created, involving time and costs that hardly match the rules of the competitive market [50]. In
contrast, complex patterns and tooling required for casting may be fabricated in a matter of hours
for RC processes, and final castings can be produced in a matter of days [51]. For the “pattern-
less” RC processes, the lead-time could be even shorter. In addition, RC enables the real-time
development and convenient validation of the designs and the casting process [52]. This ensures a
22
concurrent engineering approach and minimizes the peril of late modifications of the ultimate
production tools [53].
In recent years, direct AM of metal parts has experienced intensive research and
development, especially through the powder bed fusion AM process. However, there are still
some barriers to the wide adoption of direct metal AM, such as cost, quality, range of materials,
and size limitations [54]. In contrast, RC still holds advantages in many aspects. First, RC is able
to produce isotropic metal parts, just like traditional castings, which cannot be achieved by many
direct AM processes. Second, RC, as an inherent casting process, is able to use a much wider
range of metals and alloys compared to direct metal AM processes. Third, many RC solutions,
such as sand casting with 3D-printed sand molds, can produce complex cast parts in a few days
with a cost of merely a few hundred dollars, whereas direct AM of the same part could take
longer time and cost many thousands of dollars. In addition, metal AM using powder bed fusion
requires support structures to prevent warping, which can be avoided in some RC solutions by
using the self-supported AM processes, i.e., binder jetting. Last, but not least, RC processes can
be quite flexible in the size of metal products depending on the casting process and how AM is
integrated. For example, sand casting using 3D-printed cores can produce metal castings as large
as traditional castings, whereas the maximum build volume of powder bed fusion systems is on
the order of 0.03 m3 [54].
1.6. Thesis Overview
This thesis aims to exploit the design freedom of cast parts for RC. Specifically, the
objective in this thesis is to develop knowledge-based design guidelines for two RC processes: (1)
sand casting with molds fabricated via 3D Sand Printing (3DSP) and (2) investment casting using
material extrusion–fabricated wax-like patterns. The thesis also tries to obtain optimal cast part
designs by applying topology optimization followed by geometric modification based on relevant
23
AM and casting design rules. In Chapter 2, a study is presented on the metal part design for
3DSP-based sand casting. It analyzes the unique benefits provided by 3DSP for sand casting and
explains how this AM process can impact the sand cast part design rules. In Chapter 3, another
study is presented on the design of material extruded wax patterns for investment casting. It
explores the feasibility of using wax-like filament to fabricate quality patterns for investment
casting and summarizes the design considerations involved in the pattern design process. For each
study, a general design framework and detailed design guidelines are proposed and validated
through a case study. In Chapter 4, findings of this thesis are summarized and final conclusions
are drawn.
24
This Chapter is a manuscript submitted to the International Journal of Metalcasting (IJMC)
Chapter 2
REDESIGN OF TRADITIONAL METAL PARTS FOR 3D SAND
PRINTING USING TOPOLOGY OPTIMIZATION
Additive Manufacture of sand molds, also called 3D Sand Printing, introduces an
efficient mold and core fabrication process free from the need for pattern making. The capability
of 3D Sand Printing to create highly complex mold structures provides great design freedom for
the geometry of cast metal parts. There is a need to thoroughly understand the opportunities and
restrictions of 3D Sand Printing and rethink part design for sand casting. This study explores a
general design framework for 3D Sand Printing–based metal casting with a focus on the
development of new cast part design guidelines under such a framework. Particularly, we take
advantage of Topology Optimization during the cast part design. New part design rules are
presented and compared with traditional sand casting rules. A case study is conducted in which an
existing metal component is redesigned using the proposed design rules. The redesigned part is
successfully cast with a 3D-printed mold and demonstrates an improvement in mechanical
performance with reduced weight.
2.1. Introduction
AM builds parts from 3D model data layer by layer, which is distinctly different from
conventional manufacturing such as subtractive processes (i.e., milling or drilling), formative
processes (i.e., casting or forging), and joining processes (i.e., welding or fastening) [55]. Since
the inception of AM industry in the early 1980s, the world has witnessed a boom in the
development of various AM processes in seven categories: vat polymerization, sheet lamination,
25
material extrusion, material jetting, binder jetting, powder bed fusion, and direct energy
deposition. As AM technology evolves, the application of AM has also expanded from original
“rapid prototyping” to rapid tooling and direct production of end-use products. The fast market
growth of AM industry is not surprising. According to Wohlers Report 2016, the AM industry
Surpassed $5.1 Billion in 2015 with an impressive compound annual growth rate (CAGR) of
26.2% over the past 3 decades [5].
AM has a number of advantages over traditional processes, such as improved
manufacturability, minimum material waste, reduced tooling cost, shortened lead-times and
supply chains, among others. AM is creating new processes, new applications, and new products
beyond the capability of conventional methods. Meanwhile, AM can be incorporated into a wide
range of traditional manufacturing processes to bring them into a completely new era.
Sand casting is a metal casting process with thousands of years of history still widely
used today. Over 70% of all metal castings are produced via a sand casting process [33].
Depending on the complexity of the cast part, traditional sand casting process can be very time-
consuming. A pattern is first made, normally of wood or dimensionally stable synthetic materials,
to represent the cast part or casting. A sand mold is formed by packing the pattern with sand
either by hand or machine, and the mold cavity is created by withdrawing the pattern out of the
sand. If a cast part has internal features, then sand core(s) molded in a core box should be added
into the mold. Then molten metal is poured into the mold cavity and allowed to solidify. AM
techniques can be applied in traditional sand casting in mainly three ways to achieve a rapid sand
casting solution [49]. First, patterns can be directly 3D-printed, and multiple AM processes have
been successfully used to serve this purpose, such as Laminated Object Manufacturing (LOM),
Fused Deposition Modelling (FDM) and PolyJet. Second, AM generated models can be used as
tools for molding sand cores and other necessary sand pieces. Third, more recently, sand mold
26
can be directly printed via a binder jetting process, also named as 3D Sand Printing (3DSP),
which leads to a novel pattern-less casting process [49].
Figure 2-1 illustrates the mold printing process of 3DSP, in which a binder is printed onto
a bed of sand layer-by-layer, causing the sand to bond where the binder is deposited. 3DSP is a
revolutionary technique with which highly complex sand molds and cores can be easily produced
without the need for pattern and core box fabrication. The post-processing for 3DSP simply
removes the loose sand using compressed air or a vacuum. As the mold fabrication method
changes, the design rules for sand casting change too. For example, some geometric
considerations for the traditional sand casting process, such as draft and undercuts for cast parts,
straight parting lines, and regular rigging structures, are no longer design constraints. On the other
hand, the geometric flexibility enabled by 3DSP provides an opportunity of using more advanced
tools, e.g., topology optimization (TO), for the design of both 3D-printed molds and cast parts.
Figure 2-1: The 3DSP process via binder jetting [56]
TO is a design approach to determine the optimal layout of a part within a design space,
for a given set of loads, boundary conditions, and constraints, with the goal of maximizing the
performance of the geometry. TO is a powerful tool to generate light-weight parts with improved
mechanical performance. However, the optimized parts often have highly complex geometries
that are extremely difficult and expensive to fabricate using traditional methods. Fortunately, TO
27
relates well with AM due to its “free-form” aspect, which does not require any extra tooling to
produce complex geometries.
However, 3DSP-based casting is an indirect AM process in which the 3D-printed
products are used as a tool for the traditional manufacturing process. This means both AM
constraints and casting design rules should be considered during the casting design. It is
important to understand what specific design considerations are involved and how they can be
combined with advanced design approaches, e.g., topology optimization, for this new casting
method. The design process for a 3DSP-based casting can be divided into two stages: (1) cast part
design and (2) mold design. A general design framework for these two design stages is shown in
Figure 2-2.
28
Figure 2-2: Two design stages of a 3DSP-based casting process: (a) cast part design; (b) mold
design.
In the current study, we only focus on the cast part design stage and not the mold design
stage. Our objective is to explore the opportunities and considerations in the design freedom of
cast part while using 3DSP. The next section is a review of the traditional sand casting rules, the
“state-of-the-art” of 3DSP, and a review of applications of topology optimization in sand casting
and AM. In the methodology section, we introduce design guidelines for sand casting with 3D-
printed molds, focusing on the explanation of new cast part design rules. Particularly, we explore
how to implement topology optimization into the cast part design for this new casting process.
Then, a detailed case study is presented on the redesign and casting of a metal component
following the new design rules. The last section summarizes the important findings and
conclusions.
2.2. Literature Review
Traditional sand casting offers limited design freedom in cast metal parts due to inherent
design and processing constraints. Two major categories of part design rules should be followed
for conventional sand casting: (1) rules for controlling casting defects and (2) rules for mold
making or pattern withdrawal. Casting defect–related rules, such as minimum wall thickness,
uniform sections, fillets, and intersections, are aimed to assure complete filling and good casting
quality. Mold making–related rules refer to part design constraints for the success of mold
fabrication. For instance, to withdraw a pattern from the molding sand without causing any
damage, the pattern should not have certain features, e.g., undercuts, and must be designed with a
certain amount of draft. Table 2-1 illustrates some common cast part design rules. If a casting has
undercuts or does not tolerate draft, cores should be integrated into the mold to form the desired
mold cavity. Cores are typically separate pieces of sand mixed with binder, which are
29
traditionally molded in core boxes. The efficient fabrication and maintenance of elements that
form the sand mold, including patterns, cores and core boxes, is a major challenge in sand casting
production. Although computer numerically controlled (CNC) machining improves the efficiency
and reliability of this form of tooling production, it is still time-consuming and requires a
considerable amount of working material to get to the net tooling shape [36]. Therefore, part
design for traditional sand casting naturally tends to have as less complexity as possible.
Table 2-1: Important part design rules for traditional sand casting [45].
In contrast, 3DSP provides engineers with a cost-effective and time-efficient mold and
core fabrication process for the casting of complex metal parts that are traditionally impossible to
achieve. Cellular structures and sandwich panels have been successfully cast in 3D-printed sand
molds [57]. Unlike direct metal AM processes, e.g., powder bed fusion, which often suffer from
material limitations [54], 3D-printed molds via 3DSP can be used to cast a wide range of metals
30
and alloys. Another advantage of this process over direct metal AM is its ability to produce
isotropic metal parts. Metal specimens cast with 3D-printed molds have shown similar properties,
such as mechanical performance and hardness, to those created using traditional sand casting
techniques [58].
The broad application of 3DSP in industry is promising. Binder jetting is a rather fast and
low-cost technology. Almaghariz et al. [56] analyzed the tooling and fabrication costs for 3DSP
and conventional mold making. 3DSP showed superior cost-effectiveness at lower production
volume or when the casting has high complexity. More recently, Hackney and Wooldridge [59]
successfully applied 3DSP in the automotive industry, resulting in considerable cost and time
savings with improved quality and product flexibility.
In many industries there is always a need and interest in the design of lightweight and
structurally optimized designs [60-64]. In the field of traditional sand casting, topology
optimization techniques have been applied to redesign the cast parts [65] and riser designs [66,
67] for pouring material saving and casting quality improvement. However, limitations in the
manufacturability of conventional processes largely hinder the broader application of these
techniques. To overcome such challenges, there has been a lot of effort to develop TO approaches
integrated with manufacturing constraints. For traditional sand cast part design, researchers have
incorporated several casting constraints into the TO process, such as connectivity control [68],
minimum wall thickness [68], and pattern withdrawal [68-76]. Yet, given the number of
traditional sand casting design constraints, it is not only unlikely to find a single TO solution that
can take into account all of the constraints, but also difficult to fully embrace the benefits of TO.
In fact, only AM can take full advantage of TO. Seppala and Hupfer [77] topologically
optimized a low-pressure turbine guide vane and suggested the result can be manufactured only
via AM due to its complex interior structures. TO has been successfully used to redesign a
number of mechanical components for different AM methods. Rezaie et al. [78] took advantage
31
of TO results to improve a Messerschmitt-Bölkow-Blohm (MBB) beam with square holes
produced by material extrusion. Another good example is the well-known GrabCAD GE Bracket
Challenge [79], which involved the topology optimization of an engine bracket for Direct Metal
Laser Sintering (DMLS). For the same AM process, Maranan et al. [80] redesigned an upright on
SAE Formula student racecar and achieved reductions in material consumption, build time and
cost.
3DSP, as a binder jetting AM process, also provides a good opportunity for using TO for
the cast part design, although TO does not directly work on the 3D-printed molds. Industry has
started to explore the use of TO in conjunction with 3DSP and have obtained promising results. A
cast part used as a device in the woodworking industry in Finland was topologically optimized
and the final castings saved up to 38% of the weight [81]. Another case study showed that 3DSP
plus TO could design cast parts with up to 33% weight reduction while maintaining the structural
requirements demanded by their application [82]. However, no systematic analysis of new part
design rules has been developed so far, particularly, what casting and 3DSP constraints should be
considered while using TO remains unclear. Studies have attempted to incorporate AM
constraints into the TO process [80, 83-86]. Unfortunately, most of the current research focuses
on the optimization of overhang structures for minimum support material, which is not a
constraint for the self-supported binder jetting process. It seems timely to move a step further
based on current findings and develop knowledge-based guidelines for 3DSP-based cast part
design.
2.3. Methodology
A proposed design flowchart for sand casting with 3D-printed molds is shown in Figure
2-3. The cast part design involves following important steps:
32
Figure 2-3: Design flowchart for sand casting with 3D-printed molds.
33
2.3.1. Step 1: Prepare the 3D CAD model for a metal part
The cast part design process starts with a solid CAD model of a metal part. Potential
castable metal parts using 3D-printed molds include existing castings from traditional processes
and metal parts that are originally not suitable for conventional sand casting due to poor
manufacturability or high tooling cost. The solid CAD model can be prepared in any 3D CAD
software, e.g., SolidWorks (Dassault Systèmes SolidWorks Corporation, Waltham, MA), or
obtained through reverse engineering. Important part information, including material properties,
assembly, load conditions, and boundary conditions, must be identified before topology
optimization.
2.3.2. Step 2: Perform TO for the initial CAD model
A number of commercial and educational software tools, such as OptiStruct, Inspire,
ATOM, and TOSCA, are available to perform topology optimization [32]. Despite differences in
user interface and capability, these tools share the same fundamental workflow of TO. The first
step of TO is to identify the design space and the non-design space. The design space describes
the region that needs to be optimized. The non-design space is the region that should not be
modified, such as critical features and fixed boundaries. The setup for a common TO task is
similar to the setup for an FEA, including defining material properties, loads, boundary
conditions, interactions between components, and meshing. Figure 2-4 illustrates an example of
the TO setup for a triangle bracket. Before running the optimization task, objective(s) and
constraint(s) are defined. For a mechanical loading application, the typical objective is to
minimize the compliance or maximize the stiffness of the system, and the most important
constraint is the volume fraction of the optimized geometry to the original geometry [22]. The
volume constraint setup in a TO task mainly depends on how much reduction in weight is
desired.
34
Figure 2-4: Topology optimization setup and result of a triangular bracket.
2.3.3. Step 3: Refinement and revision of the optimization output
The output geometry of TO is a collection of discretized elements with a coarse surface
finish. Therefore, refinement is often preferred to acquire a smooth and more finely-detailed final
design. Moreover, further revisions are required to design the geometry for sand casting.
Depending on the design tools, the refinement and revision could be done differently. Figure 2-5
gives an example of getting a smooth and revisable truss from a TO output geometry using
SolidThinking Inspire (Altair Engineering, Troy, MI).
Figure 2-5: Refinement and revision using Inspire: (a) a coarse truss on a TO output geometry;
(b) refined truss structure 1; (c) refined truss structure 2.
35
2.3.4. Step 4: Apply design rules
Design considerations in three important aspects—part requirements, sand casting rules,
and 3DSP constraints—must be kept in mind during the revision, which are explained next.
(1) Part requirements
The revised geometry should meet certain part requirements that are either based on
common sense or pre-defined by the designer. The part requirements for a casting could be three-
fold:
• Part rationale: The design itself as a mechanical part must make sense. For instance,
TO only modifies the design space, not the non-design space; thus, it is possible to
output an unnatural geometry with material in the design space disconnected with the
non-design space. In such cases, connections must be manually added by the designer
to form a valid part.
• Geometric requirements: The revised design should meet, if any, pre-defined
requirements in part geometry, such as volume reduction, maximum number of holes,
and specific thickness for a certain section.
• Aesthetics: Aesthetic requirements often refer to symmetricity, organic appearance,
etc.
(2) Sand casting rules
3DSP impacts many of the traditional sand casting part design rules, especially rules
related to conventional mold making, such as undercuts and draft. It can significantly improve
part design freedom and enhance the efficiency of current sand casting processes. Designers are
now able to focus mostly on the part functionality rather than worrying about its castability. At
the same time, it allows casting of metal parts in near-net-shape regardless of complexity, thereby
minimizing material waste and post-processing. The rules for cast parts are modified as follows
for 3DSP.
36
• Undercuts: Traditional sand cast parts tend to avoid undercuts on side surfaces with
respect to the drawing direction. Some undercuts can be provided by cores, but the
addition of cores will increase the tooling cost substantially and cause potential core
displacement and/or dislodge in the mold assembly. Figure 2-6 gives an example of a
pipe nipple with undercuts in all axes, which is nearly impossible to cast using
traditional methods. However, this part is fully castable using the 3DSP-based casting
process.
Figure 2-6: A metal pipe nipple not able to cast traditionally due to undercuts.
• Draft: The need for draft on conventional cast parts requires extra effort in
designing and careful planning for any machining or assembly tolerance. In
addition, draft alters the orientation of side surfaces and geometry of part
sections, which could require trade-offs with uniformity of wall thickness.
An example of non-uniform wall thickness due to draft is shown in Figure 2-
7a. A recommended redesign of this part for a 3DSP-based casting is shown
in Figure 2-7b.
37
Figure 2-7: Redesign of a housing for 3DSP-based casting: (a) current cast part with draft; (b)
redesign.
Traditional part design rules related to casting defects and post-processing, such as wall
thickness, uniform sections, intersections, rounded edges, and machining allowance, should still
be followed for 3DSP-based casting. However, 3DSP promotes an improved implementation of
some design rules during part design.
• Datum: Casting datum features, e.g., pilot holes for drilling, are important for the
convenience of post-processing and accuracy of machining. Conventionally, datums
generally cannot be placed on side surfaces because they are likely to become
undercuts. In contrast, 3DSP provides the advantage of casting metal parts with
datum features on any surface (see Figure 2-8).
38
Figure 2-8: Redesign of a U-bracket for 3DSP-based casting: (a) current cast part with no pilot
holes on the side surfaces; (b) redesign with pilot holes.
• Uniform section: The uniformity of sections in sand cast parts are influenced by
multiple factors, such as part requirements, the complexity of cast part design,
manufacturability, and aforementioned undercuts and draft. Figure 2-9a shows the
conventional manufacturing process for an aluminum hinge bracket. The heavy
section on this bracket is much thicker than its base because the removal of material
inside it will require the addition of cores that will further increase the tooling cost
for the process. As a result, a riser must be added right on top of it to compensate the
shrinkage porosity. While using 3DSP, this heavy section can be easily redesigned to
a lighter but improved geometry without a need of any cores and risers (see Figure 2-
9b).
39
Figure 2-9: Redesign of a hinge bracket for 3DSP-based casting: (a) current cast part with a
heavy section; (b) redesign with uniform wall thickness.
• Rounded edges: Rounded edges and fillets are important for castings to avoid the
occurrence of hot spots and prevent stress concentrations and tears. However, adding
rounded edges to a cast part might also increase the complexity of its cores. For
example, traditional casting edges near core prints are often not rounded because that
will lead to more complex designs for core extension. This challenge can be easily
overcome by using a 3D-printed mold because core and mold complexity is free for
3DSP. Figure 2-10 shows the current design of a cast pipe and the improved redesign
for 3DSP.
40
Figure 2-10: Redesign of a pipe elbow for 3DSP-based casting: (a) current cast part with a core;
(b) redesign with a core.
• Intersections: When multiple sections intersect at one point, the best strategy to avoid
hot spots is to core the heavy intersection with a center hole; however, in
conventional sand casting design, this rule is sometimes inconvenient to apply
because of the need for an extra core. If such a center hole is produced by a part of
the mold instead of a core, then there is a risk that it might be damaged while
drawing the pattern from the mold, especially when the hole has a long structure.
Figure 2-11a shows an existing wheel frame that is cast with spokes intersecting at a
solid heavy hub even though a center hole is naturally needed for the structure and
good for casting. In comparison, 3DSP can easily fabricate the center hole as a core
or a part of the sand mold (see Figure 2-11b).
41
Figure 2-11: Redesign of a wheel frame for 3DSP-based casting: (a) current cast part with a thick
solid hub; (b) redesign with a center hole.
Table 2-2 summarizes a comparison of the cast part design rules between conventional
and 3DSP-based processes.
Table 2-2: Improvements of the use of 3DSP on traditional sand casting rules.
42
(3) 3DSP constraints
Because the shape of a sand mold cavity is determined by the geometry of a cast part,
designers need to pay attention to not only the casting geometry itself but also its surrounding
regions left for the sand mold. Namely, the cast part must be designed in a way that its
corresponding mold could be fabricated. The geometry of the sand mold is limited by the
constraints of the 3DSP process. Some of the most relevant constraints are:
• Build volume: 3DSP, as an AM system, has a build chamber that constrains the
maximum volume of the printed molds. If the size of one single mold exceeds a
machine’s build volume, then multiple molds might be printed separately and
assembled together to cast large metal products.
• Minimum feature size: Minimum feature size is the smallest size of a feature, such as
a wall, that can be accurately 3D-printed and remain durable for use. The wall-to-
wall thickness and holes on the cast part must have a size larger than the minimum
features size of the 3DSP system. Thin sections without support should be avoided
due to their fragility during cleaning, handling, and pouring.
2.3.5. Step 5: Perform FEA on the revised design
Geometrical revisions of the TO output might trade off its mechanical performance.
Thus, to ensure that the final design meets the original design goals and requirements, FEA
should be carried out to validate the revised design.
2.3.6. Step 6: Analyze FEA results
If the FEA result does not meet all design goals and requirements, then designers need to
revise the design more or repeat TO until a satisfactory result is obtained.
2.3.7. Step 7: Finalize the optimized geometry
The optimized geometry will be the expected geometry of the final cast part after post-
processing. This geometry will undertake necessary modifications to finish the pattern design. It
43
is worth mentioning that a pattern in 3DSP-based casting is merely a virtual geometric part in
CAD that is used for subsequent rigging and mold design as shown in Figure 2-3.
2.4. Case Study
2.4.1. TO and FEA validation
A case study is presented in which a metal component was topologically optimized for a
simple mechanical loading application and cast with a 3D-printed sand mold. Design rules for
3DSP and casting constraints were implemented. Mechanical testing was performed on the final
casting to further validate the design process. The case study was based on the Alcoa airplane
engine bracket challenge [87] but with modified material and loading condition. This bracket is a
common component on control surfaces (stiff plates) fastened by four high-strength bolts (see
Figure 2-12). The CAD file of the bearing bracket with a bearing was directly downloaded from
www.grabcad.com.
Figure 2-12: Alcoa Aircraft Bracket Challenge [87]: (a) design envelope and (b) boundary
conditions.
TO was performed in Abaqus CAE 6.14 (Dassault Systèmes SIMULIA, Johnston, RI)
using the Abaqus Topology Optimization Module (ATOM). ATOM is a SIMULIA TOSCA–
based product combining structure optimization and Abaqus FEA [88]. The target bearing bracket
was imported into Abaqus CAE as a STEP file. In this case, the non-design space included four
bolts and the region interacting with the bearing. The rest of the bracket was defined as the design
44
space. To simplify the TO setup, the rigid spherical bearing was removed from the assembly.
Instead, to simulate the movement of the bracket when a load is applied to the bearing, a kinetic
coupling interaction was added between the surface holding the bearing and its spherical center.
This interaction could preserve the shape of the surface and force the surface to move with its
center. In this way, loads could be directly applied to the center point.
In the original Alcoa challenge, the bracket was placed under three load cases (see Figure
2-13a). In our case study, we assumed that the bracket was under only one load 𝐹TO, which has a
direction shown in Figure 2-13b. For the convenience of comparing the performance of the
bracket before and after topology optimization, we arbitrarily let 𝐹TO have a value such that the
target bracket has a safety factor of 1.0. The fixed boundaries were the non-design space where
four bolts were mounted. Class 30 gray cast iron was selected as the material of the bracket for
both TO and casting. Because Class 30 gray cast iron is a brittle material, its mechanical
behaviors were defined by using both the Elasticity model and Cast Iron Plasticity model in
Abaqus CAE. Material properties such as density, Young’s modulus, Poisson’s ratio, and
hardening curves under tension and compression were manually input into Abaqus.
Figure 2-13: (a) Alcoa challenge original loading conditions [87]; (b) Assumed loading direction
in this case study.
TO was performed on the bracket with a goal of reducing its volume by 60%. The output
geometry was extracted as an STL file (see Figure 2-14b). The refinement and revision were
performed in SolidThinking Inspire 2016 and SolidWorks 2016. A solid body was generated
45
based on the extracted mesh by using the PolyNURBS tool in Inspire. PolyNURBS is a powerful
technique to create smooth freeform solid bodies from meshes. Geometrical revisions were done
afterward on this solid body using the same tool based on previously proposed redesign rules,
including undercut, draft, uniform section, rounded edges, wall-to-wall thickness, and hole size
(see Figure 2-15).
Inspire had limited CAD capability to efficiently create sharp and accurate features that
can be integrated with the solid body generated by PolyNURBS. Therefore, after the revision, the
solid geometry was saved as a STEP file and then imported into SolidWorks. Loft, flat surfaces,
accurate cuts, and fillets were added to finalize the redesign (see Figure 2-14c). The mechanical
performance of the optimized geometry was evaluated via FEA in Abaqus CAE and compared
with the target geometry (see Figure 2-16). According to the FEA results, the optimized design
achieved a 30% improvement in the safety factor with a 50% reduction in product volume
compared to the target design. The design and simulation results are summarized in Table 2-3.
Figure 2-14: (a) Bearing bracket design envelop; (b) TO output geometry; (c) optimized design.
46
Figure 2-15: Check for geometry based on redesign rules.
Figure 2-16: FEA of (a) the target design and (b) the optimized design.
Table 2-3: Comparison of the performance of the target and the optimized designs.
2.4.2. Rigging design and casting
The pattern and rigging design was completed in SolidWorks and validated via
solidification simulation in SOLIDCast 8 (Finite Solutions Inc, Butler County, Ohio). For the
convenience of post machining the final casting, all bolt holes were sealed, and four blind holes
with 3 mm diameter and 1 mm depth were placed to mark the hole locations. Similarly, the hole
originally holding the bearing was also replaced by a through hole with 7 mm diameter. In
47
addition, 4 mm of machining tolerance was added to the bottom of the bracket, which was
designed to be milled after casting to leave a smooth flat surface. The last step of the pattern
design was to add casting shrinkage allowance. The rigging system, including a sprue, a runner,
an ingate, and a riser, was then designed and attached to the pattern geometry (see Figure 2-17a).
The initial parameters of the rigging were obtained by using the Gating and Riser Design Wizards
in SOLIDCast (see Figure 2-17b). The main purpose of solidification simulation was to predict
potential casting defects, especially shrinkage porosities resulting from hot spots. The rigging
design was adjusted until all hot spots on the cast part were shifted to the riser location.
Figure 2-17: The (a) rigging design and (b) solidification simulation for the optimized bracket.
After the pattern and rigging design, sand mold assembly, including a cope, a drag, and a
core were created in SolidWorks. The mold and core were 3D-printed by an ExOne S-Max™
system (ExOne, North Huntingdon Township, PA) in The University of Northern Iowa Metal
Casting Center (UNI MCC). After printing, the loose sand powder was removed from the mold
using compressed air (see Figure 2-18a). ASTM A48 Class 30 gray cast iron was melted in a
furnace to 2200 °C and poured into the mold cavity. After the metal had completely solidified and
cooled, the casting was cleaned, and the riggings were removed. The cast bracket was further
processed on a CNC machine for bolt hole drilling and bottom surface milling. The final cast part
is shown in Figure 2-18b.
48
Figure 2-18: (a) 3D-printed sand mold; (b) the final cast part.
2.4.3. Mechanical testing
Mechanical testing was performed on a QTest Elite 100 tensile testing machine (MTS
Systems Corporation, Eden Prairie, MN). A pair of fixtures were designed and manufactured to
hold the gray cast iron bracket between the tensile grips. The base of the bracket was fastened on
one fixture by four bolts and the bracket front was pin connected with the other fixture (see
Figure 2-19). All the auxiliary components were made of steel much stronger than gray cast iron.
The response of the bracket during the testing is shown in Figure 2-20. FEA was also performed
to predict the failure of the bracket under the mechanical testing conditions and compare with the
physical testing. The failure location predicted in FEA is correct as shown in Figure 2-21. In
addition, the FEA result and the mechanical testing result were found to have a good match in the
ultimate tensile strength (UTS) and the displacement of the bracket (see Table 2-4). In
conclusion, the FEA results obtained were reliable and the design rules applied in this case study
were validated.
Figure 2-19: (a) Fixtures designed for the mechanical testing; (b) setup of the tensile testing
machine.
49
Figure 2-20: The response of the bracket in the mechanical testing.
Figure 2-21: The failure location of the bracket in (a) FEA simulation and (b) the mechanical
testing.
Table 2-4: Comparison of the UTS and displacement between the FEA and mechanical testing
results.
2.5. Discussion
This is, to our knowledge, the first study on the application of TO and design guidelines
to part design for sand casting via 3D-printed molds. The AM of sand molds is enabled by a
binder jetting process, called 3D Sand Printing (3DSP), thereby introducing a new era of design
50
and casting of metal. A new cast part design process for the 3DSP-based casting was proposed.
Part design rules and manufacturing constraints were discussed and compared with those of
conventional casting processes. To summarize, designers no longer need to worry about many of
the traditional design constraints related to mold making, namely, draft and undercuts. Other rules
considering wall thickness, rounded edges, intersections, etc. are still valid in the new casting
process; however, because of the use of 3DSP, these traditional rules can be applied more
conveniently and more cost-effectively. In the case study, TO and revisions following the new
design rules were successfully applied to redesign a grey cast iron bearing bracket. Despite a 50%
reduction in the bracket weight, a 30% improvement in mechanical performance was observed for
the optimized design according to FEA, whose validity was supported by the results from the
mechanical testing on the actual cast bracket. In this case, the cost for the 3D-printing of the sand
mold was $80 estimated by UNI MCC, and the estimated lead-time of the entire design-AM-
casting process was only 2 days. It should be noted that multiple cores and molds were printed
along with the molds for this study. Hence, the AM cost is amortized over multiple molds. In
comparison, casting of the same bracket following traditional process requires multiple complex
cores because the bracket has many holes and undercuts; as a result, it can take weeks to prepare
a mold and cores, not to mention the high tooling cost related to the fabrication of the pattern and
core boxes.
2.6. Future Work
The current chapter applied casting and 3DSP constraints after TO. Future work should
include the development of TO algorithms with integrated manufacturing constraints. Due to
almost unlimited geometric freedom of molds that can be printed via 3DSP, efforts also need to
be focused on the investigation of unconventional mold design, such as new structures of sprue
and runners, irregular parting surfaces, and the optimization of risers.
51
This Chapter is a manuscript submitted to the International Journal of Metalcasting (IJMC)
Chapter 3
REDESIGN AND TOPOLOGY OPTIMIZATION OF TRADITIONAL
METAL PARTS FOR INVESTMENT CASTING WITH 3D-PRINTED
WAX PATTERNS
Traditional investment casting, also known as lost-wax casting, suffers from high tooling
cost and long lead-time during fabrication of wax patterns. In recent years, rapid investment
casting based on AM become widely accepted in industry due to rapid production of patterns
without any tooling requirements. Direct AM production of wax patterns is not only more cost-
effective for low volume production, but also capable of creating freeform and highly complex
geometries that are extremely difficult or too expensive to cast conventionally. Such advantages
provide unlimited opportunities in the cast part design and enable the use of advanced design
tools such as topology optimization (TO). In this study, we develop knowledge-based design
guidelines for rapid investment casting through novel integration of TO with design for
investment casting and design for AM principles. Specifically, TO routine based on Solid
Isotropic Material with Penalization (SIMP) method using Abaqus Topology Optimization
Module (ATOM) has been developed for both material extrusion and investment casting
constraints. Part design optimization is demonstrated for this new technique and is validated by a
case study using heat-treated ASTM A216 WCB cast carbon steel parts. AM patterns for
investment casting were produced using commercially available wax filaments. Wax material
extrusion parameters are identified for optimal AM accuracy and surface finish.
3.1. Introduction
52
Additive Manufacturing (AM), commonly known as Three-dimensional Printing (3DP),
is defined as the process of joining materials to make objects from 3D model data, usually layer
upon layer, as opposed to traditional subtractive manufacturing methodologies (i.e., milling or
drilling) [1]. After more than 20 years of research and use, a number of AM processes have been
developed which can be divided into 7 categories: vat polymerization, sheet lamination, material
extrusion, material jetting, binder jetting, powder bed fusion, and direct energy deposition. As the
accuracy and the versatility of the AM processes have improved, the focus of the industry is
shifting from “Rapid Prototyping” to “Rapid Manufacturing” [84].
AM offers a number of advantages over traditional manufacturing processes, such as part
consolidation, geometric manufacturing freedom, reduction in material waste and tooling cost,
etc. It has been suggested that such advantages can lead to new drivers, i.e., economic,
environmental, or performance related, for industries to use AM for part production [89].
According to Wohlers Report 2016, the AM industry Surpassed $5.1 Billion in 2015, showing an
annual growth rate of 25.9% that year [5].
To date, a wide range of materials can be used in AM processes, including metals,
ceramics, polymers, composites, and biological systems [54]. Although direct AM of metal parts
is quite recent and under intensive research and development, indirect methods have been used
for a much longer time through the combination of AM and traditional metal casting, indeed
investment casting (IC) [16].
Investment casting, also called lost-wax casting, has a traditional manufacturing process
shown in Figure 3-1. Traditional IC starts from the design and tooling of a pattern die. Sacrificial
patterns are fabricated by injecting wax into the die and then often assembled onto a wax sprue to
form a cluster. Paraffins and microcrystalline waxes are the most widely used waxes, and they are
often used in combination due to their complementary properties [38]. The next step is shell
building, where the cluster is repeatedly dipped and coated in a ceramic slurry and followed by
53
stucco application. After the wax is completely removed by heating and burn-out, molten metal is
poured into the shell and cooled. The final step is to break off the shell and separate individual
castings for post-processing. IC, as a popular casting process, is capable of producing quality near
net shape metal parts with high geometric complexity and acceptable tolerances [90]. However,
the economic benefits of traditional IC are limited to mass production due to high tooling cost and
long lead-time during wax pattern fabrication, especially when a pattern has a complex geometry.
Hence, the application of AM in investment casting process to produce metal cast parts, namely
rapid investment casting (RIC), has been considered as a perfect solution to such drawbacks
thanks to its independence of geometric complexity.
Figure 3-1: Schematic of the traditional investment casting process [41].
AM provides a cost-effective means to rapidly fabricate complex patterns for investment
casting. The use of AM for pattern fabrication also enables the use of more advanced design
approaches, e.g., TO, to improve pattern design. TO is a powerful tool to generate light-weight
parts with improved mechanical performance. Such parts often have geometries too complex to
fabricate using traditional methods. By combining TO with AM, there is an opportunity for rapid
54
fabrication of IC patterns with improved geometries and features beyond the capability of
conventional ways.
Although a part fabricated by AM has virtually infinite geometric freedom, each AM
process has its own design considerations that need to be taken into account, such as minimum
feature size, need for support, surface finish, material strength, operation cost, among others.
Designers need to select a process that can best serve the purpose of a casting in this case. More
important, a pattern, regardless of whether it is fabricated via traditional tooling or AM, must
meet the requirements of IC for castability and casting quality. It is of importance to understand
all design considerations and manufacturing constraints of the entire RIC process at the pattern
design stage. Hence, we propose a general pattern design framework for a RIC process (see
Figure 3-2).
Figure 3-2: Framework of pattern design for a RIC process.
Material extrusion of commercial wax-like filaments recently available in the market
enables the convenient fabrication of patterns for investment casting. By using this process, we
investigate the application of TO in the wax pattern design. The final objective of this work is to
55
develop knowledge-based design guidelines for RIC using 3D-printed wax patterns. In the
following section, we review the advantages and drawbacks of current RIC processes and the
application of optimization techniques in both traditional and rapid investment casting. In the
methodology section, we propose a design flowchart for RIC with 3D-printed wax patterns
including the use of topology optimization on the initial design. Important rules are introduced to
refine the optimized design with material extrusion and IC constraints. We then present a case
study to validate the design process and rules. Conclusions and important findings are
summarized in the final section.
3.2. Literature Review
Investment casting (IC) is useful for fabricating complex metal parts when they are either
too costly or time-consuming to mass produce using other manufacturing processes, e.g., sand
casting. The main drawback of the IC process is the high investments in both time and cost for
the tooling of wax patterns, especially for those with complex geometries. This major limitation
can be overcome by the integration of AM technology in the IC process [91]. Rapid investment
casting (RIC) based on AM can easily fabricate complex patterns and allow a significant amount
of time and cost savings [92, 93]. Thanks to such benefits, the application of RIC ranges from the
casting of jewelry, sporting equipment and medical implants to high-performance parts utilized in
injection molding, die casting and the automotive and aerospace industries [90].
To date, a variety of RIC solutions have been developed by industrial and academic
researchers, and RIC solutions involving pattern fabrication can be classified into two categories:
(1) direct AM of patterns and (2) indirect AM of patterns. For the direct AM of patterns, both wax
and non-wax parts can be directly 3D-printed as patterns for IC [37, 49, 94]. In the indirect AM
approaches, IC patterns are often produced using AM-fabricated molds or soft tooling, i.e., room
temperature vulcanized (RTV) silicone rubber molding, fabricated from 3D-printed master
56
patterns [95-97]. In comparison, the direct AM approach has advantages over the indirect
approach when the casting geometry is complex or has internal cavities due to the risk of
damaging the mold during the removal of patterns [98].
The first use of AM-fabricated patterns in traditional IC dates back to 1989 [99]. Since
then almost all commercialized AM techniques and systems have been tried to produce IC
patterns with varying success. Among them, Selective Laser Sintering (SLS), Stereolithography
(SLA), Fused Deposition Modelling (FDM), Model Maker II (MM II) and ThermoJet have
shown feasibility when producing wax patterns that can be directly used for IC [37, 94, 100-102].
But most of these wax patterns were found to be so brittle that transporting them to foundries
could cause damage [94]. Non-wax patterns fabricated by different AM systems have also been
employed for RIC, such as epoxy QuickCast patterns via SLA [103] and acrylonitrile butadiene
styrene (ABS) patterns via FDM [104]. Non-wax patterns generally have superior rigidity and
strength compared to wax, which allows casting of thin walls and finishing operations for
improved surface quality. However, non-wax patterns tend to cause new problems such as shell
cracking, incomplete pattern burning out, and residual ash [90].
In contrast, novel wax-based thermoplastic polymers seem to be a potential solution for
pattern material for IC. This type of material has a low melting point like conventional wax but
remains adequate mechanical strength and durability at room temperature. This wax-like material
can be easily manufactured as a filament to 3D-print patterns via material extrusion, one of the
most popular and affordable AM processes. When heated, the pattern material will melt and flows
out of the shell like hot paraffin wax, leaving a near-perfect mold for casting. Another potential
advantage of using wax-like filament is that material extrusion, e.g., FDM, has been found to
provide accuracies that are equal to or better than SLA and PolyJet and better than those of SLS
[105].
57
Besides lead-time and production cost reduction, material saving and part optimization
are also important goals in industry. TO is an advanced tool to achieve such goals. Although TO
has been well developed for decades, it has not been widely applied in industry due to the limited
manufacturability of TO parts with conventional fabrication processes. However, TO relates well
with AM due to its “freeform” aspect, which does not require any extra tooling to produce
complex geometries generated by TO.
For RIC, early studies focused on optimizing the infill structure of 3D-printed IC
patterns. Zhu, et al., [106] applied topology optimization in the design of inner pattern structure to
solve the crack problem of the ceramic shell in SLA based RIC. For the same purpose, a topology
optimization procedure in combination with thermo-mechanical finite element analysis was
introduced by Gu, et al. [107] into the lattice structure configuration design for RIC patterns.
More recently, industries have started to explore the optimization of entire cast part topology for
RIC [108, 109]. Yet, limited information can be found regarding how to integrate topology
optimization, AM and IC constraints into the AM-fabricated pattern design process. It is the time
to exploit the design freedom of this RIC process and develop step-by-step design rules for
guidelines.
3.3. Methodology
The pattern design for a RIC process always starts with a 3D computer-aided design
(CAD) model representing the geometry of a metal part. The 3D model could be produced in
CAD software, e.g., SolidWorks, or obtained through reverse engineering of 3D-scanned point
cloud data. In addition, important part and application information, such as casting material,
loads, and boundary conditions must be known. A proposed design flowchart for AM-fabricated
IC patterns is shown in Figure 3-3, which involves the following important steps.
58
Figure 3-3: Design flowchart for 3D-printed IC patterns.
3.3.1. Step 1: TO of the initial CAD
The first step is the TO of the initial cast part geometry, which could be done by using
commercial or educational software tools such as Altair OptiStruct, Abaqus ATOM (Simulia
TOSCA) or SolidThinking Inspire [32]. The first step of TO is to identify the design space and
the non-design space, referring to the region that needs to be optimized and that should remain
unchanged, respectively. The setup for a common TO task includes defining material properties
and interactions, adding loads and boundary conditions, and determining objective(s) and
constraints. For a mechanical loading application, the typical objective is to minimize the
compliance or maximize the stiffness of the system, and the most important constraint is the
volume fraction [22]. Figure 3-4 illustrates an example of the TO task for a cantilever beam.
59
Figure 3-4: Topology optimization of a cantilever beam.
3.3.2. Step 2: Redesign of the optimization output
The second step is the refinement and revision of the TO output geometry. Refinement is
necessary because the TO results are often a collection of discretized elements with coarse
surface finishes. Revisions are needed to modify the result for a geometry suitable for fabrication
and casting. The redesign procedure could vary depending on the capabilities of the different
design tools. Designers need to have a comprehensive understanding of the advantages and
disadvantages of different design tools and determine the best option for the redesign. Figure 3-5
gives an example of making a smooth truss from a TO output geometry using the PolyNURBS
tool in SolidThinking Inspire. Typically, three important considerations should be taken into
account simultaneously during redesigning, including part requirements, IC design rules, and
design for material extrusion AM principles. Detail explanation of the three considerations and
corresponding redesign rules are given in steps next.
60
Figure 3-5: Generation of a smooth solid truss from TO output using Inspire PolyNURBS.
(1) Part requirements
The final redesign should meet some fundamental requirements for a mechanical part
design. The part requirements for a casting can be divided into two categories:
• Part rationale: TO outputs are generated relying solely on a mathematical
optimization process. Therefore, there is no guarantee that the resulting geometries
make physical sense for practical applications. For instance, TO outputs sometimes
may have isolated elements disconnected from the main body. The disconnection of
elements could occur within the design space or between the design space and the
non-design space. In such cases, designers need to make the decision to either
remove the isolated elements or manually connect them to the main body for the
integrity of the part geometry.
• Geometric requirements: The redesign should be subject to user-defined requirements
of the part geometry. The most important geometric requirement for the redesign
might be the volume or weight fraction constraint. Other common requirements in
part geometry design could refer to symmetricity of structure, and form of shape
(mechanical vs. organic).
(2) IC design rules
IC is able to produce more complex metal parts than other casting methods, e.g., sand
casting. For example, no draft angle has to be factored into IC because wax shrinkage makes it
easy to remove them from pattern dies. Yet, a few design rules should still be followed for good
casting quality.
• Wall thickness: If the molten metal needs to travel a long distance in a shell, then the
thickness of the cavity must be kept above a minimum thickness to avoid incomplete
61
casting. The wall thickness for investment casting can be as low as 0.5 mm in some
cases due to a high-temperature shell mold. At the same time, walls with large
thickness should be avoided to reduce the solidification time and the casting volume.
• Uniform section: For any casting processes, the uniformity of sections in cast parts is
important to prevent uneven cooling and occurrence of hot spots. Abrupt changes in
section thicknesses and the intersection of many sections at one point should be
avoided. If this is not possible, then the transitions must be kept gradual and smooth.
• Rounded edges: Sharp edges and corners on castings might cause stress
concentrations and tears, as well as hot spots; therefore, they should be replaced by
rounded edges and fillets.
• Holes: To ensure a successful shell building process, IC has special requirements on
the maximum ratio of hole depth to the hole diameter for patterns. Depending on the
hole size, the allowable maximum ratios range from 2:1 to 6:1 and 1:2 to 2:1 for
through and blind holes, respectively. All hole edges should be rounded to increase
the core strength and prevent shell fracturing during the metal pouring.
• Directional solidification: In IC, a sprue often functions as a feeder for all castings
attached to it through gates. Therefore, it is preferred that cast parts and gates are
designed to induce a directional solidification. Ideally, a cast part should allow only
one heavy section where a gate can be conveniently placed during pattern design and
removed after casting, and the cast part should have a geometry with its thinnest
sections placed farthest from the gate.
(3) Design for material extrusion rules
Material extrusion, like all other AM systems, has constraints that influence the printing
results. In addition, wax-like filament is clearly different in its properties from other common
62
material extrusion thermoplastics such as PLA and ABS. It is important to understand both
limitations of the process and the material and take them into account during the pattern design
and 3D-printing. Here, we present some important design for AM rules related to the 3D-printing
of wax-like patterns via material extrusion:
• Build volume: Material extrusion is one of the most widespread and inexpensive AM
systems, but normally with small-to-medium build volumes. A wax-like pattern for
IC must be able to fit in the build chamber of a material extrusion AM system.
• Extrusion temperature: For a material extrusion process, the temperature of the
extruder nozzle is an important factor that affects the printing quality. Ideally, it
should be kept in a range that is high enough to form sufficient bonding between
layers (see Figure 3-6) and low enough to avoid filament “oozing”. The optimal
extrusion temperature might differ from one case to another because of the variations
in filaments, printers, printing speeds, room temperatures and some other factors.
Before the printing of patterns, trials should be done to find a proper extrusion
temperature one should use.
Figure 3-6: Influence of extrusion temperature on the surface finish of a 3D-printed wax pattern:
(a) under proper temperature, (b) under insufficient temperature.
63
• Dimensional accuracy: Dimensional accuracy is a measure of a 3D printer’s ability to
fabricate parts with expected dimensions. Dimensional accuracy in x, y, and z axes
can be easily tested by comparing the actual and expected dimensions of a 3D-printed
cube (see Figure 3-7). For a precise IC process, a 3D printer must be calibrated
before printing wax-like patterns.
Figure 3-7: Testing of dimensional accuracy by 3D-printing a cube.
• Printing speed: The printing speed of a 3D printer defines the velocity (i.e., in mm/s)
of the extruder nozzle while printing. A faster printing speed reduces the total build
time, but it is often associated with lower print quality and accuracy. Given that the
quality of an IC pattern’s surface is more important than that of its inner structure, the
3D printing process should have a lower shell printing speed but allow a high
printing speed for infill and support.
• Minimum feature size: Minimum feature size is the smallest size of a feature, such as
a thin wall, that can be accurately 3D-printed and durable enough for use. Designed
features on an IC pattern must have a size larger than the minimum feature size of the
material extrusion system. Minimum feature size, as well as many other printing
parameters, can be tested by printing well-designed benchmarks. Figure 3-8 gives an
example of a testing benchmark free for download from www.thingiverse.com.
64
Figure 3-8: A benchmark for testing 3D printing features such as minimum feature size,
minimum wall thickness, surface finish, bridging and overhangs: (a) 3D model, (b) printed part.
• Layer height: The layer height influences the build time and surface quality. To print
a pattern with a good surface finish, the layer height should be kept as small as
possible. However, it should be noted that if the layer height is too small, then the hot
nozzle might re-melt and deform the nearby printed wax when it squeezes the top
layer (see Figure 3-9), especially considering that the wax-like material solidifies
much slower than other common material extrusion materials, e.g., PLA.
Figure 3-9: Quality of prints (a) with proper layer height and (b) when the layer height is too low.
• Support: Material extrusion is an AM process that needs support structures for large
overhangs. The rule of thumb for any AM process that needs support is to design a
part that needs as little support structure as possible. Except for the benefits of
material savings and ease of post-processing, such a rule is particularly important for
65
the material extrusion of wax-like filament. One reason is that the supported surfaces
of a wax-like part are often deformed and have a rather coarse finish (see Figure 3-
10b). In addition, the wax-like material is not as strong or rigid as PLA, and it needs
a much longer time to get to the maximum rigidity after extrusion, which makes the
support structures unstable while printing. Therefore, support, if needed, should have
a higher density and avoid tall and thin structures. Never build a heavy or complex
section based only on support structures.
Figure 3-10: Testing of support for wax-like filament: (a) an arch structure printed with support,
(b) bottom surface after the support is removed.
• Shell thickness: Because material-extruded wax-like objects might have small holes
and gaps on the surface, thick shells are generally recommended to reduce the
chance of slurry infiltration (see Figure 3-11) into the pattern body during shell
building.
66
Figure 3-11: Defects of a cast part resulting from slurry infiltration.
• Infill density: Generally, 100% infill density is preferred for 3D-printed wax-like
patterns, especially those with complex geometries, to prevent slurry infiltration.
However, if wax coating will be applied to seal up the pattern surfaces, then a lower
infill density is acceptable.
• Build direction: Selection of build direction, or part orientation, is a very important
and often the initial step for any AM process. It influences a printed part in many
aspects, such as the build time, printing accuracy, surface finish, support structures,
and mechanical strength. Figure 3-12 illustrates the impact of build direction on a
truss of a 3D-printed part. Designers need to trade off multiple printing features and
decide the best build direction based on the geometry of and requirements for a
printed part. For an IC pattern, high printing accuracy and good surface finish might
be the most important considerations. The area of surfaces that need support must be
minimized because of inadequate printing quality on those surfaces.
Figure 3-12: Influence of build direction to a 3D-printed truss.
67
• Warping and curling: Warping or curling is a common problem encountered in
material extrusion due to the thermal deformation of the printing material. As the
extruded thermoplastics cool down, they contract and induce stress along the print’s
lateral surfaces, especially at sharp corners. Such stress pulls the top layers inward
and causes the printed part to deform upward. Wax-like filament has almost no
warping issues when the printed part has no or small overhang angles. However,
when there are large overhangs, curling occurs (see Figure 3-13) and needs to be
manually flattened layer by layer to achieve a good surface quality. Such curling
issues seem hard to eliminate during the printing stage due to the slow solidification
speed of the wax. Instead, minimizing overhangs and sharp corners during the pattern
design stage could be a strategy for controlling curling.
Figure 3-13: Curling of wax layers.
3.3.3. Step 3: FEA validation
Due to the geometric modifications, the mechanical performance of the redesigned cast
part could differ from that of the TO output. Thus, to make sure a final casting meets its design
goals and requirements, FEA should be performed for design validation. If the FEA result could
not meet all design goals and requirements, then designers need to revise the design further or
even repeat TO until a satisfactory result is obtained.
68
3.3.4. Step 4: Patterning of the optimized design
After FEA validation, necessary allowances are added onto the optimized geometry to
finalize the pattern design. If any post-processing, such as finishing, milling, and drilling, is
needed on the cast part, then machining allowances should be added to the corresponding part
region. In addition, shrinkage allowance must be used to compensate for the shrinkage of
solidified metal during cooling. Distortion allowance might also be needed for irregular cast
geometries to correct the distortions of the cast part induced by stresses developed when the solid
metal is cooled. For the convenience of cluster assembly, sometimes the pattern together with its
gate(s) and runner(s), could be fabricated as an integrated part.
3.4. Case Study
3.4.1. TO and FEA validation
In this case study, a metal benchmark was topologically redesigned for a simple
mechanical loading application and fabricated via wax material extrusion. TO, design for AM and
IC principles were implemented. The 3D-printed wax parts were used as sacrificial patterns to
produce final castings through investment casting. The metal benchmark used here had the same
design envelop and boundary conditions as the Alcoa airplane engine bracket [87] but modified
part material and loading condition. For the convenience of mechanical testing validation, we
assumed that the target bracket was under only one horizontal load 𝐹 which equals to the
maximum load the initial bracket can stand before yield (safety factor = 1.0) (see Figure 3-14).
ASTM A216 WCB cast steel was chosen as the material of the target bracket for both topology
optimization and casting.
69
Figure 3-14: (a) The target design envelope and (b) load and boundary conditions of the bearing
bracket.
The CAD model of the bearing bracket was imported as a STEP file into Abaqus CAE
6.14. To simplify the bracket model, the rigid spherical bearing was removed from the assembly
and the horizontal load was directly applied to the geometric center of the surface originally
interacting with the bearing. A kinetic coupling interaction was added between the surface and its
center to preserve the shape of the surface and force the surface to move with its center under
loading. The fixed boundaries were defined as the four holes where the bolts were mounted.
TO was performed on the bracket geometry using Abaqus Topology Optimization
Module (ATOM), a tool combining SIMULIA TOSCA structure optimization and Abaqus FEA.
The non-design space included four bolt holes and the region originally holding the bearing. The
rest of the bracket geometry was defined as the design space. Three topological optimized
geometries were generated with 40%, 55%, and 70% volume reductions, respectively (see Figure
3-15a). The output geometries were extracted as STL files and individually imported into
SolidThinking Inspire 2016 for redesigning. By using PolyNURBS, a powerful tool to create
free-form smooth geometries, refined structures were generated based on the extracted meshes.
Bridging structures were manually created to connect the isolated bolt holes to the main bodies.
Geometrical revisions were performed afterward following the previously mentioned redesign
70
rules, including wall thickness, rounded edges, hole size and directional solidification (see Figure
3-16).
After the major revisions were completed, the solid geometries were saved in a STEP
format and then imported into SolidWorks 2016. Mechanicals shapes, including lofts, flat
surfaces, and accurate cuts were added with fillets to finalize the bracket redesigns (see Figure 3-
15b). FEA for all redesigns were then performed in Abaqus CAE for mechanical performance
evaluation. Figure 3-17 showed the comparison of FEA results between the target design and
three redesigns.
Figure 3-15: (a) TO output geometries and (b) the final redesigns.
71
Figure 3-16: Check for geometries based on redesign rules.
Figure 3-17: Comparison of safety factor of the target design and redesigns: (a) 1.0, (b) 1.3, (c)
1.4 and (d) 1.2.
3.4.2. Gating design and casting
Pattern design was then performed for the redesigned bracket geometries in SolidWorks.
A 3-mm machining allowance was added to the bottom of each bracket for milling the bottom flat
after casting. All bolt holes were replaced by blind holes with 3-mm diameter and 1-mm depth as
72
makers for hole drilling. Similarly, the spherical cavity on each bracket for its bearing was also
replaced by a through hole with 5-mm diameter. In addition, a 2% global shrinkage allowance for
cast steel was applied to all geometries. To find the best location to add gates, solidification
simulations were carried out for all brackets by using Finite Solutions SOLIDCast 8 (see Figure
3-18). Gates were all added on flat surfaces close to hotspots with taper to induce directional
solidification. A pattern for the target bracket geometry was also designed. However, because
casting processes do not favor sharp edges, before adding the allowances and the gate, a 1-mm
fillet was added to all edges on the target bracket geometry. The final pattern designs for four
brackets are shown in Figure 3-19.
Figure 3-18: Solidification simulation for (a) target with 1-mm fillet, (b) redesign 1, (c) redesign
2 and (d) redesign 3.
73
Figure 3-19: The pattern design for bracket geometries: (a) target with 1 mm fillet, (b) redesign
1, (c) redesign 2 and (d) redesign 3.
A commercial wax-like filament, MOLDLAY (Kai Parthy CC-Products, Koeln,
Germany), was used to fabricate the patterns on a MakerBot Replicator+ 3D printer (MakerBot
Industries, LLC, Brooklyn, NY). MOLDLAY has good dimensional stability and adequate
rigidity at room temperature. When being heated to 270 ℃, this wax-like material melts and
flows out of the shell like hot paraffin wax, leaving a near-perfect mold for casting. The printing
parameters were determined based on the aforementioned design for AM rules combined with
pilot testing results (see Table 3-1). An optimal build direction was chosen to minimize the need
for support structures and to ensure good printing quality for most of the part surfaces. The
printed patterns (see Figure 3-20a) were then packed and shipped to Tech Cast Inc for investment
casting. To avoid any ceramic infiltration in the patterns, a thin, low viscosity wax was used to fill
all major holes in the outer shell of the patterns. The patterns were then sealed with polyurethane.
The amount of polyurethane used was carefully controlled to be just enough to smooth the
surface finish; therefore, its effect on the pattern dimensions could be negligible. A shell mold
was successfully built by using these 3D-printed patterns without any shell cracking or residue
ash issues. After the castings were cut and cleaned, heat treatment was applied including a
normalizing process (925 ℃ for 1.5 hrs followed by an air quench) and a tempering process
74
(675 ℃ for 2 hrs followed by an air cool). The final cast steel brackets after post-processing are
shown in Figure 3-20b. A comparison of all designs in safety factor, design volume, and physical
weight are summarized in Table 3-2.
Table 3-1: Parameters for the material extrusion of wax-like patterns.
75
Figure 3-20: (a) 3D-printed wax-like patterns; (b) final cast carbon steel brackets after post-
processing.
Table 3-2: Comparison of the performance of the target design and the optimized designs.
3.5. Discussion
This is, to our knowledge, the first study on the application of TO and design guidelines
to pattern geometry design for RIC. Despite adequate geometric complexity allowed by
traditional IC, the high cost of pattern die tooling constrains the application of conventional IC to
mass production. By using AM technologies, complex patterns can be rapidly fabricated with
reduced tooling cost and production time.
In order to make the best use of the geometric design freedom provided by this method, it
is important to fully understand the manufacturing constraints involved in this RIC process. In
this study, we developed design guidelines for wax IC patterns fabricated via material extrusion.
TO was applied and followed by geometric redesign based on a set of design rules for AM and
IC. A commercial wax-like filament was tested for the optimal printing quality and used for the
fabrication of patterns. Cast metal parts were successfully obtained by investment casting using
these 3D-printed patterns. In spite of a reduction in weight, all redesigns were shown via FEA to
have better mechanical performance than the target design.
The important conclusions of this study include:
• By combining TO of cast parts and AM of wax patterns, there is significant potential
76
for the casting of lightweight and structurally optimized metal components for
mechanical applications.
• Design for AM and IC principles play a crucial role in the design of wax patterns for
RIC. Designers need to trade off different rules and determine the optimal design
procedure to achieve the desired design requirements of patterns.
• Based on an estimate provided by Tech Cast Inc, material extrusion of wax-like
patterns can reduce both the cost and lead-time of an IC process when production
volume is low. In our case study, the cost of the RIC process, including the 3D-
printing of the wax-like patterns, post-processing, and casting, was approximately
$100 for each bracket; and the lead-time of the wax-like pattern fabrication was less
than 2 weeks. In contrast, design and fabrication of traditional pattern dies for the
same brackets can take 40-50 weeks, and cost approximately $40000 in total.
Limitations of this RIC process include:
• Material extrusion is a relatively slow AM process; thus, this RIC solution is not
ready yet for mass production.
• Material extruded wax patterns have lower geometric consistency, dimensional
accuracy, and surface quality compared to injection molded wax patterns and often
need post-processing such as support removal, finishing, and wax coating.
• Current wax filaments suffer from a few drawbacks, such as slow solidification,
insufficient rigidity, and curling, thereby limiting the design freedom and the printing
quality of patterns.
3.6. Future Work
Due to the limitations of current TO design tools, the pattern design process in this study
applied IC and AM constraints after TO generation of the design geometry. There is a need for
77
new design approaches that can perform TO with manufacturing constraints integrated into the
optimization process. In addition to the geometric freedom provided by the use of material
extrusion to fabricate wax-like patterns for RIC, future investigations should validate other
benefits of this process. Systematic comparison of cost-effectiveness and time-efficiency should
be made between conventional and AM-based IC processes for different pattern complexities and
production volumes. Such comparison can help designers for IC process decision-making. Future
efforts should also be put on the development of wax filament with ideal thermal and mechanical
properties for IC pattern AM.
78
Chapter 4
CONCLUSIONS AND FUTURE WORK
4.1. Conclusions
Rapid Casting (RC) provides a solution that combines the advantages of both additive
manufacturing (AM) and traditional metal casting processes. On one hand, the use of AM
provides traditional metal casting with advances such as increased product complexity, easier part
design and optimization, reduced tooling cost, and shortened lead-time. On the other hand, rapid
casting is often cheaper than direct metal AM and can produce isotropic metal parts. The cast part
design rules for traditional metal casting have become well-known standards, but to date limited
information exists about the rules for RC processes. It is of importance for design engineers to
have comprehensive part design guidelines to follow to maximize the results of RC processes.
This thesis provides the first known investigation into the application of topology
optimization (TO) to part design for RC processes. Two RC processes are described in this work:
(1) sand casting with 3D-printed mold via 3D Sand Printing (3DSP) and (2) investment casting
(IC) using 3D-printed wax-like patterns via material extrusion. General part design guidelines are
proposed, including detailed analysis of related design rules. These design guidelines can be
applied to any sand or investment cast parts for mechanical loading applications. The major
findings of this thesis include:
• 3DSP is a revolutionary technique that introduces a pattern-less sand casting process.
The sand mold fabricated via 3DSP is able to produce metal castings with improved
geometric freedom. First, traditional pattern-making related geometric constraints are
eliminated. Second, 3DSP can easily fabricate complex sand cores together with
molds, which leads to a fast and cost-effective method to form complex internal part
79
geometries. However, attention needs to be paid to the wall-to-wall thickness and
hole size on a cast part because they are constrained by the minimum feature size of
the 3DSP process.
• It is feasible to investment cast quality metal parts using patterns fabricated via
material extrusion of wax-like filament. Selecting proper 3D printing parameters is
crucial for getting good patterns in terms of surface finish, geometric accuracy, and
part strength. Slurry infiltration can be prevented by applying sealing operation using
wax or polyurethane after printing. In addition, no shell cracking issues have been
observed so far.
• The implementation of TO in the cast part design for RC enables the casting of metal
parts with reduced weight yet improved mechanical performances. However, TO
outputs are generally not ready for casting but need geometric refinement and
revision based on corresponding manufacturing constraints.
• Different RC processes could have different design constraints depending on the AM
and casting processes adopted. Before designing, it is important to have a thorough
understanding of a specific RC process and the constraints related to it. It should be
noted that different RC processes still share some mutual design rules such as
uniform wall thickness, rounded edges and shrinkage allowance. These rules are
typically inherent rules for all metal casting processes.
Meanwhile, this thesis has the following limitations:
• This thesis focused on part design freedom provided by AM for the traditional
casting process. It does not take into consideration the impact of AM in some other
aspects, e.g., casting quality.
• Although TO is a mature design approach, using current TO and CAD software to
80
design parts for casting is still time-consuming. Multiple commercial tools are used
iteratively to get acceptable final designs, which might lead to cost and time
justification problems.
• The DfAM rules for material extrusion of wax-like patterns are summarized based on
experiments using a currently available wax filament. Therefore, some of the
conclusions might not be applicable for future wax products.
4.2. Future Work
Based on the findings and limitations of the current thesis, future research should further
explore the future opportunities, benefits, and needs of RC and address some existing problems:
• 3DSP provides almost unlimited geometric freedom for sand molds and cores.
Investigations are needed to explore the benefits of unconventional mold designs,
such as irregular parting surfaces, and novel rigging systems, for optimization of the
casting process.
• Future work should continue to seek other possibilities to optimize the print quality
of wax-like patterns, such as reducing surface holes and gaps by increasing overlap
between printing toolpaths, and using dual extruder systems to printed wax patterns
supported by stronger and more stable materials.
• There is a need for a TO algorithm with integrated RC manufacturing constraints.
Despite the existence of a few TO algorithms under casting constraints, those rules
are neither effective enough to use nor suitable for RC processes. Also, more
advanced design tools should be developed in the future to reduce the current design
steps and time.
• AM can be integrated into a casting process in at least three ways: (1) direct AM of
patterns, (2) AM of tooling for patterns, and (3) direct AM of molds. Given the
81
number of different AM methods and casting processes, there are potentially a wide
variety of RC solutions with different manufacturing constraints. Therefore, more
research should be done to understand how to update or expand the current design
rules for other RC processes.
• In-depth comparisons are needed for different RC processes based on part
complexity, tooling cost, and production time to fully understand their advantages
and disadvantages. This will be helpful for the development of decision-making
strategies for design engineers to choose the best RC solution for a specific casting
problem.
82
REFERENCES
[1] ASTM, "Standard Terminology for Additive Manufacturing Technologies," in ASTM
F2792-10e1, ed. West Conshohocken, PA: ASTM International, 2012.
[2] I. Gibson, D. Rosen, and B. Stucker, Additive manufacturing technologies: 3D printing,
rapid prototyping, and direct digital manufacturing. New York, NY: Springer, 2014.
[3] A. Gebhardt, Rapid Prototyping–Rapid Tooling–Rapid Manufacturing. Cincinnati, OH:
Hanser Publications, 2007.
[4] S. L. Ford, "Additive manufacturing technology: Potential implications for US
manufacturing competitiveness," Journal of International Commerce and Economics,
2014.
[5] T. Wohlers, Wohlers report 2016: Gloabal Reports. Belgium: Wohlers Associates, Inc,
2016.
[6] T. Wohlers, Wohlers report 2017. Belgium: Wohlers Associates, Inc, 2017.
[7] T. Wohlers, Wohlers report 2012. Belgium: Wohlers Associates, Inc, 2012.
[8] B. N. Turner, R. Strong, and S. A. Gold, "A review of melt extrusion additive
manufacturing processes: I. Process design and modeling," Rapid Prototyping Journal,
vol. 20, pp. 192-204, 2014.
[9] J.-P. Kruth, M.-C. Leu, and T. Nakagawa, "Progress in additive manufacturing and rapid
prototyping," CIRP Annals-Manufacturing Technology, vol. 47, pp. 525-540, 1998.
[10] J.-P. Kruth, "Material incress manufacturing by rapid prototyping techniques," CIRP
Annals-Manufacturing Technology, vol. 40, pp. 603-614, 1991.
[11] G. Manogharan, M. Kioko, and C. Linkous, "Binder jetting: a novel solid oxide fuel-cell
fabrication process and evaluation," JOM, vol. 67, pp. 660-667, 2015.
83
[12] B. Stucker, "Additive manufacturing technologies: technology introduction and business
implications," in Frontiers of Engineering: Reports on Leading-Edge Engineering From
the 2011 Symposium, National Academies Press, Washington, DC, Sept, 2012, pp. 19-21.
[13] Custompart.net, "3D Printing-Three Dimensional Printing (3DP). Retrieved from:
http://www.custompartnet.com/wu/3d-printing," ed, 2017.
[14] S. H. Huang, P. Liu, A. Mokasdar, and L. Hou, "Additive manufacturing and its societal
impact: a literature review," The International Journal of Advanced Manufacturing
Technology, pp. 1-13, 2013.
[15] Y. Huang, M. C. Leu, J. Mazumder, and A. Donmez, "Additive manufacturing: current
state, future potential, gaps and needs, and recommendations," Journal of Manufacturing
Science and Engineering, vol. 137, p. 014001, 2015.
[16] D. Kochan, C. C. Kai, and D. Zhaohui, "Rapid prototyping issues in the 21st century,"
Computers in industry, vol. 39, pp. 3-10, 1999.
[17] C. C. Seepersad, "Challenges and opportunities in design for additive manufacturing," 3D
Printing and Additive Manufacturing, vol. 1, pp. 10-13, 2014.
[18] J. Gausemeier, N. Echterhoff, M. Kokoschka, and M. Wall, "Thinking ahead the Future
of Additive Manufacturing–Analysis of Promising Industries," Study for the Direct
Manufacturing Research Center, Paderborn, 2011.
[19] E. Doubrovski, J. Verlinden, and I. Horvath, "First steps towards collaboratively edited
design for additive manufacturing knowledge," in Solid Freeform Fabrication
Symposium, 2012, pp. 891-901.
[20] A. Azman, F. Vignat, and F. Villeneuve, "Evaluating Current CAD Tools Performances
in the Context of Design for Additive Manufacturing," no, vol. 44, pp. 1-7, 2014.
84
[21] H. Bikas, P. Stavropoulos, and G. Chryssolouris, "Additive manufacturing methods and
modelling approaches: a critical review," The International Journal of Advanced
Manufacturing Technology, vol. 83, pp. 389-405, 2016.
[22] M. P. Bendsoe and O. Sigmund, Topology optimization: theory, methods, and
applications. New York, NY: Springer Science & Business Media, 2004.
[23] J. Deng and W. Chen, "Design for structural flexibility using connected morphable
components based topology optimization," Science China Technological Sciences, vol.
59, pp. 839-851, 2016.
[24] L. A. Schmit, "Structural design by systematic synthesis," in Proceedings of the 2nd
conference on electronic computation, ASCE, New York, 1960, pp. 105-122.
[25] G. Rozvany, "Aims, scope, methods, history and unified terminology of computer-aided
topology optimization in structural mechanics," Structural and Multidisciplinary
Optimization, vol. 21, pp. 90-108, 2001.
[26] O. Sigmund and K. Maute, "Topology optimization approaches," Structural and
Multidisciplinary Optimization, vol. 48, pp. 1031-1055, 2013.
[27] M. P. Bendsøe and O. Sigmund, "Material interpolation schemes in topology
optimization," Archive of applied mechanics, vol. 69, pp. 635-654, 1999.
[28] O. Sigmund and J. Petersson, "Numerical instabilities in topology optimization: a survey
on procedures dealing with checkerboards, mesh-dependencies and local minima,"
Structural and Multidisciplinary Optimization, vol. 16, pp. 68-75, 1998.
[29] A. Shukla and A. Misra, "Review of Optimality Criterion Approach Scope, Limitation
and Development in Topology Optimization," International Journal of Advances in
Engineering & Technology, vol. 6, p. 1886, 2013.
85
[30] K. Svanberg, "The method of moving asymptotes—a new method for structural
optimization," International journal for numerical methods in engineering, vol. 24, pp.
359-373, 1987.
[31] S. Wang, "Krylov subspace methods for topology optimization on adaptive meshes,"
University of Illinois at Urbana-Champaign, Urbana, IL, 2007.
[32] I. Ferguson, M. Frecker, T. W. Simpson, and C. J. Dickman, "Topology Optimization
Software for Additive Manufacturing: A Review of Current Capabilities and a Real-
World Example," in ASME 2016 International Design Engineering Technical
Conferences and Computers and Information in Engineering Conference, Charlotte,
North Carolina, 2016, pp. V02AT03A029-V02AT03A029.
[33] W. A. Haviland, Maya settlement patterns: a critical review. New Orleans, LA: Middle
American Research Institute, Tulane University, 1966.
[34] Custompart.net, "Sand Casting-Equipment-Mold. Retrieved from:
http://www.custompartnet.com/wu/SandCasting," 2008.
[35] Acast, "Sand Casting. Retrieved from: http://www.acast.com.au/products/sand-casting,"
2017.
[36] W. Wang, H. W. Stoll, and J. G. Conley, Rapid tooling guidelines for sand casting. New
York, NY: Springer Science & Business Media, 2010.
[37] C. Chua, C. Feng, C. Lee, and G. Ang, "Rapid investment casting: direct and indirect
approaches via model maker II," The International Journal of Advanced Manufacturing
Technology, vol. 25, pp. 26-32, 2005.
[38] R. Prasad, "Progress in investment castings," in Science and Technology of Casting
Processes, ed Rijeka, Croatia: InTech, 2012.
[39] A. Handbook, "Volume 15 Casting," Materials Park: ASM International, 2008.
86
[40] M. P. Groover, Fundamentals of modern manufacturing: materials processes, and
systems. Hoboken, NJ: John Wiley & Sons, 2007.
[41] Rajasekaran, "Steps involved in investment casting process. Retrieved from:
http://www.mechscience.com/investment-casting-process-concept-of-investment-casting-
process-principle-of-investment-casting-process-lost-wax-process/," ed, 2015.
[42] J. J. Beaman, J. W. Barlow, D. L. Bourell, R. H. Crawford, H. L. Marcus, and K. P.
McAlea, "Solid freeform fabrication: a new direction in manufacturing," Kluwer
Academic Publishers, Norwell, MA, vol. 2061, pp. 25-49, 1997.
[43] J. G. Bralla, Design for manufacturability handbook, 2nd ed. New York, NY: McGraw-
Hill, 1999.
[44] S. R. Gupta, H. S. Shan, and P. Kumar. Manufacturing Processes I-Metal Casting.
Retrieved from "http://nptel.ac.in/courses/112107144/" [Online].
[45] ASM International, Casting design and performance. Materials Park, OH: ASM
International, 2009.
[46] National Programme on Technology Enhanced Learning. Design for Manufacture and
Assembly (DFMA)-Sand casting. Retrieved from
"http://nptel.ac.in/courses/107103012/module2/lec1.pdf" [Online].
[47] International Standards Organization, "Castings - System of Dimensional Tolerances and
Machining Allowances, Second Edition, ISO 8062," 1994.
[48] Spokane Industries. Precision Castings Division-Cost Drivers and Design Considerations
for Investment Casting. Retrieved from
"https://www.spokaneindustries.com/PDF/DesignGuide.pdf" [Online].
[49] M. Chhabra and R. Singh, "Rapid casting solutions: a review," Rapid Prototyping
Journal, vol. 17, pp. 328-350, 2011.
87
[50] R. Singh and J. Singh, "Comparison of rapid casting solutions for lead and brass alloys
using three-dimensional printing," Proceedings of the Institution of Mechanical
Engineers, Part C: Journal of Mechanical Engineering Science, vol. 223, pp. 2117-2123,
2009.
[51] M. Maguire, M. Baldwin, and C. Atwood, "Fastcast: Integration and application of rapid
prototyping and computational simulation to investment casting," Sandia National Labs.,
Albuquerque, NM (United States)1996.
[52] S. S. Gill and M. Kaplas, "Comparative study of 3D printing technologies for rapid
casting of aluminium alloy," Materials and Manufacturing Processes, vol. 24, pp. 1405-
1411, 2009.
[53] A. Ramos, C. Relvas, and J. Simoes, "Vacuum casting with room temperature
vulcanising rubber and aluminium moulds for rapid manufacturing of quality parts: a
comparative study," Rapid Prototyping Journal, vol. 9, pp. 111-115, 2003.
[54] W. E. Frazier, "Metal additive manufacturing: a review," Journal of Materials
Engineering and Performance, vol. 23, pp. 1917-1928, 2014.
[55] B. P. Conner, G. P. Manogharan, A. N. Martof, L. M. Rodomsky, C. M. Rodomsky, D.
C. Jordan, et al., "Making sense of 3-D printing: Creating a map of additive
manufacturing products and services," Additive Manufacturing, vol. 1, pp. 64-76, 2014.
[56] E. S. Almaghariz, B. P. Conner, L. Lenner, R. Gullapalli, G. P. Manogharan, B.
Lamoncha, et al., "Quantifying the role of part design complexity in using 3D sand
printing for molds and cores," International Journal of Metalcasting, vol. 10, pp. 240-
252, 2016.
[57] D. Snelling, Q. Li, N. Meisel, C. B. Williams, R. C. Batra, and A. P. Druschitz,
"Lightweight metal cellular structures fabricated via 3D printing of sand cast molds,"
Advanced Engineering Materials, vol. 17, pp. 923-932, 2015.
88
[58] D. Snelling, C. Williams, and A. Druschitz, "A comparison of binder burnout and
mechanical characteristics of printed and chemically bonded sand molds," in SFF
Symposium, Austin, TX, 2014.
[59] P. M. Hackney and R. Wooldridge, "3D sand printing for automotive mass production
applications," International Journal of Rapid Manufacturing, vol. 6, pp. 134-154, 2017.
[60] K. A. James, G. J. Kennedy, and J. R. Martins, "Concurrent aerostructural topology
optimization of a wing box," Computers & Structures, vol. 134, pp. 1-17, 2014.
[61] L. Krog, A. Tucker, M. Kemp, and R. Boyd, "Topology optimization of aircraft wing box
ribs," in 10th AIAA/ISSMO multidisciplinary analysis and optimization conference,
Albany, New York, 2004, pp. 1-11.
[62] J. W. Chang and Y. S. Lee, "Topology optimization of compressor bracket," Journal of
mechanical science and technology, vol. 22, pp. 1668-1676, 2008.
[63] P. Wu, Q. Ma, Y. Luo, and C. Tao, "Topology Optimization Design of Automotive
Engine Bracket," Energy and Power Engineering, vol. 8, p. 230, 2016.
[64] X. Li and L. Wang, "Based on Topology Optimization Method the Cab Mount Bracket
Lightweight Design," in Applied Mechanics and Materials, 2012, pp. 1292-1297.
[65] Q. Xia, T. Shi, M. Y. Wang, and S. Liu, "A level set based method for the optimization of
cast part," Structural and Multidisciplinary Optimization, vol. 41, pp. 735-747, 2010.
[66] R. Tavakoli and P. Davami, "Optimal riser design in sand casting process by topology
optimization with SIMP method I: Poisson approximation of nonlinear heat transfer
equation," Structural and Multidisciplinary Optimization, vol. 36, pp. 193-202, 2008.
[67] Z. Li, Z. Mao, and W. Li, "Optimization design of riser based on particle swarm
algorithm," Zhuzao(Foundry), vol. 54, pp. 176-178, 2005.
89
[68] L. Harzheim and G. Graf, "A review of optimization of cast parts using topology
optimization: II—topology optimization with manufacturing constraints," Structural and
multidisciplinary optimization, vol. 31, pp. 388-399, 2006.
[69] G. Allaire, F. Jouve, and G. Michailidis, "Casting constraints in structural optimization
via a level-set method," in 10th world congress on structural and multidisciplinary
optimization, Orlando, Florida, 2013.
[70] J. P. Leiva, B. C. Watson, and I. Kosaka, "An analytical directional growth topology
parameterization to enforce manufacturing requirements," in Proc. of 45th
AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Material Conference,
Palm Springs, CA, 2004.
[71] A. R. Gersborg and C. S. Andreasen, "An explicit parameterization for casting constraints
in gradient driven topology optimization," Structural and Multidisciplinary Optimization,
vol. 44, pp. 875-881, 2011.
[72] M. Zhou, R. Fleury, Y.-K. Shyy, H. Thomas, and J. Brennan, "Progress in topology
optimization with manufacturing constraints," in Proceedings of the 9th AIAA MDO
conference AIAA-2002-4901, 2002.
[73] K.-T. Zuo, L.-P. Chen, Y.-Q. Zhang, and J. Yang, "Manufacturing-and machining-based
topology optimization," The international journal of advanced manufacturing
technology, vol. 27, pp. 531-536, 2006.
[74] J. K. Guest and M. Zhu, "Casting and milling restrictions in topology optimization via
projection-based algorithms," in Proceedings of the ASME 2012 International Design
Engineering Technical Conference & Computers and Information in Engineering
Conference, Chicago, IL, USA, August, 2012, pp. 12-15.
90
[75] O. Schmitt, J. Friederich, S. Riehl, and P. Steinmann, "On the formulation and
implementation of geometric and manufacturing constraints in node–based shape
optimization," Structural and Multidisciplinary Optimization, vol. 53, pp. 881-892, 2016.
[76] Q. Xia, T. Shi, M. Y. Wang, and S. Liu, "Simultaneous optimization of cast part and
parting direction using level set method," Structural and Multidisciplinary Optimization,
vol. 44, pp. 751-759, 2011.
[77] J. Seppälä and A. Hupfer, "Topology optimization in structural design of a LP turbine
guide vane: potential of additive manufacturing for weight reduction," in ASME Turbo
Expo 2014: Turbine technical conference and exposition, 2014, pp. V07AT28A004-
V07AT28A004.
[78] R. Rezaie, M. Badrossamay, A. Ghaie, and H. Moosavi, "Topology optimization for
fused deposition modeling process," Procedia CIRP, vol. 6, pp. 521-526, 2013.
[79] W. Carter, D. Erno, D. Abbott, C. Bruck, G. Wilson, J. Wolfe, et al., "The GE Aircraft
Engine Bracket Challenge: An Experiment in Crowdsourcing for Mechanical Design
Concepts," in Solid Freeform Fabrication Symposium, University of Texas at Austin,
Austin, Texas, 2014, pp. 1402-1411.
[80] V. Maranan, T. W. Simpson, T. Palmer, and C. J. Dickman, "Application of Topology
Optimization and Design for Additive Manufacturing Guidelines on an Automotive
Component," in ASME 2016 International Design Engineering Technical Conferences
and Computers and Information in Engineering Conference, 2016, pp. V02AT03A030-
V02AT03A030.
[81] E. Komi, P. Kokkonen, J. Virta, P. Puukko, and M.-K. Sini, "Simulation, optimisation
anddesign of a 3D printed sand mould for a cast metal component. ," In: Paasi J,
editor.Towar. a new era Manuf. Julkaisija, pp. 72-80, 2017.
91
[82] ExOne, "Webinar: A Case Study in Optimizing Casting Design Using 3 D Printing;
2017," ed: AFS Metalcasting Television, 2017.
[83] M. Langelaar, "Topology optimization of 3D self-supporting structures for additive
manufacturing," Additive Manufacturing, vol. 12, pp. 60-70, 2016.
[84] A. T. Gaynor and J. K. Guest, "Topology optimization for additive manufacturing:
considering maximum overhang constraint," in 15th AIAA/ISSMO multidisciplinary
analysis and optimization conference, Atlanta, GA, 2014, pp. 16-20.
[85] M. Leary, L. Merli, F. Torti, M. Mazur, and M. Brandt, "Optimal topology for additive
manufacture: a method for enabling additive manufacture of support-free optimal
structures," Materials & Design, vol. 63, pp. 678-690, 2014.
[86] D. Brackett, I. Ashcroft, and R. Hague, "Topology optimization for additive
manufacturing," in Proceedings of the solid freeform fabrication symposium, Austin, TX,
2011, pp. 348-362.
[87] Alcoa Fastening Systems & Rings. (2015). Airplane Bearing Bracket Challenge.
Retrieved from https://grabcad.com/challenges/airplane-bearing-bracket-challenge.
[88] Y. Hahn and J. I. Cofer, "Study of Parametric and Non-Parametric Optimization of a
Rotor-Bearing System," in ASME Turbo Expo 2014: Turbine Technical Conference and
Exposition, Düsseldorf, Germany, 2014, pp. V07AT28A001-V07AT28A001.
[89] T. Wohlers, Wohlers Report 2010: additive manufacturing state of the inudstry: annual
worldwide progress report. Belgium: Wohlers Associates, Inc., 2010.
[90] C. Cheah, C. Chua, C. Lee, C. Feng, and K. Totong, "Rapid prototyping and tooling
techniques: a review of applications for rapid investment casting," The International
Journal of Advanced Manufacturing Technology, vol. 25, pp. 308-320, 2005.
92
[91] S. Pattnaik, P. K. Jha, and D. B. Karunakar, "A review of rapid prototyping integrated
investment casting processes," Proceedings of the Institution of Mechanical Engineers,
Part L: Journal of Materials: Design and Applications, vol. 228, pp. 249-277, 2014.
[92] C. Lee, C. Chua, C. Cheah, L. Tan, and C. Feng, "Rapid investment casting: direct and
indirect approaches via fused deposition modelling," The International Journal of
Advanced Manufacturing Technology, vol. 23, pp. 93-101, 2004.
[93] M. Maguire, M. Baldwin, and C. Atwood, "Fastcast: Integration and application of rapid
prototyping and computational simulation to investment casting. Retrieved from
"https://www.osti.gov/scitech/servlets/purl/378877"," Sandia National Labs.,
Albuquerque, NM (United States)1996.
[94] P. M. Dickens, R. Stangroom, M. Greul, B. Holmer, K. Hon, R. Hovtun, et al.,
"Conversion of RP models to investment castings," Rapid Prototyping Journal, vol. 1,
pp. 4-11, 1995.
[95] S. Rahmati, J. Akbari, and E. Barati, "Dimensional accuracy analysis of wax patterns
created by RTV silicone rubber molding using the Taguchi approach," Rapid Prototyping
Journal, vol. 13, pp. 115-122, 2007.
[96] A. Pereira, J. Pérez, J. Diéguez, G. Peláez, and J. Ares, "Design and manufacture of
casting pattern plates by rapid tooling," Archives of Material Science, vol. 29, pp. 63-7,
2008.
[97] C. Chua, K. Hong, and S. Ho, "Rapid tooling technology. Part 1. A comparative study,"
The International Journal of Advanced Manufacturing Technology, vol. 15, pp. 604-608,
1999.
[98] P. Kumar, I. Ahuja, and R. Singh, "Application of fusion deposition modelling for rapid
investment casting–a review," International Journal of Materials Engineering
Innovation, vol. 3, pp. 204-227, 2012.
93
[99] P. Greenbaum and S. Khan, "Direct investment casting of rapid prototype parts: practical
commercial experience," in Proceedings of 2nd European Conference on Rapid
Prototyping, Nottingham, UK, 1993, pp. 77-93.
[100] C. Hull and P. Jacobs, "Stereolithography and Quickcast: Moving towards Rapid
Tooling, 3D Systems," 1995.
[101] R. Sundaram, "Investment Casting Using Stereolithography Case of Complex Objects,"
1996.
[102] D. Dimitrov, K. Schreve, N. De Beer, and P. Christiane, "Three dimensional printing in
the South African industrial environment," South African Journal of Industrial
Engineering, vol. 19, pp. 195-213, 2008.
[103] P. Jacobs, "Stereolithography 1993: epoxy resins, improved accuracy, and investment
casting," in Coupling Technology to National Need, 1994, pp. 41-53.
[104] C. Gouldsen and P. Blake, "Investment casting using FDM/ABS rapid prototype patterns.
Retrieved from
"file:///C:/Users/wangj/Downloads/InvestmentCasting_FDM_Patterns.pdf"," 1998.
[105] T. Grimm, "Fused deposition modelling: a technology evaluation," Time-compression
technologies, vol. 11, pp. 1-6, 2003.
[106] J. H. Zhu, W. H. Zhang, and X. J. Gu, "On the topology optimization design for the
stereolithography based investment casting model," in Advanced Materials Research,
2010, pp. 1464-1467.
[107] X. Gu, J. Zhu, and W. Zhang, "The lattice structure configuration design for
stereolithography investment casting pattern using topology optimization," Rapid
Prototyping Journal, vol. 18, pp. 353-361, 2012.
[108] R. Sabish, "Development of a Topological Optimized Hinge Arm through Combination
of ALM-and Investment Casting Processes. Retrieved from
94
http://www.altairatc.com/EventPage.aspx?event_id=37&name=3D+Printing+Design+Pro
cess," in 2015 European Altair Technology Conference, Paris, France, 2015.
[109] Voxeljet, "TOPOLOGY OPTIMIZATION LEADS TO A PRIZE-WINNING CAST
PART. Retrieved from https://www.voxeljet.com/company/news/topology-optimization-
leads-to-a-prize-winning-cast-part/," ed, 2016.