Air Quality in the Barnett Shale - Part 28: Why Plot 2 used Carbonyl Disulfide

I was a bit curious as to why the FWLN Report sent to the FWISD included one model for carbon disulfide (Plot 1) and another model for carbonyl sulfide (Plot 2).  Everything up to this point had focused on the health concerns for carbon disulfide.

So I looked closer at the FWLN findings:

...and:

And then I looked at the statements printed on a green background in the report:

Okay...so it is all about the carbon disulfide.  So why produce Plot 2 for carbonyl sulfide?

Well duh!  Carbon disulfide was not detected by TO-15 in 46 out of 48 samples (see post) and ASTM Method D5504-08 detected carbon disulfide but had to qualify it as an artifact (see post).

But sample QLA5DC-1 did not have that inconvient qualifier:

So as I'm reading this and looking at the data, my mind is forming the question of; if there is no carbon disulfide detected in the BSEEC samples collected from different sites over different days, and carbon disulfide is the contaminant of concern, why spring carbonyl sulfide into the mix and continue to direct attention to carbon disulfide?

Doh!  Then it hit me.  FWLN is looking at carbonyl sulfide as a "byproduct" of carbon disulfide.  In other words, if you find carbonyl sulfide there must have been - or is - carbon disulfide!  Very astute FWLN!  Only you probably didn't figure on me checking into this, did you?

Yes, indeed, carbonyl sulfide is a "byproduct" of carbon disulfide.  It is produced when carbon disulfide is purposely manufactured.  Lets see what the EPA has to say on this:

You know what else the EPA has to say about carbonyl sulfide?

In the United States, the reported concentration of carbonyl sulfide in the air in rural areas was 0.27-0.80 microgram/m3.  Levels were 1.17 micrograms/m3 in Philadelphia, PA; 1.21 micrograms/m3 in Wallops Island, VA; and 1.37 micrograms/m3 in Lawton, OK.  Reported concentrations of carbonyl sulfide over salt marshes, which are major natural sources, range from about 60 to 180 micrograms/m3, and 14 to 19 micrograms/m3 over the ocean (HSDB 1994).

Well if I am going to be objective here, I need to follow this lead....

Okay...so finally we have something that indicates that we may have an issue with a contaminate in the Barnett Shale area above normally expected background.  And even if you subtract the maximum amount attributed to the artifact, in sample QLA5DC-1 you would still have 105 ug/m3.  That's more than comes off a salt marsh based on the EPA's research.

So lets work with this.  But first we need to make a few points clear.  First the presence of carbonyl sulfide in the ambient air samples collected has nothing to do with carbon disulfide.  And second, the non-detection of carbon disulfide in the BSEEC report using method TO-15 indicates that carbon disulfide is not present.

So the question I can reasonably ask now is carbonyl sulfide a concern? 

Now we are back to the one sample - low n's - issue.  However, carbonyl sulfide was detected above the artifact concentration of 25 ug/m3 in 6 of the 18 samples indicating a possible concentration above the levels found in an urban environment.  So lets assume that 130 ug/m3 (53 ppbv) is a valid number for carbonyl sulfide in the ambient air collected by BSEEC.

What does the BSEEC Report have to say about this:

Looks like the BSEEC folks got their units mixed up. This means that I need to check to see if the AMCV they report is in ppbv or ug/m3.  And guess what I find when I check?  There is no AMCV for carbonyl sulfide.  So where did "1800 ppbv" come from?

Well if I can find the acute ReV that would be the AMCV.  Search....nothing.  What about the RfC?  Checking the IRIS website....nothing, in fact they state:

The health effects data for carbonyl sulfide were reviewed by the U.S. EPA RfD/RfC Work Group and determined to be inadequate for the derivation of an inhalation RfC. The verification status for this chemical is currently not verifiable. No EPA documentation presently exists for this chemical.

Okay...so where did the BSEEC folks get a short-term AMCV of 1800 ppbv?  Well as of right now I have no idea.  I did find a short term ESL value that the TCEQ has derived in a 2008 memo:

So even at 130 ug/m3 (53 ppbv) the amount detected is well below the health-based ESL concentration of 540 ppbv which is 70% of the AMCV.....540 / 0.3...that equals 1800...hey...that's the AMCV the BSEEC folks reported!  

Okay, that gives us a value we can defend as health-based, assuming TCEQs POD and UFs are correct...they seem well supported.  And if the carbonyl sulfide ESL is listed for odor at 55 ppbm, it makes sense that the health-based threshold would be higher:

The short-term ESL is the lowest value of acute odor-, vegetation- and health-based ESLs. (Factsheet)

And at the highest carbonyl sulfide concentration detected - 53 ppbv - that amount is 10 times lower than the health-based ESL of 540 ppbv and more than 33 times lower than the ambient air concentration - the true health-based threshold - of 1800 ppbv.

Y'all double check my math (remember math makes my head hurt) to make sure I'm seeing this right....

Lab reports 53 ppbv...TCEQ states health-based ESL is 540 ppbv...at 540 ppbv the AMCV would be 1800 ppbv...okay...so the carbonyl sulfide in the Barnett Shale samples is well below health based levels.

So what does Dr. Sattler and the FWLN have to say about the carbonyl sulfide in their Report?

Huh?  How could the oil & gas source that produced sample QLA5DC-1 produce an emission rate that produces a model that that predicts concentrations one mile from the source that are "6 times greater than the health benchmark for carbonyl sulfide?"  At 1800 ppbv that would mean one mile out you would have concentrations at 10,800 ppbv?  Even if she incorrectly used the health-based ESL of 540 ppbv, that would show concentrations at 3,240 ppbv one mile from the source.

How is that physically even possible?

Maybe I'm missing something here, how can directly downwind from the source  - 50 -75 feet away - we find an actual concentration 130 ppbv but are able to get a concentration significantly higher further from the source?  How can the calculated emission rate from this actual value extend downwind one mile from the source at a concentration 6 times greater than what produced it at 50-75 feet away?  The same emission rate is being used, correct?  Doesn't Gaussian dispersion show the highest downwind concentrations nearest the source?

Wait a minute...Dr. Sattler is calculating an air concentration in and around the source (red area on her plot) of 655 ug/m3.  That's six times higher than the actual concentration detected within the red colored area around the source.  Now I'm really confused.  I mean the emission rate is staying constant, correct?  And the downwind concentration used to "back in" to obtain the emission rate is 130 ug/m3.  So with that "backed-in" emission rate, you would calculate 130 ug/m3 for the red area.  Yet in Plot 2, that area has a concentration of  655 ug/m3:

Okay...so lets assume Dr. Sattler's maximum 1 hour concentration for carbonyl sulfide is possible.  Would that maximum level pose a health-based concern?

• The highest concentration Dr. Sattler reports in Figure 1 is 655 ug/m3 or 267 ppbv.
• The maximum concentration of 267 ppbv is less than the ESL of  540 ppbv.
• 267 ppbv is less than one half the amount of the health based ESL
• 267 ppbv is less than six times the health based ambient air concentration of the AMCV.

So why would Dr. Sattler and the FWLN insist that the concentrations one mile from the source could be as high as "6 times the health benchmark for carbonyl sulfide?"  Well lets look at Dr. Sattler's Report on Plot 2 to see what's up about this:

The 1-hour Texas Commission on Environmental Quality (TCEQ) Effects Screening Level (ESL) for carbonyl sulfide is 135 µg/m3; one of the sample concentrations exceeded this level.

.... ah...that's why!  Dr. Sattler is using an odor ESL threshold for the "health benchmark" that is being exceeded:

...and if you will recall, "The short-term ESL is the lowest value of acute odor-, vegetation- and health-based ESLs."

So even if Dr. Sattler's model is correct, the carbonyl sulfide in Plot 2 does not pose and adverse health risk.

And if that is true...and the fact that 46 our of 48 BSEEC TO-15 samples showed non-detect for carbon disulfide, and the two samples that were positive were less than 0.3 ppbv... then there is no health based concern for carbon disulfide or for carbonyl sulfide and the requirement to have a one-mile setback based on the health concerns for these two contaminants is not warranted or substantiated by any of the data provided in the FWLN or BSEEC or ERG reports.

And just so we are clear on this.  The need to provide - or not provide - a setback for O&G operations is not being argued here.  The basis for a one mile setback recommendation due to health-based concerns regarding carbon disulfide and/or carbonyl sulfide is.

And that's all I can really say about this....

Next Post: Air Quality in the Barnett Shale - Part 29: Where should we go from here?

.

Air Quality in the Barnett Shale - Part 27: Plot 2; Artifact or the real deal?

Well apparently my attempt at humor and self-depreciation as a means to entertain and inform on the complex - and really dry - world of health, air monitoring, and air dispersion modeling made me look like a fool in the eyes of the FWLN.

So today I am writing my blog while dressed in a long sleeve shirt - with tie - slacks, and my recently polished dress shoes (black to match my belt according to my wife).  And to complete this outfit of visual competency and authority, I have put on my white lab coat, the one with my name embroidered over the pocket!

Hopefully this will get my street cred' back.  You know, nothing says "scientist" like a white lab coat!  In fact, look at what these two guys at Columbia Analytical Services are wearing:

You know what else you can find on their website?  This:

Do you know what "COS" and "CS2" is?  Well since I am wearing a white lab coat, I do.

• COS = Carbonyl sulfide
• CS2 - Carbon disulfide

Do those two chemicals sound familiar to you?  Well they do to me:

This post today is only going to focus on Plot 2 of the FWLN report.  Plot 2 is based on the concentration of carbonyl sulfide found in sample QLA5DC-1.  

According to the white lab coats at Columbia Analytical Services, there are two factors in play when a Tedlar bag is used to collect a gas sample.

1. Concentration range is from ppm to percent (1 percent = 10,000 ppm)
2. Artifacts may be produced which include carbonyl sulfide and carbon disulfide

If you will recall from my last post, artifact formation is a concern and was noted by the laboratory.  You can read more about issues with Tedlar bags at this website.  And when you look at the BSEEC report you will see that they used Tedlar bags:

...and artifacts were acknowledged:

...and the concentrations were reported as ppbv (which are 1000 times smaller than a ppm):

Now all of this give me pause - and when I am wearing my white lab coat - my scientist-senses start to tingle (analogous to Spider-Man's spider-sense telling him that something is wrong) - or it could just be my cell phone set on vibrate.  Anyway....

So what's the issue?  Seems that Plot 2 which was used in the FWLN shows "multiples [that] were 6 times the health benchmark for carbonyl sulfide."  

And how did they arrive at this "6 times" number?  Using a "backed in" emission rate calculated by Dr. Sattler as described in her 3/7/11 draft report titled "Dispersion Modeling of Carbonyl Sulfide Emissions from a Natural Gas Compressor Station near Lake Arlington."  (link not available).

In this report Dr. Sattler describes the sample concentrations for carbonyl sulfide she used as follows:

...which corresponds to these samples in the BEECE report:

...which are associated with these two laboratory reports:

Upwind Sample

Downwind Sample

Now there are two factors seen above that get my spider-senses....err my white lab coat senses buzzing.  The first is that there is no data qualifier "+" for the carbonyl sulfide quantification as there is for carbon disulfide.

Note 3/14/11: Spoke with the lab this morning.  the reason that there is no qualifier is that they set the highest number attributable to an artifact at 25 ppb.  This qualifier number is dependent on the Tedlar bag used.  So the quantity of 130 ppbv may be only as high as 105 ppbv if the bag did produce artifacts of carbonyl sulfide.  By the way, I just noticed that the laboratory that performed the analysis was the same lab with the white lab coat guys that produced the table I used above.  It's a small world.

Was this an oversight or were they able to exclude the possibility of an artifact in this sample?  The reason that I question this is that the possibility of the carbonyl sulfide being an artifact is noted in samples:

QA52DW-1, EMR3DE-1, EMR3DW-1, CAP4UC-1, CAP4DC2-1, CAP4DC2-DUP1, DLB6U-1, DLB6DE-1, DHH7U-1, CLP9UC-1, XRA4UE-1, and XRA4DE-1

In other words, out of 18 samples 12 identify the carbonyl sulfide as a "possible Tedlar bag artifact."  I will be contacting the laboratory next week to see if I can find out if this was in error.

The second reason I question the validity of these carbonyl sulfide results is that even though the laboratory may feel quite confident in the numbers they have reported for ASTM Method D5504-08 for sulfur compounds:

...while wearing my white lab coat, I went and read the ASTM method.  

...and here is what I found out:

...so we have to assume that both the sampler and the laboratory understood this and took the proper precautions.  Okay, so lets assume they did.  What next?

...well the test is for the gas itself...may not be appropriate for ambient air samples...still...rang is 0.01 ppm...what's that in ppb?....if 1000 ppb = 1 ppm, the range is down to 10 ppb...okay that means the 130 ppb is 130/1000 = 0.13 ppm....kind of close to the method's limit....okay....odd though, the lab's MRL states they can "confidently" go down to 0.0078 ppm.

So once again we are looking at the identification and quantification of a contaminant of concern that is questionable.  I am unsure why ASTM  D5504-08 was utilized in the first place.  Most likely it was to look for sulfur contaminants that are common to natural gas and are not picked up by EPA Method TO-15.  In any case, the use of a Tedlar bag was possibly the wrong choice if carbon disulfide and carbonyl sulfide were to be reported.

Hey....but wait a second....isn't TO-15 appropriate for carbon disulfide and carbonyl sulfide?

And didn't BSEEC run a TO-15 analysis on sample QLA5DC-1 from which Dr. Sattler obtained the quantity for her backed in emission rate calculation for her carbonyl sulfide model shown in plot 2?

...which means there should be a lab report for this sample showing the quantity of carbon disulfide and carbonyl sulfide in the same air on the same day.  And this sample is the correct method and does not have any footnotes calling into question its validity, such as a "possible artifact."  So looking at the lab results for sample QLA5DC-1 we see:

Doh!  They didn't analyze for carbonyl sulfide!  Had they tested for carbonyl sulfide with TO-15 it would have been much more definitive and I could have ended this post right here and now stating that the carbonyl sulfide reported was indeed an artifact. If only it would be that easy....

Still, there is good evidence to think it is - most likely - an artifact.  

The TO-15 method has a detection limit for carbon disulfide of 0.236 ppbv.  The TO-15 results for 46 one hour samples show non-detect and two samples show carbon disulfide at less than 0.61 ppbv.  Now compare this to the detection of carbon disulfide with ASTM Method D5504-08 which found carbon disulfide in 12 out of 21samples with a high concentration of 11 ppbv.

And with my white lab coat on - complete with embroidered name over the pocket - I am able to look at this and boldly state that Plot 2 in the FWLN report is based on a positive identification of carbonyl sulfide and a quantitative amount that is suspect.

And if Peter Parker (aka: Spider-Man) - who himself is a scientist - was looking at this, he too would see not a criminal number, but a number that is suspect.  Call it his spidey-sense, objectivity, or just plain-ol' scientific common knowledge, but that's what any good scientist would conclude - even one that uses self-depreciation and comic book characters to try and drive home his point.

The thing is, we scientists are only as good as the numbers we use and the place we get our data from.  And because of this, the air dispersion model that generated Plot 2 that was based on the results from sample QLA5DC-1, which was derived using ASTM Method D5504-08 from samples collected in a Tedlar bag which - according to other scientists wearing white lab coats - are designed for samples in the ppm range and are also known to produce artifacts of COS and CS2, which - while wearing my white lab coat - I have identified as the two culprits of concern in the FWLN report regarding the need for set-backs; carbonyl sulfide and carbon disulfide, makes the model's predictions of "multiples [that] were 6 times the health benchmark for carbonyl sulfide" suspect.

So what are we left with now?  

If the FWLN report is based on Plot 1 for carbon disulfide which was derived from one single sample collected on one single day (see post) and Plot 2 is based on carbonyl sulfide derived from using ASTM Method D5504-08, then nothing definitive, conclusive, alarming, substantiating, or realistic regarding set-back distance can be ascertained.

Next Post: Air Quality in the Barnett Shale - Part 28: Why Plot 2 used Carbonyl Disulfide