Optimizing Sand control at New Haven Foundry.
The demands of the automotive sector place a premium on surface finish and dimensional reproducibility for its casting suppliers. Due to these demands, foundries must continuously re-examine their existing production practices such as sand control and molding to ensure the required casting properties are being met today and will be met in the future.
For New Haven Foundry, a gray iron shop located in New Haven, Michigan, the demands of the future prompted self-evaluation. The foundry produces 160,000 tons/yr of automotive components such as cylinder heads, flywheels and bearing caps via horizontally parted green sand molding. In anticipation of new demands from its automotive customers, including the addition of new brake rotor component production, the foundry decided to begin optimizing its sand system in 1999.
The focus of this optimization was low-cost upgrades. The foundry was looking for opportunities for increased control of sand system properties rather than an upgrade of technology. This article takes a look at New Haven's sand system and the optimization it underwent.
The sand system at New Haven Foundry consists of two 200-ton/hr mixers equipped with controllers feeding 2 jolt squeeze molding lines. The molding lines are producing 300 molds/hr at a flask size of 22.5 x 39 x 9 in. The majority of the foundry's cores are hotbox, but the foundry also produces coldbox cores.
The base aggregate of the sand used by New Haven is a four-screen lake sand at 55 AFS grain fineness number (GFN) for molds and 48 AFS GFN for cores. A small quantity of finer bank sand (70 AFS GFN) is added to the system to increase the amount of fines (particularly 140 mesh). This controls the casting surface finish and balances the core sand going into the system to maintain the required AFS GFN. Besides bank sand, new sand is not added to the system other than the large amount of sand entering from core breakdown. A mechanical system was set up to remove core sand from a shakeout deck.
The bond for the molding sand is added in the form of a pre-blend containing Western bentonite, Southern bentonite and seacoal. The system operates at a methylene blue clay level of 6-7%. Loss on ignition (LOI) is maintained between 3.2-4%. The fineness of the washed molding sand is 55 AFS GFN. New Haven maintains the balance of working bond vs. total bond to guard against expansion defects.
In order to expand its product mix to accommodate brake rotor production, New Haven required an upgrade to its sand system that was focused on increasing sand flowability by optimizing sand properties and reducing variations. Brake rotors require high sand flowability for optimal casting quality. The molding sand, when operated at the target levels of 6.6% methylene blue clay and 38% compactability, produces optimum casting quality. When clay level or compactability run above current specifications, the flowability is not optimal for producing molds. With this in mind, New Haven determined that the critical areas to control in its sand system were the water addition, preblend addition (bond and carbon), bank sand addition, core sand influx and the working bond.
Sand Control Program Implementation
The first step in the sand quality improvement program was to ensure collection of the necessary daily (hourly) and weekly sand test data using proper testing procedures. Once the necessary equipment and testing were in place, the foundry's sand technicians were trained to ensure reliability of the data collected. With equipment and operator variations reduced to a minimum, the sand test data then can be relied upon to reflect the true variations in the sand required to control it. While conventional sand testing methods (other than compactability testing) do not allow for optimal real time control of the sand system due to the time required to complete the tests, one must realize that careful analysis of the sand testing data produces useful information to assess overall sand system performance.
With this data collected, any major sources of variation can be identified. Once they are identified, the sand system can be upgraded to eliminate the major sources of variation, and eventually to fine-tune the sand system. While on-line control systems such as those for compactability control and bond addition are necessary, a good sand laboratory is still recommended to verify and complement these systems.
New Haven Foundry began to employ printouts in the lab and developed a computer-based on-line monitoring system for the sand system controller results in house. This PC monitoring is performed in the sand room and in the plant metallurgist's office. The foundry focuses on the following for monitoring:
* control of the water;
* control of the bond and carbons addition;
* control of the new and return sand additions.
Inadequate water control is one of the major causes of green sand-related casting defects. New Haven utilizes mullers equipped with controllers that are sometimes unable to timely finetune water additions. The foundry attempted to study and understand the shortcomings of the control system and set in an offset in one of the controllers to provide consistent sand properties (compactability and strength) from both its mixers. Since the water and bond additions were based on the controller readings, this step was critical to fully understanding the control systems and making the necessary adjustments to suit the individual mixers. This was verified by the sand test data gathered by the sand laboratory. The problem was that the foundry still had a divergence on the controllers. Due to production constraints in the past, the foundry accepted this divergence as common and made adjustments as best as possible even though the laboratory results indicated that the compactability and green strength properties trom eacn mixer were different.
However, during this sand optimization project, this issue was revisited after comparing additional laboratory tests to the results on the controllers. Compactability of both mixers needed to be verified and controlled based upon laboratory data, and the controllers needed to be offset to produce similar laboratory results. Now, the compactabilities displayed on the two monitors of the controllers are different, but they produce the same laboratory compactabilities and green strengths. The laboratory tests are more accurate and should be the baseline.
Water Cooling PLCs
With any automated controller, certain criteria must be met for the system to function optimally. If the return sand entering the mixer varies greatly in the amount of clay, temperature or moisture level, most controllers are not capable of producing perfect sand on every batch. Depending upon the set up of the controller, multiple batches of the sand may need to be produced to even out the variations and return the sand to a target level.
At New Haven, some variation in bond and carbon level occurred when running jobs without cores vs. high-core jobs. In addition, sand temperature was not always consistent, which also could throw off the controller.
One of the problems with sand temperature arises when the rotary screen of the return sand becomes plugged with wet sand. (This happened at New Haven when one of the molding lines was not running and the sand cooling system added water to the return sand as if both lines were running.) When this happens, the foundry would alert the sand room that the sand was too wet, and the sand room would reduce or turn off the cooling water. But this results in hot sand entering the mixers, and the controller then would need to readjust itself again.
To eliminate this problem, the foundry installed a PLC system on the water cooling system that takes into account the depth of the sand on the belt, upstream sand temperature (before water addition) and downstream sand tempeature (after water addition).Since initial implementation and tweaking of this system, plugging of the rotary screen has been minimized. In addition, significant improvements in compactability control, consistency in bulk density, moisture, permeability, green compression and methylene blue clay content of the sand have been realized (Table 1). Comparisons of the 180 sets of data before and after the PLC system show reductions in standard deviations up to 40% for these properties. Histograms (Fig. 1) show how the repeatability of the sand system properties (as it relates to compactability) increased dramatically with the addition of the PLC.
Pre-Blend Addition System
As system performance was assessed further at New Haven Foundry, changes also were required on the pre-blend addition system. The system was originally a screw feed mechanism that did not allow for accurate additions. Weigh adders were installed on both mixers to ensure more accurate amounts of bond were added to each batch.
Laboratory testing also discovered that the seacoal level of the pre-blend was too low for a sand system producing iron castings. The result for the foundry was a rougher casting finish then it desired. To compensate, New Haven had to shotblast its castings a second time, increasing cost and reducing throughput time. A simple change in clay/carbon ratio in the pre-blend brought the LOI level up to 3.8% and minimized the need to re-blast castings.
Core Sand Dilution
Early during the development of a sand control program, New Haven also was experiencing problems with drawing deep pockets and mold breakage. The problem with drawing pockets was due to brittle sand from excessive core sand dilution, Friability and cone jolt toughness tests confirmed this problem.
Initially, when the molding sand was first tested for these properties, friability was 18% and cone jolt toughness was 25 jolts at a methylene blue clay level of 6%. According to past research performed on core sand dilution, more than 10% friability is considered surface brittle for a molding sand and can lead to erosion- and inclusion-related problems. In addition, less than 40 cone jolts is considered a brittle sand from a bulk brittleness standpoint, and makes the sand prone to difficulty. Gradually, New Haven increased the methylene blue clay level to its current 6.8%, producing 35 jolts and about 8-9% friability (Fig. 2). The intensive mixing provided by the sand mixers along with additional cumulative mulling through core sand removal has made this possible.
Core Sand Removal
Although the friability of the sand was satisfactory, cone jolt toughness was still less than 40 jolts at times (especially when methylene blue clay was too low), indicating a brittle molding sand. To combat this problem without raising the clay level too high (and reducing flowability), the foundry worked to remove core sand from the system at shakeout.
Core sand dilution is a major problem for many foundries as today's core binders break down rapidly and thoroughly at shakeout. Depending on the binder system, degree of burnout and mold atmosphere, the core sand grains can be coated with a layer of residual binder and/or lustrous carbon. Residual binder on the core sand grains is not always a problem, but if the coating is lustrous carbon, the clay may have difficulty adhering to the grain when this core sand cycles back into the mixer. The lustrous carbon acts as a lubricant making it difficult to bond the sand with the bentonite. In addition, decomposition of the core binder when exposed to heat in the casting process causes lustrous carbon and other organics to be absorbed into the clay interlayers, causing a "waterproofing" of the bentonite that interferes with normal hydration and dehydration in the molding/casting process. Even in the case of core sand grains with no residual coating or a residual coating that does not interfere with bonding, if the a mount of core sand input is excessive, brittle sand results simply from too much new material input into the system, which results in less cumulative mulling.
At the start of the optimization, the core sand dilution into the molding sand system was estimated at 600 lb/ton or double the recommended level (300 lb/ton of metal poured for iron sand systems) of new material (as new or core sand). To combat this, perforations were made in the shakeout deck to selectively remove residual core sand. The particular spot chosen was determined by taking samples from the beginning of the shakeout deck to the end and testing methylene blue clay level. At the first point in the shakeout deck, the methylene blue clay level closely matched that of the molding sand, but further down the length of the deck, methylene blue clay was less than 0.2%, and the sand grains appeared coated with a black/silvery layer of lustrous carbon. Perforations then were made in the shakeout deck at this point to remove this sand from the system into a bin that is weighed and emptied routinely.
Although these perforations are helping to reduce core sand dilution; the system must be continuously maintained by cleaning the perforations when they become clogged. Future plans involve placing a perforated slide gate on the deck to vary the level of core sand removal depending on the level of cored jobs being run.
Beyond sand system performance, the success of removing core sand from the sand system also has had a positive affect on the foundry's bottom line. Molding sand is no longer removed at shakeout with core sand (reducing disposal costs), so the green sand will receive the benefits of cumulative mulling. As a result, the foundry has been able to reduce its new bank sand additions to the sand system, which were used to adjust the AFS GFN to the required level. The bank sand now is added primarily to control the permeability and to help the sand flow into deep pockets.
Dust Collector Cleaning
New Haven also began regular monitoring of its dust collector system (on the sand system) and buildup in its tubes.
Previously, no set frequency to monitoring the dust collector existed. The collector tubes would become clogged after extensive use and fines would not be pulled from the sand system. As a result, the permeability of the green sand would decrease.
New Haven has set up regular monitoring of the dust collector system to coincide with its 140 screen for determining bank sand additions. This has provided further freedom to control permeability.
Control of Incoming Material at New Haven
Another way New Haven Foundry is ensuring a controlled molding sand system is by verifying the properties of incoming raw materials. For incoming sand, the tests include 25-micron clay content, AFS grain fineness number and distribution, acid demand value (ADV) and pH. For the base Western and Southern bentonites and a sample of the pre-blend, tests are performed for soluble and leachable calcium and magnesium. These tests determine whether the pre-blend is being properly prepared. Most foundries neglect these tests on incoming bentonite and/or pre-blend. Some may check methylene blue requirement/ion exchange capacity, but ion exchange capacity and/ or clay level checks on a pre-blend are not adequate indicators of consistency.
Many bentonites have similar ion exchange capacities as measured by methylene blue, yet they perform quite differently in the molding system. For example, while a Western and Southern bentonite can have the exact same methylene blue requirement, the Southern bentonite will develop faster green strength (less mulling energy required) and will produce lower hot strength and durability than the Western bentonite. This is due to a difference in exchangeable ions, which are measured with the calcium and magnesium tests. The important point is that by measuring soluble calcium and magnesium and leachable calcium, bentonites and pre-blends can be checked in a manner significant to the properties they will produce in the molding sand system. If these values are constant, the bentonite or pre-blend can be considered consistent, and can be expected to perform consistently in the sand system. This control is useful when adjustments to the pre-blend composition become necessary with the changing sand to metal ratio brought about by the changing product mix.
Table 1. Comparison of Properties Before and After PLC Was Installed on the Water Cooling System Properties Before PLC (6/22/00-8/7/00) Average Permeability (AFS permeability) 108.97 Green Compressive Strength (psi) 20.85 Moisture (%) 2.89 Methylene Blue Clay (%) 6.59 Compactability (%) 38.96 Weight of Compactability Sample (lb) 222.96 Properties Standard Deviation Permeability (AFS permeability) 7.18 Green Compressive Strength (psi) 1.5 Moisture (%) 0.23 Methylene Blue Clay (%) 0.37 Compactability (%) 4.75 Weight of Compactability Sample (lb) 17.32 Properties After PLC (9/1/00-10/16/00) Average Permeability (AFS permeability) 109.31 Green Compressive Strength (psi) 20.93 Moisture (%) 2.81 Methylene Blue Clay (%) 6.57 Compactability (%) 38.05 Weight of Compactability Sample (lb) 227.23 Properties Standard Deviation Permeability (AFS permeability) 5.59 Green Compressive Strength (psi) 1.22 Moisture (%) 0.16 Methylene Blue Clay (%) 0.32 Compactability (%) 2.82 Weight of Compactability Sample (lb) 10.87 Properties Change in Standard Deviation (%) Permeability (AFS permeability) -22.14 Green Compressive Strength (psi) -18.67 Moisture (%) -30.43 Methylene Blue Clay (%) -13.51 Compactability (%) -40.63 Weight of Compactability Sample (lb) -37.24