Ceramic foam filtration: no longer a band-aid.

Inside This Story:

* By switching to a 15-mm thick foam filter in 15 PPI, a French foundry reduced the filtration area of a crankshaft by 15%, thickness by 16% and filtration costs by 25%.

* After the introduction of the redesigned gating system to include a foam filter, the scrap rates for a shell-molded crankshaft at a German foundry was reduced to less than 0.1%.

* Through simulation modeling, a French foundry discovered the important impact that the gating system design can have on finished casting properties.

In the 1980s, the use of ceramic foam filtration was thought of as an emergency solution to be used only under exceptional circumstances. It may serve as a temporary fix, but was not regarded as a permanent solution. Because of advancements in the past two decades, ceramic foam filters have been transformed from band-aids into state-of-the-art devices that support the manufacturing of high-quality automotive and engineering components at optimum cost.

Part of the reason for the transformation is because the metalcasting industry has undergone a shift in philosophy. While the idea used to be that inspection and testing would be enough to ensure a quality product, the new thought is that quality must be built into the component.

But because casting processes are inherently full of variables, the need for techniques to combat inconsistency is essential. Automated monitoring and adjustment of many process steps can counter the variables. Or, fail-safe security features can be designed into the casting process--that's where filters come in.

This article outlines some casting production cases, showing how yield, quality and productivity were improved in European foundries through the use of filtration.

FRANCAISE de MECANIQUE, DOUVRIN, NORTHERN FRANCE

The crankshaft production process

The production of pearlitic ductile iron crankshafts began at the firm's foundry in Douvrin, France, in 1970 by using a cluster of four castings (Fig. 1). The original production technique incorporated a sand riser and spinner with the sodium silicate process and has continually improved over the past 20 years. The improvements were made by doing exploratory trials with 50 x 100 x 22 in. ceramic foam filters in 10 PPI, followed by larger scale industrial testing on one crankshaft type.

[FIGURE 1 OMITTED]

The mold is produced from four 90-mm thick, double-faced jacket sub molds which are then assembled to create a single mold with six castings (Fig. 2). After assembly, the mold is turned on its end and poured vertically from the bottom to ensure minimum turbulence.

[FIGURE 2 OMITTED]

But in 1990, the inability to reduce casting prices faster than competitors proved to be a major problem. During this period, investing in, purchasing or modifying production tools had little chance of success. Productivity improvements were seen as the only way to deal with the situation. However, for this approach to succeed, the riser head and the gating system spinner that had been successful for the past 20 years had to be removed. So, from 1995-98, the challenge was taken up to move from four to six cavities in the mold.

A cluster of six crankshafts ranging from the lightest to the heaviest types was produced in the existing 500 x 700-mm mold box. Both the spinner and the riser were successfully eliminated, and the application of two ceramic foam [65.sup.2]x 18 in. filters in 10 PPI (Fig. 3) enabled rapid mold filling with low turbulence.

[FIGURE 3 OMITTED]

Following an extended evaluation program, these filters were replaced with two 15-mm thick, [60.sup.2] filters in 20 PPI. The switch reduced the filtration area by 15%, thickness by 16% and filtration costs by 25%.

On the heels of the project's success, a new challenge was presented--to use the same production technique to produce eight cavities in the same mold without risers. The new project also proved successful, and since 2001, the eight-cavity layout has progressively replaced the family of six-cavity layouts. This resulted in a doubling of productivity at a cost of no more than a new set of pattern equipment.

HALBERG, SAARBRUCKEN, GERMANY

Pushed to the Limit

The casting of a shell molded crankshaft for a Volkswagen five-cylinder engine proved to be a tricky one for Halberg. The casting is designed to reduce weight. But in order to have the required torsion fatigue properties, it had to be of the highest qualities and completely defect-free. The stringent requirements pushed the ductile iron crankshaft and the foundry to new performance limits.

Due to the complex geometry of the machined crankshaft (Fig. 4), pairs of castings were poured upright in a vertically-parted shell mold supported by metal shot. The gating system used is illustrated in Fig. 5. The basic layout design had been successfully applied to other crankshafts produced with the foam filters, which were positioned at the base of the downsprue just after the junction with the runner to avoid direct impact by the first metal. The system also included a slag flotation chamber above the filter lace to keep it free of coarse slag.

[FIGURES 4-5 OMITTED]

Although the material properties were within specification, some castings failed to pass torsion fatigue testing. Examination of the failed castings revealed inclusions in the area of fracture just on or under the casting surface.

Engineers considered the possibility that the combination of high ferrostatic pressure and iron velocity were forcing slag through the filter, affecting 60% of the castings. The slag was sticking to the mold surface and became incorporated into the casting skin. However, the tests showed skin-like layers of reaction product that were too big to have been pressed through the filter. It seemed more likely they were being formed behind the filter.

Multiple scanning electron microscope analyses confirmed the presence of complex oxide inclusions. With its 10 counterweights, the new design had a significantly larger surface area, which increased the access of oxygen to the molten metal during mold filling.

Numerous melting and process changes failed to solve the problem so it was decided to concentrate on investigating the characteristics associated with mold filling. A cut away mold was photographed during filling, revealing significant turbulence and high velocity in the area where the defects were occurring. The mold filling sequence was studied with a metal flow simulation program using the actual gating system dimensions with a downspure to runner to ingate ratio of 1:1.5:0.72. The simulation confirmed the existence of a high ingate exit velocity during mold filling.

The gating system dimensions were modified so that the ratio of downsprue to runner to ingate became 1:1.06:1.15. By changing the ratios, the high initial ingate velocity was significantly reduced. The pattern plate was modified and the mold fill sequence of a sectioned mold was again visually observed. The filling pattern was now quieter and less turbulent. A test batch of castings was produced and all of them passed fatigue testing.

Before the investigation, up to 20% of the castings were being scrapped. After the introduction of the redesigned gating system, the scrap rate was reduced to 0.1%. However, such a gating system ratio could not have been applied without the security provided by the use of foam filtration, which ensured that reaction products and slag were securely retained by the filter. The technical conclusion to be drawn from this work is that without proper attention to other production parameters, it is possible that the maximum benefits of foam filtration may not be obtained.

FRENCH DUCTILE IRON FACILITY

A Finely-Tuned Operation

This French metalcasting facility ships 70,000 tons/yr. of finished ductile iron castings, most of which are safety critical automotive components. This firm has steadily improved its standards of ductile iron melting, metallurgical control and general process cleanliness so it was able to apply more finely structured foam filters to virtually all its castings and further reduce scrap levels.

Exploratory trials by the foundry had indicated the feasibility of applying filters with a finer structure to its production, but before full implementation, a one-month evaluation trial was carried out to assess the viability of doing it without increasing other mold filling defects. Table 1 shows the foundry's scrap levels in the eight previous months when using 10 PPI 18-mm thick foam filters, compared with scrap levels resulting from the trial use of 20 PPI 15-mm thick foam filters. No negative effects were found. In fact, scrap and costs were reduced while performance improved.

But some problems did arise while attempting to produce as-cast ferritic ductile steering knuckles at a lower Si level. The foundry was using the 2.8% nominal Si level but wanted to get down to 2.4%. At the higher Si level, uneven hardness was present because of the hardening effect on the ferrite matrix. The layout of the gating system is illustrated in Fig. 6 with 12 castings arranged randomly to maximize the use of the pattern plate.

[FIGURE 6 OMITTED]

Microstructural evaluation confirmed that some castings had higher levels of residual pearlite. These affected castings could be linked to specific positions in the cluster, leading to the conclusion that the pattern of silicon distribution could be linked to the way in which the cavities filled. A mold fill simulation was performed, giving a clear indication of uneven mold filling.

The gating system was redesigned in the area in front of the filter to reduce the bulk of iron adjacent to three cavities. This caused them to cool more slowly and remain above the critical transformation temperature upon shakeout (Fig. 7). As a result of the revised design, the size of both the filter and chamber could be reduced. The filters also were moved closer to the downsprue to reduce the heat input into the nearest casting, which had the lowest nodule count. The system was changed from ingate-controlled with the runner in the cope and ingates in the drag to a downsprue-controlled system with the complete runner system located in the drag and ingates off the top of the runner. The system also was opened up to correspond with a gating ratio of downsprue to runner to ingate of 1:1.1:1.2.

[FIGURE 7 OMITTED]

Another mold fill simulation showed significant improvement in the mold fill pattern, which was a result of careful proportioning of the gating system cross-sections.

Although the starting point was to solve a metallurgical problem, the use of simulation enabled more features of casting production to be examined and understood, leading to a greater appreciation of the effect of a number of process variables on finished casting properties. The specific conclusions that can be drawn from this work are the important impact that the gating system design can have on finished casting properties. And equally as important, it was the use of filtration to safely and reliably solve the slag retention problem that gave the technical freedom to change the gating system design and solve the problem.

OPENING UP WITH FILTRATION

While it may be ceramic foam filters' most practical use, they can do more than simply stop the flow of unwanted inclusion materials into the casting. The entire casting process can be improved as well.

The filters allow metalcasters to avoid metal damage and cut down on the number of defective zones in a casting. In the pouring process, ceramic loam filters act as devices to close off the cast component. But to the metalcasting process as a whole, they can have the opposite effect. When combined with the optimization of pattern layouts, ceramic foam filters can open up metalcasters to a wider range of gating system designs.

Table 1. Evaluation of the effect of scrap levels under
different filters.

                          % Scrap due to inclusions

Ductile iron           8 previous months   1 month trial
component                 with 10 PPI       with 20 PPI
                             18 mm             15 mm

Suspension arm R             1.09              0.81
Suspension arm L             1.75              2.10
Suspension arm R             2.85              1.71
Suspension arm L             1.66              1.76
Bearing                      5.36              1.97
Bearing                      3.18              1.59
Differential housing         0.77              0.47
Hub                          1.59              0.06

                            % Mis-run and marking

Ductile iron           8 previous months   1 month trial
component                 with 10 PPI       with 20 PPI
                             18 mm             15 mm

Suspension arm R             0.38              0.16
Suspension arm L             0.81              0.40
Suspension arm R             0.81              1.86
Suspension arm L             0.73              1.18
Bearing                      2.18              0.80
Bearing                      2.30              0.72
Differential housing         0.62              0.53
Hub                          0.27              0.00

For More Information

"Improving Productivity in Ductile Iron Castings--Some Ideas to Improve Casting Yield," S. Kannan and N.T.R. Downes, 2003 AFS Transactions paper No. 03-081.

Kenneth C. Taylor has more than 40 years of foundry experience and has been associated with ferrous filter application since 1990. He currently serves as the International Marketing and Technology Manager for Foseco Europe.