Leaping microstructure hurdles with process modeling: beyond simply identifying shrink porosity and fill problems, numerical tools can predict more accurately defects at a microstructure level to help achieve better material properties.

Casting process modeling technology has come a long way since the early and mid-1980s, when the design engineer worked with just a 2-D model. The early days focused on the simple quest of identifying hot spots in the casting. As the CAD and numerical simulation software packages evolved, quick changes to the design of the casting could be made with relative ease.

Most of the casting simulation packages today can handle solidification and fluid flow in the mold with relative ease, and the role of such tools continues to grow, both in the design of the component and optimization of the casting process to produce a functional part under intended loading. The advancement in computer hardware available to run these programs with multiple CPU configurations allows a fast turnaround time by increasing the speed of solution, even on large models.

The focus now is to accurately predict and correct more in-depth elements of the process, such as stress and deformation, micro and gas porosity and as-cast mechanical properties. Knowing the answers to these questions early in the manufacturing process affords significant time and cost savings to the casting purchasers and the casting facility.

Stress Detective

The main goal of the casting process is to approach and achieve net shape. Large deformation or distortion in the part requires more rework, such as hot pressing, even after a heat treating operation. To aid in the understanding of why parts deform or develop residual stresses, some casting process modeling software now can derive thermally-induced stresses, which include effects of strength or constraint on the mold. For permanent mold casting, simulating the strength of molding components has been proven necessary in determining the amount of distortion in the casting, which is the main prevention against significant warpage in the mold itself.

By allowing the stress analysis to be fully coupled with the thermal and fluid calculation, a clearer understanding of the physical phenomenon is obtained. Comparing the mushy zone location and the evolution of stresses while cooling indicates problems such as hot tearing and cracking, which occur when the stresses in the local region go beyond the yield stress. Thus, a tight coupling of the thermal and stress calculation is required.

Another typical phenomenon involves the gap formed between the casting and the permanent mold as the casting solidifies and shrinks. When this gap develops, there can be a significant reduction in heat transfer from the casting to the mold components where contact is lost. Conversely, there may be locations near the cores that, due to the shrinkage of the casting onto the core, create an increasing contact pressure and rate of heat transfer from casting to core. Accurately tracking the heat transfer results is a better indication of surface shrink.

A multi-body mechanical contact algorithm is employed to compute the contact and gap formation between the casting and tool parts. Contact between different tool parts also is considered. An augmented Lagrangian type method is used in the contact algorithm, and an additional automatic penalty number adjustment technique is implemented. This enhances the stability and robustness of the contact computation algorithm.

In the first example of this analysis (Fig. 1), a simple T-shaped casting of A356 in a H13 permanent mold is modeled. The effective interface heat transfer coefficient at two different points on the casting is plotted. The top curve is from a point experiencing increasing contact pressure as the casting contracts. The middle curve is from a point where a gap is opening up between the casting and mold, assuming air is present. The bottom curve is from that same point, but assuming a vacuum.

[FIGURE 1 OMITTED]

The large variation in the coefficient illustrates the importance of accounting for local conditions. In addition, this example illustrates the value of reverse coupling of the mechanical deformations with the energy solution. This effect can be seen in Fig. 2, where the heat flux contours are plotted. The heat flux is greatest where the contact pressure is highest..

[FIGURE 2 OMITTED]

Post ejection springback and further relaxation after mold removal also can be tracked, completing the necessary cycle to determine final part shape (Fig. 3). Finally, heat treatment processes may define the final net shape. By appropriately simulating the thermal loading on the part during heat treatment, any additional shape change may be captured, either independent of the casting process or as a continuance of the deformation and residual stress evolution formed from casting.

[FIGURE 3 OMITTED]

Predicting Microstructure 

Hot spots and macro porosity prediction was the first goal of casting software, leading to the generic classification of these tools as "solidification software." With the exponential advances in hardware, more computing power has allowed analysis to reach resolution at the microscopic level. Now these predictions can be made at a dendritic level, thereby significantly increasing the accuracy of the solution.

In addition, metallurgical research has evolved to determine the physical phenomena involved in advanced solidification processes. Thus, micro porosity prediction includes porosity variance due to hydrostatic pressure in gravity castings, intensification effects in high pressure die castings or low pressure permanent mold castings and the evolution of gasses, such as hydrogen content in aluminum.

Microstructure prediction algorithms should calculate automatically the microstructures based on the composition of the alloy. This is achieved by linking the module with thermodynamic databases. Depending upon the chemical composition, the microstructure module automatically detects the phases that will appear and the type of microstructure that should be computed.

One of the models used as the basis of microstructure prediction is a Gaussian distribution model that defines the relationship between the number of nuclei and the undercooling, which is shown as a distribution curve in Fig. 4.

[FIGURE 4 OMITTED]

Proper Mechanics

Cost savings in castings are achieved by producing the lightest part possible to perform the intended job. Today, an increasing trend exists in the industry toward alloys that provide increased strength over traditional alloys. In order to determine whether the casting process produces a part with proper as-cast mechanical properties, microstructure prediction is required.

Slight variances in elemental composition can signal if the change has a significant effect on final mechanical strength. Additionally, optimized cooling rates, including those after shakeout or ejection, can produce desired mechanical properties in as-cast parts, thus making possible the production of castings that may not require a heat treating operation.

By understanding the relationship among material composition, cooling conditions and the microstructural evolution of mechanical properties, the casting engineer and part designer can consolidate efforts to produce the best part for the job. If more strength is needed and is developed during the process, then design allowances, such as larger thicknesses, more mass and larger fillets, may not be needed, and a more efficient part design and cost savings for the metalcasting facility can be achieved. Additionally, these analysis-derived mechanical properties can be fed back into the structural analysis to understand how the as-cast part will perform.

As an example, a sample plate was cooled in the mold to room temperature, whereas an identical model was cooled with the variation that shakeout occurred just after solidification once the solid state temperature was achieved (Fig. 5). As a result, these different cooling conditions created different microstructure evolution conditions (Fig. 6). The slowly cooled plate developed a much lower tensile strength than that of the quickly cooled plate, which corresponds with experimental results.

[FIGURES 5-6 OMITTED]

In the chosen example (Fig. 7), as the carbon equivalent (CE) equals 0.767, there is no peritectic reaction. The peritectic occurs when 0.09% < CE < 0.53%. Only a primary phase occurs, which is austenite forming during solidification. The primary dendrite radius corresponds to the primary grain size. Larger dendrite radii are located in the lower cooling rate areas of the casting. The primary solid fraction corresponds to the fraction of the primary phase (Fig. 8).

[FIGURES 7-8 OMITTED]

With the Grain

Grain structure calculations may be processed, allowing prediction of crystallographic orientations. The orientation and number of grains produced is directly related to the stress and fracture properties of the part, becoming a critical determinant of the proper processing of certain highly-loaded parts, such as aircraft turbines or compressor blades.

The results of a grain structure calculation, shown in Fig. 9, illustrate with different colors the varying grain crystallographic orientations. Depending on the required component's "in-service" performance, the grain size and orientation may become key criteria that need to be investigated.

[FIGURE 9 OMITTED]

Adi Sholapurwalla and Sam Scott, ESI Procast, Bloomfield Hills, Michigan 

For More Information

"Mandating IHTC Through Casting Simulation," D. Moore, K. Mohler, X. Wan and R. Pehlke, MODERN CASTING, Aug. 2005, p. 30-33.

"Keeping Up With Process Simulation Technology," MODERN CASTING staff report, MODERN CASTING, Aug. 2004, p. 26-29.

"Foundries Use Casting Simulation to Offer Optimized Components," T.C. Midea, D. Schmidt, D. Goettsch, R. Anaparti, and B. Gruidl, Engineered Casting Solutions, Winter 2000, p. 38-40.

Adi Sholapurwalla is an engineering manager and Sam Scott is a technical support coordinator for ESI Group, Bloomfield Hills, Mich.