What causes electromagnetic pump clogging problems?

In the 1990s at the Ford Windsor Aluminum Plant, Windsor, Ontario, Canada, (now Nemak Canada, Windsor Aluminum Plant) clogging problems began to develop with the electromagnetic pump used to deliver molten 319 aluminum to the mold in the Cosworth casting process the plant was utilizing. The clogging was due to solid deposits that precipitated from the melt.

An investigation initially revealed that the problem could be partially attributed to the precipitation of insoluble aluminum-titanium-silicon (AlTiSi crystals from the melt, which could be corrected through the lowering of Ti levels to 0.10 wt%). Unfortunately, solving the problem of crystal formation only marginally improved pump performance.

The buildup of solid deposits continued to restrict the efficiency of melt flow through the pump passages. On average, the 319 alloy pump would become unusable after 500,000 lb of aluminum (10 days of casting for the plant, 6000 castings). When this occurred, the pump would be removed from the melt and rebuilt, increasing costs and reducing productivity. These conditions weren't acceptable and the foundry looked for a new solution.

Cosworth Process & Pump Usage

The Cosworth Process for casting uses low-pressure filled, precision sand molds to produce high-strength castings. An electromagnetic pump with a computer-controlled flow rate is employed to fill the mold. The pump is designed to draw melt from the middle of the furnace, avoiding the entrapment of oxides and precipitates from the furnace and improving casting quality.

The original Cosworth Process employed by Cosworth Castings, Worcester, England, pumped melt through the bottom of a zircon mold until the filling process was complete. Pump pressure would be maintained until solidification was complete. At its research facility in Windsor, Ford engineers developed a high production process using a rollover technique in which the zircon mold is rotated 180 degrees after filling so that it can be removed from the filling station while the metal is still molten.

In order to determine the reasons for the pump clogging, the foundry set up an experiment to test the variables and establish a new shop-floor procedure to ensure smooth and efficient production.

Experimental Procedure

A castable ceramic "brick" was removed from the orifice of the clogged electromagnetic pump. This brick was cut into several sections and photographed from a variety of angles. Visual inspection revealed that the passage running through the brick was clogged with an unknown material. It also revealed that some of the clogging residue had diffused through the brick's boron nitride (BN) coating. The BN coating itself was not damaged.

The surface of the clogging residue was examined with a scanning electron microscope (SEM) equipped with an energy dispersive spectrometer (EDS) and a light optical microscope (LOM). The cleanliness of the 319 aluminum alloy melt at Windsor also was evaluated using the Porous Disk Filtration Apparatus (PoDFA) technique.

Experiment Results

The rapid buildup of residues in the pump passage channels and the orifice indicates that the electromagnetic pump in question was clearly not functioning efficiently under the production conditions to which it had been exposed. When the blocked ceramic passages used for 319 melt processing were sectioned, the BN coating remained intact, while the deposit residue was separated in many areas. When manual detachment of the deposit residue from the BN coating was attempted, the coating interface adhered to the deposit.

Figure 1 summarizes the micro chemical (EDS) analyses of the 319 alloy pump residues. The elements present in the residue phases can be put in order of frequency from high to low as follows: Al, O, Sr, Cl, Mg, Cu, Si, Ca, C, Na, Fe, S, Mn, Zn and Ti.

Examination of the pump residue showed that it was composed of a large number of complex and randomly mixed phases without any apparent structural directionality that could be linked to the electromagnetic field and/ or the layered buildup of the deposit (Fig. 2). Among the most important phases identified in the residue were:

* a number of diverse phases containing [Cl.sub.2] (Figs. 3-6);

* both Sr and Mg-rich phases (spinels) were seen in various areas of the blockage materials (Figs. 3-6);

* oxide particles were commonly observed (Figs. 3-6);

* carbon contamination was noted in selected areas (Figs. 3 and 4);

* at least one AlTiSiMg crystal (see Fig. 7) as well as several insoluble AlTiSiMg crystals were found in the blockage residue.

The "mechanical" adherence and diffusion of reactive elements such as Sr and [O.sub.2] into the BN resin were observed in certain areas of the residue interface between the ceramic brick and its BN coating (Figs. 8 and 9).

Electromagnetic Field

There is no metallographic evidence that the electromagnetic field contributed to the formation of the orifice residue phases. All three techniques used for the SEM sample preparation revealed no structural directionality and/or the presence of a layered structure, which would have been observed had the magnetic field caused precipitation of the phases from the melt (Fig. 2). Unfortunately, the available literature contains no information regarding the magnetic properties of insoluble particles at the temperature of liquid aluminum alloy. Although these properties are well established for many elements and materials at room temperature, it is difficult to speculate on the effects of an electromagnetic field on the formation of the blockage residue.

Rotary Chlorine Degassing 

At the 319 alloy melt temperature, [Cl.sub.2] degassing causes some reactive elements to form a vapor (such as [BeCl.sub.2]), while others form liquid droplets (such as [CaCl.sub.2], [LiCl.sub.2], [MgCl.sub.2] or solid particles (such as NaCl and [SrCl.sub.2]). It is important to note that these chloride compounds are relatively stable and can survive several hours in holding furnaces.

A high level of [Cl.sub.2] observed in the pump residue indicates that some gaseous, liquid and solid products of degassing can flow along the furnace launder system and react with other elements and compounds, both from the melt and the refractory material. These compounds can further react with the pump residue, the air and the sand resin while entering the mold. Finally, they can react with Sr and Mg.

The importance of the last set of reactions is substantial. The [Cl.sub.2] rotary degassing operation used at Windsor (utilizing a mixture of nitrogen and chlorine up to 5 wt%) partially depletes the melt of reactive elements such as Mg. The analysis showed that the Mg level drops from 0.35 to 0.33 wt% following this procedure.

Last, Mg from the melt has a tendency to form a thin film around [Cl.sub.2] bubbles that prevents [H.sub.2] from being absorbed. Consequently, the degassing process is impeded by excessive amounts of this element.

Strontium Addition

Because the degassing operation depletes the melt of Sr, an addition of masteralloy was performed at the charge furnace. Masteralloy was added in the form of 10wt%-Sr waffles that were placed on the melt surface. During this operation, the formation of stable Sr oxides and spinels (MgAlO) is possible. These oxides may result from the actual addition of the waffles and/or because of masteralloy impurities.

Some of the smallest particles could pass through the pump filter while others could be deposited on the opposite side during the drainback procedure.

Some residue particles made of 319 alloy (up to 0.8 mm in dia) are found in spinel envelopes and large agglomerates of Sr-oxide films. These envelopes are made of particles up to 0.5 mm in diameter and/or of tangled films (up to 1.2 mm in length). This indicates that in the presence of oxygen, Sr additions promote the formation of spinels and oxide particles (Fig. 8).

The presence of two spinel particles inside the 319 alloy droplet having a diameter of approximately 0.8 mm proves that the A1MgO phase was formed prior to entering the pump passages (Fig. 7). This droplet is surrounded by "pure" spinel, which is surrounded by Sr and Ca-rich spinel. In addition, Fig. 7 suggests that the droplet must have a lower solidification temperature than the large AlTiSiMg particle next to it (brighter phase) because it penetrated through the cracks of the AlTiSiMg particle.

The chemical composition and morphology of some of the Sr rich crystals indicates that they most likely precipitated and grew out from the Sr-supersaturated matrix (Figs. 9 and 10).

Carbon and Oxygen's Role

The casting procedure used at Windsor drains a small amount of liquid metal back from the mold after filling. In the gating system, liquid metal containing Sr and other reactive elements such as Ca and Mg is exposed to the air. Because oxygen is not soluble in molten aluminum, it rapidly reacts with these trace elements and forms complex compounds.

The drain-back flow rate is lower than the mold filling rate. Thus, conditions are favorable for any preexisting insoluble particles in the liquid metal to attach themselves to the rough brick wall, despite the fact that it is coated with BN.

Figure 11 indicates that diffusion of Sr and 02 occurs between BN crystals. The rate of deposition accelerates after the first layer of deposits is formed because other insoluble particles can more easily adhere to previous ones. It should be noted that because of the small amount of metal involved in the drain-back procedure, the overall metal cleanliness at the plant is not adversely affected.

During the process of mold filling and drain-back, it is possible for the melt to react with the air and with the organic binder that contains carbon. This can result in the formation of oxides and spinels, as well as gas(es) containing carbon and oxygen particles. Some of these particles have the opportunity to further react with other elements and compounds when they are drained back.

SEM/EDS analysis revealed the presence of carbon and oxygen in some parts of the spongy residue structure (Fig. 13). This suggests that carbon and/or oxygen gas was a factor in their formation.

Examination under high magnification reveals relatively smooth cavity walls covered by small crystals rich in Al, O, Sr, Mg, C and Cl. The fracturing present in these walls most likely occurred as a result of shrinkage during solidification and gas escape (Fig. 11).

The presence of titanium crystals in the blockage residue suggests that the Ti level in the 319 melt was on the borderline that allows for the formation of such crystals (Ti less than or equal to 0.09 wt%) (Fig. 7).

PoDFA Investigation

The results of the PoDFA investigation suggested that an intensive diffusion of reactive elements into simple inclusion compounds took place during the degassing process and during Sr addition. The PoDFA test conducted at WAY indicated high metal cleanliness even for very high Sr content samples. However, a PoDFA test sample has a mass of approximately 1.5 kg and is taken from the skimmed melt surface, while metal passing through the pump is far more massive and has been partially exposed to the mold resin and air (during the drain-back procedure). Therefore, the phases (or lack thereof) found on the PoDFA filter were not precisely representative of the phases found in the actual blockage residues.

BN Coating of Pump Passages

SEM/EDS fracture analysis of the ceramic brick and its BN coating revealed that a layer of Sr and 0 up to 1.7 mm thick was formed on the surfaces of the coating and the brick (Figs. 2 and 11). This suggests that the BN coating was removed during rodding and the melt was allowed to flow through the unprotected brick.

Clogging Causes

The 319 aluminum alloy pump residue was composed of a large number of complex and randomly mixed phases without any apparent structural directionality that could be linked to the electromagnetic field and/or the layered buildup of the deposit. The presence of compounds containing [Cl.sub.2], O, C, Sr and Mg indicates that the 319 alloy pump residue likely resulted from the following processes:

1. Degassing with substantial amounts of [Cl.sub.2]. Therefore, there is a need to optimize the ratio of [Cl.sub.2] to Ar used for melt degassing in order to minimize the formation of Cl-rich compounds that contributes to deposit formation. A reduction from the present level of Cl2 to less than 0.5% [Cl.sub.2] is suggested. This must be done with care, however, because [Cl.sub.2] reduction will maximize the retention of elements like Mg and other alkali elements. This may lead to increased casting porosity.

At Windsor, the virtual elimination of [Cl.sub.2] degassing was possible because of the very high metal cleanliness. In plants where the melt is exposed to extensive amounts of potential pollutants, this suggestion may not be viable. In this case, Windsor implemented the recommendation and eliminated [Cl.sub.2] from its degassing operation. This single change dramatically improved the pump life. At the time of publication, several electromagnetic pumps have been running for as long as 6 months without the need for cleaning or rebuilding (versus approximately 10 days prior to [Cl.sub.2] elimination).

2. Addition of Sr at the charge furnace and the presence of Mg in the original ingots. This suggests that Sr levels must be carefully optimized in order to allow for Si modification without creating excessive deposit formation.

3. Melt contamination by [O.sub.2] during masteralloy addition as well as during the mold filling and drain-back operation. It is possible that oxides are released from the masteralloy itself or are introduced during the alloy addition process. This problem could likely be minimized if the current method of 10% Sr masteralloy addition (adding waffles at the surface of the melt) was replaced with continuous wire feeding under a protective atmosphere. Finally, the replacement of 10% Sr masteralloy with 90% Sr alloy might remedy the problem, however, melt temperature would have to be reduced to a maximum of 700EC to make this viable. Windsor also has eliminated the use of strontium, providing an overall improvement to its products.

4. Melt microcontamination by O and C as a probable consequence of zircon sand resin decomposition that occurs during the filling of the mold and the drain-back procedure. However, the extent of this contamination is too small to warrant corrective action.

In some areas of the ceramic brick, reactive elements appeared to have diffused through the BN coating into the resin. A pure BN brick would avoid this, and hence might he more resistant to clogging. The cost of the design and installation of such an accessory would be the defining factor in determining the viability of this solution.

[FIGURE 1 OMITTED]

This article was adapted from a paper in the 2000 AFS Transactions (00-22). For more information, contact the AFS Dept. of Special Publications at 800/537-4237 ext. 247.

For More Information

Visit www.moderncasting.com to view the entire 2000 AFS Transactions paper (00-21), "Electromagnetic Pump Clogging in a Foundry Producing 319 Aluminum Alloy Automotive Castings: Causes and Solutions," J.H. Sokolowski, C.A. Kierkus, B. Brosnan and W.J. Evans, AFS, Des Plaines, Illinois (2000).

"The Formation of Insoluble Ti (AI, Si)3 Crystals in 356 Alloy Castings and Their Sedimentation in Foundry Equipment: Causes, Effects and Solutions," J.H. Sokolowski, C.A. Kierkus, B. Brosnan and W.J. Evans, AFS, Des Plaines, Illinois (2000).

"New Molten Metal Delivery System for Cold Chamber Die Casting Introduced," R.M. Slepian, JA. Ciesar, C.C. Alexion and B.D. Ottinger, Die Casting Engineer, May 1993, pp 44-50.

About the Authors

Bernard (Buzz) Brosnan was one of the Ford engineers that developed and implemented the high production Cosworth casting process at the Windsor Research facility and then at the Windsor Aluminum Plant. Dr. Jerry Sokolowski has been a associate proefessor at the Univ. of Windsor since 1993 and is the NSERC/Ford-Nemak/Univ. of Windsor Industrial Research Chair in Light Metals Castng Technology. Christopher Kierkus worked as a research associate at Windsor on this project.