Updates to EPA Testing Methods
EPA Published a final rule in August, 2016 which included a clarification to the usage of Method 320 (FTIR) for quantifying VOC for tests required by NSPS Subpart JJJJ. The clarification essentially requires testers to perform and submit all method required QAQC along with the test results. The purpose of this paper is to provide an explanation of what these steps entail, and highlight the difficulties in properly applying them to a VOC summation approach. Ultimately, Alliance Source Testing agrees with EPA that it is no small task to perform these steps. We maintain that the simplest, most cost-effective, and appropriate approach is using one of the traditional FID methods, mirroring the analyses used to set the limits in the subpart.
As many facility owners and source testers are aware, EPA published a rule on August 30, 2016 containing numerous changes to EPA Test Methods. These changes were proposed in September 2015, but due to extensive feedback, they took almost one year to finalize. Although not opposed by Alliance Source Testing, one of the most contested changes was the proposal to remove FTIR from 40 CFR Part 60 subpart JJJJ for the calculation of total VOC by summation. Of the 40 proposed changes, this removal was the only change that received comments along with the request for a public hearing, which was held at EPA in RTP, NC in October 2015. EPA’s final decision and explanation is listed in Federal Register / Vol. 81, No. 168 / Tuesday, August 30, 2016 / Rules and Regulations Section F, pages 59801, 59802, (explanation and response to comments) and 59809 (updated rule). EPA quotes included in this paper are found in this section. The full rule with explanation can be found here:
Failure to comply with testing requirements can result in a demonstration of non-compliance. As always, the burden for demonstrating compliance with EPA regulations falls upon the facility owner and/or operator. In general, when EPA finds that a violation has occurred, the agency has the authority to issue civil and potentially criminal enforcement actions.
Environmental civil liability is strict. Violations can be issued simply based on the existence of non-compliance. It does not take into consideration the responsible party’s knowledge of the regulation. EPA’s civil enforcement options include administrative compliance orders, notice of non-compliances and civil penalties. There are a number of factors that influence the amount of civil penalties, and they can range from $500 to $37,500 per day per violation. If knowing violations of the law are discovered, it can lead to EPA issuing criminal enforcement.
A Flame Ionization Detector… counts every single carbon atom in a reliable and predictable manner.
EPA Methods 25A, 18, ALT-096, and ALT-106 all utilize a flame-ionization-detector (FID) which has been the principal method of total VOC quantification for decades. An FID oxidizes the entire sample and measures the ions which are produced at a fixed ratio with the number of carbons that are oxidized. In other words, all hydrocarbons get burned and then counted on a carbon basis. There are some lower responses (i.e. not 1:1 ratio of carbons) in cases where “heavily oxygenated” carbons are involved, but still the approach counts every single carbon atom in a reliable and predictable manner. The original regulation was written and the limits were determined using FID technology.
When performing Method 25A, a FID analyzer is calibrated using propane, and therefore, the concentration of total VOCs is reported as ppm of propane. The difference between Method 25A and Method 18 with an FID is the inclusion of a column that separates the hydrocarbons into individual species based primarily on molecular weight. This allows for a more detailed analysis with speciated concentrations, but it can also be operated with a propane calibration (thus yielding the same results as Method 25A).
EPA Method 320 is an entirely different type of analysis. FTIR is an optical measurement based on the absorption of infra-red (IR) light passed through the sample. A unique absorption pattern – the spectrum – is produced for each IR-active compound present in the sample. The basis for calibration is a library of reference spectra stored on the computer. Using software, the sample spectra are compared to the reference library, and the concentrations of included compounds are determined. The analyst must identify the compounds of interest as well as the presence of any interfering compounds; for JJJJ, interfering compounds are typically CO2 and water as well as other VOCs. This is a great approach for methane, ethane, formaldehyde, and some other light hydrocarbons. However, a significant issue arises as the hydrocarbon compounds become larger, or there are multiple known and potentially unknown species present. Infra-red absorbance is a result of molecular deformation such as bond stretching, bending, vibrating, etc. The absorption pattern or spectrum that emerges is based on the polarity of the compound, and the way different bonds behave. Note that all bonds behave differently based on what they are attached to, which is why each compound can produce a unique spectrum.
The shortcoming of FTIR is when multiple hydrocarbon species of varying sizes are present in the same matrix. The bonds between two carbons in the middle of a butane or a pentane molecule begin to look very similar. As the number of C-C and C-H bonds increases, the spectrum loses individual resolution of absorbance, and turns into a so called “hydrocarbon blob” where individual species are indistinguishable.
An FTIR manufacturer may claim sub-ppm detection limits for the compounds individually, but that measurement is of a clean gas mixture of the target in an IR inactive gas (such as nitrogen) with no interfering compounds. When preparing an analysis by FTIR, the analyst must build a method that includes references for every individual molecular species that is analyzed, as well as every species that is present as an interferant. Specific analytical regions of each spectrum must be identified to use for the qualification and quantitation of each species. In the case with overlapping hydrocarbons, the best regions overlap, making it difficult, if not impossible to properly attribute the response to the correct species. Further, a VOC that is present but not included in the references can throw the entire set of values into disarray, resulting in significant over-counting or under-counting of molecules. When the results are then converted to a propane basis as required by JJJJ, this error can be significantly magnified.
EPA explicitly states… FTIR QA is complete if all VOC are proven to be accounted for
EPA decided to continue to allow Method 320 summation of VOCs but with two caveats.
- All VOCs must be accounted for.
- Analysis must be supported by Method 320 validation criteria.
The first caveat was a response to the common claim that 95% of the VOCs have been identified, so Method 320 VOC summation is good enough. EPA explicitly states in the decision “We have not stated that 95 percent of the VOC emissions are the target goal for such a list.” Later in the decision explanation EPA expands to say “The other commenter recommended only allowing FTIR if the QA is complete and accurate and if all VOC are proven to be accounted for. We are swayed by this commenter’s support for complete QA/QC of data and stipulation that all VOC are proven to be accounted for.”
A proper analysis would require this validation for every VOC component included
The second caveat involves the validation criteria. EPA states “…for Method 320, report results from sections 8.6.2, 9.0, and 13.0” This requirement calls the entire approach into question. Sections 8.6 and 9.0 are straightforward – they define the QA spike, which is performed once per source or condition for the target compound or surrogate However, section 13.0 describes the validation testing required to prove that there is no bias in the transport and analysis of each individual component. A validation test run is required, which consists of collecting twelve spiked and twelve un-spiked sets of data for the compound of interest. Depending on the sampling rate of the system, this process would take between 60-120 minutes per compound. At a minimum, there are 10-15 compounds of potential VOCs. However, the list is significantly more extensive when accounting for the requirement of proof that 100% of the VOC has been analyzed. A proper analysis would require this validation for every VOC component included in the analysis to verify that, if those compounds were present, the FTIR configuration and analytical methods would be sufficient for analysis, and the sample system would be capable of transporting them without bias.
This is not simply a data validation where a spectroscopist reviews residuals, calculates uncertainties, and performs a spot check of calculated results. This presents a significant cost burden to first purchase all of the VOC gases with tracers and then to perform this entire validation set. It is important to note that once a validation has been performed, it is acceptable to use it on multiple sources of a similar matrix, but the applicability of the “similar matrix” is limited across engine types. Rich Burn and Lean Burn engines have substantially different moisture and hydrocarbon profiles, meaning as far as a hydrocarbon analysis goes, these are not similar matrices. Additionally, different plays produce gas of differing composition, which altars the final VOC content in the post combustion exhaust. At a minimum, different plays would require separate validations, and likely even different stations within the regions could have different profiles and require new validations. Further, makeup air temperature and humidity has an impact on the possible organics, making quite an extensive list. As EPA originally wanted to remove the FTIR option altogether, if reports are submitted without the required validations, it is entirely possible that they will be rejected with regulators citing these two caveats.
In the words of EPA,
…the process of quality assuring emissions data measured by FTIR in accordance with Method 320 or ASTM D6348-03 is not a trivial matter. Calibration checks and matrix spiking of target compounds, including the “most difficult to recover” compound (as required by Method 320), is both challenging and time consuming due to the need to rule out interferences that may be caused by the emissions gas matrix while working to individually quantify each VOC in that matrix. In summation, we do not agree that the use of FTIR for quantification of total VOC is quick, easy or less expensive to conduct when compared with the use of Method 25A.
The original regulation was written and the limits were set using FID technology.
EPA has indicated that they will review the decision to allow EPA method 320 for JJJJ testing when they return for the periodic risk and technology review analysis for the industry sector.
At that point they will decide to remove the option altogether, or continue to allow it with the quality requirements they have put in place. With proven technology readily available, there is little motivation or explanation to continue to include FTIR for this application. Method 25A provides the most complete and correct analysis of total VOCs with well-defined parameters. Generally the best way to comply with a regulation is to use the same analytical method that was used to develop the regulation, unless one exists that is inherently better. In this case, due to the challenges of speciating hydrocarbons by FTIR and the need to prove how well it is working, Method 320 for total VOC is neither a better nor simpler approach, and stands to provide significantly different results.
About the Author
Mr. Jordan Laster holds a BS in Chemical Engineering from Widener University in Chester, PA. He worked and learned for four years alongside Dr. Grant Plummer (Ph.D Duke University, 1985), the principal author of the EPA FTIR Protocol addendum which details QA/QC procedures for Method 320. Mr. Laster has been performing FTIR analysis for air emissions since 2007 and currently oversees a team of seven spectroscopists at Alliance Source Testing performing detailed analysis for a wide-array of customers. Among his primary duties is the promotion of responsible FTIR within air emissions testing.
About Alliance Source Testing
Alliance Source Testing is a privately held company that provides air emissions testing services to industrial, power generation and oil & gas facilities throughout the United States. Alliance currently owns and operates 4 FTIRs in a variety of industrial settings and has offices in Decatur, AL; Little Rock, AR; Lewisville, TX; Pittsburgh, PA and West Chester, PA.