Variables identified for optimal coreless induction melting.

Recognizing the impact of these variables may trigger operational changes to better control your melt power consumption.

In a time when cost reduction is paramount, capital equipment investments in molding, materials handling and coremaking all offer attractive routes for long-term competitiveness. Naturally, the necessary expenditure tends to limit such activities.

However, most foundries can dramatically reduce a major portion of their costs - energy - without spending a single capital equipment dollar, simply through proper optimization of their induction melting equipment.

It has been estimated that foundries are only operating their induction furnaces at 50-80% of their optimal efficiency, wasting valuable energy dollars daily. By understanding what constitutes optimal melting procedures, however, dramatic reductions in energy consumption can be realized. For instance, a foundry melting 1000 tons/month could easily reduce its monthly melting costs by as much as $5/ton simply by altering its melting practices.

Under a research contract from the Electric Power Research Institute (EPRI), the Cast Metals Institute of the American Foundrymen's Society conducted a study relating charge makeup materials and other melting practice variables to the total power in kilowatt-hours (KWh) to prepare liquid metal for foundry applications.

Variables

Four major variables were studied that were considered the most important in determining power required for melting. They include:

* charge makeup (virgin, virgin/revert, revert);

* furnace cover (on, off);

* power application (step, full);

* furnace condition (hot, medium, cold).

Experimental Procedures

Charge Materials - Combinations of virgin and revert material (100% virgin, 50/50 virgin/revert and 100% revert) were used to make test castings and pigs (about 50 lb each). The metallic virgin material for the gray iron, WCB (plain carbon steel), and 8630 (low alloy steel) heats was low carbon steel punchings (about 2 x 1 x 0.125 in.), while the CF-8 (stainless steel) virgin material was either 1/2 in. thick pieces of plate, 15-lb half round ingots, or 20 lb 2-3/4 in. diameter continuously cast bar. Revert materials were primarily 50 lb pigs cast from previous heats or gating systems from test castings (keel blocks and test plates). Ferroalloys and carbon were granular in form (no larger than 1/8 in.), aluminum was in shot form, and nickel in briquette form.

Equipment - Melting was conducted in a 500-lb coreless induction furnace using an alumina crucible backed with dry rammed MgO refractory [ILLUSTRATION FOR FIGURE 1 OMITTED]. Three times during the study a new crucible was installed due to cracking and metal penetration. Power readings were taken from the furnace control monitor during melting.

Melting Practices - Use of virgin materials for gray iron, WCB, and 8630 required melting of steel punching scrap and additions of other charge materials previously described. Aim chemistries are shown in Table 1. Layers of carbon and ferroalloys were added to the furnace and charged with the remaining punching scrap on top. Mechanical ramming of the punchings during melting was necessary to prevent bridging. Final additions were made just prior to raising the melt to pouring temperature. In the case of CF-8, the virgin material was prealloyed, and only final additions to maintain chemistry were made before pouring.

Except for four heats melted by the heel practice, batch melting was used exclusively. The furnace was emptied for each heat and then, depending .upon the furnace condition called for during the next heat, the next complete charge was added at the appropriate time. For the "hot" furnace, the next charge was added immediately after emptying the furnace. For the "medium" furnace condition during the CF-8 runs, a waiting period of 1-1/2 hr occurred before charging the next heat. For the "cold" condition, an overnight or longer wait was used before melting the next heat.

Experimental Design - Initial runs were made on a [TABULAR DATA FOR TABLE 1 OMITTED] screening basis, followed by use of Taguchi-type fractional factorial experiments to evaluate the importance of the four variables, and their interactions. A total of 34 heats were made. During the study it was found that crucible life and sequence of melting was also important in at least part of the study. Factors such as these can be evaluated as covariants in the analysis. Additional comments on the variables and their effects are presented later.

Two basic Taguchi orthogonal array designs were used to set up runs and analyze the data.

Results

Power consumption for all heats made was measured by recording the time taken to arrive at pouring temperature, and the KWh reading at that time. This would be most representative of power consumption since the primary concern was with melting rather than tapping and pouring.

Early findings in the study showed some factors that confounded results and comparisons between heats for the total KWh in melting. The first such factor was installation of a new crucible after penetration made replacement necessary. This occurred first at the seventh run, then at two later runs. One of the new crucibles was necessary because of a power failure to the foundry just before pouring.

Table 2. Power Consumption for Alloys.

               Gray Iron    WCB/8630    CF-8

No. of Runs         6           16        12
Total KWh        1543         5226      3466
Avg. KWh          257          327       289
Std. Dev.        26.3         36.5      24.8
Variance        694.1       1332.7     614.8

On installation of the first new crucible a 42% increase in power was noted [262 KWh (for a 500 lb melt) for run 6 vs. 371 KWh for run 7]. A generally decreasing power was then noted for subsequent runs (crucible sequence) for the entire set of 34 heats.

A new crucible installation at the 21st run again showed a large increase (29%) over the previous heat. The last new crucible installation at the 25th run showed a smaller increase (10%), a drop, then a slight increase. An analysis for this effect was studied in an "invert" design run that showed that crucible sequence was important - each use resulting in lower energy consumption. A possible explanation is that backup refractory for the first six runs was not MgO, but rather an alumina mix having a lower thermal conductivity. Replacement with MgO having a higher thermal conductivity would increase heat losses and require more energy in subsequent runs.

Gray Cast Iron

First heats made using 100% virgin for gray iron showed a much greater amount of power (21%) needed to melt this material than that needed for 100% revert (299/288 KWh vs. 241/245 KWh). This is due to the fact that significantly more time is required to raise the carbon level using a carbon raiser than pig. In addition, coupling of the charge is not as efficient. As a result, 100% virgin was not included in the design. A 50/50 virgin/revert was used instead, which better represents typical foundry practice.

Analysis of these results showed a hot furnace used 2% less power than a cold furnace (229/245 KWh vs. 241 KWh). Also, melting with the cover on used 3% less power than cover-off (229/241 KWh vs. 242/245 KWh). Least power was used for a 50/50 virgin/revert charge, with the cover on.

WCB and 8630

There wasn't any significant difference between WCB and 8630 in the initial results, and therefore they were combined in the analysis and discussion. Analysis showed large interaction effects, the most significant being a three-factor interaction. Least energy use occurred with the combination of a hot furnace, virgin charge and step power when not considering the crucible sequence. Taking the crucible sequence into account shows a two-factor interaction to be most important - lowest power with virgin charge and step power.

Grouping together comparable experimental conditions for each of the four variables and averaging them shows the following impact on energy use:

* a negligible difference between virgin and revert (1% greater for revert);

* slight difference between full and step power (3.5% greater for full);

* a 3.8% difference between hot vs. cold furnace (greater for cold);

* a 13% difference between cover off and on (greater for cover off).

CF-8

In instances where replicate heats were made, power was reproducible within about 5%. (Comparison of KWh for three replicate runs showed 310,299 and 317 KWh, respectively, and two other replicate runs showed 288 and 272 KWh, respectively).

Furnace condition and charge are the most important factors. The next most important is the charge mix (ingot or plate, 50/50 ingot, plate/revert). The plate alone used more energy than either the AOD or VOD ingot. As expected, the warmer the furnace (hot vs. medium vs. cold), the lower the energy.

Cover use didn't make a significant difference in melting this alloy. As compared to WCB, lower heat losses are probably due to a combination of lower temperature for the CF-8, the presence of a slag cover which could affect emissivity losses significantly, and the non-magnetic nature of CF-8.

Alloy Effects on Power

Three distinct levels of power consumption were measured - one for each alloy as shown in Table 2. As expected, the KWh increases with the melting point of the alloy.

Heel Melting

One series of heats was processed using heel melting rather than batch melting to determine differences in power consumption for the two practices. In this series, 500 lb of CF-8 was melted using full power and cover on. When molten at the pouring temperature of 2900F (1593C), about 300 lb was tapped and poured into pig molds. [TABULAR DATA FOR TABLE 4 OMITTED] Then, 300 lb of revert pig was immediately charged into the hot furnace and full power with cover on was applied. When molten again at 2900F, another 300 lb was tapped, and the process repeated. Two additional taps were made; one at about 300 lb and the last at 500 lb for a total weight (charged, melted and poured) of 1500 lb. KWh readings were recorded at tap temperature and when charging was started, and after tapping, for a total of 866 KWh.

Table 3. Power Consumption for Heel and Batch Melting 1500 lb of
CF-8.

               Heel Melting     Batch Melting
                                  1       2

KWh               830            862     894
Time (Min)        345            361     380

Total power for heel melting was then calculated by subtracting the energy used between tapping and recharging from the total 866 KWh measured. Total power for batch melting was calculated by adding the energy used, up to tap temperature, in the two series of runs where three heats were made in one day. The comparison between the two practices is shown in Table 3.

Conclusions

Variables of charge material, power application, cover use, and furnace condition were found to be significant in determining the power consumption during coreless induction melting of ferrous alloys. The degree of significance was found to be dependent upon the alloy being melted. Results for major variables are summarized in Table 4 and discussed below.

Charge Materials

In gray iron, virgin material use resulted in significantly higher energy use (22% more) than for revert, probably because of the requirement to get carbon in solution and its associated time and temperature. For the WCB/8630, no significant effect was shown since adequate coupling to the induction field was achieved. Use of 1/2 in. thick 304 plate required more power (9.8% more) than either a mixture of plate and ingot or pig, or 100% ingot or pig. Coupling in this case was considered better for the more densely packed crucible containing ingot and/or pig.

Furnace Cover

In melting WCB/8630, use of a furnace cover reduced energy consumption by 12%. For the gray iron and CF-8, there was only a small difference (about 2% less for cover on), which isn't easily explained. Lower emissivity losses are due to the lower temperatures for these alloys compared to WCB, and the degree of slag cover present could also contribute to this effect.

Furnace Condition

Except for the gray iron, this factor is significant, especially as an interaction term with the charge. Considering averages for hot, medium and cold furnaces, the KWh for each condition resulted in 264,287 and 305 KWh, respectively, for the CF-8 (15.4% less for the hot furnace).

In the WCB/8630, a hot furnace and virgin charge resulted in the least energy consumption (3.5% less). For the CF-8, a hot furnace and charge, and their interaction, were the most significant factors.

Power Input

For the WCB/8630, step power gave lower energy consumption than full power (2% less), but primarily as an interaction term between virgin charge and a hot furnace. Considering the averages only (for comparable runs) for this' material without the factorial design, full power showed an average of 338 KWh vs. 327 KWh for step power. The difference found wouldn't offset the productivity losses resulting from a longer step power melt cycle, however.

Heel vs. Batch Melting

Comparison between these two practices showed that heel melting required less energy input than batch melting for 1500 lb of CF-8 revert. Heel melting used 830 KWh while batch melting used 862/894 KWh.

This article was adapted from a report published by the EPRI Foundry Office (847/427-9060).