LOST FOAM CASTING OF PERIODIC CELLULAR MATERIALS ...
Transcript of LOST FOAM CASTING OF PERIODIC CELLULAR MATERIALS ...
LOST FOAM CASTING OF PERIODIC CELLULAR MATERIALS
WITH ALUMINUM AND MAGNESIUM ALLOYS
by
Samson Shing Chung Ho
A thesis submitted in conformity with the requirements
for the degree of Master of Applied Science
Graduate Department of Materials Science and Engineering
University of Toronto
Copyright by Samson Shing Chung Ho (2009)
ii
Lost Foam Casting of Periodic Cellular Materials with Aluminum and Magnesium Alloys
M.A.Sc. , Samson Shing Chung Ho (2009)
Department of Materials Science and Engineering
University of Toronto
This study investigates the possibility of fabricating periodic cellular materials (PCMs)
via the lost foam casting (LFC) process using aluminum alloy A356 and magnesium alloy AZ91.
This approach combines the structural efficiency of PCM architectures with the processing
advantages of near-net-shape LFC. An initial feasibility study fabricated corrugated A356 panels.
This was followed by a study of casting variables such as pattern design, vacuum assistance, and
alloying additions in order to improve the fillability of the small cross-section struts. Finally,
integrated pyramidal sandwich panels having different relative densities were subjected to
artificial aging treatments and subsequently tested in uniaxial compression. The A356 PCMs
experienced a continuous increase after yielding while the AZ91 PCMs exhibited strut fracture
after peak strength. The results showed the compressive yield strengths of this study are
comparable with those previously reported PCMs produced by different fabrication methods.
iii
Acknowledgements
I would like to express the deepest appreciation to Professor Hibbard and Professor Ravindran for
their encouraging guidance and endless support during the course of my work. Their mentoring has
provided insight into ongoing scientific research and a rapidly changing industry. Without their guidance
and persistent help, my dedication and best efforts for this work would not been possible.
I wish to thank my colleagues Francesco D’Elia, Abdallah Elsayed, Ken Lee, Sophie Lun Sin,
Lukas Bichler and Brandon Bouwhuis for their friendship. The technical support from Alan Machin,
Joseph Amankrah and Sal Boccia is greatly appreciated.
I would like to thank the Natural Science and Engineering Research Council of Canada and the
University of Toronto for their financial support.
Finally I am grateful to my parents for their never-ending care, my brother for his friendship and
my girlfriend Joey for her endless support, throughout the years.
iv
Table of Contents
Abstract
Acknowledgements
Published Work
List of Tables
List of Figures
List of Symbols
List of Abbreviations
ii
iii
vi
vii
ix
xiii
xiv
Chapter 1: Introduction 1
Chapter 2: Literature Review 6
2.1. Periodic Cellular Materials (PCM) Topologies 6
2.2. Fabrication Methods in Wrought and Cast Periodic Cellular Materials 8
2.2.1. Investment Cast Periodic Cellular Materials 9
2.3. Types of Commercial Casting Methods 12
2.4. Lost Foam Casting (LFC) Process 13
2.4.1. Lost Foam Casting Advantages 15
2.4.2. Defects in Lost Foam Casting 16
2.4.3. Challenges in Magnesium Alloy Thin-Walled Lost Foam Casting 19
Chapter 3: Experimental 22
3.1. Foam Core Preparation 22
3.1.1. Corrugated Core Fabrication 22
3.1.2. Pyramidal Core Fabrication 23
3.1.3. Integrated Sandwich Fabrication 24
3.2. Lost Foam Casting 26
3.3. Post-Casting Characterization 28
3.4. Experimental Plan 30
Chapter 4: Results and Discussion 31
4.1. Initial Feasibility Study 31
4.2. Effect of Casting Conditions on LFC PCMs 34
4.2.1. Sprue and Runner Designs 35
4.2.2. Vacuum Assisted LFC 38
4.2.3. Alloying Additions 43
v
4.2.4. Preliminary Compressions Test Results 51
4.3. Integrated Pyramidal Sandwich 55
4.3.1. A356 PCMs Compression Test Results 56
4.3.2. AZ91 PCMs Compression Test Results 61
4.3.3. Lower and Upper Bound Model 66
Chapter 5: Conclusions 74
References 75
Appendix 80
A.1. Schematic Diagram 81
A.2. Phase Diagrams 82
A.3. Grain Fineness Number 83
A.4. Integrated Pyramidal Sandwich Compression Test Results 84
A.5. Previous Studied PCMs and Metal Foams 86
vi
Published Work
This thesis is based on the following publications:
a) Ho, S., Bichler, L., Hibbard, G.D. and Ravindran, C. 2008. Synthesis-Structure
Relationships in Cast Magnesium Periodic Cellular Materials, Magnesium Technology, TMS
2008: New Orleans, L.A.
b) Ho, S., Ravindran, C. and Hibbard, G.D. 2008. Lost Foam Casting of Magnesium Periodic
Cellular Materials, Processing and Fabrication of Advanced Materials XVII: New Delhi,
India, (Paper AC-136), pp.309-317.
c) Ho, S., Ravindran, C. and Hibbard, G.D. 2009. Production of Magnesium Thin-Wall
Cellular Castings Through Lost Foam Casting, AFS Transactions, Vol. 117, (Paper 09-051),
pp. 857-865.
d) Ho, S., Ravindran, C. and Hibbard, G.D. 2009. Fabrication of Periodic Cellular Materials
with Lost Foam Casting in Magnesium Alloy AZ91, Scripta Materialia (in preparation).
e) Ho, S., Ravindran, C. and Hibbard, G.D. 2009. Fabrication of Integrated Pyramidal
Sandwich Panels with Lost Foam Casting in Aluminum Alloy A356, Materials Science and
Engineering A (in preparation).
vii
List of Tables
Table 2-1. Fabrication details and compressive properties of various cast PCMs. 11
Table 2-2. Thermal degradation of EPS and PMMA foam pattern [Shivkumar,
1994].
19
Table 3-1. Dimensions of the corrugated sandwich core. 23
Table 3-2. Dimensions of the pyramidal lattice core. 24
Table 3-3. Dimensions of the integrated pyramidal sandwich. 25
Table 3-4. A356.2 Aluminum alloy chemical composition in wt% [House of
Metals, 2008].
27
Table 3-5. AZ91D Magnesium alloy chemical composition in wt%. 27
Table 3-6. The heat treatment schedule for A356 [Davis ed., 1993] and AZ91
[Avedesian and Baker ed., 1999].
29
Table 3-7. Etchants used in microscopic examination of A356 [Keller, 1948] and
AZ91 [Maltais et al., 2004] alloys.
29
Table 4-1. The casting parameters for the corrugated core assembly. 31
Table 4-2. The dimensions of the corrugated core assembly. 31
Table 4-3. Microhardness measurements as a function of distance from the
downsprue.
34
Table 4-4. The casting parameters for the vertical sprue lattice core assembly. 35
Table 4-5. The dimensions of the vertical sprue lattice core assembly. 36
Table 4-6. The dimensions of the vertical sprue sandwich assembly. 37
Table 4-7. The casting parameters for the vertical sprue sandwich assembly. 37
Table 4-8. The casting parameters for the horizontal sprue pattern assembly. 38
Table 4-9. The dimensions of the horizontal sprue pattern assembly. 38
Table 4-10. The casting properties for the narrowed vertical sprue pattern assembly. 39
Table 4-11. The dimensions of the narrowed vertical sprue pattern assembly. 39
Table 4-12. Summary of casting conditions: the vacuum level, pouring temperature,
and number of coatings for the eight different casting conditions.
40
Table 4-13. Summary of porosity and microhardness characterization. 43
Table 4-14. The casting properties for the narrowed vertical sprue pattern assembly. 44
Table 4-15. The dimensions of the narrowed vertical sprue pattern assembly. 44
Table 4-16. Measured chemical composition by spark emission spectroscopy for
each casting condition in wt%.
45
Table 4-17. Summary of casting parameters for integrated sandwich. 55
Table 4-18. The dimensions of integrated sandwich pattern assembly. 56
viii
Table A-1. AFS sieve number and the multiplying factor Mi [Rao, P.N., 1999]. 83
Table A-2. Yield strength and peak strength of A356 and AZ91 PCMs in compressions. 84
Table A-3. Experimental measured yield strength (σ yield PCM) and peak strength
(σ peak PCM) of A356 and AZ91 PCMs in compressions.
85
Table A-4. Previous studies of cast PCMs, summarizing the alloy, architecture, bulk
metal density (ρs), relative density ( ρ~ ), PCM’s density (σ peak PCM),
PCM’s specific peak strength (σ peak PCM / ρ PCM), PCM’s yield strength
(σ yield PCM) and PCM’s specific yield strength (σ yield PCM / ρ PCM).
86
Table A-5. Previous studies of aluminum PCMs, summarizing the alloy, architecture,
bulk metal density (ρs), relative density ( ρ~ ), PCM’s density (σ peak PCM),
PCM’s specific peak strength (σ peak PCM / ρ PCM), PCM’s yield strength
(σ yield PCM) and PCM’s specific yield strength (σ yield PCM / ρ PCM).
87
Table A-6. Previous studies of magnesium metal foams, summarizing the alloy,
architecture, bulk metal density (ρs), relative density ( ρ~ ), PCM’s
density (σ peak PCM), PCM’s specific peak strength (σ peak PCM / ρ PCM),
PCM’s yield strength (σ yield PCM) and PCM’s specific yield strength
(σ yield PCM / ρ PCM).
88
ix
List of Figures
Figure 1-1. The sandwich panel consists of a core between a pair of face sheets
where hfs = thicknesses of the face sheets, hc = thicknesses of the
core, L = the sandwich length, b = the sandwich width and htotal =
the sandwich height [after Ashby, 2005].
1
Figure 1-2. Pin-jointed frame with (A) bending-dominated mechanism and (B)
stretch-dominated mechanism [Ashby, 2005].
2
Figure 1-3. Material-property chart of Young’s modulus and density [Ashby,
2005].
3
Figure 2-1. Honeycomb core structures: (A) hexagonal, (B) square and (C)
triangular shaped [Wadley, 2006].
6
Figure 2-2. Corrugation structures: (A) triangular, (B) diamond and (C) navtruss
[Wadley, 2006].
7
Figure 2-3. Lattice architectures: (A) tetrahedral, (B) pyramidal and (C) kagome
[Wadley, 2006].
7
Figure 2-4. Normalized strength with respect to relative density for different
cellular architectures [Wadley, 2006].
8
Figure 2-5. Investment cast lattice sandwich [Chiras et al., 2002]. 10
Figure 2-6. The series of major steps in Lost Foam Casting process [dos Santos et
al., 2007].
15
Figure 2-7. Average metal velocity window with different defects [Hess, 2004]. 17
Figure 2-8. Lost foam casting defects: (A) cold laps and (B) misruns [Foseco,
1991].
17
Figure 3-1. Schematic diagram of corrugated sandwich core. 22
Figure 3-2. Three-dimension schematic model of pyramidal lattice core. 23
Figure 3-3. Two-dimension schematic diagram of pyramidal lattice core. 24
Figure 3-4. 10 mm-thick foam panel is cut with (A) the hot-wire-cutting template
into (B) pyramidal lattice pattern.
24
Figure 3-5. Schematic diagram of integrated pyramidal sandwich showing the
integration of node thickness to face sheets.
25
Figure 3-6. 20 mm-thick foam block is carved with (A) hot-wire-cutting template
into (B) integrated pyramidal sandwich.
26
Figure 3-7. Schematic representation of experimental plan. 30
Figure 4-1. As-cast corrugated core with pouring temperatures of (A) 710 °C and
(B) 730 °C.
32
Figure 4-2. Cross-sectional node geometry from the as-cast corrugation core (a,b)
poured at 710 °C and (c,d) poured at 730 °C.
32
Figure 4-3. Typical porosity at the centre (A) and edge (B) of the corrugated core 33
x
at a pouring temperature of 730 °C.
Figure 4-4. Typical microstructure of as-cast AZ91D corrugated core and etched. 33
Figure 4-5. Pyramidal lattice core pattern (A) ready for casting and as-cast
samples (B).
35
Figure 4-6. Pyramidal sandwich pattern (A) ready for casting, A356 as-cast
sample (B) and AZ91 as-cast sample (C).
36
Figure 4-7. Pyramidal sandwich assembly with top feeding sprue (A) ready for
casting, AZ91 as-cast sample (B), altered pyramidal sandwich
assembly with tapered top feeding sprue (C) and AZ91 as-cast
sample (D).
38
Figure 4-8. Pyramidal sandwich assembly with narrow and lengthen sprue (A) and
AZ91 as-cast sample (B).
39
Figure 4-9. The particular casting condition for the eight samples (summarized in
Table 4-12) had a significant effect on the extent of fillability.
Complete pattern was obtained in samples S5, S6, and S8.
41
Figure 4-10. Figure 4-10. Percentage of truss core (left) and face sheets (right)
filled for the eight different casting conditions (Table 4-12).
41
Figure 4-11. Casting map summarizing the extent of face-sheet and truss core
filling.
42
Figure 4-12. SEM images showing exposed porosity on polished surfaces of sample
S1 (A), sample S3 (B), sample S6 (C), and sample S7 (D).
43
Figure 4-13. The assembled coated pattern (A) and sandwich panel in close up (B). 45
Figure 4-14. Schematic representation showing the position of thermocouples in the
pattern.
45
Figure 4-15. The final casting of magnesium PCMs with different alloying
additions.
46
Figure 4-16. Temperature (left) and time (right) profiles: T1-T3 (top of panel) and
B1-B3 (bottom of panel).
48
Figure 4-17. Solidification time (left) and cooling rate (right) profiles: T1-T3 (top
of panel) and B1-B3 (bottom of panel).
48
Figure 4-18. Optical micrographs (at same magnification, see scale bar from
sample s6) and thermal analysis of the castings.
49
Figure 4-19. The measured porosity levels with Archmedes principle. 50
Figure 4-20. Hardness of the cast samples. 51
Figure 4-21. Typical compression test coupon (2×3.5 pyramidal truss core cells). 52
Figure 4-22. Typical uniaxial stress strain curves showing the mechanical
behaviour with and without sample preparation edge effects.
52
Figure 4-23. Progressive node/face sheet fracture during compression testing
(image labels correspond to the testing points shown on Fig. 4-22).
53
xi
Figure 4-24. Comparisons between different AZ91 PCMs in compressions. 53
Figure 4-25. Integrated sandwich assembly (A) and AZ91 as-cast sample (C). 55
Figure 4-26. Stress-strain curves (left) and rate of change in stress with strain
(right) for typical A356 PCMs in compression with different
relative density ρ~ .
56
Figure 4-27. Struts morphology initially (A) and after compression loaded (B). 57
Figure 4-28. Stress-Strain curves for typical solutionized (left) and solutionized and
aged (right) A356 PCMs in compression with different relative
density ρ~ .
57
Figure 4-29. A356 PCM’s yield strength (0.2 % offset) with respect to relative
density: as-cast (left), solutionized (middle), and solutionized and
aged (right).
58
Figure 4-30. A356 PCMs’ strains at which they yield at 0.2 % offset with respect to
relative densities after different heat treatments.
58
Figure 4-31. A356 PCMs’ modulus with respect to strains after different heat
treatments.
59
Figure 4-32. A356 PCMs’ modulus with respect to strains and different relative
densities: as-cast (left), solutionized (middle), and solutionized and
aged (right).
59
Figure 4-33. A356 etched microstructures in as-cast (A), solutionized (B), and
solutionized and Aged (C).
60
Figure 4-34. A comparison of different aluminum alloy PCMs and the current
studied LFC PCMs.
60
Figure 4-35. Stress-strain curves (left) and change in strain (right) for typical AZ91
PCMs in compression with different relative densities ρ~ .
61
Figure 4-36. Struts morphology initially (A), after peak (B), expanded view of the
fractured struts after peak (C) and after stress plateau (D).
61
Figure 4-37. Strain-strain curves for typical solutionized (left) and solutionized and
aged (right) AZ91 PCMs in compression with different relative densities
ρ~ .
62
Figure 4-38. AZ91 PCM’s yield strength (0.2 % offset) with respect to relative
density ρ~ : as-cast (left), solutionized (middle), and solutionized and
aged (right).
62
Figure 4-39. AZ91 PCMs’ strains at which they yield at 0.2 % offset with respect to
relative densities after different heat treatments.
63
Figure 4-40. PCMs’ strains at peak stress with respect to relative densities after
different heat treatments.
63
Figure 4-41. PCM’s modulus with respect to strains after different heat treatments. 64
Figure 4-42. PCMs’ modulus with respect to strains and different relative densities: 64
xii
as-cast (left), solutionized (middle), and solutionized and aged
(right).
Figure 4-43. AZ91 etched microstructures in as-cast (A), solutionized (B), and
solutionized and Aged (C).
65
Figure 4-44. A comparison on yield strength with respect to density between
magnesium foams and present study of AZ91 LFC PCMs.
65
Figure 4-45. Lower bound model based on minimum strut cross-section thickness,
tc.
66
Figure 4-46. Stress-strain curve of A356 square compression blocks with different
heat treatment conditions.
67
Figure 4-47. A356 sample block after compression (A) and magnified surface
feature of the block (B).
67
Figure 4-48. A356 square compression block`s surface with crack propagations. 68
Figure 4-49. Stress-strain curve of AZ91 square compression blocks with different
heat treatment conditions.
69
Figure 4-50. AZ91 sample block’s fractured cross section after compression (A)
and its magnified surface (B) showing transgranular fractures with
parallel plateau and ledges.
69
Figure 4-51. AZ91 square compression block fracture surface after 45° fracture. 70
Figure 4-52. Compressive yield strength of A356 and AZ91 compression blocks
with different heat treatments.
71
Figure 4-53. The comparisons between summarized experimental A356 PCMs’
yield strengths with lower bound model: as-cast (left), solutionized
(middle), and solutionized and aged (right).
72
Figure 4-54. The comparisons between summarized experimental AZ91 PCMs’
yield strengths with lower bound model: as-cast (left), solutionized
(middle), and solutionized and aged (right).
72
Figure A-1. Schematic diagram of corrugation sandwich core with dimensions. 81
Figure A-2. Aluminum-silicon phase diagram (left) and magnesium-silicon phase
diagram (right) [Nayeb-Hashemi and Clark, 1988].
82
Figure A-3. Aluminum-magnesium phase diagram [Nayeb-Hashemi and Clark, 1988]. 82
xiii
List of Symbols
α Mg-Al Phase
Β Mg17Al12 Phase
δ Face Angle of Indenter
θ ° Strut Angle
ξ Metal Loss Coefficient
ρ~ Core Relative Density
ρl Density of Liquid Metal
ρp Density of EPS
ρ PCM Density of the PCM Core
ρs Density of the Bulk Material
ρw Density of Water at Room Temperature
σys Compressive Yield Strength of Bulk Material
σys PCM Compressive Yield Strength of PCM
a Section Thickness of Pattern
a/l Strut Aspect Ratio
A0 Minimum Cross-Section Area of the Strut
Am Cross-Section Area of Pattern
APCM Cross-Section Area of PCM Under Compression
As Cross-Section Area of Sprue
b Sandwich Width
Cl Specific Heat of Liquid Metal
d Mean Diagonal of Indention
dc Node Thickness
dT/dt First Derivative Cooling Curve
E~
Young’s Modulus of the PCM Core
sE Young’s Modulus of the Bulk Material
g Gravitation Constant
h Heat Transfer Coefficient
H Latent Heat of Liquid Metal
xiv
hc Thicknesses of the Core
HE Decomposition Energy of EPS
hfs Thicknesses of the Face Sheets
hs Sprue Height
htotal Sandwich Height
l Sandwich Length
lc Length of Characteristic Chill Zone
Lf Metal Flow Length
Lp Penetration Load of Indenter
Mm Mass of Sample
Mm+w Mass of Water with Sample
' Number of Struts with Each PCM Under Compression
Pp Back Pressure Due to Gaseous EPS
Rc Radius of Curvature
tc Web Thickness
Tm Melting Temperature
To Mould Temperature
∆T Superheat Temperature
VE Metal Velocity
List of Abbreviations
ABS Acrylonitrile Butadiene Styrene
EPS Expanded Polystyrene
LFC Lost Foam Casting
pcf Pounds Cubic Feet
PCM Periodic Cellular Materials
PMMA Polymethyl Methacrylate
PS Polystyrene
Chapter 1: Introduction
1
1. Introduction
Hybrid materials are the combination of a monolithic material with another, or a monolithic
material with open space; they are designed to have their own set of properties that neither component can
offer alone [Wadley, 2006]. Hybrids include composites (fibrous or particulate), sandwich materials,
lattice structures and segmented structures [Ashby, 2005], and can reach new regions of material-property
space. The present study focuses on sandwich panels with lattice structure core.
A sandwich panel (as shown in Fig. 1-1) is an example of a hybrid with a combination of a
specially designed core that is layered between a pair of face sheets. Typically, the face sheet material is
required to be stiff and strong since it must carry most of the applied load. The core material can either be
a monolithic material or a hybrid material itself that is lightweight but strong and stiff enough to resist the
applied stresses [Ashby, 2005]. The purpose of the core is to minimize the mass near the sandwich
panel’s centroid, which contributes to higher specific bending stiffness and strength.
Figure 1-1. The sandwich panel consists of a core between a pair of face sheets where hfs = thicknesses of
the face sheets, hc = thicknesses of the core, L = the sandwich length, b = the sandwich width and htotal =
the sandwich height [after Ashby, 2005].
A variety of cellular architectures can be used as the sandwich core. The most commonly seen
architectures are metal foams and honeycomb structures. Recently, lattice truss or periodic cellular
materials (PCMs), have been studied as potential sandwich cores. These materials have internal struts
oriented in a regular three-dimensional architecture. In metal foams, the stiffness is relatively low because
their geometries permit bending of the cell edges which collapse when the externally applied load is
Chapter 1: Introduction
2
transmitted through the internal joints. This can be seen in the pin-jointed frame analogy in Fig. 1-2A.
Metal foams are therefore called bending-dominated structures [Ashby, 2005]. In contrast, lattice trusses
have higher structural efficiency than metal foams. In Fig. 1-2B, the lattice truss analog has a transverse
strut between the pin-jointed frames. As the load is applied, the force causes tension along the horizontal
strut before the cell collapses. These lattice trusses are often called stretch-dominated structures. For the
same material and density, lattice trusses will have a higher stiffness compared to the foam structure. This
performance difference is shown in Fig. 1-3, where the stretch-dominated structures retain higher stiffness
than bending-dominated structures with decreasing relative density.
Figure 1-2. Pin-jointed frame with (A) bending-dominated mechanism and (B) stretch-dominated
mechanism [Ashby, 2005].
With less than 20 % of the volume occupied by metal, the open space in PCMs can serve
additional functions such as: heat exchange by fluid flow within the core [Lu, 1999], thermal insulation
by isolating heat transfer between the face sheets [Wadley, 2006], acoustic damping by absorbing noise
and vibration within the core [Wadley, 2006], and impact resistance by absorbing energy during
architecture collapse [Zhang and Ashby, 1992].
Chapter 1: Introduction
3
Figure 1-3. Material-property chart of Young’s modulus and density [Ashby, 2005].
The potential applications for hybrids can be found in transportation industries and storage
systems. Karmann GmbH has developed the technology to produce sandwich body panels with cost-
effective performance that offer as much as an order of magnitude higher stiffness than similar steel body
parts with half the weight [Ashby et al., 2000]. The sandwich body panel consists of outer aluminum
skins bonded to titanium-hydride-expanded aluminum foam core. This structure can be used as firewalls,
roof panels and trunk panels in the car body shell. Another type of aluminum foam core (ALPORAS),
developed by Shinko Wire Company Ltd, provides sound absorbing capability that reduces highway
noise [Ashby et al., 2000]. Another type of open-celled aluminum foam (ERG DUOCEL) has been
developed for pressures tanks that requires constant temperature and pressure in moving systems [Ashby
et al., 2000]. The aluminum foam can prevent the tank from dimension change and baffle motion fluid
from affecting normal operations.
Chapter 1: Introduction
4
Several methods have been developed to fabricate PCMs. From the solid state, PCMs can be
fabricated by metal fabric layup [Sypeck and Wadley, 2001; Queheillalt and Wadley, 2005] and the
deformation forming of expanded or perforated metal sheets [Sypeck and Wadley, 2002; Bouwhuis and
Hibbard, 2006; Kooistra and Wadley, 2007]. PCMs can also be fabricated with template solidification
from the liquid state by investment casting [Deshpande and Fleck, 2001; Wallach and Gibson, 2001;
Chiras et al., 2002; Zhou et al., 2004; Li et al., 2008]. The newly developed lost foam casting (LFC) has
been utilized in manufacturing industries, and may provide an alternative PCM fabrication approach to
investment casting.
LFC was initially developed in 1962 [Goria et al., 1986] and has been commercially available
since the 1980s [Rodgers, 1988], providing a niche casting approach for products that are not ideal for
other casting methods. LFC is a near-net-shape method because it can produce components with complex
geometries and open cavities. In this process, molten metal is poured over the expanded polymeric foam
pattern, the heat from the molten metal degrades the foam into liquid and gas products by endothermic
reaction. The degradation product escapes through foam’s ceramic coating to the surrounding sand. The
molten metal fills the cavity and shape of the pattern as it solidifies; the metal velocity is an important
parameter since it determines the overall casting quality.
With increased interests in developing lightweight and high-performance materials for structural
applications, magnesium becomes a good candidate because of its lowest density among the structural
metals. While much research has been done to develop the LFC technology for aluminum alloys,
magnesium alloys have only been cast successfully in research environment [Bichler et al., 2003; Marlatt
et al., 2003]. The fillability of magnesium alloys with LFC becomes a challenge because of magnesium’s
lower density and heat content compared to aluminum, which results in lower metal velocity during
casting. Aluminum alloys offer high castability, while magnesium alloys may offer enhanced weight
specific material properties.
Chapter 1: Introduction
5
The main objective of this thesis was to use the lost foam casting method to produce magnesium
and aluminum alloy PCMs for the first time. Since PCMs have never been fabricated by LFC, the first
objective was to conduct an initial feasibility study. The second objective was to adjust the processing
parameters to enhance the PCM casting quality. The final objective was to determine the effect of relative
density and precipitation hardening on the mechanical properties of aluminum and magnesium alloy
PCMs.
This thesis is organized as follows: Chapter 2 presents a literature review of periodic cellular
materials and lost foam casting; Chapter 3 provides experimental procedures in foam core fabrication, lost
foam casting preparations and post-casting characterization; Chapter 4 presents the results and discussion
of the present study; Chapter 5 concludes the findings from this thesis.
Chapter 2: Literature Review
6
2. Literature Review
2.1 Periodic Cellular Architectures
There are three broad classes of architectures that have been studied in periodic cellular metals.
They are honeycomb, prismatic (corrugation) and lattice truss. The closed-cell honeycomb configurations
can be hexagonal, triangular, and square shaped (examples shown in Fig. 2-1). Their unit cells are
repeated in two dimensions that are oriented normal to the face sheets. If triangular configuration in a
honeycomb is rotated by 90° about the face sheet’s axis, a prismatic structure (corrugation) is then formed
as seen in Fig. 2-2. The cell structure is partially open in one direction. Diamond and navtruss
configurations are also possible by alternating the orientation and plate width. These architectures
contribute to anisotropic mechanical behavior; the stiffness of the panels is different when loading in
longitudinal or transversal directions. The open space volume in PCM can be further increased by
aligning slender beams instead of plates in different configurations as the core. Fig. 2-3 shows tetrahedral,
pyramidal and kagome lattice architectures. These structures have free flowing channels in two or three
directions. They become very mechanically efficient as the relative density decreases [Wadley, 2006].
Figure 2-1. Honeycomb core structures: (A) hexagonal, (B) square and (C) triangular shaped [Wadley,
2006].
Chapter 2: Literature Review
7
Figure 2-2. Corrugation structures: (A) triangular, (B) diamond and (C) navtruss [Wadley, 2006].
Figure 2-3. Lattice architectures: (A) tetrahedral, (B) pyramidal and (C) kagome [Wadley, 2006].
The strength of the hybrid structure with respect to its bulk material (structural efficiency) is
largely determined by the particular cellular architecture, seen in Fig. 2-4. The structural efficiency
improves from foams, to corrugations, to honeycombs and finally to lattice trusses by moving towards the
top left corner of the figure. It is important to note that the structural efficiency of pyramidal solid trusses
matches that of square honeycomb as the relative density (structural density divided by material density)
reduces. The stiffness of the structure with respect to its density for stretching-dominated structure is
∝
s
PCM
s
PCM
E
E
ρρ
Eqn. 1-1,
and bending-dominated structure is
2
∝
s
PCM
s
PCM
E
E
ρρ
Eqn. 1-2,
where PCME = Young’s modulus of the structure, sE = Young’s modulus of the bulk material, PCMρ =
density of the structure and sρ = density of the bulk material [Ashby, 2005]. Overall, there is less
reduction in the cellular architectures’ stiffness with decreasing density for lattice trusses than for
conventional metal foams.
Chapter 2: Literature Review
8
Figure 2-4. Normalized strength with respect to relative density for different cellular architectures
[Wadley, 2006].
2.2 Fabrication Methods in Wrought and Cast Periodic Cellular Materials
There are several methods that have been developed for creating cellular materials. These can be
classified into vapour, liquid or solid state approaches. The PCM’s architecture depends on the type of
fabrication methods and (precursor or sacrificial) materials used. The most common methods for foam
fabrications are foaming solidification with gas evolution or gas injections. The most common methods
for lattice truss periodic cellular materials are metal fabric layup and deformation forming of expanded or
perforated sheets. In the metal fabric layup method, individual metal mesh layers are stacked together and
joined to face sheets with transient liquid phase (brazing) method [Sypeck et al., 2002; Tian et al., 2007].
In the pressed expanded metal method, a rolled metal sheet is punched in different lattice architectures
and plastically deformed with a die into a three-dimensional architecture [Sypeck and Wadley, 2001;
Bouwhuis and Hibbard, 2006; Bouwhuis et al., 2008]. Most recently, electrodeposition on rapid-
Chapter 2: Literature Review
9
prototyped polymer cores has been used to create nanocrystalline periodic cellular lattice materials
[Gordon et al., 2009]. Thus far, the only cast method developed for PCM fabrication is templated
solidification with investment casting; a polymer lattice is used as a sacrificial template and removed to
create the negative mould for subsequent investment casting [Chiras et al., 2002; Li et al., 2008].
2.2.1 Investment Cast Periodic Cellular Materials
In investment casting, the sacrificial pattern is created in wax or polymer and joined with a runner
and gating system. The whole pattern is then coated multiple times with ceramic casting slurry and heated
to remove the wax before filling with liquid metal. Cast periodic cellular materials have been fabricated
(see Fig. 2-5) using rapid-prototyped acrylonitrile butadiene styrene (ABS) or injection moulded
polystyrene (PS) as sacrificial lattice patterns in tetrahedral [Deshpande and Fleck, 2001; Chiras et al.,
2002], pyramidal [Wallach and Gibson, 2001], and kagome [Wang et al., 2003] architectures. Table 2-1
summarizes the cast PCMs studies conducted to date. Note that in certain cases, the physical properties of
the sacrificial materials and process parameters (injection moulded polystyrene and injection moulded
polymer) were not clearly explained in the literature as they were supplied by third parties such as Meka
Mouldings Ltd. and Jonathan Aerospace Materials Corporation (JAMCORP). The types of casting alloys
consist of copper, aluminum and titanium. High fluidity casting alloys such as Al-Si and Cu-Be are used
with higher than ideal relative density ( ρ~ < 5 %) because the casting process requires higher pressure
feeding of molten metal before solidification occurs. This becomes increasingly difficult as the strut
cross-section area is reduced and increases the possibility for casting defects.
Chapter 2: Literature Review
10
Figure 2-5. Investment cast lattice sandwich [Chiras et al., 2002].
Chapter 2: Literature Review
11
Table 2-1. Fabrication details and compressive properties of various cast PCMs.
# Note: the core architecture is not sandwiched between solid face sheets but perforated face sheets.
* Note: the core relative density stated is calculated from the overall relative density (includes face sheet) given from the source in parentheses.
‡ Note: the value was not given but is calculated based on the given strut dimensions from the source.
Ψ Note: the strut aspect ratio is the diameter to length ratio of the strut.
Authors Alloy Process Sacrificial
Material Architecture
Strut
Angle
Strut
Aspect
RatioΨ
Core
Relative
Density
Compressive
Strength
Yield Peak
θ° a/l ρ~ (%) σ PCM (MPa)
Deshpande and
Fleck, 2001
Al-7Si-0.3Mg
Investment
Cast
Injection
Moulded
Polystyrene
Tetrahedral#
55°‡ 0.071 8
‡
4.2 7.3
Cu-4Si-1.4Zn 4.2 10.5
Wallach and
Gibson, 2001 (443) Al-4Si-0.2Fe
Injection
Moulded
Polymer
Pyramidal# 45° 0.059 6.2
* (10) 13.5 18
Chiras et al., 2002
Cu-2Be
Rapid-
Prototyped
ABS
Tetrahedral
55°‡
0.041
2
4.0 4.8
Wang et al., 2003 Kagome 0.045 3.8 4.8
Zhou et al., 2004
(516.1) Al~3Mg~1Si-1Fe~1Mn Injection
Moulded
Polymer
Pyramidal#
55°‡ 0.053
‡ 6.5
* (13)
2.9 5.9
(518.0) Al~8Mg-1.8Fe-0.35Si 3.8 5.6
(A356.0) Al~7Si~0.4Mg 3.5 5.8
Li et al., 2008
Ti-6Al-4V
Vacuum
Investment
Cast with
Hot
Isostatic
Pressing
Wax 51°‡
0.047‡ 5.4
* (13)
19 22
0.052‡ 5.5
* (16)
29 31
Ti-6Al-2Sn-4Zr-2Mo 35 41
Chapter 2: Literature Review
12
2.3. Types of Commercial Casting Methods
Some of the commonly used casting processes in the foundry industry include sand, permanent
mould and die casting. The preferred casting process for a component depends on various restrictions,
including a component’s geometry, production volume, design tolerance and alloy characteristics. These
constraints help to determine the most economical and feasible process for manufacturing.
Sand casting uses bonded sand to create drag and cope (top and bottom halves of the mould) with
removable core for pattern cavity [Monroe, 1992]. After liquid metal filling the space between sand
moulds and solidified, the as-cast product is removed by breaking the sand mould and machined to finish.
This is an economical process that has both low capital and maintenance cost for mould making [Monroe,
1992]. Since the initial cost is small for sand mould making, the quantity can be flexible as well.
However, the method can only provide rough surface finish and simple geometry within the mould, and
extensive machining is generally required.
Permanent mould casting uses a steel mould that is machined to the desired shape of the casting.
Casting defects can be minimized by heating the steel mould during casting. The as-cast product is
ejected from the mould with hydraulic power. If the design requires an open cavity within the product,
multiple moulds are necessary to create multiple casting components that are assembled later. This
process provides excellent surface finish, but yet it requires high initial cost in steel machined mould and
design limitations [Monroe, 1992].
Die casting uses a steel mould with liquid metal injection at high pressure. This process can
produce thin-walled castings having excellent surface finish that are difficult to fabricate with
conventional casting methods [Avedesian and Baker ed., 1999]. This process is suitable for high-level
production because of the reusability of the mould.
Chapter 2: Literature Review
13
While there is no single process that is suitable for all requirements, each process has its unique
advantages and disadvantages. In addition to the conventional casting methods discussed above, lost foam
casting (LFC) has recently been developed for manufacturing aluminum components. LFC can be utilized
for niche applications, providing some features that other conventional casting methods cannot provide.
The benefits of this process will be further discussed below.
2.4. Lost Foam Casting Process
Before actual casting of liquid metal can occur in the lost foam process, a series of manufacturing
steps are required as shown in Fig. 2-6. Expanded polystyrene (EPS) is the most commonly used foam
pattern, it can be produced by bead pre-expansion into polystyrene (PS) beads that are ready for
moulding. PS precursors are formed from ethyl benzene through an aluminum catalyst with benzene and
ethylene obtained from crude oil and natural gas [Shivkumar, 1994]. Ethyl benzene is then converted to
styrene at high temperature with nitrogen gas and iron catalysts. It forms polystyrene when exposed to a
peroxide catalyst and polymerized in a water solution [Goria et al., 1986]. These unexpanded beads have
a density of 600 g/l (38 pounds per cubic feet (pcf)) and they are expanded 20~50 times with heat at
100 ºC until the desired density is reached [Kanicki, 1985].
Moulding process occurs after the pre-expanded beads have been stabilized. They are injected
into tooling machine and are ready to form the pattern sections. As they are blown into the tool cavity, air
escapes through a venting position to achieve a proper fill. Steam is then used to soften and expand the
beads again so that they begin to fuse [Goria et al., 1986]. After the steam cycle, the foam is cooled
rapidly and aged to avoid shrinkage. Prototype parts can be carved from foam blocks and assembled by
hand gluing.
Various pattern pieces are joined to create the finish model pattern which is then joined to
specially designed runners and sprue system using hot melt glue. The final model assembly is submersed
into refractory ceramic slurry, spun and air dried between 40~60 ºC, under low-humidity. Ceramic slurry
Chapter 2: Literature Review
14
in a water-based solution consists of silica, alumina, zircon, chromite or complex aluminosilicates
[Monroe, 1992]. The purposes are: first to prevent molten metal penetrate through the pattern to the
surrounding sand; second, to allow the foam decomposition to escape; and third, to retain the pattern
geometry without distortion [Monroe, 1992]. Coating permeability is important in controlling the metal
velocity because the rate of filling depends on the backpressure from gaseous foam residue. Increasing the
rate of transport through the coating (air flow rate) can minimize the backpressure. The metal front has a
concave profile with low-permeability coating, transporting liquid products to the side for ease of removal
[Liu et al., 1997]. Coating permeability to gases is important for iron castings that have relatively high
casting temperatures. The metal velocity should be low enough to minimize foam residue entrapment by
controlling coating permeability and thickness [Green et al., 1998]. For aluminum castings, the coating
requires more liquid permeability for liquid foam residue [Monroe, 1992]. A uniform coating is necessary
as coating thickness can affect the casting quality. The primary control method is to determine the dry
coating weight of foam pattern.
The assembly is then embedded within the casting pit with refractory sand and compacted with
vibration to fill the assembly’s cavity. The lost foam process is finally ready for casting. Automatic
pouring is generally used to avoid pour-to-pour variations because if the pouring operation is interrupted,
metal flow will stop advancing and retreat backwards. The rate of pouring should be sufficiently slow to
allow the foam to decompose, but also fast enough to prevent the mould from collapsing [Bast et al.,
2004]. Vacuum can also be applied during casting if necessary; in this case the whole flask is sealed with
a plastic cover layer and connected to a vacuum pump. As the metal fills the mould, the foam collapses at
~100 ºC, and decomposes into liquid residue at 165 ºC [Gallois et al., 1987]. It then depolymerizes at
316 ºC and decomposes into gas at 576 ºC [Gallois et al., 1987]. The plastic residue that is trapped inside
the casting can cause porosity, cold laps or carbon defects [Gallois et al., 1987]. The casting solidifies as
the heat from the liquid metal escapes through the assembly’s coating into the sand. After solidification,
the pattern is taken out and shaken to remove any adhering sand. Finally, the pattern assembly is
sectioned and machined from the runner and sprue system before quality inspection can be done.
Chapter 2: Literature Review
15
Figure 2-6. The series of major steps in LFC process [dos Santos et al., 2007].
2.4.1. Lost Foam Casting Advantages
In conventional sand casting, pattern design with internal cavities requires a core and extensive
passage that are removed after casting. The core can also shift during casting, which results in variation of
internal passages and wall thickness. In contrast, LFC only requires a small-hole passage to fill the
internal cavities with unbonded sand. This allows the designer to create complex geometries and, the time
and cost for core production and removals are not necessary. Using unbonded sand also reduces the
amount of cleaning time required [Rodgers, 1988]. This flexibility can reduce some of the design
limitations and cost between [Rodgers, 1988; Rodgers, 1985].
In addition to requiring a core for casting internal cavities, conventional moulding technology
often imparts limits on the mould design and accuracy of sand and permanent mould casting. For
example, the gap between upper and lower moulds during casting produces parting lines and flashes
Chapter 2: Literature Review
16
which result in extra machining costs. This parting line from conventional casting methods can also limit
the orientation of the pattern in the moulds, number of castings in flask, placement of gates and risers, and
cause mould misalignments.
LFC is a near-net-shape process because it minimizes post-processing time. The foam patterns
can be glued together from many layers to create complex shapes. For instance, a product that requires ten
die casting parts which are assembled together with fasteners can be replaced with a single LFC casting.
An engine block is a good example that typically requires on the order of 100 different parts; these
different components can be integrated into one final assembled foam pattern for LFC [Donahue, 1990].
Large volume parts that require extensive coring and machining operations show the biggest benefits of
lost foam casting [Monroe, 1992]. Finally, LFC can be used to produce: sand cast components without the
need of bonded sand and core removal; permanent mould cast components without geometry limit and
extensive machining; or die casting components by integrating many parts into one single casting. These
advantages allow LFC to be applicable in producing steel and aluminum components such as brake rotors,
water pumps and engine components.
2.4.2. Defects in Lost Foam Casting
The advantages of LFC are only attainable, however, if the processing parameters are controlled
precisely. These processing parameters can affect the amount of casting defects, which can be minimized
by carefully adjusting the polymer degradation characteristics with polymer foam, foundry sand and
refractory coating properties. These properties are tailored according to the type of alloys used (e.g. steel,
aluminum and magnesium).
In lost foam casting, higher or lower metal velocity does not necessarily mean better quality
castings. There is an optimal range of average metal velocity (Fig. 2-7) for each alloy that can produce a
defect-free casting [Hess, 2004]. If the velocity is slower than optimum, metal would solidify and cause
cold laps or misruns (see Fig. 2-8) and heat from the metal would cause the pattern surface to collapse
Chapter 2: Literature Review
17
before any metal filling occur. Cold laps form between two premature solidified flows of metal merging
together; misruns are more severe forms of cold laps where the discontinuity extends throughout the
casting [Beeley, 2001]. In contrast, backpressure and turbulences from the liquid metal would form
porosity and cold folds, and penetration defects when the velocity is too high.
Figure 2-7. Average metal velocity window with different defects [Hess, 2004].
Figure 2-8. Lost foam casting defects: (A) cold laps and (B) misruns [Foseco International Ltd., 1991].
The decomposition products from the foam pre-form can also be a source of casting defects. In
addition to the typically used polystyrene (PS), polymethyl methacrylate (PMMA) has been developed for
lost foam casting to reduce the carbon residue. PS contains benzene side chains that are highly stable and
difficult to thermally decompose [Brenner et al., 1990]. On the other hand, the PMMA monomer unit
contains three less carbon atoms than PS and two oxygen atoms to help carry away carbon residue as
carbon monoxide.
Chapter 2: Literature Review
18
The decomposition mechanisms between PS and PMMA are different. PS decomposes by slow
random scission process; it degrades as a liquid and often remains trapped inside the casting [Brenner et
al., 1990]. Bromide-based additive can be used in PS to decompose PS molecular weight quickly and its
liquid residue is less viscous [Molibog, 2002; Hess et al., 2003]. This modified PS has a significant effect
on thin patterns where the surface area to perimeter is low because it is important to decompose quickly
when metal temperature is low and solidification time is short. PMMA decomposes by a rapid unzipping
process; its product residue is in the gaseous state which escapes quickly. Due to the difference in
decomposition mechanism between PS and PMMA, PMMA has only ~30 % viscous residue (rest in
gaseous product) at 750 °C, compared to ~60 % viscous residue in EPS [Shivkumar, 1995]. This viscous
residue can remain after the metal fills the mould, so it is very important in removing viscous residue
rapidly for EPS. In addition to decomposition mechanism, the energy required for heat of
depolymerization of PMMA is 14 kcal/mol compared to 17 kcal/mol in PS, this difference lowers the
thermal gradient between pouring temperature and metal feeding temperature within the casting [Brenner,
1989]. The differences in thermal degradation properties between EPS and PMMA are summarized in
Table 2-2. Even though the glass transition, collapsing and melting temperatures are higher with PMMA,
the starting, peak and end temperatures for volatilization are lower, resulting in a longer period for
volatiles to escape. The heat of degradation required for PMMA is also lower which requires less heat
energy from the molten metal compared to EPS.
Chapter 2: Literature Review
19
Table 2-2. Thermal degradation of EPS and PMMA foam pattern [Shivkumar, 1994].
Thermal Degradation Properties EPS PMMA
Glass transition temperature (°C) 80-100 105
Collapsing temperature (°C) 110-120 140-200
Melting temperature for collapsed beads (°C) 160 260
Starting temperature of volatilization (°C) 275-300 250-260
Peak temperature of volatilization (°C) 400-420 370
End temperature of volatilization (°C) 460-500 420-430
Heat of polymerization (J/g) 648 578
Heat of degradation (J/g) 912 842
Gas yield at 750 °C (cm3 (STP)/g) 230 273
Gas yield at 1300 °C (cm3 (STP)/g) 760 804
Density and the level of cell bonding (fusion) are the main control properties of the foam. The
density is largely influenced by the size of air cells [Monroe, 1992]. Low-density foams are more
desirable for the casting process because it minimizes foam residue trapped within casting which can lead
to porosity and carbon defects. But at low density, the stiffness of foam is low, allowing higher chance for
distortion.
2.4.3. Challenges in Magnesium Alloy Thin-Walled LFC
In order to minimize casting defects, it is important to understand and control the metal velocity.
Foam properties, pattern dimensions and alloy properties influence the metal velocity as shown in
equation 2-1 [Shivkumar et al., 1987; Pan and Liao, 2000]:
Eqn. 2-1
where VE = metal velocity (cm/s), As = cross-section area of sprue (cm2), Am = cross-section area of
pattern (cm2), ξ = metal loss coefficient, g = gravitation constant (cm/s
2), hs = sprue height (cm), Pp =
back pressure due to gaseous EPS (g/cms2), ρl = density of liquid metal (g/cm
3). From a pattern design
perspective, the metal velocity would increase by enlarging the cross-section area ratio of the sprue
relative to pattern, this same would occur by increasing the sprue height. The metal velocity would
−
+=
g
Phg
A
AV
l
p
s
m
sE ρξ
21
1
Chapter 2: Literature Review
20
decrease with back pressure due to higher gaseous foam products and the use of lightweight alloy. This
lower metal velocity results in reduced flow length as shown in equation 2-2 [Shivkumar, 1987; Pan and
Liao, 2000]:
Eqn. 2-2
where Lf = metal flow length (cm), ρL = density of liquid metal (g/cm3), a = section thickness (cm), VE =
metal velocity (cm/s), h = heat transfer coefficient (W/cm2°C), Tm = melting temperature (°C), To = mould
temperature (°C), H = latent heat (J/g), Cl = specific heat of liquid metal (J/g°C), ∆T = superheat
temperature (°C), ρp = density of EPS (g/cm3), lc = length of characteristic chill zone (cm) and HE =
decomposition energy of EPS (J/g). The flow length is affected by the physical and thermal properties of
the liquid metal, casting parameters and foam properties. The flow length would decreases: with liquid
metal that has low density, and low latent heat and low specific heat; with foam that has high density and
requires high decomposition energy.
When low-melting temperature magnesium or aluminum alloys are used in lost foam casting, the
foam pattern does not decompose completely into gas residue [Tseng et al., 1992]. The foam decomposes
partly into liquid as the pattern is filled with metal; some of the liquid residue can permeate through the
coating but the rest is trapped during solidification. The mould filling rate is a combination of foam
decomposition rate and foam product removal rate. In aluminum lost foam casting, the higher pouring
temperature does not necessarily increase the fill rate because the metal velocity is reduced by the
increasing backpressure at the thin sections [Lawerence et al., 1998]. The major challenge in casting these
structures is the ability to fill the whole pattern before the fraction of solids increase to the level where the
metal flow stops and solidifies.
Using flow length as an indicator, aluminum alloy A356 was found to be more fluid than
magnesium alloy AZ91 for the same superheat in gravity casting [Sadayappan et al., 2006]. The poor
fluidity of magnesium alloy is attributed to its low density, low heat capacity and low latent heat [Shin et
( )
−∆+
−= E
lc
pf
m
ELf H
l
LTc
H
TTh
aVL
ρ
ρρ
ο1
22~
Chapter 2: Literature Review
21
al., 2005]. The results from Sadayappan et al. shows AZ91 with shorter flow length compared to A356 in
either gravity or vacuum lost foam casting. Under gravity casting with the thickness of the pattern
reduced to half of the original, the flow length of AZ91 decreased to one third compared to only one half
in A356. The difference between AZ91 and A356 becomes more significant when increasing the cross-
section-area ratio of gating to sprue system. In addition, AZ91 could only achieve similar flow length to
A356 at the same vacuum level when the thickness of the pattern was doubled.
Chapter 3: Experimental
22
3. Experimental
3.1. Foam Core Preparation
3.1.1. Corrugated Core Fabrication
The starting pre-forms were 10 mm-thick expanded polystyrene (EPS) boards with 0.026 g/cm3
(1.6 pcf) density. A milling machine was used to fabricate the initial corrugated panel. EPS was machined
with a 60° chamfer tool bit, the rotational speed and feeding rate were adjusted to 2700 rpm and
50 mm/min respectively to prevent the foam beads from rupturing. A series of parallel cuts were made on
both sides of the EPS boards to obtain a triangular corrugated structure. Fig. 3-1 shows a schematic
representation of the corrugated structure; the height of the core, hc, was constrained by the starting
thickness of the as-received foam; the thickness of the node, dc, was set equal to the web thickness, tc; and
the panel angle, θ, was chosen to be 60º. Since the average EPS bead size was ~1 mm, a minimum web
thickness of 3 mm was selected to ensure approximately three bead diameters through each segment of
the EPS cross-section area for structural integrity. The core dimensions are summarized in Table 3-1. The
relative density, ρ~ , was determined to be 51 % using the following equation (Appendix A-1):
( )
+
−
−
−=
θθ
θρ
sintan
2
tan1~
2
cccc
cc
tdhh
dh
Eqn. 3-1
Fig. 3-1, Schematic diagram of corrugated sandwich core.
Chapter 3: Experimental
23
Table 3-1. Dimensions of the corrugated sandwich core.
Dimensions
Height of the core, hc 10.0 mm
Thickness of the node, dc 3.0 mm
Web thickness, tc 3.0 mm
Corrugation angle, θ 60 º
Relative Density, ρ~ 51 %
3.1.2. Pyramidal Core Fabrication
A pyramidal lattice design was chosen for the second round of pattern design in order to reduce
the relative density of the as-cast core, see Fig. 3-2. 10 mm-thick EPS boards having density of
0.021 g/cm3 (1.3 pcf) and 10 mm-thick expanded copolymer Probeads 70 boards (30 % PS and 70 %
PMMA) having 0.024 g/cm3 (1.5 pcf) density were used as the starting pre-forms. All the dimensions
were the same as that used in the corrugated patterns with the addition of radius of curvature at the node
(Fig. 3-3) as summarized in Table 3-2. Instead of the milling technique previously used, the hot-wire
cutting method was utilized with a guiding template (Fig. 3-4A). The current in the nickel chromium
resistant wire was adjusted to ~5 A with a (10 A-capacity) power supply in order to provide enough heat
to locally melt but not burn the foam. A series of parallel cuts were made on both sides to obtain a
triangular corrugated structure. The foam panel was then rotated 90º for another series of parallel cuts to
obtain the pyramidal lattice core (Fig. 3-4B). The relative density decreased from 51 % to 22 % by adding
the second dimension of corrugation to form the pyramidal lattice.
Figure 3-2. Three-dimensional schematic model of pyramidal lattice core.
Chapter 3: Experimental
24
Table 3-2. Dimensions of the pyramidal lattice core.
Dimensions
Height of the core, hc 10.0 mm
Thickness of the node, dc 3.0 mm
Thickness of the strut, tc 3.0 mm
Strut angle, θ 60 º
Radius of Curvature, Rc 3.2 mm
Relative Density, ρ~ 22 %
Figure 3-3. Two-dimensional schematic diagram of pyramidal lattice core.
Figure 3-4. 10 mm-thick foam panel is cut with (A) the hot-wire-cutting template into (B) pyramidal
lattice pattern.
3.1.3. Integrated Sandwich Fabrication
The pyramidal lattice approach had fillability issues because of the hot melt glue used between
the struts and the face sheets. The integrated pyramidal approach eliminated this problem by fabricating
both the pyramidal lattice and the face sheets from a single block of polymer foam. The dimensions
stayed the same as the previous pyramidal lattice design summarized in Table 3-3. The thickness of the
node, dc was eliminated by integrating the lattice core with face sheet (Fig. 3-5). Four different strut
Chapter 3: Experimental
25
thicknesses were studied; tc had values of 1.5 mm, 2.0 mm, 2.5 mm and 3.0 mm, which led to relative
densities of approximately 11 %, 12 %, 14 % and 16 %.
Figure 3-5. Schematic diagram of integrated pyramidal sandwich showing the integration of node
thickness to face sheets.
Table 3-3. Dimensions of the integrated pyramidal sandwich.
Dimensions
Height of the core, hc 10.0 mm
Height of the face sheets, hfs 5.0 mm
Thickness of the node, dc N/A
Thickness of the strut, tc 1.5, 2.0, 2.5, 3.0 mm
Strut angle, θ 60 º
Radius of Curvature, Rc 3.2 mm
Relative Density, ρ~ 11, 12, 14, 16 %
The hot-wire-cutting method was used to section the 33 mm-thick expanded copolymer Probeads
70 foam board, having 0.024 g/cm3 (1.5 pcf) density, to a 20 mm starting thickness. Then a slender rod
was used to pierce through the foam to create room for passing a resistant nickel chrome wire. The hot
wire cut through the foam following the guide template (Fig. 3-6A); this process was repeated on two
orthogonal directions to obtain the integrated pyramidal sandwich structure (Fig. 3-6B).
Chapter 3: Experimental
26
Figure 3-6. 20 mm-thick foam block is carved with (A) hot-wire-cutting template into (B) integrated
pyramidal sandwich.
3.2. Lost Foam Casting
Once the desired foam board pre-forms were prepared, they were ready for coating. The coating
solution (HA Styroshield 6450) was mixed to uniformity with a specific gravity of 45 Baume (water was
added as needed) by a propeller. The pattern was dipped into the coating solution, taken out and excess
solution was spun away. The moist-coated pattern was hung dry in a circulating hot air oven at 60 ºC for
one day. If thermal analysis was necessary, a pin-sized hole was made on the pattern for inserting a K-
type thermocouple.
The coated pattern assembly was then placed below a metal pouring cup in a silo that was filled
with refractory sand. The whole silo was vibrated horizontally at 1 g acceleration for 30 s for compaction
to allow the sand to fill the pattern cavity. If thermal analysis was needed, the K-type thermocouples were
connected to a data acquisition unit.
In this project, aluminum alloy A356 and magnesium alloy AZ91 were chosen for lost foam
casting. A356.2 (‘.2’ refers to recycled) aluminum alloy and is widely used in sand and permanent mould
casting. It has excellent castability, good resistance to hot cracking and shrinkage and also exhibits good
fluidity [Davis ed., 1993]. The chemical composition was measured by the supplier with arc emission
spectroscopy, results given below in Table 3-4. AZ91 magnesium alloy is commonly used in die casting
but can also be used in sand, permanent mould and investment casting. Among different magnesium
Chapter 3: Experimental
27
alloys, it has high purity and good corrosion resistance [Avedesian and Baker ed., 1999]. The exact
chemical composition used in this study was also measured with arc emission spectroscopy, summarized
in Table 3-5.
Table 3-4. A356.2 Aluminum alloy chemical composition in wt% [House of Metals, 2008].
Si Mg Ti Fe Cu Zn Mn Pb Al
7.08 0.41 0.17 0.08 < 0.01 < 0.01 < 0.01 0.001 Balance
Table 3-5. AZ91D Magnesium alloy chemical composition in wt%.
Al Zn Mn Si Cu Fe Ni Be Mg
8.51 0.58 0.23 0.043 0.003 0.062 0.0014 0.0006 Balance
Before casting, the ingots were preheated at 200 ºC for two hours to remove any moisture. A steel
crucible with the dried ingots was placed in an electric resistance furnace for ~1.5 hour until the ingots
melted and the temperature reached ~800 ºC. If the furnace was used to melt magnesium, a CO2 cover gas
was required to avoid any oxidation and ignition. For aluminum melt, a degasser and flux were added at
~50 ºC and ~20 ºC above the pouring temperature.
Once ready for casting, the melt was poured directly into the pouring cup, (casting time was ~10s
for a volume of ~1000cm3). A fume hood was placed over the pouring cup to vent some of the gas foam
pyrolysis products. A data acquisition unit recorded the changes in resistance from the K-type
thermocouples every 0.02 s for 900 s until the casting solidified and cooled. If an applied vacuum was
necessary for casting, a plastic cover was placed over the sand silo and pouring cup with a vacuum pump
connected at the bottom of the silo. The required vacuum level was reached by adjusting the side vent
value to let air enter the silo.
Chapter 3: Experimental
28
3.3. Post-Casting Characterization
Once cooled, the casting was taken out from the sand silo, and loose sand and refractory coating
were removed. The casting was then sectioned from the sprue with a vertical band saw. Samples were
prepared for uniaxial compression testing by machining the outer face sheet surfaces parallel. A cutting
speed of ~265 rpm using a 1” four flute cutter at a feed rate of ~0.001 in/rev was found to be suitable for
machining both the aluminum and magnesium casting.
The porosity of the casting was measured using the Archimedes method [e.g. Bendick, 1995], the
density can be calculated by the substance’s change in weight in a liquid due to buoyant force, which is
equal to the weight of the displaced liquid. The mass of the sample was first measured on a scale in air;
the weight of water in a light container is measured as well; at the end, the weight of both liquid and the
sample submerges mid-level inside by suspending in a fishing wire. The volume of the sample can be
found from the weight difference of the sample with buoyancy force in water. Then the porosity of the
sample can by found with the known theoretical density in the following equation:
)(100
%wwm
wm
s
s MM
Mporosity
−−=
+
ρρ
ρ Eqn. 3-2
where ρs = theoretical density of sample (g/cm3), Mm = mass of sample (g), ρw = density of water at room
temperature (0.9982 g/cm3) and Mm+w = mass of water with sample (g).
Heat treatment was applied to as-cast A356 and AZ91 to determine the effect of microstructure
on the mechanical properties of the PCMs. The heat treatment temperature and time summarized in
Table 3-6 for aluminum alloy A356 and magnesium alloy AZ91 were used in this study. The air furnace
was preheated and adjusted to the required temperature by monitoring the temperature with a K-type
thermocouple in the centre of the furnace. The samples were placed in the centre of the air furnace with a
heat soaking time of 5~10 minutes. When the required heat treatment was over, the samples were taken
out and quenched in moving air.
Chapter 3: Experimental
29
Table 3-6. The heat treatment schedule for A356 [Davis ed., 1993] and AZ91 [Avedesian and
Baker ed., 1999].
Alloy Solutionized Solutionized and Artificially Aged
A356 540 ºC for 12 h 540 ºC for 12 h followed by 155 ºC for 4 h
AZ91 413 ºC for 16 h 413 ºC for 16 h followed by 168 ºC for 16 h
The mechanical properties of the PCMs were tested in uniaxial compression. A Shimadzu AG-1
tensile / compression machine (with a 50 kN load capacity and internal strain gage) was used. The test
was performed at constant strain rate of 1 mm/min and the force was measured every 0.05 s. A unload-
and-reload cycle was performed during testing in order to determine the PCM loading modulus at ~0.75
of the sample yield strength. The force and displacement were converted to stress and strain by the
theoretical PCM area and PCM core height.
Table 3-7. Etchants used in microscopic examination of A356 [Keller, 1948] and AZ91 [Maltais
et al., 2004] alloys.
Alloy Etchant Procedure
A356 2 ml HF (48 %), 3 ml HCl, 5 ml
HNO3 and 190 ml water
Immerse for 8 to15 seconds, wash in warm water
and blow dry.
AZ91 15 ml acetic acid, 10 ml water
and 75 ml ethanol
Immerse for 30 seconds, wash in ethanol and blow
dry.
Rockwell scale E (HRE) and Vickers (HV) hardness measurements were performed on samples
mounted in epoxy and diamond polished surfaces. The hardness value was computed based on the change
in depth or width of the indentation with the preset load. HRE hardness value is measured by using a 1/8”
steel ball indenter with 10 kg minor and 100 kg major load to create an indent on the surface for 10 to
15 s [Voort ed., 1999]. The hardness value is inversely proportional to the indenter penetration depth,
each increment of hardness corresponds to 0.002mm of penetration depth. HV hardness value is measured
by using a pyramidal shaped indenter with 100g load to create a diamond shaped indent on the surface for
10 to 15 s; HV can be calculated from the following [Voort ed., 1999]:
22
8544.1)2/sin(2
d
Lp
d
LpHV ==
δ (Eqn. 3-3)
Chapter 3: Experimental
30
where d = mean diagonal (mm), Lp = penetration load (kgf) and δ = face angle (136º). ASTM standard
E384-05 [Mayer ed., 2007] was followed by spacing each indentation with a minimum 2.5 diagonal
widths apart or with the edge of the sample. A minimum of 20 hardness measurements were completed;
an average and standard deviation values are given in the Results and Discussion Section.
3.4 Experimental Plan
A schematic representation of the Experimental Plan is shown in Fig. 3-7.
Figure 3-7. Schematic representation of experimental plan.
Initial Feasibility Study
Fabrication and Casting of the
Corrugated Core
Pyramidal Sandwich Fillability Study
Fabrication and Casting:
i, Sprue and Runner Designs
ii, Vacuum Assisted LFC
iii, Alloying Additions
Integrated Pyramidal Sandwich
Fabrication and Casting:
i, Effect of PCM Architecture
ii, Effect of Microstructure
Thermal Analysis
Fillability Analysis
Porosity Measurement
Hardness Measurement
Optical Characterization
Metallography
Hardness Measurement
Thermal Analysis
Porosity Measurement
Uniaxial Compression
Lower and Upper Bound
Model
Chapter 4: Results and Discussion
31
4. Results and Discussion
4.1. Initial Feasibility Study
The initial investigation uses a simple triangular corrugated panel to determine the feasibility of
casting PCM structures with LFC. The pouring temperatures of 710 °C and 730 °C, within typical AZ91
magnesium alloy casting temperature, were repeated twice as summarized in Table 4-1. Table 4-2 also
shows the dimensions of the corrugated core assembly. Fig. 4-1 presents the overview of the as-cast
panels, the backside of the panels (pointing into the page) were sectioned from the side-feeding sprues.
Both cases had excellent overall fillability but with minor corner misfills furthest from the sprue. Overall,
this morphology is similar to the typically observed convex profile of the advancing metal front through
the EPS in pressureless LFC [Liu et al., 2002]. There was relatively little difference in the panels between
the two casting temperatures, but the 710 °C cast sample displayed slightly better surface finish and
sharper corners.
Table 4-1. The casting parameters for the corrugated core assembly.
Pouring Temperatures (°C) Foam Properties Alloy
710 EPS with 0.021 g/cm
3 (1.3 pcf) AZ91D
730
Table 4-2. The dimensions of the corrugated core assembly.
Web Thickness tc (mm) Relative Density ρ~ (%) Pattern Size (mm) Sprue Dimensions (mm)
3 51 155 X 150 270 X 50 X 50
Chapter 4: Results and Discussion
32
Figure 4-1. As-cast corrugated core with pouring temperatures of (A) 710 °C and (B) 730 °C.
The panels were sectioned at different locations to identify defects within the casting as seen
in Fig. 4-2. From the cross-section area of the panels, the morphologies of the node were not as sharp
as predicted in the schematics. The curvature of the panels and the cavity found near the node (Fig. 4-
2D) showed the morphology of the foam beads (approximately 1mm in diameter) and the effects of
foam machining. During fabrication, the foam beads were torn off from the foam panel by the rotating
chamfer tool. This resulted in foam beads delaminating from each other or caused dimples on the
surface by extra foam beads shredding off.
Figure 4-2. Cross-sectional node geometry from the as-cast corrugation core (A,B) poured at 710 °C and
(C,D) poured at 730 °C.
The cross-sections were polished to examine the level of porosity within the casting. Typical
examples are shown from the centre of the web (Fig. 4-3A) and at the edge (Fig. 4-3A) from the
730 °C cast sample. There was no significant difference between the two pouring temperatures. By the
Chapter 4: Results and Discussion
33
morphology of the porosity indicated that a large percentage was shrinkage caused by rapid
solidification from low-melting magnesium alloy with incomplete decomposition of foam products.
The permeability of the coating, the foam physical property and the volume-to-surface-area ratio of
the pattern can influence the rate of filling, and foam degradation and removal of the decomposition
products [Lawerence et al., 1998]. These parameters could be adjusted later to minimize the level of
casting porosity.
Figure 4-3. Typical porosity at the centre (A) and edge (B) of the corrugated core at a pouring
temperature of 730 °C.
The polished sample was etched to reveal the typical dendrite morphology within LFC AZ91’s
microstructure as seen in Fig. 4-4. As-cast AZ91 exhibits a eutectic microstructure with lighter regions of
primary α and darker regions of Mg17Al12 within the eutectic microstructure, which matches with what is
found from literature [Avedesian and Baker ed., 1999].
Figure 4-4. Typical microstructure of as-cast AZ91D corrugated core and etched.
Chapter 4: Results and Discussion
34
Microhardness measurements were obtained on the node cross-sections at regular intervals from
the sprue (10 mm, 50 mm, 90 mm, and 130 mm) and values are summarized in Table 4-3. There was no
apparent change in microhardness across the casting, nor any significant difference between the two
casting temperatures 710 °C (72 ± 7 HV) and 730 °C (76 ± 8 HV). Overall, these values were consistent
with those previously reported for as-cast AZ91D [Avedesian and Baker ed., 1999].
Table 4-3. Microhardness measurements as a function of distance from the sprue.
The initial casting study showed that it was possible to produce simple corrugated panels with
magnesium by LFC. It also showed that an alternative foam pattern fabrication technique would be
necessary to avoid damage to the foam beads. There was overall good fillability at both the 710 °C and
730 °C pouring temperatures but the corners of the panels showed the fill limit with the 3 mm web
thickness. Finally, the porosity and microhardness values found from this study indicated similar casting
properties to as-cast AZ91D. From the success in producing simple corrugated panels, the next step was
to create more complex and lower-relative-density PCMs by LFC and to characterize them in terms of
their casting qualities and mechanical properties.
4.2. Effect of Casting Conditions on LFC PCMs
In this study, the effects of sprue and runner designs, vacuum assistance and alloying additions on
the fillability of PCMs were studied. Different sprue and runner designs were developed to determine the
optimum filling of the pattern. The fillability of the pattern can be improved by delivering liquid metal
with minimum turbulence and heat loss. Different vacuum levels were applied during casting to increase
the metallostatic pressure, which improves transportation of foam degradation products and results in
Pouring Temperature (°C) 710 730
Distance from Sprue (mm) 10 50 90 130 Avg 10 50 90 130 Avg
Average Hardness (HV) 72 68 78 69 72 76 72 72 83 76
Standard Deviation (HV) 9 5 8 6 7 9 7 9 6 8
Chapter 4: Results and Discussion
35
improved flow length. Different levels of alloying additions were also studied with AZ91D to widen the
solidification range and improve fluidity which also improves the flow length of the pattern.
4.2.1. Sprue and Runner Designs
The new pyramidal lattice pattern helped to further reduce the relative density (from ρ~ = 51 % to
22 %). The same side feeding with vertical sprue design was used as initial feasibility study, but
additional side runners was implemented to allow flow to lattice pattern in the centre (see Fig. 4-5A). The
casting conditions and pattern assembly dimensions are summarized in Table 4-4 and Table 4-5.
Figure 4-5. Pyramidal lattice core pattern (A) ready for casting and as-cast samples (B).
Since the lattice pattern has many 3 mm-thick struts and there was no direct metal flow from the
runner to each strut, this pattern was considered to be very difficult to fill. Therefore, aluminum alloy
A356 was first used with a casting temperature of 730 °C instead of AZ91 due to A356’s good castability
with LFC. As seen in Fig. 4-5B, the casting failed to fill, the side runners offered insufficient volume for
metal flow before rapid solidification. The combination of small cross-section area runners and 60º turn
struts resulted in lower metal velocity and shortens the flow length. From this failed sprue and runner
design, it was determined that struts would need to be filled by providing more direct metal flow.
Table 4-4. The casting parameters for the vertical sprue lattice core assembly.
Pouring Temperature (°C) Foam Properties Alloy
730 EPS with 0.021 g/cm3 (1.3 pcf) A356
Chapter 4: Results and Discussion
36
Table 4-5. The dimensions of the vertical sprue lattice core assembly.
Web thickness tc (mm) Relative Density ρ~ (%) Pattern size (mm) Sprue dimensions (mm)
3 22 190 X 150 250 X 50 X 50
In order to improve fillability of the lattice pattern, a pair of face sheets was attached to the lattice
core to create a complete sandwich panel. The pair of face sheets can help provide metal flow directly to
each strut and eliminate post casting joining steps of the sandwich core to face sheets. The same sprue and
runner design (see Fig. 4-6A and Table 4-6) was used to determine the improvement in fillability due to
the additional face sheets. A356 was used at the same pouring temperature (730 °C) as AZ91,
summarized in Table 4-7. As Fig. 4-6B shows, there was good overall fillability on A356 sandwich
panel’s face sheets but there were numerous unfilled struts in the lattice core. The same pattern was cast
in AZ91 as well to determine the difference in flow ability between the two alloys (see Fig. 4-6C). AZ91
casting showed rather limited fillability on the face sheets and very low fillability within the core.
Figure 4-6. Pyramidal sandwich pattern (A) ready for casting, A356 as-cast sample (B) and AZ91 as-cast
sample (C).
In order to produce A356 cast sandwich panel, it was necessary to adjust minor casting
parameters to help minimize misfilled struts. Since the contact between face sheets and nodes of the struts
was obtained by hot melt glue, the adhesion was important for fillability but the use of hot melt glue also
decreases the metal velocity by having a lower vapourization rate than foam [Wang et al., 1990; Hess,
Chapter 4: Results and Discussion
37
2004]. In contrast, AZ91 cast sandwich panel might require a different sprue and runner design that
improves pressure feeding within the pattern to solve the major fillability issue.
Table 4-6. The dimensions of the vertical sprue sandwich assembly.
Web thickness tc
(mm)
Face sheet
thickness hfs (mm)
Relative Density
ρ~ (%) Pattern size (mm)
Sprue dimensions
(mm)
3 5 22 150 X 150 250 X 50 X 50
Table 4-7. The casting parameters for the vertical sprue sandwich assembly.
Pouring Temperature (°C) Foam Properties Alloys
730 EPS with 0.021 g/cm3 (1.3 pcf)
A356
AZ91
A horizontal sprue design was developed to provide more direct metal flow to the sandwich panel
with the help of gravity and straight through feeding design. The dimensions of the sprue were the same
as the previous design; the pouring cup attaching on the sprue and the whole design was simply rotated
90º to the top, see Fig. 4-7A. This design was cast with AZ91 at 730 °C in order to keep the other casting
variables constant. The casting conditions and pattern dimensions are summarized in Table 4-8 and Table
4-9. Fig. 4-7B shows that the pouring cup had collapsed on the sprue by its own weight. The impact from
falling liquid metal has also caused the core to collapse and metal penetration through the ceramic shell to
the supporting sand. The collapse was caused by insufficient compaction of the refractory sand because
the sand could not fill the middle of the sandwich core properly from the sides only. The sprue design was
also alternated by tapering towards the sandwich panel, see Fig. 4-7C. The tapered sprue design with
smaller pouring cup was expected to reduce the chance of core collapse and metal penetration through the
ceramic shell. As shown in Fig. 4-7D, although the pouring cup did not collapse onto the pattern, the
tapered sprue design did not prevent pattern collapse within the truss core. The major flaw to this design
was the insufficient flow of refractory sand to the sandwich panel below the horizontal sprue. The
fillability problem was not reduced by this horizontal sprue design either; neither the face sheets nor core
struts were completely filled.
Chapter 4: Results and Discussion
38
Figure 4-7. Pyramidal sandwich pattern assembly with top feeding sprue (A), AZ91 as-cast sample (B),
altered pattern assembly with tapered sprue (C) and AZ91 as-cast sample (D).
Table 4-8. The casting parameters for the horizontal sprue pattern assembly.
Pouring Temperature (°C) Foam Properties Alloys
730 EPS with 0.021 g/cm3 (1.3 pcf) AZ91
Table 4-9. The dimensions of the horizontal sprue pattern assembly.
Web thickness
tc (mm)
Face sheet thickness
hfs (mm)
Relative Density
ρ~ (%)
Pattern size
(mm)
Sprue dimension (mm)
3 5 22 170 X 170 (Square) 250 X 50 X 50
(Tapered) 170 X 20 X 40
From the various sprue and runner designs, vertical sprue produced the castings without pattern
collapses and refractory sand filling issues, a pair of face sheets helped to improve filling to the truss core
by producing direct flow into each strut.
4.2.2. Vacuum Assisted LFC
Since the horizontal sprue design impeded the flow of refractory sand and sand compaction, a
vertical sprue, having smaller cross-sectional area, was used, see Fig. 4-8A. The narrow vertical sprue
tended to minimize the amount of foam on the wall from the hollow sprue, which reduced the amount of
heat extracted from the liquid metal during foam degradation. The casting conditions and pattern
Chapter 4: Results and Discussion
39
dimensions are summarized in Table 4-10 and Table 4-11. The casting, as shown from Fig. 4-8B, did not
show significant improvement in fillability compared to previous casting from Fig. 4-6C. The uneven
fillability between the face sheets could be affected by the amount of hot melt glue used in between each
contact area with the sprue.
Figure 4-8, Pyramidal sandwich assembly with narrow and lengthen sprue (A) and AZ91 as-cast
sample (B).
Table 4-10. The casting properties for the narrowed vertical sprue pattern assembly.
Pouring Temperature Foam Properties Alloy
730°C EPS with 0.021 g/cm3 (1.3 pcf) AZ91
Table 4-11. The dimensions of the narrowed vertical sprue pattern assembly.
Web thickness tc
(mm)
Face sheet
thickness hfs (mm)
Relative Density
ρ~ (%)
Pattern size (mm) Sprue dimension
(mm)
3 5 22 170 X 170 250 X 40 X 40
A range of casting conditions were tested in terms of pouring temperature and applied vacuum
level (Table 4-12) to improve the fillability. Samples S1, S2 and S3 were cast at a pouring temperature of
730 °C and at vacuum levels ranging from 0 to 40 kPa. Samples S4 to S6 were cast at a pouring
temperature of 750 °C with vacuum levels ranging from 20 to 40 kPa. Finally samples S7 and S8 were
cast at 750 °C and vacuum levels of 40 and 50 kPa, respectively, but were double coated with the
refractory slurry.
Chapter 4: Results and Discussion
40
Table 4-12. Summary of casting conditions: the vacuum level, pouring temperature, and number
of coatings for the eight different casting conditions.
Sample Vacuum level (kPa) Temperature (°C) Number of coatings
S1 0
730
1
S2 14
S3 40
S4 20
750
S5 30
S6 40
S7 40 2
S8 50
The particular casting condition had a significant effect on the extent of face sheet and truss core
filling. Samples cast at 730 °C exhibited both incomplete core and face sheet filling, regardless of the
vacuum level (Fig. 4-9: S1, S2, and S3). Increasing the pouring temperature to 750 °C generally increased
the fillability (Fig. 4-9: S4, S5, and S6) and complete filling could be obtained for the 30 kPa (S5) and 40
kPa (S6) vacuum levels. However, applying vacuum also had the effect of inducing metal penetration
through the ceramic shell, which resulted in poor surface finish and sand adhesion. Double coating the
EPS pattern (S7 and S8) was attempted to create a thicker ceramic shell in order to prevent metal
penetration. The core filling of sample S7 was significantly reduced compared to the equivalent single
coated sample (S6). Complete face sheet and core filling could be achieved in the double coated condition
by increasing the vacuum level to 50 kPa (S8). However, the surface finish was only marginally
improved. Fig. 4-10 and Fig. 4-11 summarize the face sheet and core filling for all eight samples. As
would be expected, it was significantly easier to completely fill the face sheets than the pyramidal core.
Chapter 4: Results and Discussion
41
Figure 4-9. The particular casting condition for the eight samples (summarized in Table 4-12) had a
significant effect on the extent of fillability. Complete pattern was obtained in samples S5, S6, and S8.
0
20
40
60
80
100
S1 S2 S3 S4 S5 S6 S7 S8
% F
ille
d T
ru
sse
s
0
20
40
60
80
100
S1 S2 S3 S4 S5 S6 S7 S8
% F
ille
d F
ace
Sh
ee
t
Figure 4-10. Percentage of truss core (left) and face sheets (right) filled for the eight different casting
conditions (Table 4-12).
A B C D
E F G H
S1 S2 S3 S4
S5 S6 S7 S8
Chapter 4: Results and Discussion
42
0
20
40
60
80
100
60 70 80 90 100
% F
ille
d T
ru
sse
s
% Filled Face Sheets
S1: gravity cast at 730°C
S2: 14kPa vacuum at 730°C
S4: 20kPa vacuum at 750°C
S6: 40kPa vacuum at 750°CS5: 30kPa vacuum at 750°C
S3: 40kPa vacuum at 730°C
S7: 40kPa vacuum at 750°C 2 coats
S8: 50kPa vacuum at 750°C 2 coats
Figure 4-11. Casting map summarizing the extent of face-sheet and truss core filling.
Porosity and microhardness measurements were obtained from three different positions in the S1,
S3, S6, and S7 samples. Sample S1 showed a slight increase in porosity as distance increased from the
sprue where the cooling rate was slowest; this is likely related to proper metal feeding in the near sprue
region before interdendritic solidification. The overall porosity values were similar at ~3 % for the S1, S3
and S7 samples (Table 4-13). The fact that sample S7 had a lower porosity than sample S6 may be related
to the extra slurry coating given to S7, which would be expected to reduce the rate of heat transfer to the
surrounding sand. However, sample S6 also exhibited some sand adhesion due to metal penetration
through the ceramic shell, which may have an effect on the measured porosity value. The morphology of
the porosity was studied by scanning electron microscopy of polished surfaces; similar pore
morphologies, which were consistent with interdendritic porosity, were seen for all four samples (Fig. 4-
12). There was no clear evidence of spherical porosity in any of the samples which might indicate gas
porosity.
Microhardness measurements were obtained after following ASTM E92-82 [Mayer ed., 2007] on
face sheet cross-sections at 3 different positions in each of the 4 samples. The values are summarized in
Table 4-13. There was little difference in the average microhardness between samples cast using the four
Chapter 4: Results and Discussion
43
different conditions and the measured values are consistent with the reported hardness value of 70 HV for
as-cast AZ91D [Avedesian and Baker ed., 1999].
Table 4-13. Summary of porosity and microhardness characterization.
Sample Porosity (%) Hardness (HV)
S1: cast at 730°C in gravity 3.4 ± 1.9 66 ± 4.9
S3: cast at 730°C in 40 kPa vacuum 2.8 ± 1.3 66 ± 4.5
S6: cast at 750°C in 40 kPa vacuum 7.0 ± 3.1 65 ± 5.3
S7: cast at 750°C in 40 kPa vacuum with 2 coats 3.5 ± 1.7 65 ± 6.5
Figure 4-12. SEM images showing exposed porosity on polished surfaces of sample S1 (A), sample S3
(B), sample S6 (C), and sample S7 (D).
Fillability was a significant issue in magnesium LFC with PCMs, the reduced cross section of the
struts resulted in rapid solidification. Using higher casting temperatures and with an applied vacuum, it
was possible to completely fill the PCM pattern at the expense of some sand adhesion. This sand adhesion
prevented any accurate mechanical property to be obtained from the samples.
4.2.3. Alloying Additions
Since the vacuum assist LFC did not improve PCM fillability significantly, pure aluminum was
added to improve castability by broadening the solidification range [Avedesian and Baker ed., 1999] or
100µm
(a) (b)
(c) (d)
Chapter 4: Results and Discussion
44
aluminum-silicon master alloy was added to increase fluidity [Sheng et al., 2009]. Probeads 70 foam was
used instead of EPS (see Table 4-14) to improve the fillability by lowering the volatile temperature and
lowering the decomposition energy [Shivkumar, 1994], the pattern size was slightly shortened and the
sprue was slightly lengthen (see Fig. 4-13 and Table 4-15) to minimize misruns as seen previously. In the
first series of experiments, pure aluminum was added to AZ91D to increase the aluminum content by
~3 wt% (close to maximum solubility of α Mg-Al phase, ~11.5 wt% Al) and by ~14 wt% (closer to
eutectic composition of α and β phase, 32 wt% Al). In a second series of experiments, various amounts of
Si (~0.5, ~1.0 and ~1.5 wt%) were added to AZ91D using Al-50%Si master alloy. See Table 4-16 for the
chemical composition resulted from alloying additions, the bold values were altered by the alloying
additions.
Thermal analysis was performed using thermocouples positioned at six locations in the pattern, as
indicated in Fig. 4-14, to study the cooling and solidification characteristics of each alloy. The cooling
curves were recorded for all alloy compositions. The first derivative (dT/dt) of these curves was also
determined (Fig. 4-18). Thermal analysis was also used to determine solidification time and velocity
within the panel.
Table 4-14. The casting properties for the narrowed vertical sprue pattern assembly.
Pouring Temperature Foam Properties Alloy
750°C Probeads with 0.025 g/cm3 (1.5 pcf) AZ91
Table 4-15. The dimensions of the narrowed vertical sprue pattern assembly.
Web thickness tc
(mm)
Face sheet
thickness hfs (mm)
Relative Density
ρ~ (%) Pattern size (mm)
Sprue dimension
(mm)
3 5 22 85 X 170 450 X 40 X 40
Chapter 4: Results and Discussion
45
Figure 4-13. The assembled coated pattern (A) and sandwich panel in close up (B).
Table 4-16. Measured chemical composition by spark emission spectroscopy for each casting
condition in wt%.
Mg Al Zn Mn Si Cu Fe =i Be
S1 90.62 8.51 0.58 0.23 0.043 0.003 0.062 0.0014 0.0006
S2 87.10 12.1 0.55 0.21 0.021 0.003 0.0069 0.0013 0.0008
S3 76.06 23.14 0.50 0.19 0.053 0.005 0.0378 0.0017 0.0009
S4 89.38 9.23 0.57 0.23 0.572 0.003 0.0077 0.0013 0.0011
S5 89.57 8.81 0.57 0.21 0.824 0.003 0.0052 0.0013 0.0011
S6 89.11 8.76 0.55 0.18 1.389 0.003 0.0060 0.0012 0.0006
Figure 4-14. Schematic representation showing the position of thermocouples in the pattern (top: T1, T2,
T3 and bottom: B1, B2 and B3).
Fig. 4-15 shows photographs of the as-cast structure for each alloy composition. For the S1
(reference sample), the face sheets were completely filled, while many struts were unfilled in the outer
edge of the casting. With an addition of 3 wt% Al (S2), the casting showed slightly more filled struts on
the outer edge as compared with S1. Further increase in Al content (S3) resulted in the complete filling of
1 2 3
1 2 3
Top
Bottom
Chapter 4: Results and Discussion
46
the casting. However, severe metal penetration through the ceramic shell was seen inside the sandwich
core.
With the addition of 0.6 wt% Si to AZ91D in S4, the casting showed slightly improved filling of
the internal struts as compared with S1. The extent of filling was similar to that seen for S2. As the weight
addition of Si increased from 0.6 to 0.8 wt% (S5), there was no noticeable change in filling. With 1.4 wt%
Si addition (S6), only one corner of the casting did not fill. With the highest Al (15 wt%) or Si (1.4 wt%)
additions, the entire casting was improved significantly in terms of filling.
Figure 4-15. The final casting of magnesium PCMs with different alloying additions.
Fig. 4-16 represents the maximum temperature and the time to reach this temperature as a
function of the location in the casting. The melt was poured at 750 °C and it was observed that the
temperature decreased to ~700 °C as the melt reached the bottom of the sprue (B3), as shown in Fig. 4-16
on the left. As the melt entered from the sprue and flowed into the sandwich panel, the maximum
temperature dropped to ~620 °C along the panel adjoining to the sprue (T2) and (B2). The temperature
decreased even further to ~580 °C as the melt continued to flow through the 5 mm-thick panel and into
the 3 mm-thin struts. The decrease in temperature was attributed to the heat transfer from the metal to the
S3: AZ91D + ~15 wt% Al
S2: AZ91D + ~3wt% Al S1: AZ91D
S4: AZ91D + ~0.6 wt% Si
S5: AZ91D + ~0.8 wt% Si S6: AZ91D + ~1.4 wt% Si
Chapter 4: Results and Discussion
47
surrounding materials and thermal degradation of the foam. The decrease in temperature from T3 to T2
was almost twice that from T2 to T1. Assuming that the heat loss to the surrounding materials was similar
across the panel, and that the distances between the thermocouples were the same, the temperature loss
results were attributed to the fact that a greater amount of heat from the melt between the sprue and
sandwich panel is required to degrade the hot melt glue, as compared with that required to degrade the
foam.
The order of filling at the different locations can be determined by the time taken for each to
reach the maximum temperature (Fig. 4-16 on the right). As the melt was poured, it first entered the
bottom of the sprue (B3) and then filled to the top of the sprue (T3). The melt slowly spread to the panel
between T2 and B2 in 3 s. Then, the melt flowed across the panel to the edge between T1 and B1. The
difference in time needed to reach the maximum temperature with respect to the distance between the top
and bottom parts of the panel corresponded to a melt flow velocity of ~1.66 cm/s at the bottom of the part
(B2 to B1), as compared to ~1.06 cm/s at the top (T2 to T1). The difference in velocity could be due to
the pressure gradient between the top and bottom parts of the panel.
Fig. 4-17 shows the solidification time and cooling rate at different locations in the casting. Due
to the asymmetrical feeding of the sprue design, there was a difference in cooling rate and solidification
time across the sandwich panel because part of the heat from the melt was transferred to the surrounding
material at regions 2 and 3, see Fig. 4-14. As the melt flowed across the panel, the melt temperature
dropped significantly and had a very short solidification time of ~50 s at the edge (T1 and B1) compared
with more than ~250 s at the panel along the sprue (T2 and B2) shown in Fig. 4-17 on the right. The
cooling rate was significantly different at the edge (T1 and B1) as compared with that in the sprue (T2
and B2) and panel near the sprue (T3 and B3), see Fig. 4-17 on the left. The fast cooling rate at the edge
of the face sheet prevented sufficient feeding of the struts.
A section of the sprue from each casting was examined for changes in microstructure with the
alloying addition. As-cast AZ91 contains a eutectic microstructure with lighter region of primary α-Mg
Chapter 4: Results and Discussion
48
and darker region of β- Mg17Al12 phase [Avedesian and Baker ed., 1999] which is consistent with S1
(contains small amount of porosity) shown in Fig. 4-18. The corresponding cooling and first derivative
curves are shown next to the microstructure to illustrate the consistency in casting conditions between
samples. S2 alloy (3 wt% Al addition) exhibited a similar microstructure as S1. With further addition of
Al (S3), there was a large increase in the amount of β, Mg17Al12 phase compared to reference sample.
Significant spherical porosity was also observed throughout the sample. A possible source of gas porosity
would be severe gas absorption [Davis ed., 1993] with high aluminum content and at the high superheat
(~370 °C) but without proper hydrogen removal procedure. Mg2Si has a dark grey chinese script
morphology found within magnesium alloys [Avedesian and Baker ed., 1999] which is consistent with
S4, S5 and S6 alloys. These alloys also showed small amounts of porosity in all samples.
500
600
700
800
T1 T2 T3 B1 B2 B3
Maxim
um
Tem
pera
ture
(°C
)
Thermocouple Location
6(A) Temperature
0
5
10
15
20
T1 T2 T3 B1 B2 B3
Tim
e to reach the m
axim
um
te
mpera
ture
(s)
Thermocouple Location
6(B) Time
Figure 4-16. Temperature (left) and time (right) profiles: T1-T3 (top of panel) and B1-B3 (bottom of
panel).
0
50
100
150
200
250
300
350
400
450
T1 T2 T3 B1 B2 B3
Solidific
ation T
ime (s
)
Thermocouple Location
7(A) Solidification Time
0
1
2
3
4
T1 T2 T3 B1 B2 B3
Cooling R
ate
(°C
/s)
Thermocouple Location
7(B) Cooling Rate
Figure 4-17. Solidification time (left) and cooling rate (right) profiles: T1-T3 (top of panel) and B1-B3
(bottom of panel).
Chapter 4: Results and Discussion
49
Considering now the cooling curves in Fig. 4-18, each sample contained one peak at ~430 °C
which was consistent to the temperature formation of the β- Mg17Al12 phase in Mg-Al phase diagram
[Avedesian and Baker ed., 1999]. The temperature of formation for Mg2Si is higher (1085 °C) [Avedesian
and Baker ed., 1999] than the temperature of the melt which explained why no peak was detected during
casting. The initial formation of α Mg-Al phase could not either be detected because when the melt
flowed across the panel, the melt temperature was already below the AZ91 liquidus (630 °C) [Avedesian
and Baker ed., 1999].
S4: AZ91D + ~0.6 wt% Si
Mg17Al12
Mg2Si
S2: AZ91D + ~3 wt% Al
S3: AZ91D + ~15 wt% Al
Mg17Al12
S1: AZ91D Reference
Mg17Al12
Mg17Al12
100 µm
Chapter 4: Results and Discussion
50
Figure 4-18. Optical micrographs (at same magnification, see scale bar from sample S4 or S6) and
thermal analysis of the castings.
Fig. 4-19 presents the porosity levels measured by Archimedes method for each alloy. All of the
casting except S3 had ~5 % porosity. The high porosity measured in S3 could be due to hydrogen gas
from the relatively high content of aluminum without the use of aluminum degasser, as explained
previously. S6 has the best fillability based on the least number of unfilled struts and comparable porosity
level with S1. S6 represents the best compromise in term of castability.
Figure 4-19. The measured porosity levels with Archimedes principle.
S5: AZ91D + ~0.8 wt% Si
Mg2Si
Mg17Al12
Mg2Si
Mg17Al12
S6: AZ91D + ~1.4 wt% Si
100 µm
Chapter 4: Results and Discussion
51
Hardness measurements were performed following ASTM E92-82 [Mayer ed., 2006] on the sprue
of each sample. The results are presented in Fig. 4-20. A hardness value of ~77 HRE was found for S1
and was comparable to previously published data (HRE 75) [Avedesian and Baker ed., 1999]. Hardness
was found to increase with the addition of 3 wt% Al (S2). It was impossible to do accurate hardness
measurement on the S3 sample, due to the presence of significant gas porosity. With Si addition (S4, S5
and S6), there was no apparent change in hardness as compared with AZ91D.
Figure 4-20. Hardness of the cast samples.
Aluminum and silicon additions were used in this project to improve fluidity of AZ91D
magnesium alloy. The results suggested that 1.4 wt% silicon addition showed the best compromise in
terms of porosity and fillability. With this composition, it was possible to produce magnesium PCMs with
very thin sections.
4.2.4. Preliminary Compressions Test Results
Preliminary mechanical testing in uniaxial compression was conducted on filled sections of the
gravity cast (S1) from section 4.2.2. since any vacuum assist LFC PCMs had sand adhesion due to metal
penetration through ceramic shell. A typical test coupon is shown in Fig. 4-21 (2×3.5 pyramidal truss core
cells). The compression result is shown in Fig. 4-22, the stress first increased linearly with strain, then
Chapter 4: Results and Discussion
52
reached a maximum (peak strength) and finally decreased to a plateau. The series of decreasing peaks
after the initial maximum corresponded to strut member fractures and delaminations between the struts
and the face sheets, the onset of which was observed during loading. Two distinct types of failure were
observed after the initial elastic region. In the first case, failure of the struts occurred at a peak strength of
~10 MPa (10.2 MPa for the sample shown in Fig. 4-22). In the second case, failure occurred at only
~5 MPa. The peak strength in this case corresponded to early fracture at the node / face sheet interface
(Fig. 4-23); the nodes that had failed were only partially complete, suggesting that this type of failure
mechanism is an artifact of sample preparation. Fillability was a significant issue in the lost foam casting
of magnesium alloy PCMs; the reduced cross section of the struts resulted in rapid solidification.
Figure 4-21. Typical compression test coupon (2×3.5 pyramidal truss core cells).
Figure 4-22. Typical uniaxial stress strain curves showing the mechanical behaviour with and without
sample preparation edge effects.
Chapter 4: Results and Discussion
53
Figure 4-23. Progressive node/face sheet fracture during compression testing (image labels correspond to
the testing points shown on Fig. 4-22).
With pure aluminum additions, S2 was chosen for compression loading because it did not have
the level of porosity found in S3. With aluminum-silicon additions, 1.4 wt% silicon added AZ91 (S6)
showed the best compromise in terms of porosity and fillability. The previous gravity cast PCM without
edge effect (as S1) was used to compare as typical stress-strain curve (see Fig. 4-24) for the AZ91 PCM
with AZ91+ 3 wt% Al and AZ91 + 1.4 wt% Si. There were some variations in the characteristic of the
loading curves among the samples within the same casting. The average of 3 peak strengths in each
conditions were presented as 8.7 MPa ± 2.1 for S1, 9.2 MPa ± 1.8 for S2 and 8.2 MPa ± 1.5 for S6.
Figure 4-24. Comparisons between different AZ91 PCMs in compressions.
Sample-to-sample variability from within each casting was larger than the effect of alloying
addition on the mechanical properties and may be due to casting defects (porosity) or the foam structure
A B
C D
Chapter 4: Results and Discussion
54
defects (uneven strut thickness). In addition, the defects caused by the hot melt glue could cause
delaminations between the core strut during loading. Therefore, the peak stresses between the different
alloy compositions could not be meaningfully compared.
Overall, this part of the study used a light alloy approach to achieve a high specific strength ratio
at a high relative density of 22 %. The ~8.7 MPa peak strength of AZ91 PCMs corresponded to a specific
strength of 23 MPa/(g/cm3), which is comparable to the specific strength of 24 MPa/(g/cm
3) for
investment cast Cu-2%Be (having a relative density ~2.25 %) [Chiras et al. 2002].
Chapter 4: Results and Discussion
55
4.3. Integrated Pyramidal Sandwich
In this study, a new pattern was developed to eliminate the use of hot melt glue between the face
sheets and struts. This design reduced the amount of heat extracted and gas evaporated while filling the
struts to improve the fillability of the pattern. A lower strut thickness could then be used without misrun
issues. This study used 4 different strut thicknesses (1.5 mm, 2.0 mm, 2.5 mm, 3.0 mm) to determine the
degree of fillability and the effect of relative density (11 %, 12 %, 14 % and 16 %) on the compressive
mechanical properties. For each strut thickness design, a special template was machined to produce four
foam patterns that were attached to each sprue (see Fig. 4-25A) for casting in A356 aluminum and AZ91
magnesium alloys (see Fig. 4-25b). The casting parameters were held constant (Table 4-17) and the only
variable in this study was the strut thickness (see Table 4-18) resulting in different relative densities. Four
castings were produced for each alloy, each with a different relative density. Within each casting, four
sandwich panels were sectioned and some were heat treated (solutionized or solutionized and aged) to
determine differences in mechanical properties. Then each sandwich panels were sectioned to 2-by-2
struts for uniaxial compressive loading.
Figure 4-25. Integrated sandwich assembly (A) and AZ91 as-cast sample (B).
Table 4-17. Summary of casting parameters for integrated sandwich.
Pouring Temperature Foam Properties Alloy
750 °C Probeads with 0.021 g/cm3 (1.5 pcf) AZ91, A356
Chapter 4: Results and Discussion
56
Table 4-18. The dimensions of integrated sandwich pattern assembly.
Web thickness tc
(mm)
Face sheet
thickness hfs (mm)
Relative Density
ρ~ (%)
Pattern size (mm) Sprue dimension
(mm)
1.5, 2.0, 2.5, 3.0 5 11, 12, 14, 16 ~70 X ~60 350 X 40 X 40
4.3.1. A356 PCMs Compression Test Results
The typical compressive loading curves for A356 PCMs are shown in Fig. 4-26 (left). As the
PCMs compressed and passed their yield strengths, the strut members continued to plastic deform and
their cross-section area increased with no observable crack initiated or propagated (Fig. 4-27B). During
compression, the PCMs were unloaded and loaded again to determine if there was any change in PCMs’
loading modulus as they were plastically deformed. During the entire compression, A356 PCMs showed
continuous and linear increase in stress but the rate of increase in stress reduced to a steady rate (Fig. 4-
26) after yielding. The photographed strut morphology showed the surface texture of the as-cast strut was
largely influenced by surface texture of the foam pattern. The surface of the as-cast strut had great surface
details, the gaps between poorly fused foam bead could be seen.
Figure 4-26. Stress-strain curves (left) and rate of change in stress with strain (right) for typical A356
PCMs in compression with different relative density ρ~ .
Chapter 4: Results and Discussion
57
Figure 4-27. Strut morphology initially (A) and after compression testing (B).
A summary of the different compressive yield strengths of the A356 PCMs as a function of
relative density is shown in Fig. 4-28; the A356 PCMs’ yield strength increased linearly with relative
density due to increased strut cross sections. Heat treatments were also applied to some of the PCMs prior
to compression as an attempt to improve their compressive properties by modifying the A356
microstructure. The compressive stress-strain curves of the solutionized, and the solutionized and aged
samples were similar to as-cast as shown in Fig. 4-28, when the PCM yield strength increased with
relative density linearly. A356 PCMs’ yield strengths were summarized with different relatives and heat
treatments in Fig. 4-29. The solutionized PCMs (Fig.4-29 middle) demonstrated slightly lower yield
strength compared to the as-cast PCMs (Fig. 4-29 left), and the solutionized and aged PCMs (Fig. 4-29
right) showed the highest yield strengths out of all three.
Fig 4-28. Stress-Strain curves for typical solutionized (left) and solutionized and aged (right) A356 PCMs
in compression with different relative density ρ~ .
Chapter 4: Results and Discussion
58
Figure 4-29. A356 PCM’s yield strength (0.2 % offset) with respect to relative density ρ~ : as-cast (left),
solutionized (middle), and solutionized and aged (right).
As-cast, and solutionized and aged A356 PCMs’ strains increased linearly with relative density at
which they yield at 0.2 % offset as shown in Fig. 4-30. In contrast, the solutionized PCMs behaved as
opposite, the strains decreased linearly with relative density. A series of unloading and reloading during
compression were applied to determine PCMs’ unloading modulus as they were deformed. The unloading
modulus increased with strains and deformations but there was no clear difference between heat
treatments found in Fig. 4-31. Fig. 4-32 showed no clear relationships between PCM’s relative density in
each heat treatment conditions. The linear increase in unloading modulus was related to the amount of
deformations, strain or change in strut angle with deformation introduced to PCMs.
Figure 4-30. A356 PCMs’ strains at which they yield at 0.2 % offset with respect to relative density after
different heat treatments.
Chapter 4: Results and Discussion
59
Figure 4-31. A356 PCMs’ modulus with respect to strains after different heat treatments.
Figure 4-32. A356 PCMs’ modulus with respect to strains and different relative densities: as-cast (left),
solutionized (middle), and solutionized and aged (right).
Three etched microstructures were presented in Fig. 4-33 to show the difference after heat
treatments. Fig. 4-33A had typically dendritic structure found in as-cast A356, the dark regions are silicon
precipitates. Fig. 4-33B had the typical microstructure found after solutionized heat treatment, there was
less amount of silicon precipitates (dark shade regions) found between the dendrite arms (light grey
regions). Fig. 4-33C showed the change of more finely dispersed silicon precipitates around the structures
after solutionized and aged.
Chapter 4: Results and Discussion
60
Figure 4-33. A356 etched microstructures in as-cast (A), solutionized (B), and solutionized and Aged (C).
Different aluminum alloys in investment cast PCMs [Desphande and Fleck, 2001; Zhou et al.,
2004], extruded PCM [Queheillalt et al., 2008] and stretch formed PCMs [Kooistra et al., 2004] were
compared with the present results presented in Fig. 4-34. The yield strengths from aged PCMs [Kooistra
et al., 2004; Queheillalt et al., 2008] demonstrated higher increase in slope with relative density compared
to the rest. The present study had higher PCM density due to thicker strut designs. The trend of increasing
yield strength with relative density was consistent between present study and the rest. Yet the aged PCMs
in present study did not show as much increase as seen in data from Kooistra et al., 2004.
Figure 4-34. A comparison of different aluminum alloy PCMs and the current studied LFC PCMs.
Chapter 4: Results and Discussion
61
4.3.2. AZ91 PCMs Compression Test Results
In contrast, AZ91 PCMs behaved very differently beyond yielding during compression compared
to A356 PCMs. In as cast condition, the compression results from AZ91 PCMs were presented in stress-
strain curves in Fig. 4-35 (left), the cracks initiated (Fig. 4-36B) and propagated within the struts, and
fractured at peak stress on the plane normal to the aligned strut (Fig. 4-36C). After the stress dropped
rapidly, it plateau as the fractured core compressed until face sheets were deformed then stress rose again
(Fig. 4-36D). The rate of change in stress was rapid and unsteady after the peak load, this can be seen in
Fig. 4-35 (right). This might be affected by the surface detail left behind from the poor fused foam beads
acting as stress concentrations for crack initiation. AZ91 PCMs were solutionized or solutionized and
aged as well, the compression results showed similar behaviours as as-cast AZ91 PCMs, shown in Fig. 4-
37.
Figure 4-35. Stress-strain curves (left) and change in strain (right) for typical AZ91 PCMs in compression
with different relative densities ρ~ .
Figure 4-36. Struts morphology initially (A), after peak (B), expanded view of the fractured struts after
peak (C) and after stress plateau (D).
Chapter 4: Results and Discussion
62
Fig 4-37. Strain-strain curves for typical solutionized (left) and solutionized and aged (right) AZ91 PCMs
in compression with different relative densities ρ~ .
AZ91 PCMs’ yield strengths increased with relative density as summarized in Fig. 4-38, the line
of best fits showed a linear relationship similar to A356 as-cast PCMs. Heat treatments were also applied
to AZ91 PCMs as well, the yield strengths are higher with solutionized and aged, followed with as-cast
and solutionized samples similar to A356 PCMs. Solution heat treatment able to improve the maximum
toughness and with following artificial aging, the maximum hardness and yield strength increased but
lowered the ductility [Avedesian and Baker ed., 1999].
Figure 4-38. AZ91 PCM’s yield strength (0.2 % offset) with respect to relative density ρ~ : as-cast (left),
solutionized (middle), and solutionized and aged (right).
In Fig. 4-39, the amount of strains introduced to AZ91 PCMs when they yield at 0.2 % offset
increased with relative density for as-cast and solutionized and aged samples. Yet solutionized samples
behaved in opposite, the amount of strains decreased with relative density when they yielded. The same
Chapter 4: Results and Discussion
63
behaviours were found in A356 PCMs as well. In contrast, AZ91 PCMs deformed and collapsed after
certain peak stresses which were not found from A356 PCMs. The amount of strains required for peak
stress increased with relative density as seen in Fig. 4-40 for all three heat treatment conditions. The
solutionized PCMs required very different amount of strains for yielding and peak stresses as relative
density increases. In yielding, the amount of strains required decreased but the amount of strains required
increased with peak stress.
Figure 4-39. AZ91 PCMs’ strains at which they yield at 0.2 % offset with respect to relative densities
after different heat treatments.
Figure 4-40. AZ91 PCMs’ strains at peak stress with respect to relative density after different heat
treatments.
Unlike A356 PCMs, AZ91 PCMs’ unloading modulus were scattered across the different strains
and there were no correlations between heat treatments from Fig. 4-41. In Fig. 4-42, each heat treatment
Chapter 4: Results and Discussion
64
also showed no clear relationship between unloading modulus with strains on different relative densities.
This might be affected by the crack initiation and propagated around 0.05~0.1 strain at peak stress. Fig. 4-
43 showed the difference in microstructures with different heat treatments; the as-cast microstructure
showed typical equiaxed dendritic microstructure with lighter region of primary α-Mg matrix and darker
region of β- Mg17Al12 phase [Avedesian and Baker ed., 1999]; the solution treatment dissolved all the β-
Mg17Al12 lamella around grain boundaries but all its spheroidal precipitates remained within the grains
which showed similar microstructure as found literature [Fujii et al., 2007]; the aged treatment caused all
the β- Mg17Al12 to reappeared as larger clusters along grain boundaries also found similar to the literature
[Fujii et al., 2007].
Figure 4-41. AZ91 PCM’s modulus with respect to strains after different heat treatments.
Figure 4-42. AZ91 PCMs’ modulus with respect to strains and different relative densities: as-cast (left),
solutionized (middle), and solutionized and aged (right).
Chapter 4: Results and Discussion
65
Figure 4-43. AZ91 etched microstructures in as-cast (A), solutionized (B), and solutionized and Aged (C).
Since there was no magnesium PCMs produced and studied to date, a comparison of different
magnesium foam was made with the current study in Fig. 4-44. AZ91 Foams [Yamada et al., 1999;
Yamada et al., 2000] has very low density but the yield strengths are relatively low compared to LFC
PCMs. The mechanical difference between the present studied PCMs and previously studied metal foams
is due to its stretch-dominated mechanisms and bending-dominated mechanisms respectively [Ashby,
2005]. The effectiveness of the foam structure decreases with high relative density foams [Wen et al.,
2004] compared to bulk yield strength of ~90 MPa in AZ91 and ~21 MPa in pure magnesium [Avedesian
and Baker ed., 1999].
Figure 4-44. A comparison on yield strength with respect to density between magnesium foams and
present study of AZ91 LFC PCMs.
Chapter 4: Results and Discussion
66
4.3.3. Lower and Upper Bound Model
A theoretical model was developed to predict the compressive property of PCMs, assume the
PCM’s strut would failure at the minimum cross section regions by yielding (see Fig. 4-45). This lower
bound model was derived by using the material yield strength with oriented compressive force and cross-
section area with the following formula,
φ
σσ
SI'A
'A
PCM
ys
PCMys
0= (Eqn. 4-1)
where σys PCM = compressive yield strength of PCM, σys = compressive yield strength of material, A0 =
minimum cross-section area of the strut, ' = number of struts with each PCM under compression, APCM =
cross-section area of PCM under compression, φ = Strut angle (60º) and tc = strut thickness. And the
minimum cross-section strut area can be found from the following equations as well,
φφφ 2
2
2
2
0
42
2
2
TA'
t
SI'
t
SI'
tA ccc −= (Eqn. 4-2)
Figure 4-45. Lower bound model based on minimum strut cross-section thickness, tc.
Chapter 4: Results and Discussion
67
The material compressive yield strength could be found from literature but with variations
depending on the processing parameters. Therefore, individual square blocks (length to diameter ratio
below 2 to avoid buckling [Faupel and Fisher, 1981]) were machined from the PCM’s casting sprue to
determine the closest compressive property of A356 and AZ91 with different heat treatments. The same
stress-strain curve behaviour can be seen in A356 square compression blocks (Fig. 4-46) with A356
PCMs. The aluminum blocks would yield and deformed continuously with increased stress until the
equipment’s load limit. Within the tested compressive stress, the aluminum blocks showed cracks
initiated and propagated on the block’s surface and the height was shortened with width expanded as seen
in Fig.4-47 and Fig. 4-48.
0
200
400
0 0.2 0.4
ε
σcomp (MPa)
A356 As Cast
A356 Solutionized
A356 Solutionized and Aged
Figure 4-46. Stress-strain curve of A356 square compression blocks with different heat treatment
conditions.
Figure 4-47. A356 sample block after compression (A) and magnified surface feature of the block (B).
Chapter 4: Results and Discussion
68
Figure 4-48. A356 square compression block`s surface with crack propagations.
The results from the stress-strain curves of AZ91 compression blocks (Fig. 4-49) confirmed the
similar deformation behaviour found in AZ91 PCMs, the sample would deformed to a certain strain and
fractured. AZ91 square compression block would failure by shear at 45º to the loading axis. The fracture
surface (Fig. 4-50 and 4-51) demonstrated many parallel plateau and ledges that matched to transgranular
cleavage found in [Hertzberg, 1996]. It was transgranular rather than intergranular cleavage fracture
because the crack did not follow 3-dimensional grain morphologies. This was different from A356 square
compression block, A356 did not fracture and only showed minor cracks initiated and propagated on the
surface (Fig. 4-47 and 4-48).
Chapter 4: Results and Discussion
69
0
200
400
0 0.1 0.2 0.3
ε
σcomp (MPa)
AZ91 As Cast
AZ91 Solutionized
AZ91 Solutionized and Aged
Figure 4-49. Stress-strain curve of AZ91 square compression blocks with different heat treatment
conditions.
Figure 4-50. AZ91 sample block’s fractured cross section after compression (A) and its magnified surface
(B) showing transgranular fractures with parallel plateau and ledges.
Chapter 4: Results and Discussion
70
Figure 4-51. AZ91 square compression block fracture surface after 45° fracture.
The yield strengths from A356 and AZ91 compression blocks with different heat treatments are
shown in Fig. 4-52, the solutionized AZ91 and A356 compression blocks were found to be higher than as-
cast conditions (Fig. 4-52), which was conflicted with their PCMs.
Chapter 4: Results and Discussion
71
Figure 4-52. Compressive yield strength of A356 and AZ91 compression blocks with different heat
treatments.
With the compression block results, the lower bound model was developed with the determined
yield strength with different heat treatments in AZ91 and A356. Fig. 4-53 showed the difference between
experimental data found in A356 PCMs and lower bound model found from Eqn. 4-1 and Eqn. 4-2. Both
line of best fits showed similar slopes which proved that the change in strut thickness was related to yield
strength required for its derived minimum cross-section area. The gaps between the experimental data and
lower bound model can be explained by difference between the theoretical and actual strut thickness. The
gaps were not consistent between the three different heat treatments, the solutionized PCMs tended to be
lower than predicted; their yield strength has reduced significantly compared to their bulk samples.
Similar trend is shown in Fig. 4-54, AZ91 with increasing yield strength with relative density. The gap
between the experimental data and lower model was more consistent throughout different heat treatments
compared to A356. By comparing the yield strength of compression blocks to PCMs in A356 and AZ91,
the compression blocks showed slightly higher yield strengths with A356 relative to AZ91 but PCMs’
yield strengths between the two alloys show very similar values.
Chapter 4: Results and Discussion
72
Figure 4-53. The comparisons between summarized experimental A356 PCMs’ yield strengths with lower
bound model: as-cast (left), solutionized (middle), and solutionized and aged (right).
Figure 4-54. The comparisons between summarized experimental AZ91 PCMs’ yield strengths with
lower bound model: as-cast (left), solutionized (middle), and solutionized and aged (right).
An upper bound model was also developed by predicting PCM yield strengths with solid
compressive yield strength results and average strut thickness. The average strut thickness accounted by
redistributing the whole strut volume with radius of curvature into uniform thickness throughout the
struts. Since the strut thickness is uniform, upper bound model predicts stress required yield for that cross
section at any location on the strut. The light grey trend lines in Fig. 4-53 and Fig. 4-54 show the
difference in compressive yield strength results if the strut morphology becomes more efficient at the
same relative density. The curvature of radius in the strut morphology affects the predicted compressive
yield strength greatly. The experimental PCM compressive yield strength would improve greatly if the
curvature of radius could be minimized.
The mechanical properties of PCMs are affected by bulk material’s properties but not in a linear
fashion. And the difference in strut thickness and strut morphology between the measured and actual can
Chapter 4: Results and Discussion
73
widen the gap between results. The compression PCM results can be more consistent if an automated or
more precise process in fabricating truss pattern in foam for LFC. Overall, integrated pyramidal sandwich
design improved the limit in producing thin strut without fillability issue.
Chapter 5: Conclusions
74
5. Conclusions
Magnesium pyramidal lattice sandwich structures were successfully fabricated for the first time
using the lost foam casting process. A significant challenge was filling the thin struts of the pyramidal
lattice. The following conclusions regarding PCM fabrication by LFC can be made:
i. Vacuum casting caused metal penetration through the ceramic coating due to high
surface area of the PCM architecture.
ii. A vertical sprue with side feeding design showed the best compromise between fillability
and pattern collapse.
iii. Al or Si alloying addition can be used to aid fillability of the truss core.
iv. The integrated pyramidal sandwich design eliminated the use of hot melt glue during
sandwich panel assembly and minimized casting defects.
Mechanical testing of the as-cast and heat treated Al and Mg alloy micro-trusses showed that the
compression yield strength of the PCM architectures increased with relative density and artificial aging.
The strength of aluminum alloy PCMs increased beyond the yield strength, while magnesium alloy PCMs
underwent a peak stress due to strut fracture.
From the results of this thesis, the following future work is suggested: improve dimensional
accuracy from alternative technique in foam pattern fabrication; minimize the casting defects in filling
PCM patterns by modifying foam material; develop periodic cellular structures other than pyramidal
lattice with lost foam casting.
75
References
Ashby, M.F. 2005. Hybrids to Fill Holes in Material Property Space. Philosophical Magazine 85: 3235-
3257.
Ashby, M.F., Evans, A., Fleck, N.A., Gibson, L.J., Hutchinson, J.W. and Wadley, H.N.G. 2000. Metal
Foams. Boston: Butterworth Heinemann.
Avedesian, M.M. and Baker. H. ed. 1999. American Society for Metals Specialty Handbook: Magnesium
and Magnesium Alloys. Materials Park: American Society for Metals International.
Bast, J., Aitsuradse, M. and Hahn, T. 2004. Advantages of Low Pressure Lost Foam Casting Process. AFS
Transactions 112: 1131-1144.
Beeley, P. 2001. Foundry Technology. Oxford: Butterworth Heinemann.
Bendick, J. 1995. Archimedes and the Door of Science. Bathgate: Bethlehem Books.
Bichler, L. and Ravindran, C. 2005. Observations on Fillability and Metal Velocity of AZ91E Magnesium
Alloy Cast by the LFC Process. AFS Transactions 113: 857-866.
Bichler, L., Ravindran, C. and Machin, A. 2003. Novel Experiments in LFC of Magnesium Alloys. Light
Metals, 42nd
Annual Conference of Metallurgists of CIM, Vancouver, Canada, 501-513.
Bouwhuis, B.A. and Hibbard, G.D. 2006. Fabrication, Microstructure and Mechanical Properties of
Pyramidal AA3003 – Periodic Cellular Metal. Processing and Fabrication of Advanced Materials.
Cincinnati, OH, 15: 87-101.
Bouwhuis, B.A., Tang S.K. and Hibbard G.D. 2008. Process-Microstructure-Property Relationships in
AA3003 Expanded Metal Periodic Cellular Truss Cores. Composites: Part A 39: 1556-1564.
Brenner, J. 1989. 4th Annual Evaporative Foam Pattern Casting Conference. Rosemont, IL.
Brenner, J., Frederick, P. and Moll, N. 1990. PMMA Molded Foams for Use in the Ferrous EPC Process.
AFS Transactions 95: 667-670.
Chiras, S., Mumm, D.R., Evans, A.G., Wicks, N., Hutchinson, J.W., Dharmasena, K., Wadley, H.N.G.
and Fichter, S. 2002. The Structural Performance of Near-Optimized Truss-Core Panels. International
Journal of Solids and Structures 39: 4093-4115.
Davis, J.R. ed. 1993. American Society for Metals Specialty Handbook: Aluminum and Aluminum Alloys.
Materials Park: American Society for Metals International.
Deshpande, V.S. and Fleck, N.A. 2001. Collapse of Truss Core Beams in 3-Point Bending. International
Journal of Solids and Structures 38: 6275-6305.
Donahue, R. 1990. Futuristic Foam Cylinder Blocks Cast at Mercury Marine – A Design and Process
Marriage, American Foundry Society/ SME Expendable Pattern Process Seminar.
76
dos Santos, D., Vroomen, U. and Buhrig-Polaczek, A. 2007. ADI Lost-Foam: Synergy of Process and
Material. Advanced Engineering Materials 9: 259-264.
Faupel, J.H. and Fisher, F.E. 1981. Engineering Design. New York: John Wiley & Sons.
Foseco International Ltd. 1991. Lost Foam Casting Defect Identification Chart, London, Ontario.
Fujiii, K., Kawabata, T., Matsuda, K. and Ikeno, S. 2007. Changes in the Mechanical Properties and
Microstructures of AZ91 Cast Mg Alloy Caused by heat Treatment. Materials Science Forum 561-
565: 311-314.
Gallois B., Behi, M. and Penchal, J. 1987. Polystyrene in Full Mold Casting of Gray Iron. AFS
Transactions 95: 579-590.
Gordon, L., Bouwhuis, B., Suralvo, M. and Hibbard, G. 2009. Micro-Truss Nanocrystalline Ni Hybrids.
Acta Materialia 57: 932-939.
Goria, C., Del Gaudio, G., Caironi, G., Silva, G. and Selli, M. 1986. Molding of Iron Casting with
Evaporative Polystyrene Foam Patterns. Casting World winter: 41-51.
Hertzberg, R.W. 1996. Deformation and Fracture Mechanics of Engineering Materials. New York: J.
Wiley & Sons.
Hess, D.R. 2004. Comparison of Aluminum Alloys and EPS Foams for Use in the Lost Foam Casting
Process. AFS Transactions 112: 1161-1174.
Hess, D.R., Askeland, D.R. and Ramsay, C.W. 2003. Influence of Bead Chemistry on Metal Velocity and
Defect Formation in Aluminum Lost Foam Castings. AFS Transactions 111: 1265-1278.
House of Metals. 2008. Toronto, Canada.
Kanicki, D.P. 1985. New Technologies Shaping Foundries of the Future, Modern Casting Oct: 29-32.
Keller, F. 1948. Metal Handbook: The Metallography of Aluminum Alloys, Materials Park: American
Society for Metals International.
Kooistra, G.W. and Wadley, H.N.G. 2007. Lattice Truss Structures from Expanded Metal Sheet.
Materials and Design 28: 507-514.
Kooistra, G.W., Deshpande, V.S. and Wadley, H.N.G. 2004. Compressive Behavior of Age Hardenable
Tetrahedral Lattice Truss Structures Made from Aluminum. Acta Materialia 52: 4229-4237.
Korner, C., Hirschmann, M., Brautigam, V. and Singer, R.F. 2004. Endogenous Particle Stabilization
During Magnesium Integral Foam Production. Advanced Engineering Materials 6: 385-390.
Lawerence, M.D., Ramsay, C.W. and Askeland, D.R. 1998. Some Observations and Principles for Gating
of Lost Foam Castings. AFS Transactions 106: 349-356.
Li, Q., Chen, E.Y., Bice, D.R. and Dunand, D.C. 2008. Mechanical Properties of Cast Ti-6Al-4V Lattice
Block Structures. Metallurgical and Materials Transactions A 39A: 441-449.
77
Li, Q., Chen, E.Y., Bice, D.R. and Dunand, D.C. 2008. Mechanical Properties of Cast Ti-6Al-2Sn-4Zr-
2Mo Lattice Block Structures. Advanced Engineering Materials 10: 939-942.
Liu, Z., Hu, J., Wang, Q., Ding, W., Zhu, Y. and Chen, W. 2002. Evaluation of the Effect of Vacuum on
Mold Filling in the Magnesium EPC process. Journal of Materials Processing Technology 120: 94-
100.
Maltais, A., Dube, D., Fiset, M., Laroche, G. and Turgeon, S. 2004. Improvements in the Metallography
of As-cast AZ91 Alloy. Material Characterization 52: 103-119.
Marlatt, M., Weiss, D.J. and Hryn, J.N. 2003. Developments in Lost Foam Casting of Magnesium. AFS
Transactions 111: 1-8.
Mayer, V.A. ed. 2006. Annual Book of ASTM Standard: Section 3: Metals Test Methods and Analytical
Procedures. Materials Park: ASTM International.
Molibog, T.V. and Littleton, H. 2002. Degradation of Expanded Polystyrene Patterns. AFS Transactions
110: 1483-1486.
Monroe, R.W. 1992. Expendable Pattern Casting. Des Plaines: American Foundry Society.
Nayeb-Hashemi, A.A. and Clark, J.B. 1988. Phase Diagrams of Binary Magnesium Alloys. Materials
Park: American Society for Metals International.
Pan, E.N. and Liao, K.Y. 2000. Study on the Filling Behavior of the EPC A356 Alloy Castings. AFS
Transactions 108: 751-760.
Queheillalt, D.T. and Wadley, H.N.G. 2005. Cellular Material Lattices with Hollow Trusses. Acta
Materialia 53: 303-313.
Queheillalt, D.T., Murty, Y. and Wadley, H.N.G. 2008. Mechanical Properties of an Extruded Pyramidal
Lattice Truss Sandwich Structure. Scripta Materialia 58: 76-79.
Rao, P.N. 1999. Manufacturing Technology: Foundry, Forming and Welding. New Delhi: Tata McGraw-
Hill.
Rodgers, R.C. 1988. Robinson Foundry Automates Lost Foam. Foundry Management and Technology
mar: 30.
Rodgers, R.C. 1995. Growth in Evaporative Pattern Casting. Foundry Management and Technology may:
24-28.
Sadayappan, M., Thomson, J.P. and Sahoo, M. 2006. Casting Fluidity of Magnesium Alloy AZ91 in
Gravity and Low Pressure Casting. AFS Transactions 114: 747-754.
Sheng, S.D., Chen, D. and Chen. Z.H. 2009. Effects of Si additions on Microstructure and Mechanical
Properties of RS/PM (Rapid Solidification and Powder Metallurgy) AZ91 alloy. Journal of Alloys
and Compounds 470: 17-20.
Shin, S.R., Han, S.W. and Lee, K.W. 2005. Gas Pore Formation in Lost Foam Casting of AZ91H Mg
Alloy in Comparison with A356 Al alloy. Materials Transactions 46: 2204-2210.
78
Shivkumar, S. 1994. Modeling of Temperature Losses in Liquid Metal During Casting Formation in
Expendable Pattern Casting Process. Materials Science and Technology (UK) 10: 986-992.
Shivkumar, S. and Gallois, B. 1987. Physico-Chemical Aspects of the Full Mold Casting of Aluminum
Alloys, Part I: The Degradation of Polystyrene. AFS Transactions 95: 791-800.
Shivkumar, S., Yao, X. and Makhlouf, M. 1995. Polymer-Melt Interactions During Casting Formation in
the Lost Foam Casting. Scripta Metallurgica et Materialia 33: 39-46.
Sypeck, D.J. and Wadley, H.N.G. 2001. Multifunctional Microtruss Laminates: Textile Synthesis and
Properties. Journal Materials Research 16: 890-897.
Sypeck, D.J. and Wadley, H.N.G. 2002. Cellular Metal Truss Core Sandwich Structures. Advanced
Engineering Materials 4: 759-764.
Tian, J., Lu, T.J., Hodson, H.P., Queheillalt, D.T. and Wadley H.N.G. 2007. Cross Flow Heat Exchange
of Textile Cellular Metal Core Sandwich Panels. International Journal of Heat and Mass Transfer
50: 2521-2536.
Tseng, C.H.E. and Askeland, D.R. 1992. Study of the EPC Mold Filling Process Using Metal Velocity
and Mass and Energy Balances. AFS Transactions 100: 519-528.
Voort, G. F. V. ed. 1999. Metallography, Principles and Practice. Materials Park: American Society for
Metals International.
Wadley, H.N.G. 2006. Multifunctional Periodic Cellular Metals. Philosophical transactions of the Royal
Society of London A 364: 31-68.
Wallach, J.C. and Gibson, L.J. 2001. Mechanical Behavior of a Three-Dimensional Truss Material.
International Journal of Solids and Structures 38: 7181-7196.
Wang, J., Evans, A.G., Dharmasena, K. and Wadley, H.N.G. 2003. On the Performance of Truss Panels
with Kagome Cores. International Journal of Solids and Structures 40: 6981-6988.
Wen, C.E., Yamada, Y., Shimojima, K., Chino, Y., Hosokawa, H. and Mabuchi, M. 2004.
Compressibility of Porous Magnesium Foam: Dependency on Porosity and Pore Size. Materials
Letters 58: 357-360.
Yamada, Y., Shimojima, K., Sakaguchi, Y., Mabuchi, M., Nakamura, M., Asahina, T., Mukai, T.,
Kanahashi, H. and Higashi, K. 1999. Processing of an Open-Cellular AZ91 Magnesium Alloy with a
Low Density of 0.05 g/cm3. Journal of Materials Science Letters (UK) 18: 1477-1480.
Yamada, Y., Shimojima, K., Sakaguchi, Y., Mabuchi, M., Nakamura, M., Asahina, T., Mukai, T.,
Kanahashi, H. and Higashi, K. 2000. Effect of Heat Treatment on Compressive Properties of AZ91
Mg and SG91A Al Foams with Open-Cell Structure. Materials Science and Engineering A280: 225-
228.
Zhang, J. and Ashby, M.F. 1992. The Out-of-Plane Properties of Honeycombs. International Journal of
Mechanical Sciences 34: 475-489.
79
Zhou, J., Shrotriya, P. and Soboyejo, W.O. 2004. On the Deformation of Aluminum Lattice Block
Structures: From Struts to Structures. Mechanics of Materials 36: 723-737.
80
Appendix
A.1. Schematic Diagram
A.2 Phase Diagrams
A.3. Grain Fineness =umber
A.4. Compression Test Results
A.5. Previously Studied PCMs and Metal Foams
Appendix
81
A.1 Schematic Diagrams
Figure A-1. Schematic diagram of corrugation sandwich core with dimensions.
Total Area
θθθθθ sintan
2
tantansin
ccccccc tdhdhdtBase +
−=
−+−= and chHeight =
+
−=
θθ sintan
2 cccc
tdhhAreaTotal
Triangular Open Space
θtan
cc dhBase
−= and cc dhHeight −=
( )θtan
2
cc dhAreaTriangular
−=
Relative Density
( )
+
−
−
−=
θθ
θρ
sintan
2
tan1~
2
cccc
cc
tdhh
dh
Appendix
82
A.2. Phase Diagrams
Figure A-2. Aluminum-silicon phase diagram (left) and magnesium-silicon phase diagram (right) [Nayeb-
Hashemi and Clark, 1988].
Figure A-3. Aluminum-magnesium phase diagram [Nayeb-Hashemi and Clark, 1988].
Appendix
83
A.3. Grain Fineness =umber
The refractory sand particle size can affect the LFC decomposition product permeability. The
average refractory sand size was determined using the American Foundry Society’s grain fineness
number (AFS GFN). From a representative sample, different grit-size mesh sieves were used to separate
different sand particles by size after shaking for 15 minutes. The retained sand weight from each sieve
was used to determine the grain distribution. The final AFS GFN number was calculated using Table 10
and the following formula:
∑∑=
i
ii
f
fMGF' (Eqn. A-1)
Where Mi = Multiplying factor and fi = weight fraction of sand left on sieve.
Table A-1. AFS sieve number and the multiplying factor Mi [after Rao, P.N., 1999].
US Series equivalent
No. (ASTM)
Mesh
opening
(mm)
IS Sieve no.
(µm)
Multiplying
factor
6 3.327 3.35 3
12 1.651 1.70 5
20 0.833 850 10
30 0.589 600 20
40 0.414 425 30
50 0.295 300 40
70 0.208 212 50
100 0.147 150 70
140 0.104 106 100
200 0.074 75 145
270 0.053 53 200
Pan - - 300
Appendix
84
A.4. Compression Test Results
Table A-2. Experimental measured yield strength (σ yield PCM) and peak strength (σ peak PCM) of A356 and AZ91 PCMs in compressions.
Relative
Density
ρρρρ (%)
σσσσ yield PCM σσσσ peak PCM
A356 AZ91 AZ91
As-cast Solutionized Aged As-cast Solutionized Aged As-cast Solutionized Aged
10.5
4.0 2.6 4.5 5.2 5.0 5.2 12.0 12.3 9.4
5.4 5.1 6.4 5.7 4.3 3.4 11.2 9.4 5.5
4.8 4.9 5.1 7.7 5.5 8.1 17.8 13.1 14.1
4.0 6.4 5.2 5.9 12.2 8.0
Average 4.7 4.1 5.6 6.2 5.0 5.7 13.7 11.7 9.3
St. Dev 0.7 1.1 1.0 1.3 0.5 1.9 3.6 1.6 3.6
12.3
6.5 5.4 6.7 8.2 6.7 8.0 16.7 16.3 14.1
6.9 5.8 15.6 8.5 5.8 11.4 20.1 15.0 23.9
7.2 6.3 11.3 7.5 6.4 9.5 16.1 15.5 17.7
5.6 8.1 7.2 7.1 15.3 13.8
Average 6.9 5.8 10.4 8.1 6.5 9.0 17.6 15.5 17.4
St. Dev 0.4 0.4 3.9 0.5 0.6 1.9 2.1 0.5 4.7
14.0
6.2 4.7 12.3 9.1 7.5 9.4 23.9 22.6 19.6
8.0 6.2 8.8 8.6 6.5 10.4 19.9 20.9 19.9
6.2 6.3 8.1 8.9 8.1 10.4 25.0 19.6 24.6
6.3 10.2 5.7 11.3 16.0 17.2
Average 6.8 5.9 9.8 8.9 6.9 10.4 23.0 19.8 20.3
St. Dev 1.0 0.8 1.8 0.3 1.1 0.8 2.7 2.8 3.1
15.7
13.6 9.0 12.9 10.9 9.2 10.0 29.3 23.2 27.0
10.7 8.6 13.6 10.0 8.0 11.2 26.1 23.7 20.6
8.8 7.2 10.4 10.6 9.5 11.0 25.1 24.9 23.3
5.7 14.4 7.2 10.0 24.9 16.6
Average 11.0 7.6 12.9 10.5 8.5 10.5 26.8 24.2 21.8
St. Dev 2.4 1.5 1.7 0.5 1.1 0.7 2.2 0.8 4.4
Appendix
85
Table A-3. Experimental measured yield strength (σ yield) and peak strength (σ peak) of A356 and AZ91 square blocks in compressions.
σσσσ yield σσσσ peak
A356 AZ91 AZ91
As-cast Solutionized Aged As-cast Solutionized Aged As-cast Solutionized Aged
109 128 193 95 91 128 307 291 306
113 136 211 97 105 131 316 293 306
119 142 220 92 110 138 316 299 309
Average 114 135 215 95 108 135 313 296 308
St. Dev 4.9 7.1 6.6 2.6 3.4 5.3 5.0 4.2 1.9
Appendix
86
A.5. Previous studied PCMs and Metal Foams
Table A-4. Previous studies of cast PCMs, summarizing the alloy, architecture, bulk metal density (ρs), relative density ( ρ~ ), PCM’s density
(σ peak PCM), PCM’s specific peak strength (σ peak PCM / ρ PCM), PCM’s yield strength (σ yield PCM) and PCM’s specific yield strength (σ yield PCM / ρ PCM).
Authors Alloy Architecture
ρρρρs ρ~ ρρρρ PCM σσσσ peak PCM σσσσ peak PCM
/ ρρρρ PCM σσσσ yield PCM
σσσσ yield PCM
/ ρρρρ PCM
g/cm3 % g/cm
3 MPa
MPa /
(g/cm3)
MPa MPa /
(g/cm3)
Deshpande and Fleck, 2001 Cu-4Si-1.4Zn Tetrahedral 8.8 8 0.70 10.5 14.9 4.2 6.0
Chiras et al., 2002 Cu-2Be Tetrahedral 8.26 2.25 0.19 4.8 25.8 4.0 21.5
Wang et al., 2003 Cu-2Be Kagome 8.26 2.25 0.19 4.8 25.8 3.8 20.4
Li et al., 2008
Ti-6Al-4V
Pyramidal
4.46 5.4 0.24 22.0 91.3 19.0 78.9
4.46 5.5 0.25 31.0 126.4 29.0 118.2
Ti-6Al-2Sn-
4Zr-2Mo 4.54 5.5 0.25 41.0 164.2 35.0 140.2
Appendix
87
Table A-5. Previous studies of aluminum PCMs, summarizing the alloy, architecture, bulk metal density (ρs), relative density ( ρ~ ), PCM’s density
(σ peak PCM), PCM’s specific peak strength (σ peak PCM / ρ PCM), PCM’s yield strength (σ yield PCM) and PCM’s specific yield strength (σ yield PCM / ρ PCM).
Authors Alloy Architecture
ρρρρs ρ~ ρρρρ PCM σσσσ peak PCM σσσσ peak PCM
/ ρρρρ PCM σσσσ yield PCM
σσσσ yield PCM
/ ρρρρ PCM
g/cm3 % g/cm
3 MPa
MPa /
(g/cm3)
MPa MPa /
(g/cm3)
Deshpande and
Fleck, 2001 Al-7Si-0.3Mg Tetrahedral 2.68 8 0.21 7.3 34.0 4.2 19.6
Wallach and
Gibson, 2001
(443)
Al-4Si-0.2Fe Pyramidal 2.69 6.2 0.17 9.1 53.5 N/A N/A
Zhou et al., 2004
(516.1) Al-3Mg-1Si-1Fe-
1Mn
Pyramidal
2.65 5.3 0.14 5.9 42.0 2.9 20.6
(518.0) Al-8Mg-1.8Fe-
0.35Si 2.57 5.3 0.14 5.6 41.1 3.8 27.9
(A356) Al-7Si-0.4Mg 2.68 5.3 0.14 5.8 40.8 3.5 24.6
Kooistra et al.,
2004 AA6061 Tetrahedral 2.7
8.3 0.22 20.7 92.5 16.2 72.5
5.5 0.15 11.4 76.8 9.3 62.4
3.7 0.10 6.1 61.4 5.7 57.4
3 0.08 4.0 49.6 3.0 37.5
2 0.05 2.8 52.2 2.4 44.3
8.3 0.22
N/A N/A
5.2 23.2
5.5 0.15 2.4 16.2
3.7 0.10 1.1 10.8
3 0.08 1.0 11.7
2 0.05 0.8 14.4
Queheillalt et al.,
2008 AA6061 Pyramidal 2.7 6.2 0.17 11.1 66.1 10.1 60.2
Appendix
88
Table A-6. Previous studies of magnesium metal foams, summarizing the alloy, architecture, bulk metal density (ρs), relative density ( ρ~ ), PCM’s
density (σ peak PCM), PCM’s specific peak strength (σ peak PCM / ρ PCM), PCM’s yield strength (σ yield PCM) and PCM’s specific yield strength (σ yield PCM / ρ PCM).
Authors Alloy Architecture
ρρρρs ρ~ ρρρρ PCM σσσσ peak PCM σσσσ peak PCM /
ρρρρ PCM σσσσ yield PCM
σσσσ yield PCM /
ρρρρ PCM
g/cm3 % g/cm
3 MPa
MPa /
(g/cm3)
MPa MPa /
(g/cm3)
Yamada et al.,
1999
AZ91 Foam
1.81
3.0 0.053 0.132 2.480 0.113 2.111
3.1 0.055 0.135 2.428 0.118 2.127
3.0 0.054 0.105 1.962 0.083 1.540
3.0 0.053 0.083 1.545 0.063 1.186
2.8 0.051 0.069 1.354 0.058 1.140
Yamada et al.,
2000
3.7 0.066 0.143 2.153 0.136 2.053
3.4 0.062 0.124 2.018 0.114 1.851
3.4 0.062 0.106 1.714 0.102 1.649
3.4 0.062 0.112 1.817 0.105 1.703
3.5 0.063 0.081 1.292 0.074 1.175
Korner et al., 2004 65.7 1.19 230.6 193.78 19.8 16.6
56.9 1.03 238.9 231.9 14.0 13.6
Wen et al., 2004 Pure Mg 1.738 65 1.13 12.0 10.6 12.0 10.6
45 0.79 12.0 15.3 10.0 12.8