UNITED STATES ENVIRONMENTAL PROTECTION AGENCY WASHINGTON, D.C. 2046 FEB 24, 1992 OFFICE OF AIR AND RADIATION MEMORANDUM SUBJECT: Use of Long Term Rolling Averages to Limit Potential to Emit FROM: John B. Rasnic, Director Stationary Source Compliance Division Office of Air Quality Planning and Standards TO: David Kee, Director Air and Radiation Division Region V This is in response to your memorandum dated September 17, 1991 and several other recent requests for clarification of the guidance entitled "Limiting Potential to Emit in New Source Permitting" (signed by Terrel Hunt and John Seitz, dated June 13, 1989). My staff met with your staff on these issues in October, 1991, and in response has surveyed the Regional offices for any additional suggestions for clarification of the policy. It was our understanding last fall that a direct response to your September memorandum would not be responsive to your needs. However, we have since determined that guidance on application of the "Potential to Emit Policy" to the nine source categories listed in your September 17 memorandum is warranted. To that end, this memorandum gives guidelines for determination of whether to allow long term rolling averages for the nine source categories. Our survey of the Regions helped us to formulate our response regarding the nine source categories. As we have stated before, we believe that each case in which a source seeks to restrict its potential to emit by imposition of long-term production limits (i.e., limits that exceed one month) must be independently evaluated. Therefore, the facts of a specific case may lead to a different response, and the availability of a 12 month rolling average for the nine listed sources is not automatic. As you know, the potential to emit policy allows use of long-term rolling averages in any case where a source experiences "substantial and unpredictable" annual variations in production. Thus, it is the burden of the source to demonstrate the need for flexibility. In no event shall a source be allowed longer than an annual average rolled less frequently than a month. 2 General responses regarding the acceptability of long term averages for the nine source categories follows: 1. Agricultural production such as harvesting or food processing where part of a year the equipment is idle (sugar beet processing facilities). Rolling averages have been allowed for seasonal food processors. 365 day rolling averages are appropriate given the uncertainties of operating schedules. Each case should be examined, however, for predictability, and alternative limits may be set that would not be as long as an annual rolling average. One option would be requiring a monthly production limit of zero for the off months, and a higher limit for the operating months. 2. Asphalt manufacturing in northern latitudes when there is no winter demand. Units for which normal operating rates vary greatly (e.g., seasonal processes or batch mode operations) may be allowed longer averaging times. Such sources may be requested to document the historic unpredictability of their operations. Some regions do allow for longer averages where seasonal variations or climatological conditions affect the operation of the source. Any seasonal variations should be examined for predictability, and alternative limits may be set without using a longer averaging period. Again one option would be requiring a low production limit in the winter and higher production levels in the summer. 3. Emergency standby units. This is a good example of a source that would qualify for a long term average. An emergency standby unit is the classic example of unpredictability. 4. Limit on oil usage in a boiler which can accommodate oil and natural gas but is on interruptable status during winter (commercial or institutional boilers). Typically, this type of source may qualify for a longer rolling average. However, each case must be examined for predictability and it is recommended that sources demonstrate a history of unpredictable variations. 5. A printing press which has a surge in demand for Christmas season greeting cards. Generally, this type of source may be able to predict such a seasonal increase in demand. However, if the source is able to demonstrate historical substantial unpredictability, it may be allowed a rolling average. You may also consider requiring a 3 monthly average with higher monthly production/operation in the busy season than in the off season, such that annual emissions are still below the applicability threshold. 6. Quarrying or mining activities which may be interrupted by winter weather. Generally, this type of activity allows for use of a long term rolling average, unless it is determined that such interruptions are predictable. 7. Plants where there may be variations in production due to unpredictable orders or contracts. Proof of historic substantial unpredictability should be provided to justify a long term rolling average. 8. Units used occasionally for testing new products or experimentation. Generally, rolling averages may be allowed for this type of source. But again, this depends on the definition of "occasionally" for an individual plant. 9. Natural gas pipeline compressor stations with load variations depending on the seasonal variations in fuel demand in different parts of the network. Unpredictable seasonable variations due to climatological conditions may very well support use of a long term rolling average. When a determination is made that a rolling average is warranted due to substantial and unpredictable variation in production, the question of enforceability must be addressed. As we have discussed, a 365 day rolling average allows for short term enforceability of production or operation limits while allowing for long term data to be considered. When such a long term average is used, we believe that it is reasonable to require permit conditions which provide for interim limits that ensure compliance and enforceability during the first year. The method used to provide interim limits and the need to do so should be determined on a case by case basis, considering how close the allowable emissions would be to the applicability threshold, and how closely the enforcing agency believes monitoring is warranted for the particular source. You have indicated to us that some sources have suggested taking a requirement in the permit to use available data from the past year to average at the start of operations. This may weigh in favor of allowing a 365 day rolling average. 4 In response to your broader request for additional examples of how to apply the Potential to Emit Policy, we hope to discuss this issue at our annual NSR workshop. Our initial contact with other Regions indicated that the policy is adequate and allows sufficient flexibility for the permitting authority. If you have any questions concerning our response, please contact Clara Poffenberger at FTS 678-8709. cc: Gary McCutchen, NSR Section, AQMD (MD-15) William L. MacDowell, Region V Ron VanMersbergen, Region V Rachel Rinehart, Region V Karen Schapiro, AED Julie Domike, AED William Tyndall, OGC Jeffrey Renton, OGC