innovative environmental solutions,inc.

Technical Memorandum

Date: 	August 11, 2009

To:  	Stephanie Meadows/API

	Matt Todd/API

From:	Tom McGrath

Subject:  Review of Anomalous EPA Method 323 Formaldehyde Emission
Measurements

Recent emission test results from EPA Region 8 have raised questions
regarding the reliability of Method 323 “Measurement of Formaldehyde
Emissions from Natural Gas-Fired Stationary Sources—Acetyl Acetone
Derivitization Method”.  As a result, this Method was not included in
the proposed revisions to the RICE NESHAP (40 CFR 623 Subpart ZZZZ) as
an option for formaldehyde emission measurements.  In the docket for the
proposed rule, document EPA-HQ-OAR-2008-0708-0020.pdf discusses
questionable test results from the Conoco Philips Sunnyside and Argenta
compressor stations.  EPA Region 8 notes that the measured emissions are
much lower than the engine rates published by the engine manufacturer.

Method 323 withdraws a sample from the stack and “bubbles” that
sample through an impinger train (i.e., specialized glassware)
containing water.  Formaldehyde is absorbed in the water, which is
chemically treated and analyzed.  Method 323 is an example of a “wet
test method” and important sampling and analysis parameters for wet
methods include the total sample volume, sample time, sample rate
(volume divided by time), and amount (i.e., weight or mass) of the
species collected.  When reviewing test reports for wet methods, there
are many test parameters, analytical results, and related calculations
that warrant review.  Sample rate, sample volume, and recovered mass of
the subject species (i.e., formaldehyde for Method 323) are key starting
points for the test report review.  

Innovative Environmental Solutions (IES) reviewed the subject emission
test reports and determined that the sample rate during the tests was
approximately an order of magnitude higher than the maximum sampling
rate allowed by the test method (i.e., maximum rate of 0.4 liters per
minute (lpm) versus test rate of approximately 4 lpm).  Because the
sample rate was well above the method specification, it is very likely
sample breakthrough occurred (i.e., formaldehyde was not completely
absorbed in the impinger solution) and the measurements were biased low.

Select pages from the Argenta Compressor Station test report are
included below and show the incorrect sample rate.  Key data are labeled
to facilitate presentation of the data from Run #1 on Unit E007. 

A – is the Run #1 sample volume = 0.227 m3 = 227 liters;

B – is the mass of formaldehyde measured in the Run #1 sample = 0.9078
mg;

C – is the calculated Run #1 stack concentration = 4.10 ppmv; 

D – is the Run #1 emissions reported in the summary of results = 4.10
ppmv; and 

E – lists the sample volume (0.227 m3/227 liters) and the sampling
stop and start time: 9:49 – 8:49 = 60 minutes which is the typical
sample period for Method 323.

A sample rate of 227 liters/60 minutes = 3.8 liters per minute is
calculated based on items A and E.  This is about 10 times the maximum
sample rate allowed by the test method (0.4 lpm).  IES had follow-up
conversations with test personnel to discuss whether data transcriptions
could have occurred in the report – i.e., the sample volume and rate
were different than reported.  Those conversations confirmed the higher
sample rate.

Similar data review and calculation for other test runs indicates that
similar sampling rates were utilized for all test runs on all the tested
units.  Due to this sampling error, these measurements would be expected
to result in formaldehyde emissions with a similar low bias. 

Anomalous EPA Method 323 Measurements Review	August 11, 2009	

  PAGE  6 

  PAGE  1 

 

