Uncertainties and Limitations

Although quality assurance plans are in place to ensure the best results, there are uncertainties and limitations to consider when evaluating the Minnesota air toxics emission inventory. Some limitations are common to air toxics emission inventories in all states and some are specific to Minnesota. For example, in all inventories not all pollutants are included because some emission factors are missing or emission factors are of poor quality, resulting in unrepresentative emission estimates.

There are uncertainties specific to Minnesota. First, the primary concern in the point source inventory is a lack of source-specific emission information from some facilities holding an individual total facility permit. Since chemical species use varies from one facility to the other, the MPCA prefers to collect material usage and composition data from these facilities to estimate emissions. This is particularly important for those facilities using solvents such as in surface coating, solvent cleaning, and printing processes. Facilities with individual total facility permits are usually large, representing a majority of emissions from point sources 

But Minnesota does not have a rule to require these point sources to report air toxics emissions. In order to overcome the deficiency, MPCA staff sent a letter to 477 of these facilities requesting their emissions. A total of 330 facilities responded, including refineries, large utilities, and other manufacturing facilities. The majority of non-responding facilities only had combustion processes/ hot asphalt mixing processes and did not have any source-specific information. Although 5 taconite ore processing facilities didn’t report air toxics emissions, the emission estimation methods and templates were established based on source-specific stack tests and data analyses. Air toxics emissions could be calculated with the throughput data reported to the MCEI. Besides those facilities that were either closed or changed permit status, there are 17 facilities whose emissions could not be estimated without reporting.

Second, the MPCA recognizes that air toxics emission data reported by facilities may be based on an incorrect assumption that all purchased or used materials are emitted. Actually, in many cases, these materials largely react or are consumed in the industrial process. For example, we conducted a special QA/QC study for 4,4'-methylenediphenyl diisocyanate (MDI). The MPCA staff contacted facilities that showed MDI emissions in the 2002 EI, 2005 EI, or that showed MDI emissions in the EPA’s 2002 National Emission Inventory (NEI) to verify the emissions. As a result of this investigation, the state point source MDI emissions dropped from 19.81 tons to 2.93 tons while the emitting sources increased from 20 to 51. 

Third, MPCA staff could not estimate point source air toxic emissions for facilities with certain types of registration permits. There are 450  Option B (aq-ei3-01) and 865  Option D (aq-ei3-05) registration permit criteria emission inventory facilities in Minnesota.  These facilities do not report process level throughput data and have no Source Classification Code (SCC) assigned to them.  Without this information, staff could not estimate air toxics emissions for these facilities. Although the MPCA collected data from some Option D facilities and some other facilities may report to the Toxics Release Inventory, most of these small point sources had to be treated more generally as nonpoint sources in the 2005 emission inventory.  For the 2005 emission inventory, 99% of the Option D facilities (261) that emitted more than 5 tons of VOC reported their air toxics emissions.  Facilities with other types of registration permits cannot as easily provide air toxics emissions data because, unlike the Option D registration permit, their permit does not require tracking of air toxics emissions.

Fourth, uncertainties are introduced due to scarce information on control efficiencies for air toxics.

Fifth, a number of emission factors were developed using detection limits or half of the detection limits when the measurements were lower than detection limits.  This approach tends to over-estimate emissions.

Sixth, activity levels for some nonpoint sources and nonroad equipment were allocated from national totals which might not represent the actual local activities.