

TECHNICAL MEMORANDUM

TO:	Bill Schrock, Allison Costa, U.S. EPA/OAQPS/SPPD

FROM:	Eastern Research Group, Inc. (ERG)

DATE:	August 2, 2018
--------------------------------------------------------------------------------

--------------------------------------------------------------------------------
SUBJECT:	CAA Section 112(d)(6) Technology Review for the Solvent Extraction for Vegetable Oil Production Source Category

This memorandum summarizes the results of an analysis ERG conducted on behalf of the U.S. Environmental Protection Agency (EPA) to identify developments in practices, processes, and control technologies that have occurred since promulgation of the National Emission Standard for Hazardous Air Pollutants (NESHAP) for the Solvent Extraction for Vegetable Oil Production Source category. This analysis is part of the EPA review efforts in accordance with section 112(d)(6) of the Clean Air Act (CAA). 

This memorandum is organized as follows:
      1.0	Introduction
      2.0	Background for the Solvent Extraction for Vegetable Oil Production Source Category
      3.0	Developments in Practices, Processes, and Control Technologies
      4.0	Control Technology Cost and Emissions Reductions
      5.0	Summary
      6.0	References
      Appendix A  -  List of Vegetable Oil Production Processes and Facilities
      Appendix B  -  Results of RACT/BACT/LAER Clearinghouse Query
      Appendix C  -  Process Characteristics for Model Facilities 
      
Introduction
      Section 112 of the CAA requires EPA to establish technology-based standards for listed source categories that are sources of hazardous air pollutants (HAP). These technology-based standards are often referred to as maximum achievable control technology (MACT) standards. Section 112 also contains provisions requiring the EPA to periodically revisit these standards. Specifically, paragraph 112(d)(6) states:

      (6) REVIEW AND REVISION.  -  The Administrator shall review, and revise as necessary (taking into account developments in practices, processes, and control technologies), emissions standards promulgated under this section no less often than every 8 years.

      To comply with this CAA requirement, the EPA conducted a technology review for the solvent extraction for vegetable oil production MACT standard. For the purposes of conducting the technology review, the EPA considers "developments" in practices, processes, and control technologies to be:

 Any add-on control technology or other equipment that was not identified and considered during development of the original MACT standards.
 Any improvements in add-on control technology or other equipment (that were identified and considered during development of the original MACT standards) that could result in additional emissions reduction.
 Any work practice or operational procedure that was not identified or considered during development of the original MACT standards.
 Any process change or pollution prevention alternative that could be broadly applied to the industry and that was not identified or considered during development of the original MACT standards.
 Any significant changes in the cost (including cost effectiveness) of applying controls (including controls the EPA considered during the development of the original MACT standards).
Background for the Solvent Extraction for Vegetable Oil Production Source category
Source Category and Source Category Emissions
      The current NESHAP for solvent extraction for vegetable oil production was proposed on May 26, 2000 (65 FR 34252), promulgated on April 21, 2001 (66 FR 19006), and codified at 40 CFR part 63, subpart GGGG. The NESHAP regulates facilities that are major sources of HAP and that produce crude vegetable oil and meal products by removing oil from eight listed oilseeds (soybean, cottonseed, canola (rapeseed), corn germ, sunflower, safflower, peanuts, and flax) through direct contact with an organic solvent. Vegetable oil production that does not use an organic solvent or that does not use one of the listed oilseeds is not subject to the current NESHAP. Facilities that refine or process existing (received) vegetable oil are also not subject to the current NESHAP. 
      
      At the time of the original NESHAP rulemaking, there were 106 vegetable oil production facilities using hexane-based extraction solvent. EPA determined that all the facilities were major sources and EPA initially estimated that these facilities emit 27,400 tons of n-hexane per year. Since that time, the number of facilities subject to the NESHAP has decreased due to consolidation within the industry. Per review of available emissions data, permits, and consultation with industry associations, we have identified 89 vegetable oil production facilities that are major sources of HAP using hexane-based extraction solvent. Appendix A lists the names of the facilities currently subject to the NESHAP. Hexane emissions from these facilities totaled 13,500 tons in the 2014 National Emissions Inventory (NEI).

      The affected sources at a facility utilizing solvent extraction for vegetable oil production are the emission points which may potentially release n-hexane, a HAP, which is utilized as a solvent for the extraction. The EPA does not consider n-hexane classifiable as a human carcinogen; however, long-term human exposure from inhalation of n-hexane is associated with a slowing of the peripheral nerve signal conduction, which may cause numbness and muscular weakness, as well as changes to the retina which may cause blurred vision. Short-term exposure to n-hexane is associated with adverse health effects including irritation of the eyes, mucous membranes, throat and skin, as well as impairment of the central nervous system including dizziness, giddiness, headaches, and slight nausea. Because all facilities are using a solvent that consists of an n-hexane/hexane isomer blend, n-hexane is the only HAP emitted from the solvent extraction of vegetable oils. 
      
      The extraction process is the same for all eight types of oilseeds subject to subpart GGGG. In each case, oilseeds are crushed, conditioned, and rolled into flakes that are mixed with the solvent in an extractor. The oil is then dissolved in the solvent. Following this step, the oil-solvent solution is separated from the flakes and heated to evaporate the solvent. The flakes are separately desolventized and toasted. The evaporated solvent is then condensed, recovered, and reused in the process. The desolventized meal is also dried and cooled as a separate product. All vegetable oil extraction facilities operate some type of solvent collection and recovery system for the recovery of solvent, although the solvent recovery equipment configuration varies from facility to facility. The solvent recovery system collects process gas streams from key process units including extractors, desolventizer-toasters or combined desolventizer-toaster/desolventizer-coolers (DTDC), meal dryers and coolers, process evaporators, oil/solvent distillation columns, and wastewater evaporators. The solvent collection and recovery system then routes the gathered process gas streams to a recovery device that is usually a packed-bed mineral oil scrubber and may include condensers, solvent distillation systems, and solvent storage tanks. Hexane emission points in vegetable oil production facilities generally include the solvent recovery process main vent, meal dryer and meal cooler vents, residual emissions from crude meal and crude oil, equipment leaks, evaporation from equipment and storage tanks, and process wastewater. Recovery of the solvent significantly reduces the costs associated with the extraction and production of vegetable oils. As such, solvent recovery equipment is in many cases regarded as integral to the process and not treated as a pollution control device. In addition to collection and recovery systems, facilities may also use source reduction techniques.
Summary of Existing MACT 
      
      Due to the variability in process and solvent recovery equipment, the current NESHAP restricts plant-wide hexane emissions from each affected facility rather than requiring individual controls at each emission point. The current NESHAP includes emission limitations based on the number of gallons of HAP lost per ton of oilseeds processed. Facilities demonstrate compliance by calculating a compliance ratio comparing the actual HAP loss to the allowable HAP loss for the previous 12 operating months. Allowable HAP loss is based on acceptable oilseed solvent loss ratios provided in the rule in gallons per ton for new and existing sources. Compliance is demonstrated when the facility's calculated compliance ratio is less than one (i.e., the actual HAP loss is less than the calculated allowable HAP loss). Determination of compliance with the requirements of subpart GGGG requires the facility to keep records of the amount of hexane purchased, used, and recovered from the oilseed extraction process, the amount of oilseed processed, and the volume fraction of each HAP exceeding one percent in the extraction solvent used. Facilities may also adjust their solvent loss to account for cases where solvent is routed through a closed vent system to a control device that is used to reduce emissions to meet the standard. This approach allows industry the flexibility to implement the most cost-effective method to reduce overall HAP loss for individual operations. 

      During the development of the solvent extraction for vegetable oil production NESHAP, the EPA utilized two years of monthly data relating to solvent losses in gallons with respect to tons of oilseed processed.  For existing sources, EPA determined the MACT floor for each of the 12 oilseed or process operations as the average of the HAP loss performance levels corresponding to the top performing 12 percent of sources or the top five for oilseeds for operations with fewer than 30 sources. For new sources, the MACT floor was based on the performance level corresponding to the top-ranking source. The MACT solvent loss allowable is a facility-wide "bubble" over all potential sources of n-hexane emissions. Table 1 presents the solvent loss limits established in the MACT, expressed in terms of gallons of solvent loss per ton of oilseed processed.
      
      
   Table 1. Oilseed Solvent Loss Factors for Determining Allowable HAP Loss
                               Type of oilseed process
                     Oilseed solvent loss factor (gal/ton)
      
                               Existing sources
                                  New sources
Corn Germ, Wet Milling
      0.4
      0.3 
Corn Germ, Dry Milling
      0.7
      0.7 
Cottonseed, Large
      0.5
      0.4 
Cottonseed, Small
      0.7
      0.4 
Flax
      0.6
      0.6 
Peanuts
      1.2
      0.7 
Canola (Rapeseed)
      0.7
      0.3 
Safflower
      0.7
      0.7 
Soybean, Conventional
      0.2
      0.2 
Soybean, Specialty
      1.7
      1.5 
Soybean, Combination Plant with Low Specialty Production
      0.25
      0.25 
Sunflower
      0.4
      0.3

	The EPA amended the rule on September 1, 2004 (69 FR 53338) to allow for an additional compliance option that acknowledged that new low-HAP extraction solvents were introduced and in use by some facilities in the affected industry. Due to the low HAP level in the extraction solvents, facilities using this solvent would always be in compliance due to having a compliance ratio of zero. The amended rule reduced the requirements for facilities using the low-HAP extraction solvent option such that it is no longer necessary for facilities to measure the production-related parameters to determine compliance with the NESHAP. The rule continues to require these facilities to complete the necessary record keeping and reporting requirements to assure that the solvent used meets the low-HAP criteria.
Summary of Previously Considered Control Techniques
      
      To assess the MACT floor in the initial rulemaking, EPA developed model plants with emissions equal to and greater than the MACT floor emission limit, then identified and assigned potential control techniques capable of achieving the MACT floor. The control techniques previously considered included: 
      
 Installation of additional desolventizing trays in the desolventizing-toaster; 
 Installation of a counter-current desolventizer;
 Installation of an oil dryer in the oil distillation system; 
 Installation of a refrigerated condenser on the main vent; 
 Venting standing and working losses from fixed-roof storage tanks to the solvent recovery system; and 
 Implementation of a leak detection and repair (LDAR) program for fugitive equipment leaks. 
            
      The EPA also considered a "beyond-the-floor option", which would have required a catalytic incinerator to control the HAP emissions in the combined exhaust from the meal dryer and cooler vents. A fabric filter would also have been required to remove particulate matter in the exhaust stream prior to entering the catalytic incinerator. However, the EPA rejected this option in the final rule because of significantly higher costs per ton of emission reduction. 
Sources of Available Control Technology information
      To identify developments that would be appropriate to consider for control of hexane emissions from vegetable oil production facilities, we considered several sources of information, including: 

          Air permits and related permitting documentation (applications, inventories, or consent decrees).
          EPA's Reasonably Available Control Technology (RACT)/Best Available Control Technology (BACT)/Lowest Achievable Emission Rate (LAER) Clearinghouse data.
          Subsequent regulatory development efforts.
          Literature search and review. 
         
      This section discusses each of these sources and the developments in practices, processes, and control technologies that we identified, if any. 
Air Permits and Related Permitting Documentation
      ERG searched State and Federal websites for major and minor source air permits and related documentation issued to vegetable oil solvent extraction operations. The operating permits included Title V operating permits, synthetic minor operating permits, recent BACT/Prevention of Significant Deterioration (PSD) permits, and other construction permits, where available. Additional permit documentation included permit applications, supporting documents and inventories, and consent decrees. ERG reviewed these materials to compare emissions limitations, configurations, and operating practices between each facility. ERG also reviewed permit materials for any State-specific regulations regarding HAP emissions from solvent extraction operations more restrictive than subpart GGGG.
      
      A review of available Title V permits and documentation for solvent extraction facilities shows there are no new emission sources of HAP at vegetable oil processing plants which were previously unregulated. Although individual facility configurations may vary, all facilities continue to use a mineral oil system, which is composed of an absorber or scrubber and may be combined with evaporators, condensers, refrigerated condensers, solvent distillation systems, strippers, heat exchangers, and wastewater reboilers as part of the solvent recovery system used to meet the solvent loss factors required by the NESHAP. A review of Title V permits revealed 14 facilities also implement LDAR programs to reduce fugitive emissions. Facilities may also use cyclones, baghouses, dust collectors, or oil suppression systems from associated oilseed receiving, hulling, milling, flaking, pelleting, or loadout operations, but these control devices are associated with control of particulate emissions and are not adequate for hexane recovery. 
      
      ERG identified at least one process technology, applicable only to specialty soybean processing operations, that was not previously considered during the development of the 2001 NESHAP in a review of permits and a related consent decree for Archer Daniels Midland (ADM) (April 9, 2003). The consent decree applied control plans to 26 vegetable oil extraction plants in 10 States for reduction of VOC emissions from soybean, corn germ, sunflower, canola, and cottonseed processing facilities. The control plan required ADM to implement process improvements, including the installation of additional condensers, and to establish a VOC solvent loss ratio (SLR) for affected facilities. The consent decree further required ADM to pilot use of a Vacuum-Assisted Desolventizing System (VADS) on a single vegetable oil production process (VOPP) line at one of its specialty soybean processing facilities, and to evaluate the performance criteria of the VADS. The VADS technology is a new process technology discussed further in section 4.0 of this memorandum (Developments in Practices, Processes, and Control Technologies).
RACT/BACT/LAER Clearinghouse Database
      Under the EPA's New Source Review (NSR) program, companies planning to build a new facility or modify an existing facility must obtain an NSR permit if their operation will cause criteria air pollutant emissions to increase by a specified amount. The NSR permit is a construction permit that generally requires the company to minimize air pollution emissions from the new or modified facility by changing processes to limit emissions of air pollutants and/or installing air pollution control equipment.

      The terms "RACT," "BACT," and "LAER" are acronyms for different program requirements relevant to the NSR program. RACT, or Reasonably Available Control Technology, is required for existing sources in areas that are not meeting national ambient air quality standards (non-attainment areas). BACT, or Best Available Control Technology, is required for new or modified major sources in attainment areas. LAER, or Lowest Achievable Emission Rate, is required for new or modified major sources in non-attainment areas.

      BACT and LAER (and sometimes RACT) are determined on a case-by-case basis, usually by State or local permitting agencies. The EPA established the RACT/BACT/LAER Clearinghouse, or RBLC, to provide a central database of air pollution technology information (including past BACT and LAER decisions contained in NSR permits) to promote information sharing among permitting agencies and to aid in future case-by-case determinations. However, the data in the RBLC are not limited to sources subject to RACT, BACT, and LAER requirements. Noteworthy prevention and control technology decisions and information may be included even if they are not related to past RACT, BACT, or LAER decisions. 

      The RBLC contains over 5,000 air pollution control permit determinations that can help identify appropriate technologies to mitigate most air pollutant emission streams. The EPA designed the clearinghouse to help permit applicants and reviewers make pollution prevention and control technology decisions for stationary air pollution sources, and includes data submitted by several U.S. territories and all 50 States on over 200 different air pollutants and 1,000 industrial processes.

      We searched the RBLC database for the Vegetable Oil Manufacturing process category (70.300) and oilseed-specific subcategories (70.310, 70.320, 70.330, 70.350, and 70.390). We searched for the pollutant "hexane" to identify facilities that may have installed control technologies specifically to reduce hexane emissions, as well as the pollutant "VOC" to identify facilities where VOC is regulated as a surrogate for hexane. We also searched the RBLC database for the keywords "extraction", "desolventizer", "DTDC", "scrubber", and "adsorber" to identify facilities that may have extraction operations and identify facilities that have installed control technologies to reduce emissions.  

      The RBLC database search identified 21 active facilities with vegetable oil extraction operations. Seventeen (17) of these facilities have established BACT limits for the solvent loss ratio that are more stringent than the SLR provided by GGGG. A review of these facilities did not reveal any new emissions reduction practices, processes, or control technologies for hexane or VOC in current use. All 21 facilities reported the use of a solvent recovery system with mineral oil scrubber or absorber, 12 facilities reported the use of one or more condensers (in combination with a mineral oil scrubber or absorber), and 13 facilities indicated use of an LDAR program to monitor and control fugitive emissions. EPA considered all of these process technologies and control practices previously under the NESHAP for emissions reductions. However, at least one facility identified and evaluated the use of a cryogenic condenser installed after the mineral oil absorber as a potential commercially available control option in their determination of BACT. Additionally, some facilities re-evaluated the use of catalytic incineration for control of exhausts from meal dryers and coolers as potentially available control options. Section 4.0 of this memorandum (Developments in Practices, Processes, and Control Technologies) discusses the use of a cryogenic condenser and the use of catalytic incineration. Appendix 2 presents the relevant results of the RBLC search.
Subsequent Regulatory Development 
      The EPA promulgated the Solvent Extraction for Vegetable Oil Production NESHAP on April 21, 2001. Since that time, EPA has developed air toxics regulations for additional source categories that emit organic HAP from similar types of emission sources to those included in the vegetable oil production category. We have identified and reviewed these similar NESHAP regulations to identify potential developments in practices, processes, and control technologies used to control emissions that may be applicable for use at vegetable oil production facilities using solvent extraction. In particular, we reviewed several NESHAP with analogous manufacturing processes including solvent recovery, promulgated or revised after April 21, 2001.  A description of the standards reviewed, and their requirements, follows.

       NESHAP for the Pharmaceuticals Production Industry (40 CFR 63 subpart GGG). The EPA promulgated this rule in 1998 and revised the rule in 2014. This rule applies to major source facilities which produce pharmaceutical products. The rule requires that HAP emissions be controlled for the following emissions points: storage tanks, process vents, equipment leaks, wastewater collection and treatment systems, and cooling towers. Facilities must control HAP by meeting an emissions limit or control efficiency requirement, and a source can use emissions averaging to meet the emissions standards. This standard requires the reduction of organic HAP emissions by venting emissions through a closed-vent system to any combination of control devices or recovery devices, such as absorbers, carbon adsorbers, condensers, flares, boilers, and process heaters. The control or recovery device must reduce inlet emissions of HAP by 95 weight-percent or greater, or to outlet concentrations less than or equal to 20 parts per million by volume (ppmv) as Total Organic Carbon (TOC). There is also an alternative, pollution prevention-based standard that requires a reduction in the use of HAP solvents during the manufacturing process.
   
       NESHAP for Miscellaneous Organic Chemical Manufacturing (MON) Sources (40 CFR part 63 subpart FFFF). The EPA originally proposed this rule in 2002 and finalized the rule in 2006. This NESHAP established emission limits and work practice standards for new and existing MON process units, wastewater treatment and conveyance systems, transfer operations, and associated ancillary equipment located at major sources of HAP. This NESHAP requires that affected equipment control any HAP vented from these sources by routing the vapors to a control device or recovery device that reduces emissions of total HAPs by 98 percent or to a concentration of 20 ppmv. A control device may include, but is not limited to, absorbers, carbon adsorbers, condensers, incinerators, flares, boilers, and process heaters. The rule also provides an alternative, pollution prevention-based standard that requires reductions in the amounts of toxic air pollutants used during the manufacturing process. 
   
       NESHAP for Paper and Other Web Coating Sources (40 CFR 63 subpart JJJJ).  The EPA proposed this rule 2000 and finalized the rule in 2002. The rule applies to facilities that coat paper and other web substrates. The paper and other web coatings source category emits HAP such as: toluene, methanol, methyl ethyl chloride, ethylene glycol, xylenes, phenol, methylene chloride, glycol ethers, hexane, methyl isobutyl ketone, cresols, cresylic acid, dimethyl formamide, vinyl acetate, formaldehyde, and ethyl benzene. MACT for these facilities includes reducing HAP emissions by 95 percent for existing web coating operations and 98 percent for new web coating. The rule requires the use of a capture and control system where the control device may be a solvent recovery device or oxidizer. Web coating operations may also reduce emissions by using pollution prevention measures.
   
       NESHAP for Site Remediation (40 CFR 63 subpart GGGGG). The EPA promulgated this rule in 2003 and revised the rule in 2006. The 2006 final rule applies to major sources where remediation technologies and practices are used at the site to clean up contaminated environmental media (e.g., soil, groundwater, or surface water) or certain stored or disposed materials that pose a reasonable potential threat to contaminate environmental media. This regulation requires emissions controls and/or requirements for work practices for three groups of emission points: process vents, remediation material management units (tanks, containers, surface impoundments, oil/water separators, organic/water separators, drain systems) and equipment leaks. The MACT includes reducing HAP from process vents and remediation material management units by routing the vapors to a control device that reduces emissions of total HAPs by 95 percent or to a concentration of 20 ppmv. Subpart GGGGG requires an LDAR program for equipment (e.g., pumps, compressors, valves, connectors) involved in remediation.
   
       NESHAP for Halogenated Solvent Cleaning (40 CFR 63 subpart T). The EPA finalized this rule in 1994 and revised the rule in 2007. The rule requires batch vapor solvent cleaning machines and inline solvent cleaning machines to meet emission standards reflecting the application of the MACT. The rule limits solvent emissions by setting facility-wide annual solvent cleaning emission limits in kg per year. Facilities determine compliance by maintaining solvent consumption records and conducting materials balance calculations of overall solvent emissions.
      
      Each of these standards identify HAP emission limits or efficiency standards and allow for compliance using solvent recovery or control devices, materials balance calculations, and pollution prevention practices. EPA previously considered or currently allows all of these control practices under 40 CFR 63, subpart GGGG for emissions reductions. 
Review of Literature

      ERG conducted a literature review to identify additional developments and advancements in preventing and controlling HAP emissions from solvent extraction of vegetable oils. A majority of the literature reviewed discussed current abatement technologies, such as installation of mineral oil scrubbers and additional condensers or the use of counter-current desolventizers, and process improvements such as leakage monitoring or improved collection of escaping vapors from process systems, storage tanks, and handling areas in exhaust ventilation for subsequent treatment and solvent recovery. However, in most cases, ERG found insufficient detail in the available literature to determine if advancements had occurred since consideration of these options during the original subpart GGGG rulemaking. 
      
      The report titled Guidance on VOC Substitution and Reduction Activities Covered by the VOC Solvents Emissions Direction (European Commission, 2009) noted that the three most common methods of VOC control from solvent extraction in Europe include: 1) a condenser, separator, and wastewater reboiler, 2) mineral oil scrubber, and 3) cryogenic condensation. The first two technologies are routinely used within U.S. vegetable oil extraction plants subject to subpart GGGG to meet emissions standards. The report noted that cryogenic condensation is more common in European installations.
      
      Several reports and studies referred to emerging technologies that have been tested for vegetable oil extraction, including supercritical fluid extraction (European Commission, 2009; Reverchon and Marco, 2006), enzyme-aided aqueous extraction (Barnes, 2015; Campbell et al., 2011; Dijkstra, 2009; European Commission, 2009; Latif et al., 2008), ultrasonic assisted extraction (Li et al., 2004; European Commission, 2009), and osmotic shock (European Commission, 2009). These technologies use non-HAP solvent methods for the extraction of a variety of oils. Section 4.0 of this memorandum (Developments in Practices, Processes, and Control Technologies) provides further evaluation of these technologies. 
Developments in Practices, Processes, and Control Technologies
Identified Control Measures for Solvent Extraction from Vegetable Oil Production

      This section discusses any identified developments in control measures, work practices, or operational procedures that were identified during the review and the technological feasibility of these measures for application in the vegetable oil production industry.

Add-on Control Technology or Other Equipment Not Identified and Considered During MACT Development

      As described in sections 3.2 and 3.5 of this memorandum, ERG's review identified the use of cryogenic condensation to reduce emissions of hexane and VOC from the main vent in vegetable oil extraction operations. Cryogenic condensation is an add-on abatement technology that EPA did not previously identify during the development of the 2001 NESHAP. Cryogenic condensers work similarly to refrigerated condensers in that they rely on a cooling agent for the reduction of the condenser temperature. However, a cryogenic control system uses liquid nitrogen as a cooling agent to reduce the temperature of the condenser, which may achieve temperatures from -160 °F to as low as -350 °F; typical refrigerated condensers using chlorofluorocarbons or hydrofluorocarbons range from -30 to -150 °F (U.S. EPA, 2017; U.S. EPA, 2001). The lower temperatures achieved by a cryogenic condenser result in greater condensation and removal of solvent from the exhaust stream. Cryogenic condensation systems are generally best suited for low flow rates (<1,000 standard cubic feet per minute (scfm)), high inlet solvent vapor concentrations (> 1,000 ppmv), and low moisture content (Trembley and Begata, 2014), and are therefore suited for the conditions of the mineral oil absorber main vent. Cryogenic condensation efficiency typically exceeds 99 percent for VOC emission reduction (European Commission, 2009). 
      
      No VOPPs currently use cryogenic condensation within the United States; however, cryogenic condensation is in use in Europe and has been evaluated as a control in at least two BACT/LAER reviews, as they are already used to control similar sources and emissions (i.e., VOC) as those that exist at VOPPs. The prior BACT/LAER analyses reviewed the condenser as a polishing step after the mineral oil absorber. Therefore, EPA considers the use of cryogenic condensation to be a technologically feasible control option. 
      
Improvements in Add-On Control Technology or Other Equipment (That Was Identified and Considered During MACT Development)

      As discussed in section 2.3 of this memorandum, EPA evaluated the use of a catalytic incinerator to control HAP emissions from VOPPs as a "beyond-the-floor" option during the development of the Solvent Extraction for Vegetable Oil Production NESHAP promulgated in 2001. The incinerator would control the combined exhaust from the meal dryer and cooler vents. EPA included a fabric filter in this evaluation for the removal of particulate matter in the exhaust stream prior to entering the catalytic incinerator. 
      
      Catalytic incinerators contain a bed of active catalyst material that facilitates the overall combustion reaction. In a catalytic incinerator, the waste stream may be either preheated directly (using auxiliary fuel) or indirectly by heat exchange with the oxidizer's post-combustion gas. The heated gas then passes over the catalyst bed. The catalytic bed has the effect of increasing the reaction rate and promotes oxidation at lower reaction temperatures than in other thermal incinerator units, which requires less auxiliary fuel. Meal dryers and coolers in vegetable oil production operations typically have high flow rates and low inlet concentrations of hexane, however, there can be significant variability in the volume and concentration during normal operation as well as during process upsets, malfunctions, and shutdown. Catalytic incinerators can and have been used effectively at low inlet loadings (1 ppmv or less). However, the types of compounds that can be oxidized are limited due to the poisoning or clogging effect that some compounds, including particulates, have on the catalyst. Catalytic oxidation is best suited to streams with low variation in the type and concentration of VOC, and where catalyst poisons or other fouling contaminants are not present.
      
      Catalytic oxidation is not currently used in VOPPs in the United States. However, as discussed in section 3.2 of this memorandum, catalytic incineration has recently been re-evaluated as a control in some BACT/LAER reviews, based on its use for control of similar sources and emissions. Catalytic oxidation has not been selected as a viable control option at VOPPs for several reasons. First, vent gases from meal dryers and coolers that would be ducted to the incinerator would cover a wide range of volumes and solvent concentrations, which would impede the efficiency of the oxidizer. Additionally, the exhaust streams of the dryers and coolers in solvent extraction plants generally contain compounds that would contribute to fouling of the catalyst bed. The amount of particulate in the exhaust gas during normal operation is likely to cause plugging of the inlet screens or catalyst bed of the oxidizer. The exhaust from the meal dryers and coolers also contain a small amount of aerosolized oil, as well as sulfur compounds that occur naturally in soybeans and other oilseeds. Although the addition of a fabric filter or other high efficiency filtration system may reduce particulates in the exhaust stream, the aerosolized oil and sulfur compounds cannot be similarly removed and would contribute to fouling of the catalyst bed. The aerosolized oils may also cause carbonization of the oxidizer chamber that could result in a loss of control efficiency. Therefore, it is unclear that the use of catalytic incineration would result in reliable emissions reductions over time and the potential for fouling of the catalyst bed would need to be considered in the cost estimate.
      
      Another concern for  catalytic incineration in solvent extraction facilities is related to the safety of operations. The presence of fugitive hexane vapors at vegetable oil processing plants presents a fire and explosion hazard, and normal shutdown procedures (including purging hexane from process units), process upsets, and malfunctions may result in near lower-explosive limit (LEL) conditions in the meal dryer and cooler exhaust. For example, in facility shutdowns, as each system is purged, the concentration is reduced from greater than 100 percent of the upper explosive limit (UEL) through the explosive range to less than 10 percent of the LEL. Due to the flammability of hexane, the National Fire Prevention Association (NFPA) sets a standard for solvent extraction plants, NFPA 36, that requires that all ignition sources be at least 100 feet from the extraction process and requires all potential ignition sources be equipped with approved devices to prevent flashbacks into the process area. We anticipate that these requirements could further limit the installation of a catalytic incinerator at individual facilities due to space and property constraints.  
      
      Based on the technical and safety concerns identified, EPA considers the use of catalytic incineration, even with the use of a fabric filter, technically infeasible for meal dryers and coolers at VOPPs. 
      As discussed in section 3.1 of this memorandum, as part of a review of permit materials, EPA also identified the installation of additional condensers and/or condenser upgrades as a control technique for control of emissions of HAP from VOPPs. The condensers included extractor condensers (installed following the extractor and prior to the vent condenser) and once-thru cold water condensers (following the vent condenser and prior to the mineral oil absorber), which were required as part of a 2003 consent decree for ADM. The installation of additional condensers as a control technique was considered at the time of MACT development. ADM evaluated the emission reduction benefits of the condenser installation at multiple facilities, including seven conventional soybean VOPPs, two large cottonseed VOPPs, one canola and small cottonseed VOPP, one corn germ and sunflower VOPP, and three multiseed VOPPs. Each facility installed either an extractor condenser, a cold-water condenser, or both as part of the consent decree control plan. The condensers were installed in 2004 and the company provided an evaluation of the emissions reductions in 2005. The evaluation provided by the company indicated that the condenser upgrades resulted in minimal emissions reductions, and in some cases reflected no measurable emissions reductions benefits, particularly for multi-seed plants. Therefore, although many facilities may install extractor or cold-water condensers as part of an overall facility plan to help meet the SLR, these upgrades do not appear to provide significant emissions reductions, and are not evaluated further in this analysis.   
      
      We identified no additional improvements or considerations of add-on control or abatement technologies that were previously considered during MACT development. 

Work Practices and Procedures Not Identified and Considered During MACT Development

      ERG identified no additional work practices or procedures that were not already identified and considered during MACT development. 

Any Process Change or Pollution Prevention Alternative that could be Broadly Applied that was not Identified and Considered During MACT Development

      ERG identified several new process technologies that reduce or avoid HAP emissions during this review. As described in section 3.1 of this memorandum, ERG identified the use of vacuum-assisted desolventizers at specialty soybean production facilities in a review of a 2003 consent decree and permits issued following the promulgation of the 2001 rule. 
      
      Vacuum-assisted desolventizing technology is only in use by a limited number of specialty soybean facilities and is only applicable to the specialty soybean production process. Specialty soybean manufacturing varies from conventional soybean manufacturing in that the product is intended for human consumption and is therefore processed at lower temperatures to minimize the denaturation of proteins. In specialty soybean manufacturing, flakes are desolventized using either flash desolventizing, which relies on exposing solvent-laden flakes to superheated solvent vapors for a matter of seconds, or VADS. The vacuum-assisted stripper-cooler process relies on a vacuum to reduce the boiling point of the solvent, which results in an increased migration of hexane from the flakes at lower temperatures. The use of lower temperatures results in a less complete desolventization for specialty soybean products, therefore, solvent losses from specialty soybean operations are therefore generally greater than in conventional desolventizing. 
      
      The consent decree reviewed in this analysis directed the use of VADs technology to reduce emissions of VOC from a single specialty soybean VOPP line, and provided for an evaluation of the performance criteria of the VADS prior to installation on additional VOPP lines. The VADs technology pilot was intended to achieve a 90 percent reduction in VOC emissions from the specialty soybean lines.  The VADs was constructed on the VOPP line in 2004 and the facility provided a review of its evaluation in 2005. The evaluation provided by the facility indicated that although emissions reductions were achieved, there was not a substantial emission reduction benefit and the 90-percent reduction goal was not met.
      
      Currently, 14 VOPPs in the Solvent Extraction for Vegetable Oil Production source category produce specialty soybean proteins or a combination of specialty and conventional soybean proteins. Industry-provided data indicate that only four of the 14 VOPPs produce specialty soybeans using VADS, and that most VOPPs using VADS also have non-VADS equipped lines using the same extraction and recovery systems. These include one facility with VADs that operates intermittently (4-6 days per month) when the line is used for specialty processing, one facility that operates VADs on 3 of 4 specialty lines, one facility with VADS on a single specialty line and collocated with a conventional line, and one facility with VADs on a single specialty line. Because all but one of these facilities also has non-VADS equipped lines using the same oil extraction and solvent recovery systems, there is not sufficient solvent loss data that is fully representative of VADs performance. Further, in a review of the RBLC, a 2011 BACT review (RBLC ID IN-0150) of the use of VADs for a specialty soybean processing facility indicated that VADS have not been recently applied in this industry or similar source categories and were no longer commercially available. Therefore, although a limited number of VOPPs are using VADs, this technology is not considered broadly applicable to other specialty soybean facilities or other VOPP facilities at this time. 
      
      As discussed in section 3.4 of this memorandum, ERG identified several additional emerging technologies in internet searches and a review of available literature. These processes included supercritical fluid extraction, enzyme-assisted aqueous extraction, ultrasonic assisted extraction, and osmotic extraction. These are alternative extraction methods which avoid or reduce the use of HAP-based solvents. 
      
      Supercritical fluid extraction involves the extraction of vegetable oils using supercritical fluids such as carbon dioxide. The carbon dioxide is liquefied under pressure and then heated to the point that it is a supercritical fluid. The carbon dioxide acts as a solvent but is more easily removed than hexane from the product through simple depressurization. The process results in much higher solvent yields but is energy intensive due to the high pressure which must be maintained. (European Commission, 2009). Although supercritical fluid extraction has been considered in pilot projects for production of biodiesel, it is not currently in use in any VOPPs in the United States or elsewhere. 
      
      In enzyme-assisted aqueous extraction, enzymes are used to degrade the cell walls with water as the primary solvent. The enzymes may be designed to have a specific mode of action, but cellulase, hemicellulose, pectinase, and proteases are the most favorable enzymes (Kalia et al, 2001). The process results in a higher quality oil and protein. There is currently one known pilot plant for enzymatic oil extraction, located in Denmark. (European Commission, 2009)
      
      ERG identified two additional technologies in research studies, including ultrasonic-assisted extraction, a process that involves the use of ultrasonic waves to break open cell walls to accelerate the extraction of oil in the existing solvent-based process (European Commission, 2009; Li, 2002), and osmotic shock extraction, which requires a reaction at osmotic pressure to force cells in a solution to rupture (European Commission, 2009). There are no known pilot plants for these technologies. 
      
      Although supercritical fluid extraction, enzyme-assisted aqueous extraction, ultrasonic assisted extraction, and osmotic extraction have been studied for use in vegetable oil extraction applications, they are not used at any existing solvent extraction plants and are considered novel technologies that are not yet technologically feasible. 
Summary of Developments in Practices, Processes, and Control Technologies that are Considered Technologically Feasible

      After review of State and Federal air operating permits, the RBLC, recent regulatory determinations, and relevant literature, we identified the use of a cryogenic condenser after the solvent recovery system main process vent as a technically feasible control technology for reducing HAP emissions from VOPPs. The EPA did not previously consider the use of a cryogenic condenser after the solvent recovery system main process vent during the development of subpart GGGG, however, this control has been identified as a technologically feasible control option in use in European installations, as well as included in recent BACT reviews.  Therefore, given the feasibility of this technology, we are considering the use of a cryogenic condenser in VOPPs in the United States to increase the recovery of hexane from the exhaust stream. Section 5.0 of this memorandum includes a discussion of the costs for these technologies.
Cost and Environmental Impacts
      As discussed in section 4.2 of this memorandum, ERG identified a cryogenic condenser installed after the mineral oil absorber main vent as a feasible control option for VOPPs. ERG estimated the costs for this control option based on the development of model scenarios which represent various vegetable oil production facilities. Section 5.1 of this memorandum provides the model methodology. Section 5.2 of this memorandum includes the control option costs. Section 5.3 of this memorandum describes the cost impacts.
      
Development of Model Scenarios for Estimation of Control Costs

      For estimation of control costs, ERG developed six model scenarios to represent the solvent recovery system main process vent conditions at several vegetable oil processing operations. ERG developed model scenarios for the following operations:
      
      1) Conventional soybean operations (3 models). 
      2) Cottonseed operations. 
      3) Corn germ. 
      4) Specialty soybean. 

	We assigned each scenario process characteristics for the solvent recovery system main process vent that would be generally representative of similar operations in the source category. For conventional soybean operations, we developed three scenarios representing varying solvent loss characteristics in order to better represent the range of values of existing facilities. Baseline emissions were then developed for each of the six scenarios for evaluation of cost-effectiveness. The following subsections of this memorandum identify the parameters selected and discuss the estimation of baseline emissions for each model.

Selection of Process Parameters for Model Scenarios

      As discussed in section 5.1 of this memorandum, we developed six model scenarios to represent the processes and emissions in the vegetable oil production source categories. We selected process parameters for each of six scenarios (three conventional soybean, one cottonseed, one corn germ, and one specialty soybean model) based on a review of reported 2014 NEI stack parameters, facility permits, manufacturer's materials (Crown Iron Works Company, 2007), and review of existing literature and materials developed in the 2001 NESHAP (Zukor and Ali, 2000a, 2000b). We selected a set of general operating characteristics for all model scenarios; Table 2 lists these general characteristics. We assumed that a commercial grade hexane solvent (0.64 volume fraction of n-hexane) would be used in all scenarios. 
      
Table 2. General Operating Characteristics for All Model Scenarios
                                Characteristic
                     Conventional Soybean (all scenarios)
                                  Cottonseed
                                   Corn Germ
                               Specialty Soybean
Seed Production Rate (tons/day)
2,600 
1,100
1,100
3,000
Days of Operation/Year
330
330
330
330
Meal Fraction [a]
0.81
0.19
0.449
0.162
[a] 2012 Soya & Oilseed Bluebook, Soyatech, March 26, 2012.
      
      For each model scenario, we developed additional process characteristics for the solvent recovery system (mineral oil absorber) main process vent. For the main process vent, the process characteristics assigned include the exhaust temperature, flow rate in actual cubic feet per minute (acfm), and the solvent concentration of the main vent exhaust stream. We selected the exhaust temperature and flow rates for each emission point for each scenario based on a review of data reported to the 2014 NEI, facility permits, and review of existing literature and materials (Zukor and Ali, 2000a, 2000b). We estimated the solvent concentration in the main vent exhaust stream as a percentage of the LEL of n-hexane in air. Several VOPP currently monitor the solvent concentration in the main vent as a percentage of the LEL. The LEL of n-hexane in air is 1.1 percent by volume. The percent LEL we assigned to each model is based on review of facility permits and BACT reviews, manufacturer specifications, and review of existing literature.  
      
      Appendix C provides the parameters assigned to each model scenario. 
      
Calculation of Baseline Emissions for Model Scenarios

      Following establishment of the process characteristics of each model scenario, we calculated baseline emissions estimates for the main process vent for each model. ERG used the baseline emission estimates to estimate the emissions reductions and the cost effectiveness of the control options evaluated in each scenario (see sections 5.2 and 5.3 of this memorandum). 
      
      In each model scenario, we estimated the baseline emissions of hexane for the main vent based on the exhaust flow rate of the main vent (acfm), the concentration of the solvent in the exhaust stream, and the density of the gas stream (adjusted based on the exhaust temperature, using the Ideal Gas Law). As discussed in section 5.1.1 of this memorandum, the solvent concentration in the main vent exhaust stream was estimated as a percentage of the LEL of n-hexane in air (see Appendix C for the percent LEL assigned to each model).  ERG used the following equation to estimate emissions for each model:
      
EQUATION 1: Main vent (tons hexane/year) = 
      Flow (ft3/min) x 60 min/hr x hrs/year x Hexane LEL (1.1%) x % of LEL x density (lb/ft3) x 1 ton/2000 lbs

      The baseline emissions for each model scenario are listed in Table 3.

Table 3. Baseline Emission Rates by Model Scenario
                                Model Scenario
                    Main Vent Hexane Emissions (tons/year)
Conventional Soybean 
(Plant 1.1)
                                      11
Conventional Soybean 
(Plant 1.2)
                                      35
Conventional Soybean 
(Plant 1.3)
                                      112
Cottonseed (Plant 2)
                                      34
Corn Germ (Plant 3)
                                      33
Specialty Soybean (Plant 4)
                                      37

Control Cost Methodology

       To evaluate the potential control costs for a cryogenic condenser on the main vent, ERG reviewed detailed costs provided from BACT reviews conducted for a facility in the state of Indiana. The costs provided in the BACT review were based on a vendor quote for a Linde Cirrus Cryogenic Condenser and using the methodology from EPA's Air Pollution Control Cost Manual (Sixth Edition, January 2002), Section 3, Chapter 2 (Refrigerated Condensers). We estimated the cryogenic condenser vendor quote was in 2007 dollars based on the date of the BACT review.
      
      ERG estimated the total capital investment (TCI) for each model scenario by updating equipment costs from 2007 to 2017 dollars using The Chemical Engineering Plant Cost Index. Direct costs included purchasing foundation and supports, handling and erecting the structures, electrical and piping work, insulation for ductwork, painting, and site preparation. Indirect costs included engineering, construction and field expenses, contractor fees, start-up, performance tests, and contingencies. ERG also assumed that cryogenic condenser capital costs correlate with the flow rate of the exhaust gas of the main vent. Therefore, ERG adjusted the TCI for each of the six model scenarios based on the flow rate of the main vent exhaust in each scenario and applied the six-tenths relationship shown in equation 1.
                                          
                    Costb = Costa * (Flowa / Flowb)[0.6]		(Eq. 1)
Where:
      Costa	=	Vendor quote = $1,326,746 ($2017)
      Flowa 	= 	Flow rate from vendor quote, 225 actual cubic feet per minute (acfm)
      Flowb 	=	Flow rate for model scenario
      
      Using the hourly assumptions provided in the prior BACT analysis, ERG estimated direct annual costs (DAC) with updated labor rates and utility costs. Utilities include the liquid nitrogen required for the condenser, and this cost was adjusted from the vendor quote by assuming a linear relationship between liquid nitrogen cost and the flow rate of the exhaust gas of the main vent. The liquid nitrogen cost was also adjusted from $2007 to $2017 using The Chemical Engineering Plant Cost Index We adjusted labor rates to reflect May 2017 National Industry-Specific Occupational Employment and Wage Estimates for NAICS 311200 - Grain and Oilseed Milling.. 
      
      Indirect annual costs (IAC) include overhead, administrative charges, property taxes, insurance, and capital recovery. ERG recalculated the capital recovery factor assuming an interest rate of 4.75 percent over 10 years. To account for hexane recovery, ERG used the baseline emissions to estimate potential solvent recovery and assumed a cost of $2.70/gallon hexane. ERG adjusted the indirect annual costs and hexane recovery for each of the six model scenarios based on the flow rate of the main vent exhaust in each scenario. Total annual costs (TAC) include the direct annual costs and indirect annual costs minus the savings of hexane recovered. Table 4 below includes the TCI, DAC, IAC, TAC, and hexane recovery costs savings (HR).
      
Table 4. Total Capital Cost and Total Annual Costs to Reduce Hexane Emissions at VOPPs Using Cryogenic Condenser
                                   Facility
                           Total Capital Investment
                              Direct Annual Costs
                             Indirect Annual Costs
                             Hexane Recovery (HR)
                            Total Annual Cost (TAC)
Conventional Soybean (Plant 1.1)
                                   $815,602 
                                      $1,154,229 
                                         $387,374 
                                   $11,457 
                                      $1,530,145 
Conventional Soybean (Plant 1.2)
                                  $1,413,325 
                                      $2,259,561 
                                         $487,754 
                                   $36,454 
                                      $2,710,861 
Conventional Soybean (Plant 1.3)
                                  $1,873,761 
                                      $3,364,894 
                                         $565,078 
                                   $116,654 
                                      $3,813,318 
Cottonseed (Plant 2)
                                  $1,480,118 
                                      $2,406,939 
                                         $498,971 
                                    $35,413 
                                      $2,870,497 
Corn Germ (Plant 3)
                                  $1,413,325 
                                      $2,259,561 
                                         $487,754 
                                   $34,371 
                                      $2,712,944 
Specialty Soybean (Plant 4)
                                  $1,326,746 
                                      $2,075,339 
                                         $473,214 
                                   $38,537 
                                      $2,510,015 


Cost Effectiveness

       After estimating the total capital investment (TCI), total annual costs (TAC), hexane recovery, and hexane emissions reductions for installation of cryogenic condenser for each model scenario, we determined the cost-effectiveness, assuming a 99.9 percent reduction. ERG calculated the cost-effectiveness using TACs and emissions reductions for each model scenario. 

Table 5. Cost Effectiveness of Cryogenic Condenser for Model Scenarios 
Model Scenario
                             Conventional Soybean 
                                  Cottonseed
                                   Corn Germ
                               Specialty Soybean
                                    Total 
                                (All Scenarios)

                                   Plant 1.1
                                   Plant 1.2
                                   Plant 1.3
                                    Plant 2
                                    Plant 3
                                    Plant 4
                                       
TCI ($)
                                                                        815,602
                                                                     1,413,325 
                                                                     1,873,761 
                                                                     1,480,118 
                                                                     1,413,325 
                                                                     1,326,746 
                                                                     8,322,877 
TAC ($/yr)
                                                                     1,530,145 
                                                                     2,710,861 
                                                                     3,813,318 
                                                                     2,870,497 
                                                                     2,712,944 
                                                                     2,510,015 
                                                                    16,147,779 
Emissions Reductions (tpy)
                                                                            11 
                                                                            35 
                                                                           112 
                                                                            34 
                                                                            33 
                                                                            37 
                                                                           262 
Cost-Effectiveness ($/ton)
                                                                       139,243 
                                                                        77,531 
                                                                        34,082 
                                                                        84,511 
                                                                        82,293 
                                                                        67,906 
                                                                        61,694 

CONCLUSIONS
      This analysis identified one potential control technology for application in vegetable oil production facilities. ERG identified the use of a cryogenic condenser after the main vent as an add-on control option not previously considered during the development of subpart GGGG. This analysis found that the use of a cryogenic condenser on the main vent is not cost effective for reduction of HAP. Finally, this analysis found no direct evidence of any additional significant changes in the practices, processes, and control technologies that may be used by vegetable oil production facilities. 
 


References
Barnes, L. 2015. Ammonia, Hydrochloric Acid, Hydrogen Sulfide, N-hexane, Nitric Compounds, and Sulfuric Acid in the Food Processing Industry. Great Lakes Regional Pollution Prevention Roundtable. 
Campbell, K. A. Glatz, C.E., Johnson, L.A., Jung, S., de Moura, J. M. N., Kapchie, V., and Murphy, P. 2011. Advances in Aqueous Extraction Processing of Soybeans. Journal of the American Oil Chemists' Society, Volume 88, Issue 4: 449 - 465. 
Crown Iron Works Company. 2007. Desolventizer-Toaster-Dryer-Cooler. Roseville, MN. Available at: http://crownironasia.com/userimages/DTDC%20Main1.pdf. Accessed June 2018.
Dijkstra, A. 2009. Recent developments in edible oil processing. European Journal of Lipid Science and Technology, 111:855-864.
European Commission  -  DG Environment. 2009. Guidance on VOC Substitution and Reduction for Activities Covered by the VOC Solvents Emissions Directive, Final Report. (Directive 1999/13/EC). Available at: http://rs.subsport.eu/images/stories/pdf_archive/legislation/ 23_guide_document_vegetable_oil.pdf. Accessed April 2017.
Guinn, J. Domestic Quality Standards and Trading Rules and Recommended Export Contract Specifications for U.S. Soybeans and Products. U.S. Soybean Export Council. Available at: https://ussec.org/wp-content/uploads/2015/10/Guinn_Quality_Standards_Trading_Rules2002.pdf
Kalia, V.C., Rashmi, S., and Gupta, M. 2001. Using Enzymes for Oil Recovery from Edible Seeds. Journal of Scientific & Industrial Research, 60: 298-310.
Latif, S., Diosady, L., and Anwar, F. 2008. Enzyme‐assisted aqueous extraction of oil and protein from canola (Brassica napus L.) seeds. European Journal of Lipid Science and Technology, 110: 887-892.
Li, H., Pordesimo, L., and Weiss, J. 2004. High intensity ultrasound-assisted extraction of oil from soybeans. Food Research International, Volume 37, Issue 7: 731-738. 
Reverchon, E. and De Marco, I. 2006. Supercritical fluid extraction and fractionation of natural matter. The Journal of Supercritical Fluids. Volume 38, Issue 2: 146-166.
T. Roque, M. Correia, and R. Carvalho. 2013. Analysis of the Hexane Loss in a Vegetable Oil Extraction Unit. Available at: https://fenix.tecnico.ulisboa.pt/downloadFile/1126295043834814/Artigo_ TeresaRoque69452.pdf. Accessed June 2018.
Soyatech, LLC. 2012. Soya and Oilseed Bluebook. Available at: https://issuu.com/fpratt/docs/bluebook/174. Accessed June 2018.
Trembley, John, and Oscar Betata. "Using Cryogenic Condensation to Control Organic Solvent Vapor Emissions." Process Cooling, January 10, 2014.  Available at: https://www.process-cooling.com/articles/87491-using-cryogenic-condensation-to-control-organic-solvent-vapor-emissions. Accessed April 2017.
U.S. EPA, 2018a. 2018. RACT/BACT/LAER Clearinghouse. Available at: http://cfpub.epa.gov/RBLC/.

U.S. EPA, 2018b. 2014 National Emissions Inventory, version 1. Available at: https://www.epa.gov/air-emissions-inventories/2014-national-emissions-inventory-nei-data

U.S. EPA, 2017. EPA Air Pollution Control Cost Manual, Sixth Edition. EPA/452/B-02-001. Available at: https://www3.epa.gov/ttncatc1/dir1/c_allchs.pdf

U.S. EPA, 2002. EPA Air Pollution Control Cost Manual, Seventh Edition. Chapter 2, Refrigerated Condensers. Available at: https://www.epa.gov/sites/production/files/2017-12/documents/refrigeratedcondenserschapter_7thedition_final.pdf

U.S. EPA, 2001. EPA Technical Bulletin: Refrigerated Condensers for Control of Organic Air Emissions. EPA 456/R-01-004. Available at: https://www3.epa.gov/ttnchie1/mkb/documents/refrigeratedcondensers.pdf

C. Zukor and T. Ali. 2000a. Final Process and Emission Characteristics of Vegetable Oil Production Model Plants. Alpha-Gamma Technologies, Inc. to Vegetable Oil NESHAP Project File. Docket No. A-97-59, Category IV-B, Document Number IV-B-6.

C. Zukor and T. Ali. 2000b. Final Model Plant Cost Estimates for Above the MACT Floor Control Option. Alpha-Gamma Technologies, Inc. to NESHAP: Solvent Extraction for Vegetable Oil Production Project File. Docket No. A-97-59, Category IV-B, Document Number IV-B-2.


 Appendix A. Vegetable Oil Production Facilities Subject to the NESHAP 
                                   Facility 
                                     State
                              Oilseeds Processed 
                             Number of VOPP Lines
Bunge North America - Decatur
                                      AL
Soybean (conventional)
1
Cargill, Inc. - Guntersville
                                      AL
Soybean
1
Sessions Company, Inc
                                      AL
Peanut
1
Planters Cotton Oil
                                      AR
Cottonseed 
1
Riceland Foods, Inc.
                                      AR
Soybean (conventional), Rice
1
Adams Specialty Oils
                                      CA
Soybean, canola, safflower, sunflower 
1
J.G. Boswell Company
                                      CA
Cottonseed, safflower
1
ADM - Valdosta
                                      GA
Soybean (conventional), Large Cottonseed
2
Cargill - Gainesville
                                      GA
Soybean (conventional)
1
Golden Peanut - Dawson
                                      GA
Peanut 
1
ADM Bioprocessing - Clinton
                                      IA
Wet corn milling
1
ADM Soybean Processing  -  Des Moines
                                      IA
Soybean (conventional)
1
Ag Processing, Inc - Eagle Grove
                                      IA
Soybean (conventional)
1
Ag Processing, Inc - Emmetsburg
                                      IA
Soybean (conventional and specialty)
1
Ag Processing, Inc - Sergeant Bluff
                                      IA
Soybean (conventional)
1
Ag Processing, Inc - Mason City
                                      IA
Soybean (conventional)
1
Ag Processing, Inc - Manning
                                      IA
Soybean (conventional)
1
Ag Processing, Inc - Sheldon
                                      IA
Soybean (conventional)
1
Bunge North America, Inc 
                                      IA
Soybean (conventional)
1
Cargill, Inc.  -  Des Moines 
                                      IA
Soybean (conventional) (Closed)
1
Cargill, Inc. - Eddyville
                                      IA
Wet corn milling
1
Cargill, Inc. - Cedar Rapids (East)
                                      IA
Soybean (conventional) 
1
Cargill, Inc. - Cedar Rapids (West)
                                      IA
Soybean (specialty)
1
Cargill, Inc. - Sioux City
                                      IA
Soybean (conventional)
1
Cargill, Inc. - Iowa Falls
                                      IA
Soybean (conventional)
1
CHS Oilseed Processing
                                      IA
Soybean (conventional and specialty)
1
ADM - Quincy
                                      IL
Soybean (conventional)
2
Archer Daniels Midland Co  -  Decatur (East Plant)
                                      IL
Soybean (specialty)
1
Archer Daniels Midland Co  -  Decatur (West Plant)
                                      IL
Wet corn milling, Soybean (conventional)
2
Bunge Milling - Danville
                                      IL
Wet corn milling, Soybean (conventional)
2
Bunge North America, Inc. - Cairo
                                      IL
Soybean (conventional)
1
Cargill, Inc. - Bloomington
                                      IL
Soybean (specialty)
1
Solae  -  Gibson City
                                      IL
Soybean (conventional and specialty)
1
Incobrasa Industries Ltd
                                      IL
Soybean (conventional)
1
Ingredion Inc. - Argo Plant
                                      IL
Wet corn milling 
1
Viobin USA 
                                      IL
Soybean (conventional and specialty)
1
Archer Daniels Midland - Frankfort
                                      IN
Soybean (conventional)
1
Bunge Ltd - Morristown
                                      IN
Soybean (conventional) (Closed)
1
Bunge North America (East), Ltd
                                      IN
Soybean (conventional)
1
Cargill, Inc. - LaFayette
                                      IN
Soybean (conventional and specialty)
1
Consolidated Barge and Grain Co
                                      IN
Soybean (conventional), Dry corn milling
1
Louis Dreyfus Agricultural Industries LLC 
                                      IN
Soybean (conventional and specialty)
1
Rose Acre Farms, Inc.
                                      IN
Soybean (conventional)
1
Ultra Soy of America
                                      IN
Soybean (conventional)
1
Bunge Oilseed Processing Plant -Emporia
                                      KS
Soybean (conventional)
1
Cargill, Inc. - Wichita
                                      KS
Soybean (conventional)
1
Northern Sun - Goodland
                                      KS
Sunflower, Canola 
1
Owensboro Grain
                                      KY
Soybean (conventional)
1
Bunge Corporation - Destrehan
                                      LA
Soybean (conventional)
1
Perdue Salisbury Feed and Grain - Salisbury
                                      MD
Soybean (conventional and specialty)
1
Zeeland Farm Soya
                                      MI
Soybean (conventional)
1
ADM - Mankato
                                      MN
Soybean (conventional and specialty)
1
ADM  -  Red Wing
                                      MN
Soybean (conventional)
1
Ag Processing Inc - Dawson
                                      MN
Soybean (conventional)
1
CHS Fairmont
                                      MN
Soybean (conventional)
1
CHS Hallock - Kennedy
                                      MN
Canola (rapeseed)
1
CHS Oilseed Processing - Mankato
                                      MN
Soybean (conventional and specialty)
1
Minnesota Soybean Processors
                                      MN
Soybean (conventional), canola (rapeseed) 
1
ADM Soybean Processing - Mexico
                                      MO
Soybean (conventional)
1
Ag Processing Inc. - Saint Joseph
                                      MO
Soybean (conventional and specialty)
1
Cargill, Inc.  -  Kansas City
                                      MO
Soybean (conventional)
1
Prairie Pride, Inc.
                                      MO
Soybean (conventional)
1
Delta Oil Mill
                                      MS
Cottonseed (Closed)
1
Express Grain Terminals LLC
                                      MS
Soybean (conventional), Wet corn milling, and Other 
1
Cargill, Inc. - Fayetteville
                                      NC
Soybean (conventional)
1
Cargill, Inc. - Raleigh
                                      NC
Soybean (conventional) (Closed)
1
Perdue Farms Inc.  -  Cofield 
                                      NC
Soybean (conventional)
1
ADM Northern Sun Division - Enderlin
                                      ND
Soybean (conventional), sunflower
1
ADM Processing - Velva
                                      ND
Canola (rapeseed)
1
Cargill, Inc.  -  West Fargo
                                      ND
Soybean (conventional)
1
ADM - Fremont
                                      NE
Soybean (conventional)
1
ADM Soybean Processing - Lincoln
                                      NE
Soybean (conventional)
1
AGP Ag Processing, Inc. - Hastings
                                      NE
Soybean (conventional)
1
Cargill Corn Milling NA - Blair
                                      NE
Wet corn milling
1
Archer Daniels Midland - Fostoria
                                      OH
Soybean (conventional)
1
Bunge North America - Bellevue 
                                      OH
Soybean (conventional and specialty)
1
Bunge Oilseed Processing - Delphos
                                      OH
Soybean (conventional and specialty)
1
Cargill Soy Processing - Sidney
                                      OH
Soybean (conventional)
1
Producers Cooperative Oil Mill
                                      OK
Canola, sunflower, peanut, corn germ (Closed)
1
Archer Daniels Midland Soybean Division - Kershaw
                                      SC
Soybean (conventional)
1
Hartsville Oil Mill - Darlington
                                      SC
Cottonseed, peanut
1
South Dakota Soybean Processors
                                      SD
Soybean (conventional and specialty)
1
Archer Daniels Midland Company - Memphis
                                      TN
Large Cottonseed
1
Cargill, Inc. - Memphis
                                      TN
Wet corn milling
1
ADM/Southern Cotton Oil Co - Lubbock
                                      TX
Large Cottonseed 
1
ADM/Southern Cotton Oil Co - Richmond
                                      TX
Large Cottonseed, Corn Germ
1
Pyco Industries Inc.  -  Avenue A 
                                      TX
Cottonseed
1
Pyco Industries Inc.  -  East 50[th] 
                                      TX
Cottonseed (Closed)
1
Valley Co-op Oil Mill
                                      TX
Cottonseed
1
Perdue Farms Incorporated  -  Chesapeake Grain
                                      VA
Soybean (conventional)
1

 Appendix B. Practices, Processes and Control Technologies Identified for Solvent Extraction for Vegetable Oil Operations, Query of the RBLC Database (December 2016)

RBLCID
Facility Name
Date of Last Determination
Oilseed
Process Name
Pollutant
Control Method
Emission Limits
GA-0062
ARCHER DANIELS MIDLAND COMPANY - VALDOSTA
9/6/2002
SOYBEAN (CONV.),
COTTONSEED
VEGETABLE OIL PRODUCTION
VOC
CONDENSER AND MINERAL OIL HEXANE SCRUBBER, 
LEAK DETECTION AND REPAIR (LDAR) PROGRAM
Compliance with SLR limits of NESHAP; solvent consumption and soybean production limits 
IA-0029;
IA-0053
CARGILL, INC - EDDYVILLE
12/18/2001
WET CORN MILLING
CORN OIL EXTRACTION
HEXANE
MINERAL OIL SCRUBBER SYSTEM
Compliance with SLR limits of NESHAP and Plantwide lb/day VOC, rolling 365-day limits 


2/20/2002

MINERAL OIL ABSORBER
VOC






BUILDING ASPIRATOR







EXTRACTION AND D-T ASPIRATION



IA-0085
BUNGE NORTH AMERICA
5/7/2007
SOYBEAN (CONV.)
SOYBEAN OIL EXTRACTION
VOC
MINERAL OIL ABSORBER 
Overall SLR of 0.178 GAL SOLVENT/T SOYBEAN; 0.16 TON GAL VOC/T; 0.2 GAL HAP/T; 12-MTH ROLLING, 
IA-0103
AG PROCESSING SERGEANT BLUFF
3/23/2016

SOYBEAN (CONV.)
SOYBEAN OIL EXTRACTION
VOC
MINERAL OIL SCRUBBER; also operates under MACT subpart GGGG, 40 CFR 63.2850(e)(2))
0.145 GAL SOLVENT LOSS/T SOYBEAN, 
0.2 GAL HAP/T; 12-MTH ROLLING;
Production and solvent throughput limits 
IA-0111
DES MOINES SOYBEAN PROCESSING PLANT
7/6/2016
SOYBEAN (CONV.)
EXTRACTOR AND DESOLVENTIZER TOASTER DRYER COOLER;
EQUIPMENT LEAKS
VOC
MINERAL OIL ABSORPTION SYSTEM AND GOOD OPERATING PRACTICES; LDAR MONITORING SYSTEM
0.14 GAL VOC/T SOYBEAN; 12-MTH ROLLING; Total HAP = Compliance ratio <=1.00 (Consistent with MACT)  -  
IL-0067
ARCHER DANIELS MIDLAND COMPANY (EAST PLANT)
10/28/2002
SOYBEAN, SPECIALTY
EXTRACTION-OIL, MAIN VENT
EXTRACTION-OIL, SPECIALTY SOYBEAN PLANT, OVERALL
EQUIPMENT LEAK, OIL EXTRACTION

VOC
VACUUM-ASSISTED DESOLVENTIZER-COOLER, CONDENSER AND MINERAL OIL SCRUBBER -  SUBJECT TO   REQUIREMENTS FOR INLET TEMP, OIL FLOW RATE, OIL TEMP AND PRESSURE DROP;
LDAR PROGRAM
Limits solvent consumption and soybean production, sets 10.4 LB VOC/T SOYBEAN; 180-DAY ROLLING AVERAGE




POINT AND FUGITIVE FINAL (WHOLE FACILITY)


MINERAL OIL ABSORBER, LEAK DETECTION AND REPAIR PROGRAM (LDAR)
BACT applies facility wide VOC limit of 0.503 GAL/T SOY OIL based on NESHAP SLRs for new source specialty soybean and existing conventional soybean, individual VOC limits for extractor, meal dryer
IL-0125
ADM QUINCY
 06/30/2017
SOYBEAN, CONV.
VEGETABLE OIL PRODUCTION PROCESS
VOC
Limits total surface area of solvent recovery system, provides for MOS with 95% control, sets residence time for DT and requires 4 recovery trays, requires LDAR 
0.175 gal/ton soybeans processed
IN-0150
LOUIS DREYFUS AGRICULTURAL INDUSTRIES LLC
8/13/2013
SOYBEAN, CONV.
SOYBEAN OIL EXTRACTION PLANT AND MEAL DRYER AND COOLER
VOC
COMBINED CONDENSER AND MINERAL OIL SCRUBBER SYSTEM, LDAR PROGRAM
Sets individual VOC limits for Mineral Oil Scrubber and Meal Dryers/Coolers and Overall Facility-wide SLR: 0.141 GAL/T SOYBEAN

IN-0209
CONSOLIDATED GRAIN AND BARGE CO.
6/8/2016
SOYBEAN, CONV.
EXTRACTION SYSTEM
VOC
MINERAL OIL ABSORBER
0.048 LB VOC/TON




DTDC COOLER


0.152 LB VOC/TON




DTDC DRYERS


0.152 LB VOC/TON




OVERALL SOLVENT LOSS RATIO


0.19 GAL VOC/TON 
MN-0065
ADM - MANKATO
1/22/2007
SOYBEAN, CONV.
SOYBEAN OIL EXTRACTION
HEXANE
MAIN VENT W/ CONDENSER, MINERAL OIL ABSORBER, AND LDAR.
0.15 GALVOC /TON 
MN-0092
CHS HALLOCK
05/02/2017
RAPESEED
CANOLA OILSEED PROCESSING
VOC
MINERAL OIL SCRUBBER, GOOD SOLVENT RECOVERY PRACTICES, LDAR
0.29 gal/ton of
canola oilseed

MO-0075
AG PROCESSING, INC.  -  ST. JOSEPH
8/30/2007
SOYBEAN, CONV.
REFINERY PLANT AND OIL EXTRACTION PROCESSES
HEXANE
Evaporators, condensors, MOS; LDAR requirements; solvent storage tanks routed to solvent recovery; vapor recovery tray located below sparge tray of Desolventizer Toaster
Sets facility SLR to 0.145 gal solvent/ton
MO-0082
ARCHER DANIELS MIDLAND-MEXICO
4/1/2015
SOYBEAN, CONV.
SOYBEAN OIL EXTRACTION
VOC
USE OF EVAPORATORS AND CONDENSATION FOR SOLVENT RECOVERY AND UNCONDENSED VAPORS ROUTED TO A MINERAL OIL ABSORBER.  SOLVENT; STORAGE - BREATHING AND WORKING LOSSES ROUTED TO SOLVENT RECOVERY SYSTEM;  PROCESS, FUGITIVE - LDAR PROGRAM, CHILLER TO OPERATE FROM APRIL TO OCTOBER
0.15 GAL SOLVENT/T SOYBEAN, 12 MTH ROLLING; 0.171 GAL SOLVENT LOSS/T DURING SSM PERIODS 
ND-0027
WEST FARGO OILSEEDS PROCESSING PLANT
10/16/2012
SUNFLOWER
CANOLA
FLAX
EXTRACTION AND REFINING
VOC
CONDENSERS AND MINERAL OIL SCRUBBER
0.23 GAL VOC/TON; 12 MTH ROLLING 
*NE-0024
CARGILL - BLAIR PLANT
12/2/2015
WET CORN MILLING
CORN GERM OIL EXTRACTION PROCESS
HEXANE

Complies with NESHAP SLR limits 
NE-0048
ARCHER DANIELS MIDLAND - FREMONT
2/4/2009
SOYBEAN
SOYBEAN OIL EXTRACTION
VOC
MINERAL OIL SCRUBBER W/ SOLVENT RECOVERY CONDENSER, LDAR PROGRAM
0.165 GAL SOLVENT LOSS/T SOYBEAN,12-MTH ROLLING 
*NE-0059
AGP SOY
8/18/2015
SOYBEAN
SOYBEAN EXTRACTION PROCESS
VOC
MINERAL OIL ABSORBER (Includes observations for leaks and corrective action)  
0.145 gal solvent/ton soybean 
-Complies with GGGG for SSM
OH-0251
CENTRAL SOYA COMPANY INC.
7/24/2008
SOYBEAN, SPECIALTY AND CONV.
EXTRACTION OPERATION (CONVENTIONAL)
HEXANE

CONV.  -  3 CYCLONES, CONDENSER, AND ABSORBER

SPECIALTY  -  3 BAGHOUSES AND CONDENSER

0.388 GAL/T rolling 6-mo. weighted AVG (applies to specialty and conventional soybean lines), Complies with NESHAP SLR Limits





EXTRACTION OPERATION (SPECIALTY W/ HEXANE)



*OK-0156 (Note: Construction has never been initiated for this facility)
NORTHSTAR AGRI IND ENID
12/6/2016
RAPESEED
VOC STORAGE (HEXANE)
VOC
MINERAL OIL SCRUBBER, VENT CONDENSER

0.29 GAL SOLVENT LOSS/TON, 12-MONTH ROLLING 




EXTRACTION







WASTEWATER EVAPORATOR







CRUDE MEAL EMISSIONS

DESOLVENTIZER/ TOASTER
157 DEGREES, 1 HR AVG




DRYER/COOLER

DESOLVENTIZER





EQUIPMENT LEAKS

LDAR PROGRAM (NFPA 36)

PA-0308
PERDUE AGRIBUSINESS LLC/MARIETTA
05/05/2016
NEW SOYBEAN OIL EXTRACTION FACILITY
EXTRACTION  PROCESS
VOC
Good Operating Practices;
LDAR
0.125 gal/ton of soybeans solvent loss ratio 





MEAL DRYER







MEAL COOLER



SC-0118
ARCHER DANIELS MIDLAND CO. - KERSHAW FACILITY
3/30/2015
SOYBEAN, CONV.
SOYBEAN OIL EXTRACTOR
HEXANE
MINERAL OIL SCRUBBER, COLD WATER CONDENSER, AND EXTRACTOR
CONDENSER
0.18 gallons of hexane loss per ton of
soybeans processed
VA-0327
PERDUE GRAIN AND OILSEED, LLC
7/12/2017
SOYBEAN, CONV.


LDAR PROGRAM
0.18 gallons solvent/ton of beans processed. Upon startup of the new extractor the SLR shall not exceed 0.152 gallons solvent/ton of beans processed. 


 Appendix C. Process Characteristics FOR Model Facilities 

The following table presents the process characteristics for the solvent recovery system main process vent for six model scenarios that would be generally representative of similar operations in the source category.

                                Model Scenario
                               Temperature (°F)
                               Flow Rate (acfm)
                      Hexane Outlet Concentration (ppmv)
                   Baseline n-Hexane Emissions (tons/yr)[e]
1.1 Conventional soybean 1[a]
                                      90
                                      100
                                     2200
                                      11
1.2 Conventional soybean 2[b]
                                      90
                                      250
                                     2750
                                      35
1.3 Conventional soybean 3[c]
                                      90
                                      400
                                     5500
                                      112
2 Cottonseed[b]
                                      150
                                      270
                                     2750
                                      34
3 Corn germ[b]
                                      116
                                      250
                                     2750
                                      33
4 Specialty soybean[d]
                                      100
                                      225
                                     3300
                                      37
[a] Parameters based on data reported to the 2014 NEI, permit data, state modeling, and review of existing materials developed in the 2001 NESHAP (Zukor and Ali, 2000a, 2000b). Outlet concentration based on 20% of LEL.
[b] Parameters based on data reported to the 2014 NEI, permit data, and review of 2001 NESHAP materials. Outlet concentration based on 25% of LEL.
[c] Parameters based on data reported to the 2014 NEI, permit data, and review of 2001 NESHAP materials. Outlet concentration based on 50% of LEL.
[d] Parameters based on data reported to the 2014 NEI and permit data. Outlet concentration based on 30% of LEL.
[e] See section 5.1.2 of this memorandum for calculation of baseline n-hexane emissions.



