Clearing the air on emission estimates: over-estimating emissions can cost your foundry money.

The Federal Operating Permit Program (Title V) of the 1990 Clean Air Act (CAA) Amendment requires many foundries to compile comprehensive air emissions inventories. These Title V requirements have posed a problem for foundries preparing air emissions inventories. Foundrymen who don't have the answers turn to consultants for help with these issues, but they generally do not have all the answers either. Although the basic steps of compiling an air emissions inventory are relatively straightforward, the choice of which information to use - and when to seek alternate information - is not always clear. Developing and maintaining accurate and detailed emissions information can help you avoid regulatory compliance and legal liability problems, and can be an invaluable tool when facing the potential requirements for additional emissions restrictions. Following are two case studies that were presented at the ninth annual AFS Environmental, Health and Safety Conference in Indianapolis.

General Considerations

As an operator of three steel foundries across the country, ESCO Corp., Portland, Oregon has studied the effects of overestimating emissions and realized the benefits of critically evaluating its emission estimates. Working closely with consultants and state environmental regulators to develop comprehensive air emissions inventories for its facilities, ESCO "revisited" a number of original process emissions and found it was overestimating emissions. Subsequent refinements were made, which resulted in significant decreases in estimated facility-wide air emissions. According to ESCO Environmental Engineer Dan Cutugno, some of these decreases in turn had important effects on the potential regulatory obligation of its foundries.

Little information is available about air emissions solely pertaining to foundries. Commonly used emission estimates (AP-42, AIRS SCC) are expressed as an amount of a pollutant generated per unit of raw material consumed, or product produced.

"These emission factors are often reported as a range of values and can be traced to data of questionable quality," Cutugno said. "Like most other industries compiling Title V emissions inventories, the typical foundry relies mostly on published emission factors, unless source test data or other information specific to the subject facility is available."

Often, so much emphasis is placed on completing the Title V application that foundries fail to realize the importance of strategic considerations related to the reported emissions estimates. This is especially the case when foundry personnel rely on consultants to compile inventories, then have little or no involvement in the selection and rationalization of emission estimates, Cutugno said.

"In addition to the technical complexities of developing an accurate facility air emissions inventory, certain legal considerations also play a role," Cutugno said. "There seems to be a consensus that the potential legal liabilities related to under-estimating air emissions are so great that over-estimation of emissions is the safe way to proceed. While it is generally appropriate to take a conservative approach to an emissions inventory, it is also prudent to take advantage of opportunities for refining emission estimates based on site-specific knowledge, reasonable assumption and common sense."

The Trigger Effect

There are several potential adverse consequences to over-estimating air emissions. The first is that regulations that shouldn't apply to a facility may become applicable because estimated air emission levels "triggered" a particular regulation. According to Cutugno, over-estimation of volatile organic compound (VOC) emissions commonly results in an over-estimation of hazardous air pollutants (HAPs). Such a scenario could conceivably push a facility over the threshold for a major HAP source.

Another regulatory trigger that could be exceeded involves state-implemented Reasonable Achievable Control Technology (RACT) requirements for facilities that emit VOCs at annual rates that exceed limits specified in the particular regulation.

An additional negative impact of over-estimation involves permitting fees. The Title V fee structure is based on a dollar amount paid per ton of pollutant emitted. According to Cutugno, ESCO operates foundries in Oregon and Mississippi, where the current Title V permit fees are $31 and $16 per ton, respectively. By correcting its emission over-estimates, ESCO expects to save several thousand dollars per year in base permitting fees.

Examples of Over-estimation

Following are some specific examples of air emissions over-estimation, and the alternatives that resulted in more reasonable, defensible emission estimates for ESCO. The results of changes made are summarized in Table 1. The table shows that by using revised emission factors ESCO Corp. reduced its estimates by 20-95%, cutting down on costly permitting fees.

1. Carbon Monoxide Emissions from Electric Arc Furnaces - ESCO originally used AP-42 information for this source, expressed as pounds of carbon monoxide (CO) emitted per ton of metal poured (pptp). The AP-42 value is expressed as a range, from 1-37 pptp, and state regulators required that the maximum of the range be used in the emission inventory. Subsequent source tests conducted on one of ESCO's five electric arc furnaces indicated a range of CO emissions from 1.7-12.2 pptp. ESCO was again required to apply the highest value, but the new emission factor resulted in a decrease in estimated CO emissions of 66%. ESCO was also permitted to utilize the source test data from the one arc furnace to estimate CO emissions from three of its five furnaces. At the maximum production levels anticipated, this resulted in a decrease in CO emissions from 854 tons/year to 281 tons/year.

2. Particulate Matter (PM) Emissions from Pouring and Cooling Operations - The AP-42 factor for PM emissions from this source is 4.2 lb per ton of metal poured (pptp). In conjunction with a prevention of significant deterioration (PSD) analysis, ESCO conducted source testing on a roof vent to determine the actual amount of PM being emitted to the atmosphere from pouring and cooling. This testing indicated that 2.52 lb of PM per hour was leaving the facility through the roof vents. ESCO then used a conservative annual hours of operation estimate to derive the total tons of PM per year from pouring and cooling.

At the maximum production level anticipated for this facility, the revised emission factor resulted in an annual decrease in PM emissions from 61 tons/year to 14 tons/year.

3. Settling Factors in Particulate Matter Emission Estimates - Published emission factors often do not take into account that, depending on the process and the nature of the PM generated, a significant portion of the generated emissions can "fall out" by gravity. This should not be included with the remaining portion that is actually introduced [TABULAR DATA FOR TABLE 1 OMITTED] into the atmosphere. This is particularly relevant to indoor processes generating fugitive emissions, or to processes such as grinding or sand handling that can generate relatively heavy PM that readily settles out indoors. At least one important air emissions reference for foundries includes particulate settling rates for a number of common processes. When appropriately used with approval from a regulatory agency, these factors will have a significant positive effect on estimated PM emissions.

For example, ESCO applied a 20% settling factor to the AP-42 emission factor for scrap and charge handling. The literature-reported settling rate of 80% was not used because much of the handling is done outdoors, where the potential for wind-blown dust is greater. Nevertheless, the conservative factor chosen by one facility reduced estimated PM emissions by several tons per year.

4. VOC Emission Estimates Based on Mass Balance Calculations - "Chemical consumption" data in conjunction with Material Safety Data Sheet (MSDS) information is typically used to calculate VOC emissions from a variety of processes such as core washing, painting and miscellaneous chemical usage. This "mass balance" approach assumes that the fraction of VOCs present in the product is ultimately emitted to the atmosphere. There are several aspects of the mass balance approach that can contribute to over-estimation of VOC emissions. "The accuracy of this entire approach hinges on the accuracy of the consumption information. Steps must be taken to ensure that this information is as precise as possible, and that changes in consumption can be tracked closely," Cutugno said.

ESCO originally reported a maximum of 72 tons of VOCs per year from alcohol-based core washing. However, more detailed consumption information was later obtained indicating that the earlier figure was a significant over-estimation. The revised emission factor and resultant emissions estimate of 24 tons/year represented a decrease in estimated VOC emissions of 67%, and savings of approximately $1525 per year in permitting fees.

5. VOCs from Sand Reclamation - ESCO's original approach to estimating emissions from this source conservatively assumed that, after molding and coremaking, any organic compounds remaining in the sand (based on AFS Toxic Release Inventory guidance) survived the pouring, cooling and shakeout process. It also assumed that they were subsequently released in their entirety during sand reclamation. ESCO has subsequently concluded that this assumption is overly conservative for three reasons.

First, it is unlikely that volatile binder constituents in the sand would survive in an intact state at the extreme temperatures with which the sand is subjected to during the casting process. Second, ESCO's sand reclamation is non-thermal, and utilizes only physical processes such as screening and air classification. Consequently, any organic constituents that were not released at the high pouring and cooling temperatures would not be released during physical sand reclamation. Lastly, ESCO conducted organic vapor screening of the air exhaust of the sand reclamation baghouses, and significant levels of VOCs were not detected.

As a result of these considerations, the revised inventory assumes that only 5% (rather than 100%) of the organic binder constituents in the pre-casting sand are released during sand reclamation, Cutugno said. This resulted in a decrease in maximum annual VOC emissions from this source from 16.5 tons/year to less than 1 ton/year, a 95% decrease.

Calculation Emissions

An alternate method for developing emissions estimates is based on the Occupational Safety and Health Administration (OSHA) time weighted average (TWA) for the breathable fraction of particulates in the foundry that are [TABULAR DATA FOR TABLE 2 OMITTED] not otherwise regulated. By using the OSHA respirable dust limit and design air flows, foundries can produce results in emission estimates for specific processes that range from 4-200 times lower than estimates based on published emission factors. When the analysis of actual OSHA air sampling data is complete, the range is expected to be 200- 1000 times lower. This is the case at John Deere Foundry in Waterloo (JDFW), Iowa, a gray and ductile iron foundry with three automated molding lines. According to Randall McDougall, the environmental engineer at JDFW, the plant is expected to save about $21,650 each year in Title V emission fees using the OSHA limit.

Depending on the type of core process used, mold line pouring emissions may include benzene, phenol, C[O.sub.2], CO, NO, isocyanates and fine particulate matter (P[M.sub.10]). Because the off-gassing materials are being released over a relatively large area (the pouring floor), the capture and control of the emissions are neither required nor economically feasible. All mold line exhaust gases are therefore exhausted out of the building through general ventilation vents and fans. The total exhaust volume for JDFW is 312,300-320,000 standard cu ft/min (scfm). Shakeout and cooling emissions are essentially the same as the molding and pouring lines with the exception that the process covers even more floor space than the pouring operations. The total exhaust volume for shakeout is 32,000 scfm. The emissions for the cooling area range from 339,399-512,000 scfm. All of the emission from shakeout and cooling are exhausted out of the building through general ventilation. Table 2 shows how the emission estimation process at JDFW was simplified by using existing monitoring and volume flow data as opposed to tracking numerous production throughput streams.

"The fugitive nature of the JDFW's emissions, and the fact that they are emitted to the atmosphere through building ventilation exhaust fans made quantification using continuous monitor systems (CEMS) and stack testing impractical," McDougall said. "The mass balance method also was not practical because fumes emitted from the sources were created as a by-product of the outgassing from hot castings or the pouring of hot metal into the molds. The net result is that there is no reliable means of balancing raw material input with the remaining material when the production process is complete."

The unreliability of the aforementioned methods has caused foundries to rely on published EPA or other emission factors, which are equally unreliable, as they result in unrealistically high values. At JDFW, production throughput from each area is potentially counted several times to avoid under-estimating emissions. This "double counting" causes the facility to pay substantially higher emission fees than if the emissions were estimated more accurately. JDFW began seeking a readily available, more accurate and verifiable estimating method. After evaluating and rejecting various criteria, officials selected the OSHA respirable dust levels for "particulates not otherwise regulated" (PNOR).

The foundry discovered that if monitored respirable dust levels didn't exceed the 5 milligram per cu meter (mg/[m.sup.3]) 8-hr TWA limit on the floor of the foundry, then the particulate emissions exhausted through the building ventilation 37 ft above the foundry floor wouldn't contain any higher levels. JDFW already monitors respirable dust levels in the facility as required by OSHA regulations.

The 8-hr TWA is monitored by personal dust samplers worn by floor workers for an entire 8-hr shift. Because the TWA is an average for the 8-hr period, it is representative of average emissions on an hourly basis. While the actual monitoring data has not been analyzed, the emission factor development used the OSHA dust limit at 5 mg/[m.sup.3] as the worst possible concentration.

Emissions estimated using the TWA limit are calculated by multiplying the design volume flow of all building exhaust fans in the areas where the processes are located by the 5 mg/[m.sup.3] TWA limit. The product is then multiplied by the department's actual annual hours of operation. For example, in Mold Line 801 at JDFW, the total volume is 691,399 scfm, or 19,580 cu meters per min. In 1994, the same mold line operated a total of 4315 hr. By multiplying the total volume flow by the annual hours of operation and making the appropriate unit conversions, total P[M.sub.10] emissions for the three subject processes in Mold Line 801 were 27.9 tons per year. In this particular example, emissions will be 21.5 times lower than the published emission factors for the three Mold Line 801 processes, McDougall said.

The next step in the process of developing this emission calculation method is to refine the emission estimates by evaluating actual monitoring data to determine a more accurate actual emission rate. It is believed that this analysis will show that the fumes formed will be roughly proportional to metal throughput. Variability in binders and oils used for preparation of the cores and/or molds may cause some minor deviation from strict proportionality of emissions. If the assumption that fume emissions are proportional to metal throughput is proven correct, then actual monitored particulate concentrations in Mold Lines 802 and 803 will be substantially lower than actual monitored concentrations in Mold Line 801. Consequently, it can be said that more accurate P[M.sub.10] emissions estimation can be achieved using building ventilation flow rate applied to the OSHA maximum respirable dust limit, or the actual OSHA respirable dust monitoring data.

The science (and "art") of air emissions inventories continues to evolve. Foundries that rely on air emissions inventories for regulatory permitting and reporting purposes should periodically revisit and critique the methodologies being used to estimate air emissions.