The science of process design for casting has developed relatively slowly until recently. Rule-of-thumb and empirical procedures, along with much trial and error, characterized this discipline over the past century. In the past decade, computer simulation of the process has come into general use, so casting engineers have a more scientific basis upon which to design. However, even with simulation, the design process itself has tended to be basically a trial-and-error search for an “adequate” design.
The state of the art in simulation is now changing even more rapidly. In August 2001, the first commercial system for optimizing the casting process was released. Based on a marriage of Altair Engineering’s OptiStruct optimization software and Finite Solutions’ SolidCast simulator, the new system — called OPTICast — offers casting engineers a rational and automated design methodology that was not previously available.
Casting engineers struggle with physics to produce a good, high-quality part. One of the basic complications is that when liquid metal cools and freezes, it contracts and becomes dense. If the casting process is not designed correctly, then this contraction will result in holes, or “shrinkage porosity,” within the metal part. Such a defect can cause a cast part to fail under stress or even interfere with the function of the part if machining operations expose the defect.
The usual solution is to provide a shape attached to the casting that can serve as a delivery source of liquid metal during solidification, “feeding” metal into the casting to compensate for the contraction. These shapes “risers” or “feeders” are removed from the casting net shape by cutting or grinding.
This situation creates something of a dilemma for the casting engineer. Larger risers tend to ensure the absence of shrinkage defects in the casting. However, risers add to the cost of producing a part: the metal in the risers is not part of the finished casting, but it must be melted, poured, removed, and handled as part of the process. Therefore, the smaller the risers, the lower the cost of producing a casting.
Given today’s extremely competitive environment, it is essential to find the “perfect” riser size, not only to keep defects from forming but also to hold extraneous metal cost to a minimum. The OPTICast system does that - and more.
On the optimization path
The initial design of the OPTICast system began in early 2000 as a project sponsored by the Edison Materials Technology Center (EMTEC), an Ohio organization whose mission, in part, is to encourage technology development and application in materials processing industries, including foundries. EMTEC provided the funding for the project and brought together the participants, whose mission was to link casting simulation and optimization technology to further improve the casting design process.
To accomplish that goal, Altair and Finite Solutions’ team members developed an interface to the Altair optimization software and created a method to automatically adjust model geometry and initial conditions based on changes suggested by the optimization system. Developers considered other issues, such as identifying which features of the model would be made available as “design variables” and which outputs could be used for “objective functions” and “constraints.”
For example, developers designated the size of the riser, initial material temperatures and pouring time as key design variables. They determined that one of the most useful combinations of potential constraints and objective functions involved specifying the level of predicted shrinkage porosity within the casting as a constraint and asking the system to maximize the process yield - defined as the ratio of the net weight of the casting to the weight of total metal poured - as the objective function. This results in the system finding the least amount of metal that will produce a cast part with a desired quality level.
OPTICast in action
The optimization software speeds up the casting design process, as demonstrated in the following example, in which the casting designer’s goal is to decide if the current design has resulted in a “good part” at the lowest possible cost.
First, the user identifies the model features that are to be considered design variables. Here, one of the large side risers, (shown in red), has been selected. The minimum and maximum scale factors are specified, defining the design space for this riser. The user also needs to indicate a “pin point,” which is a point on the selected shape that will stay at a constant location while the shape expands and contracts.
Once these items are specified, the next feature (one of the other risers) is selected as an additional design variable, and the same information is entered. This continues until all risers have been selected.
The user then optionally selects one or more constraints and the objective function from a list of system outputs. Once these selections have been made, the optimization project is launched. All that the user must do is to click a menu selection. The process after that point is automatic.
The software runs a series of process simulations with changes in the design variables. The system automatically modifies the model each time, and results are compared to the constraint(s) and objective function.
Once the optimization is complete, the final model is available for display and postprocessing of results within SOLIDCast. In addition, the progress of the optimization can be examined either graphically or by loading a spreadsheet showing progressive values of design variables and output data.
Seeing is believing
One of the reports OPTICast generates is a plot of progressive values of the objective function. The process yield in the example was increased from an initial value of 63% to more than 80% by running 15 process simulations. This yield increase was achieved by downsizing the risers to about 73% of their initial size.
In this case, the amount of metal saved per casting poured was about 100 lb (45 kg); total material savings were 75,000 lb/year. The annual cost savings for production of 750 parts amounted to about $9,000. What’s more, annual energy savings due to the reduction in the amount of metal required to be melted were about 27,000 kW/hr, with a projected cost savings in kW/hr of $2,700/year.
For verification of the optimization results, the user can examine the final optimized model. For example, a time plot shows progressively solidified areas within the casting and risers. Another plot demonstrates calculated metal feeding patterns due to contraction of the metal.
From one relatively simple example, you can see that optimization has the potential to significantly reduce costs and energy usage while maintaining or improving quality in the casting process. The total amount of time required to define this project, run the optimization, and analyze results was less than half a day, yet the annual payback was more than the initial cost of the software. Given the number and variety of parts cast in the world industry, the application of optimization to casting process design has tremendous potential.
In addition to yield improvement, it is possible to approach quality maximization for critical parts through proper selection of constraints and the objective function. For example, the strength of the material in a casting is related to the cooling rate of the metal while the casting is solidifying: higher cooling rates generally translate into better material properties (higher yield and ultimate strength). Optimizing a casting for maximum cooling rate, while constraining the formation of internal defects, can result in a part that can withstand higher loads - and that performs better over its lifetime.
There have been a number of successful applications of OPTICast in a variety of casting processes. EMTEC program manager Percy Gros says, “We have member companies that have expressed their desire to immediately make great use of this new OPTICast product.”
For the metalcasting industry, optimizing the process design means that is no longer necessary for the design engineer to examine each process simulation and determine whether the process is complete, and if not, what should be changed. Using optimization, the engineer can spend his/her time more productively on problem definition. This nudges the design engineer, perhaps unconsciously, into taking a more rational approach to the overall design process. Ultimately, it will result in more successes. And, the companies that adopt optimization technology will be better competitors in the global market.
Lawrence E. Smiley is president of Finite Solutions Inc., Cincinnati. Before launching the company in 1993, he worked with Reliable Castings Corp., Bishopric Products Co., and Republic Steel Corp.