Smoothed Particle Hydrodynamics - Application
Validation study of gravity die casting
Gravity die casting (GDC) processes are capable of making
complicated high integrity components, such as wheels, cylinder heads,
engine blocks and brake callipers, at lower cost than most other casting
methods. Cycle times in gravity die casting are shorter than in the sand
casting process that increases the quantity of castings produced per unit
time. Surface finish and internal quality (particularly pertaining to
porosity) are better using this process. Improvements to both product
quality and process productivity can be brought about through improved die
design. These include developing more effective control of the die filling
and die thermal performance.
Experimental setup
Water analogue modelling is widely used in gravity and
low pressure die casting to study the characteristics of the molten
aluminium flow in die cavities. To investigate the flow pattern in a
cavity with typical features found in dies (ie sudden transitions in
section thicknesses, bends, branches, etc) a transparent die has been
built. Figure 1a shows the cavity geometry and Figures 1b and 1c show two
of the designs of the gate and runner system which were tested under
different flow conditions and gate sizes. These are the two models
considered in this paper and they are called Model 1 and Model 2. Several
blocks of clear Perspex 20mm thick were used to assemble the mould between
two clear flat support plates.
A constant head tank situated above the sprue was used
to supply the flow of coloured water into the cavity. Four nozzles with
different diameters were located above the sprue to produce four different
flow rates in the cavity. The water flow was initiated by a stopper
connected to a solenoid valve. The outflow for each nozzle has been
calibrated and it was found that the flow rate was constant.
The flow in the cavity was recorded for a number of
different flow rates and gate sizes. A conventional Panasonic VHS video
camera has been used to record the real time images of the flows. At a
later stage, the video images were converted to digital format with the
use of a Matrox Mystique frame grabber and software on a Pentium II
personal computer.
|
|
Transparent die models.
(a) Die cavity geometry. The lengths are in mm and the third
dimension is 20 mm thick.
(b) Model 1: Bottom gating, single gate.
(c) Model 2: Vertical orientation, bottom gating.
|
Model 1 comparisons
Two-dimensional SPH simulations have been
performed for both of the models shown above; here results are presented
for Model 1. (Results for Model 2 can be found in [1].) For this model,
the diameter of the inlet nozzle is 9.5 mm and the velocity of fluid flow
through the nozzle is kept constant at 1.02 m/s giving a flow rate of
0.072 l/s. Under this arrangement, the filling time for the cavity of the
die is about 4 seconds.
For the SPH simulations, the filling processes were
assumed to be isothermal and the fluid properties (density and viscosity
at the temperature of 20oC) used were those of water. A
resolution of 22 particles per centimetre were used. For Model 1, the
boundary of the die requires 3266 particles. To simulate the inlet
boundary condition, a row of particles having the inlet velocity that
spans the width of the nozzle is generated at regular intervals. As the
SPH simulations are two-dimensional, the width of the inlet is computed
from the radius r of the nozzle in mm according to  r2/20.
The number of particles increases steadily with time as fluid enters the
die through the inlet nozzle. The total number of particles is around
52000 near the end of the simulation.
In the figures below, the SPH results are compared with
both the experimental results and also the results obtained using
MAGMAsoft, a 3D solidification and fluid flow package commonly used in the
die casting industry. The fill patterns obtained numerically and
experimentally at selected time instances. In the experimental images, the
number near the top right hand corner is the time in seconds. At 1.2 s,
the fill pattern in the die cavity is symmetric with respect to a vertical
through the gate. The SPH result compares very well with experiment apart
from two minor discrepancies. The first discrepancy is the presence of a
void in the runner in the SPH result. The second discrepancy is that the
SPH result shows the fluid spreads out thinner and wider than the
experiment on the horizontal section of the die cavity to the sides of the
gate. The MAGMAsoft result shows a higher fountain above the gate than the
experiment. It also shows a higher fluid level in the sprue.
From 1.6 s to 2.0 s, the filling process shows a
sloshing motion from side to side in the rectangular section of the die
above the gate. This sloshing motion is caused by the recirculation of the
fluid from the jet hitting the body of fluid above the gate. The SPH
results reproduce this motion and the levels of fluid reached in the
various part of the die very well. The MAGMAsoft results also reproduce
these features but exhibit some voids in the lower horizontal section of
the cavity above the gate. Apart from a small void in the right half of
the lower horizontal section of the cavity at 1.6 s, there is no void
present in the experimental results for 1.8 s and 2.0 s.
As the amount of fill increases, the sloshing motion in
the rectangular section of the cavity is increasingly dampened down. By
2.6 s, this motion is barely evident as the fluid surface in the
rectangular section of the cavity above the gate is now almost flat. Both
the SPH and MAGMAsoft simulation results reproduce this feature well.
However, the MAGMAsoft result shows a lower fluid level in the sprue than
the experiment. By 3.4 s, the rectangular section above the gate is
completely filled and the horizontal section of the C-shaped cavity to the
right is beginning to be filled. This is reproduced very nicely by both
SPH and MAGMAsoft simulations. The SPH simulation slightly over-predicts
the height of water in the sprue whereas the MAGMAsoft result slightly
under-predicts this level.
| MAGMAsoft |
Experiment |
SPH |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Comparison of
experimental and numerical results at selected times during the
filling of Model 1. |
Conclusions
The agreement of the SPH simulations with
experiment and also with the MAGMAsoft results is very good, and
demonstrates that the SPH method is able to predict a significant
proportion of the detailed features of the free surface shapes. In
general, the natural free-surface capability of SPH allows it to reproduce
a selection of these fine features better than MAGMAsoft. The fact that 2D
SPH simulations compare so well with experiment implies that the fluid
flows in the two dies studied in this paper are essentially 2D.
The experiments and MAGMAsoft simulations described above were
undertaken at CSIRO Manufacturing Science and Technology. More details
concerning this validation study can be found in [1].
References
[1] J. Ha, P.W. Cleary, V. Alguine and T. Nguyen, Simulation
of die filling in gravity die casting using SPH and MAGMAsoft,
Proceedings of the Second International Conference on CFD in the Minerals
and Process Industries (Melbourne, 1999).
|
To
top