Adaption of Parameters in FLOW-3D for the Simulation of the Aluminum Integral Foam Molding Process
This application note was contributed by Johannes Hartmann and Vera Jüchter, Department of Materials Science, Chair of Metals Science and Technology (www.wtm.uni-erlangen.de), University of Erlangen-Nuremberg
Aluminum foams show exceptional properties such as good damping and high energy absorption and mass specific flexural stiffness . The stiffness makes it especially attractive for use in load-bearing and at the same time lightweight structures. In order to increase this weight-specific stiffness and for better load transmission, a compact skin is needed , as realized in Aluminum Foam Sandwiches (AFS).
At the Chair of Metals Science and Technology at the University of Erlangen-Nuremberg, the modified die casting process "Integral Foam Molding (IFM)" has been developed in order to produce aluminum foams with an integral solid skin, a foamed core and a gradual transition region in-between (see Fig. 1 ). This process was developed from foam injection molding of polymers and is therefore appropriate for cost-effective one-step mass production of complex foam parts with compact layer. A simulation technique, described in this note, has been adapted to model this process as an aid in selecting process parameters.
Figure 1. Cross section of an aluminum integral foam with a compact skin, a transition region
with decreasing relative density and smaller pores, as well as a foamed core.
Aluminum Integral Foam Molding Technology
A certain amount of blowing agent (magnesium hydride, MgH2) is therefore placed in the runner system and the shot chamber filled with aluminum melt (the schematic process cycle is depicted in Fig. 2; the process is described in detail in ). As the piston advances, the powder is entrained in a turbulent way into the mold. In case of the technology variant "High Pressure Integral Foam Molding (HP-IFM)," the part is completely filled at a high ambient pressure as known from the standard die casting process, guaranteeing an excellent surface quality. Starting from the tempered surfaces of the mold, the melt starts to solidify to an integral solid skin. After some milliseconds – the so-called delay time – the mold is opened over a core puller system and the volume locally increased (and pressure decreased) which initiates pore growth in the still semi-solid inner region due to the thermal decomposition and hydrogen release of the magnesium hydride particles. Every blowing agent particle represents a pore nucleus from which pore growth starts until it is stopped by the counter-pressure of neighboring pores expanding simultaneously. The forming cell walls are hereby stabilized by primarily solidified particles of the aluminum alloy, so-called endogenous stabilization .
Figure 2. Schematic process cycle of "High Pressure Integral Foam Molding (HP-IFM)" of aluminum.
Figure 3. Schematic curves of decomposition of magnesium
hydride as a function of the melt temperature,
calculated by the Johnson-Mehl-Avrami approach .
Click on the image for a bigger graph.
A prerequisite for a homogeneous foam morphology in the entire volume of the casting part is a good distribution of the particles at the moment of decomposition initiation. Furthermore, the temperature of the melt during blowing agent entrainment is on the one hand decisive for the decomposition kinetics of the magnesium hydride (see Fig. 3) and on the other hand determines the amount of solid phase during foaming. Insufficient solidified alpha-grains lead to non-stabilized struts between forming neighboring pores as drainage of the melt due to capillary forces takes place leading to pore coarsening. However, a very high amount of solid phase increases the rigidity of the matrix and leads to disrupted structures by hindering spheroidization of the developing pores .
Microcellular Aluminum Integral Foams – Approaching the Process Limits
Simulation of the integral foam molding process represents a powerful tool that not only helps to investigate the mold filling properties of a new part design but can also predict particle entrainment and determine the foam evolution conditions saving cost-intensive trial series. The goal of current research is to decrease pore size while keeping the porosity level constant. Computational fluid dynamics (CFD) simulation would help get as close as possible to the current process limits or to even push them further. Improvement in foam morphology would not only lead to more homogeneous structures with smaller scatter in the mechanical properties but would also allow production of thinner parts whose mechanical properties might then be determined by finite element. This objective can only be achieved by a high particle distribution density within the melt and at the same time a totally stabilized pore growth with a decrease in coalescence phenomena.
Adapting the Simulation Parameters to Practical Integral Foam Molding Experiments
Figure 4. Adjustment of heat transfer in FLOW-3D v9.4
by comparisons of a real solidification curve (black)
to the growth rate of the solidified skin in simulation (red).
Click on the image for a bigger graph.
In order to be able to use CFD simulation for the reliable prediction of particle behavior or temperature fields, different simulation parameters have to be determined by adjusting those to match real experiments. To this end, integral foam parts were produced with varying delay times between ca. 30 and 130 ms resulting in different dense skin thicknesses where foam formation was impossible due to a solid phase fraction exceeding a certain percentage1 at the moment of mold expansion and pore growth initiation. This leads to a characteristic so-called solidification curve with an axis intercept and slope depending on the die temperature and other chosen process parameters (see Fig. 4). Simulating the casting cycle for the alloy AlSi9Cu3(Fe) by varying the heat transfer coefficient (the value for fully liquid melt as well as for fully solidified melt), the experimental solidification curve can be fitted. In order to achieve this goal, it was necessary to extend the simulation to the dosing of the melt into the shot chamber to depict the real temperature distribution before the beginning of piston movement. The temperature was locally measured in the shot chamber by placed thermocouples and could be successfully depicted in good agreement with the real data within the simulation. The same can be referred to temperature measurements at the die surface during mold filling whose evolution over time correlates well with the simulation results.
In a second step, further parameters defining the melt flow behavior such as the surface tension or the coefficient of solidification drag are adjusted by comparing simulations with different settings to experimental studies where the piston is stopped before filling the mold (see Fig. 5). As soon as the flow of the melt within the simulation is consistent with the practical tests, the parameters are set.
Figure 5. Adjustment of melt flow defining parameters such as the surface tension
by comparisons of real experiments (left) to simulations (right).
After defining the cooling as well as the flow characteristics of the melt, the entrainment of the particles is simulated. In order to adjust the simulation for correct particle/fluid-interaction, the particle-defining parameter coefficient of particle drag is fitted by comparisons to x-rayed samples where substitute particles with a higher contrast in x-ray characterization to aluminum than magnesium hydride are entrained, e. g., copper or iron particles (see Fig. 6). The simulation results fit quite well with the experiments so that a reliable forecast of particle distribution as a function of process parameters can be deduced.
Figure 6. Adjustment of parameters influencing particle/melt-interactions by comparisons of x-rayed samples
(left); produced by the entrainment of copper particles) to simulations (right).
Altogether it could be demonstrated that FLOW-3D can be an important instrument to investigate potential weak points in the fabrication of new integral foam parts before their actual production. In that way, a successful filling and blowing agent distribution without cold flow or dead zones can be assured. Furthermore, thanks to the correct depiction of temperature fields to be expected, the formation of compact skin and decomposition properties of magnesium hydride (and so the pore formation conditions) can be deduced. This offers the potential to define the process parameters to satisfy customer requests with regard to integral foam structures.
 C. Körner, R. F. Singer, Adv. Eng. Mater. 2000, 2 (4), pp. 159-165.
 C. Körner, in Integral Foam Molding of Light Metals – Technology, Foam Physics and Foam Simulation, Springer, Berlin, Heidelberg, Germany 2008.
 H. Wiehler, C. Körner, R. F. Singer, Adv. Eng. Mater. 2008, 10 (3), pp. 171-178.
 J. Hartmann, A. Trepper, C. Körner, Adv. Eng. Mater. 2011, 13 (11), pp. 1050-1055.