Cooling and Shakeout Emissions Control: Practical Solutions to Real Problems.

A system audit performed at GM's Defiance, Ohio foundry provides practical solutions to emissions buildup in ducts, hoods and dust collectors.

Sand molding and coremaking operations in the foundry present a unique set of problems for the designers and operators of emissions control systems. Ever-tightening standards have required foundries to address and solve the particulate matter emissions problems in the pouring, cooling and shakeout areas through a variety of methodologies, each with varying degrees of success. The installation of medium-energy scrubbers, cartridge collectors, baghouses and high-energy scrubbers have not always solved the problem of high opacity from these sources.

Many of the control systems installed have a serious problem with the buildup of flammable materials in the hoods, duct systems and collectors. The plugging of equipment and the fires that occur in these systems due to the buildup present hazards to the operators and result in high operating costs for these systems.

This article provides information on solutions developed to address these emissions control problems based on work performed at GM's Defiance, Ohio green sand foundry's pouring, cooling and shakeout areas. This project was undertaken in response to GM Defiance's realization that it couldn't keep the bags in the collectors on its mold cooling and shakeout stations from premature plugging with the rapid buildup of these condensable organic materials in the collectors. The maintenance costs associated with short bag life are driving the efforts to reduce the buildup to manageable rates (see sidebar "GM Uses Batch, Continuous Seeding to Improve Dust Collector Filter Life" on p. 34).

Emissions Generation

As the metal is poured in the mold, the rapid increase in temperature drives reactions that generate emissions of a complex mixture of organic compounds. The use of seacoal in the green sand mold mix provides gas generation and reducing atmospheres within the mold crucial to casting quality. Partial pyrolysis of coal generates a true "vegetable soup of organic compounds, which either condense within the green sand, vent from the mold or release at cope pickoff and shakeout of the castings.

Additional organic matter emissions are generated from the use of cores and a number of green sand additives. These additives and compounds include flour, wax, mold release agents, the partial or complete products of resin, carrier and additive combustion, and the mix of core organics. The amount of additives and compounds will vary with amount of core material used in the mold.

In the pouring area, mold emissions can be characterized as metal oxides, soot, and small amounts of sand and organic materials. In the cooling area, emissions can be characterized as smoke, condensed organic materials, products of combustion and steam. In shakeout, emissions can be characterized as smoke, sand, dust, metal oxides, steam and condensed organic materials.

In ventilation systems, smoke is defined as particulate matter less than one micron in diameter. For the systems studied for this article, the visible portion of the emissions from the mold and cores fits this definition. It is believed that some of this material is an aerosol and some of it is a mist, either of which can fit this definition. The subsequent configuration and operating characteristics of the molding line and the capture and control system will then determine the relative severity of the problems related to the emissions generated.

For green sand molds that do not require cores, there typically are no vents in the molds and little buildup in the hoods and duct system. The introduction of cores into the mold typically requires the addition of vents to prevent gas buildup within the mold. The number and placement of vents is designed to relieve the gas buildup, as required for good casting quality.

The organic materials driven out of the cores vary with core type and resin supplier. In a comparison of the hotbox and coldbox cores, the amount of solids left in the core for binding the sand will be assumed to be equivalent and the carrier solvents will be assumed to contribute to the gas generation. The use of hotbox cores theoretically produces fewer emissions in the molding operations because the organic solvents are driven out in the coremaking process. The use of coldbox cores introduces more petroleum-based solvents into the molding process, which will mix and become soluble with a portion of the organic material driven out of the green sand.

The introduction of larger core packages limits the amount of green sand that is heated to high temperatures by direct contact with the iron, This results in a shift in the fume generation mechanism and a shift in the amount of condensable organic generated. This shift may appear to be an increase because the physical character of the solvents driven out of the cores is quite different from the volatiles driven out of the green sand. Whether a true increase in condensable organic exists or not, many operators report an increased organic buildup with increased coldbox core usage.

Duct Fires

Fresh products of the pyrolysis reactions are not only "smelly" and "sticky," but are highly reactive in air and readily condense and polymerize. As the materials are deposited in the duct, they are soft and runny, but as they sit, these coal/tar-like substances polymerize to form hard coatings. These coatings are harder to ignite than the fresh condensate because of their low vapor pressures, but once they are ignited, the burning layers of built-up condensate develop a lot of heat and smoke.

This coating generates a number of serious problems for the foundry operator. Fires resulting from an accumulation of this material that gets ignited can be quite severe and difficult to extinguish. These fires can be ignited from transient temperature spikes, burning foreign materials in the system, hot work on or in the vicinity of the duct system, static discharge, or the heat generated by the polymerization reactions.

Duct fires are currently fought with water hoses and with permanently piped water sprinkler or deluge systems to douse and flood the fires. This method is messy and can be somewhat dangerous depending on the duct configuration and access. It has been reported that carbon dioxide gas has been used with limited success to initially douse the fire, however, the fires reportedly flare up again as soon as air is reintroduced.

Reduction of the buildup and avoidance of the fire is clearly the best solution, but limited information exists on a means to reduce buildup. Adsorbent materials co-injected into the gas stream appear to mitigate the buildup and, in some cases, the inclusion of nonflammable materials in the buildup that does occur may help to reduce the severity of the fires that do occur. Also, mechanical cleaning (scraping, chipping and water-blasting) does not work on organic materials once they form a solid coating on the duct walls.

Another method used to clear this buildup is the controlled burning of the buildup in the ducts (with the pre-notification of your environmental agency). If properly monitored and controlled, this method is effective and can be done without serious damage to the duct system. However, the addition of control devices onto the duct systems renders this control strategy impractical since the controlled burning could jeopardize the control device.

Organic Buildup in Ducts, Hoods

Observations made inside the cooling tunnels on green sand molding lines indicate that the amount of organic coating in the hoods and the ducts tied to those hoods is directly related to the amount of visible smoke emitted from the molds under the hoods. The emission of smoke from a green sand mold is directly related to the amount of gas vented from that mold and the amount of smoke that is consumed by the flame present at the discharge point for the vented gases.

In a system where there are no vents from the mold, the only visible smoke emitted in the cooling process is from the pouring basin and the parting line of the flask. On systems where there are vents to relieve gas from the mold, there are usually flames present on the vents at the pouring station; but as the mold moves down the line, these flames eventually extinguish. The point at which the flames are extinguished has been observed to be the point at which the nature of the emissions changes from light "soot"-type material to a semi-liquid buildup.

Some cooling tunnels have been designed to maximize and distribute the open area of the gas off-take from the hood. This design modification was intended to reduce the burning of tramp materials being drawn into the duct system and the possibility of a fire. Observations made in these hoods also show a buildup of a sooty ash-like material instead of the wet organic material prevalent in other tunnels. It appears that the increased area of the off take has reduced the drafts within the hood to the point that the flames on the vents remain lit long. On other systems where these flames are blown out by the ventilation system, the amount of buildup is noticeably greater once the majority of flames are out. When this flameout occurs, it has a profound impact on the amount of buildup in the collection and control system from that point until shakeout.

Once there is a significant amount of organic material in the gas stream, the accumulation of organic material in the duct systems appears to be directly related to the temperature and velocity of gas drawn through the duct system. It is difficult to characterize specific physical properties of the organic material emitted from the mold. The condensation temperatures for the organic materials of interest may range from 100-500F (38-260C). As these materials condense, they produce the visible smoke from this process. At exhaust system temperatures [ambient to 200F (93C)], these compounds may be liquid or solid. This smoke is composed of fine particulate matter that refracts light to produce the visible plume.

In some industrial applications where organic buildup in the ducts has been a problem, the addition of heat to the duct system has greatly reduced the buildup and fires. The use of direct-fired gas burners is an effective way to heat a duct system, and in applications where an afterburner is used, represents little cost penalty. Typical temperature levels required to eliminate the buildup are in the 350-450F (177-232C) range.

For the foundry applications where the control device is a bag or cartridge collector, this is not a viable option. The heated organic materials would penetrate the bags and condense as a visible aerosol in the discharge stack. The ducts would stay clean but the opacity problem would return.

Another option to reducing the buildup in ducts is minimization of air flow. Since cooler ducts result in greater amounts of organic buildup, then one way to reduce the amount of material buildup is to run the duct system hotter. The most efficient way to accomplish this is to reduce the amount of air drawn into the collection system. Some designers of cooling line hood systems use a guideline of 200 CFM flow/linear foot of conveyor (CFM/ft). This rule of thumb results in excessive amounts of tramp air introduced into these systems. Calculation of the amount of air required to stay below explosive conditions for the specific mold lines encompassed by this study indicates that this process could be safely controlled by as little as 35 CFM/ft.

Wet Scrubber Performance

Wet dust collectors are installed on many applications in the foundry. The medium-energy, spin-vane-style scrubbers, operating at a 6-8 in. W.C. pressure drop, do an excellent job on particulate matter above 5[micro], but these scrubbers do not perform well on particulate below this size. These types of units do little to reduce the emission of the submicron particulate, which is responsible for the opacity seen from some foundry operations.

Higher energy scrubbers operating in the 12-16 in. W.C. range were studied and tested in a shakeout application. The results of this study indicate that a multi-throat venturi, followed by a mist eliminator is not capable of removing particulate matter below a few microns in size and does little to reduce opacity.

One installation of a high-energy scrubber 70-80 in. W.C. was reviewed. This system also is incapable of consistent performance in removing the opacity-producing particulate matter.

Fabric Filter Buildup, Blinding

It is a common practice to seed a new set of bags or cartridges in the collector prior to placing the unit in service. On some of the collectors installed on mold line applications, it has been found that periodic and even continuous application of seed material can greatly increase filter life and serviceability of the collector (see sidebar "GM Uses Batch, Continuous Seeding to Improve Dust Collector Filter Life").

For the fabric filters servicing the shakeout area with significant particulate matter loading, the filters tend to fail quickly due to the buildup of organic matter in the fabric. The organic buildup on the filters tends to wet the fabric of the filter, blinding the air passages through the filter media. After 69 months, total plugging of the filter occurs.

A layer of seed material can act as a collection surface and adsorb some of this liquid particulate matter. The layer of seed material will eventually become saturated itself and stick to the filter surface if not removed. Once removed by the cleaning cycle, the filter can be reseeded and the process repeated. In some instances, the increase in filter life has ranged from in excess of 100x to as low as a 4x increase in service life. This is not a "cure all" for all filter life problems. This increase in life is valid strictly for applications where the wetting of the filter with the organics has been observed.

The seed material used to date has been powdered limestone. This material has a high surface area for adsorption of the organics and is chemically inert. The operations studied utilize a wet dust removal and transport system, which services both the wet and dry dust collector systems. The use of materials that would become reactive in water has therefore been carefully avoided. On dry systems, the use of some other material may provide some cost savings.

Control of the seeding and cleaning of these collectors has been carefully coordinated. It has been found that in some severe applications that the filters can last as little as a few days before being totally blinded by organic buildup. The solution has been to seed continually in one case and to retard the pulse-jet cleaning in other cases until there can be fresh seed applied immediately after the unit is pulsed. The operating cost of the seeding systems is more than balanced by the reduction in filter costs and labor to change the filters.

GM Uses Batch, Continuous Seeding to Improve Dust Collector Filter Life

GM's Defiance, Ohio green sand foundry found itself faced with a real problem. In July 1997, it had installed two new dry dust collectors on its mold lines to control the emissions from cooling sections and soon realized that it couldn't operate them like the collectors serving other foundry operations. The problem was that the bags in these dust collectors were showing filter lives ranging from days to weeks. With the cost and scheduling problems associated with the reduced filter lives and the requirement of dust collector operation for regulatory purposes, a solution had to be found.

All casting operations are conducted using rotary mechanical iron pourers (RMIPs) and the cooling and shakeout operations are totally enclosed tunnel conveyors and enclosures. The fume collection from the RMIPs is controlled by fume collection hoods tied to either a fabric filter or a straight stack. Fume collection from the cooling tunnels is controlled by either fabric filters, medium-energy scrubbers or straight stacks. Fume control from the shakeout is controlled by medium-energy scrubbers, bag houses, cartridge collectors and a high-energy scrubber.

The dust collectors were 60,000 cfm baghouse with 750 spun polyester filters, on-line pulsing based on pressure differential and 5:1 air/cloth ratios. One collector covered the exhausts from the mold cooling room and tunnel, and the other just covered the tunnel. The first collector had a filter life of I month while the second lasted 4 days.

Within the first month the collectors were online, they displayed several problems:

* decreasing exhaust capacity due to reduced filter permeability from organic condensation;

* oily/tar-like organic buildup in the ductwork and interior walls;

* dramatic pressure differential increases resulting in filters that were rendered useless in as little as four days;

* filters adhered to cages being discarded with the bags.

An investigation turned up some possible solutions:

* elevate and maintain the air temperatures in the ductwork above the dew point of mold emission vapor;

* investigate and evaluate alternative filter bag designs and materials;

* incinerate emission vapor prior to introduction into the dust collector;

* investigate the role that mold combustion plays in organic condensation from vapor;

* evaluate dust collector seeding in continuous and/or batch operations. The choice was to investigate seeding methods in which hot oil vapors are exhausted and later condensed on collection filters. This testing could be performed at a low cost.

The seed is a ground limestone material (2-4 [micro]) with good adsorption properties and a minimal dropout problem with duct velocities of 3000-4000 fpm. The material also is inert, which allows disposal of it in a dust collector water system. This seed material is injected into the duct upstream of the dust collector, where it is mixed with emission gases. This mixture enters the collector and coats each filter bag. The layer of seed adsorbs organic condensation before it is adsorbed into the filter fabric. The pressure drop across the filters increases as the seed material is applied. The dust collector then must be pulsed. The organic "goo"/seeding material mixture pulses off the filter bags and falls into the screw conveyor. Then, fresh seeding material re-coats the filters and the process repeats itself.

A batch seeding operation was set up for the first collector. A small positive pressure fan was installed and ducted to the tie-in point in the dust collector ductwork. Then, the limestone is injected into the collector's airstream via the negative pressure side of the fan. The pressure differential actuated pulsing controls were set so that the collector would not pulse while it was online. At the end of each week, the fan is shut off and the collector is pulsed down to remove the filter cake. To prepare the collector for the next operation, the seed material was injected onto the filters until the pressure drop across the filters reached 4 in. indicating the presence of sufficient cake material for one week of operation. During the week, the fan is left on at all times to hold the cake in place.

A continuous seeding operation was set up for the second dust collector. A metered feeding system and fan were connected to the dust collector duct to provide a consistent amount of limestone. The pressure differential of the pulsing controls was set at a narrow range (4 and 5 in. SP) and the automatic metering feeder was used to inject the seed into the duct at a preset rate. The capital cost of this installation was higher than the previous method, but much of the manual labor was eliminated. The decision whether to batch or continuous seed a dust collector lies in the operating schedules, existing filter life, seed/utility costs and spent seed disposal costs.

The result for GM was that the filter life in the first dust collector went from 1 to 18 months and in the second from 4 days to 9 months. Evidence shows that filter life is limited only by the infiltration of the fines from the seeding material into the filter interstices, not solely organic buildup, as was previously the limit.