Casting troubleshooting via solidification modeling.

A 2-D modeling software program lit the path for a nonferrous Canadian foundry in solving its feeding problems.

Today, solidification is a well-known and marketed tool available to the foundryman. While some believe that in coming years, almost all first-world foundries will be using solidification software, only a minority of foundries are using solidification modeling regularly. In fact, most small and medium-sized foundries do not use modeling at all. It is even a matter of contention whether--today or in 10 years--solidification modeling will be called for in all situations.

The problem is bluntly posed in economic terms: Do the expected savings in trial runs and reduced scrap (additional indirect benefits may also be considered) balance the real cost of solidification modeling for a definite application? For a given casting, this cost depends mainly on the sophistication of the software used. People tend to forget personnel costs are always greater than the hardware/software amortizement.

To place things into perspective, let us say that for the manganese bronze yoke casting in Fig. 1, the cost of a complete solidification modeling study on a scientific two-dimensional (2-D) or 2-1/2-D software is roughly $1000 while a complete 3-D filling and solidification analysis could cost 5-10 times as much.

Unquestionably, more accurate 3-D filling and solidification packages should be used when dealing with mass production of castings in a tightly controlled manufacturing environment like automotive foundries. However, the refinement and accuracy of these simulations would rarely be justified in a jobbing foundry.

This article presents a case study from Fonderie St.-Romuald, a medium-sized nonferrous foundry in the Canadian city of St.-Romuald. Each month, this foundry typically produces 5-10 new patterns as single replacement parts or as 200-500 unit series. These parts are cast in a large variety of copper and aluminum alloys in quality ranging from commercial standards to stringent military specifications requiring 100% radiographic inspection. Under such a variety of conditions, using solidification modeling may be very judicious at times, yet unjustified on a short run for relatively "easy" castings.

Case Study

Over a one-year period, Fonderie St.-Romuald produces about 800 manganese bronze yokes (C86300) of the type shown in Fig. 1. After machining of the shaft, these castings are assembled to other components to form a power transmission switching device.

The original matchplate allowed the filling of four cavities through side risers fed by sprue and runners. Subsequent machining often revealed a porous region at the junction of the shaft with the main body of the casting. This defect was the cause of a 10-15% rejection rate.

Various attempts to solve the problem--such as increasing the riser diameter, lowering the pouring temperature and increasing the size of the riser heel--did not substantially improve the situation.

Since this 5-lb net weight part was "mass produced" in the context of Fonderie St.-Romuald, foundry officials invested time and energy into a solidification modeling study of the casting.

AFSolid

Although truly 2-D, the 2-1/2 AFSolid Software allows foundrymen to take into account the "depth" of a given section of a 2-D drawing by assigning a "thickness factor" to each section (depending on its cooling modulus). This simplified treatment, pioneered in the early days of solidification modeling, applies a thickness factor of 0.5 to a cylindrical section when a factor of 1 is assigned to a section of great depth (such as an extrusion).

As indicated in Fig. 4, this program assigns different thermal properties to the casting (dark blue), olivine green sand (light blue), silica core (green) and riser insulating sleeve (purple). Several situations were performed for different liquid metal initial temperatures, riser sizes and critical solid fraction (such as the solid fraction of the mushy zone above which no feeding can take place). In all cases, a hot spot near the casting's shaft/fork junction was revealed on the side of the riser.

Figure 5 vividly shows this hot spot for an initial temperature of 1940F (1060C) and a critical solid fraction of 0.5. When a similar procedure was repeated with a top riser, directional solidification was maintained in the critical zone over the entire solidification process.

It must be emphasized that the initial temperatures of the mold and casting were assumed to be uniform and symmetrical at the end of filling (such as at the onset of the simulation). That is, the thermal asymmetry introduced by the side gating prior to complete filling was not accounted for.

Effect of Side Gating

Using a side gate in the filling process introduces a thermal asymmetry in the liquid metal (mainly in the mold materials). In spite of the convective mixing, temperature gradients exist in the liquid metal at the end of the filling process. These gradients are almost impossible to evaluate without a solution to the fluid flow problem, which is only available on sophisticated 3-D packages.

However, early overheating of the mold materials near the ingate will have the stronger impact on solidification asymmetry, especially for long filling times. This effect may easily be taken into account by applying the hot metal to the entrance section of the mold/core cavity for a time corresponding to the filling process, then saving the temperature field in the mold materials and using it as an initial condition to the final simulation.

When treated in this manner for a filling time of eight seconds, the solidification pattern of Fig. 7 was obtained, indicating the risk of a shrinkage is still present if the casting is gated from the side.

Results

The most favorable rigging of the yoke casting involves filling and feeding from the shaft section of the casting, where initial temperature gradients in the liquid (not accounted for in the simulation) will enhance even more the directional solidification exhibited in Fig. 6.

Although foundrymen might be appalled by top-pouring this type of alloy, premium-quality castings of highly oxidizable alloys have been consistently produced when proper flow control and filtration are provided at the riser-casting junction.

X-ray analysis confirmed this solution, revealing no shrinkage occurred when the casting was filled and fed from the casting's axis of symmetry.

Limits & Possibilities

The success mentioned above shouldn't obscure the fact that solidification modeling is not a panacea--it can only address problems related to feeding. Unfortunately, many other problems exist in the foundry. Also, setting up a faithful model of the physical situation is more difficult than it seems.

Besides a proper knowledge of the foundry process, personnel experienced in solidification modeling is necessary to obtain reliable results. This is why it's difficult for a foundry engineer to approach solidification modeling as just another task to tack on to his/her day-to-day labors. Using external services might be a viable avenue, especially for smaller foundries, as long as modeling is handled by a competent foundryman.

Using solidification modeling in the same way as a calculator is not only useless but sometimes very misleading. However, when exploited in a proper manner, solidification modeling will:

* reduce scrap and improve scheduling:

* optimize cost to quality ratio;

* help question erroneous longstanding beliefs;

* improve the foundryman's credibility and boost motivation;

* help designers and patternmakers come up with casting geometries adapted to the foundry process.