Printer Friendly

Optimizing casting yield using computer simulation.

Foundry engineers tested the usefulness of casting simulation by producing a steel casting with a higher safety margin, less turbulent filling and improved yield.

Casting simulation software programs use thermodynamics and physics to mathematically describe the filling and solidification of metal in a mold cavity. Over the past few years, the thermophysical data for metals, molding materials and feeding systems has improved significantly, thus resulting in accurate predictions of feeding characteristics and casting soundness. This translates to better quality, less scrap and more profit for foundries.

To test the capabilities of computer simulation, a team of foundry engineers attempted to make accurate predictions of filling and feeding performance in both a conventional feeding system and modified system (sleeve with integral filter). Using a first principles computer simulation program, the engineers set out to evaluate the filling characteristics of both methods and to evaluate the performance of the feeding systems.

Filling predictions showed that the modified unit would fill the casting evenly, without turbulence, while the conventional gating system was predicted to be very turbulent and likely to produce reoxidation inclusions. Even though feeding predictions between the two configurations were similar, the modified design was predicted to have a slightly higher safety margin. Reduced turbulence combined with integral filtration means that application of the modified design would result in improvements in casting quality.

Several castings were produced to compare the two methods, and the actual filling and solidification results agreed with computer prediction. Casting appearance, quality and feed safety margin were improved, and significant cost savings were realized as casting yield was improved by 11%. The tests have resulted in an accurate set of thermophysical data for metal, molding material, feeding systems and heat transfer coefficients.


The casting that engineers studied was a two-on adapter casting poured with low-carbon alloy steel at approximately 2900F (1593C). The molding material is green sand and the cores are ester silicate-bonded sand.

Filling and solidification computer predictions were made for two configurations: a conventional gating system and a modified pouring unit on the original feeder basin that replaced the downsprue and gating system. Engineers applied a Magmasoft simulation program that uses a finite volume meshing scheme, runs on a Unix-based workstation, and employs Navier-Stokes flow solvers to describe the filling characteristics of the system at any given point in time. The software's k-[Epsilon] equations describe flow turbulence, and for solidification, Fourier heat transforms describe the solidification profile as a function of time.

The conventional gating system consists of a downsprue with two separate ingates, one for each casting in the mold, and a single 3 x 6 in.-thick exothermic feeding system - a riser sleeve topped with an exothermic/insulating hot topping compound. The finished casting weighed approximately 20 lb, and the total pour weight for each mold was approximately 89 lb. This included 20 lb of metal in the gating system, 29 lb in the riser and base and 40 lb of castings. The overall casting yield was approximately 45%.

The second design employs a 3 x 6-in. direct-pouring unit. The process combines feeding systems technology with a ceramic foam filter that reduces turbulence and metal reoxidation within the mold during filling. The total pour weight for each mold with the modified gating system was 71 lb. This includes 31 lb of metal in the riser and base and 40 lb of castings. The overall casting yield was approximately 56%, an 11% improvement over the conventional gating system.



The improvements in filling characteristics of the modified design, as compared to the conventional gating system, are dramatic. Both designs were poured in approximately 5 sec.

Figure 1a shows the conventional gating design at 1 sec into pouring at a cross-section through the adapter. The gray shades represent metal temperature, with the lighter shades representing the hottest temperatures. The black lines are the flow vectors and indicate metal flow velocity and direction. Clear areas represent unfilled sections.

After 1 sec, the runner bar is nearly full of metal, and metal has just begun to enter the casting cavity. The velocity in the runner bar is high, and a significant amount of the metal in the runner bar encounters a wall and must turn 90 [degrees] before entering the ingate. A 90 [degrees] flow change typically results in a significant amount of turbulence and unsteady flow behavior, and this is evident at the casting ingate. It is likely that reoxidation inclusions are formed at this point.

Figure 1b is an X-ray view of the casting after 2 sec of filling. At this point, the runner bar is full and is operating as designed. Formation of inclusions has likely been reduced or eliminated, however, the metal entering the casting cavity is not filling the casting uniformly. The metal flows to the point socket end of the casting, deflects off of the mold wall and enters the small flange part of the casting cavity via this indirect path. Turbulence and rapid heat loss are both present in the small flange portion of the casting. In addition, metal is beginning to spill into the riser well and cool rapidly.

Figure 1c shows X-ray views of the casting after 3 sec of filling. The flow rate is still very high and is still turbulent, especially where the flow directly impinges on the mold wall and is forced to splay 90 [degrees]. Notice that 60% of the fill time has elapsed, and a significant portion of the casting cavity is still unfilled. Worse still, sand inclusions due to metal impinging on and washing the mold wall are likely.

Figure 1d shows the filling profile after 4 sec, and the flow is finally becoming more stable and beginning to evenly fill the casting. However, note that the casting still has not been filled and that the riser is being filled before the casting is completely full. Any inclusions generated in the filling process would likely remain in the casting, rather than progress to the riser and riser basin.

By contrast, the modified filling profiles are well-behaved. Figure 2a is an X-ray view of the casting with the modified design after 1 sec of filling. Metal is shown filling the casting cavity immediately. There appears to be some initial splashing of the metal due to the momentum effects of the metal stream after it enters the riser well and riser neck. This is because the break-off riser necks from the conventional method are used as ingates in the modified method. Nonetheless, this minor flow irregularity does not affect the quality or integrity of the casting.

Figure 2b shows an X-ray view of the casting after 2 sec of fill time. The flow vectors are small and point upward, indicating that the casting is filling evenly from the bottom. Figure 2c shows the casting after 3 sec, and again the casting is filling evenly and quietly. Turbulence is not present. Finally, Fig. 2d shows the casting after 4 sec of filling, and again, the fill profile is well-behaved. In addition, the casting cavity is nearly filled after 4 sec; thus, the last second of fill time is devoted to filling the riser. This is the optimal filling scenario, because the hottest metal ends up in the riser.

Because of the inherent filtration benefits and the observed filling benefits, the computer results predicted that the modified castings would be produced with significantly less inclusions than conventionally gated castings.


The improvements in solidification and feeding characteristics are subtle but significant. The temperature distributions between the two designs are similar because the castings are poured in only 5 sec. However, the modified design shows a distinct advantage because the hottest temperature in the casting at the end of filling is at the riser neck contact. Compare this with the conventionally gated castings, in which the hottest temperature occurs at the end of filling in the ingate, which is the furthest point from the riser. This is due to the superheating effects of the ingates.

In an X-ray comparison of the casting solidification time for both the conventional gating and the modified designs [ILLUSTRATION FOR FIGURE 3 OMITTED], because the castings are small and the filling time is relatively short, the solidification profiles between the two designs are similar. As in the previous comparison, the differences are subtle, but significant. Clear areas solidified within 200 sec, while shaded areas solidified between 200 and 700 sec. The light-gray areas of the casting solidified last.

In both cases, the last place to solidify in the casting was the riser neck contact. However, the solidification gradient from the tip of the casting to the riser contact is slightly steeper and does not extend as far toward the tip of the casting for the modified design. For these reasons, the modified method has a more favorable solidification profile than the conventionally gated method.

Because the modified method introduces hot metal directly into the mold cavity, metal heat loss associated with conventional gating systems is avoided. As a result, the adapter castings made with the modified method are less prone to cold metal-related defects.

Additionally, the positive effects of this specific configuration could be dramatic for larger castings of this type and configuration, which would likely benefit from longer pouring and solidification times.


The castings are predicted to be sound for both configurations. Figure 4 is an X-ray comparison of macroporosity or feeding predictions. Clear areas are predicted to be sound, while the feed pipe is defined by the shaded areas.

The 3 x 6-in. direct-pouring unit is predicted to have a feed safety margin of 25.6%, while the 3 x 6 in.-thick exothermic has a predicted feed safety margin of 12.5%. Therefore, the modified casting has an 13.1% additional safety margin.



Castings were produced using both of the above configurations, and some differences in appearance were noted. Occasionally, castings produced with the original gating system had small inclusions at the extreme ends of the castings. Also, cold metal lapping is a frequent defect.

Due to the limited number of castings produced and the effective conventional gating design, a large difference in the inclusion levels in the castings was not established. However, it is anticipated that casting quality is improved in terms of fewer small inclusions and better surface finish due to improved filling for the modified castings.

Feeding Comparison

Risers from each configuration were sectioned to determine the actual feeding results. The modified process has more feed safety margin than the conventional process. In addition, computer results using accurate metal, molding material and feeding system thermophysical data accurately predicted actual feed safety margin results.

When comparing the predicted feed safety margin and the actual sectioned feeder for the 3 x 6-in. thick exothermic riser used in the conventional gating system, the computer predicted 12.5%, while the actual feed safety margin was 15%. This is excellent agreement and verifies that the computer program and the database for metal, molding material and feeding systems is accurate. The accuracy is especially important because the feeding system data is for a complex, exothermic sleeve.
Table 1. Comparison of Gating Method Production for a 5000-lb Ladle

Approximate 5000 lb ladle Existing Practice Direct

Number of casting molds in heat 58 73
Number of castings in heat 116 146
Shipped weight of single casting 20.05 lb 20.05 lb
Total shipped weight of castings 2326 lb 2927 lb
Weight of metal in riser and base 28.85 lb 31.10 lb
Weight of gate 20.30 lb 0 lb
Pour weight of mold 89.25 lb 71.20 lb
Ladle heat weight 5037 lb 5022 lb
Charge weight (H/0.94[at]6.0% loss) 5358 lb 5343 lb

Figure 5 is a comparison of the predicted feed safety margin and the actual sectioned feeder for the modified pour unit. The computer predicted 25.6%, while the actual feed safety margin was 26.3%. Again, this is excellent agreement.

Finally, comparisons between the two sectioned feeders shows that the modified casting has 11.3% more feed safety margin than the conventionally risered configuration.


The modified system appears to provide significant process and yield advantages for this casting, and other castings with similar configurations. Table 1 shows the basic differences between the two methods. Assuming a 5000-lb ladle and a full heat devoted to this casting, the 116 castings could be produced using the conventional method (58 molds) and 146 using the modified process (73 molds). The total shipped weight of castings would improve from 2326 lb to 2927 lb. Casting melt productivity could be improved approximately 26% by converting to this method. This means that if eight furnace heats are required for a particular order of this casting with the conventional gating system, only six are required to produce the exact same number of castings using the modified method.

Also, a sand and time savings of approximately 40% is projected. Elimination of the conventional gating system should allow for a reduction in sand-to-metal ratio, and, possibly, the ability to increase the number of castings per mold from two to four.

Finally, the cleaning room should experience a large reduction in upgrade hours with the modified casting. Castings made using this method will have reduced mold, ladle and reoxidation inclusions. Additional savings will be realized through the elimination of the gate contacts.

This article was adapted from a paper (99-144) presented at the 1999 AFS Casting Congress and is available from the AFS Library at 800/537-4237.
COPYRIGHT 1999 American Foundry Society, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1999, Gale Group. All rights reserved. Gale Group is a Thomson Corporation Company.

Article Details
Printer friendly Cite/link Email Feedback
Author:Outten, John F.
Publication:Modern Casting
Article Type:Cover Story
Geographic Code:1USA
Date:May 1, 1999
Previous Article:Simulation software in action: five users share their experiences.
Next Article:1999 casting simulation software survey.

Related Articles
Numerical modeling of castings in the production process.
S.E. foundrymen position themselves for the future.
Iron casters discuss keys to optimizing inoculation.
Simulation software in action: five users share their experiences.
1999 casting simulation software survey.
Simulation Case Histories: Solving Problems while Optimizing Processes.
Eliminating Modeling 'Trial & Error' with Casting Process Optimization.
Computer model diecasting shot sleeves to predict distortion.
Casting process modeling software for windows opens door for diecasters.

Terms of use | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters