Air Toxics Emissions Inventory Protocol for the Great Lakes States

Quality Assurance/Quality Control Plan

3.0 General Quality Assurance Procedures

This section describes certain general inventory development quality assurance/quality control (QA/QC) procedures that address the following areas:

• Responsibilities of the QA/QC Coordinator;
• Emission factor data quality rating criteria; and
• Documentation.

Section 4.0 presents specific QA/QC checks.

3.1 Responsibilities of QA/QC Coordinator

Identifying a QA/QC coordinator is considered to be a general QA procedure. The QA/QC coordinator is the focal point of the QA/QC process, responsible for ensuring that checks are done (details of these responsibilities are provided in Section 4.0).

The following QA/QC checks must be conducted by the QA/QC Coordinator:

• Ensure that completeness checks described in Section 4.0 are being conducted by appropriate inventory personnel.
• Assess the Project Documentation File once per month. The information, in a representative random sample of the source category files, should be checked for completeness against existing database information.
• Review the means of communication between personnel for consistency and adequacy.
• Review all added, updated, or deleted information for completeness and accuracy.
• Evaluate the communications within the state toxics inventory structure, identify deficiencies, and recommend more effective communications, when applicable (at least once per quarter).

3.2 Air Toxics Emission Factor Quality Rating Criteria

It is anticipated that most air toxics emissions estimates contained in the regional repository will be derived using emission factors (either "generic" or approved source-specific factors). The "quality" of the resulting emission estimates depends to a large extent on the quality of the emission factor used. This section describes the procedures that must be used to rate the quality of these emission factors.

A table summarizes the Factor Information Retrieval System (FIRE) (see Section 2.0 of the Protocol for a brief overview of FIRE) rating criteria, which are based on AP-42 (United States Environmental Protection Agency [U.S. EPA 1985]) factor rating criteria with the addition of a "U" value factors for air toxics. A method for estimating and rating supplemental factors derived from information contained in SPECIATE and AP-42 has been developed for those factors not included in FIRE/GLC (FIRE/GLC is a version of the FIRE system developed specifically for the Great Lakes Commission that includes only the compounds listed in the Protocol). Although it is not recommended that SPECIATE be used to prepare air toxics emissions estimates, in the absence of any other information it may be better to estimate emissions for a given pollutant/source category using these supplemental factors than to provide no estimate at all. Any emissions estimates derived in this manner should be viewed with extreme caution and should not be used to develop compliance programs. SPECIATE data were evaluated for use in developing emission factors for inclusion in FIRE. The absence of an emission factor in FIRE for a given compound, and the ability to generate an emission estimate from a combination of AP-42, SPECIATE, and critical pollutant emission estimates information, further indicate that the resulting factor (and any emission estimate derived from such a factor) is of poor quality/reliability. In any event, the development of rating criteria that include speciation-derived emission factors allows the Great Lakes States to assign data quality ratings to all of the target compound emission factors, including those derived from SPECIATE.

3.2.1 FIRE/GLC Quality Rating

Emission factor data quality ratings used in FIRE and in FIRE/GLC were assigned based on the rating developed by the U.S. EPA procedures (U.S. EPA, 1992c). The FIRE repository system has two quality indicators:

• A rating for the quality of the source test from which the emission factor came (data quality); and
• A rating for the emission factor quality (factor quality).

The data quality rating is used only in the U.S. EPA's FIRE repository system and indicates the quality of the source test or data in the reference. The factor quality rating is used in both the repository and distribution systems and indicates the quality of the emission factor derived from the source test or other reference data. (FIRE/GLC contains a repository and a distribution database. The repository contains all emission factors for a given compound Source Classification Code/ Area and Mobile Source [SCC/AMS] association; the distribution database contains the best estimate.)

The rating for the data quality has a range from A to D. Factor quality has a range from A to E and U. A "U" rating for unratable factor quality was specified by the U.S. EPA to be used for air toxics emission factors based on engineering judgment, source tests with certain deficiencies, lack of supporting documentation, and other reasons. A "U" does not necessarily imply poor quality factors, but may only mean those data have not been reviewed to determine their actual data rating.

3.2.2 Factor Quality Rating for non-FIRE/GLC Sources

The method developed to generate emission factors and emission factor ratings for target compound sources with no emission factors in FIRE/GLC (non-FIRE/GLC sources) uses a combination of the source's AP-42 emission factor data and its SPECIATEspeciation profile rating (based on the specific rating assigned to each SCC/AMS code associated with a given profile) for a source/ device/ process. SPECIATE contains data on the relative amounts of constituents of total organic gases (TOG) and particulate matter (PM), including a rating of the profile quality (Section 4.0 of the Protocol contains a sample profile). A combination of AP-42 and SPECIATE data can be used to derive an emission factor and an emission factor rating.

When using a combination of AP-42 criteria pollutant factor ratings and SPECIATE speciation profile ratings to derive a target compound emission factor, the resulting data quality rating needs to be downgraded due to the problems associated with using speciation profiles to develop air toxics emissions factors and emissions estimates. Speciation profiles, particularly for TOG, were not developed to derive air toxics emission factors, but for other purposes (i.e., photochemical or receptor modeling). If a profile has several toxic constituents, it is difficult to assign different data quality ratings to each of the resulting derived emission factors because the analytical techniques for each compound will typically have varying sensitivities and detection limits.

Furthermore, in many instances, the volatile organic compound (VOC) emission factor must first be converted to a TOG factor by upwardly adjusting VOCs to include methane and/or formaldehyde, both of which may not have been included when the VOC emission factor was originally derived, but are included in the speciation profile (see Section 4.0 of the Protocol). These adjustments add an additional degree of uncertainty to the resulting air toxics emission factor, which should be reflected in the factor's data quality rating. Moreover, additional variability is expected if a factor is derived from a criteria pollutant emission factor and a speciation profile than if it were directly developed from source testing.

For these reasons, the highest rating used for an emission factor derived using a speciation profile is "C" (i.e., the product of an "A"-rated AP-42 criteria pollutant factor and an "A"-rated profile). Most of the target compound emission factors derived from a combination of AP-42 and SPECIATE have been assigned a data quality rating of "U," which is consistent with the quality of these factors (i.e., "Unratable"). A table summarizes the resulting data quality ratings when using different combinations of AP-42 and speciation profile ratings. In most instances, a target compound emission factor derived from a combination of AP-42 and a speciation profile has been assigned either a "Poor" or "Unratable" rating, reflecting the relatively large degree of uncertainty associated with using this approach.

Using this method, the target compound emission factors for non-FIRE/GLC sources can be rated using information contained in AP-42 and SPECIATE without referring to primary reference materials. At the same time, the states can access the reference material for certain key emission factors and refine the initial ratings derived directly from the above-mentioned databases. This allows the Great Lakes States to quickly and cost-effectively rate the target compound emission factors using a rating criterion that is internally consistent with the one currently used to rate criteria pollutant emission factors in AP-42, while affording the states the opportunity to refine these ratings at a later date.

The following discussion describes additional issues regarding the rating of target compound emission factors for area sources and on-road motor vehicle sources.

3.2.3 Area Source-Specific Emission Factor Data Quality Ratings

Some important area source categories can be found in FIRE/GLC; emission factors and corresponding data quality ratings for sources that are not included in FIRE/GLC but that map to SCC codes (e.g., auto refinishing, halogenated solvent cleaning, and dry cleaning) and speciation profiles (speciation profiles are associated with SCC codes in SPECIATE) can be developed using the SPECIATE/AP-42 approach (see the Protocol and Section 3.1 of this appendix). Certain other area source categories do not directly map to an SCC code/speciation profile, but can be represented by one. For example, gasoline marketing emissions may be represented by point source loading rack emissions, and residential oilcombustion may be represented by commercial-institutional external combustion emissions. In these cases, the data quality rating will be slightly lower than that explained previously. Since there is not a direct map between these area source categories and the SCCs, the data quality rating should be downgraded.

For a few cases (such as pesticides and industrial process cooling towers), mapping area source categories to point source categories may not be feasible. In these cases, alternative sources of emission factors need to be sought. If the emission factors are derived based on calculations or engineering judgment, a data quality rating of "U" (consistent with FIRE data quality ranking) is assigned. For example, in the case of cooling towers, a mass balance-based equation has been derived to estimate the toxic emissions from industrial process cooling towers (Radian Corporation, 1991). Therefore, it may be possible to assign an air toxics emission factor for cooling towers based on default cooling tower design specifications. In such a case, a data quality rating of "E" is assigned.

3.2.4 On-Road Mobile Source-Specific Emission Factor Data Quality Ratings

Emissions from on-road motor vehicles are generally obtained from the product of emission factors (expressed in units of mass of pollutant per vehicle mile traveled [VMT]) and vehicle activity data. Air toxics emission factors for on-road motor vehicles should be obtained primarily using the SPECIATE/AP-42 approach (see Section 4.0 of the Protocol). The speciation profiles in SPECIATE were recently supplemented using additional data from the California Air Resources Board (ARB) and the Auto/Oil Air Quality Improvement Program results. The ARB examined the impact of toxic emissions from on-road motor vehicles in the creation of their 1990 Motor Vehicle Toxics Control Plan (ARB, 1990). Specifically, this plan examines the cancer risk posed by motor vehicle toxic emissions. Five compounds, none of which are GLC hazardous air pollutants (HAPs), were found to account for approximately 98% of the cancer cases due to motor vehicle toxic emissions in California. The Auto/Oil Program was an extensive examination of the benefits, costs, and impacts of a variety of automotive fuels, including current generationgasoline, reformulated gasoline, and alternative fuels such as methanol. Exhaust and evaporative speciation data from this program were used to update SPECIATE. In addition, states should also review any new speciation profiles and air toxics emission factors developed by the U.S. EPA Office of Mobile Sources (OMS).

The SPECIATE/AP-42 approach should be used to rate emission factors for on-road motor vehicles. For toxic VOC emissions, the data quality ratings of the criteria pollutant emission factors (from MOBILE5) and the speciation profiles (from SPECIATE) should be taken into account. The MOBILE5 emission factor model is developed and maintained by U.S. EPA (see Section 7.0 of the Protocol), based on hundreds of exhaust and evaporative emissions tests. However, this model may underestimate VOC emissions from on-road motor vehicles. Based on the rating criteria outlined previously, a data quality rating of "B" was assigned to the VOC emission factors estimated by MOBILE5.

3.3 Project Documentation

The ability to independently reproduce an emissions inventory is an important measure of the success of a QA/QC program. Therefore, the information and methods used to generate the inventory must be documented by each state. This documentation must be kept in the Project Documentation File and held for five years by each participating state. Information that must be included in the Project Documentation File is categorized as follows.

Documenting the Inventory as a Whole
• Names of persons involved in developing the inventory, including an organization chart and the responsibilities of each person;
• Location of the Project Documentation File;
• Description of the completeness review;
• Area included (e.g., list of counties);
• Emittants included;
• Time interval of emissions estimates;
• Description of any training provided to emissions inventory staff;
• Reference for any documents prepared when planning the inventory;
• Description of the state permit and inventory data management systems and other files used to identify and locate sources;
• Listings and other data generated during the source identification phase of the project;
• Description of the database used to maintain the emissions data and estimates;
• Identification of those sources/devices/processes that were not included in the inventory and an explanation for not including them; and
• Log of dates for each checkpoint in the inventory process.

Documenting each Source/Device/Process/Emittant Calculation, Including Downstream Control Devices and Release Devices
• Facility ID, facility name, facility owner, and device ID(s) (if a facility source);
• Source, device, and process codes, if used;
• Area/mobile source name (if an area or mobile source);
• Area source cutoff criteria (if an area source);
• Device name;
• Configuration of devices (i.e., how each facility device is physically connected to downstream devices);
• SCC or AMS codes;
• Names of those who prepared the emissions estimate for this source/device/ process;
• Names of those who reviewed the estimate;
• Methods used:
  • Direct measurement,
  • Process simulation software,
  • Mass balance,
  • Engineering calculation,
  • Source-specific emission factor, and
  • Generic emission factor;
• Description of how the method was applied, including a sample calculation if appropriate (present the actual calculation if not documented elsewhere);
• Description of the source of the emission factor or speciation factor if it is not directly obtained from FIRE/GLC, SPECIATE or the latest version of MOBILE, or not directly applicable to the subject process (e.g., partial SCC code);
• Type(s) and source(s) of activity data acquired and used and/or how the activity data were calculated, including any sample calculations as appropriate (present the actual calculation if not documented elsewhere);
• Description of activity data collection, including any survey forms, survey response rate, and method used to reach non-responders;
• Description of how rule penetration and rule effectiveness adjustments were handled;
• Description of updates that are made and why;
• Description of facility and area source reconciliation (if an area source), including a sample calculation as appropriate (present the actual calculation if not documented elsewhere);
• Method to account for all occurrences of this source/device/process (i.e., completeness) (present the adjustment calculations [e.g., scale-up] if not documented elsewhere);
• Description of the reasonableness check(s) used; and
• Description of any subsequent updates, including changes identified on other QA/QC documents.

Included with the Above Documentation
• Calculation worksheets;
• Sample and completed survey forms;
• Lists of facilities, including mailing lists; and
• Any other written data/documentation.

Documenting Each QA/QC Check Made
• Name/description of check;
• Subject of check (e.g., specific facility, specific area source);
• Who made the check;
• Date the check was performed;
• Changes made to data as a result of the check (note: this information should be included in the "description of any subsequent updates" portion of the documentation for the specific source/ device/ process/emittant); and
• Date the change was made.

In addition to the above items, any automated QC checks should generate a printed report listing sufficient information to identify the nature and location of the possible error, and provide room on the report to record who reviewed and corrected the potential error and on what date the correction was entered into the database.

A "Recommendation for Corrective Action" form similar to that shown must be made available to and used by all emissions inventory staff to document any data quality problems not handled in a routine manner elsewhere. These forms must be stored in the Project Documentation File.