Fighting defects with improved gating: four metalcasting firms overcame casting defects by augmenting the gating systems.
Four metalcasting facilities were faced with inclusion defect problems in their castings. However, altered gating systems allowed them to overcome their dilemmas, reduce scrap and improve production.
This article details case studies from these green sand casting facilities and how problems were solved with gating.
Neenah Foundry Co.
Neenah Foundry Co.'s Plant 3 ductile iron casting facility, Neenah, Wis., was experiencing defects on a 20-lb. (9.07-kg) ductile iron round housing. The heavy automotive casting measured 8.5 in. (21.5 cm) in diameter with a center opening 2.5 in. (6.35 cm) in diameter. The casting featured six "scallop"-type bosses 1.5 in. (3.81 cm) in diameter and 2.125 in. (5.4 cm) thick on the perimeter of the part. These bosses were divided into two segments, three on each side, by a thinner section 2 in. (5.08 cm) wide and 1.063 in. (2.7 cm) thick.
The job was run as two-on configuration on a vertically parted green sand automated molding machine with the two pattern impressions on the same horizontal plane. The patterns were radially positioned with three of the bosses at both the 6 and 12 o'clock locations for optimum feeding by a riser placed at 12 o'clock. An iron filter was located at the base of the down runner, which then fed two horizontal runner bars, one beneath each pattern impression. Each cavity then was filled via one ingate at the 6 o'clock position (Fig. 1).
[FIGURE 1 OMITTED]
After 15 production runs over six months and a total of 17,000 castings, the overall scrap percentage was at 7.4% with 6.1% due to inclusions. Slag and dross were determined to be the cause of inclusions, even though several runner systems were found to be free of any such inclusions. The focus initially was placed on the filling characteristics (such as turbulence) in the mold cavities.
It was believed that turbulence during mold cavity filling was due to ingate placement directly on the three bosses. As a result, the cavities were rotated 90[degrees] to place the "thin" sections of the casting at 6 and 12 o'clock, so the ingate would be in a more compressed and uniform section (Fig. 2). In turn, this would reduce the tendency of turbulence and inclusions as the mold filled.
[FIGURE 2 OMITTED]
Rotating the patterns to relocate the ingate was successful. Since the change, more than 14,000 castings have been produced, and scrap levels have decreased to 0.6% for inclusions and 2.2% overall. Further, without any modification to the riser or its placement, the casting remained sound.
Neenah also had been experiencing inclusion difficulties with a 40-lb. (18-kg) ductile iron "hub type" casting measuring 11.5 in. (29.21 cm) in diameter. The flange on the part was 1.25 in. (3.175 cm) thick and contained six equally spaced scallops that reduced its thickness to 0.625 in. (1.58 cm). There were two feed paths 180[degrees] apart connecting the flange to the center hub for feeding purposes.
The pattern was mounted as one-on configuration on a vertically parted green sand automated molding machine. The feed paths were radially located at the 6 and 12 o'clock locations to allow for a riser to be placed at the 12 o'clock position. An iron filter was placed vertically just ahead of a bottom horizontal runner bar. Two ingates then were placed from the runner bar to the casting (Fig. 3).
[FIGURE 3 OMITTED]
During 10 production dates, the rejection rate for the 4,200 pieces cast was at 8.9% in which 5.5% resulted from inclusions. The inclusions then were defined further as re-oxidized slag and dross. Several runner systems and ingates were pulled at shakeout and blasted for analysis. The gating was found to be clean and free from any defects, which led the staff to focus on the filling characteristics in the mold cavity itself.
Based on the results of the automotive round housing casting, it was decided to change the ingate location. The existing ingates were positioned half on the "thick" and half on the "thin" area of the flange. It was believed that this change in section size caused turbulence during filling, resulting in the slag formation. The ingates were repositioned directly in front of the thin, or "scalloped," area of the flange (Fig. 4).
[FIGURE 4 OMITTED]
Like the first part, Neenah found success with the relocated ingates. More than 1,580 parts have been made with inclusion scrap at 0.8% and overall scrap at 1.5%.
Deeter Foundry, Lincoln, Neb., began experiencing shrink in gray iron thick-walled castings after beginning to use steel in its cupola charge. It was determined that shrink was occurring in the ingates (Fig. 5). The casting was produced via a horizontally parted green sand molding machine.
[FIGURE 5 OMITTED]
Adding risers was an option but considering the added cost in the finish room, the staff chose to try casting with a lower pour temperature to lessen the chance of shrinkage. After switching to lower pour temperatures, Deeter began to notice a snake-like series of inclusions (Fig. 6), and similar conditions started to appear on other castings. Data indicated that the defects were linked to lower pour temperatures and low manganese and high sulfur levels in the iron, causing a re-evaluation of the most effective options to rid the castings of the defects.
[FIGURE 6 OMITTED]
Despite the potential for increased cleaning costs, risers were added to the gating systems, and molds were poured at higher temperatures. As a result, most of the inclusions were eliminated.
Grede Foundries-Iron Mountain Foundry
At its Iron Mountain Foundry, Kingsford, Mich., Grede Foundries encountered cope surface defects on gray iron hydraulic valve bodies for the fluid power industry produced on a horizontally parted green sand molding machine. The castings ranged from 13-25 lbs. (5.9-11.3 kg) with multiple cavities in the mold and each having pour weights ranging from 113-154 lbs. (51.2-69.5 kg). These defects were determined to be mold sand inclusions located on the cope face of the castings. Historically, these castings ran at a 3-5% scrap level, which caused an unacceptable number of returns for sub-surface porosity.
After evaluating the situation, it was determined that the current pour rate of 5-6 lbs. (2.2-2.7 kg) per second was too slow and that the molding sand started to break down before the cavities filled with molten iron. This caused the sand to settle on the cope surface of the castings, thus causing the porosity.
A modified runner system on a sample part was used to pour the casting at 8-9 lbs. (3.6-4.08 kg) per second versus the 5-6 lbs. (2.2-2.7 kg) per second rate, a 24% increase in pouring rate. The modified runner system then was used on a similar part, in which the pouring rate was increased by 26%--from 5.7 lbs. (2.58 kg) to 7.7 lbs. (3.5 kg) per second.
The same principle was applied to a third part in which the pouring rate was increased by 30% (from 5-6 lbs. [2.2-2.7 kg] to 7-8.5 lbs. [3.1-3.9 kg]). As shown in Fig. 7, the increased pouring rates due to the modified runner improved the overall quality of the castings.
[FIGURE 7 OMITTED]
The three parts that were used for this case study currently run at 0.5-2% scrap with a significant improvement in returns (Fig. 8).
[FIGURE 8 OMITTED]
Grede Foundries-New Castle Foundry
Grede's facility in New Castle, Ind., encountered "fisheye" defects on automotive castings, which were cast in molds produced from 34 x 48-in. (86 x 122-cm) flasks by an automated horizontally parted green sand molding machine. When it was first detected, the defect only occurred for one in every 1,000 scrap castings. It was assumed that the defect was caused by a piece of core sand on the mold face that had burned, giving off gas in a localized area.
However, the defect increased and became a significant percentage of scrapped castings.
After contacting several colleagues about the defect, New Castle received two main suggestions--the defect was caused by "oil-dry" type absorbent material getting into the sand system or by exothermic sleeves. However, the defects that were described to go along with these did not match the defect on Grede's castings (Fig. 9).
[FIGURE 9 OMITTED]
Attempts were made to screen the sand and pull out various "chunks" of debris, including core butts, clay balls and exothermic sleeve pieces. These pieces then were molded against the face of a pattern and castings poured. The worst defect the chunks caused was a slight scar on the casting surface.
The mulling equipment was examined, including cleaning the floor and resetting plows and wheels. Clay and water additions also were reduced to try and prevent any clay balls and the fisheye defect, but nothing improved.
The staff noticed the defect increased immediately after it ran high copper, pearlitic grades of iron. Another colleague then identified the cause of the defect as some residual from the exothermic sleeves in the green sand. The residual material had to be combined in the green sand, and not at the mold surface for the defect to occur.
The sleeve manufacturer overcame the defect by changing its sleeve formulation to a "fluorine-free" recipe. By examining scrap data and the approximate time that the defect started to appear, New Castle found that the errors seemed to coincide with a change made by the sleeve supplier in the formulation of the sleeves.
The casting firm tested the change before the larger number of defects, but did not discover any difference in the castings' performance or an outbreak of fisheye defects.
Further, most of the exothermic sleeves were being used on jobs (mostly pearlitic iron grades) transferred to New Castle from sister plants. As the staff worked to reduce downtime involved in grade changes, it started scheduling all these parts together in a one-to-two-day production schedule for pearlitic grades. However, the sand system was being shocked with a plethora of exothermic sleeves for that period of time.
The sleeve supplier offered alternative sleeves while it continued to investigate the cause of the problem. One alternative was to switch back to the original formulation, with which New Castle had only rare occurrences of the defect. The other was to try an insulating recipe sleeve instead of an exothermic sleeve. The latter would require testing and possible PPAP submission of each part number to ensure that shrink would not be introduced into the casting.
There were 20 patterns that used sleeves that were considered medium-to-high-volume parts. After reviewing several gating systems on these patterns, the facility's engineering department noticed most of these parts were considerably over-risered. Often, the sleeve was perched on top of an already substantial sand riser. It was decided that each of the 20 patterns would be reviewed individually and the sleeves either eliminated or replaced with insulating sleeves. Initially, the staff started to work on the highest volume jobs and highest sleeve users. As it eliminated the sleeves or replaced them with insulating sleeves, the fisheye defect began to disappear. At the end of the investigation, New Castle had successfully changed or eliminated the sleeves on all 20 parts.
The fisheye defect still appears occasionally, caused by some of the other jobs with exothermic sleeves, but nothing that causes any significant defects.
The yield on all these jobs has been significantly improved. In many cases, the cost of the sleeve has been saved, and there has been an overall cost savings per casting.
RELATED ARTICLE: Finding other means to reduce inclusions.
Augmenting sand mold gating systems isn't the only way Deeter Foundry, Lincoln, Neb., solves inclusion defect problems. The metalcasting firm also found success in eliminating inclusion defects by improving sand and molten metal formulas.
Deeter experienced metal penetration on a large frame casting, of which voids were found in the mold surfaces. These appeared to have been created by molding sand lumps from one sand system that were coming out of both mullers. The second sand system did not have lumps despite using the same premix and new sand bins as the problematic system. This revealed that raw materials were not the cause. The staff checked for clay balls but found the Methylene Blue Clay and LOI/Volatiles were the same as the system sand.
It was found that lumps started to appear on the sand surface 15-20 sec. into the mull cycle. However, timing the sequence of adding water, return sand, and new sand and bond revealed no errors.
Deeter then focused on trends in its green sand tests and found that the level of fines had dropped. Although the percent moisture of the green sand was normal, further investigation found that the moisture of the sand out of the cooler was lower than readouts indicated (1.2% actual versus 2.0% setting). The casting firm concluded that the missing water was being compensated at the muller, but it was not effectively tempered in comparison to being added in the cooler. Fines in green sand tend to hold water. With drier sand, more fines were taken up by the dust collector aggravating the condition.
The firm changed the probes to obtain accurate moisture level feedback out of the cooler and added them to its calibration schedule. The new sand was set back temporarily to allow fines to build up again. The clay lumps and penetration disappeared, and the casting finish improved.
Deeter overcame another inclusion problem when it experienced sand and slag defects on manhole covers. The staff tried to duplicate the defect by putting a teaspoon of core sand in the ingates of three molds and additions of green sand to ingates of three other molds. No inclusions were found in either group.
The trial was repeated with the sand (some lumps as large as kernels) placed inside the mold rather than the gate. The results showed that some molds were inclusion-free, while only those with the largest lumps of sand showed inclusions, revealing that improperly blown-out molds or friable sand was not the problem.
At this point, the finishing department began to report frames with inclusions on the bearing surface, the cope side.
Deeter concluded that slag/oxides and gas were forming in the mold at low pour temperatures. As the iron level rose in the mold, gas was trapped under the horizontal surfaces. The minimum pour temperatures on affected castings were raised, and the rate of defects was reduced. However, inclusions remained.
Scrap data then was pulled for one casting that consistently suffered from inclusions. A spreadsheet was prepared to see if any correlation existed between the sand data and the percent slag scrap, but nothing was found.
Deeter looked next at iron chemistry. The mixture of manganese, copper and tin was not perfect, but the staff noticed that when the total of these three was low, the scrap was higher. It was concluded that the low levels of residuals in the iron were not the cause of slag but indicated the physical characteristics of the charge material, which was shredded steel thinner than the steel punch-outs used at other times.
The thinner material, with greater surface area-to-weight ratio, allowed more oxidation and, thus, more slag. This was not seen in trial runs with shredded steel but research into the slags suggested that this was an issue at lower pour temperatures. At higher temperatures, the oxides break down, recombining with other elements going into solution in the iron. Trials then were run using a special low-temperature flux in the ladle and bulkier material for the steel charge. With the flux and larger scrap, the slag levels were much lower than previously, and the result was a major improvement in the appearance of its castings.
This article was adapted from a paper (04-101) presented at the 108th Metalcasting Congress, Rosemont, Ill.
A.J. Huff is the technical director for Foundry Advanced Clay Technologies, Three Rivers, Mich. John McCutcheon has worked 27 years at Neenah Foundry, Neenah, Wis., where he is the casting engineer, responsible for gating and risering. Jeff Bianco is the factory manager at Grede Foundries Inc.--Iron Mountain Foundry, Kingsford, Mich. Godfrey Sergeant is the technical director at Grede Foundries Inc.--New Castle Foundry, New Castle, Ind. Mary Wolfe is process manager for Deeter Foundry, Lincoln, Neb.
For More Information
"Effect of Gating Design on Mold Filling," M. Masoumi, H. Hu., J. Hedjazi and M.A. Boutorabi, AFS Transactions (05-152).
A.J. Huff, FACT, Three Rivers, Michigan John McCutcheon, Neenah Foundry Company, Neenah, Wisconsin Jeff Bianco, Grede Foundries Iron Mountain Foundry, Kingsford, Michigan Godfrey Sergeant, Grede Foundries New Castle Foundry, New Castle, Indiana Mary Wolfe, Deeter Foundry, Lincoln, Nebraska
|Printer friendly Cite/link Email Feedback|
|Date:||Jul 1, 2005|
|Previous Article:||2005 foundry of the Year Hayes Lemmerz Montague. By increasing automation and improving efficiencies and quality, this plant has grown with its...|
|Next Article:||Practical tips on gating iron castings.|