Xcel Energy Comments to the Draft Performance Specification 18 

Xcel Energy would like to the United States Environmental Protection
Agency and thank the OAQPS staff for allowing us to contribute and
comment on the Draft Performance Specification 18 specifications and
test procedures.  Please accept our comments to this project.

1.	It is based on our experiences of operating an HCl CEMS for 3 years
that allow me to comment that Dynamic Spiking will not work well as a
substitute for Relative Accuracy Test Audits (RATA) on extractive
systems.  On a recent RATA, we experienced a leak in the flange that
supports the CEMS probe and calibration header that went undiscovered
until we were performing the annual RATA test.  This leak was
undetectable during daily calibration and would not be uncovered without
a stand alone RATA sampling train.  As I understand the plumbing design
of dynamic spiking, it would also not detect this type of leak, which
would allow an inaccurate CEMS to operate in full compliance with PS
18.  The flange that had this leak was impossible to examine or access
without taking the boiler offline and taking the entire CEMS system
apart.  It is not a task that can be part of any routine maintenance
schedule.    

2	We have performed many Method 26A tests on our emissions units and
have found it to be a very durable, well-designed, and accurate sample
train that provides quality data.  We have found it to be a practical
tool to evaluate the performance of our HCl monitors and have been
utilizing it during our annual performance testing.   It provides us
accurate and high quality data that we have compared to our HCL CEMS. 
This Draft PS 18 requires duplicate train samplings.  This is both
problematic and not necessary.  Many sample locations are not designed
for duplicate train testing; they may have only one port for testing. 
The RATA does not need a duplicate train to insure any more quality of
the RM data; the leak check requirements in the method will insure
sample integrity and once you have the HCl in solution, it is stable. 
We employ many of the EPA sample trains in our testing regimen; of any
train, this one least needs duplicate train QA.  The paired train will
not insure any more quality of the RATA testing regimen.  If the RM data
is of poor quality, the RATA will fail.  

3.	Since on-site analysis for Method 26A is impractical, the presumption
is that we would be performing 12 runs for every RATA.  Dual trains
will not provide any more confidence in relative accuracy.  It is just
an added expense, nearly doubling the cost.  A disharmonious result of
the dual train agreement due to a random event requires that run must be
excluded.  If that event occurred during a single train RATA, I would
exclude that one from the RATA population as it would disagree with the
CEMS.  That same random event has twice the chance of occurring when
dual trains are employed.  And it is not excluding 1 out of 12 runs
that is the crux of the matter, but excluding 1 out of the 3 that I am
allowed that makes this a magnified problem.  The in-run disagreement
will not be uncovered until the testing firm is off the stack for a week
or two as analysis is performed offsite.  To summarize, excluding a run
due to dual train disagreement is equivalent to an inaccurate run and
will be one the three runs that will be excluded for the RATA.  And now
there is twice the chance of that happening, for twice the cost.  

4.	The duration of the sample time extended to 1 hour for Method 26 is
not necessary.  If the laboratory quantitation limit is low enough, the
21-minute RATA run-time is more than adequate.  We suggest that PS 18
allows the tester to define the testing time that meets their
quantitation limits requirements with a minimum of a 21 minute run. As
the analysis is going to be performed off-site, it should be expected
the tested facility will ask for more than the minimum samples to
improve the chance for passing the RATA. It can be expected that many
will require 12 runs for any PS18/Method 26A RATA.  If the 1 hour runs
are required, this will require two days of testing for a RATA when it
is preferable to see this testing completed in one day.  It can and
will be as accurate if the sample duration is left to best practices and
a good knowledge of laboratory detection limits. Reaching the
quantitation limit or a minimum of 21 minutes will suffice for Method
26A.  All of these requirements seem to be biased against the Method 26A
train in favor of other technologies as a reference method.  I am unsure
why this apparent bias exists.

5.	P.S. 18 must take into account the possibility of inlet CEMS
monitoring.  Some sources with HCl limits also comply with a 95% control
efficiency.  If continuous monitoring becomes required for units that
use control efficiency, PS 18 must have flexibility to operate under
inlet conditions.  For example, the 2 diameter upstream/ 0.5 diameter
downstream may not be practical for inlet sites.  Some of the span
requirements may need to be addressed as the inlet concentrations can
range from 50 to 5000ppm, as the concentrations can be quite variable at
the inlet. 

6.	I am trying to visualize (From section 8.3.2) “70 percent of the
path within the inner 50 percent of the stack or duct cross-sectional
area.”  Could a drawing or diagram be included with this?

7.	Finally, in section 8.2 it appears there is a word missing between
“accessible” and “where”

Thank you for allowing Xcel Energy to contribute to this process.

Regards,

John F. Buresh

Senior Environmental Analyst

Xcel Energy

