5.19
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Air Quality Planning and Standards
Research Triangle Park, North Carolina 27711
MEMORANDUM:
----------
DATE: May 3, 1985
SUBJECT: NSR Advisory Memorandum #1: TSP PSD Increment Consumption in
North Carolina
FROM: Gary McCutchen, Senior Engineer
New Source Review Section, SIB, CPDD (MD-15)
TO: Mike Trutna, Chief
New Source Review Section
In response to an October 29, 1984, request from Archie Lee, Region IV,
to you regarding whether a change in the North Carolina State implementation
plan consumes increment, I have reviewed a September 20, 1984, letter from -
Bill Johnson, Chief, Air Quality Section, North Carolina Division of
Environmental Management, as well a paper by Robert Wooten and a December
21, 1984, Federal Register notice dealing with this SIP change.
The position taken by North Carolina appears to be this:
(1) Operators of several power plant boilers have indicated that they
are unable to meet the State's original particulate mass emission limits
(ranging from 0.10 to 0.14 pounds per million Btu).
(2) In June 1979, the State granted a 3-year variance, setting interim
limits of 0.20 to 0.25 pounds per million Btu, with the requirement that the
utilities conduct a series of stack tests which would provide data for
setting permanent limits.
(3) The State proposes permanent limits of 0.10 to 0.25 pounds per
million Btu, which are generally lower than the variance limits but higher
than the original limits.
(4) The State alleges that PSD increment is not consumed by these new
permanent limits because there has been no change in actual particulate
matter emissions from these units. North Carolina based this statement on
the following reasons:
(a) The general equipment design is the same as when originally
installed or improved.
(b) The companies have good incentive to maintain their equipment
and operate it properly because of a mass-opacity correlation used to find a
365-day emission total to compare to an annual limit. (Note: EPA proposed
2
no action on the annual limits in the December 21, 1984, Federal Register
because such limits do not protect short-term increment.)
(c) The coal burned before and during the variance is of similar
quality and its quality can be monitored at will.
(d) The new limits are set at levels appropriate to what the
control equipment can meet on a continuous compliance basis if properly
operated and maintained.
North Carolina submitted the Wooten paper to support that there has been
(and would be) no increase in actual emissions from these utilities as a
result of the new limits.
It is EPA's position that a PSD increment-consuming emissions increase
would result from the new emissions limits and that the acceptability of
this increment consumption must be determined before the new limits could be
approved. This position is presented in the attached February 25, 1985,
memorandum from Bozof and Schneeberg to Archie Lee.
There are several additional issues raised by the State's proposal.
First, the State evidently concludes that these utilities cannot meet the
original limits. However, as Wooten points out on page 12 of his paper
(although in a negative tone), these units at one time emitted at rates low
enough to provide support for an NSPS of 0.03 pounds per million Btu. There
is no evidence given in the Wooten paper that any effort was made to improve
operation or maintenance of the control equipment to attempt to at least
meet the original emissions limits, much less the NSPS levels. Acceptance
of the current control levels should not have occurred without a careful and
detailed study of the reasons for the decrease in efficiency, particularly
the possibility of improved maintenance.
Second, the concept of continuous (opacity) versus intermittent (stack
test) compliance is a good one, but much more extensive attempts than North
Carolina's have failed to find a workable mass-to-opacity correlation. The
State has not indicated in the Wooten paper:
(1) What the continuous opacity limit would be,
(2) How this opacity limit would differ from the usual opacity limits
placed on boilers,
(3) How it would protect the short-term increments, since it appears
to be intended for use with the State's annual emissions limits,
(4) How the opacity limits were derived from the mass emissions
limits.
Third, the Wooten paper presents a lengthy statistical approach for
determining the new emissions limits, but what it boils down to is to take
all of the test data (including results which exceeded even the variance)
and add a safety factor to contain a limit that none of the units would
exceed, then call this the new limit. At best, this is an odd approach to
standard setting.
3
Finally, the State contends that actual emissions would not increase
because:
(1) The control equipment is the same. Operators, of course, could do
many things that would decrease the control equipment effectiveness,
including turning the equipment off.
(2) Day-to-day compliance would be encouraged by the opacity limits.
Since there are usually opacity limits on boilers anyway, this argument
doesn't seem appropriate. In addition, the opacity limits referred to are
for an annual limit rather than the short-term limits.
(3) The coal burned is (and presumably will be) of similar quality.
First, the Wooten paper compares a factor alpha, which is pounds sulfur per
million Btu divided by the percent ash. According to the paper, a higher
alpha should improve precipitator performance, all else being equal. Wooten
concludes that there has been "little practical change" in alpha over the
years; I disagree. Roxboro's alpha decreased over 25 percent, from 0.072 to
0.053. The H. F. Zee alpha increased 30 percent from 0.059 to 0.077.
Changes in alpha have, therefore, occurred. In addition, the State
indicates that it can monitor coal quality at will, but implies that it is
not doing so. Therefore, coal quality could change (and has changed)
greatly.
(4) The new emissions limits are more appropriate to what the control
equipment can meet on a continuous compliance basis. I can understand why
this is considered a reason for believing that actual emissions rates have
not increased, at least when comparing emissions under the new limits to
emissions under the variance, but this has little to do with whether
increment is consumed if allowable, rather than actual, emissions limits are
used for PSD increment consumption calculations. Of course, as stated
earlier, there is nothing in the Wooten paper which supports the underlying
assumption that the test data presented by the State represent the best
control that can be achieved by the units tested.
Despite the above-mentioned quirks in the State's submission, EPA has
elected to accept the SIP revision providing North Carolina conducts an
increment consumption demonstration in accordance with the February 25
Bozof/Schneeberg memo. However, telephone conversations with Roger Pfaff,
Region IV, on April 1, 1985, and Lee Daniel, North Carolina, on April 2,
indicated that there were still some issues on this. A meeting with Roger
and Ken Woodard was scheduled for April 9, 1985, to resolve the issues.
At the April 9 meeting, three issues were identified and resolved:
(1) Annual Average Increment Consumption. In calculating "future"
(post-SIP revision) emissions, should actual or allowable capacity and
operating hour values be used? Since these are existing sources with data
on operating hours and capacity utilization (btu per hour) and since there
are no changes or modifications contemplated by this SIP revision which
would affect or influence these operating characteristics, it would seem
reasonable to assume that the same operating patterns would occur in the
future. Thus, "future" emissions calculations would most closely resemble
"actual" future emissions if present operating parameters were assumed to
continue in the future.
4
Although agreeing with the above, we concluded as a group to require
calculation of allowable levels, not actual levels, of emissions. The
definition of actual emissions states that when actual emissions cannot be
determined (and how can we determine actual 1986, 1987, etc., emissions?),
then allowable emissions are to be used. Allowable emissions, of course,
would use maximum (100% capacity and 8760 hours per year unless there were
enforceable constraints on the source. Therefore, we concluded that for
annual emissions (and annual TSP increment consumption), future emissions
would be represented by allowable emissions. It should be noted that this
does represent a departure from the previous Region IV thinking of using
maximum actual operating hours and rates instead of allowable.
Actual emissions would be calculated using the average of all valid
test results. Capacity would be based on the average btu per hour for each
boiler, calculated by taking an average yearly btu heat input and dividing
this yearly total by 8760 hours per year to obtain an average btu per hour
heat input rate. The average yearly btu heat input would be based on two
years of data representative of normal unit operation during the baseline
year. The first two years to be looked at would be the baseline date year
and the preceding year. If no baseline date has been triggered, then
increment consumption calculations are not necessary.
Note that this method of calculating actual emissions provides the same
tons-per-year (tpy) emissions rate that we would obtain if we simply
averages the tpy for two representative years. However, by obtaining an
average per hour "actual" emissions rate, we have an emissions rate that can
be input into the model, since we have resolved the question of what "hours
per year" to use.
The approach outlined above provides the maximum amount of PSD
increment consumption consistent with the Federal regulations and conforms
with the intent of the August 7, 1980, preamble to the PSD rules. Of
course, a source which meets this maximum test of increment consumption is
evaluated on the basis of actual emissions when the next PSD source applies
for a permit, so the use of the more conservative maximum increment
consumption approach does not in the long run artificially limit growth in
an area.
Another interesting sidelight to the question of emissions calculations
is the policy of determining compliance with long-term (annual) National
Ambient Air Quality Standards (NAAQS). A 1983 letter from Sheldon Meyers to
Richard E. Grusnick (copy attached) specifies modeling at maximum capacity,
defined as the allowable emission rate and the "statistical maximum
operating date based on the last three years of operation." The latter
cautions, however, that the "three year" concept can be affected by economic
conditions, and that operating rates should "truly reflect the rates that
can be expected during good economic times." This policy was confirmed in a
March 25, 1985, memorandum from G. T. Helms to Winston A. Smith, with
additional explanation.
As a result of the different objectives on the PSD and NAAQS
calculations maximizing PSD increment consumption and maximizing actual
emissions, respectively, the calculations used to determine "actual"
emissions are quite
5
different. We should make certain that New Source Review personnel remain
aware of these differences and that the correct approach is used.
(2) 24-Hour Increment Consumption Calculations. The February 25
Bozof/Schneeberg memo specifies that the baseline 24-hour boiler emissions
are calculated using actual emissions and "assuming maximum actual operation
over any typical 24-hour period during the two years prior to baseline
triggering or other representative figure." As detailed as this instruction
seems, it still leaves room for several interpretations because of the use
of "typical." Suppose, for example, that the source actually operated at
100% capacity during one 24-hour period over a two-year time span. This
constitutes the maximum actual operation level, but certainly isn't typical.
The problem, of course, is at what frequency does the maximum actual
operating rate become typical? Would five 24-hour periods at 100% capacity
be acceptable as typical? If not, would 10 or 50 or even 100, or would it
take operation at a certain level at least 50% of the time to be called
typical?
We tentatively decided at the April 9 meeting to use the maximum actual
operating rate unless that rate was so unusual as to constitute the
equivalent of circumvention. As a rule of thumb, one would expect to see
such maximums occur at least 5 percent of the total 24-hour operating time
periods (which means nonoperating time periods don't count in making this
determination). This conforms with an earlier Region IV policy
determination (copy attached July - 31, 1981, summary of policy
determinations; Reference #2.18, item #4, in the Region IV New Source
Review: PSD Nonattainment Policy Reference Guide), which specifies use of
the maximum 3-hour and 24-hour emissions rates. The use of the 5% guideline
is intended only to rule out the possibility that a source could
deliberately operate only a few times at very high rates in order to
decrease increment consumption at some future time. Of course, this affects
the amount of increment consumed. The higher the "actual" operating rate
used, the less increment consumed in comparison to the allowable (future)
operating rate. Since we would not be accepting the highest actual rate
without question, our interpretation is at least as stringent as prior
policy.
(3) Use of Test Data. The State of North Carolina has indicated that
the actual emission rate should be based on the highest of the test results
available. Region IV has insisted that the best estimate of emissions rates
is the average of all valid test results. The best rationale North Carolina
could develop was that the high value was "likely" to occur (or to have
occurred) during at least some periods of time.
We felt that the only rationale for use of high values would have to be
based on a parallel with the use of a maximum operating rate. However, the
actual emission rate estimate itself does not appear to have been intended
to be the highest value found when a stack is tested, but to consist of a
value as close as possible to actual emissions. The best estimate of actual
emissions is, of course, an average value, so we decided to continue to
insist on use of the average of all valid test results. The 24-hour actual
emissions level would therefore be based on:
6
(1) An emission factor based on the average of all valid test results,
(2) Maximum 24-hour heat input rates for the 24-hour increment
consumption calculation,
(3) Average 2-year heat input rate (in btu per hour) for the annual
increment consumption rate.
Attachments
cc: NSR Network
R. Bauman
T. Helms
Notebook Entries: 6.19