I would like to use this time to discuss some issues of which I have been made aware and which were found to be related to information provided by a citizens action group. In my investigation I found there were several claims that when combined sounded as if there were significant problems with Method 25 relating to the asphalt industry.
One of the issues was the higher recovery temperature of samples, which they took as possibly creating a negative bias based on information passed to them by former asphalt industry employees. While there was such an issue in recent memory, it has been resolved and there is honestly no indication that it would have any significant historical bias effect on the results.
Another issue was largely based on pictures of sample consoles that they suspected may not have met the specifications of the method. While it has not been documented as being used in any of the asphalt plant monitoring programs, there is a high probability of such variations in some of the sampling systems being used.
Section 6.1.2 of Method 25 requires the use of a 25mm (15/16) Gelman filter holder with a 303 SS body and a 316 SS support screen for the filter. The filter requirements are specified in section 7.1.3 and are meant to be comparable to the filter used in Method 5. This requirement is to prevent the bias from any carbon containing particulate matter. As a result, Method 25 defines an NMOC as whatever carbon contained in a gas stream of 121 C that passes through the specified 25 mm filter contained within the specified holder and into the condensate trap. Anything not passing through said trap configuration at the specified gas stream temperature is not classified as an NMOC. Thus, the critical aspects are the filters, the configuration of the holder and the temperature of the gas stream. Any potential change in these variables could change the definition of NMOC between sources. There is a second definition of NMOC as defined in a flammable gas stream, which eliminates the heated aspect, but these types of sources would not be expected to have much in the way of particulates with which to create a bias.
While there appears to be a variation from the filter holder specified in the method by some sample consoles, I could not confirm or deny whether the trains which use the modified filters have been used in the monitoring of asphalt plants. There are still some Nutech, Clean Air Engineering, and various other custom made consoles still in use around the country. Some of those consoles could have been modified away from the method requirements even though the original designs were intended to follow the method as written. This is something the state regulatory agencies are going to have to investigate for themselves and probably on a case by case process. I suspect such an investigation will be followed by some input at the federal level and a change from that point forward in either the design of the console or as an approved alternative design. I would not expect this to cause a systemic bias, but there is really no data available to me either way. I know there were questions concerning variations of the specified 25 mm filters before the recent standardization to the Method 5 equivalent as discussed in prior newsletters (http://www.triangleenvsvcs.com/newsletters/2017/3/28/march-2017 ), so there is a chance the design could also have an impact in some cases. Since the specified ASTM methodology by which the Method 5 filter was initially determined has been withdrawn, doing a direct comparison between the different designs of filters may be even more problematic in the future.
To better explain the problem with filter comparison I would like to use data on the various filters from Advantec MFS, which also manufactures a direct equivalent filter holder to the Gelman specified in the method. All of the various filters are rated to retain 99.95% 0.30 micron DOP aerosol, but there is a wider variation for the nominal particle retention sizes. The two grades of flat glass fiber filter show 0.60 micron and 0.50 micron respectively, which may be even closer than this appears, but the glass fiber thimble grade show a 1.0 micron rating. The difference in particle size retention from the same manufacturer while also meeting the aerosol requirement sets up the question as to what effects this particle size difference could have on NMOC measurements for different sources. There is an even greater difference when you look at the PTFE filter media where a particle retention size of 3 or 4 micron is listed for the same 99.95% aerosol retention. We have seen an effect from differing particulate size selection in certain sources and this indicates that even with a clear aerosol comparison there still may be significant differences in the retention of particles for those filters and thus the different sources.
The situation may not be impacted by just these issues, but given part of this concern could go back at least a decade or more, the changes in the interim would make it nearly impossible to reach such a clear determination of a specific impact. In fact, even if there was an intensive review that showed some level of possible bias, that would not necessarily hold true by 2017 because of those interim changes.
For example, prior to 2010 there were Byron Hydrocarbon analyzers in operation performing Method 25 analyses on the canister fraction and the recovered trap fraction. I believe the latest Byron model used was designed to analyze for multiple compounds with multiple runs. It has been a few years since I have even seen one so my memory is all I have on this. Such a system would tend to meet the method requirements since technically all that is required to be reported in the tank fraction is the NMOC and for the trap fraction only the CO2 is required to be reported. We on the other hand analyzed both fractions for CO, CO2, CH4, and NMOC, which allowed us to more accurately monitor the various types of sources.
The issue of probe and filter temperature measurements in the sample console was one of the first major issues we uncovered after we purchased the equipment of Research Triangle Laboratories over two decades ago. There were complaints about the Nutech design in relation to the length of time it took for the probe and filter box to heat the gas stream to the required temperature in the method. To solve those complaints there was a change made to the design to add an additional thermocouple were an average between the gas stream temperature, the one specified in the method, and the air temperature inside the filter box was registered on the read out. Since the filter box air temperature reaches the set point much faster than does the gas stream being heated by that box, the average of the two showed compliance with the method requirements much sooner than was actually the case. This meant that the gas stream could slowly heat up to the correct temperature or possibly not. There was really never any way to know because the readings were adulterated. As I remember we saw improved consistency with all of the sample consoles after this problem was corrected. The sample consoles we constructed all measured only the gas stream temperatures as the method required. It was speculated that if the temperature of the filter increased slowly from run to run that might cause a change in the reported NMOC from some of the sample from prior runs passing through the filter as it heated to the correct temperature instead of passing through during the initial sampling. There is such a possibility just as there is a similar possibility of the gas stream never making it to the correct temperature in some sources and thus never being measured in the same manner as other sources both similar and different to it.
I know from my time at Research Triangle Laboratories, almost three decades ago, that not all of the modifications to the method are ever mentioned in a report or any other documentation, such as the addition of a second trap within the recovery system to allow the NMOC from the original trap to be transferred to the second trap during the CO2 flush. Even the small variations such as this can cause problems if you do not compare the results to the standard method specifications. The cumulative effect of small changes is exactly what I fought in setting up our systems to get back to what the method stated, but with better QA/QC where ever possible. Every one of the changes were discussed with someone at the EMC and compared to an unchanged system. The “new” and better column sets looked great until they were compared to the original set and failed. The cleaning of the trap and doing a QC analysis at a higher temperature than the recovery sounded great and it was. The difference in the two was the confirmation of the benefit was evident in the latter case and the failure of the columns was documented in the former through that comparison process. Other modifications, such as the use of dual traps during sampling, which were an approved modification are also not specifically mentioned in the comments on the report. The report generally does have the both of the trap identifications listed on the COC and on the data sheets for the individual samples, but it does require them to be sought rather than being brought to the attention to the reader of the report.
While it seems there is nothing that would force any of the prior data to be invalidated, as far as my limited investigation has found, it should create a higher level of review for future compliance testing. There have been some changes over the period in question which complicates any data mining for a single point comparison. There have been issues with trap recovery temperatures, which I believe have also been resolved and will be correctly monitored in the future. The console designs have some clear modifications based on the information available and may have more significant variations once a complete review is completed on them. The known modification to the filter and housing would not be expected to be a problem requiring a repeat of prior testing, which was also the determination when the Nutech console problem was initially discovered, but if other issues are uncovered the combination of them all may change that view. Nothing I have found is limited to only asphalt plants, so other sources such as printing and coating operations may also be impacted. Nor were these issues contained within any single state. These issues are applicable in all of the states just as it includes all of the industries which use Method 25 for compliance testing. Is it a critical issue from a science perspective? Probably not, but there may be additional problems with some sources. Is it a problem for a proscriptive method such as Method 25? Yes, it would give a reason for any group to question the validity of the results especially if any other modifications are uncovered in the investigation process.
Prior versions of the newsletters can be found on our website here:
Triangle Environmental Services, Inc.
P.O. Box 13294 122 US Hwy 70 E
Research Triangle Park, NC 27709 Hillsborough, NC 27278
(919) 361-2890 (800) 367-4862 Fax: (919) 361-3474