July 2017

Triangle NoTES

 

July 2017

 

This is a very important newsletter to inform everyone of an immediate change to our procedures. I will discuss the nuts and bolts of why and what we believe was the cause but the critical information is that our Method 25-C analysis has been changed effective immediately.

 

We will no longer report CO2 and CH4 as a standard part of the Method 25-C analysis nor will we be able to add CO to the standard report for the same reasons. The reporting of these compounds seemed like a good QA measure to allow anyone to see the impact of the correction of the NMOC to an air free basis by also showing that impact on the CO2 and CH4 from the same sample as reported by the Method 3-C. The problem is that the procedures which were required to accommodate this additional reporting under the Method 25-C have been found to have an impact on the reported NMOC concentrations in some cases. As the NMOC is the critical “compound” of interest we had to make a change to the Method 25-C to protect this portion of the report.

 

We additionally reported CO, CO2, and CH4 in the Method 25 report, which was the basis for our Method 25-C report and were carried over, since it used the same basic analytical system. We had later dropped the CO reporting from Method 25-C, unless requested, because of the constant reporting of a non-detect. The catalysts on the analyzer are documented by the initial performance tests at the 1% level, which meant the measured concentration of a standard also had to be at that 1% level. Therefore, we could not accurately report the results for any compound above 1% at the analyzer without some qualifier flag. With these calibration range limits there sometimes had to be additional dilutions of the sample in order for us to report the CO2 and CH4 under Method 25. Much larger dilutions were almost always needed for the Method 25-C. Given the requirements of measuring the tank pressures to the nearest mmHg and the limitation of the pressure measuring devices, this also caused the dilution to involve a partial evacuation of the gas in the tank to allow for the added diluent. This 1% maximum concentration limit of CO2 ensures the automated analytical system we use has a set endpoint for turning the back flush valve to prevent the possibility of CO2 being reported as NMOC. All of this was established for the Method 25 analysis prior to the Method 25-C use of that system. Our standard procedures were that we significantly diluted the sample between the Method 3-C and Method 25-C analysis to ensure the calibration limits were met, but due to the results of this recent investigation they are now changed.

 

The problem with those procedures was not evident until we performed some internal research around our 4th of July holiday. We had a couple of earlier projects where the reported concentrations of NMOC seemed out of line with historical reports. As is our policy, we assumed we had made some type of error and tried to document it. When we could not document any error, it became clear something was changing in the sample, but we did not know what the cause may have been for the changes.

 

We know from our work supporting the research of Dr. Morton Barlaz at NC State in the years prior to there being a Method 25-C, there is an established pattern for landfill gases. The experiments took core samples from existing fills and created “average” landfill compositions in chemical digesters which were monitored at set intervals over time. We had proposed what would essentially become the Method 25-C analysis, less the air correction, as a simple and effective way to measure the emissions of the landfill over time. There were a few TO-14 type analysis performed as well, but their cost and relative level of ineffectiveness for this use saw fewer and fewer performed. The scans for the compounds outside of the target list were actually more beneficial than the target lists were. Since the landfills can have anything and everything in them the scans tended to show how many different compounds there actually were. After the use of Method 25-C there were again a number of GCMS scans performed on landfill gas samples from the field. These scans showed the gases emitted from a landfill move more toward the heavier molecular weight compounds as the fill ages. The years of monitoring indicate the environmental conditions will affect both the gas and total NMOC production. Outside of the more recognized seasonal swings, significant changes of barometric pressure and temperature can impact the results from one day to the next. Early in the development we discovered there was a problem with mass dilution of the tank for cleaning, which indicated those higher molecular weight compounds were not being removed from the tanks by the normal type of cleaning for Method 25 samples.

 

What we believed may have been happening was that for whatever reason we were seeing a sample

from an emission band containing a compound or compounds which had a vapor pressure that fell into a range that caused a change. The evacuation of the sample tank during dilution would allow a portion of the compound(s) to return to the vapor state and that dilution factor would increase the concentration more significantly. This was a hard hypothesis to support without data so we used what unused samples we had in house for our research. Many of our clients take a spare sample to be used in case of a problem with their primary sample and the spares we had not yet processed for cleaning gave us the base on which we could test. Due to the smaller number of potential samples we knew we could not generate any data on the percentage of samples with impact, but we were not concerned with that. We were looking for any impact and if we found it we would make changes accordingly. We had a spare from one of the questionable sets of data so we expected a higher probability of seeing an impact.

 

Since the effect we were seeking should not be evident in the initial dilution for shipping or any dilution not requiring evacuation, we started with minimal dilution and progressed with greater and greater dilutions. Since there is a variable sample size, etc. which could impact the calculation of concentrations, we ignored that aspect and worked only from the change of the measured area counts for each sample between dilutions. Using the initial area counts from the least diluted level as a base we could increase the dilution levels, calculate the theoretical area counts from the initial based on this dilution and compare our measured to the theoretical. For example should we have a sample with 10000 area counts and we dilute it by a factor of 2 we would expect to see 5000 area counts in the second analysis. If we were to see anything higher it would indicate the possibility of the effect we were suspecting. For some of the samples there were no significant changes to the area counts, but in a couple there were very significant changes over the range of dilutions. The worst case saw disproportional increases in areas counts ending with almost a tripling of the areas counts from the initial. This shows that not only is there a variable probability of the impact from sample composition to sample composition but also a variable effect of that impact based on the actual level of dilution and how that dilution was achieved. Larger dilutions or more smaller dilutions could theoretically show very different effects on the final concentrations reported.

 

Looking at all of the data very closely we also noticed that the NMOC area counts for the samples with the changes seemed to show an increase in area counts from injection to injection but stabilized after a few injections. This may be due to the loss in pressure from the injection allowing more of the compounds to go to the gas phase until the maximum pressure is reached or some other mechanism. The changes were not sufficient to impact the %RSD limits required in methods, but it did seem to be related.

Since all we needed to make adjustments to our procedures was a single sample with significant changes that worst case was sufficient, but with the second less significantly changed sample there was no question that we needed to immediately adjust the procedures concerning sample dilution.

 

I was somewhat shocked that almost exactly 20 years ago I had presented a paper at the AWMA “Measurement of Toxic and Related Air Pollutants” detailing the known problems with Method 25-C, including that with high molecular weight compounds and the difficulty in cleanup. At the time we assumed the tank clean up issue would only result in a slight negative bias to the sample, but now it seems it can cause either a positive or a negative bias depending on the situation. At the time the folks at the EMC did not think it was a critical issue since there was a mathematical average done for the set of probe samples and the compositing in the field had the potential to be at least as variable. I also added my suggestion to hybridize Method 25 with Method 25-C to make the field sampling a little easier. This hybridization would use an ice water bath for the condensate trap to knock out those higher molecular weight compounds which seem to be the cause of the problems we have seen. Dry ice for the trap would probably be overkill given the amount of data generated over the years without any traps and the greater ease of sampling with an ice water bath trap in the more isolated areas would be a clear benefit as well.

 

I have previously noted the problem concerning the use of different column sets. This played a large part in not recognizing this problem earlier because most differences noted were between laboratories and the more simple explanation was probable column differences. Given the column issue never seems to completely go away and can have a significant negative bias, this is still a viable concern. I believe the last time I heard of a major laboratory having to make the change to the columns required in Method 25-C was 8-9 years ago, which was over 10 years after the impact of different columns were known. Making sure the correct columns are used is difficult since there are limits on the sources of analyzers and most of the analyzers are set up for the alternative columns. The actual conversion from the alternative columns can take months. I know of one manufacturer which took some 3-4 months to make the conversion for a single analyzer to meet the Method 25/25-C requirements.

 

With the lack of a large data set and without much more research we cannot draw any firm conclusions about the causes or effects for any particular samples, but we have enough information to warrant the changes we are making to try to mitigate any impacts on the reported concentrations. From this point forward, the Method 3-C analysis will provide the concentrations of the O2, N2, CO2, and CH4 in this type of samples and the Method 25-C will only report the NMOC as corrected to an air free basis. There has still not been an official change to adjustment formula in the CFR to use either O2 or N2 for the adjustment under the method so we are still relying on the instructions by the EMC on the subject from nearly a decade ago, which indicated the formula in the method would eventually be changed accordingly.

 

If anyone has any questions or would like a more detailed explanation, please do not hesitate to contact me. I view this as an extremely important issue for a lot of reasons.

 

Wayne Stollings

Triangle Environmental Services, Inc.

Wstollings@aol.com

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