Richard MASc S2021
Richard MASc S2021
Richard MASc S2021
Christopher Richard
A Thesis
in
The Department
of
Mechanical Engineering
September 2020
©Christopher Richard, 2021
ii
CONCORDIA UNIVERSITY
School of Graduate Studies
complies with the regulations of the University and meets the accepted
standards with respect to originality and quality.
Signed by the Final Examining Committee:
Abstract
The Relationship Between Lattice Structure Topology and Rapid Investment
Casting Performance
Christopher Richard
Acknowledgements
I would like to acknowledge everyone who helped me during my degree.
My parents and friends for their unconditional support. I would like to thank
the Birks Family Foundation for there support throughout the entirety of my
education with special regards to Mr. J. Birks and Mrs. G. Carozza.
I would like to thank my supervisor Tsz-Ho Kwok for his incredible support,
ensuring I have all the equipment necessary to complete my Master’s and his
encouragement and tutelage.
v
Contents
List of Tables ix
1 Introduction 1
2 Related Work 7
2.1 Casting Technology . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Additive Manufacturing . . . . . . . . . . . . . . . . . . . . . . 9
2.3 Lattice Structures . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.4 Applications of Rapid Investment Casting Lattice . . . . . . . . 13
4 Results 28
4.1 Mold Flow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.2 Cast 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
4.3 Cast 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.4 Mechanical Properties . . . . . . . . . . . . . . . . . . . . . . . 32
4.5 Voids and Grain Structures . . . . . . . . . . . . . . . . . . . . 34
5 Discussion 38
5.1 Design Guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . 40
6 Conclusion 43
vii
List of Figures
List of Tables
Chapter 1
Introduction
Currently, the most studied metal AM process is SLM. SLM has many positive
aspects, which make it a topic of interest, but also some limitations. According
to Aboulkair et al. [5], one of the most common defects in SLM parts in general,
irrespective of the material, is porosity. Maconachie et al. [6] also mentioned
that SLM fabrication of lattice structures is understood to result in manufac-
turing defects. Beyond that metallic additive manufacturing methods cannot
be used for mass manufacturing of complex parts. This is due to the high cost
associated with the process. Alternatively multiple patterns for RIC can be
made at a very low cost and depending on the geometry can be made hollow to
save on material usage and print time. According to Leuders et al. [7], When
comparing RIC to SLM, fatigue loadings are still a challenge for SLM, and the
same is true for selective sintering (SLS). Li et al. [8] demonstrated the presence
of residual stresses in most metal AM processes. The residual stresses are due
to high temperature gradient and rapid cooling. Given that RIC is a hybrid
AM and solidification process, it is less prone to residual stresses. In addition
of the three manufacturing processes, investment casting (IC) is already well
established in many industries that would benefit from a shift towards RIC. RIC
has similar capabilities as SLM. Both methods have their respective uses and
place in the additive manufacturing space, but RIC performs better in some ap-
plications. One example is in the manufacturing of aluminum alloy structures,
which is a challenge for SLM to process because aluminum is reflective, and the
laser has difficulties melting aluminum without causing defects [5]. Also, RIC
2 Chapter 1. Introduction
does not require sintering, which often leads to considerable shrinkage. The
use of SLM for lightweight cellular structures has been heavily iterated and
researched. The result is a vast trove of knowledge regarding materials, param-
eters, and design methodologies for metallic additive manufacturing of cellular
lattice structures. Although RIC can also be used to manufacture complex lat-
tice structures with unprecedented design freedom, little work has been done
to advance the design methodology for RIC of lattice structures. The limited
research includes using fused filament fabrication (FFF) to produce low-cost
patterns [9], studying the effect of cross-sectional shape of struts on mechanical
properties [10], and finding the optimal filling direction for honeycomb struc-
tures [11]. However, there is no current overarching analysis of the effect that
lattice topology has on casting performance and what topological properties
play the most extensive role in minimizing casting defects. Lattice topology
is defined as the properties of a lattice structure in 3D that remain the same
regardless of deformations such as twisting and stretching. For strut-based lat-
tice topologies these properties are the bounding volume, unit cell density, and
strut shape/distribution/connectivity. This analysis is required to avoid casting
defects in lattice structures to faithfully apply them to metal parts. There are
many benefits to RIC over other manufacturing processes. RIC can produce
parts with complex geometries such as: thin sections, cavities and complex in-
ternal lattice structures. Beyond that, RIC requires very little post processing
and it can produce parts with great dimensional accuracy and surface finish.
Finally, RIC components do not suffer from poor fatigue strength so it can
be used in industries like aerospace, dentistry and biomedical engineering, etc.
This work will present the background information on IC and RIC and then
delve deeper into the current open areas of research. Some areas of research
show more potential towards improving the manufacturing process as a whole
and these will be presented with the intention of identifying the most lucrative
research directions in RIC.
Given that RIC is a hybrid manufacturing method with both additive and
solidification processes, its design considerations stem from both methods. To
Chapter 1. Introduction 3
Laser Beam
Laser
Resin Tank
UV LED Bulb
d) e)
Fillament
Fillament Roll
Resin Tanks
Temperature Control
Resin Jets Nozzle
UV Led Bulb
Build Platform Build Platform
benefits of these 3D printing processes are that they can create really high
quality patterns with great surface finishes. The RIC process can be seen in
Fig 1.2. Conventional manufacturing methods impose geometric restrictions
this is not the case for RIC. IC uses a sacrificial pattern to form a mold. This is
done by encasing the pattern in refractory material. The pattern then undergoes
a burnout process leaving a mold cavity in the shape of the pattern. Molten
metal can then be poured into the mold cavity creating a metal part with the
same geometry as the pattern. Design for IC has its own set of considerations:
feeding direction, gating and feeding system design, and pattern topology. The
IC performance depends heavily on the pattern and gating system topology
and its inherent mold flow. Like many processes, topology optimization can
be applied to casting. Topology optimization for casting relies primarily on
the following properties: feed paths, flow velocity and flow connectivity. Feed
paths are used to ensure directional solidification moving from thin to thick.
Risers can be placed to allow for directional solidification that sweeps from the
extremities of the mold cavity toward the riser. Flow velocity can be optimized
to ensure molten metal does not prematurely solidify. Finally flow connectivity
can be controlled to avoid turbulence or hot spots. Minimizing the number
of intersections of metal flows will allow the casting to cool at similar rates
throughout, this will avoid the potential for hot spots which result in localized
shrinkage voids. Studying the effect of lattice topology on RIC performance
will help draw out the optimization objectives in order to be able to eventually
apply topology optimization to the casting of lattice structures. When it comes
to the design of lattice structures for RIC, the lattice topology is one of the most
critical factors that affects printability, castability, and mechanical properties.
Different objectives may have contradicting requirements on the topology, and
the research question here is: what lattice topological properties have the most
significant impact on the overall performance in RIC, and how to improve them?
In this work, a few properties are hypothesized, including the relative strut
size, number of joints, joint valence, and strut angle, significantly affect the
casting performance. These properties can be evaluated using a variety of lattice
Chapter 1. Introduction 5
topologies and structures. The relative strut size is the strut size divided by the
unit cell width, which is a property that will remain the same regardless of unit
cell width. The number of joints refers to the total number of points within
the cell where struts connect. Joint valence refers to how many struts connect
at a given joint. Strut length is the distance traveled by the metal flow in the
lattice cell. Strut angle is the angle of the strut from the filling direction of the
lattice structure. These properties were chosen as they relate to the definition
of lattice topology. Relative strut size and strut shape directly affect unit cell
density. Number of joints, joint valence and strut angle directly affect strut
distribution/connectivity.
Refractory 3D printed
pattern Broken shell Metal casting
material
slurry
Furnace
Removal from
part tree and
feeding
Shell Mold
work is to expand the limited research on lattice design for RIC as there is no
overarching analysis on the effect of lattice topology on casting performance.
The contributions of this work include:
• The properties are compared and analyzed with the test results, and a set
of design guidelines for RIC is created.
• Based on the analysis, new lattice structures are designed and optimized
for RIC.
Chapter 2
Related Work
complex design elements such as small cross sections and high dimensional ac-
curacy [13]. Given the nature of the process, IC design is principal to good mold
flow [1,14–17]. Poor gating system design could lead to casting defects, this be-
haviour can be seen in Fig. 2.1. During the optimization of casting homogeneity,
the gating system’s quality can be the most important factor affecting casting
defects occurrence [18]. One way to improve casting performance is using a
novel parabolic conical-helical sprue which reduces surface turbulence in metal
below critical velocity during mold filling [19]. Poor gating system design could
also lead to rough surface finish and accuracy, and a design where the size of
sprue and runner is unbalanced will produce unstable molten metal flow [16].
Current research is focused mainly on iterating gating system design op-
timization in the hopes of reducing the likelihood of casting defects. Gating
system design remains an iterative laborious process. There exist design guide-
lines for the process, but limited research on applying computational methods
to automate the process. One exception is the use of computing and data-
driven methods for gating and feeding system design in IC. Yu et al. present a
data-driven framework coupled with the RBF optimization method for gating
system design. The results of the optimization can be seen in Fig 2.2, it can be
observed that all the defects occur in the feeders and not the cast part.
From the optimizations the gating system’s diameter was found to be the
most influential on the volume of average shrinkage porosity [2]. This knowledge
2.2. Additive Manufacturing 9
can be added to the ever growing design methodology for IC. Conventional IC
relies primitive technologies for pattern making but AM is the most popular
option to replace the wax used in pattern making [20]. The only difference
between IC and RIC is the pattern making method, RIC relies on AM for
pattern making. This allows for RIC to take advantage of the full design freedom
of AM as it does not need to worry about pattern removal from the mold like
its wax alternatives. Let it be noted that for traditional IC the wax pattern is
melted whereas for RIC the resin pattern is burned for removal. This means
that the wax in IC can be reused to a certain extent, this is not the case for
RIC resin. There also exist wax 3D printing processes that can be used for RIC,
these processes are a bit slower and the patterns are more delicate than resin
for the molding process. For the 3D printed wax, the wax could not directly be
reused, it would have to be reprocessed for the 3D printer.
precision casting patterns. MJM can also 3D print using photo-curable resins,
but has the benefit of printing in wax materials as well, it can also create much
larger parts, with resolutions slightly higher than SLA and DLP. The resolution
of 3D prints are highly dependent on the machine and materials used. FDM
can be used create much larger parts than its resin based alternatives. Given
the low cost and large build volumes, a lot of research has been done to improve
its surface finishes and dimensional accuracy. Kumar et al. [28] attempted
to improve the surface finish of 3D printed patterns by observing the effect
of varying process parameters such as: geometric volume to area ratios, wax
coated or uncoated patterns, orientation, mold thickness and material grade
in order to achieve high dimensional accuracy when casting a hip joint. They
concluded, a thin coating of wax increases accuracy of the patterns being made.
Higher volume to area ratios and pattern orientations of 90 degrees lead to
higher accuracy. They were able to achieve the permissible tolerances grades
determined by the ISO standard. Hafsa [29] and Ibrahim et al. [30] compared
the use of two AM methods, MJM and FFF for creating RIC patterns. The
authors found that utilizing different internal structure for their patterns had
an effect on surface roughness. The main drawback of using FFF for IC is that
much effort is required to improve the quality of the printed patterns. Presently
FFF can not achieve the quality of patterns that DLP, SLA and MJM can. For
small scale IC, given the low cost, high speed and resolution of DLP, it is a
excellent choice for pattern making. For larger parts MJM seems to be the best
process available at the moment [13].
surprise that fields such as: dental, biomedical and aerospace utilize this tech-
nology. Pattnaik et al. [49] stated that IC is superior to other casting practices
and for this reason it can be applied to many applications such as: making au-
tomobile components, aircraft engines, jewelry, statues, prosthetics, computer
hardware, electronics hardware, radar and machine tool components. With the
unbounded design freedom provided by RIC and the isotropic nature of IC it
can be used for many manufacturing applications requiring high accuracy. The
following are some of the recent RIC advancements in the fields of biomedical
engineering, dentistry and aerospace engineering.
IC as of late has showed a lot of promise for the manufacturing of implants
in the biomedical field, given that components are generally individualized.
Beyond that RIC is known for its ability to create complex 3D parts with
excellent surface finishes which lends itself well to create lightweight components
that don’t lack in strength and interface well with the human body. Singh
et al. [50] proposed the use of fused filament fabrication (FFF) for printing
biomedical implants for use in IC. The study was focused on controlling surface
roughness using three factors of the IC process such as slurry layers, slurry
viscosity and dry time of primary layer. The research found that all three
properties affected the surface quality of the cast hip joint and the research
optimized the finish to micro-level resulting in a reduction of post processing
operations. Singh et al. [51] later reviewed all the current research focused on
improving the surface finish of FFF 3D printed patterns through parameter
selection of IC and AM for biomedical implants. FFF is the AM method of
choice for pattern making in the biomedical field due to its ability to create
larger parts. One of the issues that remains to be resolved is the poor surface
finish associated with FFF.
Dentistry has been using IC since 1897, Pattnaik et al. [48] stated that many
dental laboratories employ the lost wax casting process. Given that every den-
tistry partial denture is unique to the anatomy of the patient, IC is perfect for
this application. The conventional IC process has been replaced with RIC and
instead of having crowns fit using molds, the process has been replaced by a
2.4. Applications of Rapid Investment Casting Lattice 15
Given these blades are 50 percent lighter than the nickel-based alternative. The
authors were able to qualify this production technology. In aerospace how much
a component weighs is of great importance, being able to create components
with more complex internal geometries instead of using assemblies can lead
to significant weight savings and better performance. It seems that complex
internal structures in aerospace have mostly been made using gelcasting, it
would be interesting to further investigate the use of RIC for this process.
17
Chapter 3
This section outlines the parameters and setup used for the theoretical and
experimental tests. The characterization methods and lattice structures used
in the tests are also detailed.
The AM machine used in this study is the 3D Systems FabPro 1000, a DLP
printer with a resolution of 65 microns in the X and Y directions and 30-50
microns in the Z direction. The material used for this printer is the FabPro
Proto GRY plastic resin. Proto GRY is a prototyping resin manufactured by
3D systems for prototyping. This resin was used due to its great printability
when compared to casting resins. For this material, the Z direction can only
achieve a resolution of 50 microns.
The casting machine used is the Neutec® J-2R™, and the flask is the 4” di-
ameter and 6” tall Neutec® SuperPerf™ flanged flask. A high strength plaster
– the Ransom and Randolph Ultra-Vest Maxx – is used as the mold material.
The plaster is prepared with the CIMO St. Louis 92 - 4KG digital vacuum
investment mixer. Heating of the mold is done in the Lucifer L17-K12 Furnace.
Two casting materials are used in the experiments. One is the recycled 70-30
brass with a density of 8.73 × 103 g/mm3 , and the other one is the recycled 6061
aluminum with a density of 2.7 × 103 g/mm3 . The DenPlus Basic Eco Sand-
blaster is applied in post-processing, and the glass beads used in sandblasting
have a size of 50 microns.
18 Chapter 3. Materials and Methods
Samples are cut using a fret saw to perform grain structure analysis on
the casts. Next, they get mounted in Bakelite using the Bueller SimpliMet
3000 compression press. The samples get lapped using 300, 400, 600, 800, and
1200 grit sandpapers. The samples get mirror-polished using alumina powder.
Finally, they are etched using 200ml distilled water, 10gm Ferric chloride, and
50ml Hydrochloric acid. The void and grain structure analyses are done using
the Keyence VHX-6000 microscope.
The overall RIC process is illustrated in Fig. 3.1, and the necessary details are
given in the following.
a) b) c) d)
Pattern
Furnace
Build Platform Pattern Tree
Flask Mold
Pattern
Mold Base
Resin Tank
e) f)
A pattern has the shape of the object to be cast, so the computer-aided model
(CAD) model is used to make the pattern. In RIC, the pattern is produced by
3.2. Manufacturing Process 19
an AM machine, in this work the DLP printer is used (see Fig. 3.1(a)). After
slicing the CAD model, DLP produces the pattern by using a projector to cure
(solidify) complete layers of liquid resin at a time. The build head lowers into the
resin tank, leaving one layer thickness between the head transparent and tank.
The projector then exposes a mask of UV light through the transparent bottom
of the tank. This mask is the two-dimensional (2D) profile of the current layer.
After the layer curing, the tank tilts to detach the print from the build plate
before moving up a layer thickness. This process repeats until the completion
of the full 3D part. The main benefit of DLP is that it can create very high-
quality patterns with great surface finishes [57]. Alternate AM processes that
can be used for pattern making are MJM, SLA, and FFF see Fig. 1.1. DLP can
achieve more complex overhangs using less supports thanks to it curing layer-
by-layer instead of point-by-point and thus having better self-support capability
of the material. Also, it is a lot faster than MJM and SLA while achieving a
comparable resolution. FFF does not produce high enough quality parts for
making delicate lattice structure patterns.
The same manufacturer offers a casting resin. The benefit of casting resin
is that it creates less stress and ash in the mold during burnout. However,
the casting resin was not rigid enough and did not print well. It also printed
a lot slower than the prototyping resin, i.e., four times slower. Although the
prototyping resin requires a higher temperature to be burnt out, it has sig-
nificantly better printability, which was important as all the lattice topologies
tested needed to be self-supporting to ensure there was no need for internal
support removal.
The printed pattern gets placed on a mold base and a steel flask is placed on
top of the mold base. The perforated flask is then covered in masking tape to
avoid spilling the plaster, see Fig. 3.1(b). The plaster then gets weighed based
on the manufacturer’s specifications and the volume of the flask and pattern.
The plaster is placed in the mixer, and vacuum is pulled. Water is then added
20 Chapter 3. Materials and Methods
TEMPERATURE (°F)
1000
800
600
400
200
0
0 2 4 6 8 10 12
TIME (H)
to the mixer based on the manufacturer’s specifications, and the plaster gets
hydrated 38%. The mixer then mixes the plaster for 7 minutes, at which point
a knob is pulled to pour the plaster into the mold. Finally, the mold is vibrated
to remove any bubbles from the plaster for another 7 minutes, see Fig. 3.1(c).
The mold then gets removed from the mixer and is left for 10 minutes to dry.
The mold containing the resin pattern then gets burnt out (Fig. 3.1(d)) with
the custom burnout profile in Fig. 3.2.
3.2.3 Casting
For the casting experiment, the casting machine gets preheated to 1038◦ C, and
the mold gets preheated to 5388◦ C. 150% weight of 70-30 brass was weighed
and added to the casting machine’s crucible for melting. The brass weight is
calculated based on the density of 70-30 brass and the pattern volume. The
reason of 150% mass was chosen is to ensure that there is an excess of metal,
eliminating the lack of molten metal as a cause for casting defects. The recy-
cled brass was melted down in the casting machine using argon shielding with
a flow of 8L/min. Once the casting machine hit the melting temperature of
1038◦ C, the recycled 70-30 brass was added to the machine, causing the ma-
chine’s temperature to drop a few hundred ◦ C. Once the temperature rose back
to 1038◦ C and the metal was molten, the flask was introduced to the flask cham-
ber. Vacuum was then pulled for the flask chamber before pulling the lever that
3.3. Lattice Designs 21
introduces the molten metal to the mold, see Fig. 3.1(e). The mold was then
left under vacuum for 4 minutes to remove dissolved gases and fill the mold.
After 4 minutes, the vacuum pump was turned off, and the mold was left to cool
in the casting machine for 10 minutes. After 10 minutes the mold was removed
from the machine and left to air cool. The cooling temperatures and times are
optimal based on prior testing; these temperatures and time caused the least
defects and stress cracking. Finally, once the mold had cooled to 200◦ C, the
mold was quenched in a bucket of room temperature water.
For casting experiments involving 6061 aluminum, the whole process is ba-
sically the same. The only difference is that the casting machine preheat tem-
perature is 649◦ C, and the mold preheat temperature is 315◦ C.
3.2.4 Post-processing
The majority of the plaster dissolves away from the quenching process. After
quenching, the remaining plaster caught in small details gets sandblasted at 90
Psi, see Fig. 3.1(f). This removes minimal amounts of material without affecting
geometric accuracy. Once all the plaster is removed from the sample, the feeder
gets removed using a fret saw.
3.3.1 Set 1
In the first experiment, the set of lattice topologies includes rhombic, kelvin,
cubic, and octet-truss, as shown in Fig. 3.3(a-d). These topologies were chosen
as they are commonly used structures in AM, and they vary a lot in topology
while remaining open celled and self-supporting. Each lattice unit cell is a
22 Chapter 3. Materials and Methods
a) Z
e)
n7.00mm
2 Units
(20mm)
7.00mm
b)
2.50mm 2.00mm
5.00mm
6 Units
(60mm)
c)
2 Units
(20mm)
d)
10 × 10 × 10 mm cube, and the green arrow on the wireframe cells is the Z-axis,
which is the filling direction. Their connectivity are shown in order to better
understand how they connect in 3D. All the topologies have a straight circular
strut cross-section and a cubic packing strategy. The scope of topologies was
narrowed in order to limit the amount of possible topologies. All four topologies
have a constant volume and the strut size is changed to achieve this. The density
of the unit cells is kept constant at 20%. The strut size needed to achieve
a density of 20% for different lattice topologies were obtained through trial
and error using Autodesk Inventor. The strut sizes for these topologies range
from 1.3732 mm to 1.98020 mm. To better grade their filling performance and
proneness to casting defects, the structures to be cast contain 2×2×6 unit cells
as shown in Fig. 3.3(e), and they were all fed via a 7 mm cylindrical feeder.
3.3. Lattice Designs 23
3.3.2 Set 2
For the second experiment, the main goal is to use the first experiment’s ob-
servations to create lattice topologies more optimized for casting performance.
Rhombic and octet-truss are being passed along from the previous experiment
to serve as benchmarks compared to other publications and the previous experi-
ment. Along with those two topologies, two more have been proposed. The four
unit cells are rhombic, octet-truss, proposed cell, and hourglass. The structures,
as well as their overall dimensions and orientation, can be seen in Fig. 3.4. The
hourglass structure has a balance of vertical, horizontal, and 45◦ struts. This
structure is used to evaluate further the effect of strut angle distribution on
casting performance. The second proposed structure is one that was created
purely for good casting performance. It has a low number of joints, low joint
valence; most of its struts are 45◦ , and it has a large relative strut size. The
hypothesis is that this combination of properties will lead to a better casting
performance.
The unit cell size for this experiment is 5 × 5 × 5 mm. This was chosen
to observe more casting defects without exceeding the dimensions of the flask.
In addition, the casting material was changed to 6061 aluminum. Due to the
higher solidification shrinkage of aluminum compared to the brass, set 2 is a
more challenging test beyond the geometry setting. This too will contribute to
the ability to observe more casting defects to better grade the lattice topologies.
The strut sizes for these samples ranged from 0.7 mm to 1.2 mm. This finer strut
size will increase the likelihood of premature melt solidification which will more
significantly differentiate the different topologies’ performance. The structures
to be cast contain 2 × 2 × 6 unit cells as shown in Fig. 3.4(e) In order to feed
the larger number of unit cells; the feeder size is being increased to 12 mm,
which tapers from 15 mm. The 15 mm diameter was chosen to interface with
the sprue base for the mold.
24 Chapter 3. Materials and Methods
a) Z
e)
5 Units
(25mm)
b) n 15.00mm
1.50mm
4.00mm
n 12.00mm
20.00mm
7.0mm
2.00mm
c) 1.00mm
5 Units
(25mm)
5 Units
(25mm)
d)
3.4 Characterization
The lattice structure geometry for the mold flow simulations is the same as
the one to be cast in Fig. 3.3(e). Since the complexity of the second experi-
mental structures was increased based on the results of the first. They were too
complicated to simulate on a standard PC, and thus, this simulation was not
conducted.
The mold flow simulation was performed using the parameters listed in Ta-
ble 3.1. The simulation was a gravity process with the fill parameter set as
a constant liquid level on the sprue. The casting method is IC, and a shell
mold was chosen to simplify the simulation. The shell thickness of 50 mm is
quite large and comparable to the casting experiments’ flask mold. The 7 mm
cylindrical feeder fed the molten metal as in Fig. 3.3(e). The elements utilized
by the mold flow simulations are tetrahedral, and the software chose their size.
Table 3.1: Mold flow Altair Inspire Cast 2019.3 material properties
Mechanical simulations were performed using Ansys Workbench 19.2. The ma-
terial properties used for the simulations were 70-30 brass found in Table 3.2.
Two static structural, mechanical simulations were performed per lattice sam-
ple: tension/compression and shear. Bending was omitted because if a lattice
topology performed well under tension and compression, it also performed well
under bending since bending is a combination of the two [58]. For the lattice
topologies observed, the focus was lightweight, rigid topologies. Therefore the
topologies were only loaded right up till the onset of plastic deformation. The
results were only used to determine the equivalent tensile and shear modulus.
Given that the only deformation observed was plastic and this Ansys simulation
26 Chapter 3. Materials and Methods
f)
g)
c) e)
Figure 3.5: Loading conditions tensile: a) Displacement b) Force c)
Fixed support. Loading conditions shear: d) Force e) Fixed support
f) Displacement g) Displacement
did not take into account struts colliding with each other, the compressive simu-
lations were the same as the tensile simulations, so they are not reported. Also,
two large blocks of material were added to both ends of the lattice structure to
ensure the loads are applied uniformly to the structure. For tensile loading, a
force was applied to the top face 15 kN at a time. A fixed support was added
to the bottom face, which restricts movement in every direction. Finally, a dis-
placement constraint was added to the top face that only allows displacement
in the direction of the applied load. For shear loading, a force was applied to
the front face 15 kN at a time. A fixed support was also added to the bot-
tom face, which restricts movement in every direction. Finally, a displacement
constraint was added to both inner faces that only allowed displacement in the
applied loading direction. The geometry for the mechanical tests was simplified
as shown in Fig. 3.5, this was done to save on computation time. The purpose
of the simulations was not to determine the exact deformation or maximum
principal stress in the structures but rather to serve as a comparison tool for
the different topologies to establish which ones have good cast-ability without
sacrificing mechanical performance. The loading conditions and fixed supports
3.4. Characterization 27
as well as the geometry utilized for the simulation, can be seen in Fig. 3.5. The
observed results for mechanical performance are the max deformation in the
applied load direction and the max principal stress. From these results, stress-
strain graphs can be generated, and the resultant modulus for tensile and shear
of the structure can be determined. These moduli will be used to compare the
mechanical performance of the different lattice topologies.
Table 3.2: Ansys 70/30 [Cu/Zn] brass material properties
Chapter 4
Results
The results of the mold flow simulation can be seen in Fig. 4.1. The 3D
plots of only the rhombic and kelvin structures are presented because both the
cubic and octet-truss samples solidified before filling according to the simula-
tion, so their results were not plotted. The three observed casting properties
that showed the largest deviation across the lattice topologies were filling time,
porosity, and cold shuts. Firstly, from the comparison of filling time, it can be
seen that the rhombic structure fills fastest, and the filling time does not differ
much in the x and y direction, only in the filling direction. Secondly, a porosity
of 20% was chosen to be unacceptable and highlighted. The kelvin structure
has repeatable porosity located at the horizontal struts (90◦ from filling direc-
tion). The rhombic structure shows almost no porosity of 20% in the body, but
just at the top. This could be caused by the highly directional filling, which
4.2. Cast 1 29
would cause the feeder to solidify last. Finally, cold shuts were observed. The
magnitude of cold shuts in the kelvin sample is higher than that of the rhombic
sample. Cold shut locations for these two structures were consistently located
on horizontal struts (90◦ from filling direction).
4.2 Cast 1
a) b) c) d)
20.23mm
a) b) c) d)
Figure 4.3: Set 1 cast structures: a) Rhombic, b) Kelvin, c) Cubic,
and d) Octet-Truss.
30 Chapter 4. Results
Rhombic
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 47.162 20.00 1.980
Printed N/A 20.23 2.013
Cast 49.000 20.19 2.077
% Fill 103.90
Kelvin
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 47.163 20.00 1.976
Printed N/A 20.21 2.090
Cast 45.840 20.15 2.047
% Fill 97.19
Cubic
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 47.142 20.00 1.629
Printed N/A 20.20 1.723
Cast 40.350 20.15 1.693
% Fill 85.59
Octet-Truss
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 47.163 20.00 1.373
Printed N/A 20.08 1.447
Cast 38.500 20.11 1.417
% Fill 81.63
lattice structures were around 1% larger in structure width than the CAD ge-
ometries. The biggest contributor to the inaccuracy in size seems to be caused
by the 3D printing process, but this could be accounted for and corrected in the
CAD models or the printer software. Because some samples are more filled than
others, the measurements may not be a perfect representation as the unfilled
samples will have a more significant variation in dimensions. The percent fill
was calculated by comparing the cast part’s mass from the CAD model to what
it is. The percentage fill is based on the weight of the cast samples with the
feeder removed compared to the CAD model’s weight. According to the per-
cent fill, the rhombic structure filled the most, followed by the kelvin, cubic and
octet-truss. The percentage fill and visual inspection are in agreement. Overall
the rhombic structure has the best casting performance. The structure seems
to have flowed well based on the number of defects. The casting performance
from best to worst was rhombic,
4.3 Cast 2
a) b) c) d)
Figure 4.4: Printed Patterns Experiment 2: a) Proposed cell b)
Hourglass c) Rhombic d) Octet-truss
All the four lattice structures in the second set are also successfully printed
without supports, molded, and cast, as shown in Fig. 4.4 and Fig. 4.5. There
was no significant variation in the samples’ weight (see table 4.1). According
to the percent fill, although the difference of the top-three was low, the most
successful structure was the hourglass proceeded by the proposed cell, rhombic,
32 Chapter 4. Results
25.02mm
a) b) c) d)
Figure 4.5: Set 2 cast structures: a) Proposed cell, b) Hourglass, c)
Rhombic, and d) Octet-Truss.
and finally the octet-truss. However, from visual inspection, it is clear that
the proposed cell structure showed the fewest visible defects, followed by the
hourglass, rhombic, and finally octet-truss. The proposed cell sample only had
one visibly unfilled strut. The hourglass sample had many visible defects present
in its vertical (0◦ struts) as well as its joints among the vertical (0◦ ), horizontal
(90◦ ) and diagonal (45◦ ) struts. The rhombic structure had many defects similar
to the first casting experiment with voids at the perpendicular (90◦ ) struts and
the high joint valence joints. Finally, the octet truss sample did not fill. From
these results, it can be seen that the two new structures designed based on the
observation from experiment 1 are indeed better.
The tensile stress vs. strain for elastic deformation can be seen in Fig. 4.6. From
the graph, it can be seen that under tensile loading, the hourglass, proposed
cell and cubic samples performed the best. These three topologies had the
highest resultant tensile modulus see Table 4.3. The resultant modulus of the
structures describes the structures’ rigidity under different loading conditions.
The rigidity of the structures under tensile loading from highest to lowest is
cubic, hourglass, proposed cell, rhombic, kelvin, and octet-truss. Similarly, the
line’s slope for shear stress against strain tells how much deformation occurs
for a given applied shear load. The resultant shear modulus of the structures
can be seen in Table 4.3. The structure’s rigidity under shear loading from
4.4. Mechanical Properties 33
Hourglass
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 12.005 25.00 1.022
Printed N/A 25.19 1.013
Cast 13.030 25.01 1.157
% Fill 108.54
Proposed Cell
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 12.004 25.00 1.111
Printed N/A 25.02 1.147
Cast 12.750 24.93 1.113
% Fill 106.21
Rhombic
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 12.005 25.00 0.989
Printed N/A 25.180 1.028
Cast 12.540 24.86 1.003
% Fill 104.46
Octet-Truss
Mass (g) Cell Width (mm) Strut Diameter (mm)
CAD 12.005 25.00 0.686
Printed N/A 25.10 0.743
Cast 11.740 20.080 1.447
% Fill 97.79
2000
1500
1000
500
0
0 0.05 0.1 0.15 0.2 0.25
350
300 Rhombic
250
200
150
100
50
0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35
Table 4.5: Lattice topology strut size and joint characteristics. Strut
size is in mm.
The grain area distribution was not determinable due to voids limiting the
ability for software to determine the grain boundaries. The porosity results can
be seen in Fig. 4.8, which shows the microscopic images at 50x magnification
without having etched the sample. The area and frequency of each void were
tabulated. The results for void ratio, number of voids, and max/min void area
are summarized in Table 4.4. A clear trend from the results can be observed.
With the increase in the joint valence, a higher void ratio is observed. The void
ratio increases from 0.21% in the 2-valence joint to 1.45% in the 4-valence joint
and finally to 3.35% in the 6-valence joint. The void ratio is based on the sum
of the void area, so the void area shows the same trend. The number of defects
also increases with the increase in joint valence. The only outlier in terms of
behavior is the max void size. The max is higher for the 4-valence joint than
the 6-valence joint. The large defect in the 4-valence joint is quite close to the
strut’s surface and could be a surface defect when looking at the structure as a
whole. Regardless of this outlier, the overall void ratio still ends up being lower
than the 6-valence one. These results show that the higher the valence number,
the worse the casting performance.
36 Chapter 4. Results
Sample 1: Sample 2:
a) 2-Valence Joint
1000μm
Outer Edge of casting
b) 4-Valence Joint
c) 6-Valence Joint
40
Frequency
30
20
1000μm
10
0
82656
1100
1500
1900
130
190
250
300
700
10
70
More
Void Area (μm²)
80
Frequency
60
40
20
0
82656
1100
1500
1900
130
190
250
300
700
10
70
More
200
Frequency
150
100
50
0
82656
1100
1500
1900
130
190
250
300
700
10
70
More
Chapter 5
Discussion
Table 4.5). Lower relative strut sizes result in slower filling times which often
results in premature melt solidification. Therefore, it is clear that the relative
strut size plays the largest role in the success of a lattice structure casting. This
ratio gives a idea of the negative effect that adding more struts for rigidity has
on the strut size and in turn the metal flow. From this the structures that
create an efficient short path for the metal to flow through can be determined.
Rhombic
90.00
Cubic
80.00
Kelvin Cell
70.00
% Total Strut Lenght
Octet-Truss
60.00
Proposed Cell
50.00
Hourglass
40.00
30.00
20.00
10.00
0.00
0° (Vertical) 35°-55° (Diagonal) 90° (Horizontal)
Absolute Value of Strut Angle from Filling Direction
Figure 5.1: Topology strut angle distribution
Next in the level of importance is the number of joints. From the mold
flow and experimental castings, it is clear that the topologies with the highest
number of joints performed very poorly. The effect can be seen in the cubic
(27), kelvin (24) and octet-truss (14) topologies. The remaining three topologies
with a joint number of 9 all performed well. To add a level of granularity to
the analysis, these three topologies can be further classified based on their joint
valence. The proposed cell structure performed better than the other two due
to its low joint valence (4.58), this trend continues for the hourglass (5.74) and
then the rhombic (8). This behaviour can be further supported by the void
analysis results. A higher void ratio can be observed as the joint valence is
increased from 2 to 4 to 6. In the 2-valence and 4-valence joints, fully formed
equiaxed zones can be seen unlike the 6-valence joint. Equiaxed grains are more
40 Chapter 5. Discussion
In order to better establish and understand the design guidelines for RIC, the
performance grading of all the topologies can be seen in Table 5.1. The topolo-
gies are graded from 1 to 4 with 4 being the best and 1 being the worst perfor-
mance. This table also includes the tensile and shear performance for reference.
The topological properties are listed in decreasing order of casting perfor-
mance from left to right. The two topologies (proposed cell and hourglass)
performed the best experimentally, as shown from the table. They scored 3 and
5.1. Design Guidelines 41
4 across the board and scored 4 in the most important categories (relative strut
size, number of joints, and joint valence). The mechanical performance was not
sacrificed due to the angle distribution, which includes vertical, horizontal, and
diagonal struts [58].
The number of joints and joint valence also affect mechanical strength. This
effect was observed in the FEA results and is supported by Li et al. [60].
The authors stated that lattice cells’ deformation mode changes from bending-
dominated to stretch-dominated with the increase of the joint valence. Stretch
dominated is the more rigid of the two behaviors. Although, rigidity can be
achieved without a high number of joints and joint valence, as Maxwell’s crite-
rion of rigidity is limited as demonstrated by Chen et al. [61]: if finite element
analysis shows a rigid lattice, then the net lattice is rigid.
The design guidelines in order of importance are as follows. The relative
strut size should be kept below 0.20. The number of joints should be kept
below 9. The max and mean joint valence of 8 or less is recommended. Finally,
for mechanical performance, the strut angle distribution should include vertical,
diagonal, and horizontal struts. The following minimum conditions must be met
for faithful use of the design guidelines.
• The lattice unit cell must range from 20 mm3 to 1000 mm3 and the strut
size should range approximately from 1 mm to 5 mm.
• The materials must be cast using some sort of assisted method such as
vacuum or centrifugal with some sort of inert shielding gas.
• The pattern must be 3D printed using a resin or wax based process with
a resolution of at least 100 µm in x, y and z.
By far from what has been tested, the proposed cell, hourglass and rhombic
topologies meets all these criteria and achieve good casting and mechanical
performance.
43
Chapter 6
Conclusion
In summary, it is clear that, much like other metallic AM processes, RIC can
take full advantage of AM’s unprecedented design freedom. RIC was success-
fully used to create a variety of lattice structures. These structures were utilized
to determine the topological lattice properties critical to casting performance.
From those results, a methodology to study the performance of RIC lattice
structures has been established. In this methodology, topological properties are
compared and analyzed using the test results. This analysis results in a set of
design guidelines for RIC. The properties established to affect casting perfor-
mance in descending importance are relative strut size, number of joints, joint
valence, and strut angle distribution. These properties differ slightly but rely
on the same logic as the design guidelines for casting. Feed paths and flow ve-
locity/connectivity play a large role in design for casting. Without controlling
these properties, hot spots, porosity and premature solidification can occur. All
four proposed topological lattice properties directly affect a lattice’s feed paths
and flow velocity/connectivity. The design guidelines developed throughout
this work differ in that they are specifically tailored towards optimizing cellular
structures. Cellular structures by nature have a high potential for flow restric-
tion or high connectivity. For this reason the design guidelines need to be more
targeted than the general casting ones. The properties deemed to have the most
significant effect on tensile and shear mechanical performance are strut angle
distribution, number of joints, and joint valence. With the design methodology,
44 Chapter 6. Conclusion
the proposed cell and hourglass topologies were created. These lattice topolo-
gies had the best overall casting and mechanical performance of all the tested
lattices. The limitations of the current work include testing only strut-based
lattice topologies. Future work could expand beyond this. Future work could
also further refine the design methodology and automate the design process
using software-driven methods.
45
Bibliography
[1] Fei Li, Yingchun Wang, Donghong Wang, Yanjie Zhao, Chengkang Qi,
and Baode Sun. Comparison of various gating systems for investment
casting of hydraulic retarder impeller with complex geometry. Proceedings
of the Institution of Mechanical Engineers, Part B: Journal of Engineering
Manufacture, page 0954405420971994, 2020.
[2] Jiangping Yu, Donghong Wang, Dayong Li, Ding Tang, Xin Hao, Shixin
Tan, Da Shu, Yinghong Peng, and Baode Sun. Engineering computing and
data-driven for gating system design in investment casting. Int. J. Adv.
Manuf. Technol., 111(3):829–837, 2020.
[3] Patrik Karlsson, Lars Pejryd, and Niclas Strömberg. Generative design
optimization and characterization of triple periodic lattice structures in
alsi10mg. In International Conference on Additive Manufacturing in Prod-
ucts and Applications, pages 3–16. Springer, 2020.
[4] Lin Cheng, Jiaxi Bai, and Albert C To. Functionally graded lattice struc-
ture topology optimization for the design of additive manufactured com-
ponents with stress constraints. Computer Methods in Applied Mechanics
and Engineering, 344:334–359, 2019.
[5] Nesma T Aboulkhair, Marco Simonelli, Luke Parry, Ian Ashcroft, Christo-
pher Tuck, and Richard Hague. 3D printing of aluminium alloys: Additive
manufacturing of aluminium alloys using selective laser melting. Progress
in Materials Science, 106:100578, 2019.
[8] C. Li, Z.Y. Liu, X.Y. Fang, and Y.B. Guo. Residual stress in metal additive
manufacturing. Procedia CIRP, 71:348 – 353, 2018.
[10] Yingjie Huang, Yingying Xue, Xinfu Wang, and Fusheng Han. Effect of
cross sectional shape of struts on the mechanical properties of aluminum
based pyramidal lattice structures. Materials Letters, 202:55–58, 2017.
[14] Jenn-Kun Kuo, Pei-Hsing Huang, Hsin-Yi Lai, and Jian-Rong Chen. Op-
timal gating system design for investment casting of 17-4PH stainless steel
BIBLIOGRAPHY 47
[15] Donghong Wang, Jinyu Sun, Anping Dong, Da Shu, Guoliang Zhu, and
Baode Sun. An optimization method of gating system for impeller by rsm
and simulation in investment casting. Int. J. Adv. Manuf. Technol., 98(9-
12):3105–3114, 2018.
[16] Maidin S, T.M. Yi, Hambali A, Suriati Akmal, Ruzy Hambali, and Zulke-
flee Abdullah. Investigation of optimum gating system design of fused de-
position modelling pattern for sand casting. J. Mechan. Eng. Sci., 11:2801–
2814, 09 2017.
[17] P.-H. Huang and M.-J. Guo. A study on the investment casting of 17-4ph
stainless steel helical impeller of centrifugal pump. Mater. Res. Innov.,
19(sup9):S9–77–S9–81, 2015.
[18] Marek Bruna, Dana Bolibruchová, Richard Pastirčák, and Anna Remišová.
Gating system design optimization for investment casting process. J.
Mater. Eng. Perform., 28(7):3887–3893, 2019.
[19] Santosh Reddy Sama, Tony Badamo, Paul Lynch, and Guha Manogharan.
Novel sprue designs in metal casting via 3D sand-printing. Addit. Manuf,
25:563–578, 2019.
[20] OMAR MOHD FAIZAN BIN MARWAH. Analysis of various rapid proto-
typing techniques for investment casting. PhD thesis, Universiti Teknologi
Malay, 2016.
[21] Jiayi Wang, Santosh Reddy Sama, Paul Carl Lynch, and Guhaprasanna
Manogharan. Design and topology optimization of 3D-printed wax patterns
for rapid investment casting. Procedia Manuf, 34:683–694, 2019.
[23] Jaspreet Singh, Rupinder Singh, and Harwinder Singh. Dimensional ac-
curacy and surface finish of biomedical implant fabricated as rapid invest-
ment casting for small to medium quantity production. J. Manuf. Process.,
25:201–211, 2017.
[25] Yoshiki Ishida and Taira Miyasaka. Dimensional accuracy of dental casting
patterns created by 3d printers. Dental materials journal, 35(2):250–256,
2016.
[26] Mario Monzon, Zaida Ortega, Alba Hernandez, Ruben Paz, and Fernando
Ortega. Anisotropy of photopolymer parts made by digital light processing.
Materials, 10(1), 2017.
[27] Pei-Hsing Huang and Wei-Ju Huang. Processing design of miniature cast-
ing incorporating stereolithography technologies. International Journal of
Mechanical, Aerospace, Industrial, Mechatronic and Manufacturing Engi-
neering, 11(8):1403 – 1406, 2017.
[28] Parlad Kumar, Inderpreet S. Ahuja, and Rupinder Singh. Effect of process
parameters on surface roughness of hybrid investment casting. Progress in
Additive Manufacturing, 1(1):45–53, Jun 2016.
[32] Jason Walker, Evan Harris, Charles Lynagh, Andrea Beck, Rich Lonardo,
Brian Vuksanovich, Jerry Thiel, Kirk Rogers, Brett Conner, and Eric Mac-
Donald. 3d printed smart molds for sand casting. Int. J. Metalcast.,
12(4):785–796, 2018.
[33] Yashkumar Patel, Ankit Kahar, and Dhaval Doshi. Technical Review of
Additive Manufacturing technique in Patternless Casting Manufacturing.
Int J S Res Sci. Tech., page 6, 2018.
[35] Oraib Al-Ketan, Reza Rowshan, and Rashid K Abu Al-Rub. Topology-
mechanical property relationship of 3D printed strut, skeletal, and sheet
based periodic metallic cellular materials. Addit. Manuf., 19:167–183, 2018.
[36] Zhen Wang, Zhiyu Xiao, Ying Tse, Chuanshou Huang, and Weiwen Zhang.
Optimization of processing parameters and establishment of a relationship
between microstructure and mechanical properties of slm titanium alloy.
Optics & Laser Technology, 112:159–167, 2019.
50 BIBLIOGRAPHY
[37] Li Tang, Qingbiao Zhang, Keshan Liang, Xiqing Zhao, and Zhixiong Zhang.
Discrete optimization of internal part structure via slm unit structure-
performance database. Metals, 8(1):45, 2018.
[38] Lina Yan, Li Ping Zhao, and Gavin Kane O’Neill. Dimensional consistency
of SLM printed orthopaedic implants designed using lightweight structures.
Transactions on Additive Manufacturing Meets Medicine, 2(1), 2020.
[39] Han Wang, Yu Fu, Mingming Su, and Hai Hao. A novel method of in-
direct rapid prototyping to fabricate the ordered porous aluminum with
controllable dimension variation and their properties. Journal of Materials
Processing Technology, 266:373–380, 2019.
[40] Dawei Li, Wenhe Liao, Ning Dai, Guoying Dong, Yunlong Tang, and
Yi Min Xie. Optimal design and modeling of gyroid-based functionally
graded cellular structures for additive manufacturing. Comput. Aided Des.,
104:87–99, 2018.
[41] Marius A Wagner, Thomas S Lumpe, Tian Chen, and Kristina Shea.
Programmable, active lattice structures: Unifying stretch-dominated and
bending-dominated topologies. Extreme Mechanics Letters, 29:100461,
2019.
[42] Enrique Alabort, Daniel Barba, and Roger C Reed. Design of metallic bone
by additive manufacturing. Scripta Materialia, 164:110–114, 2019.
[43] Ajeet Kumar, Luca Collini, Alix Daurel, and Jeng-Ywan Jeng. Design and
additive manufacturing of closed cells from supportless lattice structure.
Additive Manufacturing, page 101168, 2020.
[44] Ahmed Hussein, Liang Hao, Chunze Yan, Richard Everson, and Philippe
Young. Advanced lattice support structures for metal additive manufactur-
ing. Journal of Materials Processing Technology, 213(7):1019–1026, 2013.
[46] Renkai Huang, Ning Dai, Xiaosheng Cheng, and Lei Wang. Topology opti-
mization of lattice support structures for heat conduction in selective laser
melting. The International Journal of Advanced Manufacturing Technol-
ogy, 109(7):1841–1851, 2020.
[47] Ajeet Kumar, Saurav Verma, and Jeng-Ywan Jeng. Supportless lattice
structures for energy absorption fabricated by fused deposition modeling.
3D Printing and Additive Manufacturing, 7(2):85–96, 2020.
[50] Rupinder Singh, Sunpreet Singh, and Prince Kapoor. Investigating the sur-
face roughness of implant prepared by combining fused deposition modeling
and investment casting. Proceedings of the Institution of Mechanical Engi-
neers, Part E: Journal of Process Mechanical Engineering, 230(5):403–410,
2016.
[51] Daljinder Singh, Rupinder Singh, and Kamaljit Singh Boparai. Develop-
ment and surface improvement of fdm pattern based investment casting of
biomedical implants: A state of art review. J. Manuf. Process., 31:80 – 95,
2018.
[53] Bjorn Einar Dahl, Hans Jacob Ronold, and Jon E. Dahl. Internal fit of
single crowns produced by cad-cam and lost-wax metal casting technique
assessed by the triple-scan protocol. The Journal of Prosthetic Dentistry,
117(3):400 – 404, 2017.
[54] H Wu, D Li, Y Tang, N Guo, F Cui, and B Sun. Rapid casting of hollow
turbine blades using integral ceramic moulds. Proc. Inst. Mech. Eng. B. J.
Eng. Manuf., 223(6):695–702, 2009.
[55] Haihua Wu, Dichen Li, Xiaojie Chen, Bo Sun, and Dongyang Xu. Rapid
casting of turbine blades with abnormal film cooling holes using integral
ceramic casting molds. Int. J. Adv. Manuf. Technol., 50(1-4):13–19, 2010.
[56] Julio Aguilar, Andre Schievenbusch, and Oliver Kättlitz. Investment cast-
ing technology for production of tial low pressure turbine blades – process
engineering and parameter analysis. Intermetallics, 19(6):757 – 761, 2011.
[59] Yanpeng Wei, Bo Yu, Quanzhan Yang, Peng Gao, Zhiquan Miao,
Jingchang Cheng, and Xun Sun. Damping behaviors of steel-based kelvin
lattice structures fabricated by indirect additive manufacture combining
investment casting. Smart Materials and Structures, 29(5):055001, 2020.
[60] Chuanlei Li, Hongshuai Lei, Zhong Zhang, Xiaoyu Zhang, Hao Zhou, Pand-
ing Wang, and Daining Fang. Architecture design of periodic truss-lattice
cells for additive manufacturing. Addit. Manuf, page 101172, 2020.
BIBLIOGRAPHY 53
[61] Wen Chen, Seth Watts, Julie A Jackson, William L Smith, Daniel A Tor-
torelli, and Christopher M Spadaccini. Stiff isotropic lattices beyond the
maxwell criterion. Science advances, 5(9), 2019.