United States Environmental Protection Agency Office of Air Quality Planning and Standards Research Triangle Park NC 27711 August 1981 Air IvvEPA Guideline Series Control of Volatile Organic Compound Fugitive Emissions from Synthetic Organic Chemical, Polymer, and Resin Manufacturing Equipment Draft ------- NOTICE This document has not been formally released by EPA and should not now be construed to represent Agency policy. It is being circulated for comment on its technical accuracy and policy implications. Control of Volatile Organic Compound Fugitive Emissions from Synthetic Organic Chemical, Polymer, ^ Resin Manufacturing Equipment Emission Standards and Engineering Division Contract No. 68-02-3168 Prepared for U.S. ENVIRONMENTAL PROTECTION AGENCY Office of Air, Noise, and Radiation Office of Air Quality Planning and Standards Research Triangle Park, North Carolina 27711 August 1981 • ------- GUIDELINE SERIES The guideline series of reports is issued by the Office of Air Quality Planning and Standards (OAQPS) to provide information to state and local air pollution control agencies; for example, to provide guidance on the acquisition and processing of air quality data and on the planning and analysis requisite for the maintenance of air quality. Reports published in this series will be available - as supplies permit - from the Library Services Office (MD-35), U. S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, or for a nominal fee, from the National Technical Information Service, 5285 Port Royal Road, Springfield, Virginia 22161. 11 ------- TABLE OF CONTENTS Page List of Tables v List of Figures viii Chapter 1.0 Introduction 1-1 Chapter 2.0 Processes and Pollutant Emissions 2-1 2.1 Introduction 2-1 2.2 Facilities and Their Emissions 2-3 2.3 Model Units 2-17 Chapter 3.0 Emission Control Techniques. .' 3-1 3.1 Leak Detection and Repair Methods 3-1 3.2 Other Control Strategies 3-14 3.3 Other Considerations 3-18 Chapter 4.0 Environmental Analysis of RACT 4-1 4.1 Introduction 4-1 4.2 Air Pollution 4-2 4.3 Water Pollution 4-6 4.4. Solid Waste Disposal 4-6 4.5 Energy 4-6 Chapter 5.0 Control Cost Analysis of RACT 5-1 5.1 Basis for Capital Costs 5-1 5.2 Basis for Annualized Costs 5-3 5.3 Emission Control Costs 5-8 5.4 Cost Effectiveness 5-12 m ------- TABLE OF CONTENTS (continued) Page Appendix A. Emission Source Test Data A-l Appendix B. List of Chemicals Defining Synthetic Organic Chemical, Polymer, and Resin Manufacturing Industries B-l Appendix C. Method 21. Determination of Volatile Organic Compound Leaks C-l Appendix D. Example Calculations for Determining Reduction in Emissions from Implementation of RACT D-l iv ------- LIST OF TABLES Page Table 2-1 Fugitive Emission Sources for Three Model Units . . 2-19 Table 2-2 Uncontrolled Fugitive Emission Factors in Process Unit Equipment 2-21 Table 2-3 Estimated Total Fugitive Emissions from Model Units 2-22 Table 2-4 Average Percent of Total Fugitive Emissions Attributed to Specific Component Types 2-23 Table 3-1 Percentage of Emissions as a Function of Action Level 3-6 Table 3-2 Estimated Occurrence and Recurrence Rate of Leaks for a Quarterly Monitoring Interval 3-7 Table 3-3 Average Emission Rates from Sources Above 10,000 ppmv and at 1000 ppmv 3-9 Table 3-4 Impact of Monitoring Interval on Correction Factor Accounting for Leak Occurrence/Recurrence (For Example Calculation) 3-12 Table 3-5 Example of Control Efficiency Calculation 3-13 Table 3-6 Cost Effectiveness Versus Initial Percent of Valves Leaking in Model Units 3-16 Table 3-7 Illustration of Skip-Period Monitoring 3-19 Table 3-8 Cost Effectiveness of Quarterly Leak Detection and Repair for Typical Process Units 3-22 Table 4-1 Estimated Hourly Emissions and Emissions Reduction on a Model Unit Basis 4-3 Table 4-2 Estimated Annual Emissions and Emissions Reduction on a Model Unit Basis 4-3 Table 4-3 Emission Factors for Sources Controlled Under RACT . 4-4 Table 4-4 Example Calculation of VOC Fugitive Emissions from Model Unit A Under RACT 4-5 ------- Table 5-1 Table 5-2 Table 5-3 Table 5-4 Table 5-5 Table 5-6 Table 5-7 Table 5-8 Table 5-9 Table 5-10 Table 5-11 Table A-l Table A- 2 Table A-3 Table A-4 Table A- 5 Table A-6 Table A-7 Table A-8 Capital Cost Data Capital Cost Estimates for Implementing RACT . . . Labor-Hour Requirements for Initial Leak Repair Under RACT Basis for Annual i zed Cost Estimates Annual Monitoring and Leak Repair Labor Requirements for RACT Recovery credits Annuali zed Control Cost Estimates for Model Units Under RACT Cost Effectiveness for Model Units under RACT . . . Cost Effectiveness for Component Types in Model Unit A Cost Effectiveness for Component Types in Model Unit B Cost Effectiveness for Component Types in Model Unit C Frequency of Leaks from Fugitive Emission Sources in Synthetic Organic Chemical Units Twenty-four Chemical Process Units Screened for Fugitive Emissions Summary of SOCMI Process Units Fugitive Emissions . Average Fugitive Emission Source Screening Rates. . Sampled Process Units from Nine Refineries During Refinery Study Leak Frequencies and Emission Factors from Fugitive Emission Sources in Petroleum Refineries Comparison of Leak Frequencies for Fugitive Emission Sources in SOCMI Units and Petroleum Refineries . . Frequency of Leaks from Fugitive Emission Sources in Two DuPont Plants Page 5-2 5-2 5-4 5-5 5-7 5-9 5-11 5-13 5-14 5-15 5-16 A-4 A-5 A-8 A-9 A-10 A-l 2 A-l 3 A-l 4 VI ------- Page Table A-9 Frequency of Leaks from Fugitive Emission Sources in Exxon's Cyclohexane Unit A-16 Table A-10 Summary of Maintenance Study Results from the Union Oil Co. Refinery in Rodeo, California A-18 Table A-ll Summary of Maintenance Study Results from the Shell Oil Company Refinery in Martinez, California. . . . A-20 Table A-12 Summary of EPA Refinery Maintenance Study Results . A-21 Table A-13 Maintenance Effectiveness Unit D Ethylene Unit Block Valves A-22 Table A-14 Occurrence Rate Estimates for Valves and Pumps by Process in EPA-ORD Study A-24 Table A-15 Valve Leak Recurrence Rate Estimates A-25 Table A-16 Summary of Valve Maintenance Test Results A-26 Table D-l Uncontrolled Fugitive Emission Factors in Process Unit Equipment D-2 Table D-2 Controlled Emission Factors for Equipment Affected by RACT D-2 Table D-3 Example Calculation of Uncontrolled Emissions from an Illustrative Process Unit D-3 Table D-4 Uncontrolled Emissions from Components Affected by RACT D-4 Table D-5 Controlled Emissions from Components Affected by RACT D-4 Table D-6 Emission Reduction Expected from RACT D-5 vn ------- LIST OF FIGURES Page Figure 2-1 General schematic of process levels that make up the organic chemical industry 2-2 Figure 2-2 Diagram of a simple packed seal 2-4 Figure 2-3 Diagram of a basic single mechanical seal 2-5 Figure 2-4 Diagram of a double mechanical seal 2-6 Figure 2-5 Diagram of a double mechanical seal 2-6 Figure 2-6 Diaphragm pump 2-7 Figure 2-7 Labyrinth shaft seal 2-9 Figure 2-8 Restrictive-ring shaft seal 2-9 Figure 2-9 Mechanical (contact) shaft seal 2-10 Figure 2-10 Liquid film shaft seal with cylindrical bushing . . . 2-10 Figure 2-11 Diagram of a gate valve 2-11 Figure 2-12 Example of bellows seals 2-12 Figure 2-13 Diagrams of valves with diaphragm seals 2-13 Figure 2-14 Diagram of a spring-loaded relief valve 2-14 Figure 2-15 Diagram of hydraulic seal for agitators 2-16 Figure 2-16 Diagram of agitator lip seal 2-16 Figure 3-1 Cumulative distribution of total emissions by screening values - valves in light liquid service 3-10 Figure 3-2 Cumulative distribution of sources by screening values - valves in light liquid service 3-10 Figure C-l Calibration Precision Determination C-8 Figure C-2 Response Time Determination C-9 vm ------- 1.0 INTRODUCTION The Clean Air Act Amendments of 1977 require each State in which there are areas in which the national ambient air quality standards (NAAQS) are exceeded to adopt and submit revised state implementation plans (SIP's) to EPA. Revised SIP's were required to be submitted to EPA by January 1, 1979. States which were unable to demonstrate attainment with the NAAQS for ozone by the statutory deadline of December 31, 1982, could request extensions for attainment with the standard. States granted such an extension are required to submit a further revised SIP by July 1, 1982. Section 172(a)(2) and (b)(3) of the Clean Air Act require that nonattainment area SIP's include reasonably available control technology (RACT) requirements for stationary sources. As explained in the "General Preamble for Proposed Rulemaking on Approval of State Implementation Plan Revisions for Nonattainment Areas," (44 FR 20372, April 4, 1979) for ozone SIP's, EPA permitted States to defer the adoption of RACT regul-ations on a category of stationary sources of volatile organic compounds (VOC) until after EPA published a control techniques guideline (CTG) for that VOC source category. See also 44 FR 53761 (September 17, 1979). This delay allowed the states to make more technically sound decisions regarding the application of RACT. Although CTG documents review existing information and data concerning the technology and cost of various control techniques to reduce emissions, they are, of necessity, general in nature and do not fully account for unique variations within a stationary source category. Consequently, the purpose of CTG documents is to provide State and local air pollution control agencies with an initial information base for proceeding with their own analysis of RACT for specific stationary sources. 1-1 ------- 2.0 PROCESSES AND POLLUTANT EMISSIONS 2.1 INTRODUCTION The discussion presented in this document applies to equipment in process units which manufacture synthetic organic chemicals and polymers and resins. The equipment in process units in the synthetic organic chemical manufacturing industry (SOCMI) is similar to equipment in the polymer and resin manufacturing industry. Both industries process volatile organic compounds. Therefore, the information and discussion presented in this chapter and subsequent chapters applies equally to SOCMI plants and polymer and resin plants. The SOCMI is a segment of the chemical industry consisting of some of the higher volume intermediate and finished products. A list of these chemicals is presented in Appendix B, Table I. The polymer and resin manufacturing industries to which the discussion in this document applies are presented in a list in Appendix B, Table II. It should be emphasized that the discussion in this document are intended to apply to equipment in process units which manufacture these chemicals. Most of the SOCMI chemicals produced in the United States are derived from crude petroleum or natural gas. The ten principal feedstocks used in the manufacture of organic chemicals are produced primarily in petroleum refineries. After chemical feedstocks are manufactured from petroleum, natural gas, and other raw materials, they are processed into chemical intermediates and end-use chemicals (see Figure 2-1). Approximately 12 percent of the plants in the United States produce less than 5,000 megagrams (Mg) annually. Another 12 percent have production capacities in excess of 500,000 Mg. 2-1 ------- RAW MATERIALS (CRUDE OIL, CRUDE NATURAL GAS, ETC.) CHEMICAL REFINERIES | FEEDSTOCK | PLANTS CHEMICAL FEEDSTOCKS CHEMICAL PLANTS CHEMICAL PRODUCTS Fi gure 2-1. General schematic of process levels that make up the organic chemical industry. 2-2 ------- The polymer and resin manufacturing industry includes operations which convert monomer or chemical intermediate materials obtained from the basic petrochemical industry and the SOCMI into polymer products. Such products include plastic materials, synthetic resins, and synthetic rubbers. 2.2 FACILITIES AND THEIR EMISSIONS 2.2.1 Potential Source Characterization and Description In this document, fugitive emissions from process units are considered to be those volatile organic compound (VOC) emissions that result when process fluid (either gaseous or liquid) leaks from plant equipment. There are many potential sources of fugitive emissions in a typical process unit. The following sources will be considered in this chapter: pumps, compressors, in-line process valves, pressure relief devices, open-ended valves, sampling connections, flanges, agitators and cooling towers. These potential sources are described below. 2.2.1.1 Pumps. Pumps are used extensively in process units for the movement of organic liquids. The centrifugal pump is the most widely used pump. However, other types, such as the positive-displacement, reciprocating and rotary action, and special canned and diaphragm pumps, are also used. Chemicals transferred by pumps can leak at the point of contact between the moving shaft and stationary casing. Consequently, all pumps except the shaftless type (canned-motor and diaphragm) require a seal at the point where the shaft penetrates the housing in order to isolate the pump's interior from the atmosphere. Two generic types of seals, packed and mechanical, are currently in use on pumps. Packed seals can be used on both reciprocating and rotary action types of pumps. As Figure 2-2 shows, a packed seal consists of a cavity ("stuffing box") in the pump casing filled with special packing material that is compressed with a packing gland to form a seal around the shaft. Lubrication is required to prevent the buildup of frictional heat between the seal and shaft. The necessary lubrication is provided by a lubricant that flows between the packing and the shaft.2 Deterioration of the packing will result in process liquid leaks. 2-3 ------- End C — i Stuffing Box L * nn U CxixxixMxExl Packing / Gland rf Fnrf 1XIXDOXIXIXIX1 , / 1 r / UJd Packing \ / Possible Leak Area Figure 2-2. Diagram of a simple packed seal.3 Mechanical seals are limited in application to pumps with rotating shafts and can be further categorized as single and double mechanical seals. There are many variations to the basic design of mechanical seals, but all have a lapped seal face between a stationary element and a rotating seal ring. In a single mechanical seal application (Figure 2-3), the rotating-seal ring and stationary element faces are lapped to a very high degree of flatness to maintain contact throughout their entire mutual surface area. As with a packed seal, the seal faces must be lubricated to remove frictional heat, however, because of its construction, much less lubricant is needed. A mechanical seal is not a leak-proof device. Depending on the condition and flatness of the seal faces, the leakage rate can be quite low (as small as a drop per minute) and the flow is often not visually detectable. In order to minimize fugitive emissions due to seal leakage, an auxiliary 4 sealing device such as packing can be employed. 2-4 ------- PUMP STUFFING BOX GLAND •RING FLUID END SHAFT \ROTATING SEAL RING STATIONARY ELEMENT POSSIBLE LEAK AREA Figure 2-3. Diagram of a basic single mechanical seal. In a dual mechanical seal application, two seals can be arranged back-to-back or in tandem. In the back-to-back arrangement (Figure 2-4), the two seals provide a closed cavity between them. A seal liquid, such as water or seal oil, is circulated through the cavity. Because the seal liquid surrounds the double seal and lubricates both sets of seal faces in this arrangement, the heat transfer and seal life characteristics are much better than those of the single seal. In order for the seal to function, the seal liquid must be at a pressure greater than the operating pressure of the stuffing box. As a result some seal liquid will leak across the seal faces. Liquid leaking across the inboard face will enter the stuffing box and mix with the process liquid. Seal liquid going across the outboard face will exit to the atmosphere. ' In a tandem dual mechanical seal arrangement (Figure 2-5), the seals face the same direction. The secondary seal provides a backup for the primary seal. A seal flush is used in the stuffing box to remove the heat generated by friction. The cavity between the two seals is filled with a buffer or barrier liquid. However, the barrier liquid is at a pressure lower 2-5 ------- SEAL LIQUID. POSSIBLE LEAK INTO SEALING FLUID FLUID END w \/ o PRIMARY — V SEAL \ * — SECONDARY SEAL GLAND PLATE Figure 2-4. Diagram of a double mechanical seal (back-to-back arrangement)7 PRIMARY SEAL SUFFER LIQUID OUT IN (TOP) (BOTTOM) V SECONDARY SEAL GLAND PLATE 70-1787-1 Figure 2-5. Diagram of a double mechanical seal (tandem arrangement)8 2-6 ------- than that in the stuffing box. Therefore, any leakage will be from the stuffing box into the seal cavity containing the barrier liquid. Since this liquid is routed to a closed reservoir, process liquid that has leaked into the seal cavity will also be transferred to the reservoir. At the reservoir, the process liquid could vaporize and be emitted to the atmosphere. To ensure that VOC's do not leak from the reservoir, the reservoir can be vented to a control device. Another type of pump that has been used is the shaft!ess pump which includes canned-motor and diaphragm pumps. In canned-motor pumps the cavity housing the motor rotor and the pump casing are interconnected. As a result, the motor bearings run in the process liquid and all seals are eliminated. Because the process liquid is the bearing lubricant, abrasive solids cannot be tolerated. Canned-motor pumps are being widely used for handling organic solvents, organic heat transfer liquids, light oils, as well as many toxic or hazardous liquids, or where leakage is an economic problem.10 Diaphragm pumps (see Figure 2-6) perform similarly to piston and plunger pumps. However, the driving member is a flexible diaphragm fabricated of metal, rubber, or plastic. The primary advantage of this arrangement is the elimination of all packing and seals exposed to the process liquid. This is an important asset when hazardous or toxic liquids are handled.11 DISCHARGE CHECK VALVE INLET CHECK VALVE DIAPHRAGM PISTON Figure 2-6. Diaphragm pump. 12 2-7 ------- 2.2.1.2 Compressors. Gas compressors used in process units are similar to pumps in that they can be driven by rotary or reciprocating shafts. They are also similar to pumps in their need for shaft seals to isolate the process gas from the atmosphere. As with pumps, these seals are likely to be the source of fugitive emissions from compressors. Shaft seals for compressors may be chosen from several different types: labyrinth, restrictive carbon rings, mechanical contact, and liquid film. All of these seal types are leak restriction devices; none of them completely eliminate leakage. Many compressors may be equipped with ports in the seal area to evacuate gases collecting there. The labyrinth type of compressor seal is composed of a series of close tolerance, interlocking "teeth" which restrict the flow of gas along the shaft. A straight pass labyrinth compressor seal is shown in Figure 2-7. Many variations in "tooth" design and materials of construction are available. Although labyrinth type seals have the largest leak potential of the different types, properly applied variations in "tooth" configuration and shape can 13 reduce leakage by up to 40 percent over a straight pass type labyrinth. Restrictive carbon ring seals consist of multiple stationary carbon rings with close shaft clearances. This type of seal may be operated dry or with a sealing fluid. Restrictive ring seals can achieve lower leak rates than the labyrinth type. A restrictive ring seal is shown in Figure 2-8. Mechanical contact seals (shown in Figure 2-9) are similar to the mechanical seals described for pumps. In this type of seal clearance between the rotating and stationary elements is reduced to zero. Oil or another suitable lubricant is supplied to the seal faces. Mechanical seals can achieve the lowest leak rates of the types described here, but they are not suitable for all processing conditions. Centrifugal compressors also can be equipped with liquid film seals. A diagram of a liquid film seal is shown in Figure 2-10. The seal is formed by a film of oil between the rotating shaft and stationary gland. When the circulating oil is returned to the oil reservoir, process gas can be released to the atmosphere.18 To eliminate release of VOC emissions from the seal oil system, the reservoir can be vented to a control device. 2-8 ------- PORT MAY BE ADDED FOR SCAVENGING OR INERT-GAS SEALING INTERNAL GAS PRESSURE '//'///'/.'///'/) ATMOSPHERE Figure 2-7. Labyrinth shaft seal. 14 SCAVENGING PORT MAY BE AOOEO FOR VACUUM APPLICATION ATMOSPHERE Figure 2-8. Restrictive-ring shaft seal. 15 2-9 ------- INTERNAL GAS PRESSURE CLEAN OIL IN PRESSURE /• BREAKDOWN .' SLEEVE STATIONARY SEAT CARSON RING ATMOSPHERE CONTAMINATED OIL OUT Figure 2-9. Mechanical (contact) shaft seal. ^- CLEAN OIL IN INNER BUSHING OUTER BUSHING ATMOSPHERE CONTAMINATED OIL OUT OIL OUT Figure 2-10. Liquid film shaft seal with cylindrical bushing. 2-.10 ------- 2.2.1.3 Process Valves. One of the most common pieces of equipment in organic chemical plants is the valve. The types of valves commonly used are control, globe, gate, plug, ball, relief, and check valves. All except the relief valve (to be discussed further below) and check valve are activated by a valve stem, which may have either a rotational or linear motion, depending on the specific design. This stem requires a seal to isolate the process fluid inside the valve from the atmosphere as illustrated by the diagram of a gate valve in Figure 2-11. The possibility of a leak through this seal makes it a potential source of fugitive emissions. Since a check valve has no stem or subsequent packing gland, it is not considered to be a potential source of fugitive emissions. Sealing of the stem to prevent leakage can be achieved by packing inside a packing gland or 0-ring seals. Va'lves that require the stem to move in and out with or without rotation must utilize a packing gland. Conventional packing glands are suited for a wide variety of packing materials. The most common are various types of braided asbestos that contain lubricants. Other packing materials include graphite, graphite-impregnated fibers, and tetrafluoroethyler tetrafluoroethylene. The packing material used depends on the valve application 19 and configuration. These conventional packing glands can be used over a wide range of operating temperatures. At high pressures these glands must be 20 quite tight to attain a good seal. PACKING GLAND PACKING VALVE STEM POSSIBLE LEAK AREAS Figure 2-11. Diagram of a gate valve. 21 2-11 ------- Elastomeric 0-rings are also used for sealing process valves. These 0-rings provide good sealing but are not suitable where there is sliding motion through the packing gland. Those seals are rarely used in high pressure service and operating temperatures are limited by the seal material.22 Bellows seals are more effective for preventing process fluid leaks than the conventional packing gland or any other gland-seal arrangement.23 This type of seal incorporates a formed metal bellows that makes a barrier between the disc and body bonnet joint. An example of this seal is presented in Figure 2-12. The bellows is the weak point of the system and service life can be quite variable. Consequently, this type of seal is normally backed up with a conventional packing gland and is often fitted with a leak detector in 24 case of failure. BELLOWS BODY BONNET Figure 2-12. Example of bellows seals. 25 A diaphragm may be used to isolate the working parts of the valve and the environment from the process liquid. Tw.o types of valves which utilize diaphragms are illustrated in Figures 2-13(a) and (b). As Figure 2-13(b) shows, the diaphragm may also be used to control the flow of the process fluid. In this design, a compressor component pushes the diaphragm toward the valve bottom, throttling the flow. The diaphragm and compressor are 2-12 ------- connected in a manner so that it is impossible for them to be separated under normal working conditions. When the diaphragm reaches the valve bottom, it seats firmly against the bottom, forming a leak-proof seal. This configuration is recommended for fluids containing solid particles and for medium-pressure service. Depending on the diaphragm material, this type of valve can be used at temperatures up to 205°C and in severe acid solutions. If failure of the seal occurs, a valve employing a diaphragm seal can become a source of fugitive emissions. DIAPHRAGM DISK STEM DIAPHRAGM Figure 2-13. Diagrams of valves with diaphragm seals. 27 2.2.1.4 Pressure Relief Devices. Engineering codes require that pressure-relieving devices or systems be used in applications where the process pressure may exceed the maximum allowable working pressure of the vessel. The most common type of pressure-relieving device used in process units is the pressure relief valve (Figure 2-14). Typically, relief valves are spring-loaded and designed to open when the process pressure exceeds a set pressure, allowing the release of vapors or liquids until the system pressure is reduced to its normal operating level. When the normal pressure 28 is reattained, the valve reseats, and a seal is again formed. The seal is a disk on a seat, and the possibility of a leak through this seal makes the pressure relief valve a potential source of VOC fugitive emissions. Two 2-13 ------- potential causes of leakage from relief valves are: "simmering or popping", a condition due to the system pressure being close to the set pressure of the valve, and improper reseating of the valve after a relieving operation.29 Possible Leak Area Process Side Figure 2-14. Diagram of a spring-loaded relief valve. Rupture disks are also common in process units. These disks are made of a material that ruptures when a set pressure is exceeded, thus allowing the system to depressurize. The advantage of a rupture disk is that the disk seals tightly and does not allow any VOC's to escape from the system under normal operation. However, when the disk does rupture, the system depressurizes until atmospheric conditions are obtained. This could result in an excessive loss of product or a corresponding excessive release of fugitive emissions. 2-14 ------- 2.2.1.5 Agitators. Agitators are commonly used to stir or blend chemicals. Like pumps and compressors, agitators may leak organic chemicals at the point where the shaft penetrates the casing. Consequently, seals are required to minimize fugitive emissions from agitators. Four seal arrangements are commonly used with agitators. These are compression packing (packed seal), mechanical seals, hydraulic seals, and lip seals.31 Packed seals for agitators are very similar in design and application to the packed seals for pumps (Section 2.2.1.1). Although mechanical seals are more costly than the other three seal arrangements, they offer a greatly reduced leakage rate to offset their higher cost. The maintenance frequency of mechanical seals is, also, 32 one-half to one-fourth that of packed seals. In fact, at pressures greater than 1135.8 kPa (150 psig), the leakage rate and maintenance frequency ' 33 are so superior that the use of packed seals on agitators is rare. As with packed seals, the mechanical seals for agitators are similar to the design and application of mechanical seals for pumps (Section 2.2.1.1.) The hydraulic seal (Figure 2-15) is the simplest and least used agitator shaft seal. In this -type of seal, an annular cup attached to the process vessel contains a liquid that is in contact with an inverted cup attached to the rotating agitator shaft. The primary advantage of this seal is that it is a non-contact seal. However, this seal is limited to low temperatures and pressures and can only handle very small pressure fluctuations. Organic cherricals may contaminate the seal liquid and then be released into the 34 atmosphere as fugitive emissions. INVERTED CUP ANNULAR CUP Figure 2-15. Diagram of a hydraulic seal for agitators. 2-15 35 ------- A lip seal (Figure 2-16) can be used on a top-entering agitator as a dust or vapor seal. The sealing element is a spring-loaded elastomer. Lip seals are relatively inexpensive and easy to install. Once the seal has beer installed the agitator shaft rotates in continuous contact with the lip seal. Pressure limits of the seal are 2 to 3 psi because it operates without lubrication. Operating temperatures are limited by characteristics of the elastomer. Fugitive VOC emissions could be released through this seal when this seal wears excessively or the operating pressure surpasses 36 the pressure limits of the seal. •*— —*• 37 Figure 2-16. Diagram of agitator lip seal. 2.2.1.6 Open-Ended Valves or Lines. Some valves are installed in a systeir so that they function with the downstream line open to the atmosphere. Examples are purge valves, drain valves, and vent valves. A faulty valve seat or incompletely closed valve would result in leakage through the valve and fugitive VOC emissions to the atmosphere. 2.2.1.7 Sampling Connections. The operation of a process unit is checked periodically by routine analyses of feedstocks and products. To obtain representative samples for these analyses, sampling lines must first be purged prior to sampling. The purged, liquid or vapor is sometimes drained onto the ground or into a sewer drain, where it can evaporate and release VOC emissions to the atmosphere. 2-16 ------- 2.2.1.8 Flanges. Flanges are bolted, gasket-sealed junctions used wherever pipe or other equipment such as vessels, pumps, valves, and heat exchangers may require isolation or removal. Normally, flanges are employed for pipe diameters for 50 mm or greater and are classified by pressure and face type. Flanges may become fugitive emission sources when leakage occurs due to improperly chosen gaskets or a poorly assembled flange. The primary cause of flange leakage is due to thermal stress that piping or flanges in some services undergo; this results in the deformation of the seal between the flange faces.38 2.3 MODEL UNITS This section presents model process unit parameters. The model units were selected to represent the range of processing complexity in the industry. They provide a basis for determining environmental and cost impacts of reasonably available control technology (RACT). 2.3.1 Model Units Available data show that fugitive emissions are proportional to the number of potential sources, but are not related to capacity, throughput, 39 age, temperature, or pressure. Therefore, model units defined for this analysis represent different levels of process complexity (number of sources) rather than different unit sizes. 2.3.1.1 Sources of Fugitive Emissions. Data from petroleum refineries 40 indicate that cooling towers are very small sources of VOC emission. Differences in operating procedures, such as recirculation of process water, might result in cooling tower VOC emissions, but no data are available to verify this. The number of agitator seals in the industry is not known. Furthermore, the emission rate from agitator seals has not been measured. Since there are no data from similiar sources in other industries, no estimates of emission rate can be made. Because of these uncertainties, cooling towers and agitator seals are not included in the Model Units. 2.3.1.2 Model Units Components. In order to estimate emissions, control costs, and environmental impacts for process units on a unit specific basis, three model units were developed. The equipment components comprising the 2-17 ------- model units are shown in Table 2-1. These three model units represent the range of emission source populations that may exist in SOCMI process units. The number of equipment components for each model unit was developed from a data base compiled by Hydroscience, Inc. The data base included equipment source counts from 62 SOCMI plants which produce 35 different chemicals. These plant sites represent approximately 5 percent of the total existing SOCMI plants and include large and small capacities, batch and continuous production methods, and varying levels of process complexity. The source counts for the 35 chemicals include pumps, valves, and compressors. These counts were used in combination with the number of sites which produce each chemical in order to determine the average number of sources per site. Hydroscience estimates that 52 percent of existing SOCMI plants are similar to the Model Unit A, 33 percent are similar to B, and 15 percent are similar to C. Dcta from petroleum refineries indicate that emission rates of sources decrease as the vapor pressure (volatility) of the process fluid decreases. Three classes of volatility have been established based on the petroleum refinery data. These include gas/vapor service, light liquid service, and heavy liquid service. The split between light and heavy liquids for the refinery data is between naphtha and kerosene. Since similar stream names may have different vapor pressures, depending on site specific factors, it is difficult to quantify the light-heavy split. The break point is approximately at a vapor pressure of 0.3 kPa at 20°C. The data collected by Hydroscierce were used to estimate the split between gas/vapor and liquid service for each 44 source. In order to apply emission factors for light and heavy liquid service, it is assumed that one-half of SOCMI liquid service sources are in light liquid service. There are no data available on the actual distribution of sources in volatility ranges. It is assumed that all packed seal pumps are in heavy liquid service. This assumption is reasonable, since more volatile liquid are more suitable for mechanical seal applications, and newer process units tend to use fewer packed seals. Sampling connections are a subset, of the open-ended valve category. Approximately 25 percent of 45 open-ended valves are used for sampling connections. Emissions which occur through the valve stem, gland, and open-end are included in the cpen-ended valve category. The emission factor for sampling connection applies only to emissions which result from sample purging. 2-18 ------- TABLE 2-1. FUGITIVE EMISSION SOURCES FOR THREE MODEL UNITS Equipment component3 Pump seals Light liquid service0 Single mechanical Double mechanical Sea 11 ess Heavy liquid service0 Single mechanical Packed Valves Gas service Light liquid service Heavy liquid service Safety/relief valves Gas service Light liquid service Heavy liquid service Open-ended valves and lines6 Gas service Light liquid service Heavy liquid service Compressor seals Sampling connections Flanges Cooling towers Number Model unit A 5 3 0 5 2 90 84 84 11 1 1 9 47 48 1 26 600 _-9 of components in Model unit B 19 10 1 24 6 365 335 335 42 4 4 37 189 189 2 104 2400 ..9 model unit Model unit C 60 31 1 73 20 1117 1037 1037 130 13 14 115 581 581 8 320 7400 ..9 aEquipment components in VOC service only. b52 percent of existing SOCMI units are similar to model unit A. 33 percent of existing SOCMI units are similar to model unit B. 15 percent of existing SOCMI units are similar to model unit C. cLight liquid is defined as a fluid with vapor pressure greater than 0.3 kPa at 20°C. This vapor pressure represents the split between kerosene and naphtha and is based on data presented in reference 39. dHeavy liquid is defined as a fluid with vapor pressure less than 0.3 kPa at 20°C. This vapor pressure represents the split between kerosene and naphtha and is based on data presented in reference 39. eSample, drain, and purge valves. fBased on 25 percent of open-ended valves. Reference 1, pg. IV-3. 90ata not available. 2-19 ------- 2.3.1.3 Uncontrolled Fugitive Emission Estimates. Data characterizing the uncontrolled levels of fugitive emissions in the SOCMI are presently unavailable. However, data on this type have been obtained for the refining industry. These data are presented in Table 2-2, and represent the average uncontrolled emission rate from each of the components of a specific type in the process unit. Because the operation of the various process equipment in the SOCMI is not expected to differ greatly from the operation of the same equipment in the refining industry, the refinery fugitive emission data can be used to approximate the levels of fugitive emissions in SOCMI. The total amount of VOC emitted from fugitive sources can be estimated for each Model Unit. Total hourly emissions can be calculated by multiplying the number of pieces of each type of equipment (Table 2-1) by the corresponding hourly emission factor (Table 2-2). The total annual emissions have been calculated by multiplying the total hourly emissions for each Model Unit by the number of hours in a yeer (8,760 hours/year). These estimated annual emission rates appear in Table 2-3. The average percent of total VOC emissions attributed to each component type is presented in Table 2-4. The percent attributed to each component type is the same for each model unit. 2-20 ------- TABLE 2-2. UNCONTROLLED FUGITIVE EMISSION FACTORS IN PROCESS UNIT EQUIPMENT Uncontrolled emission Fugitive emission source factor,9 kg/hr Pumps . Light liquids With packed seals 0.12 Wtih single mechanical seals 0.12 With double mechanical seals 0.12 With no seals 0.0 Heavy liquids With packed seals 0.020 With single mechanical seals 0.020 With double mechanical seals 0.020 With no seals 0.0 t Valves (in-line) Gas . 0.021 Light liquid^ 0.010 Heavy liquid0 0.0003 Safety/relief valves Gas . 0.16 Light liquid° 0.006 Heavy liquid0 0.009 Open-ended valves Gas . 0.025 Light liquid^ 0.014 Heavy liquid0 0.003 Flanges 0.0003 Sampling connections 0.015 Compressors 0-44 e Cooling towers 13.6-1107 Agitators NAT aThese uncontrolled emission levels are based upon the refinery data presented in reference 39- bLight liquid is defined as a fluid with vapor pressure greater than 0.3 kPa at 20°C. This vapor pressure represents the split between kerosene and naphtha and is based on data presented in reference 39. cAssumes the inner seal leaks at the same rate as single seal and that the VOC is emitted from the seal oil degassing vent. dHeavy liquid is defined as a fluid with vapor pressure less than 0.3 kPa at 20°C. This vapor pressure represents the split between kerosene and naphtha and is based on data presented in reference 40. eThese levels are based on cooling tower circulation rates that range from 0.05-3.66 ms/sec (714-58,000 GPM). Reference 46- NA no data available. 2-21 ------- TABLE 2-3. ESTIMATED TOTAL FUGITIVE EMISSIONS FROM MODEL UNITSa Model unit Model unit Model unit ABC Estimated total emissions (Mg/yr) 67 260 800 aBased upon equipment component counts in Table 2-1, the uncontrolled emission factors in Table 2-2, and 8,760 hours/yr. 2-22 ------- TABLE 2-4. AVERAGE PERCENT OF TOTAL FUGITIVE EMISSIONS ATTRIBUTED TO SPECIFIC COMPONENT TYPES Component Percent of total uncontrolled emissions attributed to to component type for model units A,B,C Pump seals Light liquid service Heavy liquid service In-line valves Gas service Light liquid service Heavy liquid service Safety/relief valves Gas service Light liquid service Heavy liquid service Open-ended valves Gas service Light liquid service Heavy liquid service Compressor seals Sampling connections Flanges 12 2 26 11 23 3 9 2 4 5 2 Less than one percent. 2-23 ------- 2.4 REFERENCES 1. Erikson, D.G., and V. Kalcevic. Emissions Control Options for the Synthetic Organic Chemicals Manufacturing Industry, Fugitive Emissions Report, Draft Final. Hydroscience, Inc., 1979. p. II-2. 2. Ref. 1. 3. Ref. 1, p. II-3. 4. Ramsden, J.H. How to Choose and Install Mechanical Seals. Chem. E., 85(22):97-102. 1978. 5. Ref. 1, p. II-3. 6. Ref. 4, p. 99. 7. Ref. 4, p. 99. 8. Ref. 4, p. 99. 9. Ref. 4, p. 99. 10. Perry, R.H., and C.H. Chi 1 ton, Chemical Engineers' Handbook, 5th Ed. New York, McGraw-Hill Book Company, 1973. p. 6-8. 11. Ref. 10, p. 6-13. 12. Nurken, R.F. Pump Selection for the Chemical Process Industries, Chem. E., Feb. 18, 1974. p. 120. 13. Nelson, W.E. Compressor Seal Fundamentals. Hydrocarbon Processing, 56_( 12): 91-95. 1977. 14. American Petroleum Institute, "Centrifugal Compressors for General Refinery Service", API Standard 617, Fourth Edition, November, 1979, p. 8. Reprinted by courtesy of the American Petroleum Institute. 15. Reference 14, p. 9. 16. Ref. 13. 17. Ref. 13. 18. Ref. 1, p. 11-7. 19. Lyons, J.L., and C.L. Ashland, Jr. Lyons' Encyclopedia of Valves. New York, Van Nostrand Reinhold Co., 1975. 290p. 20. Templeton, H.C. Valve Installation, Operation and Maintenance. Chem. E., 78(23)141-149, 1971. 2-24 ------- References (continued) 21. Ref. 1, p. II-5. 22. Ref. 18, p. 147-148. 23. Ref. 18, p. 148. 24. Ref. 18, p. 148. 25. Ref. 18, p. 148. 26. Pikulik, A. Manually Operated Valves. Chem. E., April 3, 1978. p. 121. 27. Ref. 24, p. 121. 28. Steigerwald, B.J. Emissions of Hydrocarbons to the Atmosphere from Seals on Pumps and Compressors. Report No. 6, PB 216 582, Joint District, Federal and State Project for the Evaluation of Refinery Emissions. Air Pollution Control District, County of Los Angeles, California. April 1958. 37 p. 29. Ref. 1, p. II-7. 30. Cooling Tower Fundamentals and Application Principles. Kansas City, Missouri, The Marley Company, 1969. p. 4. 31. Ramsey, W.D. and G.C. Zoller. How the Design of Shafts, Seals and Impeller Affects Agitator Performance. Chem. E., 83_(18): 101-108. 1976. 32. Ref. 29, p. 105. 33. Ref. 29, p. 105. 34. Ref. 29, p. 105. 35. Ref. 29, p. 106. 36. Ref. 29, p. 106. 37. Ref. 29, p. 106. 38. McFarland, I. Preventing Flange Fires. Chem. E. Prog., 65_(8): 59-61. 1969. 39. Wetherold, R.G., et al. Emission Factors and Frequency of Leak Occurrence for Fittings in Refinery Process Units, interim report, EPA Contract No. 68-02-2665. Austin, Texas, Radian Corporation, February 1979. pp. 11-49. 2-25 ------- References (continued) 40. Radian Corporation. The Assessment of Environmental Emissions From Oil Refining. Draft Report, Appendix B. EPA Contract No. 68-02-2147, Exhibit B. Austin, Texas. August, 1979. pp. 3-5 through 3-16. 41. Ref. 1, pp. IV-1, 2. 42. Ref. 1, p. II-9-13. 43. Ref. 37,p p. 11-23. 44. Ref. 1, p. 11-10. 45. Ref. 1, p. IV-8. 46. Letter with Attachments from J.M. Johnson, Exxon Company, U.S.A., to Robert T. Walsh, U.S. EPA. July 28, 1977. 2-26 ------- 3.0 EMISSION CONTROL TECHNIQUES Sources of fugitive VOC emissions from process unit equipment were identified in Chapter 2. This chapter discusses the emission control technique which is considered representative of reasonably available control technology (RACT) for these sources. The estimated control effectiveness of the technique is also presented. 3.1 LEAK DETECTION AND REPAIR METHODS Leak detection and repair methods can b£ applied in order to reduce fugitive emissions from process unit sources. Leak detection methods are used to identify equipment components that are emitting significant amounts of VOC. Emissions from leaking sources may be reduced by three general methods: repair, modification, or replacement of the source. 3.1.1. Individual Component Survey. Each fugitive emission source (pump, valve, compressor, etc.) is checked for VOC leakage in an individual component survey. The source may be checked for leakage by visual, audible, olfactory, or instrument techniques. Visual methods are good for locating liquid leaks, especially pump seal failures. High pressure leaks may be detected by hearing the escaping vapors, and leaks of odorous materials may be detected by smell. Predominant industry practices are leak detection by visual, audible, and olfactory methods. However, in many instances, even very large VOC leaks are not detected by these methods. Portable hydrocarbon detection instruments are the best method for identifying leaks of VOC from equipment components. The instrument is used to sample and analyze the air in close proximity to the potential leak surface by traversing the sampling probe tip over the entire area where leaks may occur. This sampling traverse is called "monitoring" in subsequent descriptions. The VOC concentration of the sampled air is displayed on the instrument meter. The performance criteria for monitoring instruments and a description of instrument survey methods are included in Appendix C. 3-1 ------- The VOC concentration at which maintenance is required is called the "action level". An action level of 10,000 ppmv is considered representative of RACT. Components which have indicated concentrations higher than this "action level" are marked for repair. Emission data indicate that large variations in mass emission rate may occur over short time periods for an individual equipment component. 3.1.2 Repair Methods The following descriptions of repair methods include only those features of each fugitive emission source (pump, valve, etc.) which need to be considered in assessing the applicability and effectiveness of each method. They are not intended to be complete repair procedures. 3.1.2.1 Pumps. Many process units have spare pumps which can be operated while the leaking pump is being repaired. Leaks from packed seals may be reduced by tightening the packing gland. At some point, the packing may deteriorate to the point where further tightening would have no effect or possibly even increase fugitive emissions from the seal. The packing can be replaced with the pump out of service. When mechanical seals are utilized, the pump must be dismantled so the leaking seal can be repaired or replaced. Dismantling pumps may result in spillage of some process fluid causing emissions of VOC. These temporary emissions could be greater than the continued leak from the seal. Therefore the pump should be flushed of VOC as much as possible before opening for seal replacement. 3.1.2.2 Compressors. Leaks from packed seals may be reduced by the same repair procedure that was described for pumps. Other types of seals require that the compressor be out of service for repair. Since most compressors do not normally have spares, repair or replacement of the seal would require a shutdown gf the process. If the leak is small, temporary emissions resulting from a shutdown could be greater than the emissions from the leaking seal. 3-2 ------- 3.1.2.3 Relief Valves. In general, relief valves which leak must be removed in order to repair the leak. In some cases of improper reseating, manual release of the valve may improve the seat seal. In order to remove the relief valve without shutting down the process, a block valve may be installed upstream of the relief valve. A block valve is required upstream of the safety/relief valve in order to permit in-service replacement of the valve if it cannot be repaired. In some chemical plants, installation of a block valve upstream of a pressure relief device may be a common practice. Although allowed by ASME codes , this practice may be forbidden by operating or safety procedures of a particular company. A spare relief valve should be attached while the faulty valve is repaired and tested. After a relief valve has been repaired and replaced, it is possible that the next over-pressure relief will result in another leak. t 3.1.2.4 Valves. Most valves have a packing gland which can be tightened while in service. Although this procedure should decrease the emissions from the valve, in some cases it may actually increase the emission rate if the packing is old and brittle or has been overtightened. Plug-type valves can be lubricated with grease to reduce emissions around the plug. Some types of valves have no means of in-service repair and must be isolated from the process and removed for repair or replacement. Other valves, such as control valves, may be excluded from in-service repair by operating procedures or safety procedures. In many cases, valves cannot be isolated from the process for removal. Most control valves have a manual bypass loop which allows them to be isolated easily, although temporary changes in process operation may allow isolation in some cases. If a process unit must be shut down in order to isolate a leaking valve, the emissions resulting from the shutdown might be greater than the emissions from the valve if allowed to leak until the next scheduled unit turnaround which permits isolation for repair. Depending on site specific factors, it may be possible to repair process valves by injection of a sealing fluid into the source. Injection of sealing fluid has been successfully used to repair leaks from valves in petroleum 2 refineries in California. 3-3 ------- Fugitive emissions from open-ended valves are the result of leakage through the seat of the valve. Approximately 28 percent of valves (excluding safety/relief and .check valves) in VOC service are open-ended. They include drain, purge, sample, and vent valves. Fugitive emissions from open-ended valves can be controlled by installing a cap, plug, flange, or second valve to the open end of the valve. In the case of a second valve, the upstream valve should always be closed first after use of the valves. Each time the cap, plug, flange, or second valves is opened, any VOC which has leaked through the first valves seat will be released. These emissions have not been quantified. The control efficiency will be dependent on the frequency of removal of the cap or plug. Caps, plugs, etc. for open-ended valves do not affect emissions which may occur during use of the valve. These emissions may be caused by line purging for sampling, draining or venting through the open-ended valve. 3.1.2.5 Flanges. In some cases, leaks from flanges can be reduced by replacing the flange gaskets. Most flanges cannot be isolated to permit replacement of the gasket. Data from petroleum refineries show that flanges emit very small amounts of VOC. 3.1.3 Control Effectiveness of Leak Detection and Repair Methods There are several factors which-determine the control effectiveness of a leak detection and repair program; these include: t Action level (leak definition), • Inspection interval (monitoring frequency), • Achievable emission reduction of maintenance, and t Interval between detection and repair of the leak. Some of these factors can be estimated by using data collected from petroleum refineries. 3.1.3.1 Action Level. The action level is the VOC concentration observed during monitoring which defines a leaking component which requires repair. The choice of the action level for defining a leak is influenced by a number of important considerations. First, the percent of total mass emissions which can potentially be controlled by the monitoring and repair 3-4 ------- program can be affected by varying the leak definition, or action level. Table 3-1 gives the percent of total mass emissions affected by the 10,000 ppmv action level for a number of equipment types. The choice of an appropriate leak definition is most importantly limited by the ability to repair leaking components. The ability to repair leaking equipment from above 10,000 ppmv to below 10,000 ppmv has been demonstrated in field testing. (Table A-16, Appendix A). This repair ability has not been as well demonstrated for a 1,000 ppm leak defintion, however. Some available data do not support the conclusion that repairing leaks in the 1,000 to 10,000 ppmv range would result in an overall reduction in emissions. The nature of repair techniques for pipeline valves, for instance, is such that attempts to repair leaks below a certain level by tightening the packing gland may result in an increase in e'missions. In practice, valve packing material can become hard and brittle after extended use. As the packing loses its resiliency, the valve packing gland must be tightened to prevent loss of product to atmosphere. Excessive tightening, however, may cause cracks in the packing, thus increasing the leak rate. 3.1.3.2 Inspection Interval. The length of time between inspections should depend on the expected occurrence and recurrence of leaks after a piece of equipment has been checked or repaired. The choice of the interval affects the emission reduction achievable since more frequent inspection will result in leaking sources being found and repaired sooner. In order to evaluate the effectiveness of the quarterly monitoring interval which is considered representative of RACT, it is necessary to estimate the rate at which new leaks will occur and repaired leaks will recur. The estimates which have been used to evaluate quarterly monitoring are shown in Table 3-2. 3.1.3.3 Allowable Interval Before Repair. If a leak is detected, the equipment should be repaired within a certain time period. The allowable repair time should reflect an interest in eliminating a source of VOC emissions but should also allow the plant operator sufficient time to obtain necessary repair parts and maintain some degree of flexibility in overall plant maintenance scheduling. The determination of this allowable repair time will 3-5 ------- TABLE 3-1. PERCENTAGE OF EMISSIONS AS A FUNCTION OF ACTION LEVEL3 Fraction of mass emissions from sources with leak rates above the 10,000 ppmv action level (as %) Source type Pump seals Light liquid service 87 Heavy liquid service 21 Valves Gas service 98 Light liquid service 84 Heavy liquid service 0 Safety/relief valves 69 Compressor seals 84 Flanqes 0 3-6 ------- TABLE 3-2. ESTIMATED OCCURRENCE AND RECURRENCE RATE OF LEAKS FOR A QUARTERLY MONITORING INTERVAL Component type Estimated Percent of sources percent of leaking at quarterly components inspection from leak leaking occurrence, recurrence, initially and leaks not repaired^ Pump seals Light liquid service Heavy liquid service 23 2 2.3 0.2 Valves Gas service Light liquid service Heavy liquid service 10 12 0 1.0 1.2 0.0 Safety/relief valves Compressor seals 8 33 0.8 3.3 Flanges 0 0.0 Approximate fraction of components with a concentration greater than or equal to 10,000 ppmv prior to repair. ^Estimated that 10 percent of the initial leaks represent subsequent occurence and recurrence rate for quarterly inspections. This estimate is based on engineering judgement. 3-7 ------- affect emission reductions by influencing the length of time that leaking sources are allowed to continue to emit pollutants. Some of the components with concentrations in excess of the leak action level may not be able to be repaired until the next scheduled unit shutdown. The allowable interval before repair considered representative of RACT is fifteen days. The percent of emissions from a component which would be affected by the repair interval if all other contributing factors were 100 percent efficient is 97.9 percent. The emissions which occur between the time the leak is detected and repair is attempted are increased with longer allowable repair intervals. 3.1.3.4 Achievable Emission Reduction. Repair of leaking components will not always result in complete emission reduction. To estimate the emission reduction from repair of equipment it was assumed that leaks are reduced by repair to a level equivalent to a concentration reading of 1,000 ppmv. The average emission rates of components above 10,000 ppmv and at 1,000 ppmv are shown in Table 3-3. 3.1.3.5 Development of Controlled Emission Factors. The uncontrolled emission levels for the emission sources that are typically found in the model plants were previously presented in Chapter 2 (Table 2-2). Controlled VOC emission levels can be calculated by a "controlled emission" factor. This factor can be developed for each type of emission source by using the general expression: Controlled emission factor = Uncontrolled factor - [uncontrolled factor x emission reduction efficiency] The reduction efficiency can be developed by the following expressions and correction factors: p Reduction efficiency = AxBxCxD Where: A = Theoretical Maximum Control Efficiency = fraction of total mass emissions for each source type with VOC concentrations greater than the action level (Table 3-1, Figure 3-1). 3-8 ------- TABLE 3-3. AVERAGE EMISSION RATES FROM SOURCES ABOVE 10,000 PPMV and at 1000 PPMV9 Source type Pump seals Light liquid service Heavy liquid service In-line valves Gas service Light liquid service Heavy liquid service Safety/relief valves Compressor seals Fl anges (Y) Emission rate from sources above 10,000 ppmv (kg/hr) 0.45 0.21 0.21 0.07 0.005 1.4 1.1 0.003 (X) Emission rate from sources at 1000 ppmvb (kg/hr) 0.035 0.035 0.001 0.004 0.004 0.035 0.035 0.002 Y-X Percentage reduction 92.2 83.3 99.5 94.3 20.0 97.5 96.8 33.3 ng ng ^Average emission rate of all sources, within a source values above 10,000 ppmv. ^Emission rate of all sources, within a source type, having screening values of 1000 ppmv. 3-9 ------- 100 CO 2 90 CO to 80 to 70 3 _i 60 < i- 2 50 u. o uj (C UJ 40 20 10 0 UPPER LIMIT OF 90% ^ /CONFIDENCE INTERVAL ESTIMATED PERCENT OF TOTAL MASS EMISSIONS LOWER LIMIT OF 90% CONFIDENCE INTERVAL 10 I02 I03 I04 I05 I06 SCREENING VALUE (ppmvH LOG,0 SCALE) PERCENT OF TOTAL MASS EMISSIONS - PERCENT OF TOTAL EMISSIONS ATTRIBUTABLE TO SOURCES WITH SCREENING VALUES GREATER THAN THE SELECTED VALUE. Figure 3-1. Cumulative distribution of total emissions by screening values - valves in light liquid service.9 UPPER LIMIT OF 9S% CONFIDENCE INTERVAL IMATED PERCENT OF SOURCES LOWER LIMIT OF 9S% CONFIDENCE INTERVAL 10 10*103 I04 10s SCREENING VALUE (ppmv) (LOG|QSCALE) PERCENT OF SOURCES - PERCENT OF SOURCES WITH SCREENING VALUES GREATER THAN THE SELECTED SOURCE. Figure 3-2. Cumulative distribution of sources by screening values - valves in light liquid service.10 3-10 ------- B = Leak Occurrence and Recurrence Correction Factor = correction factor to account for sources which start to leak between inspections (occurrence) and for sources which are found to be leaking, are repaired and start to leak again before the next inspection (recurrence) (Table 3-2, 3-4). C = Non-Instantaneous Repair Correction Factor = correction factor to account for emissions which occur between detection of a leak and subsequent repair; that is, repair is not instantaneous. D Imperfect Repair Correction Factor correction factor to account for the fact that some sources which are repaired are not reduced to zero emission levels. For computational purposes, all sources which are repaired are assumed to be reduced to a 1000 ppmv emission level equivalent to a concentration of 1000 ppmv (Table 3-3). t These correction factors can, in turn, be determined from the following expressions: \ (DB-l- (3) D 1-f 'Where: n = Total number of leaks occurring and recurring over the monitoring m interval. N - Total number of sources at or above the action level (Figure 3-2). t = Average time before repairs are made (with a 15-day repair limit, 7.5 is the average used). f = Average emission factor for sources at the average screening value achieved by repair. F Average emission factor for all sources at or above the action level An example of a control effectiveness calculation is presented in Table 3-5. Support data for this calculation are presented in Tables 3-1, 3-2, 3-3, and 3-4, as well as in Figures 3-1 and 3-2. 3-11 ------- TABLE 3-4. IMPACT OF MONITORING INTERVAL ON CORRECTION FACTOR ACCOUNTING FOR LEAK OCCURRENCE/RECURRENCE (FOR EXAMPLE CALCULATION) Monitoring a Bb interval m 3 months 0.2NC 0.90 a n = Total number of leaks which occur, recur, and remain between monitoring intervals. B = Correction factor accounting for leak occurrence and recurrence. c N = Total number of components at or above the action level. 3-12 ------- TABLE 3-5. EXAMPLE OF CONTROL EFFICIENCY CALCULATION Assume: 1) A leak detection and repair program to reduce emissions from valves in light liquid service. 2) Action level = 10,000 ppmv. 3) Average screening value after directed repair = 1,000 ppmv. 4) Leak detection monitoring interval = 3 months. 5) Allowable repair interval = 15 days. 6) Number of valves having new or recurring leaks between repair intervals, r\m 0.2N (see Table 3-4). Calculations: A = 0.84 (from Figure 3-1 for a screening value of 10,000 ppmv) B = 0.9 (from Table 3-4) C = 0.979 (for 15-day interval) where: F = A(Avg. uncontrolled emission factor)9 . Fraction of sources screening 2l 10,000 ppmv = (0.84)(0.010 kg/hr)/0.11 = 0.076 kg/hr f = Emission factor at 1,000 ppmva = 0.004 kg/hr . 0.947 Overall percentage reduction =AxBxCxD (0.84) x (0.90) x (0.979) x (0.947) = 70 percent Therefore: Emission factor after control = 0.010 kg/hr - (0.70) (0.010 kg/hr) = 0.003 kg/hr Reference 3. bFrom Figure 3-2. 3-13 ------- 3.2 OTHER CONTROL STRATEGIES This section discusses two fugitive emission control strategies for valves in gas service and valves in light liquid service other than the quarterly leak detection and repair procedures discussed above. Consideration of alternative control strategies for valves is pertinent because they account for such a large percentage of the components to be monitored (about 90 percent in the model process units). However, alternative control strategies are not pertinent for other components (pumps, compressors, safety/relief valves) because these other components are relatively few in number. These strategies should be considered alternatives to quarterly leak detection and repair to allow plants the flexibility to meet a level of performance using control procedures considered most appropriate by that plant. Plants which currently have relatively few leaking valves because of good design or existing control procedures would be most likely to benefit from these strategies if they were included in regulations adopted by a State agency. Thus, these alternative control strategies might be included in State regulations as alternative standards to quarterly leak detection and repair. Before implementing one of these alternative control strategies, however, an owner or operator should be required to notify the Director of the State agency. 3.2.1 General The emission reduction and annualized cost of a quarterly leak detection and repair program depend in part on the number of valves found leaking during inspections. Since about 90 percent of the components to be monitored in a process unit are valves, most of the cost of detecting leaks in a process unit can be attributed to valves. In general, few leaks mean VOC emissions are low. Consequently, the amount of VOC emissions that could be reduced through a leak detection and repair program and the product recovery credit associated with the program would be small. As a result, the annualized cost of a leak detection and repair program for a process unit increases as the number of leaks detected and repaired decreases. 3-14 ------- On an individual component basis valves have a lower emission rate than the other components (Table 2-2) and have a percentage leak rate which is lower than most other components (Table 3-2). As the percent of valves found leaking decreases the product recovery credit decreases. The direct cost for monitoring, however, remain the same because the number of valves which must be monitored remains the same. Therefore, the cost effectiveness (annualized cost per megagram of emissions controlled) of a leak detection and repair program varies with the number of valves (or the percent of valves) which leak within a process unit. Table 3-6 presents the cost effectiveness of a quarterly leak detection and repair program for the model process units as a function of the initial percent of valves found leaking. As shown in Table 3-6, the cost effectiveness for a quarterly leak detection and repair program for valves appears reasonable for leak percentages of one percent or higher. A plant averaging one percent of valves leaking will sometimes have less than one percent of valves leaking and sometimes have more than one percent leaking. Statistically, if a plant averaged one percent of valves leaking, then the percent of valves found leaking during a random annual inspection should exceed two percent less than five percent of the time. In other words, if a random annual inspection indicated that no more than two percent of valves are leaking, the probability is greater than ninety-five percent that an average of one percent of valves leaking is actually being achieved in practice. Therefore, two percent of valves found leaking is a reasonable criterion to judge the applicability of alternative control strategies for valves. 3.2.2 Allowable Percentage Of Valves Leaking A State regulation incorporating an alternative control strategy based on an "allowable percentage of valves leaking" would require a plant to limit the number of valves leaking at any time to a certain percentage of the number of valves to be monitored. As discussed above, it appears that two percent of valves leaking represents a reasonable performance level for an allowable percentage of valves leaking. 3-15 ------- TABLE 3-6. COST EFFECTIVENESS VERSUS INITIAL PERCENT OF VALVES LEAKING IN MODEL UNITS3 Initial percent of valves leaking 0.1 0.5 1 2 5 10 20 Cost A 23,500 4,400 2,000 850 100 (150) (270) effectiveness in model B 8,500 1,400 460 35 (220) (300) (350) unit ($/Mg)b C 4,500 630 140 (HO) (300) (350) (370) aFor quarterly leak detection and repair. bAssumes value of VOC of $410/Mg. ( ) = indicates credit 3-16 ------- This type of regulation would require the owner or operator to conduct a performance test at least once a year by the applicable test method. Additional performance tests could be requested by the State. A performance test would consist of monitoring all valves in gas service and valves in light liquid service, and attempting to repair any valves which are leaking. All other components would be subject to quarterly leak detection and repair. The percentage of valves leaking would be determined by dividing the number of valves for which a leak was detected by the number of valves monitored. If the results of a performance test showed that the percentage of valves leaking was greater than the performance level of two percent of valves leaking, then the process unit would be in violation of the State regulation. Incorporating this type of alternative control strategy in the State t regulation would provide the flexibility of a performance standard. Compliance with the regulation could be achieved by the method deemed most appropriate by the plant for each process unit. The plant could implement the quarterly leak detection and repair program for valves to comply with the regulation or it could implement a program of their choosing for valves to comply with the performance level in the regulation. 3.2.3 Alternative Work Practice for Valves A State regulation incorporating an alternative control strategy for valves based on "skip-period" monitoring would require that a plant attain a "good performance level" on a continual basis in terms of the percentage of leaking valves. As discussed above, it appears that two percent of valves leaking represents a "good performance level." This type of regulation would require the owner or operator to begin with implementation of a quarterly leak detection and repair program for valves. If the desired "good performance level" of two percent of valves leaking was attained for valves in gas service and light liquid service for a certain number of consecutive quarters, then one or more of the subsequent quarterly leak detection and repair periods for these valves could be skipped. This strategy is generally referred to as "skip-period" monitoring. All other components would be subject to quarterly leak detection and repair intervals. 3-17 ------- If Implementation of the quarterly leak detection and repair program showed that two percent or less of the valves in gas service and valves in light liquid service were leaking for j_ consecutive quarters, then m quarterly inspections may be skipped. If the next inspection period also showed that the "good performance level" was being achieved, then m quarterly inspections could be skipped again. When an inspection period showed the "good performance level" was not being achieved, then quarterly inspections of valves would be reinstituted. If j_ consecutive quarterly inspections then showed again that the good performance level was being achieved, then m_ quarterly inspections could be skipped again. As mentioned above, two percent of valves leaking represents a good level of performance. Table 3-7 illustrates how "skip-period" monitoring might be implemented in practice. In this case, the "good performance level" must be met for five consecutive quarters (i=5) before three quarters of leak detection could be skipped (m=3). If the quarterly leak detection and repair program showed that two percent or less of the valves in gas service and valves in light liquid service in a process unit were leaking for each of five consecutive quarters, then three quarters could be skipped following the fifth quarter in which the percent of these valves leaking was less than the "good performance level." After three quarters were skipped, all valves would be monitored again on the fourth quarter. This strategy would permit a plant that has consistently demonstrated it is meeting the "good performance level" to monitor valves in gas service and valves in light liquid service annually instead of quarterly. Using this approach, a plant could optimize labor and capital costs to achieve the good level of performance by developing and implementing its own leak detection and repair procedures or installing valves with lower probabilities of leaking. 3.3 OTHER CONSIDERATIONS This section identifies and discusses other-considerations that a State agency may wish to address when drafting a regulation. These considerations include components which are unsafe or difficult to reach, small process units, and unit turnarounds. 3-18 ------- TABLE 3-7. ILLUSTRATION OF SKIP-PERIOD MONITORING0 Leak detection period 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Leak rate of valves during period (%) 3.1 0.8 1.4 1.3 1.9 0.6 - - 3.8 1.7 1.5 0.4 1.0 0.9 - - 0.9 - - 1.9 Quarterly action taken (monitor vs. skip) monitor monitor monitor monitor monitor monitor skip skip skip monitor monitor monitor monitor monitor monitor skip skip skip monitor skip skip skip monitor Good performance level achieved? No Yes Yes Yes Yes Yes - - No Yes Yes Yes Yes Yes - - Yes - - Yes 1 2 3 4 5b 1 2 3 4C 1 2 3 4 5b 1 2 3 4d 1 2 3 4d ai=5, m=3, good performance level of 2 percent. bFifth consecutive quarter below 2 percent means 3 quarters of monitoring may be skipped. Percentage of leaks above 2 percent means quarterly monitoring reinstituted. Percentage of leaks below 2 percent means 3 quarters of monitoring may be skipped. 3-19 ------- 3.3.1 Unsafe and Difficult to Reach Components Some components might be considered unsafe to monitor because process conditions include extreme temperatures or pressures. A State agency may wish to require less frequent monitoring intervals for these components because of the potential danger which may be presented to monitoring personnel. For example, some pumps might be monitored at times when process conditions are such that the pumps are not operating under extreme temperatures or pressures. Some valves may be difficult to reach because access to the valve bonnet is restricted or the valves are located in elevated areas. These valves might be reached by the use of a ladder or scaffolding. Valves which could be reached by the use of a ladder or which would not require monitoring personnel to be elevated higher than two meters might be monitored quarterly. However, valves which require the use of scaffolding or which require the elevation of monitoring personnel higher than two meters above permanent support surfaces might be monitored annually, for example. 3.3.2 Small Process Unit Some process units have so few components to be monitored that the cost effectiveness of a quarterly leak detection and repair program for that process unit would be high. A State agency may wish to consider such process units ''small" and exempt them from compliance with a regulation. The total cost of a leak detection and repair program would consist of the capital cost of VOC detection instruments and the cost of labor for leak detection and repair. The cost of VOC detection instruments would be the same for all sizes of process units, but the cost of labor for leak detection and repair would depend on the number of components to be monitored. As the number of components to be monitored decreases, both the labor cost and the recovery credit associated with VOC emission reduction decrease. This results in a lower total cost. However, since the cost of the VOC detection instruments is fixed, a leak detection and repair program becomes less cost effective as the number of components subject to monitoring decreases. 3-20 ------- Valves in light liquid service and valves in gas service are the greatest percentage (about 90 percent) of the components which would be subject to monitoring in a typical process unit. In addition, the number of valves in gas service and light liquid service can be used as a crude indicator of the total number of components in a process unit which would be subject to monitoring, Table 3-8 shows the emission reduction, net annualized cost, and cost effectiveness for quarterly leak detection and repair in process units with different numbers of valves. The magnitude of emission reduction and cost effectiveness of emission control suggest that implementation of a leak detection and repair program for units which have more than 100 valves in gas service and valves in light liquid service appears reasonable. Thus, States may wish to consider exempting process units with less than 100 valves in gas service and light liquid service from regulations requiring control of fugitive VOC emissions. 3.3.3 Unit Turnarounds A State agency might wish to consider a provision in their regulations which would allow the agency Director to order an early unit shutdown for repair of leaking components in cases where the percentage of leaking components awaiting repair at unit turnaround becomes excessive. 3-21 ------- TABLE 3-8. COST EFFECTIVENESS OF QUARTERLY LEAK DETECTION AND REPAIR FOR TYPICAL PROCESS UNITS ro ro Number of valves in process unit 10 50 100 200 Uncontrolled emissions (Mg/yr)* 1.3 6.3 12.6 25.2 Potential emission reduction (Mg/yr) 1.1 5.3 10.6 21.1 Total annual ized cost ($) 5,590 6,050 6,620 7,740 Net annual ized cost ($)b 5,140 3,880 2,270 (910) Cost effectiveness ($/Mg) 4,670 730 210 (40) Cost effectiveness per valve ($-valve/Mg) 467 15 2.1 (0.2) Based on model unit proportion of valves in gas service and valves in light liquid service and operating 365 days per year. bBased on VOC value of $410/Mg. ( ) = net credit. ------- 3.4 REFERENCES 1. Part UG - General Requirements (Section VIII, Division I.) In: ASME Boiler and Pressure Vessel Code, An American National Standard. New York, The American Society of Mechanical Engineers, 1977. p. 449. 7. 8. Teller, James H. Advantages Found in On-Line Leak Sealing. Journal, 77 (29):54-59, 1979. Oil and Gas Wetherold, R.G., and L.P. Provost. Emission Factors and Frequency of Leak Occurance for Fittings in Refinery Process Units. Interim Report. EPA/600/2-79-044. Radian Corporation. February 1979. p. 2. from F.R. Bottomley, Union Oil Company. Feldstein, Bay Area Quality Management Ref. 3. Valve Repair Summary and Memo Rodeo, California. To Milton District, April 10, 1979. Ref. 3. Ref. 3. Tichenor, B.A., K.C. Hustvedt, and R.C. Weber. Controlling Petroleum Refinery Fugitive Emissions Via Leak Detection and Repairs in Proceedings: Symposium on Atmospheric Emissions from Petroleum Refineries (November 1979, Austin, Texas). EPA-600/9-80-013. Radian Corporation. March 1980. pp. 421-440. 9. Ref. 3 10. Ref. 3. 3-23 ------- 4.0 ENVIRONMENTAL ANALYSIS OF RACT 4.1 INTRODUCTION The environmental impacts resulting from implementation of reasonably available control technology (RACT) are examined in this chapter. Implementing a quarterly leak detection and repair program and capping of open-ended lines with a second valve, cap, plug, or blind flange is considered representative of RACT for control of fugitive VOC emissions from equipment components in the SOCMI and in the polymer and resin manufacturing industry. Leak detection should consist of quarterly monitoring the following components in VOC service with a VOC detection instrument: pumps in light liquid service, valves in light liquid service, valves in gas service, compressors, and safety/relief valves in gas service. Pumps in light liquid service should be visually inspected weekly for indications of leaks. The VOC detection instrument and the monitoring method employed should be EPA Reference Method 21 (Appendix C) or an equivalent State method. A component should be considered in VOC service if it contains ten percent or greater VOC by weight. A VOC is any organic compound which participates in atmospheric photochemical reactions and is measured by the applicable test methods described in EPA Reference Method 21 or equivalent State method. For the purpose of this document, a light liquid is defined as a fluid with a vapor pressure greater than 0.3 kPa at 20°C. A component should be considered in light liquid service if it contacts a fluid containing greater than ten percent by weight light liquid. A component should be considered in gas service if it contains process fluid that is in the gaseous state at operating conditions. Components which have a measureable VOC concentration of 10,000 ppmv or greater should be considered leaking components. Leaking components should be repaired within 15 days of the date the leak is detected. Repair should be considered as reduction of measureable VOC concentration below 10,000 ppmv. Leaking components which cannot be repaired without a unit shutdown should be repaired at the next unit turnaround. 4-1 ------- 4.2 AIR POLLUTION Implementation of RACT would reduce VOC fugitive emissions from process units. A significant beneficial impact on air pollution emissions would result. The hourly and annual emissions from each model unit before and after control by RACT are presented (Tables 4-1 and 4-2). There would be no adverse air pollution impacts associated with RACT. 4.2.1 Development of VOC Emission Levels The uncontrolled emission factors for process unit equipment were previously presented in Chapter 2 (Table 2-2). Emission factors were developed for those sources that would be controlled by the implementation of RACT. These controlled fugitive emission levels were calculated by multiplying the uncontrolled emissions from this equipment by a control efficiency. The control efficiency is determined by several factors which are described and presented in Chapter 3. The controlled VOC emission factors for each source are presented in Table 4-3. In calculating the total fugitive emissions from model units controlled under RACT, the uncontrolled and controlled emission factors were used. These emission factors were multiplied by the equipment source inventories for each model unit. An example calculation for estimating emissions from model unit A under RACT is shown in Table 4-4. 4.2.2 VOC Emission Reduction The emission reduction expected from the implementation of RACT can be determined for each model unit. The emission reduction is the difference between the amount of fugitive emissions before RACT is implemented and the amount of fugitive emissions after RACT is implemented. These amounts are presented in Tables 4-1 and 4-2. The reduction in emissions for the model units after RACT would be implemented is 66 percent. 4-2 ------- TABLE 4-1. ESTIMATED HOURLY EMISSIONS AND EMISSIONS REDUCTION ON A MODEL UNIT BASIS. Estimated emissions (kg/hr) Average percent Level of Mode! unit reduction from control ABC uncontrolled level Uncontrolled 7.7 29.3 91.2 RACT 2.6 10.2 31.6 66 TABLE 4-2. ESTIMATED ANNUAL EMISSIONS AND EMISSIONS REDUCTION ON A MODEL UNIT BASIS. Estimated emissions (Mq/yr) Average percent Level of Model unit ~ reduction from control ABC uncontrolled level Uncontrolled 67 260 800 RACT 23 89 277 66 4-3 ------- TABLE 4-3. EMISSION FACTORS FOR SOURCES CONTROLLED UNDER RACT Uncontrolled emission source Pumps Light liquid service Valves Gas service Light liquid service Safety/relief valves Gas service Compressors Uncontrolled emission Inspection factor, interval kg/hr Quarterly Quarterly Quarterly Quarterly Quarterly 0.120 0.021 0.010 0.160 0.440 Ab 0.87 0.98 0.84 0.69 0.84 Correction factors Bc 0.90 0.90 0.90 0.90 0.90 Cd 0.98 0.98 0.98 0.98 0.98 De 0.92 0.99 0.94 0.97 0.97 Control ef f i ci ency (AxBxCxD) 0.71 0.86 0.70 0.59 0.72 Controlled emission factor, kg/hr 0.035 0.003 0.003 0.066 0.123 aFrom Table 2-2. theoretical maximum control efficiency. Reference 1. cLeak occurrence and recurrence correction factor. Reference 2. Non-instantaneous repair correction factor - for 15-day maximum allowable repair time, the correction factor is [365 - (15/2)] * 365. Reference 2. elmperfect repair correction factor, calculated as 1 - (f f F), where f = average emission rate for sources at 1000 ppmv and F = average emission rate for sources greater than 10,000 ppmv. References 1, 2. Controlled emission factor = uncontrolled emission factor x [1 - (AxBxCxD)]. ------- TABLE 4-4. EXAMPLE CALCULATION OF VOC FUGITIVE EMISSIONS FROM MODEL UNIT A UNDER RACT Number of sources in. model unit0 (N) Emission. factor, kg/hr-source (E) Emissions from sources, kg/hr (N x E) Emission Source: Pumps . Light liquid single mechanical jseal Light liquid double mechanical seal Heavy liquid single mechanical seal Heavy liquid packed seal In-line valves Gas service . Light liquid service Heavy liquid service Safety/relief valves Gas service j Light liquid service Heavy liquid service Open-ended valves Gas service , Light liquid service Heavy liquid service Compressors Sampling connections Flanges 5 3 5 2 t 90 84 84 11 1 1 9 47 48 1 26 600 Total 0.035 0.035 0.020 0.020 0.003 0.003 0.0003 0.066 0.006 0.009 0.003 0.003 0.0003 0.123 0.015 0.0003 emissions 0.175 0.105 0.100 0.040 0.270 0.252 0.025 0.726 0.006 0.009 0.027 0.141 0.014 0.123 0.390 0.180 2.583 aFrom Table 2-1. RACT emission factors include uncontrolled factors from Table 2-2 and controlled factors from Table 4-3. °Sources in VOC service. Light liquid is defined as having a vapor pressure equal to or greater than 0.3 kPa at 20°C. A component is in liquid liquid service if it contains greater than 10 percent by weight light liquid. 6Heavy liquid is defined as having a vapor pressure less than 0.3 kPa at 20°C. Open-ended valve factor is equivalent to the in-line valve factor because of capping the open end. 4-5 ------- 4.3 WATER POLLUTION Implementation of RACT would result in no adverse water pollution impacts because no wastewater is involved in monitoring and leak repair. Some liquid chemicals may already be leaking and entering the wastewater system as runoff. A beneficial impact on wastewater would result from implementation of RACT since liquid leaks are found and repaired. This impact, however, cannot be quantified because no applicable data on liquid leaks are available. 4.4 SOLID WASTE DISPOSAL The quantity of solid waste generated by the implementation of RACT would be insignificant. The solid waste generated would consist of used valve packings and components which are replaced. 4.5 ENERGY The implementation of RACT calls for an emission control technique that requires no additional energy consumption for any of the model unit sizes. A beneficial impact would be experienced by saving VOC which has been heated, compressed, or pumped. 4-6 ------- 4.6 REFERENCES 1. Wetherold, R. and L. Provost, Emission Factors and Frequency of Leak Occurrence for Fittings in Refinery Process Units. EPA-600/2-79-044, February 1979. 2. Tichenor, B.A., K.C. Hustvedt, and R.C. Weber. Controlling Petroleum Refinery Fugitive Emissions Via Leak Detection and Repair, Draft. Symposium on Atmospheric Emissions from Petroleum Refineries, Austin, Texas. 4-7 ------- 5.0 CONTROL COST ANALYSIS OF RACT The costs of implementing reasonably available control technology (RACT) for controlling fugitive emissions of volatile organic compounds (VOC) from process units are presented in this chapter. Capital costs, annualized costs, and the cost effectiveness of RACT are presented. These costs have been developed for the model units presented in Chapter 2. All costs presented in this chapter have been updated to second quarter 1980 dollars. 5.1 BASIS FOR CAPITAL COSTS Capital costs represent the total cost of starting a leak detection and repair program in existing process units. The capital costs for the implemen- tation of RACT include the purchase of VOC monitoring instruments, the purchase and installation of caps for all open-ended lines, and initial leak repair. The cost for initial leak repair is included as a capital cost because it is expected to be greater than leak repair costs in subsequent quarters and is a one-time cost. The basis for these costs is discussed below and presented in Table 5-1. Capital cost estimates for model units under RACT are presented in Table 5-2. Labor costs were computed using a charge of $18 per labor-hour. This rate includes wages plus 40 percent for related administrative and overhead costs . 5.1.1 Cost of Monitoring Instrument i The cost of a VOC monitoring instrument includes the cost of two instruments. One instrument is intended to be used as a standby spare. The cost of $4600 for a portable organic vapor analyzer was obtained from a 2 manufacturer. 5-1 ------- TABLE 5-1. CAPITAL COST DATA Cost value used in analysis Item (June 1980 dollars) Cost basis Reference Monitoring instrument 2 x 4600 9200/model unit One instrument used as a spare 2 Caps for open-ended lines 53/1ine Based on cost for 1" screw-on 3 type valve. Cost June 1980 = $35. Installation = 1 hour at $18/hour. TABLE 5-2. CAPITAL COST ESTIMATES FOR IMPLEMENTING RACT (thousands of June 1980 dollars) Capital cost item Model Unit A 1. Monitoring instruments . 2. Caps for open-ended lines (104 caps) 3. Initial leak detection and repair cost Total Model Unit B 1 . Moni tori ng i nstruments . 2. Caps for open-ended lines (415 caps) 3. Initial leak detection and repair cost Total Model Unit C 1. Monitoring instruments . 2. Caps for open-ended lines (1277 caps) 3. Initial leak detection and repair cost Total Level Uncontrolled 0.0 0.0 0.0 0.0 0.0 0.0 0^0 0.0 0.0 0.0 0.0 0.0 of control RACT 9.2 5.5 4.48 19.18 9.2 22.0 14.35 45.55 9.2 67.7 40.34 117.24 aBased on capital cost data presented in Table 5-1. bFrom Table 2-1. ""Initial leak detection and repair are treated as capital costs because they are one-time cost. 5-2 ------- 5.1.2 Caps for Open-Ended Lines Fugitive emissions from open-ended lines and valves can be controlled by installing a cap, plug, flange, or second valve to the open end. These pieces of equipment are all included in the definition of a cap for an open-ended line. The cost of a cap for an open-ended line is based on a cost of $35 for a one-inch screw-on type globe valve. This cost was supplied 3 by a large distributor. A charge of $18 for one hour of labor is added to $35 as the cost for installing one cap. Therefore, the total capital cost for installing a cap on an open-ended line is $53. 5.1.3 Initial Leak Repair The implementation of RACT will begin with an initial inspection which will result in the discovery of leaking components. The number of initial leaks is expected to be greater than the number found in subsequent inspec- tions. Because initial leak repair is a one-time cost, it is treated as a capital cost. The number of initial leaks was estimated by multiplying the percentage of initial leaks per component type by the number of components in the model unit under consideration. Fractions were rounded up to the next highest integer. The repair time for fixing leaks is estimated to be 80 hours for a pump seal, 40 hours for a compressor seal, and 1.13 hours for a valve. The repair time for fixing pump seals and compressor seals includes the cost of a new seal. These requirements are presented in Table 5-3. The initial repair cost was determined by taking the product of the number of initial leaks, the repair time, and the hourly labor cost of $18. 5.2 BASIS FOR ANNUALIZED COSTS Annualized costs represent the yearly cost of operating a leak detection and repair program and the cost of recovering the initial capital investment. This includes credits for product saved as the result of the control program. The basis for the annualized costs is presented in Table 5-4. 5-3 ------- TABLE 5-3. LABOR-HOUR REQUIREMENTS FOR INITIAL LEAK REPAIR UNDER RACT Source type Pumps (light liquid) Single mechanical seal Double mechanical seal Number of components per model unit ABC 5 19 60 3 10 31 Estimated number of initial leaks3 ABC 2 5 14 1 3 8 Repair time, hrs 80b 80b Labor-hours requi red ABC 160 400 1120 80 240 640 Valves (in-line) Gas Light liquid Safety/relief valvesd (gas service) 90 365 1117 84 335 1037 9 37 112 11 41 125 1.13 10 42 127 1.13C 12 46 141 11 42 130 Valves on open-ended lines Gas Light liquid Compressor seals 9 47 1 37 189 2 115 581 8 le 6e 1 4e 23e 1 12e 70e 3 1.13C 1.13C 40b 1 7 40 5 26 40 14 79 120 TOTAL 310 799 2241 aBased on the percent of sources leaking at _> 10,000 pom. From Table 3-2. blncludes labor-hour equivalent cost of new seal. Reference 6. Steighted average based on 75 percent of the leaks repaired on-line, requiring 0.17 hours per repair, and on 25 percent of the leaks repaired off-line, requiring 4 hours per repair. Ref. 5, p. B-12. dlt is assumed that-these leaks are corrected by routine maintenance at no additional labor requirements. Ref. 6. ^The estimated number of initial leaks for open-ended valves is based on the same percentage of sources used for in-line valves. This represents leaks occurring through the stem and gland of the open-ended valve. Leaks through the valve seat are eliminated by adding caps. 5-4 ------- TABLE 5-4. BASIS FOR ANNUALIZED COST ESTIMATES 1. Capital recovery factor for capital charges • Caps on open-ended lines • Monitoring instruments 2. Annual maintenance charges • Caps on open-ended lines • Monitoring instruments 3. Annual miscellaneous charges (taxes, insurance, administration) • Caps on open-ended lines • Monitoring instruments 4. Labor charges 5. Administrative and support costs for implementing PACT 6. Annualized charge for initial leak repairs 7. Recovery credits 0.163 x capital0 0.23 x capital13 0.05 x,capital0 $3,000° 0.04 x capital p 0.04 x capital $18/hourf 0.40 x (monitoring + repair labor)9 E (estimated number of leaking components per model unit x repair time) x $18/hrf x 1.49 x 0.163n $410/Mg aTen year life, ten percent interest. From Ref. 5, pp. IV-3,4. bSix year life, ten percent interest. From Ref. 5, pp. IV-9,10. cFrom Ref. 5, pp. IV-3,4. Includes materials and labor for maintenance and calibration. Reference 6. Cost index = 242.7 * 209.1 (Reference 7 and 8). eFrom Ref. 5, pp. IV-3,4,9,10. Includes wages plus 40 percent for labor-related administrative and overhead costs. Cost (June 1980) from Ref. 1. gFrom Ref. 5, pp. IV-9,10. Initial leak repair amortized for ten years at ten percent interest. •^References 9,10,11. 5-5 ------- 5.2.1 Monitoring Labor The implementation of RACT requires visual and instrument monitoring of potential sources of fugitive VOC emissions. The monitoring labor-hour requirements for RACT are presented in Table 5-5. The labor-hour require- ments were calculated by taking the product of the number of workers needed to monitor a component (1 for visual, 2 for instrument), the time required to monitor, the number of components in a model unit, and the number of times the component is monitored each year. The monitoring times for the various components are presented in Table 5-5. They are 0.5 minute for visual inspection, 1 minute for in-line valves and open-ended valves, 5 minutes for pump seals, 8 minutes for safety valves, and 10 minutes for 4 compressor seals. Monitoring labor costs were calculated based on a charge of $18 per hour. 5.2.2 Leak Repair Labor Labor is needed to repair leaks which develop after initial repair. The estimated number of leaks and the labor-hours required for repair are given in Table 5-5. The repair time for each component is the same as presented for initial leak repair. Leak repair costs were calculated based on a charge of $18 per hour. 5.2.3 Maintenance Charges and Miscellaneous Costs The annual maintenance charge for caps is estimated to be five percent of their capital cost. The annual cost of materials and labor for maintenance 678 and calibration of monitoring instruments is estimated to be $3000. ' ' An additional miscellaneous charge of four percent of capital cost for taxes, insurance, and associated administrative costs is added for the monitoring instruments and caps. 5.2.4 Administrative Costs Administrative and support costs associated with the implementation of RACT are estimated to be 40 percent of the sum of monitoring and leak repair labor costs. The administrative and support costs include record- keeping and reporting requirement costs. 5-6 ------- TABLE 5-5. ANNUAL MONITORING AND LEAK REPAIR LABOR REQUIREMENTS FOR RACT 01 Monitoring Number of components per model unit Source type Pumps (light liquid) Single mechanical seals Double mechanical seals Valves (In-line) Gas Light liquid Safety/relief valves (gas service) Valves on open-ended lines" Gas Light liquid Compressors seals TOTAL A 5 3 90 84 11 9 47 1 B C 19 60 10 31 365 1117 335 1037 42 130 37 115 189 5B1 2 8 Type of* monitoring Instrument Visual Instrument Visual Instrument Instrument Instrument Instrument Instrument Instrument Monitoring tlme.b m1n 5 0.5 5 0.5 1 1 8 1 1 10 Times monl tored per year 4 52 4 52 4 4 4 4 4 4 Estimated Monitoring labor- number of d hours requ1redc leaks per year A 3.3 2.2 2.0 1.3 12.0 11.2 11.7 1.2 6.4 1.3 52.6 B 12.8 8.2 6.8 4.3 49.0 44.8 44. » 4.9 25.2 2.7 203.5 C A B C 40.0 1 2 5 26.0 20.8 1 1 3 13.4 149.0 4 15 45 138.4 4 16 50 139.0 15.3 1 2 5 77.6 2 9 28 10.7 1 1 2 630.2 Leak repair Repair Leak repair labor- time, hours required6 hrs A 80b 80 80b 80 1.13f 4.5 1.13f 4.5 O9 0 1.13e 1.1 1.13e 2.3 40b 40 212.4 B 160 80 17.0 18.0 0 2.3 10.2 40 327.5 C 400 240 50.9 56.5 0 5.7 31.6 80 864.7 'Two workers for Instrument monitoring, one for visual. Reference 4. Ttonltorlng labor-hours «• number of workers x number of components x time to monitor (total 1s minimum of 1 hour) x number of times monitored per year. dFrom Table 3-2. eLeak repair labor-hours • number of leaks x repair time. Weighted average based on 75 percent of the leaks repaired on-line, requiring 0.17 .hour per repair, and on 25 percent of the leaks, repaired off-line, requiring 4 hours per repair. Ref. 5, p. B-12. It Is assumed that these leaks are corrected by routine maintenance at no additional labor requirements. Ref. 4 estimated number of leaks per year for open-ended valves 1s based on the same percent of sources used for In-Hne valves. This represents leaks occurring through the stem and gland of the open-ended valve. Leaks through the seat of the valve are eliminated by adding caps. ------- 5.2.5 Capital Charges The life of caps for open-ended lines is assumed to be ten years and the life of monitoring instruments is assumed to be six years. The cost of repairing initial leaks was amortized over a ten-year period since it is a one-time cost. The capital recovery is obtained from annualizing the installed capital cost for control equipment. The installed capital cost is annualized by using a capital recovery factor (CRF). The CRF is a function of the interest rate and useful equipment lifetime. The capital recovery can be estimated by multiplying the CRF by the total installed capital cost for the control equipment. This equation for the capital recovery factor is: 1(1 + i)n CRF= (l+i)n-l where i = interest rate, expressed as a decimal n = economic life of the equipment, years. The interest rate used was ten percent (June 1980). The capital recovery factors and other factors used to derive annualized charges are presented in Table 5-4. 5.2.6 Recovery Credits The reduction of VOC fugitive emissions results in saving a certain amount of VOC which would otherwise be lost. The value of this VOC is a recovery credit which can be counted against the cost of a leak detection and repair program. The recovery credits for each model unit are presented in Table 5-6. The VOC saved is valued in June 1980 dollars at $410/Mg.9'10'n 5.3 EMISSION CONTROL COSTS This section will present and discuss the emission control costs of implementing RACT for each of the three model units. Both the initial costs and the annualized costs are included. 5-8 ------- TABLE 5-6. RECOVERY CREDITS Model unit A B C Uncontrolled emissions, Mg/yr 67 260 800 Emissions under RACT, Mg/yr 23 89 277 Emission reduction, Mg/yr 44 171 523 Recovered product value, $/yr 18,040 70,100 214,400 Based on an average price of $410/Mg. References 9,10,11. 5-9 ------- 5.3.1 Initial Costs The cost of initially implementing RACT consists of capital costs and initial leak repair. The 'capital cost of $9200 for two monitoring instruments is the same for all model unit sizes. Caps for open-ended lines will cost $5500 for model unit A, $22,000 for model unit B, and $67,700 for model unit C. The one-time initial leak repair cost is $7812 for model unit A, $20,135 for model unit B, and $56,470 for model unit C. The total initial capital costs for implementing RACT are $19,180 for model unit A, $45,550 for model unit B, and $117,240 for model unit C. 5.3.2. Recovery Credits The value of VOC saved each year as a result of implementing RACT is included as an annual credit against the net annualized costs. The implemen- tation of RACT will result in saving $18,260 worth of VOC annually in model unit A, $69,540 worth of VOC in model unit B, and $211,100 worth of VOC in model unit.C. 5.3.3 Net Annualized Cost The net annual cost for controlling emissions is the difference between the total annualized cost and the annual recovery credit for each model unit. Net annualized control cost estimates for model units under RACT are presented in Table 5-7. Capital cost data were previously presented in Table 5-1. For model unit A, the annualized capital charges are $4280 and the total annual operating costs are $10,550. Product recovery credits total $18,260. The net annualized cost for model unit A is a negative $3436, which means that $3436 is actually gained every year by preventing loss of VOC. The annualized capital charges for model unit B are $8990 and the total annual operating costs are $18,730. The recovery credit is $69,540 per year. The net annualized cost for model unit B is a negative $41,823, which means that $41,823 is saved every year by controlling VOC emissions. 5-10 ------- TABLE 5-7. ANNUALIZED CONTROL COST ESTIMATES FOR MODEL UNITS UNDER RACT (thousands of June 1980 dollars) Cost item A Model unit B C Annualized capital charges 1. Control equipment a. Instrument 2.12 2.12 2.12 b. Caps 0.89 3.59 11.04 2. Initial leak repair 1.27 3.28 9.2 Subtotal 4.28 8.99 22.36 Operating costs 1. Maintenance charges a. Instrument b. Caps 2. Miscellaneous (taxes, insurance, administration) a. Instrument b. Caps 3. Labor a. Monitoring labor . b. Leak repair labor c. Plant and payroll overhead Subtotal Total before credit Recovery credits Net annual ized cost 3.0 0.275 0.37 0.22 0.95 3.82 1.91 10.545 14.825 18.26 (3.44) 3.0 1.1 0.37 0.88 3.66 5.9 3.82 18.73 27.72 69.54 (41.82) 3.0 3.39 0.37 2.71 11.34 15.56 10.76 47.13 69.49 216.73 (147.24) aSum of labor hours for monitoring in Table 5-5 multiplied by $18/hour. bSum of labor hours for leak repairs in Table 5-5 multiplied by $18/hour. cBased on 40 percent of monitoring labor plus leak repair labor costs. dThese costs are credits. (XXX) = net credit. 5-11 ------- Model unit C has annualized capital charges of $22,360 and total operating expenses of $47,130. The recovery credit is $216,730 per year. The net annualized cost for model unit C is a negative $147,240, which is an annual savings as a consequence of controlling fugitive VOC emissions. 5.3.4 Differences in Net Annualized Costs The cost for RACT is different for each model unit. The cost for caps for open-ended lines varies because the number of open-ended lines is different for each model unit. Because the larger model units have more components, more labor-hours are needed for monitoring and leak repair. For this reason, labor costs will increase as model unit size increases. 5.4. COST EFFECTIVENESS Cost effectiveness is the annualized cost per megagram of VOC controlled annually. The cost effectiveness of RACT for each model unit is the net annualized cost for implementing RACT divided by the emission reduction gained under RACT. The cost effectiveness of RACT is summarized in Table 5-8. The implementation of RACT on model unit A results in a net annualized cost which is a credit of $3436. The emission reduction associated with RACT is 44.5 Mg/yr. The cost effectiveness is -$77/Mg. The implementation of RACT in the case of model unit B results in a net annualized cost which is a credit of $41,820. The emission reduction associated with RACT is 169.6 Mg/yr. The cost effectless is -$247/Mg. The implementation of RACT in the case of model unit C results in a net annualized cost which is a credit of $147,240. The emission reduction associated with RACT is 528.6 Mg/yr. The cost effectiveness is -$279/Mg. A comparison of the cost effectiveness of RACT for each model unit reveals that cost effectiveness increases as model unit size increases. The strong influence of recovery credits is responsible for the increase in cost effectiveness. 5-12 ------- TABLE 5-8. COST EFFECTIVENESS FOR MODEL UNITS UNDER RACT Annual ized cost before credit ($1000) Annual recovery credit ($1000) Net annual ized cost ($1000) Total VOC reduction (Mg/yr) Cost effectiveness ($/Mg VOC) A 14.82 18.26 (3.44) 44.54 (77.0) Model unita B 27.72 69.54 (41.82) 169.6 (247) C 69.49 216.73 (147.24) 528.6 (279) '(XXX) net credit. The cost effectiveness of RACT for each component type is presented in Tables 5-9, 5-10, and 5-11 for model units A, B and C, respectively. The cost of the monitoring instrument cannot be attributed to any single type of component since all the components are monitored by the instrument. Therefore, the cost for each component does not include the cost of the monitoring instrument. The cost effectiveness for RACT for pumps and compressors is higher than other components due to the additional time required for leak repair. 5-13 ------- TABLE 5-9. COST EFFECTIVENESS FOR COMPONENT TYPES IN MODEL UNIT A on Component Pumps (light liquid) Valves Gas service Light liquid service Safety/relief valves Open-ended valves Gas service Light liquid service Heavy liquid service Compressors TOTAL UNIT (without instrument cost) TOTAL UNIT (with instrument cost) Number of components 8 90 84 11 9 47 48 1 250C 250C Annuali zed cost before credit ($)a 5,239 456 442 295 183 877 643 1,204 9,339 14,825 Annual recovery credit ($) 2,444 5,822 2,111 3,715 709 1,857 463 1,140 18,261 18,261 Net annual ized cost ($)a 2,795 (5,366) (1,669) (3,420) (526) (880) 180 64 (8,822) (3,436) Total VOC reduction (Mg/yr) 5.96 14.20 5.15 9.06 1.73 4.53 1.13 2.78 44.54 44.54 Cost effectiveness ($/Mg) 469 (378) (324) (377) (304) (194) 159 23 (198) (77) Does not include cost of monitoring instrument, unless otherwise noted. 3Cost for caps on lines only. Not monitored under RACT. "Total does not include open-ended lines in heavy liquid service. (XXX) = net credit. ------- TABLE 5-10. COST EFFECTIVENESS FOR COMPONENT TYPES IN MODEL UNIT B en i—• en Component Pumps (light liquid) Valves Gas service Light liquid service Safety/relief valves Open-ended valves Gas service Light liquid service Heavy liquid service Compressors TOTAL UNIT (without instrument cost) TOTAL UNIT (with instrument cost) Number of components 29 365 335 42 37 189 189 2 999C 999C Annualized cost before credit ($)a 9,486 1,836 1,771 1,128 698 3,532 2,535 1,240 22,226 27,720 Annual recovery credit ($) 8,852 23,595 8,421 14,178 2,923 7,466 1,833 2,275 69,543 69,543 Net annual i zed cost ($)a 634 (21,759) (6,650) (13,050) (2,225) (3,934) 702 (1,035) (49,989) (41,823) Total VOC reduction (Mg/yr) 21.59 57.55 20.54 34.58 7.13 18.21 4.47 5.55 169.62 169.62 Cost effectiveness ($/Mg) 29 (378) (324) (377) (312) (216) 157 (186) (295) (247) Does not include cost of monitoring instrument, unless otherwise noted. Cost for caps on lines only. Not monitored under RACT. Total does not include open-ended lines in heavy liquid service. (XXX) = net credit. ------- TABLE 5-11. COST EFFECTIVENESS FOR COMPONENT TYPES IN MODEL UNIT C en i-« cn Component Pumps (light liquid) Valves Gas service Light liquid service Safety/relief valves Open-ended valves Gas service ' Light liquid service Heavy liquid service Compressors TOTAL UNIT (without instrument cost) TOTAL UNIT (with instrument cost) Number of components 91 1,117 1,037 130 115 581 581 8 3,079C 3,079C Annualized cost before credit ($)a 25,882 5,560 5,491 3,503 2,130 10,868 7,790 2,779 64,001 69,490 Annual recovery credit ($) 27,782 72,213 26,072 43,890 9,086 22,952 5,633 9,106 216,734 216,734 Net annual i zed cost ($)a (1,900) (66,653) (20,581) (40,387) (6,956) (9,010) 2,157 (6,327) (149,657) (147,240) Total VOC reduction (Mg/yr) 67.76 176.13 63.59 107.05 22.16 55.98 13.74 22.21 528.62 528.62 Cost effectiveness ($/Mg) (28) (378) (324) (377) (314) (161) 157 (285) (283) (279) Does not include cost of monitoring instrument, unless otherwise noted. Cost for caps on lines only. Not monitored under RACT. otal does not include open-ended lines in heavy liquid service. (XXX) = net credit. ------- 5.5. REFERENCES 1. Letter with attachments from Texas Chemical Council to Walt Barber, U.S. EPA. June 30, 1980. 2. Purchase order from GCA/Technology Division to Analabs/Foxboro, North Haven, Connecticut. July 3, 1980. 3. Telecon. Samuel Duletsky, GCA Corporation with Dave Myer, Piedmont Hub, Greensboro, N.C. September 25, 1980. Price of 1" screw-on type valve. 4. Letter with attachments from J.M. Johnson, Exxon Company, U.S.A., to Robert T. Walsh, U.S. EPA. July 28, 1977. 5. Erikson, D.G., and V. Kalcevic. Emission Control Options for the Synthetic Organic Chemicals Manufacturing Industry, Fugitive Emissions Report, Draft Final. Hydroscience, Inc. 1979. p. IV-9. * 6. Environmental Protection Agency. Control of Volatile Organic Compounds Leaks from Petroleum Refinery Equipment. EPA-450/2-78-036, OAQPS No. 1.2-111. June 1978. 7. Economic Indicators. Chem. Eng. Vol. 86 #2. January 15, 1979. 8. Economic Indicators. Chem. Eng. Vol. 87 #19. September 22, 1980. 9. Letter from Vincent Smith, Research Triangle Institute, to Russell Honerkamp, Radian Corporation. November 30, 1979. 10. Reference 8. 11. Economic Indicators. Chem. Eng. Vol. 86 # 1. January 14, 1980. 5-17 ------- APPENDIX A. EMISSION SOURCE TEST DATA ------- APPENDIX A EMISSION SOURCE TEST DATA The purpose of Appendix A is to describe testing results used in the development of the Control Techniques Guideline (CTG) document for VOC fugitive emissions from the Synthetic Organic Chemicals Manufacturing Industry (SOCMI) and the polymer and resin manufacturing industry. The information in this appendix consists of a description of the tested facilities, and the sampling procedures and test results of VOC fugitive emissions studies in SOCMI and the petroleum refining industry. Fugitive emission sources of VOC in SOCMI and in the petroleum refining industry are similar. Considerable data exist concerning both the incidence and magnitude of fugitive emissions from petroleum refineries. Studies of fugitive emissions in SOCMI have been undertaken by EPA to support the use of emission factors generated during studies of emissions in petroleum refineries for similar sources in the Synthetic Organic Chemicals Manufacturing Industry. The results of the EPA SOCMI studies, EPA data from a study of fugitive emissions from petroleum refineries, and some industry studies of fugitive emissions are discussed in Section A.I. Section A.2 consists of the results of three studies on the effects of maintenance on reducing fugitive VOC emissions from valves in petroleum refineries and two studies on maintenance of valves in SOCMI process units. These results are included as an indication of the reduction in emissions which could be expected as a function of the designated action level, and by applying routine on-line maintenance procedures. A.I FUGITIVE EMISSIONS TEST PROGRAMS Three SOCMI test programs have been conducted by EPA. One was a study performed by Monsanto Research Corporation of a small number of fugitive emission sources in four SOCMI units. More intensive screening was performed at six SOCMI units in another study. The third EPA study of SOCMI fugitive A-l ------- emissions was a screening and sampling program conducted at twenty-four SOCMI units. The results of these studies are presented in this section. Similar types of studies have been performed by industry. This section also contains the results of an Exxon study of fugitive emissions in cyclohexane unit and a DuPont study of fugitive emissions in unidentified process units. The results of a study on fugitive emissions from petroleum refineries are also presented in this section. Data on fugitive emissions were obtained from 64 units in thirteen refineries located in major refining areas throughout the country. Data on the effects of maintenance were obtained at the last four of these refineries. These results are presented later in Section A.2 of this Appendix. A.1.1 Study of Fugitive Emissions at Four SOCMI Units Monsanto Research Corporation (MRC) conducted EPA Industrial Environmental Research Laboratory (EPA-IERL) sponsored study of fugitive emissions at four SOCMI units. The process units were monochlorobenzene, butadiene, ethylene oxide/glycol, and dimethyl terephthalate. Due to the small number of plants/processes sampled and the experimental design of this study, the results were not considered to be comparable with the results of other studies. Since the data generated by the MRC study could not be considered representative of the SOCMI and valid conclusions could not be drawn concerning the relative magnitude of fugitive emissions in the SOCMI, the results of the study were not used in the development of standards for fugitive emissions control. This study demonstrated the need for more intensive sampling and screening which was undertaken by EPA. A.1.2 Description and Results of EPA Study of Six SOCMI Units2'3'4'5 The objective of this test program was to gather data on the percentage of sources which leak (as defined by a VOC concentration at the leak interface of _> 10,000 ppmv calibrated with methane). To achieve this objective, an attempt was made to screen all potential leak sources (generally excluding flanges) on an individual component basis with a portable organic vapor analyzer. The test crews relied on plant personnel to identify equipment handling organics. Normally, all pumps and compressor seals were examined, and the percentage of valves carrying VOC which were screened ranged from 33 A-2 ------- to 85 percent. All tests were performed with a Century Systems Corporation Organic Vapor Analyzer (OVA), Model 108, with the probe placed as close to the source as possible. The results of this study are shown in Table A-l. Six chemical process units were screened. Unit A is a chlorinated methanes production facility in the Gulf Coast area which uses methanol as feedstock material. The individual component testing was conducted during September 1978. Unit B is a relatively small ethylene production facility on the West Coast which uses an ethane/propane feedstock. Testing was conducted during October 1978. Unit C is a chlorinated methanes production facility in the Midwest. This plant also uses methanol as the basic organic feedstock. OveV the last few years, several pieces of equipment have been replaced with equipment the company feels is more reliable. In particular, the company has installed certain types of valves which they have found do not leak "as much" as other valves. The individual component testing was conducted during January 1979. Unit D is an ethylene production facility on the Gulf Coast, using an ethane/propane feed. The facility is associated with a major refinery, and testing was conducted during March 1979. Units E and F are part of an intermediate size integrated petroleum refinery located in the North Central United States. Testing was conducted during November 1978. Unit E is an aromatics extraction unit that produces benzene, toluene, and xylene by extraction from refined petroleum feedstocks. Unit E is a new unit and special attention was paid during the design and startup to minimize equipment leaks. All valves were repacked before startup (adding 2 to 3 times the original packing) and all pumps in benzene service had double mechanical seals with a barrier fluid. Unit F produces benzene by hydro- dealkylation of toluene. Unit F was originally designed to produce a different chemical and was redesigned to produce benzene. A.1.3 Description and Results of an EPA Study of 24 SOCMI Units The U.S. EPA Industrial Environmental Research Laboratory coordinated a study to develop information about fugitive emissions in the SOCMI. A total of 24 chemical process units were selected for this purpose. The process units were selected to represent a cross section of the population of the SOCMI. Factors considered during process unit selections included annual production volume, number of producers, volatility, toxicity, and value of the final products. Table A-2 shows the process unit types selected for screening. A-3 ------- TABLE A-l. FREQUENCY OF LEAKS FROM FUGITIVE EMISSION SOURCES IN SYNTHETIC ORGANIC CHEMICAL UNITS (Six Unit Study) Unit A* Chloromethanes Equipment type Valves Open-ended lines Pimp seals Compressor seals Control valves Pressure relief valves Flanges Drains Number of sources tested 600 52 47 _e 52 7 30 _e Percent with screening values MO, 000 ppmv 1 2 15 6 0 3 Unit Ba Ethyl ene Number Percent with of screening sources values tested 2,301 386 51 42 126 e _e _e MO, 000 ppmv 19 11 21 59 20 Unit Cb Chloromethanes Number of sources tested 65B _e 39 3 25 _e _e _e Percent with screening values MO ,000 ppmv 0.1 3 33 0 Unit Dc Ethyl ene Number of sources tested 862 90 63 17 25 _e _e 39 Percent with screen! ng values MO, 000 ppmv 14 13 33 6 44 10 Unit Ed BTX Recovery Number Percent with of screening sources values tested MO ,000 ppmv 715 1.1 33 0.0 33f 3.0 _e S3 4.0 _e _e _e Unit Fd Toluene H)A hunber of sources tested 427 28 30 _e 44 e _e e Percent with screening values MO ,000 ppmv 7.0 11.0 10.0 11.0 'Source: Reference 6. Source: Reference 7. cSource: Reference 8. Source: Reference 9. *No data. Pump seals In benzene service have double mechanical seals. ------- TABLE A-2. TWENTY-FOUR CHEMICAL PROCESS UNITS SCREENED FOR FUGITIVE EMISSIONS Unit Type 1. Vinyl Acetate 2. Ethylene 3. Vinyl Acetate 4. Ethylene 5. Cumene 6. Cumene 7. Ethylene 8. Acetone/Phenol 9. Ethylene Dichloride 10. Vinyl Chloride Monomer 11. Formaldehyde 12. Ethylene Dichloride 13. Vinyl Chloride Monomer 14. Methyl Ethyl Ketone 15. Methyl Ethyl Ketone 16. Acetaldehyde 17. Methyl Methacrylate 18. Adi pic Acid 19. Tri chloroethylene/Perchloroethylene 20. 1,1,1-Trichloroethane 21. Ethylene Dichloride 22. Adi pic Acid 23. Acrylonitrile 24. Acrylonitrile Source: Reference 11 A-5 ------- The screening work began with the definition of the process unit boundaries. All feed streams, reaction/separation facilities, and product and by-product delivery lines were identified on process flow diagrams and in the process unit. Process data, including stream composition, line temperature, and line pressure, were obtained for all flow streams. Each process stream to be screened was identified and process data were obtained with the assistance of plant personnel, in most cases. Sources were screened by a two-person team (one person handling the hydrocarbon detector and one person recording data). The Century Systems Models OVA-108 and OVA-128 hydrocarbon detectors were used for screening. The HNU Systems, Inc., Model PI 101 Photoionization Analyzer was also used to screen sources at the formaldehyde process unit. The detector probe of the instrument was placed directly on those areas of the sources where leakage would typically occur. For example, gate valves were screened along the circumference of the annular area around the valve stem where the stem exits the packing gland and at the packing gland/valve bonnet interface. All process valves, pump seals, compressor seals, agitator seals, relief valves, process drains, and open-ended lines were screened. From five to twenty percent of all flanges were randomly selected and screened. For the purpose of this program "flange" referred to any pipe-to-pipe or tubing-to-tubing connection, excluding welded joints. Each screening instrument was calibrated at least daily. The model OVA-108 instruments, with a logarithmic scale reading from 1 ppmv to 10,000 ppmv, were calibrated with high (8,000 ppmv) and low (500 ppmv) concentration methane-in-air standards to ensure accurate operation at both ends of the instrument's range. The model OVA-128 instruments, with a linear readout ranging from 0 ppmv to 1,000 ppmv, were also calibrated with high and low concentration standards. A pre-calibrated dilution probe was required with the OVA-128 when calibrating with the 8,000 ppmv standard. The HNU Photoionization instrument, used to screen the formaldehyde process unit, was calibrated with isobutylene which has an ionization potential close to that of formaldehyde. A-6 ------- Results of the screening program at the 24 process units are summarized in Table A-3. The fugitive emission sources in the study were screened at an average rate of 1.7 minutes per source for a two-person team (or 3.4 person-minutes per source). This average screening rate includes time spent for instrument calibration and repair. Table A-4 presents screening time data on a unit- by-unit basis. These time requirements are somewhat higher than would be expected for routine monitoring because of the extensive record keeping associated with the screening project. A.1.4 Description and Results of Refinery Fugitive Emissions Study12 Data concerning the leak frequencies and emission factors for various fugitive sources were obtained primarily at nine refineries. More complete information for compressors' and relief valves' emissions was obtained by sampling at four additional refineries. 'Refineries were selected to provide a range of sizes and ages and all of the major petroleum refinery processing units were studied. The type of process units and the number of each studied in the first nine refineries are listed in Table A-5. In each refinery, sources in six to nine process units were selected for study. The approximate number of sources selected for study and testing in each refinery is listed below: Valves 250-300 Flanges 100-750 Pump seals 100-125 Compressor seals 10-20 Drains 20-40 Relief Valves 20-40 There were normally 500-600 sources selected in each refinery. The distribution of sources among the process units was determined before the selection and testing of individual sources was begun. Individual sources were selected from piping and instrumentation diagrams or process flow diagrams before a refinery processing area was entered. Only those preselected sources were screened. In this way, bias based on observation of individual sources was theoretically eliminated. A-7 ------- TABLE A-3. SUMMARY OF SOCMI PROCESS UNITS FUGITIVE EMISSIONS (Twenty-four unit study) > oo Source Type Flanges Process Drains Open Ended Lines Agitator Seals Relief Valves Valves Pumps Compressors Other8 Service Gas Light Liquid Heavy Liquid Gas Light Liquid Heavy Liquid Gas Light Liquid Heavy Liquid Gas Light Liquid Heavy Liquid Gas Light Liquid Heavy Liquid Gas Light Liquid Heavy Liquid Light Liquid Heavy Liquid Gas Gas Light Liquid Heavy Liquid (1) Number Screened 1.443 2,897 607 83 527 28 923 3.603 477 7 8 1 85 69 3 9.668 18.294 3.632 647 97 29 19 33 2 (2) X Not Screened 4.6 2.6 2.4 23.1 1.9 0.0 17.5 10.4 21.5 46.1 11.1 66.7 72.7 40.5 66.7 17.5 12.2 9.9 4.3 40.5 9.4 9.5 19.5 33.3 (3) (4) % of Screened Sources 95% Confidence Interval with Screening Values for Percentage of Sources •s. 10.000 ppmv 2 10.000 pprav 4.6 (3.6. 5.8 1.2 (0.9, 1.8 0.0 (0.0. 0.6 2.4 3.8 7.1 5.8 3.9 1.3 14.3 0.0 0.0 3.5 2.9 0.0 11.4 (] 6.4 0.4 8.8 2.1 0.3. 8.4 2.3, 5.8 0.9, 23.5) 4.4, 7.5 3.3, 4.6 0.5, 2.8 0.4, 57.9 0.0, 36.9 0.0, 97.5 0.7, 10.0 0.3. 10.1 0.0. 70.8 10.8. 12.1) 6.1. 6.8 0.2. 0.7 6.6, 11.1) 0.3, 7.3) 6.9 (0.9. 22.8) 21.0 6.1 0.0 6.0, 45.6 0.7, 20.2 0.0. 84.2 Includes filters, vacuum breakers, expansion joints, rupture disks, slight glass seals, etc. Source: Ref. 13 ------- TABLE A-4. AVERAGE FUGITIVE EMISSION SOURCE SCREENING RATES (Twenty-four Unit Study) Average Screening Number of Time Per Process Unit Type Screened Sources Source, Minutes 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. Vinyl Acetate Ethyl ene Vinyl Acetate Ethyl ene Cumene Cumene Ethyl ene Acetone/Phenol Ethylene Dichloride Vinyl Chloride Monomer Formaldehyde Ethylene Dichloride Vinyl Chloride Monomer Methyl Ethyl Ketone Methyl Ethyl Ketone Acetaldehyde Methyl Methacrylate Adi pic Acid Tri chol oroethyl ene/Perchl oroethyl ene 1 ,1 ,1-Trichloroethane Ethylene Dichloride Adi pic Acid Acrylonitrile Acrylonitrile 1,391 5,078 2,780 5,278 1,025 1,573 3,685 3,207 1,430 868 230 744 2,619 585 679 1,148 2,019 1,577 2,720 570 42 664 1,406 1,864 2.0 1.3 0.9 1.5 0.9 1.0 1.9 3.2 2.6 1.8 1.6 1.6 2.2 1.2 0.9 0.7 1.6 1.9 2.5 1.9 Total 43,182 1.7 aAverage source screening time was determined for a two-person team, one person screening with a portable hydrocarbon detector and one person recording data. Average screening time includes time spent for instrument calibration, maintenance, and repair. Source: Ref. 14 A-9 ------- TABLE A-5. SAMPLED PROCESS UNITS FROM NINE REFINERIES DURING REFINERY STUDY Refinery Number of process unit sampled units Atmospheric distillation 7 Vacuum distillation 4 Thermal operations (coking) 2 Catalytic cracking 5 Catalytic reforming 6 Catalytic hydrocracking 2 Catalytic hydrorefining 2 Catalytic hydrotreating 7 Alkylation 6 Aromatics/isomerization 3 Lube oil manufacture 2 Asphalt manufacture 1 Fuel gas/light-ends processing 11 LPG 2 Sulfur recovery 1 Other 3 Source: Ref. 15 A-10 ------- The screening of sources was accomplished with portable organic vapor detectors. The principal device used in this study was the J. W. Bacharach Instrument Co. "TLV Sniffer" calibrated with hexane. The components were tested on an individual basis and only those components with VOC concentrations in excess of 200 ppmv were considered for further study. A substantial portion of these leaking sources was enclosed and sampled to determine both the methane and nonmethane emission rates. An important result of this program was the development of a correlation between the maximum observed screening value (VOC concentration) and the measured nonmethane leak rate. Emission factors and leak frequency information generated during this study are given in Table A-6. A.1.5 Comparison of Fugitive Emissions Test Data from Refineries and SOCMI Units The results of the SOCMI studies and those of the refinery emissions study are compared in Table A-7. A.1.6 Description and Results of the DuPont Study DuPont conducted a program of VOC fugitive emission measurement from pumps and valves at two of their plants. The processes of the 5 and 10 year old plants were not revealed. The OVA-108 was used for screening (leak identification) and for leak rate determination (analysis of collected leak vapors). The leak rate was determined by taking Tedlar bags partially filled with air and enclosing the leaking valve. The hydrocarbon concentration in the bags was recorded as a function of time. Visual estimates of the initial bag volume were assumed to be ±5 percent. Dupont did not have a dilution probe and, therefore, measurements above 10,000 ppmv were not made. Analysis of the data collected indicates that no significant difference in leak rates exists between manual and automatic control valves. Significant trends were observed with changes in product vapor pressure. It also seemed that full open or closed valve seat positions resulted in lower leak rates than intermediate positions. The results of the DuPont study are shown in Table A-8. 24 25 A.1.7 Description and Results of the Exxon Study ' A fugitive emissions study was conducted by Exxon Chemical Company at the cyclohexane unit at their Baytown plant. The total number of valves, pumps and compressor seals, and safety valves was determined. For all A-ll ------- TABLE A-6. LEAK FREQUENCIES AND EMISSION FACTORS FROM FUGITIVE ..EMISSION SOURCES .IN PETROLEUM REFINERIES Percent of sources having screening values Equipment 2. 10,000 ppmv type TLV-Hexane Valves Gas service Light liquid service Heavy liquid service Pump seals Light liquid service Heavy liquid service Compressor seals (hydrocarbon service) Pressure relief valves Gas service Light liquid service Heavy liquid service Flanges Open-ended lines Gas service Light liquid service Heavy liquid service NA 10 12 0 NA 23 2 33 8 0 NA Estimated emission factor for refinery sources, kg/hr-source NA 0.021 0.010 0.0003 NA 0.12 0.02 0.44 0.086 0.16 0.006 0.009 0.0003 NA 0.025 0.014 0.003 Source: Ref. 17 A-12 ------- TABLE A-7. COMPARISON OF LEAK FREQUENCIES FOR FUGITIVE EMISSION SOURCES IN SOCMI UNITS AND PETROLEUM REFINERIES Equipment Type Valves (all) Gas Light Liquid Heavy Liquid Open-ended lines (all) Gas Light Liquid Heavy Liquid Pumps (all) Light Liquid Heavy Liquid Percent of SOCHI Sources Having Screening Values S 10.000 ppmv, OVA-108 Methane (six unit study) 11 10 17 Percent of SOCHI Sources Having Screening Values £ 10,000 ppmv, OVA-108 b Methane (24 unit study) 11.4 6.4 0.4 5.8 3.9 1.3 8.8 2.1 Percent of Petroleum Refinery Sources Having Screening Values i 10, 000 ppmv, TLV-Hexanec 10 12 0 N/A 23 2 Compressors (Gas) Pressure Relief Valves (all) 43 0 6.9 33 Gas Light Liquid Heavy Liquid Flanges (all) 3 Gas Light Liquid Heavy Liquid Process Drains (all) N/A Gas Light Liquid Heavy Liquid Agitator Seals (all) N/A Gas Light Liquid Heavy Liquid Other* N/A 3.5 2.9 0.0 4.6 1.2 0.0 2.4 3.8 7.1 14.3 0.0 0.0 N/A N/A N/A 0 N/A • 0.00 7.00 3.74 N/A N/A "Source: Ref. 18, 19, 20, 21 bSource: Ref. 22 cSource: Ref. 23. Screening with OVA-108 (methane) at 10,000 pprav is equivalent to screening with TLV (hexane) at 4,121 ppmv. Includes filters, vacuum breakers, expension joints, rupture disks, sight glass seals, etc. A-13 ------- TABLE A-8. FREQUENCY OF LEAKS3 FROM FUGITIVE EMISSION SOURCES IN TWO DuPONT PLANTS. Equipment type Valves Gas Light liquid Heavy liquid Pumps Light liquid Heavy liquid No. of leakers 47 35 11 1 1 1 0 No. of non-leakers 741 120 143 478 35 6 29 Percent leakers 6.1 23.1 7.1 0.2 2.7 14.3 0 aLeak defined as 10 ppmv or greater. Source: Ref. 26 A-14 ------- sources, except valves, all of the fugitive emission sources were sampled. For valves, a soap solution was used to determine leaking components. All leaking valves were counted and identified as either small, medium or large leaks. From the set of valves found to be leaking, specific valves were selected for sampling so that each class of leaking valves was in approximately the same proportion as it occured in the cyclohexane unit. Heat resistant mylar bags or sheets were taped around the equipment to be sampled to provide an enclosed volume. Clean metered air from the filter apparatus was blown into the enclosed volume. The sampling train was allowed to run until a steady state flow was obtained (usually about 15 minutes). A bomb sample was then taken for laboratory analysis (mass spectrometry). Table A-9 presents the results of the Exxon study. A.2 MAINTENANCE TEST PROGRAMS The results of four studies on the effects of maintenance on fugitive emissions from valves are discussed in this section. The first two studies were conducted by refinery personnel at the Union Oil Co. refinery in Rodeo, California, and the Shell Oil Co. refinery in Martinez, California. These programs consisted of maintenance on leaking valves containing fluids with Reid vapor pressures greater than 1.5. The third study was conducted by EPA. Valves were selected and maintained at four refineries. The fourth study was conducted by EPA at Unit D (ethylene unit). The study results and a description of each test program are given in the following sections. 29 A.2.1 Description and Results of the Union Maintenance Study The Union valve maintenance study consisted of performing undirected maintenance on valves selected from 12 different process units. Maintenance procedures consisted of adjusting the packing gland while the valve was in service. Undirected maintenance consists of performing valve repairs without simultaneous measurement of the effect of repair on the VOC concentration detected. This is in contrast to directed maintenance where emissions are monitored during the repair procedure. With directed maintenance, repair procedures are continued until the VOC concentration detected drops to a specified level or further reduction in the emission level is not possible. Also, maintenance may be curtailed if increasing VOC concentrations result. A-15 ------- TABLE A-9. FREQUENCY OF LEAKS3 FROM FUGITIVE EMISSION SOURCES IN EXXON'S CYCLOHEXANE UNIT Equipment Source Valves Gas Light liquid Safety valves Pump . seals Compressor sealsb Total in Unit 136 201 15 8 N/A Screened and Sampl ed 136 100 15 8 N/A Percent Emission Leaking factor(kg/hr) 32 15 87 83 100 0.017 0.008 0.064 0.255 0.264 99.8% Confidence Interval (kg/hr) 0.008 - 0.035 0.003 - 0.007 0.013 - 0.5 0.082 - 0.818 0.068 - 1.045 N/A - Not available aLeak defined as 200 ppmv or greater. Double mechanical seal pumps and compressors were found to have negligible leaks. Source: Reference 27,28 A-16 ------- The Union data were obtained with a Century Systems Corporation Organic Vapor Analyzer, OVA-108. All measurements were taken at a distance of 1 cm from the seal. Correlations developed by EPA have been used to convert the data from OVA readings taken at one centimeter to equivalent TLV readings at the leak interface (TLV-0).30 This facilitates comparison of data from different studies and allows the estimation of emission rates based on screening values-leak rate correlations. The results of the Union study are given in Table A-10. Two sets of results are provided; the first includes all repaired valves with before maintenance screening values greater than or equal to 5,300 ppmv (OVA-108), and the second includes valves with before maintenance screening values below 5,300 ppmv (OVA-108). A screening value of 5,300 ppmv, obtained with OVA at 1 cm from the leak interface, is equivalent to a screening value of 10,000 ppmv measured by a Bacharach Instrument Co. "TLV Sniffer" directly at the leak interface. The OVA-1 cm readings have been converted to equivalent TLV-0 cm readings because: 1) EPA correlations which estimate leak rates from screening values were developed from TLV-0 cm data. 2) Additional maintenance study data exists in the TLV-0 cm format. 3) EPA Reference Method 21 specifies 0 cm screening procedures. The results of this study indicate that maintenance on valves with initial screening values above 10,000 ppmv (OVA-108) is much more effective than maintenance on valves leaking at lower rates. In fact, this study indicates that emissions from valves are reduced by an average of 51.8 percent for valves initially over 5,300 ppmv while valves with lower initial screening values experienced an increase of 30.5 percent. A.2.2 Description and Results of the Shell Maintenance Study The Shell maintenance program consisted of two parts. First, valve repairs were performed on 171 leaking valves. In the second part of the program, 162 of these valves were rechecked and additional maintenance was performed. Maintenance consisted of adjusting the packing gland while the valve was in service. The second part of the program was conducted approximately one month after the initial maintenance period. It was not determined whether the maintenance procedures were directed or undirected, based on the information reported by Shell. A-17 ------- TABLE A-10. SUMMARY OF MAINTENANCE STUDY RESULTS FROM THE UNION OIL CO. REFINERY IN RODEO, CALIFORNIA9 00 All valves with initial screening values ^5300 ppmvb Source: Ref. 33. All valves with initial screening values <5300 ppmv Number of repairs attempted Estimated emissions before maintenance, kg/hr Estimated emissions after maintenance, kg/hr Number of successful repairs (<5300 ppmv after maintenance) Number of valves with decreased emissions Number of valves with increased emissions Percent reduction in emissions Percent successful repairs Percent of valves with decreased emissions Percent of valves with increased emissions 133 9.72 4.69 67 124 9 51.8 50.4 93.2 6.8 21 0.323 0.422 -- 13 8 -30.5 — 61.9 38.1 value of 5300 ppmv, taken with the OVA-108 at 1 cm, generally corresponds to a value of 10,000 ppmv taken with a "TLV Sniffer" at 0 cm. ------- Fugitive VOC emissions were measured one centimeter from the source using the OVA-108. These data have been transformed to TLV-0 cm values as were the Union data. The same methods of data analysis described in Section A.2.1 have been applied to the Shell data. The results of the Shell maintenance study are given in Table A-ll. A.2.3 Description and Results of the EPA Maintenance Study32 Repair data were collected on valves located in four refineries. The effects of both directed and undirected maintenance were evaluated. Maintenance consisted of routine operations, such as tightening the packing gland or adding grease. Other data, including valve size and type and the processes' fluid characteristics, were obtained. Screening data were obtained with the Bacharach Instrument Company "TLV Sniffer" and readings were taken as close to the source as possible. Unlike the Shell and Union studies, emission rates were not based on the screening value correlations. Rather, each valve was sampled to determine emission rates before and after maintenance using techniques developed by EPA during the refinery emission factor study. These values were used to evaluate emissions reduction. The results of this study are given in Table A-12. Of interest here is a comparison of the emissions reduction for directed and undirected maintenance. The results indicate that directed maintenance is more effective in reducing emissions than is undirected maintenance, particularly for valves with lower initial leak rates. The results showed an increase in total emissions of 32.6% for valves with initial screening values less than 10,000 ppmv which were subjected to undirected maintenance. However, this increase is due to a large increase in the emission rate of only one valve. A.2.4 Description and Results of Unit D (Ethylene Unit) Maintenance Study Maintenance was performed by Unit D personnel. Concentration measurements of VOC were made using the OVA-108, and readings were obtained at the closest distance possible to the source. The results of this study are shown in Table A-13. Directed and undirected maintenance procedures were used. The results show that directed maintenance results in more repairs being successfully completed than when undirected maintenance is used. A-19 ------- TABLE A-11. SUMMARY OF MAINTENANCE STUDY RESULTS FROM THE SHELL OIL COMPANY REFINERY IN MARTINEZ, CALIFORNIA3 I ro O Number of repairs attempted Estimated emissions before maintenance, kg/hr Estimated emissions after maintenance, kg/hr Number of successful repairs (<5300 ppmv after maintenance) Number of valves with decreased emissions Number of valves with Increased emissions Percent reduction In emissions Percent successful repairs Percent of valves with decreased emissions Percent of valves with Increased emissions March All repaired valves with Initial screening values >5300 ppmvb 161 11.08 2.66 105 161 0 76.0 65.2 100.00 0.0 maintenance All repaired valves with Initial screening values <5300 ppmv 11 0.159 0.0 — 11 0 100.0 " 100.0 0.0 April All repaired valves with Initial (March) screening values _>5300 ppmv 152d 2.95 0.421 45 151 1 B5.7 83.3 99.3 0.7 maintenance All repaired valves with initial (March) screening values <5300 ppmv lle 0.060 0.0 — 11 0 100.0 — 100.0 0.0 aSource: Reference 34. bThe value of 5300 ppmv. taken with the OVA-108 at 1 on. generally corresponds to a value of 10,000 ppmv taken with a "TLV Sniffer" at 0 cm. ""Shell reported the screening value of a]l valves which measured <3000 ppmv (<1500 ppmv-TLV at 0 cm) as non-leakers. Emissions estimates obtained from emission factors. Reference 14. Initial value of 90 of these valves was <1SOO ppm-TLV at 0 cm. 54 valves screened j>5300 (note nine valves from Initial data set not rechecked In April). eln1t1al value of 10 of these valves was <1500 ppm-TLV at 0 cm. ------- TABLE A-12. SUMMARY OF EPA REFINERY MAINTENANCE STUDY RESULTS I ro Repaired values with initial screening values >10,000 ppmv Number of valves repaired Measured emissions before maintenance, kg/hr Measured emissions after maintenance, kg/hr Number of successful repairs (<10,000 ppmv after maintenance) Number of valves with decreased emissions Number of valves with increased emissions Percent reduction in emissions Percent successful repairs Percent of valves with decreased emissions Percent of valves with increased emissions Di rected maintenance 9 0.107 0.0139 8 9 0 87.0 88.9 100.0 0.0 Undirected maintenance 23 1.809 0.318 13 21 2 82.4 56.5 91.3 8.7 Repaired values with initial screening values <10,000 ppmv Directed maintenance 10 0.0332 0.0049 -- 6 4 85.2 -- 60.0 40.0 Undirected maintenance 16 0.120 0.159 -- 15 1 -32.6 -- 93.8 6.3 Source: Ref. 36 ------- TABLE A-13. MAINTENANCE EFFECTIVENESS UNIT D ETHYLENE UNIT BLOCK VALVES 1. Total number of valves with VOC > 10,000 ppm from unit survey 121 2. Total number of valves tested for maintenance effectiveness 46 % Tested 38% UNDIRECTED MAINTENANCE 3. Total number subjected to repair attempts 37 4. Successful repairs (VOC<10,000 ppm) 22 % Repaired 59% Fo-llowup DIRECTED MAINTENANCE 5. Number of valves unrepaired by undirected 14 maintenance subjected to directed maintenance 6. Number repaired by followup directed maintenance 5 % of unsuccessful repaired by directed maintenance 36% 7. Total number repaired based on undirected 27 maintenance subset (3) above % Repaired 73% 8. Total number of repairs including leaks not 29 found before initial maintenance Total % repaired 63% Total % not repaired 37% Source: Reference 37 A-22 ------- 38 A.2.5 Description and Results of EPA-ORD Valve Maintenance Study A study was undertaken by the EPA Office of Research and Development (ORD) in order to determine the effectiveness of routine (on-line) maintenance in the reduction of fugitive VOC emissions from in-line valves. The overall effectiveness of a leak detection and repair program was examined by studying the immediate emission reduction due to maintenance, the propagation of the leaks after maintenance, and the rate at which new leaks occur for pumps and valves. Testing was conducted at six chemical plants, two for each of three chemical processes (ethylene, cumene, and vinyl acetate production). It was found that an estimated 71.3 percent (95 percent confidence limits of 54 percent to 88 percent) reduction in fugitive emissions from all valves leaking at various concentrations resulted immediately following maintenance (lasting up to six months). The 30-day rates of occurrence for valves and pumps initially screened at less than 10,000 ppm were 1.3 percent (95 percent confidence interval of 0.7 percent to 2.1 percent) and 5.5 percent (95 percent confidence interval of 2.2 percent to 10 percent), respectively, as shown in Table A-14. In Table A-15, 30-day, 90-day, and 180-day recurrence rate estimates are given along with approximate 95 percent confidence limits. Maintenance of valves in the study averaged about 10 minutes per valve. A.2.6 Comparison of Maintenance Study Results A summary of the results of the maintenance programs described in the preceding sections is presented in Table A-16. Generally speaking, the results of these maintenance programs would tend to support the following conclusions: • A reduction in emissions may be obtained by performing maintenance on valves with screening values above 10,000 ppmv (measured at the source). • The reduction in emissions due to maintenance of valves with screening values below 10,000 ppmv is not as dramatic and may result in increased emissions. A-23 ------- TABLE A-14. OCCURRENCE RATE ESTIMATES FOR VALVES AND PUMPS BY PROCESS IN EPA-ORD STUDY a,b 30-day estimate VALVES Cumene units Ethyl ene units Vinyl acetate units All units PUMPS Cumene units i\> Ethyl ene units Vinyl acetate units All units 1 2 0 1 5 18 2 5 .9 .0 .3 .3 .8 .4 .8 .5 95 percent confidence interval (0.2, (0.9, (0.0, (0.7, (0.7, (2.8, (0.8, (2.2, 5.9) 3.6) 0.6) 2.1) 20) 42) 6.2) 10) 90- day estimate 5 6 0 3 16 45 8 15 .6 .0 .8 .8 .3 .7 .1 .7 95 percent confidence interval (0.6, (2.7, (0.1, (2.0, (2.1, (8.2, (2.2, (6.6, 17) 10) 1.9) 6.0) 49) 80) 17) 27) 180-day estimate 10.8 11.6 1.5 7.4 30.0 70.5 15.6 29.0 95 percent confidence interval (1.3, (5.3, (0.3, (4.0, (4.2, (16, (4.4, (12, 30) 20) 3.8) 12) 74) 96) 32) 47) Reference 38. JA leak from a source is defined as having occurred if it initially screened <10,000 ppmv and at some later date screened >10,000 ppmv. ------- TABLE A-15. VALVE LEAK RECURRENCE RATE ESTIMATES3'b Recurrence rate 95 percent confidence estimate limits on the recurrence (percent) rate estimate 30-day 60-day 180-day 17.2 23.9 32.9 (5, (7, (10, 37) 48) 61) Reference 38. Data from 28 maintained valves were examined. Only those valves that screened greater than or equal to 10,000 ppmv immediately before maintenance and screened less than 10,000 ppmv immediately after maintenance were considered having a potential to recur. A-25 ------- TABLE A-16. SUMMARY OF VALVE MAINTENANCE TEST RESULTS Maintenance test Number of valve repairs attempted Number of successful repairs Percent repaired Union3 133 67 50.4 Shell3 March 1979 161 105 65.2 April 1979 54 45 83.3 EPA-4 refineries Directed0 . 9 8 88.9 Undirected0 23 13 56.5 Unit D (Ethylene)b Directed and undirected 46 29 63.0 EPA-ORDb Directed 97 28 TOTAL 523 295 3Ihitial screening value of >5,300 ppmv at 1 cm was used to define the population subject to repair. Repair was successful when a valve screened <5,300 ppmv at 1 cm. Before maintenance screening value of_>10,000 ppmv at 0 cm was used to define the population subject to repair. Repair was successful when a valve screened <10,000 ppmv at 0 cm. C0irected maintenance refers to a valve maintenance procedure whereby the hydrocarbon detector is utilized during maintenance. The leak is monitored with the instrument until no further reduction of leak is observed or the valve stem rotation is restricted. Undirected maintenance refers to action by plant personnel in which an assigned worker tightens the valve packing gland with a wrench to further compress the packing material around the valve stem and seat. A-26 ------- The information presented in the tables of Appendix A has been compiled with the objective of placing the data on as consistent a basis as possible. However, some differences were unavoidable and others may have gone unrecognized, due to the limited amount of information concerning the details of methods used in each study. Therefore, care should be exercised before attempting to draw specific quantitative conclusions based on direct comparison of the results of these studies. A-27 ------- A.3 REFERENCES 1. Tlchenor, Bruce A., memo to K.C. Hustvedt, U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards. Research Triangle Park, N.C. October 27, 1980. 2. Muller, Christopher, memo to files. U.S. Environmental Protection Agency, Emission Standards and Engineering Division. Research Triangle Park, N.C. January 18, 1979 (Plants A & B). 3. Muller, Christopher, memo to file. U.S. Environmental Protection Agency, Emission Standards and Engineering Division, Research Triangle Park, N.C. March 19, 1979 {Plant C). 4. Muller, Christopher, memo to file. U.S. Environmental Protection Agency, Emission Standards and Engineering Division. Research Triangle Park, N.C. March 19, 1979. (Plant D). 5. Hustvedt, K.C., trip report to James F. Durham, Chief, Petroleum Section. U.S. Environmental Protection Agency. January 5, 1979 (Plants E & F). 6. Reference 2. 7. Reference 3. 8. Reference 4. 9. Reference 5. 10. Blacksmith, J.R., G.E. Harris, and G.J. Langley, Frequency of Leak Occurence for Fittings in Synthetic Organic Chemical Plant Process Units. EPA Contract Numbers 68-02-3176-01,06/68-02-3173-02,11/68-02- 3171-01/68-02-3174-04. Radian Corporation, Austin, Texas. September 1980. 11. Reference 10. 12. Wetherold, R.G., and L.P. Provost, Emission Factors and Frequency of Leak Occurence for Fittings in Refinery Process Units. EPA-600/2-79- 044. Radian Corporation, Austin, Texas. February 1979. 13. Reference 10. 14. Reference 10. 15. Reference 12. 16. Meeting Report. Meeting held between U.S. Environmental Protection Agency and DuPont at Durham, N.C. June 12, 1979. A-28 ------- 17. Reference 12. 18. Reference 2. 19. Reference 3. 20. Reference 4. 21. Reference 5. 22. Reference 10. 23. Reference 12. 24. Fugitive loss study summary and memo from James B. Cox, Exxon Chemical Company, Baytown, Texas, to J.W. Blackburn, Hydroscience Incorporated. February 21, 1978. 25. Letter from James B. Cox, Exxon Chemical Company, Baytown, Texas, to R.T. Walsh, U.S. Environmental Protection Agency, Chemical and Petroleum Branch. March 21, 1979. 26. Reference 16. 27. Reference 24. 28. Reference 25. 29. Valve Repair Summary and Memo from F.R. Bottomley, Union Oil Company, Rodeo, Calfornia, to Milton Feldstein, Bay Area Quality Management District. April 10, 1979. 30. Reference 12. 31. Valve Repair Summary and Memo from R.M. Thompson, Shell Oil Company, Martinez Manufacturing Complex, Martinez, California. To Milton Feldstein, Bay Area Quality Management District. April 26, 1979. 32. Radian Corporation. Assessment of Atmospheric Emissions from Petroleum Refining, Final Report, Appendix B, Detailed Results. EPA Report No. 600/2-80-075C, Exhibit B. Austin, Texas. April, 1980. 33. Reference 29. 34. Reference 31. A-29 ------- 35. Air Pollution Emission test at Phillips Petroleum Company, Sweeney, Texas. EMB Report No. 78-DCM-12E, December 1979. 36. Reference 32. 37. Reference 35. 38. Langley, G.J. and R.G. Wetherold. Evaluation of Maintenance for Fugitive VOC Emissions Control. Radian Corporation. Industrial Environmental Research Laboratory. Cincinnati, OH. EPA-600/S2-81-080. May 1981. A-30 ------- APPENDIX B. LIST DEFINING SYNTHETIC ORGANIC CHEMICAL, POLYMER, AND RESIN MANUFACTURING INDUSTRIES ------- APPENDIX B LIST OF CHEMICALS DEFINING SYNTHETIC ORGANIC CHEMICAL, POLYMER, AND RESIN MANUFACTURING INDUSTRIES TABLE I: Synthetic Organic Chemicals Manufacturing Industry OCPDB No.*_ Chemical 20 Acetal 30 Acetaldehyde 40 Acetaldol 50 Acetamlde 65 AcetanlHde 70 Acetic acid 80 Acetic anhydride 90 Acetone 100 Acetone cyanohydrin 110 Acetonltrile 120 Acetophenone 125 Acetyl chloride 130 Acetylene 140 Acrolein 150 Acrylamlde 160 Acrylic add and esters 170 Acrylonltrlle 180 Adlpic acid 185 Adipon1tr11e 190 Alkyl naphthalenes 200 Ally! alchohol 210 Allyl chloride 220 Am1nobenzo1c add OCPDB No. Chemical 230 Aminoethylethanolamlne 235 p-amlnophenol 240 Arayl acetates 250 Amy! alcohols 260 Amyl amine 270 Arayl chloride 280 Aroyl mercaptans 290 Arayl phenol 300 Aniline 310 Aniline hydrochloride 320 Anisidine 330 Anisole 340 Anthranilic acid 350 Anthraquinone 360 Benzaldehyde 370 Benzamide 380 Benzene 390 Benzenedisulfonic add 400 Benzenesulfonic acid 410 Benzil 420 BenziHc acid 430 Benzole acid 440 Benzoin *The OCPDB Numbers are reference indices assigned to the various chemicals in the Organic Chemical Producers Data Base developed by EPA. B-l ------- OCPD8 NO. Chemical OCPDB No. Chemical 450 BenzonltMle 460 Benzophenone 480 Benotr-1 chloride 490 Benzoyl chloride 500 Benzyl alcohol 510 Benzyl aralne 520 Benzyl benzoate 530 Benzyl chloride 540 Benzyl d1chloride 550 B-lphenyl 560 Blsphenol A 570 Bromobenzene 580 Brononaphthalene 590 Butadiene 592 1-butene 600 n-butyl acetate 630 n-butyl acrylate 640 n-butyl alcohol 650 s-butyl alcohol 660 t-butyl alcohol 670 n-butyl amine 680 s-butyl anine 690 t-butyl anrlne 700 p-tert-butyl benzole add 710 1,3-butylene glycol 750 n-butyraldehyde 760 Butyric acid 770 Butyric anhydride 780 Butyronltrile 785 Caprolactam 790 Carbon disulfide 800 Carbon tetrabromide 810 Carbon tetrachloride 820 Cellulose acetate 840 Chloroacetic add 850 ra-chloroani1i ne 860 o-chloroanillne 870 p-chloroanH1ne 880 Chlorobenzaldehyde 890 Chlorobenzene 900 Chlorobenzoic add 905 Chlorobenzotrichlorlde 910 Chlorobenzoyl chloride 920 Chlorodlfluoroethane 921 Chlorodlfluoromethane 930 Chloroform 940 Chloronaphthalene 950 o-chloronitrobenzene 951 p-chloronitrobenzene 960 Chlorophenols 964 Chloroprene 965 Chlorosulfonlc add 970 ra-chloroto1uene 980 o-chlorotoluene 990 p-chlorotoluene 992 ChlorotHfluororaethane 1000 m-cresol 1010 o-cresol 1020 p-cresol 1021 Mixed cresols 1030 Cresyllc add 1040 Crotonaldehyde 1050 Crotonic acid 1060 Cumene 1070 Cumene hydroperoxide 1080 Cyanoacetic acid 1090 Cyanogen chloride 1100 Cyanuric add 1110 Cyanuric chloride 1120 Cyclohexane 1130 Cyclohexanol 1140 Cyclonexanone B-2 ------- OCPDB No. Chemical 1150 Cyclohexene 1160 Cyclohexylamine 1170 Cyclooctadlene 1180 Decanol 1190 D1acetone alcohol 1200 D1am1nobenzo1c acid 1210 Dichloroaniline 1215 m-dichlorobenzene 1216 o-dichlorobenzene 1220 p-d1chlorobenzene 1221 Dlchlorodifluoromethane 1244 l,2-d1chloroethane (EDC) 1240 Dichloroethyl ether 1250 Dichlorohydrin 1270 Dichloropropene 1280 D1Cyclohexylamine 1290 D1ethyl amine 1300 Diethylene glycol 1304 Diethylene glycol dlethyl ether 1305 Diethylene glycol dimethyl ether 1310 Diethylene glycol monobutyl ether 1320 Diethylene glycol monbutyl ether acetate 1330 Diethylene glycol monoethyl ether 1340 Diethylene glycol nonoethyl ether acetate 1360 Diethylene glycol nonomethyl ether 1420 Dlethyl sulfate 1430 Dif1uoroethane 1440 DHsobutylene 1442 DUsodecyl phthalate 1444 OUsooctyl phthalate 1450 Oiketene 1460 Dimethylamine 1470 N,N-d1methylan1line 1480 N,N-d1methyl ether 1490 N,N-d1methylformam1de OCPDB No. Chemical 1495 Dimethylhydrazine 1500 Dimethyl sulfate 1510 Dimethyl sulflde 1520 Dimethyl sulfoxide 1530 Dimethyl terephthalate 1540 3,5-d1n1trobenzo1c acid 1545 Dinitrophenol 1550 Dinltrotoluene 1560 Dioxane 1570 Dloxolane 1580 D1 pheny1 amine 1590 Dlphenyl oxide 1600 Dlphenyl thiourea 1610 Dipropylene glycol 1620 Dodecene 1630 Dodecylaniline 1640 Dodecylphenol 1650 Epichlorohydrin 1660 Ethanol 1661 Ethanolamines 1670 Ethyl acetate 1680 Ethyl acetoacetate 1690 Ethyl acrylate 1700 Ethylamine 1710 Ethyl benzene 1720 Ethyl bromide 1730 Ethylcellulose 1740 Ethyl chloride 1750 Ethyl chloroacetate 1760 Ethylcyanoacetate 1770 Ethylene 1780 Ethylene carbonate 1790 Ethylene chlorohydrin 1800 Ethylenediamine 1810 Ethylene dlbromide 1830 Ethylene glycol B-3 ------- OCPD8 No. Chemical 1840 Ethylene glycol diacetate 1870 Ethylene glycol dimethyl ether 1890 Ethylene glycol monobutyl ether 1900 Ethylene glycol monobutyl ether acetate 1910 Ethylene glycol monoethyl ether 1920 Ethylene glycol monoethyl ether acetate 1930 Ethylene glycol monomethyl ether 1940 Ethylene glycol monomethyl ether acetate 1960 Ethylene glycol monophenyl ether 1970 Ethylene glyco! monopropyl ether 1980 Ethylene oxide 1990 Ethyl ether 2000 2-ethylhexanol 2010 Ethyl orthoformate 2020 Ethyl oxalate 2030 Ethyl sodium oxalacetate 2040 Formaldehyde 2050 Formamlde 2060 Formic add 2070 Fumarlc add 2073 Furfural 2090 Glycerol (Synthetic) 2091 Glycerol dlchlorohydrln 2100 Glycerol trlether 2110 Glydne 2120 Glyoxal 2145 Hexachlorobenzene 2150 Hexachloroethane 2160 Hexadecyl alcohol 2165 Hexamethylened1am1ne 2170 Hexanethylene glycol 2180 Hexamethylenetetramlne 2190 Hydrogen cyanide 2200 Hydroqulnane 2210 p-hydroxybenzole add 2240 Isoamylene 2250 Isobutanol 2260 Isobutyl acetate 2261 Isobutylene 2270 Isobutyraldehyde 2280 Isobutyrlc add n_* OCPDB No. Chemical 2300 Isodecanol 2320 Isooctyl alcohol 2321 Isopentane 2330 Isophorone 2340 Isophthallc add 2350 Isoprene 2360 Isopropanol 2370 Isopropyl acetate 2380 Isopropylamine 2390 Isopropyl chloride 2400 Isopropylphenol 2410 Ketene 2414 Linear alkyl sulfonate 2417 Linear alkylbenzene 2420 Haleic add 2430 Haleic anhydride 2440 Halle add 2450 Mesltyl oxide 2455 HetanlUc add 2460 Hethacrylic add 2490 Methallyl chloride 2500 Methanol 2510 Methyl acetate 2520 Methyl acetoacetate 2530 Hethylamlne 2540 n-methylannine 2545 Methyl bromide 2550 Methyl butynol 2560 Methyl chloride 2570 Methyl cyclohexane 2590 Methyl cyclohexanone 2620 Methylene chloride 2530 Methylene d1an111ne 2635 Methylene dlphenyl d11socyanate 2640 Methyl ethyl ketone 2644 Methyl formate 2650 Methyl Isobutyl carblnol 2660 Methyl Isobutyl ketone 2665 Methyl methacrylate 2670 Methyl pentynol 2690 a -methylstyrene ------- OCPDB No. Chemical 2700 Morpholine 2710 a-naphthalene sulfonlc add 2720 a-n»phthalene sulfonic add 2730 a-naphthol 2740 0-naphthol 2750 Neopentanolc add 2756 o-nitroaniline 2757 p-nitroaniline 2760 o-nitroanisole 2762 p-n1troan1sole 2770 Nitrobenzene 2780 N1trobenzo1c add (o, m, and p) 2790 NUroethane 2791 Nltromethane 2792 Nltrophenol 2795 Nltropropane 2800 Nitrotoleune 2810 Nonene 2820 Nonyl phenol 2830 Octyl phenol 2840 Paraldehyde 2850 Pentaerythrltol 2851 n-pentane 2855 1-pentene 2860 Perchloroethylene 2882 Perchloromethyl mercaptan 2890 o-phenetidine 2900 p-phenet1dine 2910 Phenol 2920 Phenolsulfonic acids 2930 Phenyl anthranilic acid 2940 Phenylenedlamine 2950 Phosgene 2960 Phthalic anhydride 2970 Phthalimide 2973 s-picoline 2976 Piperazine OCPDB No. Chemical 3000 Polybutenes 3010 Polyethylene glycol 3025 Polypropylene glycol 3063 Proplonaldehyde 3066 Prop1on1c add 3070 n-propyl alcohol 3075 Propylamine 3080 Propyl chloride 3090 Propylene 3100 Propylene chlorohydrln 3110 Propylene dichlorlde 3111 Propylene glycol 3120 Propylene oxide 3130 Pyrldine 3140 Quinone 3150 Resorcinol 3160 Resorcylic acid 3170 Salicylic acid 3180 Sodium acetate 3181 Sodium benzoate 3190 Sodium carboxymethyl cellulose 3191 Sodium chloroacetate 3200 Sodium formate 3210 Sodium phenate 3220 Sorbic acid 3230 Styrene 3240 Succinic add 3250 Sucdnltrile 3251 Sulfanilic acid 3260 Sulfolane 3270 Tannic acid 3280 Terephthalic acid 3290 & 3291 Tetrachloroethanes 3300 Tetrachlorophthalic anhydride 3310 Tetraethyllead 3320 Tetrahydronaphthalene 3330 Tetrahydrophthalic anhydride 3335 Tetramethyllead B-5 ------- OCPDB No. Chemical 3340 Tetrainethylened1am1ne 3341 Tetramethylethylenedlamine Table II: Polymer and 3349 Toluene Manufacturing Industry 3350 Toluene-2,4-d1am1ne 3354 Toluene-2,4-d11socyanate Polyethylene 3355 Toluene dUsocyanates (mixture) Polypropylene 3360 Toluene sulfonoride Polystyrene 3370 Toluene sulfonlc adds Styrene-butadlene latex 3380 Toluene sulfonyl chloride 3381 Toluldlnes 3390, 3391 Trlchlorobenzenes & 3393 3395 1,1,1-tHchloroethane 3400 1,1,2-trichloroethane 3410 Trichloroethylene 3411 Trlchlorofluoromethane 3420 1,2,3-trlchloropropane 3430 I,l,2-tr1chloro-l,2,2-tr1f1uoroethane 3450 Tr1ethylamine 3460 THethylene glycol 3470 THethylene gylcol dimethyl ether 3480 TrUsobutylene 3490 THmethylamlne 3510 Vinyl acetate 3520 Vinyl chloride 3530 V1nyl1dene chloride 3540 Vinyl toluene 3541 Xylene (mixed) 3560 o-xylene 3570 p-xylene 3580 Xylenol 3590 Xyl1d1ne B-6 ------- APPENDIX C. METHOD 21. DETERMINATION OF VOLATILE ORGANIC COMPOUND LEAKS ------- APPENDIX C METHOD 21. DETERMINATION OF VOLATILE ORGANIC COMPOUND LEAKS 1. Applicability and Principle 1.1 Applicability. This method applies to the determination of volatile organic compound (VOC) leaks from organic process equipment. These sources include, but are not limited to, valves, flanges and other connections, pumps and compressors, pressure relief devices, process drains, open-ended valves, pump and compressor seal system degassing vents, accumulator vessel vents, and access door seals. 1.2 Principle. A portable instrument is used to detect VOC leaks from individual sources. The instrument detector is not specified, but it must meet the specifications and performance criteria contained in paragraph 2.1. 2. Apparatus 2.1 Monitoring Instrument. The monitoring instrument shall be as follows: 2.1.1 Specifications. a. The VOC instrument detector shall respond to the organic compounds being processed. Detectors which may meet this requirement include, but are not limited to, catalytic oxidation, flame ionization, infrared absorption, and photoionization. b. The instrument shall be intrinsically safe for operation in explosive atmospheres as defined by the applicable U.S.A. Standards (e.g., National Electrical Code by the National Fire Prevention Association). C-l ------- c. The instrument shall be able to measure the leak definition concentration specified in the regulation. d. The instrument shall be equipped with a pump so that a continuous sample is provided to the detector. The nominal sample flow rate shall be 1-3 liters per minute. e. The scale of the instrument meter shall be readable to ±5 percent of the specified leak definition concentration. 2.1.2 Performance Criteria. The instrument must meet the following performance criteria. The definitions and evaluation procedures for each parameter are given in Section 4. 2.1.2.1 Response Time. The instrument response time must be 30 seconds or less. The response time must be determined for the instrument system configuration to be used during testing, including dilution equipment. The use of a system with a shorter response time than that specified will reduce the time required for field component surveys. 2.1.2.2 Calibration Precision. The calibration precision must be less than or equal to 10 percent of the calibration gas value. 2.1.2.3 Quality Assurance. The instrument shall be subjected to the response time and calibration precision tests prior to being placed in service. The calibration precision test shall be repeated every 6 months thereafter. If any modification or replacement of the instrument detector is required, the instrument shall be retested and a new 6-month quality assurance test schedule will apply. The response time test shall be repeated if any modifications to the sample pumping system or flow configuration is made that would change the response time. C-2 ------- 2.3 Calibration Gases. The monitoring instrument is calibrated in terms of parts per million by volume (ppmv) of the compound specified in the applicable regulation. The calibration gases required for monitoring and instrument performance evaluation are a zero gas (air, <3 ppmv VOC) and a calibration gas in air mixture approximately equal to the leak definition specified in the regulation. If cylinder calibration gas mixtures are used, they must be analyzed and certified by the manufacturer to be within ±2 percent accuracy. Calibration gases may be prepared by the user according to any accepted gaseous standards preparation procedure that will yield a mixture accurate to within ±2 percent. Alternative calibration gas species may be used in place of the calibration compound if a relative response factor for each instrument is determined so that calibrations with the alternative species may be expressed as calibration compound equivalents on the meter readout. 3. Procedures 3.1 Calibration. Assemble and start up the VOC analyzer and recorder according to the manufacturer's instructions. After the appropriate warm-up period and zero or internal calibration procedure, introduce the calibration gas into the instrument sample probe. Adjust the instrument meter readout to correspond to the calibration gas value. If a dilution apparatus is used, calibration must include the instrument and dilution apparatus assembly. The nominal dilution factor may be used to establish a scale factor for converting to an undiluted basis. For example if a nominal 10:1 dilution apparatus is used, the meter reading for a 10,000 ppm calibration compound would be set at 1000. During field surveys, the scale factor of 10 would be used to convert measurements to an undiluted basis. C-3 ------- 3.2 Individual Source Surveys. 3.2.1 Case I - Leak Definition Based on Concentration Value. Place the probe inlet at the surface of the component interface where leakage could occur. Move the probe along the interface periphery while observing the instrument readout. If an increased meter reading is observed, slowly probe the interface where leakage is indicated until the maximum meter reading is obtained. Leave the probe inlet at this maximum reading location for approximately two times the instrument response time. If the maximum observed meter reading is greater than the leak definition in the applicable regulation, record and report the results as specified in the regulation reporting requirements. Examples of the application of this general technique to specific equipment types are: a. Valves—The most common source of leaks from valves is at the seal between the stem and housing. Place the probe at the interface where the stem exits the packing gland and sample the stem circumference. Also, place the probe at the interface of the packing gland take-up flange seat and sample the periphery. In addition, survey valve housings of multipart assembly at the surface of all interfaces where leaks can occur. b. Flanges and Other Connections—For welded flanges, place the probe at the outer edge of the flange-gasket interface and sample around the circumference of the flange. Sample other types of nonpermanent joints (such as threaded connections) with a similar traverse. c. Pumps and Compressors—Conduct a circumferential traverse at the outer surface of the pump or compressor shaft and seal interface. If the source is a rotating shaft, position the probe inlet within one centimeter of the shaft seal interface for the survey. If the housing configuration C-4 ------- prevents a complete traverse of the shaft periphery, sample all accessible portions. Sample all other joints on the pump or compressor housing where leakage can occur. d. Pressure Relief Devices—The configuration of most pressure relief devices prevents sampling at the sealing seat interface. For those devices equipped with an enclosed extension, or horn, place the probe inlet at approximately the center of the exhaust area to the atmosphere for sampling. e. Process Drains—For open drains, place the probe inlet at approximately the center of the area open to the atmosphere for sampling. For covered drains, place the probe at the surface of the cover interface and conduct a peripheral traverse. f. Open-Ended Lines or Valves—Place the probe inlet at approximately the center of the opening to the atmosphere for sampling. g. Seal System Degassing Vents and Accumulator Vents—Place the probe inlet at approximately the center of the opening to the atmosphere for sampling. h. Access Door Seals—Place the probe inlet at the surface of the door seal interface and conduct a peripheral traverse. 3.2.2 Case II-Leak Definition Based on "No Detectable Emission". a. Determine the local ambient concentration around the source by moving the probe inlet randomly upwind and downwind at distance of one to two meters from the source. If an interference exists with this determination due to a nearby emission or leak, the local ambient concentration may be determined at distances closer to the source, but in no case shall the distance be less than 25 centimeters. Note the ambient concentration and then move the probe inlet to the surface of the source and conduct a survey as described C-5 ------- in 3.2.1. If a concentration increase greater than 2 percent of the concentration-based leak definition is obtained, record and report the results as specified by the regulation. b. For those cases where the regulation requires a specific device installation, or that specified vents be ducted or piped to a control device, the existence of these conditions shall be visually confirmed. When the regulation also requires that no detectable emissions exist, visual observations and sampling surveys are required. Examples of this technique are: i. Pump or Compressor Seals—If applicable, determine the type of shaft seal. Perform a survey of the local area ambient VOC concentration and determine if detectable emissions exist as described in 3.2.2.a. ii. Seal System Degassing Vents, Accumulator Vessel Vents, Pressure Relief Devices—If applicable, observe whether or not the applicable ducting or piping exists. Also, determine if any sources exist in the ducting or piping where emissions could occur prior to the control device. If the required ducting or piping exists and there are no sources where the emissions could be vented to the atmosphere prior to the control device, then it is presumed that no detectable emissions are present. 4. Instrument Performance Evaluation Procedures 4.1 Definitions. 4.1.1 Calibration Precision. The difference between the average VOC concentration indicated by the meter readout for consecutive calibration repetitions and the known concentration of a test gas mixture. 4.1.2 Response Time. The time interval from a step change in VOC concentration at the input of the sampling system to the time at which 90 percent of the corresponding final value is reached as displayed on the instrument readout meter. C-6 ------- 4.2 Evaluation Procedures. At the beginning of the instrument performance evaluation test, assemble and start up the instrument according to the manufacturer's instructions for recommended warmup period and preliminary adjustments. If a dilution apparatus is used during field surveys, the evaluation procedure must be performed on the instrument-dilution system combination. 4.2.1 Calibration Precision Test. Make a total of nine measurements by alternately using zero gas and the specified calibration gas. Record the meter readings (example data sheet shown in Figure 21-1). 4.2.2 Response Time Test Procedure. Introduce zero gas into the instrument sample probe. When the meter reading has estabilized, switch quickly to the specified calibration gas. Measure the time from concentration switching to 90 percent of final stable reading. Perform this test sequence three times and record the results (example data sheet given in Figure 21-2). 4.3 Calculations. All results are expressed as mean values, calculated by: 7 = n 1?1 *1 where: x. = Value of thetndividual measurements = Sum of the individual values. x" = Mean value. n = Number of measurements. C-7 ------- Instrument ID Calibration Gas Data Calibration = ppmv Run Instrument Meter Difference^ ' No. Reading, ppm ppm 1. 2. 3. 4. 5. 6. 7. 8. 9. Mean Difference Calibration Precision = * 10° (1)Calibration Gas Concentration - Instrument Reading Figure C-l. Calibration Precision Determination C-8 ------- Instrument ID Calibration Gas Concentration 90% Response Time: 1. Seconds 2. Seconds 3. Seconds Mean Response Time Seconds Figure C-2. Response Time Determination C-9 ------- APPENDIX D. EXAMPLE CALCULATIONS FOR DETERMINING REDUCTION IN EMISSIONS FROM IMPLEMENTATION OF RACT ------- APPENDIX D Example Calculations for Determining Reduction in Emissions from Implementation of RACT The purpose of this appendix is to provide emission factors and procedures for computing the emission reduction associated with reasonably available control technology (RACT) for control of fugitive emissions in the synthetic organic chemical, polymer, and resin manufacturing industries. The emission factors and procedures allow the reader to compute total process unit uncontrolled emissions, uncontrolled emissions from equipment affected by RACT, controlled emissions from equipment affected by RACT, and the emission reduction from implementing RACT. The emission factors and computation procedures in this appendix should be used only where specific plant fugitive emission data are not available. Uncontrolled emission factors for each type of component are presented in Table D-l. The uncontrolled emissions from a process unit can be computed by multiplying the uncontrolled emission factor for each component type by the number of components of the given type, then summing the emissions from all component types. The average emissions from components controlled under RACT (pumps in light liquid service, compressors, valves in gas service and light liquid service, and pressure relief valves in gas service) following implementation of RACT are presented in Table D-2. The basis for these factors has already been discussed in Chapters 3 and 4. The number of components in a process unit may be estimated (at the Director's discretion) or derived from actual count. A sample calculation of uncontrolled emissions from an illustrative process unit is presented below in Table D-3. D-l ------- TABLE D-l. UNCONTROLLED FUGITIVE EMISSION FACTORS IN PROCESS UNIT EOUIPMENTa Uncontrolled Fugitive emission factor, emission source kg/hr Pumps Light liquids 0.12 Heavy liquids 0.020 Valves (in-line) Gas service 0.021 Light liquid service 0.010 Heavy liquid service 0.0003 Pressure relief valves Gas service 0.16 Light liquid service 0.006 Heavy liquid service 0.009 Open-ended valves Gas service 0.025 Light liquid service 0.014 Heavy liquid service 0.003 .Flanges 0.0003 Sampling connections 0.015 Compressors 0.44 aFrom Table 2-2. TABLE D-2. CONTROLLED EMISSION FACTORS FOR EQUIPMENT AFFECTED BY RACT Controlled Fugitive emission factor, emission source kg/hr Pumps in light liquid service 0.036 Valve in gas service 0.003 Valve in light liquid service 0.003 Gas pressure relief valves 0.066 Compressors . 0.123 aFrom Table 4-3. D-2 ------- TABLE D-3. EXAMPLE CALCULATION OF UNCONTROLLED EMISSIONS FROM AN ILLUSTRATIVE PROCESS UNIT Fugitive emission source Pumps Light liquid service0 Heavy liquid service In-line valves Gas service Light liquid service Heavy liquid service Pressure relief valves Gas service Light liquid service Heavy liquid service Open-ended valves Gas service Light liquid service Heavy liquid service Compressors0 Sampling connections Flanges Number of sources (N) 16 10 • 15°H 120d 100d 17 2, 3 12 19 17 2 34 750d Emission factor , kg/hr-source (E) 0.12 0.020 0.021 0.010 0.0003 0.16 0.006 0.009 0.025 0.014 0.003 0.44 0.015 0.0003 Total emissions Emissions from source, kg/hr (N X E) 1.92 0.2 3.15 1.2 0.03 2.72 0.012 0.027 0.3 0.266 0.051 0.88 0.51 0.225 = 11.49 kg/hre Determined by actual count or by estimate. bFrom Table D-l. Inspected under RACT. Because of the large number of components, an estimate may be appropriate. eThe expected annual emissions are 100.65 Mg/yr (8,760 hr/yr). D-3 ------- The overall emission reduction associated with RACT can be computed by finding the emission reduction for those process units maintained under RACT. The emission reduction is the difference between the emissions expected from components in the unrepaired (uncontrolled) condition and in the repaired (controlled) condition. The uncontrolled emissions are presented in Table D-4: TABLE D-4. UNCONTROLLED EMISSIONS FROM COMPONENTS AFFECTED BY RACT. Emission source Light liquid pump Gas service valve Light liquid service valve Gas pressure relief valve Compressors Number of sources (N) 16 150 120 17 2 Emission factor kg/hr-source (E) 0.12 0.021 0.010 0.16 0.44 Emissions from sources, kg/hr (N X E) 1.92 3.15 1.2 2.72 0.88 9.87 The emissions expected after implementing RACT are also presented in Table D-5: TABLE D-5. CONTROLLED EMISSIONS FROM COMPONENTS AFFECTED BY RACT. Emission source Light liquid pump Gas service valve Light liquid service valve Gas pressure relief valve Compressors Number of sources (N) 16 150 120 17 2 Emission factor kg/hr-source (E) 0.036 0.003 0.003 0.066 0.123 Emissions, from sources, kg/hr (N X E) 0.576 0.45 0.36 1.12 0.246 2.75 From Table D-2. D-4 ------- The difference between emissions before and after the implementation of RACT can be expressed as an annual emission reduction. This emission reduction is presented in Table D-6: TABLE D-6. EMISSION REDUCTION EXPECTED FROM RACT 1. Uncontrolled emissions from process unit 100.65 Mg/yr (11.49 kg/hr) 2. Uncontrolled emissions from components affected 86.46 Mg/yr by RACT (9.87 kg/hr) 3. Controlled emissions from components 24.09 Mg/yr affected by RACT (2.75 kg/hr) 4. Total emission reduction: 86.46 Mg/yr - 24.09 Mg/yr = 62.35 Mg/yr (7.12 kg/hr) 5. Percent emission reduction: 62.35 Mg/yr inn _ 100.65 Mg/yr x 10° ~ D-5 ------- |