Optimization of Gating System Design for

Die Casting of Thin Magnesium Alloy-

Based Multi-Cavity LCD Housings

B. D. Lee, U. H. Baek & J. W. Han

Journal of Materials Engineering and

Performance

ISSN 1059-9495

J. of Materi Eng and Perform

DOI 10.1007/s11665-011-0111-1

1 23
Your article is protected by copyright and
all rights are held exclusively by ASM
International. This e-offprint is for personal
use only and shall not be self-archived in
electronic repositories. If you wish to self-
archive your work, please use the accepted
author’s version for posting to your own
website or your institution’s repository. You
may further deposit the accepted author’s
version on a funder’s repository at a funder’s
request, provided it is not made publicly
available until 12 months after publication.

1 23
 Author's personal copy
JMEPEG ÓASM International
DOI: 10.1007/s11665-011-0111-1 1059-9495/$19.00

Optimization of Gating System Design for Die Casting

of Thin Magnesium Alloy-Based Multi-Cavity
LCD Housings
B.D. Lee, U.H. Baek, and J.W. Han

(Submitted April 18, 2011; in revised form November 30, 2011)

High-pressure die casting is the preferred process for manufacturing Mg-alloy components used for
numerous applications. High-pressure die casting is suitable for mass production and has the advantage of
also being suitable for accurately fashioning objects of complicated shapes. One disadvantage of high-speed
die casting is the occurrence of defects such as shrinkage or air entrainment. Gating system design must be
very effective in actual manufacturing facilities to avoid the occurrence of such defects. The objective of this
study is to present a methodology for obtaining optimal designs of 4-cavity thin electronic component
housings. The fluid behavior and amount of air entrainment caused by the overflows and air vent designs
were analyzed using a computer fluid dynamics (CFD) simulator. The effectiveness of the proposed system
was demonstrated through CFD simulations and experiments using an actual manufacturing process. Also,
the effect of vacuum systems on the porosity and mechanical properties of the castings was studied. The
volume of porosity in the casting was found to be significantly reduced using vacuum assistance during die
casting. As a result, the tensile strength and the elongation of the die casting products are improved.

quality of the die cast products. Irrespective of the shape of the

Keywords AZ91D, die casting, mechanical properties, micro-
structure, numerical simulation castings, the quality is always determined by the gating system.
In general, the system contains the arrangements and shapes of
gates, runners, sprues, overflows, and air vents. Herman (Ref 9)
discussed the effects of the molten fluid velocity and die
temperature on the molten fluid distance and quality. In the
1. Introduction process, molten fluid is injected into the die cavity, and flows
with a high velocity, commonly over 10 m/s, through gates.
Brevick (Ref 10) discovered that this turbulent flow usually
Kim et al. (Ref 1) and Shin et al. (Ref 2) discovered that
causes the entrapment of gas in the die cavity, and consequently
extended magnesium alloys have the lowest density among
induces porosities, which is a frequent reason for casting
current common alloys and can be instrumental in lowering the
rejection. Gouraly et al. (Ref 11, 12) have greatly contributed to
weight of automobiles and airplanes. Such magnesium alloys
study on the porosity in die castings. However, the quantitative
have been limited in their use due to their low ductility and
effect of porosity on mechanical properties of die castings
corrosion resistance compared to other light materials. But with
should be studied further. In addition, the die casting paper (Ref
the advancement of alloy and refining technologies, the limita-
13-16) proposed some methods to predict gas entrapment in the
tions that were described by the study reported (Ref 3-5) and
mold-filling process. In spite of the study, it is well known that
Chang et al. (Ref 6) have been overcome. Recently, researchers
the die casting process is quite complicated. However, the die
(Ref 7, 8) examined the automobile industry and discovered a
designer must determine the gating system used according to his
great interest in the applications of magnesium alloy.
own experience. This trial and error method usually results in
Magnesium alloys with HCP structures are usually produced
longer lead times and increased costs for die casters.
by casting processes for almost all their applications. Die casting
Xiong et al. (Ref 17) used numerical methods to analyze the
is a precise casting method in which molten fluid is injected at
die casting process from a thermal perspective to shorten the
high pressure into a die cavity. As soon as the molten fluid has
costly and time-consuming lead times that are associated with
filled the cavity, it solidifies by fast cooling. Die casting is
process design, tool fabrication, modification, and the trial and
widely used in the automobile, aerospace, electronics, and
error procedure involved in finding acceptable operating
household appliance industries due to its results of high strength
conditions. Some recent researchers studied 2-cavity molds
and good performance. The quality of the die cast is essentially
and thin component housings at a thickness of 1 mm. But no
determined by the die in which the metal is cast. The design of
attempts have made to study molds and component housings at
the gating system is very critical to the die because it will affect
a thickness of <1 mm. Also, vacuum systems have been
the subsequent design procedures and influence the overall
developed for heavy housings, such as seat frames and steering
wheel cores, using a computer fluid dynamics (CFD) simulator.
B.D. Lee, U.H. Baek, and J.W. Han, Division of Material Science and
Engineering, Inha University, Incheon 402-751, Republic of Korea.
The main purpose of this article is to design the runner of the
Contact e-mails: fl[email protected] and [email protected]. mold with a developed thickness of 0.3 mm and with a 4-cavity

Journal of Materials Engineering and Performance

 Author's personal copy
mold using a CFD simulator. Vacuum systems were analyzed casting. In die design, runners and overflows are the important
for a light housing with a 4-cavity mold and a thickness of aspects for reducing the amount of trapped air. The runner and
0.3 mm. Several experiments with AZ91D die castings were gate design was studied by Hirt and Sicilian (Ref 19). The
carried out. The experimental results were compared with the largest percentage area reduction should occur at the gate,
simulated results. The porosity and mechanical properties of the where an area reduction ranging from 10 to 40% is typical, as
specimens from the die castings were measured, and those studied by Yoshimura et al. (Ref 20).
measurements agreed well with the simulations. Also, the effect Finally, all the effort listed above is aimed at trying to force
of porosity on the mechanical properties was analyzed. the air out of the shot sleeve ahead of the plunger, and this
approach depends on having a path for the air to go. This means
there must be air vents and a vacuum system. The location of
2. Numerical Models the air vents is absolutely critical if the porosity is major, so the
air vents should take precedence over other factors. Above all,
because the object of this study is obtaining casting conditions
The CFD code FLOW-3D-based on a finite volume/finite
of LCD housing with a 0.3 mm thickness, if the casting fails
difference approach was used in the simulation. Two method-
with the design then the casting remains in the cavity due to the
ologies, fractional area/volume obstacle representation
quick solidification rates. Then, failure will deteriorate the
(FAVOR) and volume-of-fluid (VOF), constitute the core of
quality of the casting products. Figure 1 shows additional
the software. These methods differ from methods in most other
overflow and air vents in the original model. Here, the aim of
codes but offer many advantages, and are summarized in
these numerical simulations was to examine the predictive
Ref 18 by Barkhudarov and Hirt.
capability of the model with different overflow designs and
additional air vents. Different overflow designs and additional
2.1 Optimization Model Description and Simulation
air vents have different ways of controlling the flow of molten
Condition
fluid during filling. For the prevention of trapped air, a vacuum
When using a parametric design in gating system design, the system shown in Fig. 1(d) is considered.
designer does not need to create the geometries of the gating Jin and Kang (Ref 21) studied the location of the vacuum
system from scratch in each die design process. The main channel connections and discovered that they should be at the
approach for developing a parametric design system is to build last points to fill, and this is best defined by flow simulation.
a gating model database and to construct adequate parametric The number of cells in the mold/cavity domain is 2,981,826,
models of gating elements using a 3D CAD tool before the with the smallest cell size being 0.2 mm from the gate to the
gating system design is put into practice. In this case, a software product. Tables 1 and 2 show the physical properties and initial
package of the casting process simulation can obtain solid conditions for the simulation. SKD61 die steel was employed
modeling information with a certain file format interface, such as a mold material at an initial temperature of 200 °C. AZ91D
as STL. FLOW-3D software obtains solid modeling informa- LCD housings were produced on a hot-chamber die casting
tion by importing STL files from commercial 3D modeling machine. The solid and molten temperatures of AZ91D were
software. Figure 1 shows the geometry and mesh setup of the set at 470 and 590 °C, respectively. The cross section of the

Fig. 1 Diagram of the LCD housing: (a) original model, (b) 1st corrected model, (c) 2nd corrected model, and (d) vacuum system model

Journal of Materials Engineering and Performance

 Author's personal copy
Table 1 Thermo-physical properties of the casting and mold
Materials Property Symbol Value Unit

Casting (AZ91D) Thermal conductivity of molten fluid jl 72 W/(m Æ k)

Thermal conductivity of solid js 254 W/(m Æ k)
Specific heat of molten fluid Cl 1050 J/(kg Æ k)
Specific heat of solid Cs 1034 J/(kg Æ k)
Density of molten fluid ql 1.81 kg/m3
Density of solid qs 1.79 kg/m3
Liquidus temperature Tl 590 °C
Solidus temperature Ts 470 °C
Mold (SKD61) Thermal conductivity jf 25 W/(m Æ k)
Specific heat C 0.46 J/(kg Æ k)

Table 2 Initial conditions for the simulation (boundary conditions)

Viscous Numerical Pouring Mold
Flow mode fluid approximations temperature temperature

Incompressible Renormalized Explicit 630 °C 200 °C

group model

Plunger velocity, Plunger velocity, Thickness Volume

1st stage 2nd stage Tip diameter (products) (products)

0.14 m/s 0.83 m/s 74 U 0.3 mm 3.092E+03 m3

gate was 58.82 mm2. Finally, the initial injection speeds of the @T 1
¼ ðrAc Þ  ðjm rTm Þ ðEq 5Þ
plunger were 0.14 and 0.83 m/s. @t qCm Vc

2.2 Geometry Representation where the subscript m indicates a parameter related to the
mold and the c indicates quantities that are complements of
An advancement of the conventional finite difference method the volume and area fractions. At the metal/mold interface,
is given by adopting the FAVOR method. In this method, the heat flux, q, is calculated according to
rectangular grid cells can be partially blocked by obstacles
(Ref 19). This improves the accuracy of the numerical solution q ¼ hðT  Tm Þ ðEq 6Þ
near mold walls and allows for the use of coarser grids than in where h is the heat transfer coefficient.
standard finite difference methods studied by Sequeira et al.
(Ref 22). Since the geometry representation is less mesh-dependent, 2.3 Evaluation of Air Entrainment
the FAVOR is also referred to as a free gridding method.
For an incompressible, viscous fluid, the FAVOR equations In die casting, trapped air is usually the biggest problem
take the form: contributing to porosity because of the very turbulent flow and
the fast fill rate that is necessary in the process. If this problem
r  ðAuÞ ¼ 0 ðEq 1Þ
could be eliminated, die casting would probably become the
@u 1 1 1 prominent casting method, especially for smaller castings. In this
þ ðAu  rÞu ¼  rp þ ðrAÞ  ðlrÞu þ g ðEq 2Þ article, the trapped air amount is estimated using a CFD
@t V q qV
simulator. Air entrainment at the molten fluid surface is based
@H 1 1 on the concept that turbulent eddies raise small molten fluid
þ ðAu  rÞH ¼ ðrAÞ  ðj  T Þ ðEq 3Þ elements above the free surface that may trap air and carry it back
@t V qV
into the molten fluid. Sequeira et al. (Ref 23) found that the
where molten fluid elements that can be lifted above the free surface
Z depend on whether or not the intensity of the turbulence is enough
H ¼ CðT Þ@T þ Lð1  fs Þ ðEq 4Þ to overcome the surface-stabilizing forces of gravity and surface
tension. Turbulence transport models characterize turbulence by
In these equations, parameters of A and u are vectors. There- a specific turbulent kinetic energy Q and a dissipation function D.
fore, A is the open area fraction associated with the flow in The characteristic size of turbulence eddies is then given by
the ith direction, V is the open volume fraction, q is the den- (7).
sity, p is the pressure, u is the ith velocity component, l is pffiffiffiffiffiffi
Q3
the fluid viscosity coefficient, g is gravity, H is the fluid L ¼ 0:1 ðEq 7Þ
D
enthalpy, T and Tm are fluid temperature and mold temperature,
respectively (°C), ƒs is the solid fraction, L is the latent heat, This scale is used to characterize surface disturbances. The
and C and j are the fluid-specific heat and thermal conductiv- disturbance kinetic energy per unit volume, i.e., pressured (Pd)
ity coefficient, respectively. For the mold, the energy equation associated with a fluid element raised to a height L and with a
has the form surface tension energy based on a curvature of L is given by (8).

Journal of Materials Engineering and Performance

 Author's personal copy
Testing System equipped with Bluehill software and a 50 kN
Pd ¼ qgL þ r=L ðEq 8Þ load cell. All the specimens were cut into central rectangular
where q is the molten fluid density (kg/m3), r is the coeffi- plates from cast products. Namely, tensile specimens were cut
cient of surface tension, and g (m/s2) is the component of into perpendicular to the gas vent direction and the number of
gravity normal to the free surface. In addition, because the specimens was cut three from same locations. The tensile
potential energy is the dominant parameter compared to the samples were tested without heat treatment. All tests were
surface tension in Pd, (J), it is conceivable that the influences performed at ambient temperature (25 °C) and conducted at a
of surface tension are small. For air entrainment to occur, the strain rate of 103. Dargusch et al. (24) examined the AZ91D
turbulent kinetic energy per unit volume, Pt = qQ, must be die casting sample by XRD/SEM and identified it to have an
larger than Pd. Said in another way, it means that the turbu- a + b dual-phase microstructure. Specimens for microstructural
lent disturbances must be large enough to overcome the sur- characterization were cut from the middle section of the die
face stabilizing forces. The volume of air entrained per unit casting samples. The microstructure of the HPDC and vacuum
time, Va, is given as (9). system samples was examined by optical microscopy (OM)
with quantitative metallography. The specimens for OM were
@Va prepared by the standard technique of grinding with SiC
þ urVa ¼ Rð1  Va Þ ðEq 9Þ
@t abrasive paper and polishing, followed by etching in an
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi aqueous solution of 60 vol.% ethylene glycol, 20 vol.% acetic
where R ¼ Cair 2 ðPl  Pd Þ=q, u is fluid velocity (m/s), t is
time (s), and Cair is a coefficient of proportionality. In this acid, and 1 vol.% concentrated HNO3. For the determination of
study, Cair = 0.5, which means we assume on average that air the quality of the die casting samples, two distinct criteria were
will be trapped over about half the surface area. If Pl is less used. In the first one, the visual inspection of the amount of
than Pd, then Va is zero. porosity was analyzed by computed tomography (CT). CT is a
The air entrainment is expressed in (10). non-destructive evaluation technique for visualizing and
inspecting the internal structure of objects.
X
n
After completing the CT data acquisition and reconstruction
A¼ Vak Ffk Vfk Vck ðEq 10Þ process, we have a 3D volumetric density distribution of the
k¼1
object. Using visualization tools we can view the object from
where A is the quantity of air entrainment (in m3), Va is the vol- different perspectives. We can rotate the object, zoom into an
ume of air entrained per unit time (in m3), Ff is the fluid frac- area of interest, slice the object at any orientation (arbitrary
tion, Vf is the volume fraction, Vc is the volume of the mesh sectioning), and measure density, distance, area, and volume.
cell (in m3), and n is the aggregate number of mesh cells. In The data acquisition process also provides a complete set of
this study, the calculation of air entrainment is used as the mea- digital radioscopic images that can be used for further analysis.
sure of an evaluation studied by Dargusch et al. (24). The second criterion that we used was a density determination
of the machined part using Archimedes principle.

3. Experimental Methods
4. Results and Discussion
The materials used in this study were 200 tons of AZ91D
alloys cast by a high pressure die casting (HPDC) process. The 4.1 Numerical Simulation
composition of AZ91D is shown in Table 3. When molten fluid is injected under high pressure into the
The molten fluid and solid temperatures of magnesium were die cavity, the entrapment of gas by these flows during metal
about 578 and 430 °C, respectively. The vacuum system we used filling is one of the primary causes of porosity. Unless we
in the study consisted of a vacuum pump, a vacuum tank, and a design an appropriate molten fluid velocity and die temperature,
vacuum valve. The 1st step for the die casting process was the the gases evolved during mold filling may float through the
molten fluid moving to the injection chamber. Second, the casting and leave oxide trails. These defects are difficult to
plunger moved past the pouring hole and sealed off the die control and may not be found until castings are cleaned,
cavity. The vacuum valve was then activated and lower-than- machined, and inspected. However, the simulation of the filling
atmospheric pressure was created in the die cavity. The cavity process takes a virtual look inside the die, sees exactly how
was continuously evacuated from the beginning of the die filling defects are being generated, and gives us information about the
process to the end. In the HPDC process and vacuum system, the possible problems.
die temperature was controlled at 200 °C. The pouring temper- In the study, for reducing the trapped air in the die casting
ature was controlled at 630 °C. A variety of plunger speeds process, it is important to observe the fluid flow of magnesium
ranging from 0.14 to 0.83 m/s were used during the casting. melt. The first study involved the prediction of the filling
The tensile tests were conducted following the KS B0801 behavior of an LCD 4-cavity housing. It was about
sub-size using an Instron 5569 Universal Electromechanical 300 9 300 mm in area and had a section thickness of
0.3 mm. The filling parameters used in the simulation were:
pouring temperature, 630 °C; preheated temperature of the
Table 3 The chemical compositions of the AZ91D mold, 200 °C; 1st plunger speed in the ingate, 0.14 m/s; 2nd
Mg alloy plunger speed in the ingate, 0.83 m/s. The process parameters
used in Fig. 2(a) are the same as those used in Fig. 2(b) to (d).
Al Zn Mn Fe Si Cu Ni Mg
Figure 2 shows the mold-filling simulation results of the HPDC
AZ91D magnesium alloy in the cavity of the LCD housing.
AZ91D 8.9 0.69 0.24 0.001 0.02 0.002 0.001 Bal.
Molten fluid enters the die cavity through four gates attached to

Journal of Materials Engineering and Performance

 Author's personal copy

Fig. 2 The filling sequences of numerical simulation for: (a) original model, (b) 1st corrected model, (c) 2nd corrected model, and (d) vacuum
system model

the Y-shaped runner. It can be seen from Fig. 2 that the filling narrow and comparable to the overflow thickness, the discharge
process is stable in all filling stages. The molten fluid flows of molten fluid through the air vents on both sides occurs late in
through the runners at first and fills the gates preferably. the filling.
Finally, the cavity of the LCD housing was filled by molten It can be seen from Fig. 2(a) that molten fluid is still stable
fluid, as shown in Fig. 2(a). But the filling velocity in the cavity and the filling character is analogous to that in Fig. 2(b) to (d).
was slower than the velocity in the center because of the fast But the molten fluid has completely filled the cavity and the
cooling and great resistance in both sides of the cavity. The entire temperature distribution is different. We predicted the
filling velocity in the 2nd and 3rd cavities was 0.075 m/s faster effect of overflows and an additional air venting system. This is
than the velocity in the center. In this case, we predicted that the thought to result from the comparability of the filling rate, as
flow of unstable molten fluid will increase the amount of shown in Fig. 3. When the filling volume reaches about 90%
trapped air in the 2nd and 3rd cavities. The transition from (see Fig. 3 at 9.03E03 s), the fluid of the original model
yellow to green shows where the incoming fluid decelerates as began to decelerate sharply as it approached the overflow on the
it encounters the back filling fluid which moves more slowly. top of the cavity. Though the fluid of the corrected model is
The high temperature follow-up fluid remelted the leading low somewhat stable, the fluid at 9.03E03 s was delayed for a bit.
temperature fluid. The molten fluid substantially filled the However, the fluids of the 2nd corrected model and vacuum
cavity above about 80% at a rate of 0.008 m/s. system model resulted in a stable filling rate based on the
This is thought to result from the different molten fluid numerical simulation. In the study, the simulation results in
velocity, where the 1st plunger speed turned into the 2nd Fig. 2 and 3 are in good agreement. The simulation results in
plunger speed. Some molten fluid was discharged from the Fig. 2 indicate that it is important to design overflows and air
overflow at the top of the cavity. Because the air vents are very vents. Even though the system change with respect to overflow

Journal of Materials Engineering and Performance

 Author's personal copy

Fig. 2 Continued

and air vents affected the product of the temperature distribu-

tion, the model did not appear to exhibit any difference in the
temperature distribution.
We designed a model based on the porosity interpretation to
study the vacuum system in detail.
The molten fluid simulations considering air entrainment
were performed, and the influence of the overflow, air vent, and
vacuum systems was investigated. The calculated results for the
amount of air entrainment are contained in Fig. 4. The
entrained air was greatly reduced. After complete filling, the
total amount of air entrainment for the simulation results was
0.952, 0.892, 0.602, and 0.534 for the cases in Fig. 4(a), (b),
(c), and (d), respectively. The amount of air at the red circle had
simulated values of 0.32, 0.021, 0.018, and 0.004, respectively.
The 2nd corrected model had 58% lower values compared to
the original. Moreover, the vacuum system model showed
enhanced results compared to the 2nd corrected model by 13%. Fig. 3 Filling rate of numerical simulation for: (a) original model,
As a result, the result using the vacuum system model was 78% (b) 1st corrected model, (c) 2nd corrected model, and (d) vacuum
lower than that of the original design. Though the vacuum system model

Journal of Materials Engineering and Performance

 Author's personal copy

Fig. 4 Volume fraction analysis of air for: (a) original model, (b) 1st corrected model, (c) 2nd corrected model, and (d) vacuum system model

system of the 0.3 mm thin product did not affect the Figure 6 shows the result from using Archimedess princi-
temperature distribution, it was found to be very efficient for ple. The porosity ratio of the 2nd corrected model decreased by
removing porosity in the cavity with a vacuum system. Thus, 40% compared to the original model. It is similar to the result
by optimizing the design of the overflow with an air vent with of the simulation.
vacuum systems, a decrease in the amount of air was achieved. Figure 7 compares the simulation and the experimental
As a result, we could conclude that the original process was results with respect to the decreased ratio of porosity. Even
unsuitable for this casting, while the corrected model process though the result of the experiment data was only a little
(1st corrected model and 2nd corrected model) will lead to different compared to the simulation, it has a tendency to have a
better results. lower porosity in the cavity with the vacuum system.
A tensile test to analyze the effect of porosity on the
mechanical properties in the cavity was performed. Figure 6(c)
4.2 Experimental Validation
shows the results of the mechanical test in the LCD housing.
There are several physical processes influencing volume The tensile strength is 291, 309, 326, and 335 MPa for the
shrinkage during solidification. Because molten fluid is nearly respective models in the figure. Among the four investigated
incompressible, any loss in volume caused by shrinkage must models, the vacuum system model exhibits the most significant
be replaced by new molten fluid or by the opening of a void improvement in tensile strength and elongation by vacuum
somewhere in the casting. After molten fluid has cooled enough assistance. The reason being is that porosity affects the fracture
for its solid fraction to exceed the point of rigidity, there can be of specimens. The mechanical properties of an alloy depend on
no (or very little) additional molten fluid flow to compensate for the defects that may be present in the matrix. These defects
the shrinkage. Figure 5 shows the location of samples taken could be point, line, surface, or volume defects. Among these
and CT results used for analyzing internal defects. There were defects, volume defects (porosity, secondary phases, or inclu-
more micropores and the pores were somewhat larger in the sions) are known to be the most significant ones and may
center of the samples than at the sample surfaces. This is in dramatically affect the mechanical properties. The size and
accordance with the porosity evaluation in the literature volume of these defects are critical since they act as the
published by ASM (25). But, the internal defects in the study initiation points for porosity and also as stress risers, as
showed a combination of shrink and air porosity at the examined by Nave et al. (26).
overflow, as in Fig. 5. It can be seen from Fig. 5(a) that the It is obvious that there is a lot of porosity in the low tensile
internal defects are analogous to those in Fig. 5(b) to and (d). strength model. These porosities induce stress concentration
The reason being was that it was difficult for the CT method to and crack initialization, which result in quite poor tensile
analyze the internal defects with sizes below 60 lm. The LCD strength and the elongation of samples. It is well known that in
housing has a small defect size because it solidified quickly and the presence of defects or stress risers the components may
the casting has a thickness of <1 mm. For a better analysis of fracture at stresses far away from their nominal theoretical
the 0.3 mm thick components, an additional method, such as limits. Figure 6(b) and (c) is perfect examples of such
Archimedess principle, needs to be considered. phenomena. Also, the yield strength of Mg alloys fabricated

Journal of Materials Engineering and Performance

 Author's personal copy

Fig. 5 Location of samples taken and CT analysis results of the distribution of internal defects in the LCD housing: (a) original model, (b) 1st
corrected model, (c) 2nd corrected model, and (d) vacuum system model

by the die casting method was influenced by other factors, such 8.20 lm for the respective models given in the figure. The
as the mean size of grains and precipitates of the Mg17Al12 grain size from the 1st and 2nd corrected models is more
phase. The reason, as given in Fig. 8(a) and (c), illustrated that refined than that of the original and vacuum system models.
grain size influenced the yield strength, and the 2nd corrected This result was slightly different from the die casting process.
model had maximum yield strength of about 259 MPa among In the study, the grain size is not important because we focused
all specimens. on the entrapped air.
Figure 7(a) shows the microstructural change in the LCD Cracks are easily formed in the porosity of conventional die
housing as a function of the gating system. The grain size at the castings, because of the high stress concentration due to the
gating systems and vacuum systems is 8.13, 7.16, 7.92, and reduced effective area.

Journal of Materials Engineering and Performance

 Author's personal copy
took place due to surface defects, such as the inner porosity and
heterogeneous solidification contraction. Finally, the prece-
dence of injecting molten fluid with air entrainment and oxide
concentration improves the flow fluid with a vacuum system.
Therefore, the improved flow fluid reduces surface and inner
defects because molten fluid quickly flows out of the overflows
and air vents before solidification. Also, the fast solidification
affects the efficiency improvement of the process. Figure 9
shows the results of the produced surface of the LCD housing
by the adaptive vacuum system.
The trial results indicate that the LCD housing of the
vacuum system has a very good surface finish throughout the
entire casting.

5. Conclusions

In this study, we investigated the effect of different casting

designs and vacuum systems while casting 0.3 mm thick LCD
housings. The main conclusions of this study are summarized
below:

(1) The simulation results of cavity filling and air entrap-

ment confirmed the validity of the simulation analysis.
(2) The simulations that considered air entrainment were
performed and the influence of the gating system and
vacuum system were investigated.
(3) We compared the gating system with a filling behavior
and the porosity rate analyses in the cavity and optimum
process conditions adapted for the die casting. The re-
sults of the model show good agreement with the avail-
able experimental data.
(4) In the result of the simulation, surface defects and inner
defects of the casting products had satisfactory results in
the model that had overflow and air vents at the points
of difficult filling behavior. A special T2 model was the
best model. There were almost no defect differences
between the T2 model and the T2 vacuum model based
Fig. 6 Experimental data of LCD housing: (a) grain size, (b) on surface defect and inner defect analyses. This result
porosity with a different gate design, and (c) mechanical properties agreed with the conclusions of the analysis.
(5) However, the results of the analysis of the porosity rate
using Archimedes principle, which can detect defects
The microstructure and fracture of tensile test of the HPDC under 60 lm, were quite different. It was the same as
alloy are illustrated in Fig. 8. Generally speaking, Chen et al. the results from the simulation interpretation. Finally,
(27) showed that the extended Mg alloys consist of an Mg-rich the specimens of the T2 vacuum model indicated that
phase with a b phase located at the grain boundary. Figure 8 the vacuum-assisted die casting process resulted in a
shows the typical microstructure of the permanent mold cast reduced porosity when compared to the T0, T1, and T2
AZ91 alloy. models.
Nave et al. (28) discovered that the alloy consists of a solid (6) Also, we performed tensile testing to analyze the effect
a-Mg solution containing Al, Zn, and the divorced eutectic of inner defects on mechanical properties of LCD hous-
consists of a massive Mg17Al12 phase and a supersaturated Mg ings. Mechanical properties of LCD housings increase
solid solution. Discontinuous precipitates of Mg17Al12 are also with less porosity rates. The porosity of specimens influ-
observed in the vicinity of the massive eutectic. Dorum et al. ences the tensile strength and elongation and the T2 vac-
(29) studied the presence of zinc and discovered that a uum model had improved mechanical properties.
comparatively faster cooling process in a permanent mold (7) The result of the casting products from the adapted vac-
produces a divorced eutectic. uum system using a thick deviation was satisfactory
The thickness of the products at the start and end of die without inner defects and solidification contraction.
casting must be unchanged. This study was focused on a Also, we developed good products for 4-cavity LCD
0.3 mm thin product even though the deviation of the product housings that were 0.3 mm in thickness.

Journal of Materials Engineering and Performance

 Author's personal copy

Fig. 7 Comparison of simulation and experimental results of the decreased ratio of porosity

Fig. 8 Microstructure of LCD housing by die casting process: (a) original model, (b) 1st corrected model, (c) 2nd corrected model, and (d)
vacuum system model

Journal of Materials Engineering and Performance

 Author's personal copy

Fig. 9 Produced 0.3 mm thickness LCD housing by vacuum system

Acknowledgments 15. A. Kimatsuka, I. Ohnaka, J.D. Zhu, and T. Omichi, Mold filling
simulation of high pressure die casting for predicting gas porosity,
This research was supported by a grant from the Materials & Model. Cast. Weld. Adv. Solidif. Process. 2003, p 335–342
Components Technology R&D Program Funded by The Ministry 16. T. Sakuragi, Prediction of Gas Hole by Mold-Filling Simulation with
of Commerce, Industry and Energy, Republic of Korea. Consideration of Surface Tension, JSME Int. J. B, 2004, 47(4), p 699–708
17. S.M. Xiong, F. Lau, and W.B. Lee, Numerical Methods to Improve the
Computational Efficiency of Thermal Analysis for the Die Casting
Process, J. Mater. Process. Technol., 2002, 128(1–3), p 19–24
References 18. M.R. Barkhudarov and C.W. Hirt, Casting Simulation: Mold Filling
and Solidification—Benchmark Calculations Using FLOW-3DÒ,
1. S.K. Kim, J.K. Lee, Y.O. Yoon, and H.H. Jo, Development of Technical Report, Flow Science, Inc., Santa Fe, 1993
AZ31 Mg Alloy Wrought Process Route, Without Protective Gas, 19. C.W. Hirt and J.M. Sicilian, Proceedings of the 4th International
J. Mater. Process. Technol., 2007, 187–188, p 757–760 Conference on ShiHydrodynamics, Washington, DC, September 1985
2. B.S. Shin, Y. Kim, and D.H. Bae, Deformation Behavior of a Wrought 20. T. Yoshimura, K. Yano, T. Fukui, S. Yamamoto, S. Nishido, M.
Mg-Zn-RE Alloy at the Elevated Temperatures, J. Korean Inst. Met. & Watanabe, and Y. Nemoto, Optimum Design of Die Casting Plunger
Mater., 2008, 46, p 1–5 Tip Considering Air Entrainment, Proceedings of 10th Asian Foundry
3. A. Luo, J. Renaud, I. Nakatsugawa, and J. Plourde, Magnesium Congress (AFC10), Nagoya, Japan, May 2008 20
Castings for Automotive Applications, J Manag, 1995, 47, p 28–31 21. C.K. Jin and C.G. Kang, Fabrication Process Analysis and Experi-
4. D.M. Magers, Light Materials for Transportation Systems, Vol 53, mental Verification for Aluminum Bipolar Plates in Fuel Cells by
Center for Advanced Aerospace Materials POSTECH, Pohang, 1993 Vacuum Die-Casting, J. Power Sources, 2011, 196, p 8241–8249
5. O. Lunder, T.Kr. Aune, and K. Nisanciogtu, Effect of Mn Additions on 22. W.P. Sequeira and G. L. Dunlop, Effect of Section Thickness and
the Corrosion Behaviour of Mould-Cast Magnesium ASTM AZ91, Microstructure on the Mechanical Properties of High Pressure Die Cast
Corrosion, 1987, 43, p 291 Magnesium Alloy AZ91D, 3rd International Magnesium Conference,
6. S.Y. Chang, D.H. Lee, B.S. Kim, T.S. Kim, Y.S. Song, S.H. Kim, and Institute of Materials (UK), Manchester, England, 10–12 April 1996,
C.B. Lee, Characteristics of Plasma Electrolytic Oxide Coatings on p 63–73
Mg-Al-Zn Alloy Prepared by Powder Metallurgy, Met. Mater. Inst., 23. W.P. Sequeira, M.T. Murray, G.L. Dunlop, and D.H. StJohn, Effect of
2009, 15, p 759–764 Section Thickness and Gate Velocity on the Microstructure andMe-
7. B.S. Shin, J.W. Kwon, and D.H. Bae, Microstructure and Deformation chanical Properties of High Pressure Die Cast Magnesium Alloy AZ9I,
Behavior of a Mg-RE-Zn-Al Alloy Reinforced with the Network of a Automotive Alloys, A Special symposium, TMS AnnualMeeting and
Mg-RE Phase, Met. Mater. Inst., 2009, 15, p 203–207 Exhibition, Orlando, Florida, USA, February 9–13, 1997
8. S.H. Choi, D.H. Kim, and B.S. Seong, Simulation of Strain-Softening 24. M.S. Dargusch, G.L. Dunlop, and K. Pettersen, Proc. Magnesium
Behaviors in an AZ31 Mg Alloy Showing Distinct Twin-Induced Alloys and Their Application, Wolfsburg, B.L. Mordike and K.V.
Reorientation Before a Peak, Met. Mater. Inst., 2009, 15, p 239–248 Kainer, Eds., Warkstoff-Informationsgesellschaft mbh., Frankfurt,
9. E.A. Herman, Die Casting Dies: Design, North American Die-casting 1998, p 82–277
Association, 1992, p 15–24 25. ASM Handbook Committee: Metallography and Microstructures, ASM
10. J.R. Brevick, Die Casting Defects Causing Rejections During International, Metals Park, OH, 1985, p 425–434
Machining, Die Cast. Eng., 1997, 5, p 42–46 26. M. Nave, A.K. Dahle, and D.H. St John, The Effect of Solidification
11. C.M. Gourlay, H.I. Laukli, and A.K. Dahle, Defect Band Character- Rate on the structure of Magnesium-Aluminium Eutectic Grains, Int.
istics in Mg-Al, Al-Si High-Pressure Die Castings, Metall. Mater. J. Cast Met. Res., 2000, 13, p 1–7
Trans. A, 2007, 38A, p 1833–1844 27. Y. Chen, Q. Wang, J. Peng, C. Zhai, and W. Ding, Effects of Extrusion
12. S.G. Lee and A.m. Gokhale, Formation of Gas Induced Shrinkage Ratio on the Microstructure and Mechanical Properties of AZ31 Mg
Porosity in Mg-Alloy High-Pressure Die Castings, Scripta Mater., Alloy, J. Mater. Process. Technol., 2007, 117, p 281–285
2006, 55(4), p 387–390 28. M.D. Nave, A.K. Dahle, and D.H. St John, TMS Annual Meeting,
13. M. Mclaughlin, C.W. Kim, and A. Conrad, Flow Pattern in the Casting Nashville, TN, H.I. Kaplan, J.N. Hryn, and B.B. Clow, Eds., TMS,
Process, Mater. Des., 2004, 25, p 655–661 Warrendale, PA, March 2000, p 233–242
14. A. Kimatsuka, I. Ohnaka, J.D. Zhu, and T. Kamitsu, Mold Filling 29. C. Dørum, O.S. Hopperstad, T. Berstad, and D. Dispinar, Numerical
Simulation with Consideration of Gas Escape Through Sand Mold, Int. Modeling of Magnesium Die-Castings Using Stochastic Fracture
J. Cast. Met. Res., 2002, 15(3), p 149–152 Parameters, Eng. Fract. Mech., 2009, 76, p 2232–2248

Journal of Materials Engineering and Performance