VCT produces solutions for solid castings.

Inside This Story:

* Virtual casting technology (VCT) connects modeling simulation with casting design teams to reach the root cause of a casting defect.

* VCT methods allowed casting engineers to eliminate component errors and improve productivity more efficiently than common shop floor methods.

* Using such technology can give metalcasters the competitive edge.

As more casting suppliers compete internationally, there has been a growing pressure for them to produce components of higher quality at lower costs. One of the most efficient methods to maintain competitiveness has been casting process modeling simulation in the casting design stages. These technologies have offset costs due to a significant reduction in design lead times and scrap rates and increased design flexibility. Developments in casting simulation coupled with numerical optimization techniques have demonstrated a reduction in design leadtime up to 15% with productivity improvements up to 80%.

Despite this trend, the traditional trial-and-error approach based on metalcasting experience is still the most widely used strategy in terms of process troubleshooting. The risks of new "outside the box" process design ideas are often too high for a facility to take, and such firms go with quick "on the fly" methods on the shop floor to solve problems. As a result, modeling simulation has been hindered by insufficient computing resources and technical expertise and a lack of initial conditions that accurately translate process settings into a casting simulation.

However mathematical and physical modeling have demonstrated new approaches to reduce this risk through reliable visualizations of these ideas.

Increases in computing power, expertise and the understanding of scientific-based problem solving in recent years have made casting simulation more accepted and utilized in traditional new product and ongoing process development stages. Since the initial development of a patented virtual casting process design methodology, ongoing research efforts at Argent Metals Technology (AMT), Auckland, New Zealand, have focused on refining output accuracy of commercial simulation packages to enable a faster and more robust process development. This has been accomplished through the firm's virtual casting technology (VCT).

Virtual casting technology acts as a superset of commercial casting packages as it incorporates any simulation code and adds a critical problem-solving methodology and interfacing code for:

* optimization of tooling or process modifications in the model in order to achieve a practical mapping to a real tool for a solution.

* calibration of accurate simulations (boundary conditions, initial conditions, etc.);

The interconnectivity of this with casting engineering and tooling engineering groups in metalcasting facilities is essential to accelerate the deployment of a solution to achieve benefits faster than current "remote simulation service" practices commonly used by many firms and even used within different parts of larger firms.

VCT allows the design and casting team to go beyond just simulating the issue and understanding sensitivities of the modeling. It also can provide casting engineers an opportunity to formulate a solution with a higher level of confidence and reasoning during troubleshooting.

Recently, AMT compiled a cross-functional team comprising simulation engineers, metallurgists, casting engineers and production machine operators to utilize VCT methods and examine a casting experiencing high levels of scrap. The interconnectivity of VCT with the team allowed the engineers and designers to accelerate their solution development to achieve solutions faster than current "remote simulation service" practices that often are used in many casting facilities.

This article addresses the methodology the AMT team used to tackle the porosity issue by combining the results of the simulated modeling with real castings and implementing them into modified designs to improve productivity.

Finding the Defect

The casting investigated was a motorcycle wheel casting measuring 18 in. (45.72 cm) in diameter and 5.5 in. (13.97 cm) thick. This component was permanent mold cast with A356 aluminum alloy poured at 1,292F (700C). It had an as-cast weight of 26.45 lbs. (12 kg) and weighed 19.73 lbs. (8.95 kg) after finishing. However, during the development of the component, defects were discovered on nearly half of the castings.

Of the total 75 castings produced (according to the standard original process for this particular casting run), 32 castings were rejected after machining. High levels (up to 43%) of shrinkage porosity at the upper rim of the casting were revealed after the machining process. These defects not only ruin the cosmetic appearance of the wheel, but they also are detrimental to the structural integrity.

The location of the shrinkage porosity was found directly underneath the feeder. The initial suggestion was that it could be described as a typical under-feeder shrinkage. Under-feeder shrinkage often is observed in the casting where an oversized feeder is present and the area of casting directly underneath the feeder is overheated. A simple modulus calculation indicated the modulus ratio of this casting was at 1.39, which is well above a typical aluminum alloy ratio of 1.2. This supported the initial suggestion of underfeeder shrinkage.

However, a notable trend was revealed when the total number of defects and their individual locations from those castings were recorded (Fig. 1). An imbalance was found as the first two front ingates of the casting (regions C and I) and their adjacent areas (B, D and H, J) revealed significantly higher defect rates than the two rear ingate areas (E and F).

[FIGURE 1 OMITTED]

This immediately raised a question of whether the only contributing factor for this pattern was oversized risers. If so, it was more logical to observe evenly distributed defects because the feeder geometry was uniform all around the casting. The fact that the distribution of the defect was concentrated near particular ingates indicated that filling characteristics contributed a major role here, as well.

The investigations looked to solve the defect problem by implementing the VCT methods and developing an outline (Table 1) that integrated simulation analysis and metalcasting experience. By following this template, AMT strove to attain an optimal solution to the casting problem by utilizing the VCT strategy to determine solidification and fluid flow properties.

Simulated Solidification

Models were carefully set up applying all the necessary process information to accurately simulate the current casting process, and the simulation was run for 10 cycles.

A temperature snapshot of a casting at its 10th cycle is shown in Fig. 2. The temperature distribution at the upper rim shows region C remains substantially hotter than region A and E. Ingates were placed at both region C and E; however, the thermal analysis of the casting showed the temperature of region E, where the second ingate is located, was the same level as region A, where there was no ingate.

[FIGURE 2 OMITTED]

Solidification analysis revealed the overall solidification time of region C as much slower than the rest, while region E had the fastest solidification time (Fig. 3). When comparing these results with those of Fig. 1, it appeared excessive temperature buildup in region C could have caused a high level of shrinkage porosity. This also supports the initial suggestion that shrinkage porosity in that region was due to "underfeeder" shrinkage.

[FIGURE 3 OMITTED]

Validating Solidification

To validate this simulation analysis, secondary dendrite arm spacing (SDAS) of an actual casting wheel was measured at the same locations (Fig. 4) because there have been known relationships between SDAS and cooling rate. In this study; the average SDAS in regions 1-5 was measured in castings and compared to the simulated solidification time to verify the simulated trends only.

[FIGURE 4 OMITTED]

It was found that the overall SDAS of region C was up to 23% larger than that of region E; hence, region C had a longer solidification time (Fig. 5). This proved that the simulation coincided with the SDAS measurements and confirmed region C had the longest solidification time while region D had the shortest solidification time. The high level of shrinkage porosity in region C was due to insufficient dissipation of heat in that region, but the root cause of this still had yet to be identified.

[FIGURE 5 OMITTED]

Additionally the analysis failed to explain why both region C and E had similar ingates yet showed a significant difference in solidification time. Therefore the attention of the analysis was shifted to fluid flow to investigate the extent of effects during cavity filling to shrinkage porosity.

Analyzing Flow

When pattern filling at different ingate locations was analyzed, the molten metal most often filled through ingate 1 at region C. The filling analysis showed the metal reached the first ingate through the runner bar and started flowing into the cavity at 1.7 sec. after first pouring. It was another 1.5 sec. until the metal reached the second ingate, while up to 30% of the cavity volume had flowed through the first ingate.

Another fluid flow analysis revealed the preferential filling path was due to a pressurized flow of metal in the runner bar. A pressure plot of metal in the runner bar in Fig. 6 showed that as metal flowed in the runner bar from the pouring cup, it was pressurized due to the restrictive geometry of the runner bar.

[FIGURE 6 OMITTED]

As the pressurized stream metal reached the first ingate at region C, the ingate then became an additional open channel for a pressurized metal stream to flow. This preferential filling pattern resulted in more mass flow through ingate 1.

Consequently, more mass through this part of the die provided more heat input into the die; hence, the surface temperature at ingate 1 was higher than that of ingate 2. A snap shot of simulated die surface temperature at 30 sec. into the 10th cycle is shown in Fig. 7.

[FIGURE 7 OMITTED]

The die surface temperature of the region adjacent to the ingate, which correspondents to the upper rim of the casting, was 86-104F (30-40C) higher at ingate 1 than ingate 2. This additional mass flow through ingate 1 explains the slower solidification time of region C and differences in solidification time between region C and E.

Validating Flow

This verification of simulated solidification results can be done relatively easily by SDAS measurement. However, it is often the verification of cavity filling simulation that poses a challenge because it deals with molten metal at very high temperature in a sealed cavity.

A simple and efficient way to capture a snapshot of a real casting cavity filling can be done by pouring a small volume of metal into a cold cavity to rapidly freeze the metal as it flowed into the cavity. Although this method does not provide full visualization of entire cavity filling, a snap shot at various stages can be achieved. In this instance, the permanent mold was cooled down to 140F (60C) to promote rapid freezing, and a small volume of metal was poured. After several attempts with different volumes of metal in the pouring ladle, the correct portion of the cavity filling was achieved and compared to the simulated filling pattern in Fig. 8. This snapshot of real cavity filling visibly confirmed the simulated filling pattern at the ingates by showing that a vast majority of metal flowed through ingate 1 as predicted in the fluid flow analysis.

[FIGURE 8 OMITTED]

Putting to Practice

One of the best solutions to prevent the excessive flow into ingate 1 was found to be modifications to the gating and runner system to provide a more balanced filling pattern. However, due to the constriction in mold space, this solution was not achievable unless a new mold was manufactured. Therefore, additional cooling circuits directly targeting the hot spot in region C were designed and simulated to confirm the effectiveness of the modification. The result showed additional cooling on region C was effective to reduce the heat content in that area, which allowed for a reduced solidification time while having a minimal effect on solidification time of the other areas (Fig. 9).

[FIGURE 9 OMITTED]

This methodology was brought to a small casting trial to validate the effectiveness of the modification. It was found that only two out of 60 castings were rejected during this trial compared to 32 reject castings out of 75 in the previous casting run without the additional cooling circuit. Despite the relatively small number of samples involved in the trial, the effectiveness of the modified cooling circuit demonstrated a reduction in the hot spot in regions C and I and a reduction of casting defects from 42.7% to 6.7%.

Since these investigations were conducted, several more automotive castings have been examined utilizing the VCT methods with advanced casting process development systems to achieve the optimal output of the components' production. Future launch projects using the VCT practice also are underway, including evaluations of common production costs, such as with casting machine time. This will improve savings and allow facilities to gain the competitive edge.

This article was adapted from a paper (05-068) presented at CastExpo '05 in St. Louis.

For More Information

"Improvement of the Low--Pressure Permanent Mold Pressurization Sequence for Wheel Casting," D. Singh, J.R. Lee and Z.W. Chen. American Foundry Society Transactions (03-015).

"Optimizing Cycle Time and/or Casting Quality in the Making of Cast Metal Products," D. Singh, et al., U.S. Patent 5,298,898 (2001).

Table 1. A Problem-Solving Strategy Using VCT.

1. Calibration (Reality Check of Model)

   a. Simulation of the current process using process data and
      thermocouple history.

   b. Solidification and fluid flow simulation to identify the
      possible cause of the the problem.

   c. Solidification and filling results are validated to confirm the
      identified cause is correct.

2. Optimization (Performance Upgrade)

   a. Modifications to tooling design considering objective functions
      (such as directional solidification and cycle time) and
      constraint functions (such as restrictions in mold geometry and
      process environment).

   b. Mapping of virtual tooling material, shape, insulation and
      cooling circuit times to an actual permanent mold.

Dr. Darius Singh is the research and technology manager, Rafael Lee is a simulation engineer and Charles Klein is a casting technical specialist for Argent Metals Technology NZ Ltd., Auckland, New Zealand.