Steel foundries gather to improve casting performance.

Among the highlights of the 1998 T&O were an industry-wide survey and details on innovative acid steelmaking production improvements implemented at Harrison Steel.

The enthusiastic turnout at the Steel Founders' Society of America's (SFSA) 52nd Technical and Operating Conference proves that steel foundrymen are always eager to learn more about improving production and quality. The attendance at the meeting, held November 5-7 at the Hotel Intercontinental, Chicago, totaled 260 - the largest tumour in 15 years - and attendees represented 66 foundries.

The conference's 39 speakers, who included guest presenters from Finland, England, Argentina, Mexico and nine universities, focused on their current operations, their successes with new procedures and overcoming defects.

Attendees showed special interest in an Iowa State Univ. study aimed at improving current steel production and delivery as well as production improvements in acid steelmaking at Harrison Steel Casting Co.

DELIVERY PERFORMANCE

In an effort to decrease the material handling, work in process inventories, rework and scrap that has hurt the steel casting industry, Frank Peters and Tim VanVoorhis, assistant professors in Industrial and Manufacturing Systems Engineering at Iowa State Univ., launched a study of U.S. steel foundries.

The first step of the project, which VanVoorhis discussed in his presentation, "Current Steel Production Practices," was to identify production problems through data collection by researchers working in the foundries and mail surveys. Researchers spent 53 weeks collecting production data at 10 steel foundries, and surveys were sent to all domestic steel foundries. Each mailing included two separate surveys, one for the sales manager and one that contained additional questions about capacities, equipment and staffing for the production manager. The average annual production of the 41 respondents is 11,304 tons, and the average number of employees is 147.

Six foundries reported that their production schedule was based on due date, and 13 said that capacity was a consideration, without specifying the particular areas where capacities were taken into account. Others stated which capacities were factors: molding (11 respondents), melting (8), coremaking (4), each department (3), cleaning operations (2), outsourcing (1) and unspecified bottlenecks (1). Other schedule considerations include backlog, inventory, need, dollar value of products, order quantity, overdue orders and monthly goals. Of all respondents, 69% reported using software to assist in scheduling.

When asked to rank areas with the smallest capacities, respondents named (again ranked from most to least points) molding, heat treat, coremaking, welding, grinding, melting, inspection, blast-cleaning, machining, riser removal, arc air, pouring and shakeout.

When reviewing responses on major capital investment projects completed in the last 5 years and those that are priorities for completion within the next 5 years, researchers found that the two numbers were nearly the same (108 and 102, respectively) and they could be compared directly. Molding and coremaking equipment, which account for most projects, continue to be important, and planned cleaning room equipment upgrades nearly doubled. In addition, the planned purchases of machining equipment and melting furnaces dropped off significantly from the past 5 years.

Twenty-seven respondents, or 64%, track castings through the plant and 12, or 29%, reported the use of bar coding. Bar coding was used to track castings through the plant (3 respondents), shipping in general (2), shipments to a particular customer (2), labor reporting (1), inventory (1), final product identification (1), work order entry (1) and scheduling (1). The average number of production employees able to work in two or more areas was 46%, and the average number of people who process incoming orders is 3.4.

Differing Perceptions

Of the 21 foundries that submitted a response from the production manager and the sales manager, responses were similar, with a few exceptions. As shown in Fig. 1, sales managers paint a poorer picture of on-time performance than production managers. Both groups did tend to agree on reasons for expediting orders, areas with the smallest capacity and those most likely to cause delays.

Molding Processes

Seven respondents reported that at least 90% of their production was investment castings, 18 reported 60% nobake production and 11 reported 60% green sand. Researchers found that of the respondents, investment foundries tend to be smaller and require more labor hours per ton; green sand foundries, while similar to nobake facilities, have somewhat lower labor hours per ton and slightly poorer on-time performance [ILLUSTRATION FOR FIGURE 2 OMITTED]. Also, nobake foundries tended to list heat treat, molding and grinding as bottleneck areas, while green sand foundries listed welding and coremaking.

The study detailed the biggest areas for production and delivery improvement, and the next phase of the project will focus on improved production scheduling and innovative plant layouts.

ACID STEELMAKING

Harrison Steel, a foundry in west central Indiana that manufactures parts for off-road mining and earth moving equipment, has conducted many trials aimed at improving the quality of steel castings, and John D. Carpenter, metallurgist, shared the results of the experiments. in "Acid Steelmaking Practice Changes and Results," Carpenter discussed the foundry's use of an oxygen (O) probe, ferroaluminum (FeAl) as a pre-block, calcium carbide (Ca[C.sub.2]) as a slag reducing agent, argon (Ar) stirring, increased carbon (C) boil to reduce dissolved gas contents, an O valve, and powdered Ca-silicide injection.

Harrison Steel primarily pours Plain carbon and low-alloy steels into green sand molds from bottom pour ladles. The foundry uses one 8.5-ton and two 20-ton electric arc furnaces for acid steelmaking. The furnace lining consists of a silica brick bottom and sidewalls with a coxie sand and dextrine mixture rammed into the bottom, and the ladles are lined with bloating fireclay brick.

A standard melt contains a charge of 50% purchased scrap and 50% foundry returns. A chemistry sample is taken before O injection to determine the necessary C removal. The melter controls the O injection process in order to reach the aim block C level. Then, the heat is blocked with a combination of ferroalloys depending on the steel grade, and the final deoxidation is done during tap with additions from a hopper directly into the tapping stream. All of the acid slag is tapped into the pouring ladle for its insulating value and ease of removal.

Of the initiatives tried at Harrison, alloy wire injection, increased C removal and the controlled-flow O valve were beneficial. In addition, FeAl was beneficial at increasing recoveries but was not cost-effective, and even though Ca[C.sub.2] additions were not effective in the ladle, they were beneficial and cost effective in the furnace. Ar rinsing was unsuccessful with the acid practice, Carpenter said.

Wire Injection

At Harrison, "missed" heats - those with low chemistry - either get poured into ingots for remelting, poured back into the furnace for remaking or poured into castings that end up in the scrap pile. Problem heats, 2.41% of the heats made at Harrison in 1997, were least costly when poured back into the furnace and remade, he said. On average, pourbacks increased cost 18%, pouring ingots increased cost 42% and scrapped castings increased cost 81%. Most of the missed heats (70%) were related to control of chemistry, he said, adding that poor aluminum (Al) and manganese (Mn) control were the most prevalent problems.

Last May, Harrison began injecting Al wire to salvage heats with a low Al content. In the initial trials, the Al added wasn't sufficient - it was stratified in the ladle as measured by test bars poured through the course of the heat. The Al addition was injected in conjunction with Ca wire for its stirring effect. Using this practice, Carpenter said, within the first two weeks of implementation, three heats were salvaged.

To reduce problem heats due to Mn control, medium Ca ferromanganese wire was injected by lowering the amount of Mn addition from the block and allowing for a partial Mn addition to the ladle. This method allows for higher-than-expected recovery heats to remain within specification limits, bringing low side heats back to the desired aim point, Carpenter said. Initial trials were done using three grades, 70, 90 and 180. For all heats, 0.05 weight percent Mn was removed from the block. After the heat was tapped, a ladle chemistry was taken to find out the amount of needed Mn.

Solid Core vs. Powder Core Wire

In early 1995, Harrison tried to achieve the same macrocleanliness obtainable with solid core Ca wire with less expensive Ca-silicide powdered wire. The only noticeable differences between the two products were the change in the wire feed rate due to a different wire diameter and the change in the reaction violence, Carpenter said, adding that operators preferred the powdered product.

The two wire types were compared for differences in cleanliness using data from a 1.5-yr study on a particular casting, with the standard quality measurement system based on recording the measurement of the longest dimension of an oval that could be drawn around any weldable defects. There were two index measurements recorded for each defect type: the actual count data, or number of defects of that type, and the total length of the ovals drawn around the defects from each type. These two indices were interpreted as frequency and severity.

The study, according to Carpenter, suggest that the decreased violence of the reaction may have an influence on the recoveries of oxidizable elements. The powdered core wire was superior for reducing the amount of oxidation in the ladle, castings were less likely to require weld repair for cracks, and there was no change in cleanliness level between the two types of wire, he said.

Based on the results, Harrison switched to powdered core wire, reducing material cost for wire injection by 48.2%. based on the results of a second trial, Harrison reduced the level of Ca addition to 0.65 lb/ton of steel. The change had a slight detrimental effect on casting quality but there were increased alloy recoveries and a 27.8% reduction in the cost of Ca wire, bringing the overall cost reduction to 62.6%, Carpenter said.

Ar Stirring

Harrison initiated trials in which a porous plug in the bottom of the ladle was attached to a bottle of industrial grade Ar. The flow rate was measured with an in-line flow meter, and during the tap, the flow rate was set for 5-6 standard cu. ft/min (SCFM). The flow varied from 1-5 SCFM, and the rinsing time was set at 5 min for non-Ca wired heats. The Ca-treated heats were continuously rinsed during the injection and for 5 min after feeding. This was done with four 20-ton ladle campaigns and a total of 86 heats.

Harrison found that Ar stirring was neither beneficial nor detrimental to casting quality, Carpenter said. Further, there may have been a slight reduction in recoveries of the final deoxidizers, but statistical confidence was low. There was no statistical change in ladle lining life, and the cost of the gas used and ladle plug increased the cost of a heat 0.23%.

Ladle Ca[C.sub.2] Addition

Ca[C.sub.2] was added to the ladle during the tap, when it was half full. This caused excessive foaming of slag, but there was insufficient freeboard available in the ladle to allow the reaction to proceed. The results of these initial trials suggested that any benefits were not realized, Carpenter said.

A second set of trials involved adding incrementally higher Ca[C.sub.2] to grade 148 low-alloysteel, but still, no benefits were realized. It was obvious that the reaction of Ca[C.sub.2] with slag was occurring, but it was felt that the level of O had to be reduced in the slag and that improvements should be seen in the recoveries and/or variations of recoveries, according to Carpenter. Either the reaction wasn't sufficient in the remaining slag to show benefits or the "damage" from high FeO and MnO already had been done before the additions, he said.

Furnace Ca[C.sub.2] Addition

Based on the ladle trials, Harrison decided to try a new approach: deoxidizing slag in the furnace before tapping. Ca[C.sub.2] was added to the furnace just before the block additions. The foundry undertook two studies: one on a low-alloy grade and one on carbon steel.

An addition of 3.75-5 lb Ca[C.sub.2]/ton of steel was shoveled on the slag in the furnace after the block C level was achieved. There was essentially no benefit in the plain carbon grade, but in the low-alloy grade, recovery of Mn and silicon (Si) increased. For a total alloy material cost increase (Ca[C.sub.2] addition minus the alloy savings) of 0.25%, the foundry reduced Si variation by 40.4%, chromium (Cr) variation by 63.2% and Al variation by 44.3%, Carpenter said.

O Probe

In 1995, Harrison purchased an active O sensor, and this probe was used to measure active O at different process steps to use as reference data during different trials.

The first trials involved using a pre-block (35% FeAl ingots) to reduce active O in the furnace to 200 ppm before adding the ferroalloy block. FeAl is much denser than slag but close to the density of liquid steel, and with the chill effect when the Al was tossed into the bath, the alloy melts near the slag/metal interface, Carpenter said. This process was successful, and all 700 of the 20-ton heats produced, with the exception of one steel grade, were treated this way.

The average heat had an active O content of 333.2 ppm, and standard deviation after decarburization was 118.9 ppm. This high standard deviation led to adjusting the FeAl addition based on active O for each heat, Carpenter said. The average level of active O required 4.7 FeAl blocks for an average addition weight of 61.3 lb. FeAl increased the recoveries for all of the oxidizable elements in each of the 10 steel grades tested, but there wasn't a statistically significant change in the standard deviation of the recoveries, indicating no improvement in control, he said. Further, the cost of the FeAl addition was higher than the alloy savings for nine of the grades tested. As a result, the practice was abandoned.

Increased C Removal

In 1996, spurred by a number of "rock candy" fractures in steel grades not typically sensitive to the defect, Harrison began using a N and O analyzer to explore a reduction in dissolved N. First, baseline data on the N levels present during processing was collected [ILLUSTRATION FOR FIGURE 3 OMITTED]. N samples also were taken as pin tests from the risers of poured castings. These tests were well correlated with the test from the midpour, indicating that N pickup during pouring was fairly consistent, Carpenter said. based on the estimated costs of stream shrouding, the decision was made to investigate the N pickup associated with block additions and removing the C from the bath.

Grades containing vanadium in the charged returns and those with the highest level of medium CMn in the block addition had the highest final N content. Medium CMn was reduced to its lowest possible level, and more C was removed through a C "boil" [ILLUSTRATION FOR FIGURE 4 OMITTED].

Last May, Harrison increased the mandatory C boil from 20 points of C to 35 for all grades, except one low-carbon steel sold in the normalized condition. The overall cost of getting a heat in the ladle increased 0.15%, but since the change, there has been no "rock candy"-related cracking, Carpenter said.

O Valve

Last February, Harrison purchased an O valve to correlate the amount of O used with other variables in the process. Before installation, the relationship between the length of time for injection and volume of O injected with an [R.sup.2] value of 0.84. Following an initial problem with flow rate, the foundry improved the [R.sup.2] value to 0.95.

The valve also provided better cost tracking of the heat. Most of the cost analysis from the N study was made possible by this valve. Harrison still plans to explore the possibility of predicting the amount of O required to reach the block C level before injection.