GRAC puts aluminum filtration to the test and into practice.

The foundry shares the expertise gained from years of experience and careful testing, and details how filtering has cut its scrap rate.

Liquid metal filtration during the casting process has become a common technology in both ferrous and nonferrous casting. When properly implemented, filtration can eliminate exogenous (refractory particles, sand grains, etc.) and indigenous (oxides, slag or dross, etc.) inclusions from the final casting. A well-placed filter can also reduce metal turbulence and its resultant porosity.

Grand Rapids Aluminum Castings (GRAC) understands well the benefits of effective filtration. The 30-employee green sand and permanent mold foundry has been using filters in casting various parts for six years, trying a variety of filter styles and materials before settling on pressed cellular filters for the majority of its parts.

GRAC's success did not come from the simple purchase of the filters, however. Considerable trial and error effort was required, along with significant re-examination of the gating for many patterns. This article discusses GRAC's experiences, and details some of the important operational findings the six years have seen.

Filters and Gating

GRAC casts aluminum automotive parts, including brake housings, gas pipes, and gear blanks and boxes. Many of these parts have rigid quality requirements. The gas pipes, for example, must pass pressure tightness testing to ensure they are absolutely free from porosity, which was a common problem caused by oxide inclusions. The gear boxes receive surface machining after casting, and prior to GRAC's introduction of filtration, inclusions were causing cutting tool damage - to the tune of $300 per [TABULAR DATA FOR TABLE 1 OMITTED] tool. Filters have turned out to be the cure in the case of both of the above castings.

Table 2. Average Pouring Time (sec) with Various Filters and
Without Filter

             Filter A       Filter C       Filter B       No Filter

Part 7880      4.05           4.12           3.65            3.51
Part 6432     12.20          12.70          10.77           10.02

As stated above, successful filtration can require rethinking the gating arrangements on some castings. One of the important principles of gating system modification GRAC officials have discovered is to keep the flow rate coefficient of the gating system close to its original value. The filter setup in the gating system should not significantly change the casting chamber filling procedure or the casting solidification models. While the runner is usually recommended for filter placement, the actual setting often depends on space available in the flask and the convenience to the molding operation.

In this article, Cases 1-5 detail the gating requirements of five representative castings that GRAC pours using pressed cellular filters [ILLUSTRATION FOR FIGURE 1 OMITTED]. In each casting, the pouring temperature is around 1300F (704C), and in each gating arrangement, the choke is located at the bottom of the sprue. When the filter is located in the runner, it performs another important function besides inclusion capture: it reduces the turbulence of the metal flow in the runner and helps to prevent oxide formation within the mold.

Flow Rate and Filtration

The flow rate coefficient of a gating system is an important parameter of its design. It influences mold filling time and possibly casting solidification. When introducing a filter, it is important to keep the flow rate coefficient as close as possible to the original gating system. To explore the influence of the filter on gating flow rate, GRAC conducted tests to measure mold filling time, the main parameter representing the flow rate coefficient of the gating system. For the tests, Parts #6432 and #7880 (see Cases 2 and 5) were poured using three different types of filters, as well as without filters. Pouring temperatures were manipulated to gauge their effect on filling time.

As it turned out, pouring temperature influence on mold filling time is not significant. Decreasing the pouring temperature by 300F (148C) for both parts produced little change in filling time. Under the constant resistance to metal flow in a gating system, melt viscosity is the main determinant of filling time and, unless it approaches the alloy's liquidus temperature, viscosity doesn't change much with melt temperature.

The differences between the three filters are detailed in Table 1. Filters A and B have nearly the same open area percentage (64.8% and 61.8%, respectively), and the same cell diameter (0.091 in.). The only difference between them is the filter area: the port area of A is 1.37 sq in, compared to 1.99 sq in. for B. When replacing A with B, the filter to choke area ratio is increased from 4.9 to 7.5 (53%). Meanwhile, the mold filling time is reduced only slightly: 0.4 sec for part 7880, and 1.4 sec for part 6432, or about 10%. These results imply that the 4.9 area ratio of filter to choke is almost large enough to minimize the influence of filter resistance to metal flow. Therefore, unless the mold filling time is extremely important, it is not usually necessary to further increase the filter area under these conditions.

Filter C, which is square in shape, has a smaller cell diameter (0.059 sq in.) and a smaller open area percentage (51.4%) than the other two filters. But its port area is 1.52 sq in., compared to A (1.37 sq in.). The results (Table 2) show that the mold filling time is almost the same for these two filters. The slight increase in filling time is probably because the square shape of Filter C caused incomplete use of the filter area.

As shown in Table 2, using Filter B only increases mold filling time for both castings by about 0.1 sec, compared to the unfiltered gating systems. Filling time increased slightly more with the other two filters, implying that the filter has two types of resistance to metal flow: the area effect and the friction effect. Using filter B, we see that filter area has no effect on mold filling time, and that the minimal delay in filling is because of the interfacial friction effect of the metal passing through the filter.

Inclusion Capture

To study the inclusions captured by a pressed cellular filter, GRAC carried out scanning electron microscope analysis with an energy dispersive spectrometer. It was found that, although Filter A has a relatively large cell diameter, it still can retain both exogenous and indigenous inclusions, including sand agglomerations and indigenous compounds. Furthermore, the chemical nature of the inclusions found ran a gamut of compounds. Figure 2 shows inclusions captured by the filter, with both large and small size compounds retained at the filter entrance. Interestingly, the inclusions do not seem to be touching the filter and, although they are smaller than the cell diameters, they are still caught at the entrance to the cells.

Figure 3 gives a clearer picture of the mechanism by which these inclusions seem to be captured. There is a "filtration cake" - a mixture of dross and inclusions - formed at the filter entrance. This occurs because the dross, carried along at the front of the metal flow in the gating system, has a generally larger size than most inclusions, and is easily retained at the filter entrance. There, it helps to capture the inclusions borne by the metal behind it. Thus, it is a combination of the filter and the dross it initially traps that effectively separate inclusions from the metal.

Overall Benefits

GRAC assesses the effects of filtration with before and after comparisons of casting scrap rates. As can be seen from the various Cases, filtering has dramatically cut scrap and eliminated defects on castings where structural integrity and machinability are crucial.

The test results presented in this article led GRAC to some interesting conclusions. Continued analysis of the metal on both sides of the filter consistently reveals how well the foundry is operating. The inclusions found trapped before the filter indicate the condition of the furnace, the atmosphere in which the metal was melted and how well the metal was cared for. Analysis of the metal found after the filter shows it is usually clean, regardless of how it looked before the filter.

In light these factors, the foundry three years ago came to the conclusion that a lower cost type of cellular filter works as well for its quality requirements as the more expensive foam filters previously used. GRAC was purchasing foam filters at $1.08 a piece. The pressed cellular filters, however, cost just $0.26 a piece - a tremendous savings considering the foundry uses about 5000 filters per month. The new filters are currently being used in the gating systems, as well as in some of GRAC's direct-poured castings, and are helping to produce the highest quality castings at the lowest cost.

Continued research caused foundry officials to consider again the various theories of why filtration works - from the "filtration cake" detailed in this article to the shredding and scattering of inclusions so their wide distribution doesn't affect the final casting properties. The accepted theory at GRAC is that with the right density ratio in the metal, the denser metal passes through the cells with enough velocity that the less dense inclusions can't get in line to make their own way through the cells. Or, as GRAC personnel are fond of saying: "They can't run with the big dogs, so they stay on the porch."

Case 1.

Part: #7335 (Gas Pipe) Alloy: Al 319 Parts Cast Per Year: 4000-5000 Casting Pour Weight: 11 lb Filter Used: Filter A Scrap Rate Before Filtering: 4.0% Current Scrap Rate: 0.2%

As shown here, part #7335 is poured in an arrangement with two castings in the mold. The filter is vertically seated near the end of the runner, where the runner's cross section is enlarged. A blind riser connects the runner and the two ingates, which are opened from both sides of the riser. The cross sectional areas of gating system components are:

Sprue Bottom Area = 0.44 sq in., Runner Area = 0.84 sq. in., Filter Area = 2.15 sq in. and Ingate Area = 1.86 sq. in.

Hence, it is a nonpressurized system. The recommended ratio of filter area to choke area is 4-6 for aluminum alloys. In this gating system, the filter area is 4.9 times as large as the choke area - squarely in the middle.

Case 2.

Part: #6432 (Gas Pipe) Alloys: Al 319 Parts Cast Per Year: 4000-5000 Pour Weight; 15.3 lb Filter Used: Filter A Scrap Rate Before Filtering: N/A Current Scrap Rate: 0.2%

Like Case 1, Part #6432 is gated for two castings in one mold. The filter is located vertically in the runner area near the blind riser, and the two ingates are opened on the sides of the blind riser. The difference in the gating of this part from Case 1 is that the runner is extended after the blind riser and an open pop off is used at the end of the runner. Again, the filter area to choke area ratio is 4.9

Cases 3-4.

Part: #2526 (Gear Blank) Alloy: Al 319 Parts Cast Per Year: 4000-5000 Pour Weight: 10.76 lb Filter Used: Filter A Scrap Rate Before Filtering: 3.0% Current Scrap Rate: 0.0%

Part: #10075 (Gear Box) Alloy: Al 319 Parts Cast Per Year: 4000-5000 Pour Weight: 13.87 lb Filter Used: Filter A Scrap Rate Before Filtering: 5.0% Current Scrap Rate: 0.0%

The setting of the filters for these two parts are the same: the filter is located horizontally in the sprue, and each mold has four castings and one open riser. The filter has been placed in the sprue due to limited space in the flask. For part #2526, the gating system's cross sectional areas are:

Sprue Area = 0.44 sq in., riser Area = 0.77 sq in. and Ingate Area = 1.6 sq in.

After the runner, a riser is designed in the center of the four castings, and four ingates are open at the four corners of the square riser base.

For #10075, which originally had an oxide problem in the section near the ingate, there are two blind risers in the mold and two separated runners are used to connect the sprue with the blind risers. The gating system area ratio for this part is Sprue: Riser:Ingate = 0.44:0.56:5.25. The choke for both of these castings is downstream from the filter and the choke to filter area ratio is 4.9.

Case 5.

Part: #7880 (Gas Pipe) Alloy: 319 Parts Cast Per Year: 4000-5000 Pour Weight: 5.1 lb Filter Used: Filter A Scrap Rate Before Filtering: 5.0% Current Scrap Rate: 0.5%

Prior to filtering, the common defect encountered in this casting was leaking in the wall due to porosity-forming oxides. In the part's gating system, the filter is located vertically between the sprue base and the runner. After the filter, the runner connects the two castings. This casting has a gating area ratio of Sprue:Runner:Ingate = 3:8:4. The choke is still at the sprue bottom, with an area of 0.44 sq in. The filter area is 4.9 times that of the choke. As the casting is relatively small, no riser was designed for this casting.