Simulation software in action: five users share their experiences.

Five software users discuss how the simulation of cast components can improve molding design, reduce defects, improve quality and increase profitability.

As part of an AFS Engineering Process Design and Modeling Committee (1-F) panel presentation at the 1999 Casting Congress, five users of casting simulation programs were asked to share information on how they use software to improve metalcasting processes and casting designs.

In this article, five individuals detail how simulation software allowed them to improve a casting's design. Through their practical experiences, foundrymen can see how simulation is being used to improve casting quality and impact profitability every day on the foundry floor.

Eck Industries

Cast aluminum knuckles are increasingly used in automobiles and trucks in order to reduce weight, improve handling and improve fuel economy, and in 1997, Eck Industries, a 350-employee aluminum caster in Manitowoc, Wisconsin, was asked to cast a front steering knuckle for the Plymouth Prowler. Early development revealed difficulties in manufacturing the part.

A decision was made to produce this component in a low-pressure permanent mold, casting a left- and right-hand knuckle in each machine cycle. In light of past production difficulties and the structural loading of the casting, the company elected to do a complete fluid flow and solidification model. In addition, information on the solidification rate of the casting would help estimate dendrite arm spacing and expected mechanical properties.

The material specified was a low-Fe (0.1% maximum) version of A356 due to elongation requirements.

The casting requirements were:

* ASTM frame 2 maximum defect level;

* 35 ksi minimum tensile strength;

* 25 ksi minimum yield strength;

* 7% minimum elongation;

* tensile bars to be pulled from high-stress areas of the casting;

* 100% radiography and dye penetrant nondestructive testing.

Two fluid flow analyses of the cavity filling were based on casting geometry and modified with the addition of a feeding sprue. A post-pour metal temperature of 1202F (650C) was used in all analyses.

The process was modeled by an outside firm with cycle, coating and cooling information provided by Eck. For each analysis, 16 complete process cycles were modeled in order to find the steady state operating conditions of the die, and a final detailed solidification simulation was run after the cycles.

The initial modeling demonstrated a tendency for shrinkage defects in the ball joint ends. In subsequent analyses, the shrinkage potential in the ball joint proved stubborn. Water cooling was added to the ball joint ends and center of casting, which created a problem in the steering arm without solving the problem in the ball joint end. A modeled configuration with water cooling only in the bottom mold half showed improved results.

Engineers found an air cooling configuration that would get the ends to feed directionally, but it also required a more conductive interface between the casting and the die on the end faces of the casting. A thin graphite coating on the die in those areas provided the condition for real castings.

The modeling presented an accurate picture of the actual casting process [ILLUSTRATION FOR FIGURE 1 OMITTED]. Actual casting experience showed less sensitivity to shrinkage in the ball joint than was expected, producing sound castings over a wide range of cooling and mold coating conditions. The castings from the initial mold trial were good as soon as the mold reached operating temperature. The modeling allowed Eck to produce castings off a new tool without changing the mold gating or cooling.

The steering arm was more sensitive to shrinkage than expected, with slight deviations from the modeled parameters leading to casting defects. This suggests that additional work could be done with the process to determine sensitivity of casting quality to processing parameter variations.

GM Powertrain 

GM Powertrain's Saginaw, Michigan operation performed a simulation and analysis of the metal flow, slag movement and gate removal of a gray iron V-8 engine block. The goals of the simulation were to address performance issues in the gating design, including fill time, gate removal stresses, metal front turbulence and slag capture efficiency. In addition, the simulation was performed in an effort to increase casting quality and foundry productivity. Quality is increased if engineers can reduce the likelihood of slag entering the casting cavity by finding a more efficient slag catcher and reducing the slag generated by surface turbulence in the gating. Productivity is increased by reducing the fill time (increased line speed and lower metal temperature) and ensuring the gating breaks apart on the shakeout deck without manual assist.

The entire gating and casting geometry was modeled in the simulation. The metal stream was introduced at the ladle nozzle with a source of buoyant particles to simulate slag contamination. In addition, various regions in the flow field were identified with damaging metal surface turbulence and which promoted the entrapment of slag particles.

The simulation showed that the gating system would fill in the first 3.5 sec of the pour, during which some metal jetting and splashing of metal occurred in the system. Then the simulation predicted a quiescent metal flow front as the casting bottom filled.

The simulation of metal and slag particle movement over the entire 23-sec pour revealed that 72% of particles were captured in the gating system and 28% escaped into the casting [ILLUSTRATION FOR FIGURE 2A OMITTED]. A sorting of the slag particles' final location allowed a capture efficiency curve to be calculated for the basin and various runners [ILLUSTRATION FOR FIGURE 2B OMITTED]. The simulation showed a gating efficiency of 67% and a gating plus basin efficiency of 72%.

Stress levels in the gating under removal loads also were predicted. These results provided a baseline comparison against alternate gating designs and were an important step in improving metal quality, shorter fill time and higher metal yield while still meeting geometric and gate removal constraints.

The group concluded that baseline performance parameters should be a fill time of 23 sec, gate removal stresses of 190 megapascals, a metal front jet 4.5 cm in height and a slag capture efficiency of 67%. Alternate designs had to meet or improve on these levels before being considered for casting trials. based on these parameters, engineers altered the "slag catcher" at the front of the design as well as the configuration of the cross gates and rail runners.

Pacific Steel Casting Co.

When engineers at Pacific Steel, a 320-employee steel foundry in Berkeley, California, first looked at casting simulation software 5 years ago, they were not convinced of its practical applicability. However, both the software and hardware to run these simulation packages have progressed by leaps and bounds. The company has been using simulation for the last 3 years and is happy with the results. Simulation is used to identify and solve the problems that occur in castings, try out new ideas and optimize the production cost of the casting.

Pacific Steel performed a simulation on an insert casting - a high-production part weighing about 20 lb. The original risering configuration for this job consisted of two risers on either end of the casting in a green sand mold, and it was thought that this could be improved.

The main objectives of simulation were to increase the number of castings per mold and the yield and to reduce the riser contact area.

The first version consisted of four castings/mold, one common riser feeding two castings in an "over/under configuration" and chills on the smaller end of the casting. This configuration achieved the objectives in that:

* it increased the castings/mold, even though the mold size itself increased;

* increased yield because fewer risers per casting are required;

* decreased cleanup time because of the eliminated riser contact area.

The benefits achieved far outweighed the added costs of a "splitter core" required by the new design. Simulation analysis of this part showed a small region of "acceptable level shrink" in the part.

The first version on the shop floor, however, was not a success. Actual radiographic inspection of the part revealed an unacceptable level of shrink. Engineers tried to reconcile the differences between the simulation and reality, and after some investigation of the possible causes for this defect, they found that a smaller chill was used than what was specified.

Another computer analysis was performed to see the effect of the smaller chill. The simulation [ILLUSTRATION FOR FIGURE 3 OMITTED] confirmed the shrink in the region. It was evident that the shrink in the casting was very sensitive to the size of the chill; anything less than the recommended chill size would cause an unacceptable shrink in the casting. Hence, engineers wanted a consistent and more predictable way of making this casting.

In the second version, a back riser was added to the smaller end of the casting in place of the chills. This configuration looks similar to the original configuration, yet it still retains some of the objectives. The yield is better than the initial configuration, and the number of castings per mold is higher.

Software analysis of this configuration predicted a "shrink-free" casting. This is the configuration now in production, and Pacific Steel does not have any shrink-related problems with the casting.

This particular design and simulation process illustrates the importance of accurate modeling. The results of a simulation are only as accurate as the model that describes the casting process. Engineers must do a good job of capturing the shop floor conditions. They also must use accurate thermophysical properties for all of the materials used in the model to ensure a successful simulation.

Selee Corp.

Filter supplier Selee Corp., Hendersonville, North Carolina, decided to optimize gating design and filter sizing for a customer's casting, a 13-ft, 19,000-lb ductile iron bridge base support that is poured for 120 sec at 2600F (1427C). The original design called for eight 6 x 12 x 1.25-in. silicon carbide filters. The problem with this design was that it called for a tremendous, inefficient gating system that caused turbulent filling of the casting. Because the bridge base support is so large, the company couldn't afford to make a mistake.

The goals of the simulation were to accurately predict the outcome of casting before any metal was poured, to verify that the gating and filtration systems were accurate, to predict possible problems, and to ensure a complete pour and high-quality finished casting.

Selee received an original drawing of the support from its customer and constructed a solid model with a 3-D solid modeling package. This was imported into meshing software and a finite element-based casting simulation software package was used to generate a filling and solidification model. That simulation then was analyzed for potential problematic areas.

Engineers took a look at the original gating system and determined that it was poorly designed. The original horseshoe runner system was changed to a main runner design that covers the distribution area. The new design has eight filters and eight ingates.

The advantages of this redesign include:

* bottom fill instead of top fill, so the mold fills with lower turbulence;

* reduced runner length by 35 ft, improving yield;

* reduced possibility of sand inclusions from erosion;

* hotter pour.

Engineers took the new design to the next stage - simulation of filling - because they weren't sure that the casting would solidify the same way. The simulation predicted filling to be even and docile, but it also showed two areas of possible shrinkage [ILLUSTRATION FOR FIGURE 4 OMITTED]. Selee recommended risers at the two areas of shrinkage to combat the problem.

The customer then applied the new design and made 30 castings with zero scrap. The prediction was accurate and the process didn't require any trials with scrapped heats. The use of simulation proved to be much cheaper than pouring any metal.

Prospect Foundry

Since it entered modeling in 1994, Prospect Foundry, a 230-employee ductile and gray iron casting subsidiary of Atchison Casting Corp., has expanded its simulation capabilities in an effort to become proactive. It has fine-tuned its system for iron, molding sand and core materials through simulation, which has helped the foundry achieve more accurate results. The company tries to simulate all new jobs before running them in the shop. Today, Prospect runs approximately 15 jobs/month with a 90% success rate. Further, the company averages 5-6 simulations/part number (about 80 simulations/month).

In 5 years of solidification modeling at Prospect, engineers have had many success stories - from figuring out how to remove the shrink on its toughest jobs to making almost all new jobs shrink-free the first time.

Simulation also has helped in quoting. When trying to explain to a casting buyer why a cover-core is needed to riser into a heavy section, pictures can be sent showing the results with and without the core. This helps the buyer understand why there might be differences in the tooling quotes, and thus feel more confident that parts will be shrink-free.

Another benefit of solidification modeling is the ability to revise the design of the part to make it easier to cast. As an example, engineers had tried to remedy a problem casting in which, despite what was done with the riser and the riser neck, the path was freezing off too soon causing shrink in the castings [ILLUSTRATION FOR FIGURE 5A OMITTED]. Engineers knew that stock had to be added to the pattern to feed into the heavy section [ILLUSTRATION FOR FIGURE 5B OMITTED]. By changing the design of the casting on the computer and running the simulation, the change could be minimized and the casting would be shrink-free. Once Prospect found the change that would work, the customer was contacted, and the results were emailed so the company could see exactly what had been done to make the casting design work. Since this was an area that was already machined, the stock added to reposition the neck was not a problem to remove. Through the use of the solidification software, Prospect showed its customer that the current design could not be cast solid, justifying a design change.

The benefits to simulating the part are:

* starting the job off with fewer defects;

* improving yield and maximizing the number of impressions per mold;

* saving money on sampling costs with fewer tests to run, fewer cut castings, man-hour savings and not as many band saw blades to buy;

* increasing profits by spending less time making samples and more time producing castings.

On tough jobs, when a foundryman is unsure how to part the job or how to orient the cope to the drag, he can simulate it to see what works best before building the pattern.

This article was adapted from a panel presentation at the 1999 AFS Casting Congress.