The aluminum industry is a large consumer of energy on a per-weight basis. Improvements in refractory technology represent significant potential for energy savings. In this final of two parts, problems inherent in the molten aluminum/refractory interface are reviewed, and the performance trials of a new castable refractory formulation are presented.

Fig. 5. Furnace lined with D-1 shows no appreciable refractory damage after cold cleaning.

In Part 1 of this article (February 2006, IH) we reviewed some of the problems in the aluminum industry and how they led to the development of a castable refractory material, Formulation D-1 (D-1), that would prevent heat loss in thermal processes and clean easily after exposure to molten aluminum.

The energy efficiency of secondary aluminum production is affected greatly by the properties of the refractories used. Undesirable reactions sometimes occur inside a furnace, either within the refractory lining or adjacent to it within the vessel. These can reduce the service life of the refractory, impede production, or cause unscheduled maintenance outages.

The formation of corundum, an oxide of aluminum, within a furnace and around its components can lead to furnace inefficiencies. It can be scraped off, but this procedure often damages the refractory lining in the furnace. Also, corundum formation can cause stress fractures within the refractory matrix and further impede furnace and refractory performance.

Similarly, the formation of spinel, a mineral phase formed by the reaction of aluminum and magnesium in the presence of oxygen, can infiltrate the pores of a refractory lining and cause problems. Because spinel expands on crystal formation, it can cause the ceramic to fracture. Formation of spinel can cause the same operational problems as corundum.

The development of a new castable refractory was undertaken to alleviate these problems. It was focused on thermal efficiency and suitability for molten aluminum contact. Ease of installation was a secondary objective. The formulation that resulted was designated D-1. Some technical properties of D-1 are given in Part 1 of this article.

We tried the new formulation with a number of customers and in varied applications. Brief case studies of how the new castable performed in service are presented here.

Case Study - Comparison of Two Holding Furnaces

Customer 2 agreed to line a holding furnace with D-1. This furnace is approximately twice the size of that operated by Customer 1 (see part 1, IH February 2006). Customer 2 also agreed to line another furnace with their ‘standard' material with exactly the same configuration as the test vessel. A direct comparison between two vessels was made. The vessels are identified as DOE1 and DOE2, with DOE1 having been lined with castable D-1. Thermocouples were embedded in each furnace at identical points within the lining and data were recorded during several months of operation.

The metal temperature in each furnace averaged 1,133°F. Both hot face thermocouples reported an average temperature of 1,130°F. The shell temperature of DOE1 Formulation D-1 was 113°F on average, while DOE2 averaged 138°F. Overall, DOE1 caused more heat to be retained in the furnace, a fact also confirmed by monitoring the kilowatt usage of each furnace. This experiment was performed by Customer 2 and the results were shared with us.

For DOE1 the one week kilowatt usage average was 2,920 kWh. For DOE2 the kilowatt usage for the same period was 3,460 kWh. Therefore, the ‘standard' furnace (DOE2) used 18.5 percent more electricity than the furnace lined with D1 during a 7-day period holding the same alloy at the same temperature. If the 540 kWh per week savings is annualized, 28,080 kWh can be saved over one year. At a current electricity price of $0.0463 per kWh, the furnace lined with D-1 can save the customer about $1,300 annually. This customer has more than 40 similar furnaces on site, as well as several larger melting furnaces.

This experiment also demonstrated the improved performance of D-1 compared to the standard lining after 9 months of service. Figure 5 shows the DOE1 furnace after cold cleaning. All material build-up was easily removed from the walls. As a result, there was no appreciable damage to the refractory. In addition, no penetration of the refractory is observed. We concluded that although a furnace lined with a standard refractory material will degrade physically and thermally, a furnace lined with D-1 will maintain its physical integrity, even though bench testing exhibits lower overall strength, and will retain its optimum thermal efficiency over a longer period.

Fig. 6. Over-the-road crucible.

Case Study - Over-the-Road Crucibles

Customer 3 tested D-1 in over-the-road-crucibles. This application was intended to improve the wear rate of the vessel and therefore improve the vessel's lifetime and average efficiency. The trial is of particular interest because in this application heat is supplied by a gas burner. We wanted to see how the material performed in this environment compared to the holding furnaces, which were heated electrically. Gas heat is less controllable than its electric counterpart, but the reaction time needed to make temperature adjustments is quicker. As a result, head space temperatures tend to be higher and thermal shock becomes a concern. The head space and metal line in these vessels represent the highest wear areas.

Commonly, the crucibles are filled with metal and kept at an appropriate temperature by using a lid that contains a burner. When ready to transport, the burner is removed, the lid is sealed and the crucible is placed on the transport vehicle. Because the crucible needs to retain as much heat as possible, it is typically lined with an insulating material. Initially, the current material is more insulating than D-1. However, this is achieved by the use of porous aggregate with a large pore size. In time the refractory develops buildup, becomes penetrated, and its insulating capability is significantly reduced. Because the current refractory is relatively weak and is easily damaged during cleaning, it must be replaced often. Customer 3 wanted a material that would maintain an average insulating character over the lifetime of the crucible that was an improvement over the current refractory.

Initially only the burner/transport lids were cast of D-1. These were put into service along with lids comprised of the current material. Both were monitored for wear and average lifetime. One lid was run with a combination of the two materials, which allowed a side-by-side comparison. The standard material became significantly worn, while the D-1 material wore much less.

After successfully demonstrating its capability in the lids, D-1 was installed in the most difficult section of the crucible body, which is the top 18 inches. This is a difficult area because it is at the metal line. There is a great temperature differential between the areas above and below the metal line when the burner is on. Furthermore, this area is susceptible to corundum buildup because of the availability of atmospheric oxygen. Excessive wear in this area typically takes the vessel out of service.

Figure 6 shows the crucible body lying on its side. The division between the two materials is apparent, with D-1 comprising the top 18 inches. This photo was taken after a few runs and the vessel was turned down to observe any differences and to clean buildup that typically occurs. Our customer reported that the 1-in. band cleaned very easily while their usual material required more drastic measures to clean it. In this case, they used a mechanical device to scrape off the buildup, and scrape marks, which cause premature wear, can be observed in the material.

Shell temperatures were periodically measured between the crucible portion lined with the standard material and that lined with D-1. The standard lining started with a better insulating capability than D-1. However, over time, this advantage is lost due to penetration and loss of the standard lining. D-1 retains its structural integrity by resisting penetration and cleaning much easier. As a result, after 11 months of service in the sidewall of the crucible the area lined with the standard material is reading an average 8.4 percent higher in shell temperature compared to D-1.

Case Study - Special Alloy Production

Customer 4's application was solicited because it operates at a higher than typical temperature, approximately 1,204oC. Their alloy is atypical and very aggressive to refractories.

Because the customer's process is confidential, we were not permitted to monitor the performance of the test installation furnace. However, the customer monitored energy efficiency and shell temperatures as part of their evaluation process. The customer's normal lining lasted only six months, by which time the furnace was so penetrated that most thermal efficiency was gone.

Table 2 tabulates the measured energy usage values for each lining. Reported results indicate that the furnace lined with the standard material uses 19.9 percent more energy than with D-1. Selected temperature readings are shown in the bottom portion of the table. Actual and relative differences are shown for comparison.

Fig. 7. Hot wall installation with forms stripped.

Case Study -- Service Trial in Reverberatory Furnace

Because Customer 2 had a good experience with D-1 in their holding furnaces, they asked us to try the material in a hot wall of a 100,000 lb. melting furnace at another location. The customer agreed to cover the costs of the trial, which required 11,000 lbs of D-1 for installation in the hot wall.

A picture of the wall after stripping the forms is shown in Figure 7. This view is from inside the furnace. Aluminum can be seen sticking to the roof. The only portion that was cast in D-1 was the wall, with doorways. The discoloration is a result of the grease used as a release for the wooden forms.

This wall typically fails before any other part of the furnace for two reasons. The first is that the molten aluminum level comes close to the top of the doors. As a result, the bottom of the wall is immersed in molten aluminum at the bath temperature and the area on the outside of the wall and above the metal in the well is exposed to atmosphere at another temperature. The second is that the inside of the wall above the aluminum is exposed to the furnace atmosphere and, more importantly, is impinged by the flame from two gas burners. These are mounted on the opposite wall and expose the hot wall to a third temperature.

This puts tremendous thermal stress on certain critical areas of the wall. The current refractory composition lasts approximately 2 years before it finally breaks apart and has to be replaced. After 8 months of service the D-1 hot wall installation is holding up as well as the prior lining and shows no signs of thermal stress cracking.


This research has been successful in creating a product for aluminum applications that will improve the thermal efficiency of aluminum manufacturing operations. A unique refractory castable has been developed. The composition is such that aluminum will not adhere or penetrate the lining, corundum will not form internally, and any external corundum that forms will not strongly adhere. The micro-porous nature of the castable provides significant insulating capability and resultant reduction in energy consumption. The customers who have field tested the castable have found that cleaning operations take less time and result in less wear on the lining. IH

  • Authors McGowan and Beaulieu can be reached at Westmoreland Advanced Materials LLC, 110 Riverview Dr., Monessen, PA 15062; ph. (724) 684-5902; fax (724) 684-5962; email; or visit Formulation D-1 and variants are protected under United States Patent Application US20050049138A1, dated March 3, 2005, and in the European Union under appropriate filings.

  • Additional related material may be found by searching for these (and other) key words/terms via BNP Media LINX at aluminum, refractory, corundum, alumina, spinel, silica.