Innovations in controlling the lost foam process.

A newer understanding of variables - and how to control them - holds future promise innovative casting method.

Lost foam (LF), the process in which a refractory coated expendable foam pattern is surrounded by unbonded sand and "replaced" by the liquid metal during pouring, has found the majority of its limited use in captive engine operations.

One explanation why the potential of the 30-year-old process has not yet been realized is the long technical learning curve and the need for more established process controls.

"LF can produce dimensionally accurate and precise castings," wrote Charles Bates in Precision Expandable Pattern Casting Technology (Final Research Report Oct. 1, 1990-Jan. 31, 1993). "But it can also, without proper control and attention to process detail, produce warped, distorted and defective castings."

LF, viewed by some producers as the process of the future, offers tangible cost advantages that peak t.he interest of both casting purchasers and foundries alike. These include part consolidation and less machining costs, and relieved environmental disposal problems due to the use of unbonded sand.

According to Stratecasts, Inc., 64,000 tons of LF castings (38,000 in aluminum and 26,000 in iron) will be produced this year. By '98, the process is expected to grow to 90,000 tons (50,000 in aluminum and 40,000 in iron). And as process sharing and control technologies emerge, proponents believe the versatile casting method may grow further.

"LF is a controllable process," said Bruce McMellon, Vulcan Engineering Co., "but you must control the things available to you."

According to Robinson/Bodine-Robinson Foundry's Mike Johnston, "Almost every new foam project involves doing something no one has ever done before. You never know all you need to know, so use information and resources where you can find them."

At the AFS Lost Foam Casting Conference held in Akron, Ohio, in September, 1995, several innovations were presented to help control LF variables and learn more about their behavior. Among these were new tools for monitoring variables, and a recent discovery about molding media.

Coating Permeability

Coatings and their permeability have been well recorded as a critical factor to the success of LF. According to Charles Bates, Univ. of Alabama-Birmingham (UAB), "A LF pattern coating must allow gaseous and liquid pyrolysis products to escape the mold cavity in a timely manner since many internal and surface casting defects are due to the insufficient removal of these pyrolysis products." Erratic and inconsistent coating behavior, as noted in the Cast Metals Coalition's Spring '95 newsletter, has caused unacceptably high scrap rates in LF foundries.

General Motors, the largest U.S. LF producer, developed a simple, real-time system (using equipment already present in most foundries) to accurately check a LF coating's permeability.

GM's Gerald Kocan noted that coating permeability has repeatedly proved to be a primary factor in governing mold filling, and must be carefully controlled. Currently, coating variables such as percent-solids, viscosity, Baume and deposit weights are included in coating control programs and additional checks on pH and temperature are made in an effort to assure uniformity. However, these don't account for the makeup or morphology of the solids. Therefore, a real-time direct method measurement is the only reliable way to assure its consistency.

With a goal of designing a permeability measurement system to obtain a valid sample and measure it with reasonable accuracy, GM discovered a way to directly measure coating permeability with existing foundry permmeters. Used in sand labs, these devices measure the degree to which a standard, porous sand sample allows a known amount of gas to pass through it. Using this device for coating permeability, they reasoned, wouldn't require extra training, nor sacrifice accuracy or process relevancy.

The best coating sample would be obtained by flaking a coating sample off a pattern, but it would hinder manufacturing process control, noted Kocan. GM officials concluded a 100 x 100 stainless steel wire mesh disc of 2-9/16 in. diameter would be an adequate support medium for obtaining the sample, since the disc itself didn't register any permeability resistance.

The coating located in the pores of the mesh was assumed to be the effective check point of permeability measurement. An extrapolation of the 30.3% estimated open area of the wire mesh translates to 0.95 sq in. of coating being tested from the 2 in. diameter of the permmeter opening.

After mixers are turned off and swirling subsides, 3-5 discs - tagged with masking tape and hung from a spring clip - are lowered into the coating. After removal from the coating, the discs drip-dry for 10-15 min before being placed in a drying oven.

The adapted permmeter for coating permeability measurement is shown in Fig. 1 and 2. An adaptive flanged cylinder and cap were built to seal the sample between two soft neoprene rubber gaskets of 1/16 in. Locating pins on the cylinder flange help position the cap and gaskets accurately and two 1 in. spring clamps positioned opposite each other on the flanges provided the compressive force to form the seal. The opening of the cap allows for a visual check of the upper gasket placement in reference to the opening's perimeter.

Gauge validation trials were conducted on electric and floating drum permmeters at GM's LF foundry in Massena, New York and at the Advanced Development Laboratory in Saginaw, Michigan, respectively.

Using the Shainin Isoplot method, the trials, performed on both aluminum and iron coatings, showed that the high-low range of perm values for most sample conditions fell within 10% of the nominal reading for most group samplings, which proved this system as a tool for monitoring coating permeability. The typical turnaround time was only 30 min from dipping discs to perm results for a five-piece group; about the same length as a percent-solids test.

"Effective process control requires that attention be given to the critical factors influencing casting quality, both directly and indirectly," Kocan said. "Coating permeability, a process variable exerting considerable influence on mold filling and resultant quality in LF, can be monitored effectively in a control format without any inconvenience for manufacturing operations."

Process Control Tools

As a result of research conducted at UAB funded by AFS, the U.S. Dept. of Energy and 26 industry sponsors, new instruments have been developed to combat LF variation in three critical areas, according to Bates.

Pattern Coating - "The largest problem facing foundries in pattern coating consistency is developing an accurate method for measuring and controlling the air permeability and liquid absorption capabilities of refractory coatings," said Bates. In addition to GM's system for measuring coating permeability comes two other methods from UAB.

The air flow method uses a pressure regulator, two pressure transducers, a flow meter and a specimen holder to measure the air flow rate through a coating sample as a function of the pressure differential over the sample. The liquid flow test measures the liquid absorption rate into a coating sample as a function of time.

The values obtained from these tests can determine quantitative values for the air permeability and liquid absorption capabilities of refractory coatings, making it easier for foundries to control the properties of this critical variable.

Precision Pattern Production - Variables in type of polymer, expanding agents, steaming time/temperature, mold cooling cycle, pattern ejection time/temperature and pattern stabilization time/temperature all affect dimensional and precision accuracy of patterns. Precise pattern production depends on accurate measurement of dimensional changes, and because patterns are easily deformed, conventional measurement devices are difficult to use.

An air gauging system, using two air chambers and two critical flow orifices, was developed. Through changes in air pressure, the system accurately measures dimensional variations - without contacting the delicate pattern. This device also predicts foam pattern shrinkage as a function of time.

Sand Fill/Compaction - After coating, drying and placement in the flask, the pattern must be engulfed by densely compacted sand so that its surface area is surrounded and can support the coating during pouring. To fill cavities faster, high vibrational energy levels are often used. However, they can also contribute to pattern distortion.

A sand fill/compaction gauge has been developed that relies on the dielectric properties of the sand to gauge bulk density. Its sensor produces an appropriate voltage proportional to the amount of sand between the plates. A distortion strip, consisting of a stainless steel shim stock instrument with thin-foil type strain gauges has been developed that, when attached to the pattern, can determine the direction, extent and timing of distortion to a pattern during compaction.

These tools enable foundries to optimize compaction cycles to minimize fill times and isolate and minimize causes of pattern distortion. Patterns with difficult-to-fill internal cavities can be successfully instrumented with fill/compaction gauges to provide a measure of cavity-filling efficiency.

Dimensional Precision

Mercury Marine, a LF producer of aluminum cylinder blocks, believes its recent discoveries about LF sand media for large castings is one of "the most significant advancements in LF since unbonded sand was patented in '64."

During testing on the manufacturer's engine blocks, they discovered the drawbacks of silica sand, and realized the benefits of replacing it with a low thermal expansion coefficient sand of high heat diffusivity, such as chromite sand.

While one might not expect sand temperature to greatly influence casting dimensions, said Raymond Donahue, they discovered that in LF, hot sand makes "big" castings and cold sand makes "small" castings.

Silica is the most widely used molding material because it's readily available and inexpensive. However, as Donahue noted, silica "reduces the precision that can be delivered to that of green sand casting because at 100F (37.7C) below the aluminum's melting point, the silica sand goes through a phase transformation to a lower density sand. The volume of sand that goes through this phase transformation causes gross sand movement. Most of this movement is in the direction parallel to the axis of the flash and therefore the casting dimensions are relatively larger in this direction than in the perpendicular directions."

The importance of the sand temperature (using silica sand as an example) on the precision of a casting is illustrated in Fig. 3. The curve shows the average dimensional measurements of a LF engine block in inches as a function of the temperature of unbonded silica sand. At an 80F (26.6C) ambient temperature, the average engine block dimension was 9.53 in. As the silica's temperature increased to 160F (71.1C), the average block dimension also increased to a value of about 9.59 in., or an increase of 0.06 in.

Mercury Marine also found that the leak tightness of cast engine blocks produced using silica differs with the sand temperature. Using silica at 80F, for example, the leak rate for an in-line three cylinder engine aluminum block is three times more than that observed when using silica sand at 130F (54.4C).

Yet silica can't be successfully used at 130F because heated sand produces larger, unacceptable casting dimensions, and is thus normally used at 80F. At this temperature, however, it has been necessary to impregnate the cast block - sometimes as many as three times - to enable the cast block to pass the leak tightness requirements.

To produce castings with proper precision, Donahue concluded that LF sands must have a heat diffusivity greater than 1500 J/[m.sup.2]/[degrees] K/[s.sup.1/2] and a total linear expansion from 32F-2912F (0C-1600C) of less than 1%. Chromite, silicon carbide, mullite, carbon and olivine sands all have these physical properties.

In addition, Mercury Marine determined that these "precision" sands should have an AFS GFN of 25-33, with an AFS permeability number of 450-500. This sand is coarser and differs from traditional silica sands in fineness (silica's normal 40 GFN), and permeability (traditionally 300 AFS perm).

Shown in Fig. 4A and 4B are the results of one test using the silica sand and the low-expansion carbon sand at 80F, respectively. In Fig. 4A, the center line positions are widely scattered, resulting in engine blocks that couldn't be adequately machined. The scatter measurements of Fig. 4B (showing the same casting and location with carbon sand) is only a fraction of that in Fig. 4A.

In contrast to silica's problems at higher temperatures, using a low-expansion sand at 120F or above with the above mentioned physical properties produces leak-tight engine blocks in both the in-line three-cylinder design and the V-6. Both designs are dimensionally predictable without clean-up problems in any of the bores after machining.

They also discovered that to combat variations in sand temperature, the dimension of the pattern can be changed through aging or selecting another foam bead type. For example, a larger pattern might be desired in the winter to compensate for lower sand temperatures.

While originally qualifying carbon sand, they switched to a synthetic mullite due to carbon sand's supply shortages, and have had favorable results for more than one year, expecting all future blocks to be produced in the same manner.

Commenting on the results of the low-expansion sand, Mercury Marine/Brunswick Corp.'s Jack Reichert said that "if glued assemblies are made to the dimensional targets they are supposed to hit, the precision LF process is the most dimensionally correct and precise way to make castings of engine block size."

He also added that this sand significantly lowered machining scrap on the blocks, and gave the foundry the flexibility of moving to higher temperature sands to eliminate liquid styrene defects trapped in the casting - without affecting dimensional accuracy.

"The 'new' sand system should be desired by everyone using LF," said Reichert, "because its thermal properties significantly reduce sand expansion and thus improve the precision of LF."

RELATED ARTICLE: Lost Foam Facts

Lost Foam Foundries-North America: 117 (1995-96 Casting Source Directory).

Year Developed: bonded sand patent - 1958 by H.F. Shroyer. Unbonded sand patent - 1964 by M.C. Flemings.

Advantages: elimination of cores and parting lines, improved dimensional accuracy, part consolidation and replacement of bonded sand.

Niche: complex parts of high volume, with complex internal cavities that are machined, and assemblies that can be combined into one cast component.

Alloys Poured: Aluminum, brass, bronze, cast iron and steel.

Sands Used: silica, zircon, chromite and olivine.

Pattern Materials: expandable polystyrene (EPS), expanded polymethyl methacrylate (EPMMA), expanded polyethylene (EPE), expanded polyalkylene carbonate (EPAC).

Environmental Impact: only source of emissions during pouring is pattern degradation, and no spent binder or core butt problems exist. The possible emission of volatile organic compounds during casting cooling is avoided (T.S. Piwonka, Univ. of Alabama-Tuscaloosa, 1993 AFS EPC Conference).

Key Equipment: quality compaction table and rigid LF flasks. Heavy duty equipment provides best consistency.

RELATED ARTICLE: Lessons Learned in Lost Foam

"We've learned to be more selective of the parts we put into the process. We also cast in sand and permanent mold and are more inclined not to force a part that probably doesn't fit the parameters of LF. We look for castings that would require multiple sand cores and machining features that can be cast. We are extremely cautious of patterns that require gluing for parts that are pressure vessels. If the part geometry doesn't fit this scenario, we look at other processes. We've learned to keep the process simple. We don't over-engineer the part or process. We don't try to make the sand do things our compaction table isn't capable of. We use only two coatings, and measure only percent-solids and Baume. We've learned not to cast features to tolerances that aren't possible. We know what works for us. We don't quote or accept jobs without a high confidence level of success."

- Mark Morel, Morel Industries, Entiat, Washington

"We've learned that LF is more sensitive to change than other foundry processes. Scientific SPC applications, diligently observed, are essential. Generally, the process yields a product superior to that of other casting processes."

- Roy Shores and Sid Tankersley, Mueller Co., Albertville, Alabama

"We haven't been too successful in improving our operation with LF. We're still at a stage where a contribution is required from our sand castings. That situation, however, isn't due to technical problems primarily - we find it is surprisingly easy to produce beautiful LF castings. Opening the market is much more difficult. As a small jobbing foundry pouring aluminum and copper-based alloys, it's extremely difficult to compete on existing parts against standard casting methods. Operating in nonautomotive markets where designers have been much less exposed to the potential benefits resulting from the use of LF, we find that marketing poses the greatest challenge to our success or failure. We're learning that most existing castings are not interesting prospects for conversion to LF - it must start right at the design stage. LF has much going for it. This fact, coupled with the commitment to success brings confidence that shinier days lie ahead."

Georges Theriault, Fonderie St-Romuald, St-Romuald, Quebec, Canada

"The biggest lesson Saturn has learned in the production of quality LF castings is synergy. Without the synergy from the company president to the operating technician on the floor, the product designers to the manufacturing engineers, the supplier community, financial and other service staffs, our endeavor wouldn't enjoy the success we do today. The LF technology is a process containing many parameters. Synergistically learning the effect one parameter has on another, then balancing or optimizing those parameters to obtain the highest quality products, allowed us to become the industry standard since our inception five years ago. The biggest lesson learned that impacts quality casting production is coating. Understanding and controlling the effect coating has on your process is mandatory."

- Jim Deppler, Saturn Corp., Spring Hill, Tennessee 

"The biggest thing to watch, even more than the pattern, coating, etc., is your sand. If you don't maintain the permeability of the sand and allow the LOIs to build up, you'll start making junk castings. The gas has to get out. Subsurface defects will start showing up in machining, and if it isn't corrected you'll get surface defects and eventually nonfill. Coating can be a problem at times, but we always diluted our coatings with distilled water, and avoided any variables with city water. If you keep your sand up to snuff, you'll make good castings. Generally, sand causes defects."

- Warren Robertson, Cedar, Minnesota

"Precise and process capable tooling is extremely important. Without a good tool, you won't make good foam patterns...In both gray iron and aluminum, we've found most dimensional, metallurgical and cosmetic defects are associated with nonuniform density in the patterns. The foundry should work with the foam molder to develop a process that yields foams that make good castings, not just 'pretty' foams. Besides good, clean metal, nothing is more important than understanding and adequately controlling the coating. The coating process is so critical we will soon implement robotic dipping to improve the consistency of the coating procedures, which may potentially be the most dramatic process improvement a foundry can make."

Mike Johnston, Robinson/Bodine-Robinson Foundry, Alexander City, Alabama