Environmental Protection Technology Series
                     tVALUATION  OF  THE
CORRELATION SPECTROMETER  AS AN
                     AREA  S02  MONITOR
                    Environmental Sciences Research Laboratory
                        Office of Research and Development
                        U.S. Environmental Protection Agency
                  Research Triangle Park, North Carolina  27711

-------
                     RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development,
U.S. Environmental Protection Agency, have been grouped into
five series.  These five broad categories were established to
facilitate further development and application of environmental
technology.  Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in
related fields.  The five series are:

          1.  Environmental Health Effects Research
          2.  Environmental Protection Technology
          3.  Ecological Research
          4.  Environmental Monitoring
          5.  Socioeconomic Environmental Studies

This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series.  This series describes research performed to
develop and demonstrate instrumentation, equipment and methodology
to repair or prevent environmental degradation from point and non-
point sources of pollution.  This work provides the new or improved
technology required for the control and treatment of pollution
sources to meet environmental quality standards.
This document is available to the public through the National
Technical Information Service, Springfield, Virginia  22161.

-------
                                           EPA-600/2-75-077
                                           October  1975
           EVALUATION OF THE CORRELATION
        SPECTROMETER AS AN AREA S02 MONITOR
                         by
                   R.  B.  Sperling
          Environmental  Measurements,  Inc.
               215 Leidesdorff Street
          San Francisco,  California 94111
                     68-02-1773
                  Project Officer

                    H.  M. Barnes
Emissions Measurement and Characterization Division
    Environmental  Sciences Research Laboratory
   Research Triangle Park, North Carolina  27711
       U.  S.  ENVIRONMENTAL PROTECTION AGENCY
         OFFICE OF RESEARCH AND DEVELOPMENT
    ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
   RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711

-------

-------
CD
CONTENTS
   Section 1  -  INTRODUCTION	1-1




               BACKGROUND	1-1



               PURPOSE	1-2



   Section 2  -  SUMMARY	2-1




               EQUIPMENT	2-1



               MEASUREMENTS  	 2-1



               DATA  PROCESSING	2-1



               COMPARISON  OF RESULTS   	 2-2



   Section 3  -  CONCLUSIONS	3-1



               COSPEC INTERNAL CONSISTENCY   	 3-1



               COSPEC RELATIVE ACCURACY  	 3-2



               COSPEC PRECISION  	 3-2



               WIND  MEASUREMENT  ERRORS   	 3-3



   Section 4  -  RECOMMENDATIONS   	 4-1



               WIND  MEASUREMENT  ACCURACY  	 4-2




               COSPEC CALIBRATION  	 4-3



               CROSS STACK MONITOR RESULTS   	 4-4



               STATIONARY  COSPEC DATA  	 4-4



               COSPEC/SKY  ORIENTATION  .  .	4-5



               SAMPLING PROTOCOLS  	 4-5



               PLUME GEOMETRY	4-6



               ARCHIVAL COSPEC DATA	4-7



   Section 5  -  EQUIPMENT AND FACILITIES	5-1



               AIR QUALITY MOVING LABORATORY   	 5-1



               COSPECS   .	5-2
                                111

-------

-------

-------
                             TABLES






   I.  Barringer Research COSPEC III Specifications 	 5-3



  II.  COSPEC Calibration and Range 	 5-4



 III.  Sample Data Listing	7-4



  IV.  Sample Mass Flux Calculation   	7-7



   V.  Remote Sensing Measurement Results 	 8-4



  VI.  In-Stack Measurement Results 	 9-4



 VII.  Comparison of Results,  "20-Minute Averages"	10-3




VIII.  Comparison of Results,  "60-Minute Averages" .  .,. .  .  10-7



  IX.  Comparison of Results,  "Daily Averages" 	  10-8



   X.  Confidence Limits 	  10-14



  XI.  S02 Mass Flux Variations	•	10-15



 XII.  COSPEC-Pibal S02 Mass Flux Variations	10-17



XIII.  COSPEC-PibaL Precision	10-20
                            VI

-------
                         Section 1


                       INTRODUCTION
Environmental Measurements, Inc. (EMI) conducted a field
evaluation of the Barringer Correlation Spectrometer
(COSPEC)* at a pulverized coal fired power plant in the
southwestern United States.  The work was funded by the
Environmental Protection Agency (EPA) under Contract No.
68-02-1773. This contract was designed to contribute to the
EPA's investigation of remote sensing instrumentation and
methodology for stationary source emissions.

BACKGROUND

One of the active programs of the Environmental Protection
Agency is the investigation of remote sensing instrumenta-
tion and methodology for electro-optical techniques in which
the analyzer is physically removed (up to a thousand meters
or more) from the source being investigated.

The COSPEC is a remote sensing system designed to measure
sulfur dioxide and/or nitrogen dioxide by absorption phe-
nomena using scattered sunlight as an energy source. The
manufacturer's literature states that the instrument is
capable of monitoring S02 or N02 from a source in three
modes:

     •  In a helicopter or airplane,  flying in the
        vicinity of the source plume with the COSPEC
        viewing downward;

     •  In a stationary mode at ground level by sighting
        the tripod-mounted instrument on the plume;

     •  In an upward or vertical looking mode from ground
        level by mounting the instrument in a vehicle and
        traversing the perimeter of the stationary source
        property.

EPA personnel have already evaluated the stationary mode at
ground level.  It is the third mode,  the ground level moving
measurements, that was investigated in this contract.
*COSPEC is a registered name of the manufacturer, Barringer
Research, Ltd., Toronto, Canada; the various models are
described in Section 5.
                             1-1

-------
PURPOSE

The purpose of this contract was to determine the relative ac-
curacy of an EPA-furnished COSPEC II as a perimeter monitor for
SO,.  The accuracy of the COSPEC II was compared to total pol-
lutant burden measurements made using the compliance test
methods for SP2 concentration (Method 6, FR 23 December 1971)
and for volumetric stack gas flow rate (Methods 1, 2, 3 and 4,
FR 23 December 1971), and a cross-stack S02 continuous monitor.
In addition, comparisons were made between the EPA COSPEC and
two other COSPECs provided by the contractor and the manufac-
turer.
                              1-2

-------
                       Section 2


                        SUMMARY
EQUIPMENT

Three correlation spectrometers were used to gather field
data for a two week period at a pulverized coal fired power
plant.  A COSPEC II (Government-furnished), a COSPEC III
(supplied by the contractor) and a COSPEC IV (supplied by
the manufacturer) were installed in a moving laboratory to
make simultaneous measurements of SOj in the power plant
plume.  The measurements were made by the upward viewing
COSPECs as the moving laboratory was driven under the plume,
while in-stack measurements were being conducted.  Concurrent
winds aloft measurements were made with a balloon theodolite.

MEASUREMENTS

The testing of the instruments involved the following act-
ivities :

     •  Thirty-five Method 6 SC>2 measurements and Method 2
        stack gas velocity measurements were made over 20
        minute periods for six days.  The in-stack measure-
        ments were conducted in the single active stack by
        Pacific Environmental Services, Inc. (PES), a sub-
        contractor.

     •  The COSPEC remote sensors made 465 individual plume
        measurements.   There are a total of 42 hours of
        reported COSPEC data; 11.6 hrs. of concurrent re-
        mote/in-stack measurements resulted.

     •  A total of 120 pilot balloon measurements were
        made; they were analyzed to determine wind para-
        meters for S02 mass flux calculations.

DATA PROCESSING

The COSPEC measurements were processed by computer.  The
moving laboratory traverse routes and the analog chart
records of instrument S02 response were digitized.  The data
sets were merged into a matrix: X, Y coordinates and COSPEC
S02-  The pilot balloon (pibal) data were processed to
determine wind speeds and directions within the dispersing
plume.  The final results, the calculation of S02 mass flux,
were computed from the geography/COSPEC data matrix and the
wind speed and direction information.
                             2-1

-------
COMPARISON OF RESULTS

The remote sensor and in-stack measurements were tabulated
in common units of S02 mass flux (metric tons per day).
Comparison of the COSPEC results with the EPA reference
method (the cross-stack continuous monitor data were not
processed) required further processing of the raw data. The
COSPEC S02 fluxes were averaged over the 20 minute periods
during which in-stack data were collected; the Method 6/
Method 2 SC"2 fluxes were augmented to account for gas not
detected by the stack sampling but possibly measured by the
remote sensor.   Both intercomparisons between the two COSPEC
instruments and comparison between the remote and in-stack
methods were made using standard statistical procedures.
                             2-2

-------
(D
                           Section 3


                          CONCLUSIONS
   This COSPEC evaluation shows that the instrument, when used
   with a single balloon theodolite to measure winds, can re-
   motely measure S02 emissions with a relative accuracy of ±24%.
   Improvements in the accuracy of wind speed measurements would
   improve the relative accuracy of the remote sensing system.
   The methodology warrants further evaluation.

   The specific conclusions drawn from the analysis of the con-
   current remote sensor/in-stack measurements are:

         •  COSPEC Internal Consistency - The COSPEC produced
            consistent results, instrument to instrument;
            close agreement was obtained for COSPECs II, III
            and IV.

         •  COSPEC Relative Accuracy - The eight days of COSPEC
            S02 emission rates agreed within 1.5% of the six
            day Method 6 average; statistically, the COSPEC S02
            mass flux could differ ±24% from the reference
            method using single theodolite pilot balloon wind
            speeds and 20-minute averages.

         •  COSPEC Precision •- Eleven 20-minute samples of
            COSPEC remote sensing (four hours of plume measure-
            ments) are required to achieve a ±251 error with
            95% confidence.

         •  Wind Measurement Errors - Increased frequency and
            accuracy of measuring winds aloft would improve the
            accuracy of COSPEC remote measurment of S02 emissions.

   A discussion of each of these conclusions follows:

   COSPEC INTERNAL CONSISTENCY

   The COSPEC remote sensor, when used as an area monitor for S02,
   provides consistent results from instrument to instrument.
   Strong correlations were found (r = .90 - .95) between the
   calculated S02 emission rates for the three COSPECs tested.
   The emission rates (averaged over 20-minute periods) compared
   favorably even though individual (single-plume crossing)
   values fluctuated widely.
                               3-1

-------
These fluctuations in individual readings were caused by
fluctuations in wind speed and by possible variations in
plant conditions indicated by varying Method 6 results.  The
pilot balloon data provided average, representative wind
speeds for each remote sensing test period, not instantaneous
wind data for each COSPEC plume crossing; also, the single
theodolite system is inherently less accurate than other
methods.  However, the limitations of the single theodolite
pibal measurements did not affect the intercomparison between
instruments because the wind data were applied uniformly to
all three COSPECs.

The internal consistency of the remote sensing measurements
and the data processing procedures used in this study were
demonstrated.  Similar results were obtained with the three
instruments, independent of model.  However, closer agree-
ment was found between the more recent COSPEC III and IV
models.

COSPEC RELATIVE ACCURACY

Single COSPEC plume measurements of 5-to-90 seconds duration
did not yield valid mass flux results because of the changing
character of the plume.   Averaging 20-minutes of COSPEC plume
crossings improved the estimate of mass emissions;  but the
results were flawed by errors in determining the wind speed.
Hourly and daily averages improved the results by increasing
the number of both plume crossings and wind measurements.
The longer the averaging time the closer agreement there was
between the remote and in-stack methods.

The eight-day average S02 emission rates (in metric tons per
day) for COSPECs II,  III and IV were 52.46 MT/D, 52.99 MT/D
and 51.76 MT/D, respectively.  They agree within 1.5% of the
in-stack six-day average (adjusted to add the estimated 151
emissions not measured in the main stack) of 52.22 MT/D.
These averages, while not a direct day-for-day comparison,
indicate that the  two methods were providing comparable re-
sults in the long term.

The statistics derived from data obtained in this project
suggest that for COSPEC remote sensor data, the true percent
difference between the average COSPEC S02 mass flux and that
of the reference method (Method 6) would fall within ±241,
using 20-minute averaged data in both cases.  This "24%. re-
lative accuracy" estimate would be improved with refinements
in the wind data.

COSPEC PRECISION

The use of COSPEC data in the short term must be restricted
to averaging a minimum of 20-minutes of measurements--just
                            3-2

-------
as the in-stack data were averaged over 20-minute periods.
To determine whether significant improvement in accuracy
would occur for hourly- or daily-averaged data an extensive
statistical analysis "was performed on the paired remote/in-
stack data.

The number of 20-minute samples required to achieve various
percent error intervals at selected confidence levels were
calculated.  For example, seventeen samples are required  (in
the worst case) to be 95% confident that the average percent
error (i.e. difference between COSPEC and Method 6) differs
by no more than ±20% from the actual (i.e. long-term average)
percent difference between COSPEC and Method 6 results.
this  precision can be achieved in a single day of measure-
ments.

WIND MEASUREMENT ERRORS

The results suggest that the true relative accuracy and pre-
cision of the COSPEC measured S02 emissions is masked by
other errors, principally errors in determining wind speed
at the elevation of the plume.  The number of plume measure-
ments was large enough (465 over six days) to produce long-
term average results comparable to the stack sampling method.
A wide spread in the COSPEC data was caused by insufficient
wind information.   The present data base does not readily
allow separation of the COSPEC S02 error and the pilot bal-
loon wind speed error.  However, it can be concluded that
increased frequency and improved accuracy in measuring wind
speed within the COSPEC-sensed plume would directly improve
the accuracy of the remote sensing SOz mass flux calculations
                             3-3

-------
                        Section 4
                     RECOMMENDATIONS
The results of this project show the relative accuracy of
the remote sensing COSPEC is better for long-term than short-
term data collection periods and the true accuracy is masked
by the unmeasured fluctuations in wind speed at the eleva-
tion of the measured plume.  Other factors may have contributed
to the cumulative error of the COSPEC results, as well.  Exist-
ing information could be studied to extend the analysis and
clarify the COSPEC accuracy.  Examination of the following
eight areas is recommended:
       Wind measurement accuracy -
    •  COSPEC calibration -
       Cross-stack monitor results -
    •  Stationary COSPEC data -
    •  COSPEC/SKY orientation -
    •  Sampling protocols -
    •  Plume geometry -
    •  Archival COSPEC data -
use existing data
and new methods to
improve accuracy

check calibration
cells for accuracy

process data and
compare with COSPEC

compute S02 mass
flux and compare
with moving COSPEC
results

determine cause of
instrument error and
effect on accuracy

narrow averaging
times for statistical
analysis

correlate plume shape
to accuracy

analyze other mea-
surements for relative
accuracy.
The first five areas deserve the greatest emphasis because
they would yield the most useful new information.   The last
three are pertinent but are relatively less important.  Each
of the recommended areas is discussed separately in detail.
                           4-1

-------
WIND MEASUREMENT ACCURACY

The use of the single balloon theodolite wind measuring
system limited the accuracy of wind speed data used to
calculate SC>2 mass flux from COSPEC remote sensor data.
The variable wind conditions caused errors, also.  The
gusty winds experienced at the study site were typical
of a change-in-season wind regime.  The high winds
threatened to cancel or shorten the in-stack monitoring.
It is not surprising that the results show a significant
perturbation of the flux calculations attributable to
the gusty winds.

The problem of wind measurement is clarified when it is
understood that this study compares the accuracy of the
COSPEC and wind measurements together, with the Method 6
concentrations and the Method 2 stack gas flow measure-
ments.  The calculation of SCK emission rates (mass flux)
by both methods involves essentially two basic measurements:
the concentration of S02 and the flow rate of gas.  For
comparison the two formulas are:

     S02 Mass Flux = CMethod 6 * QMethod 2 * K      (page A - 18)

     S02 Mass Flux = CCQSPEC * V PIBAL * K * D * A  (Pa§e 7 ' 7)

The C terms are SC^ concentrations as measured by Method
6 and the COSPEC; the Q term is the Method 2 flow rate of
gas in the stack (f t-Vhr), and v is the pibal wind speed in
the plume (m/s).  The constants (K) can be ignored in the
present analysis; the D and A terms (distance, and wind/
road angle in the COSPEC equation) contribute errors, but
they are known to be on the order of ±7%.  Therefore, to
determine the relative accuracy of the COSPEC requires a
comparison of the accuracies of the concentration and flow
terms.

The three-COSPEC flux results have a high correlation; it
is assumed, therefore, that the C term is more precise
than the v term.  Furthermore, experience with the measure-
ment of wind speed with single and double theodolite sys-
tems support the conclusion that v is the least precise of
all the measured parameters.  Because the accuracy of the
wind speed in the COSPEC equation is low, the COSPEC ac-
curacy suffers.

Two types of wind speed inaccuracies occurred during this
study.  The single theodolite system was used which required
use of a rise rate table.  These assumed balloon elevations
could cause significant errors (as much as 40% compared to
double theodolite results).  In addition, the frequency of
                           4-2

-------
CD
   pibal measurements (15 per five-hour day) was not sufficient
   to provide winds aloft data for each COSPEC plume crossing.
   the use of representative wind speeds contributed signifi-
   cant errors to the flux calculations.

   It is recommended that:

        •  Other double theodolite pibal data be reviewed to
           quantify the error in the single theodolite data;

        •  Continuous tower anemometer data be used to develop
           a methodology for adjusting pibal wind speed data
           to be used in mass flux calculations;

        •  More frequent and more precise winds aloft measure-
           ment devices (double theodolite, acoustic or laser
           doppler remote sensor) be used with the COSPEC to
           determine possible improvement in accuracy.

   COSPEC CALIBRATION

   Following the field work and prior to data processing work
   the calibration cells in the COSPECs were checked for S02
   content.  The high concentration SO? cell was removed from
   the EPA COSPEC II (Serial Number 5922) and sent to the man-
   ufacturer for recertification.  The cell was reported to
   have a value of 377 ppmM, as compared to the initial value
   of 350 ppmM S02 (see Table 2).  This 1.1% increase in this
   one cell was not taken into consideration when calculating
   the COSPEC results.

   The manufacturer was asked also to retest the cells in the
   COSPEC IV (the COSPEC III was being used in a field program
   and was unavailable for testing); the results were reported
   as "no change in cell values".  However, the field cali-
   brations from the COSPEC IV suggested that eigher the low
   cell (105 ppmM) was actually some lower value or the high
   cell (395 ppmM) was higher than certified.  Prior to pro-
   cessing the COSPEC IV data the calibration curve for this
   instrument was adjusted using the field calibration data.

   Only full retesting of all calibration cells in all three
   instruments would provide sufficient information to apply
   correction factors to the calculations contained in this
   report.

   It is recommended that:

        •  All six COSPEC S02 calibration cells for the three
           instruments used in this study be recertified by
           the manufacturer;

        •  Any changes in cell S02 concentrations be analyzed
           to determine what impact they would have on mass
           flux calculations.

                              4-3

-------
©
   CROSS-STACK MONITOR RESULTS

   The cross-stack continuous monitor SC>2 data provided by the
   power plant personnel were partially processed by EMI.  During
   this procedure it was determined that there was some uncer-
   tainty about the calculation of the correction factors in the
   final results.  The instrument manufacturer recommended going
   back to the original data to recalculate the final results.
   Then SO? mass flux computations could be made, using the in-
   stack flow rates, to compare with the COSPEC results.

   It is recommended that:

        •  The cross-stack monitor S02 data be verified;

        •  Calculations of S02 mass flux be made, and compared
           with COSPEC mass flux using the same statistical
           methods applied to the Method 6 data.

   STATIONARY COSPEC DATA

   During four days of field measurements one of the three COSPECs
   was mounted on a tripod to make stationary plume measurements.
   Neither the field procedures nor data reduction are in this
   report because these experiments were not included in the
   scope of work.   The data could be reduced to compare fixed
   COSPEC flux calculations with the moving COSPEC data.

   This would be particularly interesting for those instances
   where the plume rose vertically; the COSPEC was rotated on
   the tripod, scanning from left to right across the top of the
   stack.   The S02 emission rate could be determined from these
   plume profiles by using the PES stack gas velocities.  Also,
   vertical scans were made when the plume was transported hori-
   zontally;  in these cases the emission rates could be processed
   using the pibal wind data.

   For either situation direct comparison could be made between
   the stationary and the moving measurements, which were being
   conducted simultaneously.   In the first case (the use of the
   stack gas velocity instead of the pibal wind speeds) further
   insight could be obtained into the pibal wind  speed errors.

   It is recommended that:

        •   Stationary COSPEC data be processed into S02 mass
           flux;

        •   Comparison of stationary and moving COSPEC mass fluxes
           be made to help clarify errors contributed by the
           wind speed measurements.
                              4-4

-------
COSPEC/SKY ORIENTATION

During the analysis of this large body of remote sensor data
it was noted that some of the calculated S02 mass flux values
fluctuate, from high to low numbers alternately, one traverse
to the next.  The pattern of alternating high and low fluxes
(for example, see Table 5, 14 May 1975, Events 17-21) suggests
that the COSPEC may have a sensitivity to the skyward orien-
tation of the light-gathering telescope for certain times
of day, sun angles, cloud cover, and other unknown conditions.

This COSPEC/sky orientation effect occurred about ten percent
of the time.  For the five events in the cited example (14
May) the six "high" fluxes for all three COSPECS average
46.2 MT/D; the nine "low" fluxes average 25.2 MT/D.  Neither
the high nor the low results were excluded from the sta-
tistical analysis.

A more detailed examination of the 465 flux calculations
could be carried out to determine during which time periods
this orientation effect occurred, whether the average of all
readings is a "true" value, or whether the higher or lower
calculations should be disregarded.

It is recommended that:

    •  An analysis of the apparent COSPEC sensitivity to
       sky orientation be performed to determine the extent
       to which it occurred during this study;

    •  Experimental data be gathered to measure the possible
       impact on COSPEC accuracy.

SAMPLING PROTOCOLS

The averaging times used for the COSPEC data and the in-stack
data may have affected the comparison of results.  Some lati-
tude was used in grouping the COSPEC results for averaging.
This included using COSPEC data gathered between the in-stack
20-minute tests when averaging over periods of one hour or
longer.  By narrowing the time spans used to calculate the
averages for the COSPEC results the comparisons may change;
similarly, by using results collected only during the 20-
minute in-stack test (excluding COSPEC measurements made be-
tween)  the comparisons may change.

It may be necessary to refine the statistical analysis to
make certain,  wherever possible, that the two methods were
measuring the same flux gas parcel.   Calculation of the
COSPEC position and the time for transport of the parcel
                          4-5

-------
from the stack would insure that "simultaneous" measurements
were indeed made.  This could be important because there is
reason to believe that the sulfur feed rate to the boiler
varied over short time periods.  For example, the range of
daily average sulfur in the coal was 0.79 to 1.11%; the range
of shift averages was 0.67 to 1.21%.  It is reasonable to
expect that there were short term variations in sulfur con-
tent, as well.  Therefore, COSPEC plume measurements not
truly concurrent with the Method 6 samples times perhaps
should be excluded from the analysis.

It is recommended that:

    •  COSPEC plume data not concurrent with in-stack mea-
       surements be deleted from the COSPEC averages;

    •  The statistical analysis be made to determine the
       effect of narrower COSPEC averaging times.

PLUME GEOMETRY

Many of the COSPEC plume profiles measured were non-Gaussian
in shape.  Multiple peaks for a single plume profile occurred
for three possible reasons:  in-stack vortices caused plume
bifurcation; gusty winds tended to break up the plumes; bumpy
roads caused the COSPEC to swing in and out of plumes during
some traverses.

The bifurcated (bi-modalj plume profiles occurred both at the
stack exit (measured by stationary COSPEC) and frequently
within one kilometer downwind (measured by moving COSPEC).
This phenomenon,  and the dispersion of the plume by gusty
winds,  probably caused most of the non-Gaussian profiles;
these were fully processed.  The profiles which were seg-
mented further (by the "rotating" COSPEC traveling on rough
roads)  were usually excluded from the analysis.

The plume profiles could be classified into Gaussian and non-
Gaussian categories; these could then be correlated with
the results of the flux calculations to determine which pro-
vided more accurate results.

It is recommended that:

    •  Plume profiles be categorized with respect to shape
       and etiology;

    •  Correlation of profile categories with-S02 mass flux
       be made to determine the effect, if any, on accuracy.
                           4-6

-------
ARCHIVAL COSPEC DATA

While a large quantity of remote sensor data were gathered
for this study the relative accuracy of the COSPEC could
be determined more precisely if the data base were enlarged,
To expand the data base for further analysis two sources of
information could be used:   archival data (e.g., Maryland
Power Plant Siting Program) and new COSPEC measurements at
other S02 sources.  Comparable in-stack S02 data could be
made available for either past or future COSPEC work.

For either the old or new data two important parameters
could be studied.  The accuracy of the wind speed measure-
ments could be improved using double (instead of single)
theodolites or remote wind sensors; the wind gustiness
problem could be  eliminated by selecting sites and times
of measurement where wind flow is more stable.  Having data
from more than one site would increase the confidence in
the remote sensor results.

It is recommended that:

    •  Archival COSPEC data be analyzed to determine rela-
       tive accuracy, or

    •  Additional COSPEC measurements at new sites be made
       to broaden the data base and strengthen the analysis
       of relative accuracy.
                          4-7

-------
                        Section  5
                 EQUIPMENT  AND  FACILITIES
AIR QUALITY MOVING LABORATORY

An EMI Air Quality Moving Laboratory (AQML)  was used to trans-
port the COSPEC remote sensors during the plume measurement
tests.  The AQML, shown in Figure 1, is designed to make moving
COSPEC measurements;
COSPEC evaluation.
serve as a mounting
spectrometers.  The
position behind the
 it was partially modified for this
A platform was installed in the van to
surface for two additional correlation
third COSPEC was installed in the usual
driver.  Each instrument viewed over-
head through a telescope fitted with a right-angle mirror.

To monitor ground-level S02 a Bendix Model 8300 flame photo-
metric Total Sulfur Analyzer was installed in the AQML.   The
point sampling instrument was placed, along with the hydrogen
gas needed for its operation, in the central portion of the
 van.  Air was  drawn  through  a  teflon sampling  manifold  into the
 van by  a  squirrel-cage  fan;  a  short  teflon  inlet tube,  in  turn,
 sampled this  air  stream for  analysis by  the  point  monitor.
    Figure 1.  The AQML carried three COSPECS; two
    viewed   the  sky out the right  front window, one
    through  the window behind the driver.
                              5-1

-------
The analog signals from all four instruments were recorded on
a six-pen Rikadenki strip chart recorder.  A rear-mounted
propane-powered generator supplied electrical power for all
instrumentation.

A two-person crew operated the AQML.  One person was the driver;
the data logger, sat in the rear seat and observed the six-
pen chart recorder, making notations of times and geographic
locations on the chart records.  Duties were rotated frequently
with two additional persons at the fixed field site.

COSPECS

Sulfur dioxide total burdens were measured by the three
vehicle-mounted COSPECS.  (NO? channels were available from
the COSPEC II and COSPEC III but they were not connected
for this study.)  The natural radiation of the solar electro-
magnetic spectrum is influenced by the absorption spectrum
of the target gas, sulfur dioxide, in accordance with the
Beer-Lambert law of absorption.  The correlation spectrometer,
an electro-optical instrument, detects portions of the mole-
cular absorption bands specific for this molecule.  The op-
tical unit includes a Cassegrain telescope, an Ebert-Fastie
quarter-meter dispersive element, a correlation disc assembly,
and a photomultiplier to detect light energy levels.  The
electronics of the COSPEC contain signal processing circuits
to provide an analog output signal for strip chart recorders.

Three different models of the correlation spectrometer were
used in this study.

     •  The COSPEC II, (Serial No. 5922) is a dual-gas
        monitor intended for quantitative measurement of
        sulfur dioxide and nitrogen dioxide.  It is designed
        for use in the passive mode only.  That is, it can be
       . operated exclusively by solar illumination.

     •  The COSPEC III (Serial No. 5932) is also a dual-gas
        monitor with the ability to function in the passive
        and active modes.   This latter option allows the in-
        strument to be operated with a remote modulated arti-
        ficial light source.

     •  The COSPEC IV (Serial No.  6256) is designed for single-
        gas measurements.   Improved optics allow for an in-
        creased sensitivity on the order of one magnitude
        above the II and III.   This instrument operates in
        the passive mode.

In this study all three COSPEC's were used to detect a single
gas (S0£) in the passive mode of operation.  The manufacturer's
specifications are listed in Table 1 for a dual-gas instrument.
                             5-2

-------
CD
                              Table 1

                   BARRINGER RESEARCH  COSPEC III
                 DUAL-GAS CORRELATION  SPECTROMETER
                           Specifications
TARGET GASES:

LIGHT SOURCES:

OPTICS:
DYNAMIC RANGE:

SENSITIVITY:


RESPONSE TIME:

OUTPUTS:
CALIBRATION:


CONTROLS:
MECHANICAL:
ENVIRONMENTAL:


POWER REQUIREMENTS
S02 and/or N02

Skylight

Cassegrain telescope on front turret 1° x 1'
A right-angle mirror attachment is provided
for "vertical look" operation.  A sighting
telescope is provided for alighment on the
light source or a target plume.

1-1000 ppm-M (parts per million-meter)

2 ppm-M, (threshold at 8 second integration
time)

1,2,4,8,16, or 32 seconds

• S02 Signal
• N02 Signal
• S02 AGC Voltage
• N02 AGC Voltage
Four fused-silica cells
and NO9
  Power, ON/OFF
  Signal alignment
  Meter scale change
  Zero offset adjust
  Integration time
two each for SO,
• Entrance slit trans-
  lation control
0 Calibration cell
  selectors
  Size:  71 cm x 30 cm x 43 cm  (28" x 12" x 17")
  Weight:   17.2 kg  (38 Ibs)  including
  isolators and mounting plate (provided with
  standard tripod mounting holes)

Vibration isolators are standard
Ambient temperature range: -20°C to +50°C

115 VAC, 60 hz, 18 watts
                                5-3

-------
2  away from plumes, just prior to or following a traverse.
    Notations of the time and the choice of cells used during the
    calibration and the recorder sensitivities are made on the chart
    record.  The voltage for the automatic gain control (AGC)
    circuit is also noted on the chart record to provide an in-
    dication of the changing light intensity during the day.
                              Table  2

                    COSPEC CALIBRATION AND RANGE

CALIBRATION
CELLS:
Low
High
Low+High
Linear Range *
Maximum Useful
Range *
SULFUR DIOXIDE (ppmM)
COSPEC II
S/N 5922

65
350
415
0-650
0-1600
COSPEC III
S/N 5932

75
180
255
0-600
0-2000
COSPEC IV
S/N 6256

105
395
-
0-800
0-1900
    *As determined by field tests; see discussion on page 5-5
    and Figure 3,
                                5-4

-------
On this project, calibrations were performed on each COSPEC
eight to fifteen times during every measurement day.  These
calibrations were used to determine the sensitivity of the
instrument to S02.  Due to the varying intensity of the
Rayleigh-scattered ultra-violet light, this sensitivity changes
during the day.  Therefore, the sensitivity values for each
day were plotted and a set of sensitivity curves was determined
for each instrument.  A typical sensitivity curve, that for the
COSPEC II for 8 May 1975, is reproduced in Figure 2.  The
parts per million-meters/millivolt (ppmM/mv) sensitivity for
this instrument on this day varied from approximately 0.50
ppmM/mv in the morning and evening to approximately 0.28
ppmM/mv at midday, when the ultra-violet light was most intense.
The sensitivity curves were used to derive the specific
sensitivity value for each COSPEC measurement event.

The COSPEC has a linear range from its noise limitation,
5-25 ppmM SO?, to approximately 700 ppmM.  This is a function
of the Beer-Lambert Law, and is substantiated by in-the-field
experience with the instrument.  The linear relationship of
the three COSPECS used in this study was verified on site.
High concentration SO, calibration cells were placed into the
instrument's light path to provide calibration steps as high
as 5000 ppmM.  Figure 3 shows a typical COSPEC response curve.
The linear portions of each COSPEC and the estimated maximum
useful range  (as determined from these curves of instrument
response) are listed in Table  2,

These linearity data were used to process the field results.
COSPEC data in the linear range were used as gathered; values
in excess of the linear range were corrected using the curves
generated from the field linearity tests.

In practice the non-linearity of the COSPEC was encountered
only for plume measurements as close as 500 meters from the
power plant stack.  Measurements made 500 meters to 5 kilometers
downwind were within the linear range.  Approximately 191
of the data gathered were corrected for non-linearity.

PIBALS

The measurement of wind speed and direction, essential para-
meters in the calculation of SO, mass flux, was accomplished
in two ways.  A tower-mounted Climatronics anemometer, (8.2
meters above ground level)  was located away from the immediate
influence of the power plant, within about 600  meters  of
the source;processed data were provided by power plant per-
sonnel.
                          5-5

-------
                               Figure  2


                TYPICAL COSPEC SENSITIVITY  CURVE
E
Q-
Q.
0.6



0.5



0.4



0.3



0.2
oo
z
LU
OO
                        1     »
8     9
                                COSPEC  I I S/N 5922  8  MAY  1975
                             I I     12

                             TIME (MDT)
                                                1.5
               60
               40
           OO
           CD
           <
           o


           t3   on
           O   ^0
                               Figure  3


                          TYPICAL COSPEC
                        NONLINEARITY  CURVE
                              /Extrapolated
                            /  Ii near output
                                    Actual
                                  instrument output
                                      COSPEC I I
                                      S/N 5922
                           2000       4000


                             S02 (ppmM)
                                              6000
                              5-6

-------
Single theodolite pilot balloon (pibal) measurements were made
250 meters north of the source to measure winds aloft.  The
actual stack height was 122 meters; the effective stack height
(stack height + plume rise) was 150-300 meters or more.  The
measurements were used in the field to understand the changes
in wind movement at these elevations, while the moving measurements
were in progress.  The data were used in the calculation of SC^
mass flux.

A WeatherMeasure Model BT 901 balloon theodolite was employed.
Thirty-gram weather balloons were filled using a 139-gram
filling device.  The helium-filled balloons were released during
Method 6 measurement periods and were tracked for five minutes
or more — to an altitude of approximately 1000 meters.

STACK SAMPLING

Sulfur dioxide stack emissions were measured by Pacific Environ-
mental Services Inc. using EPA Method 6.  Velocity traverses
were conducted using EPA Method 2 and moisture determinations
were performed using EPA Method 4.  This sampling took place
82 meters above ground level in already available sampling ports.
The ports were located downstream from the collection device and
blowers and upstream from discharge to the atmosphere, approximately
eight stack diameters downstream of any flow disturbances.  The
test methods are fully described in Appendix A.

All laboratory titrations were performed on site by PES person-
nel.  Bubbler contents were titrated with barium perchlorate
using Thorin as an indicator as specified by Method 6.  The
on site chemical analysis proved to be fortuitous.  Difficulties
with a vacuum pump invalidated some of the early tests; this was
discovered only after analysis of the initial stack gas samples.
After replacement of the faulty pump the stack sampling proceeded
routinely.

Additional stack measurements were made by means of a continuous
cross-stack monitor.  This device measured SC^ concentrations in
the stack gases at the same platform where the manual Method 6
stack sampling was performed.  Data from this instrument were
telemetered to the central control room of the power plant and
processed and printed on daily log sheets.  These results were
provided by power plant personnel.
                           5-7

-------
(D
                           Section 6


                       FIELD MEASUREMENTS
   REMOTE SENSING METHODOLOGY

   EMI employed procedures developed over  five  years  to  make
   moving measurements of power plant plumes  as required by  the
   contract.

   The AQML was driven around and downwind of the power  plant
   usually within a one kilometer radius  but  as far downwind  as
   five kilometers.  The objective was to  cross the plume, at  as
   many different downwind distances as possible, to  measure  the
   dispersing S02.  This was difficult to  achieve, as there  were
   limited traverse routes; this problem was  overcome, however,
   by making repetitive surveys on available  roads.

   The traversing speed varied with the distance from the source.
   Close to the power plant the speed was  kept below 30 kilometers
   per hour to allow the instruments to respond fully and to  pro-
   vide clear definition of narrow plumes. As  the AQML  was  moved
   further downwind from the source, it moved faster under the
   plume.  Because the plume is broader at the  greater distances,
   changes in overhead burdens and ground-level concentrations
   are less abrupt, and the instruments respond to them  easily.
   Decisions were made in-the-field based on  the real-time data:
   whether to repeat the plume-tracking measurement at the pre-
   sent radius or to move to a second radius  of measurement.


   WIND MEASUREMENTS

   A single pibal theodolite was used for  the entire  study.   The
   theodolite was located approximately 250 meters due north  of  the
   stack.  This was far enough away to avoid  influence by the super-
   structure of the power generating station  for most wind directions
   and speeds encountered during the study, but close enough  to  main-
   tain visual contact with the stack sampling  crew.  A  flag  system
   was used to indicate when stack sampling was in progress.  The
   pibal crew would schedule equi-spaced  balloon releases during these
   test periods.

   Typically four pibals were tracked during  each stack  test.
   The moving laboratory was actively measuring before,  during
   and after these 20-minute intervals so  that  supplementary  pibals
   were released during times when the stack  sampling apparatus  was
   being purged  and/or prepared for a new  series of test.  The measure-
   ment of winds  aloft was made an average of 15 times per day.
                                6-1

-------
Readings of balloon position were made by the theodolite
observer and written on a data sheet by a recorder.  A hand-
held calculator was used to compute wind speeds and wind
directions on site.  This gave the field crew results with
which to understand the changing meteorological conditions and
to guide the moving laboratory to the proper position for
plume measurement.

The real-time-data from the moving laboratory, of course,
provided additional data on the wind vector.  The location
of the plume and visual observation of geographic position of
the AQML provided continual feed back on the wind conditions.
However, when winds became gusty or the wind direction would
rotate rapidly, the periodic return of the moving laboratory
to the pibal site was useful in validating wind conditions.
This procedure optimized the data collection efficiency of the
moving laboratory.

ACTIVITY SUMMARY

Measurements were made at the power plant for 12 data-days;
a summary of the measurement activities is presented in Figure
4.  This chart shows time periods when the moving laboratory
was active and when Method 6 stack sampling occurred.  No
Method 6 test times are shown prior to 13 May because results
from these tests were invalidated by the faulty vacuum pump.

A total of 35 valid Method 6 tests were completed between 13
and 20 May 1975.  This resulted in 11.6 hours of concurrent
testing with the remote sensor measurements.  A total of 42
hours of remote sensor measurements were actually gathered.
This includes data collected between 8 and 12 May 1975,  during
which continuous in-stack monitor data are available.  COSPEC
measurements for 5 through 7 May were not treated because
neither the in-stack monitor nor the Method 6 determinations
were valid for these days (the in-stack monitor was undergoing
calibration and adjustment).

On a typical day 60 crossings of the plume were made by driv-
ing the moving laboratory around and down wind of the power
plant stack.   Two or three COSPEC's were mounted in the AQML
at all times:   multiple COSPEC measurements resulted from most
plume crossings.  An average distance of 150 kilometers was
covered by the moving laboratory each day.
                             6-2

-------
 DRTE
5 MRY


E MRY


7. MRY


B MRY


3 MRY


12 MRY


13 MRY


IH MRY


IS MRY


IB MRY


13 MRY


20 MRY
0
                  Figure 4

              HCTIVITY 5UMMRRY

                MRY   1375
    TI ME    < MDT >
E        12        IB


             LJU
                 CD5PEC MER5UREMENT5

            I    I METHOD E TESTS
2H
                  6-3

-------
MEASUREMENT SCENARIO

The activities of the typical day can be presented in scenario
form; a photographic review of EMI and PES work is shown in
Figures 5 and 6.

     The two measurement crews arrived at the power plant each
     morning in three vehicles:   the AQML, the pibal station
     wagon and the stack-sampling van.  The stack sampling
     train was prepared for the  first series of tests while the
     theodolite was sef up and the COSPECs were given initial
     calibrations.  The moving laboratory immediately began
     surveying around the power  plant, at a 250-meter radius, to
     determine the plume vector.   The stack sampling crew
     reached the  sampling platform by elevator and set up their
     equipment.   When they were  ready for the initial test a
     red flag was hung on the platform railing.  Communication
     by radio with the pibal crew verified the starting time of
     the first test.

     The measurement cycle was a  20-minute Method 6 sampling
     period followed by a 20-minute purge time and a second
     20-minute test.   This pair  of tests was followed by an
     interval of  30 to 60 minutes.   During the first 20-minute
     run,  the first three or four pibals were released while
     the moving  laboratory continued traversing.

     The objective during any one 20-minute experiment was
     to achieve  as many plume cross sections as possible at
     two or three downwind distances.

     Coordination between the ground and stack sampling crews
     was maintained throughout the day.   At times  conditions,
     such as changing wind directions, would cause a few
     minutes delay in the start  of the new run, but usually
     the stack sampling proceeded at its scheduled pace.   The
     moving measurements  were essentially continuous with
     multiple plume crossings during the 20-minute test times.
     Between stack sampling intervals  the AQML continued
     traversing with an emphasis  on tracing the plume as far
     downwind as  possible.
                            6-4

-------
Figure 5.  EMI crew at work.  Clockwise from upper right:  cali-
brating three COSPECS prior to installation in moving labora-
tory; digitizing base map for computer analysis of COSPEC data;
reading balloon theodolite to measure winds aloft; communicat-
ing with stack sampling crew by radio to coordinate field
measurements.
                            6-5

-------
Figure 6.   PES crew at work.  Clockwise from upper left: insert-
ing sampling probe into port; connecting impingers to probe; pre-
paring fresh set of impingers; performing on-site titration.
                                 6-6

-------

-------
        POWER PLANT STACK
               I  KM
          FIGURE 7
  TRRVER5E ROUTE MRP
  CD5PEC EYRLURTIDN
  B-20  MY 1375
ENVIRDMENTRL  MER5UREMENT5/ INC.
     5RN FRRNCI5CD/ CB 3HI1I
                              7-2

-------
(D
  Key events were evaluated for relevance to the project goals;
  events with offscale readings (because of temporarily incorrect
  chart recorder sensitivities) were eliminated from further consi-
  deration.  Approximately 50% of the events were selected for pro-
  cessing.

  The raw moving laboratory field data included the analog traces
  for SC>2 total burdens for the three COSPECS and ground-level
  concentrations from the point monitor on a single chart record.
  These records also included hand-written annotations of time and
  positions made by the data-logger, as well as instrument calibra-
  tions, time constants and weather conditions.  Figure 8 is a
  sample chart record of moving measurements.  The data were recorded
  on 12 May, 1975 in the afternoon  (1514-1516) MDT.  They were iden-
  tified as Event 36.  The digitized portion of Event 36 began at
  geography reference point No. 511 and ended at point 308.  This
  route was 720 meters east of the power plant stack.  The distance
  covered by the AQML was 310 meters.

  The three COSPEC traces show that an S02 plume was detected which
  had a peak between points 512 and 513.  The three analog signals
  tracked together (accounting for the pen offset); each instrument
  recorded a generally Gaussian distribution  with superimposed
  peaks and valleys.   The ground-level S02 monitor recorded no gas
  above 10 ppb during this two-minute plume crossing.

  To process these data reference baselines were first drawn for
  the SQ2 burden records.  The background was defined as the
  instrument output on either side of well-defined plume anomalies.
  For the example in Figure 8 the baselines were 12% COSPEC II, 31%
  COSPEC III, and 20% COSPEC IV.

  Each analog trace was digitized at geographic reference points,
  and points inserted between to provide a sampling density
  sufficient to characterize the plume anomalies.  Major assump-
  tions of this procedure are a constant velocity of the vehicle
  between indicated landmarks, and straight line interpolation
  between geographic points and inflection points.  Therefore,
  assuming straight line variations between each of the digitized
  points,  they may be joined by straight lines to recreate the
  original record.

  No correction was made for time delays caused by instrument response
  time.   The COSPEC time constant was set at one or two seconds;
  coupled with the slow moving AQML (typically 30 km/hr) the offset
  in the plume anomalies with respect to the geographic reference
  points was assumed negligible.

  MERGING THE DATA MATRIX

  After the geography and the COSPEC data were separately digit-
  ized the two sets of digitized values were merged into a matrix
                               7-3

-------
         Figure 8
TYPICAL AQML CHART RECORD
 12 May 1975, Event 36
't

1
(
(
<
1
(
(
..; 1
1
1
(
;
'•'
j
~? •"';
r c
! u~
i u«.
r £
> £—
i' C
i' t:l
»
1
1
(
1
1
t
>
r
i
i
»
5
!
9
9
9
9'
9
9
Q
..'
9
9
9
9
9
9
9
9
o' i :


i i
i





9




"/•'
-=r


' ' i • :
i k
H~




• t
i


- 1
^
r3'
k^
£
•ff
M
II!


• - - - 1
f *
>j
. 3
•• =: r
<
- . S
"r'
S
:.'!{ '.'.'. ' ;!R ! • '!!4;; ; ' :'sl i; i !t>
:. Iji i ! Bi i ' i; i i f '
-i || ; i ji| Ilj ^s_ i^^
i1.' \\ : M' n ! M ! I *•. A
\ \ Event No. l-fA/ ^
j'J \ I I Vs'
, '^.... •* 	 """I:".
llmllliifTtfTnfln^^
:i:::::::::::::::::i«;!::^i;i;;:::t

{ li-iite^r1 1 T i
> T F
^Lf' CqSPECitn.._S02BL
( I u. E ll
. l ,--COSPEC IV SO? Burden
J- 1 i
COSPEC. t S02 Burden
1 ; . >
' Ground- level ' S02 Monitor 1
TABLE 3
EHVIF
SflN F
COSPE
PR.XE
EVEN!
••IfiP JL
•"' 0 C1
a •„.' *? '-'
, 396
,394
~.' '!"j 1
t •-.' ..• .L
i • J!1 C1 • !7
, 392 •
' 394
•397
399
4 0 4
3 9 2
3;-: 0
3 b (".
•..".' O •-'
>*i i !'! !! a! ! ! 1 < «- '
I \\ i : i ; i i- i : i 1 :


ffidirl ^ 1 \ \& | *"
H.I j-fe ' i ia ' i .j. ^sa_

to) xi'"Hi-XlJ-
T g | • •• X i
T >•!' i"""'T : :°
	 := p) '2 	 ; . •=»
:::::|:::!jDIqlttzIng Mark.: '**





It .•
' * ' ^^
1 L ' )
i *
1 Time; °*
^-
	 siL::::::::--

.'ONMENTfiL MEflS.REMEMTS IHC
RHNUibUC ? JHLIFjFu-ilFl
:C EVRLLIflTIHN
:CT 181 12 MR1-' 19
" HC. 36 ' 1514 - 151
lORDS GEQG . C I SPEC
' Y • FT. SI2T
KM • PPMM
75.330 511 0
75.313 0 -14
' 75 . 295 0 . 2t
75,278 0 2G
75,268 512 • .14:-
7 £.2 47 ' i- ' 15C:
75", 233 i S6"7
7? . 220 !• 218
7 5. 2 87 !• . 299
75,193 t 218
- 75,188 51: 162
r'' b « 1 b 7 i 6 1
75. 153 ' i 2£
75,: 140 30 • 0
75 ,,098 6 £
75.048 308 t

, • 7-4 .
75
6 MDT
11 J-HPEU ill I iJbK !•:.!..:
B 882 IB 802 TE
PPMM . PPi-ltf
0 0
U • 0
0 0
1 8 :•': 1
:. ::8 ' 109
317 ,!b4
" ' 251 226
290 194
198 21.7
158 . 70
"* ' '. • J "i
8 0
6 8

-------
CD
  to  be  processed  by  computer.  An example of this matrix is
  presented  in  Table  3.  These are the same data which are shown
  in  the original  chart  records for  12 May 1975 Event 36  (Figure 8).

  The data  listing shows the X,Y  coordinates of the position at
  which  the  data were recorded; geographic reference points are
  listed next.  The zeros which appear in the geography point column
  correspond to the added marks which appear on the chart record.
  The last  three columns list the COSPEC II, COSPEC III, and COS-
  PEC IV raw data  in  ppmM.  These results were derived from the
  percent full  scale  readings taken  from the chart records, with the
  subtraction of the  reference background and multiplication by
  the sensitivity  of  the COSPEC for  that portion of the measurement
  day.   An  example of a  data point processed in this way is shown
  below:
          (66S FS  -  mFS  ) x  6.80        = 367 ppmM


  Note  on  Table  3  that  the three sets of COSPEC readings, although
  not identical  in value,  do show a close correlation as the values
  rise  from  zero (background)  to a peak and fall back to a zero.
  The peak values  for the  three instruments were 367 ppmM, 356 ppmM
  and 333  ppmM,  respectively.

  WIND  DATA  ANALYSIS

  The wind data  gathered at the study site were processed to deter-
  mine  wind  speeds within  the  COSPEC-measured plumes for use in
  the calculation  of SO? mass  flux.  The primary data used were the
  pibal measurements; the  continuous anemometer data from the low
  tower were consulted  for trends in wind speed.   ( A problem with
  the anemometer assembly  intermittently caused wind direction
  readings 180°  out  of  phase,  limiting the use of  these data.)

  The weather conditions observed at the site have been summarized
  and appear in  Appendix B.  A daily resume of wind and weather
  conditions, a  copy of the anemometer wind speed  and wind direction
  data  for the daylight hours, and a summary of the pibal measure-
  ments are  presented for  each data-day.

  The analysis of  the pibal results first required the complete
  processing of  each set of theodolite measurements for every pilot
  balloon  released.  The calculation of wind speed and direction was
  based on an assumed balloon  rise rate. The altitude of the pibal
  for each thirty-second observation was taken from the tables used
  by  the  National Weather Service and the EPA.  The complete listing
  of wind  speed  and wind direction at the various  altitudes were
  tabulated  and  summarized.

                               7-5

-------
To select a wind speed for any given event, the wind data be-
tween ground level and 600 meters were analyzed.  The heights
to be used in the calculation depended on the downwind distance
of the traverse, plume touchdown as measured by the AQML and the
overall wind speed for that period.  Typical altitudes used
were in the 100 to 300 meter range for in-close plumes without touch-
down.  Several kilometers downwind, when there was plume touch-
down, the pibal heights used in determining wind speed were
ground level to 500 meters.

The number of pibals varied throughout the day.  The pibal
nearest to the time of the plume measurements was always consulted;
if additional pibals were released within the same time period,
an average wind speed was used.  The anemometer data were consulted
for trends in wind speed.  But no quantitative results were used
because the instrument was measuring only low elevation winds.

The selection of wind directions for each measurement event was
carried out simultaneously with the wind speed determinations.
Again, where more than one pibal measurement was made, average wind
directions were calculated.  Additional wind direction data were
inherent in the plume measurements themselves.  The plume center
line was plotted on a map of the power plant and surrounding
area using the peak value from the COSPEC readings to locate the
plume axis.   These plume vectors were measured and tabulated, and
agreed with the independent pibal measurements usually within 10
degrees.   Frequently, due to the gusty nature of the winds, the
direction of the plume measured by the moving laboratory varied
considerably from that measured by the pilot balloons.  In such
cases the direction determined by the moving laboratory was used
in the calculations.

CALCULATING SO? MASS. FLUX

The calculation of the final results -- the S02 mass flux --
was done by a proprietary computer program which operates on the
geography/COSPEC data matrix and the wind speed/direction infor-
mation.   An example of the results of the calculation is shown in
Table 4.   The printout lists the segment length, the wind/road
angle and the mass flux for each segment and the mass flux for
each segment and the cumulative sums for each of three COSPECS.
The final results are expressed in metrrc tons per day, pounds
per hour, and kilograms per hour.
                              7-6

-------
                               Table   4
                     SAMPLE MASS  FLUX CALCULATION
       FLUX
                            i-2;:Mfl'f 197-5-
                       EVENT
  ••MIND =  : t-4'..l-'. M/-S  FROM  -: 305-: DEGREES
             -,'   ;:: •.'''. • •»•'•'• :-:'£OSF'EC -I r';>;*'J::.' ";';:'cbsPEC  IIlV?;-v;C;-\T;".''^GbSP^
            SEG' "'' •:•;'.:-.S.b.2"::- : !sb2^Ustf2.'' :.•:'• 302,. '  302'-: -/SQ2..r/l'l.^SLS':;^S'Q2S;:;3i3'2^
            EHGTK-  ' •'-• . A ..LYFLUX .' '^UFf-  •.•'-•'."••"•••.'•'FLUX'- ° SUM, i:^--'-'  "•''"'• B-LUXv-o.^Hft'''"
RMGLE  LENGTH-          .  .. .   	
  -    -. METERS •-'..".P.PMM. tiT'/D . .MT/D-., ' - PPHM.;,- MT/D
             1 8 •
            ' 50.:
             51
                     41
    •2.3 '-  -5S ,   ...1:06  ..:-6.. 1
0' '• -0. 6.  -/ -5.9v   '.'33"• "-1. 9
8' -  0,0 -. .. :-59-.-    •  '7-  ';-Q. 8.
0.'•.•  0-. 6    "~59~ - . •- .  •© .   -0,0
39   ~2*2
 4 '  -0,,.2
 ti    6»0.
 S  .  fi' O •
                                                                                   ~6ti
                 -FLUX: -•-:58.8:.   MT/I):
                  '• •:••'•->  -•• .--v-245t ', KG'/HR :
                  "  "   %"54'02  LBS/HR -l
                                                68.0  •:I1T/.D;-
                                                2.833: KG/HR
                                                6244 -LBS/H.R:
                                                2495-' ';KG./'HR
                                                5499 LBS/HR-.
T 0 T R L. TR R V ERSE L E N G T H = '.  '  0. 31 l< IL 0 M E T'E R S
REFERENCE. TOPE  FILE  NUMBER  (X  VRLUE)-  10
                                   7-7

-------
(D
    The  computations of SOomass flux--emission rate of the tar-
    get  power plant—were derived from the COSPEC total burdens,
    the  geography, and the wind speed according to the following
    formula:
   Mass Flux =

   Where:
C*D*a*v*K
   Mass Flux =  The rate of flow of the substance, mass per unit
                volume in milligrams per second (mg/s)
                prior to use of constant (K).

           C =  The COSPEC-supplied burden value.   Specifical-
                ly  it is the average value between two adja-
                cent X § Y points as expressed in milligrams
                per square meter (mg/M^) above background.

                For SC>2, mg/M2 is derived from parts per mil-
                lion-meters as follows:

                     ppm-M =  2.66 milligrams  per square meter
                              using:

                     ppm   =  1 unit per 1,000,000 units

                           =  1 cubic cm per cubic meter

                     Molecular volume at STP for S02 =

                              64(gram  molecular weight)
                              22.414 liters(mol volume]
                     Cubic meter = 999.972 liters

                     STP = 273°K;760 mm Hg

                     Temperature and pressure  at processing of
                     S02 calibration cell =293°,  760mm Hg.
                     as follows, for a cubic meter,
                     64(S)
                     22.414(1)
                   999.972(1)    273°(K)^2660g
                     2660%  x  M  =  2660g
                        M3              M2

                     and, for a cubic centimeter (ppm of a cubic
                     meter),
                               7-8

-------
           2660& x 10"6 = 2.66 milligrams/M2

              M2
D =  distance traveled between the two X § Y points
     used to determine the burden, as expressed in
     meters (M)

a =  The sine of the angle between the wind flow
     direction (from pibal results) and distance
     segment, D, used to adjust the segment to a
     position normal to the wind stream.  A non-
     dimensional value between 1 and 0.

v =  The speed of the wind estimated at the altitude
     of the SO? stream as expressed in meters per
     second (M/S).

     (I M/S = 3.6 kilometers per hour = 2.237 miles
     per hour).

K =  A constant to convert to a more uniform, familiar
     unit from the milligrams per second, as calcu-
     lated:

      K               Mass Flux dimensions
     10  * (mg/sec) = grams per second

     3.6 * (mg/sec) = grams per hour

     3.6*10   (mg/sec) = metric tons per hour

     8.64*10"^* (mg/sec) = metric tons per day

     7.93*10" * (mg/sec) = pounds per hour
                   7-9

-------
                         Section 8


                   REMOTE SENSOR RESULTS
TABULATED RESULTS

A total of 465 calculations of S02 mass flux were made for the
COSPEC/Method 6 comparison.  The emission rates, as measured
by each remote sensor, are presented in tabular form in Table 5.
The results are listed individually for every measurement
event that produced valid results.  Key parameters which influ-
enced the measurement and calculation of remote sensor mass
flux are presented in the table.  These include the distance
downwind of the stack at which the measurement was made (meters),
the wind speed (meters per second) and the wind direction  (degrees
from which the wind was blowing).  In addition the date, time
and event number are given, as well as indication of plume touchdown.

The S02 mass flux values in Table 5 show wide variation from one
event to the other; conversely the results show close agreement
between the two or three COSPECs.   These conditions are analyzed
and discussed in Section 10.  It is important to note, however,
that the individual values represent single plume crossings
which lasted from a fraction of a minute to, at the most,  several
minutes in duration.  Because of variations in test conditions--
principally fluctuations in wind speed—great differences  from
one set of flux numbers to the next sometimes occur.  No single
measurement, therefore, is assumed to be a valid estimate  of
S02 stack emissions for comparison with Method 6 or any other
measurement technique.  Rather it is the average of several
such measurements that is taken to be the measure of stack
emissions.  These individual measurements, therefore, are  pre-
sented as a complete listing of the processed COSPEC data  from
which the comparisons of averaged results have been made.

PLOTTED RESULTS

The results for the individual COSPECs are plotted as a function
of time to give a pictorial presentation of all data.  Figure 9
shows SOz Mass Flux plotted separately for the COSPEC II,  COSPEC III
and COSPEC IV.   Each day's data are plotted as a continuous string
even though there were intervals where no measurements were made
or where data were not processed.   Breaks in the continuous plots
do occur for days where a particular instrument was  not being
used for  moving  plume measurement.
                              8-1

-------
CD
  The central  tendency  of  the remote sensor results is expressed
  by the arithmetic means  shown on each plot.  The scatter or dis-
  persion of the  results is  expressed by the standard deviation fT*) ,
  noted on the plots  as a  percent of the mean.  The three means
  agree within 1.5%.  The  standard deviations are large (53.7%-61.4%),
  reflecting the  fluctuations of individual measurements caused by
  variations in wind  speed and other parameters.
                               8-2

-------
CD
            Figure  9


     COSPEC S02 MASS  FLUX

Individual  Plume Calculations

          8-19 May  1975


     (Data  from Table  5)
     x
     co
        150
        100
     co   50
     CN

     O
     CO
                       I	i
             8   9  12   13
                                COSPEC  I I  (S/N  5922)
                         mean = 52.46'

                           0- = 53.7$
14         15


    MAY 1975
                              16
                              19
     ~  150
     O
        100
     3  50
     O
     co
              8  9  12   13
                               COSPEC I I I  (S/N 5932)
             mean=52.99

               f =60.0?
        14
      15


MAY 1975
                      16
19
     x

     _i
     u_

     CO
     co
     CN
     O
     CO
        150
        100
        50
    COSPEC  IV (S/N  6256)

    mean  =  51.76•

         =  61.<
              8   9  12  13
                                         19
                                  8-3

-------
                                                TABLE  5


                                   REMOTE SENSING MEASUREMENT RESULTS

                                   BARRINGER CORRELATION SPECTROMETER

                                    •Individual Plume Calculations
DATE
MAY
1975
8















TIME
(MDT)
0911-0915
0915-0917
0918-0920
0920-0923
0923-0926
0926-0930
0943-0949
0949-0957
0957-1001
1318-1321
1321-1323
1323-1324
1324-1325
1325-1326
1442-1445
1445-1449
1449-1453
1453-1457
EVENT
NO.
5
6
8
9
10
11
14
15
16
34
35
36
37
38
53
54
55
56
DOWNWIND
DISTANCE
(Meters)
120
120
300
300
300
300
350
460
350
40
30
30
30
30
650
350
660
660
PLUME
TOUCHDOWN

-
_
-
-
-
• -
-
-
-
-
-
-
-
-

WIND
SPEED
(m/s)
2.1
2.1
2.1
2.1
2.1
2.1
2.3
2.3
2.3
7.6
7.6
7.6
7.6
7.6
4.9
4.9
4.9
4.9
DIRECTION
(degrees)
141
141
124
124
124
124
117
83
117
329
302
270
260
270
281
227
275
278
SO MASS FLUX (MT/D)
COSPEC
II
_
34.2
73.2
95.4
110.2
59.0
46.3
58.5
22.7
30.0
46.3
77.8
108.8
39.4
76.0
71.0
35.9
27.1
COSPEC
III
57.7
48.6
67.3
69.3
91.6
57.4
46.2
70.3
27.3
33.8
57.4
122.6
79.2
60.3
80.5
103.5
28.4
25. 2
COSPEC
IV
-
—
-
-
-
"
-
-
~
-
~
-
"
-
-

oo
I

-------
                                                TABLE 5
                                   REMOTE SENSING  MEASUREMENT RESULTS
                                   BARRINGER CORRELATION SPECTROMETER
                                      Individual Plume Calculations
DATE
MAY
1975
9








12








TIME
(MDT)
0940-0947
0947-0954
0954-1002
1002-1017
1017-1026
1427-1439
1439-1443
1511-1519
1519-1524
1030-1032
1030-1035
1035-1041
1500-1501
1501-1503
1506-1510
1510-1514
1514-1516
1516-1519
EVENT
NO.
9
10
11
12
13
28
29
31
32
5
6
7
31
32
34
35
36
37
DOWNWIND
DISTANCE
(Meters)
1500
1500
1900
2100
2100
550
800
550
550
710
660
610
50
30
760
680
720
640
PLUME
TOUCHDOWN
No
No
No
Yes
Yes
No
No
No
Yes
No
No
No
No
No
No
No
No
No
WIND
SPEED
(m/s)
4.4
4.4
4.4
4.4
4.4
0.7
0.7
2.1
2.1
14.1
14.1
14.1
14.1
14.1
14.1
14.1
14.1
14.1
DIRECTION
(degrees)
107
107
112
121
121
182
233
184
181
309
305
300
322
303
295
298
305
288
SO MASS FLUX (MT/D)
COSPEC
II
42.8
42.7
60.8
69.2
148.9
19.4
3.5
78.7
60.9
79.3
57.5
83.2
16.5
30.6
13.9
65.8
58.8
87.3
COSPEC
III
41.5
45.7
65.9
83.1
88.7
22.8
10.2
71.7
73.6
86.2
75.5
93.9
14.6
29.2
10.8
56.1
68.0
82.5
COSPEC
IV
_
-
-
-
-
-
-
-
-
83.6
34.9
94.5
8.3
13.7
6.6
61.3
59.9
83.7
CD

-------
                                                TABLE 5


                                   REMOTE SENSING MEASUREMENT RESULTS

                                   BARRINGER CORRELATION SPECTROMETER

                                     Individual Plume Calculations
DATE
MAY
1975
12






13








TIME
(MDT)
1536-1542
1555-1601
1601-1610
1622-1624
1627-1630
1630-1635
1635-1637
1640-1644
0859-0904
0904-0908
0908-0912
0928-0934
0934-0941
0941-0947
0947-0950
1105-1120
1120-1125
EVENT
NO.
42
45
46
50
52
53
54
57
6
7
8
11
12
13
14
18
19
DOWNWIND
DISTANCE
(Meters)
5300
1600
1600
740
750
890
950
330
740
720
650
1200
1100
1100
1200
830
720
PLUME
TOUCHDOWN-
NO
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
Yes
WIND
SPEED
(m/s)
11.1
8.1
8.1
10.1
10.1
10.1
10.1
11.3
2.4
2.4
2.4
5.8
5.8
5.8
5.8
6.3
6.3
DIRECTION
(degrees)
296
287
290
307
292
31'9
322
325
310
309
303
317
322
322
316
336
310
SO MASS FLUX (MT/D)
COSPEC
II
20.4
41.6
32.4
69.4
29.7
29.3
6.3
27.4
32.7
32.8
24.6
61.1
103.4
93.4
106.4
34.2
87.2
COSPEC
III
23.4
42.1
28.6
73.2
47.1
22.5
8.0
23.6
26.2
22.6
27.9
65.1
136.7
104.9
130.1
25.8
54.1
COSPEC
IV
21.4
46.3
31.1
-
'
~
53.1
29.0
24.9
28.5
48.2.
137.7
140.5
141.3
34.0
69.2
oo
I
0\

-------
                                                 TABLE 5



                                    REMOTE  SENSING  MEASUREMENT RESULTS


                                    BARRINGER CORRELATION SPECTROMETER


                                      Individual  Plume Calculations
DATE
MAY
1975
13


14
















TIME
(MDT)
1157-1200
1205-1210
1210-1213
0830-0835
0835-0838
0838-0842
0854-0856
0902-0906
0906-0908
0908-0912
0912-0915
0919-0923
0923-0926
0926-0929
0929-0934
0934-0937
0943-0950
0950-0954
0954-1002
1002-1006
EVENT
NO.
22.5
24
25
5
6
7
9
11
12
13
14
17
18
19
20
21
23
24
25
26
DOWNWIND
DISTANCE
(Meters)
1600
600
600
200
240
200
200
280
220
200
220
410
320
380
380
470
1300
1300
1800
1800
PLUME
TOUCHDOWN
No
No
No
No
No
No
No
No
No
No
Yes
No
No
Yes
Yes
No
No
No
Yes
Yes
WIND
SPEED
(m/s)
8.3
5.3
2.5
3.3
3.3
3.3
3.3
3.3
3.3
3.3
3.3
2.6
2.6
2.6
: 2.6
2.6
5.1
5.1
5.1
5.1
DIRECTION
(degrees)
282
298
298
99
127
99
90
130
122
110
122
112
122
115
115
99
98
100
110
110
SO M.-VSS FLUX (MT/D)
COSPEC
I I
60.1
58.2
55.5
38.9
32.7
68.9
23.3
39.0
45.3
39.7
23.9
17.8
45.1
26.7
35.1
22.7
34.8
16.9
52.0
33.8
COSPEC
III
16.5
63.9
48.7
39.8
43.1
71.6
29.3
41.7
62.4
50.7
13.8
25.8
52.2
31.6
41.9
36.3
46.9
18.0
65.6
43.3
COSPEC
IV
42.1
-
-
36.3
40.7
75.4
29.6
38.9
-
40.9
32.6
19.2
56.5
25.7
46.4
21.3
55.4
17.5
60.4
37.5
oo
i
-j

-------
                                                TABLE 5


                                   REMOTE SENSING MEASUREMENT RESULTS


                                   BARRINGER CORRELATION SPECTROMETER


                                      Individual Plume Calculations
DATE
MAY
1975
14















TIME
(MDT)
1006-1010
1010-1014
1014-1018
1039-1041
1041-1042
1042-1045
1134-1137
1137-1146
1146-1207
1215-1228
1230-1245
1252-1255
1255-1259
1300-1304
1304-1307
1307-1311
EVENT
NO.
27
28
29
34
35
36
40
41
42
44
47
49
50
52
53
54
DOWNWIND
DISTANCE
(Meters)
2100
1800
2000
310
200
260
180
3100
5200
5100
2500
190
140
250
250
250
PLUME
TOUCHDOWN
Yes
No
No
No
No
No
No
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
WIND
SPEED
(m/s)
5.1
5.1
5.1
7.9
4.1
4.1
4.1
2.3
2.3
2.3
1.4
1.2
1.2
1.7
1.7
1.7
DIRECTION
(degrees )
119
110
115
122
110
136
130
120
130
135
133
130
141
147
147
147
SO MASS FLUX (MT/D)
COSPIiC
II
52.9
54.8
47.1
67.3
56.5
46.6
-
-
~
-
-
-
—
-
-

COS PEC
III
75.8
65.4
60.4
81.4
73.0
52.7
58.1
27.2
35. 2
41.3
16.8
27.5
37.1
114.8
24.3
19.1
COSPEC
IV
73.2
65.8
61.7
69.1
35.8
49.0
55.7
26.5
41.2
42.9
32.3
27.9
36.4
88.5
29.2
15.1
oo
I
oo

-------
                                                TABLE 5




                                   REMOTE SENSING MEASUREMENT RESULTS



                                   BARRINGER CORRELATION SPECTROMETER



                                      Individual Plume Calculations
DATE
MAY
1975
14






15












.TIME
(MDT)
1411-1415
1415-1420
1420-1422
1422-1423
1423-1424
1424-1426
1426-1428
0842-0847
0847-0849
0849-0851
0851-0853
0857-0901
0901-0903
0910-0917
0917-0920
0920-0925
0925-0934
1030-1033
1033-1037
1037-1039
EVENT
NO.
57
58
59
60
61
62
63
4
5
6
7
9
10
12
13
14
15
20
21
22
DOWNWIND
DISTANCE
(Meters)
280
240
230
230
240
250
250
200
200
200
200
470
460
1500
1700
1700
3300
220
240
220
PLUME
TOUCHDOWN
Yes
Yes
No
No
No
No
Yes
No
No
No
No
No
No
No
No
No
No
No
No
No
WIND
SPEED
(m/s)
1.3
1.3
1.3
1.3
1.3
1.3
1.3
6.0
6.0
6.0
6.0
11.3
11.3
5.4
5.4
5.4
5.4
6.8
6.8
6.8
DIRECTION
(degrees)
207
157
182
175
154
147
195
90
98
89
92
101
83
109
109
109
103
117
123
117
SO MASS FLUX (MT/D)
•COSPEC
II

-
-
-
-
-
-
_
-
-
-
_
-
_
-
-
-
72.3
44.8
56.5
COSPEC
III
13.7
-
10.7
25.9
20.7
30.3
20.4
_
-
-
-
-
-
_
-
-
-
_
-
-
COSPEC
IV
13.7
54.7
10.3
22.6
17.3
26.7
21.1
45.9
32.2
29.8
23.5
53.4
55.0
20.7
19.2
59.2
45.3
84.4
48.5
52.6
oo

-------
                                                TABLE  5
                                   REMOTE SENSING  MEASUREMENT RESULTS
                                   BARRINGER CORRELATION SPECTROMETER
                                      Individual Plume Calculations
DATE
MAY
1975
- 15


















TIME
(MDT)
1104-1108
1127-1136
1136-1139
1139-1140
1140-1142
1142-1145
1145-1149
1149-1151
1151-1154
1154-1156
1156-1159
1255-1258
1258-1302
1302-1305
1305-1308
1308-1311
1324-1328
1328-1331
1333-1338
EVENT
NO.
25
29
30
31
32
33
34
35
36
37
38
41
42
43
44
45
49
50
52
DOWNWIND
DISTANCE
(Meters)
2200
2200
200
200
280
200
270
240
260
210
200
450
260
270
250
280
180
130
460
PLUME
TOUCHDOWN-
NO
No
No
No
No
No.
No
No
No
No
No
No
Yes
No
No
No .
No
No
Yes
WIND
SPIJ n"
(m/s)
7.3
7.3
5.3
5.3
5.3
5.3
5.3
5.3
5.3
5.3
5.3
2.3
2.3
2.3
2.3
2.3
1.0
1.0
2.2
DIRECTION
(degrees )
126
126
107
107
129
96
133
123
140
112
107
109
138
133
149
129
129
140
83
'SO MASS FLUX (MT/D)
f OS PEC
II
39.4
94.5
98.1 .
54.8
49.7
54.3 .
73.6
93.6
47.5
99.5
59.8
35.9
59.9
15.7
55.0
41.4
-
—
69.3
COSPEC
III

• -
_
-
-
-
-
-
-
--
-
-
-
-
-
-
-
-
-
COSPEC
IV
50.9
66.7
121.8
74.7
60.4
56.1
57.2
119.3
51.3
116.2
39.5
44.3
62.0
22.6
66.4
51.7
29.1
30.7
77.5
oo

-------
                                                TABLE  5
                                   REMOTE  SENSING  MEASUREMENT RESULTS
                                   BARRINGER  CORRELATION  SPECTROMETER
                                     Individual  Plume  Calculations
DATE
MAY
1975
16
















TIME
(MDT)
0853-0858
0858-0900
0900-0902
0902-0904
0904-0906
0906-0908
0908-0912
0912-0914
0914-0916
0916-0923
0923-0932
0932-0937
0937-0945'
1034-1042
1042-1043
1043-1045
1045-1047
EVENT
NO.
2
3
. 4
5
6
7
8
9
10
11
12
13
14
17
18
19
20
DOWNWIND
DISTANCE
(Meters)
200
200
200
200
460
460
400
480
460
1500
1900
2000
1900
80
80
80
90
PLUME
TOUCHDOWN-
NO
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
. . No
WIND
SPEED
O/s)
6.1
6.1
6.1
6.1
8.8
8.8
8.8
8.8
8.8
7.1
7.1
7.1
7.1
6.0
6.0
6.0
; 6 .0
DIRECTION
(degrees )
90
90
80
94
83
92
82
83
98
105
111
117
112
169
169
169
156
SO MASS FLUX (MT/D)
4m
cos PEC
ii
_
-
-

_
-
-
-
-
_
-
-
-
-
-
-

COS PEC
II I
72.9
48.4
49.0
59.4
114.2
45.7
124.5
94.2
57.1
55.6
20.9
65.8
27.2
37.5
46.5
35.3
20.9
CO SPEC
IV
71.8
49.2
41.9
61.7
147.2
39.7 '
146.1
95.2
68.5
62.9
25.8
62.7
29.6
34.7
50.8
33.9
24.0
OO

-------
                                                TABLE 5
                                   REMOTE SENSING MEASUREMENT RESULTS
                                   BARRINGER CORRELATION SPECTROMETER
                                     Individual Plume Calculations
DATE
MAY
1975
16




















TIME
(MDT)
1047-1049
1049-1051
1051-1052
1052-1053
1053-1058
1058-1100
1100-1102
1114-1118,
1118-1123
1123-1126
1126-1129
1129-1132
1240-1243
1243-1244
1244-1247
1247-1250
1310-1313
1313-1317
1317-1319
1319-1322
1322-1324
EVENT
NO. .
21
22
23
24
25
26
27
31
32
33
34
35
39
40
41
42
45
46
47
48
49
DOWNWIND
DISTANCE
(Meters)
240
230
230
240
230
240
230
630
560
580
600
560
60
70
60
120
260
230
240
240
250
PLUME
TOUCHDOWN
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
WIND
SPEED
(TH/S)
6.0
6.0
6.0
6.0
6.0
6.0
6.0
5.0
5.0
5.0
5.0
5.0
14.0
14.0
14.0
10.0
11.3
11.3
11.3
11.3
11.3
DIRECTION
(degrees )
182
165
165
156
158
179
165
153
169
165
158
173
197
169
194
185
199
175
182
190
194
".S0j MASS FLUX (MT/D)
rospiv
1 1

-
'-
-
-
-
-
_
-.
-
-
-
_
-
-
-
-
-
-
-
—
COS PEC
1 1 1
82.6
59.7
46.5
80.0
146.2
55.5
18.2
37.4
34.6
61.4
49.6
21.9
131.2
76.5
85.8
35.2
65.5
79.5
133.2
49.3
129.9
COSPEC
IV
83.5
53.6
50.2
70.4
143.0
67.4
21.9
45.8
31.0
56. .5
45.2
18.6
105.1
81.3
77.2
28.2
78.7
71.8
137.4
54.1
106.6
oo
I

-------
                                                TABLE 5


                                  .REMOTE SENSING MEASUREMENT RESULTS

                                   BARRINGER CORRELATION SPECTROMETER

                                      Individual Plume Calculations
DATE
MAY
1975
16




















TIME
(MDT)
1324-1326
1326-1330
1330-1332
1332-1335
1335-1338
1338-1340
1340-1343
1343-1346
1346-1348
1416-1416
1416-1417
1417-1417
1417-1418
1418-1419
1419-1420
1425-1426
1426-1428
1428-1430
1430-1432
1432-.1433
1433-1435
1435-1436
EVENT
NO.
50
51
52
53
54
55
56
57
58
63
63.5
64
64.5
65
66
68
69
70
71
72
73
74
DOWNWIND
DISTANCE
(Meters)
250
260
240
270
610
640
580
580
600
60
70
70
90
70
100
240
270
250
260
250
280
270
PLUME
TOUCHDOWN
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
WIND
SPEED
(m/s)
11.3
11.3
11.3
11.3
14.6
14.6
14.6
14.6
14.6
7.6
7.6
7.6
7.6
7.6
7.6
5.6
5.6
5.6
5.6
5.6
5.6
5.6
DIRECTION
(degrees)
194
199
190
203
207
212
194
194
197
197
, 179
179
153
182
153
182
2.04
190
199
190
207
203
SO MASS FLUX (MT/D)
COSPEC
II
_
.
-
-
_
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-

COSPEC
III
23.9
50.5
32.6
25.0
104.5
49.4
122.8
106.8
51.7
19.8
71.8
32.3
18.4
15.9
70.6
18.4
29.6
20.7
50.9
9. 2
27.0
24.3
COSPEC
IV
26.1
51.6
21.8
28.5
89.7
53.1
121.1
124.8
38.4
19.7
58.8
30.1
15. 0
15. 5
49. 2
11.5
21.2
23. 9
47.4
1r r\
5. 0
28. 5
22. 9
oo
I

-------
                                                TABLE 5

                                   REMOTE SENSING MEASUREMENT RESULTS
                                   BARRINGER CORRELATION SPECTROMETER
                                      Individual Plume Calculations
DATE
MAY
1975
16



19














TIME
(MDT)
1436-1437
1437-1439
1439-1441
1441-1447
1132-1134
1134-1137
1156-1158
1158-1208
1208-1209
1209-1212
1212-1214
1214-1216
1216-1219
1219-1221
1221-1230
1230-1237
1237-1239
1239-1242
1320-1334
EVENT
NO.
75
76
77
78
8
9
14.5
15
16
16.5
17
18
19
20
21
22
23
24
29
DOWNWIND
DISTANCE
(Meters)
270
300
380
300
460
400
50
30
50
30
370
360
320
280
• 550
550
670
670
330
PLUME
TOUCHDOWN
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
No
WINR
SPEED
O/s)
5.6
5.6
5.6
5.6
3.6
3.6
5.3
5.3
5.3
5.3
2.1
2.1
2.1
2.1
2.1
2.1
2.1
2.1
3.3
DIRECTION
(degrees )
203
209
236
212
244
225
205
228
201
269
234
228
216
205
251
251
269
273
220
SO MASS FLUX (MT/D)
COS PEC
II

-
-
-
99.7
51.5
27.4
30.7
60.2
46.1
8.3
27.6
55.2
30.9
19.3
14.9
33.1
16.0
44.8
i 'OS PEC
III
23.3
33.0
83.6
26.8
92.8
45.5
23.4
45.5
58.0
60.9
9.8
34.1
74.8
26.9
13.1
10.0
28.4
13.5
64.1
COSPEC
IV
21.8
31.5
89.0
36.5
-
- -
-
-
-
.
-
-
-
-
-
-
-
~

oo

-------
                                                TABLE 5
                                   REMOTE SENSING MEASUREMENT RESULTS
                                   BARRINGER CORRELATION SPECTROMETER
                                      Individual Plume Calculations
DATE
MAY
1975
19










TIME
(MDT)
1507-1511
1511-1513
1513-1516
1521-1526
1526-1532
1534-1537
1537-1547
1547-1555
1555-1607
1607-1612
1618-1622
EVENT
NO.
38
39
40
42
43
' 44
45
46
47
48
49
DOWNWIND
DISTANCE
(Meters)
320
310
420
590
590
520
1300
1300
1200
1400
400
PLUME
TOUCHDOWN-
NO
No
No
No
No
Yes
Yes
Yes
Yes
Yes
No
WIND
SPEED
O/s)
5.6
5.6
5.6
5.6
5.6
5.6
6.0
6.0
6.0
6.0
9.7
DIRECTION
(degrees )
218
216
242
256
256
248
231
228
214
243
240
SO MASS FLUX (MT/D)
COSPEC
II
25.2
61.1
56.4
36.5
58.2
105.0
111.8
21.2
58.4
87.0
143.6
COSPEC
III
28.7
63.9
80.6
28.2
62.7
133.7
113.6
25.5
62.3
86.5
135.1
COSPEC
IV

-
-
-
-
-
•-
-
-
-

oo

-------
                         Section 9


                   STACK SAMPLING RESULTS
Pertinent sections of the results of the stack sampling tests,
as reported by Pacific Environmental Services Inc.  to Environ-
mental Measurements Inc. have been extracted to form Appendix A.
They include the following:

     «  Test procedures and analytical methods

     o  Sulfur balance

     o  Equations and sample calculations

The information contained in Appendix A shows how the calcula-
tions for S02 stack concentrations and mass flux were made.

TABULATED RESULTS

A tabulation of all in-stack measurement results is presented
in Table 6.   Listed are the stack gas parameters (molecular
weight, velocity and flow rate), and sulfur  dioxide concentra-
tion (pounds per cubic foot, and ppm) and mass flux (in metric
tons per day) for each of the 35 reported runs.  The date and
time for each run number are given also.

PLOTTED RESULTS

The Method 6/Method 2 results are presented in graphic form in
Figure 10.  This plot shows the S02 mass flux calculations for all
35 runs as continuous strings of data even though runs were
irregularly spaced in time and a different number of stack tests
were performed each day.  The standard deviation of the in-stack
results is 15.2 percent (of the mean).

SULFUR BALANCE

PES performed a sulfur balance on the target power plant combus-
tion process and made other calculations to estimate the total
S02 emissions that could be expected from the pulverized coal-
fired power plant.  This analysis is presented in full in Appendix A.
As described, the Method 6 S02 determination can be expected to
be as much as 15% low.  This is principally thought to be due to
losses between the inner and outer stacks.  The samples extracted
by the sampling train were withdrawn by the probe inserted into
the inner refractory stack; no in-the-field measurements or esti-
mates were made of the quantities of S02 flowing in the annular
space between this stack and the outer mechanical supporting stack.
The SOo mass flux values presented in Table 6, therefore, are con-
sidered raw data.  The analysis performed in Section 10 adjusts the
in-stack results to allow for the apparent 15% unmeasured emissions.
                              9-1

-------
CROSS-STACK MONITOR

A cross-stack continuous monitor was in place and operating at
the target power plant during the May 1975 measurements.
Sulfur dioxide concentration data as measured by this instru-
ment were turned over to the contractor by the power plant per-
sonnel.  The data were in the form of daily computer printout
logs with various power plant operation parameters and output
variables.  At this writing there is uncertainty about the
correlation between the in-stack monitor results and the Method 6
results.  Extensive checking of the raw data and verification of
correction factors for moisture and percent oxygen must be per-
formed.  Therefore, neither results of the cross-stack measure-
ments nor comparisons with the COSPEC results are presented in
this report.
                              9-2

-------
(D
                              Figure  10


                    STACK SAMPLING SC>2  MASS  FLUX
                      Individual  Run  Calculations
                             13-20 May  1975


                           (Data  from Table  6)
      80
                            METHOD 6/METHOD 2
      40
   to
   en
   (N
   o
   co
        mean = 44.25

          f = 15.2?
           13
14
15
19
20
                              MAY  1975
                                9-3

-------
          TABLE  6
IN-STACK MEASUREMENT RESULTS
   EPA METHODS 2,  4  and 6
 Individual Run Calculations
DATE
MAY
1975
13



14





15







TIME
(MDT)
0850-0910
1104-1124
1150-1210
1330-1350
0900-0920
0945-1005
1125-1145
1212-1232
1350-1410
1430-1450
0835-0855
0915-0935
1035-1055
1115-1135
1240-1300
1320-1340
1445-1505
1525-1545
RUN
NO.
15
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
STACK GAS
•MOL. WT.
(Ib/lb-mole)
DRY
30.18
30.18
30.18
30.18
30.24
30.24
30.24
30.24
30.24
30.24
30.26
30.26
30.26
30.26
30.26
30.26
30.26
30.26
WET
29.31
29.31
29.31
29.31
29.58
29.58
29.58
29.58
29.5'8
29.58
29.57
29.57
29.57
29.57
29.57
29.57
29.57
29.57
VELOCITY
(ft/sec)
51.26
60.14
41.82
44.96
36.47
49.69
46.43
45.14
43.74
45.27
52.99
55.98
46.55
47.51
46.62
42.65
40.02
47.32
FLOW RATE
106(ftVhr)
38.8
44;9
31.5
33.6
27.6
37.3
34.6
33.4
31.3
33.3
39.7
41.8
34.7
35.2
34.4
31.5
29.3
34.7
SULFUR DIOXIDE
CONCENTRATION
10'5(lb/ft3)
9.35
6.30
6.74
13.1
13.6
11.4
10.7
12.0
13.1
11.4
10.9
11.3
12.8
13.0
11.6
13.3
13.9
13.4
(ppm)
568
396
408
792
822
689
652
729
785
690
661
691
774
790
699
806
841
812
MASS
FLUX
(MT/D)
39.5
30.7
22.9
47.9
40.8
46.2
40.3
43.6
44.7
41.3
47.2
51.5
48.3
49.8
43.1
45.5
44.3
50.5

-------
                                                TABLE  6


                                      IN-STACK  MEASUREMENT  RESULTS


                                        EPA METHODS 2,  4 and 6


                                      Individual Run Calculations
e
DATE

MAY
1975
16






19





20




TIME
fMDTI
V.1 liy A J
0830-0850
0910-0930
1030-1050
1110-1130
1235-1255
1315-1335
1415-1435
1116-1136
1155-1215
1315-1335
1355-1415
1505-1525
1545-1605
0820-0840
0900-0920
1015-1035
1055-1115

RUN
NO

34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
STACK GAS
MOL. WT.
(Ib/lb-mole)
DRY
30.46
30.46
30.46
30.46
30.46
30.46
30.46
30.44
30.44
30.44
30.44
30.44
30.44
30.42
30.42
30.42
30.42
WET
29.75
29.75
29.27
29.75
29.75
29.75
29.75
29.73
29.73
29.73
29.73
29.73
29.73
29.72
29.72
29.72
29.72

VELOCITY
(ft/sec)
42.21
33.40
42.04
44.71
36.86
36.86
44.55
47.04
44.03
53.87
50.40
47.60
55.55
54.30
53.46
47.32
41.64

FLOW RATE
106(£t3/hr)
31.5
24.8
30.8
32.7
27.0
27.0
32.7
34.6
32.4
39.5
37.0
34.9
40.7
39.9
39.2
34.5
30.4
SULFUR DIOXIDE

CONCENTRATION
10-5(lb/£t3)
14.0
12.3
12.8
12.1
12.3
12.6
11.6
13.1
12.7
11.8
11.0
12.2
11.5
13.1
. 12.8
13.0
12.6
(ppm)
849
746
775
736
748
762
704
792
771
715
666
739
699
791
774
786
761
MASS
FLUX
(MT/D)
47.9
33.3
42.9
43.1
36.2
37.1
41.3
49.4
45.0
50.6
44.3
46.4
50.9
56.9
54.6
48.9
41.7
to
I
en

-------

-------
CD
   gases was not measured by  Method  6.  The  conclusion drawn by
   the subcontractor is  that  the  S02  gas  in  the annular space be-
   tween the inner refracting stack  and the  outer structural stack
   was emitted into the  atmosphere undetected by the stack sampling
   technique.   Sulfur entering the combustion process also could
   have left as 862 through other leakage paths or as sulfur in the
   ash.

   The remote  sensing method  could be expected to measure both the
   central  stack S02 emissions and some portion of the intra-stack
   and other leakage S02  emissions.   To perform a meaningful analysis
   of the relative accuracy between  the two  methods the in-stack
   mass flux calculations  were adjusted to add the 15%.  This was
   done by  the formula:

   Adjusted In-Stack S02  Mass  Flux    Raw In-Stack S02 Mass Flux

                1001                           85%

                         ADJUSTED  =  1.18  x RAW

   The raw in-stack flux calculations (reported in Section 9) were
   multiplied  by 1.18; the adjusted S02 mass flux values are listed
   in Tables 7,  8 and 9.
                               10-2

-------
                                           TABLE  7




                                    COMPARISON OF  RESULTS



                              REMOTE SENSING AND STACK SAMPLING



                                     "20-MINUTE AVERAGES"
DATE
MAY
1975
8



9
•


12



13




TIME
(MDT)
0911-0930
0943-1001
1318-1326
1442-1457
0940-1002
1002-1026
1427-1443
1511-1524
1030-1041
1500-1519
1536-1610
1622-1644
0859-0912
0928-0950
1105-1125
1157-1213

REMOTE SENSING
NUMBER
OF
EVENTS
6
3
5
4
3
2
2
2
3
6
3
5
3
4
2
3

AVERAGE S02 MASS FLUX (MT/D)
COS PEC
II
74.4
42.5
60.5
52.5
48.8
109.1
11.5
69.8
73.3
45.5
31.5
32.4
30.0
91.1
60.7
57.9

COSPEC
III
65.3
47.9
70.7
59.4
. 51.0
85.9
16.5
72.7
85.2
43.5
31.4
34.9
25.6
109.2
40.0
43.0

COSPEC
IV
.
_
-
-
-
-
-
-
71.0
38.9
32.9
-
27.5
116.9
51.6
*

STACK SAMPLING
TIME
(MDT)
„
_
-
-
-
T
-
-
-
-
-
-
0850-0910
-
1104-1124
1150-1210
1330-1350
RUN
NO.
.
_
-
-
-
-
-
-
-
-
-
-
15
-
17
18
19
ADJUSTED S02
MASS FLUX
(MT/D)
_
_
-
-
_
-
-
-
-
-
-
-
46.6
-
36.2
27.0
56.5
*Insu££icient data for averaging,

-------
                                                TABLE  7


                                         COMPARISON  OF RESULTS

                                   REMOTE SENSING AND STACK SAMPLING


                                         "20-MINUTE  AVERAGES"
DATE
MAY
1975
14








15







16




TIME
(MDT)
0830-0856
0902-0937
0943-1018
1039-1045
1134-1207
1215-1245
1252-1311
1411-1428
-
0842-0903
0910-0934
1030-1108
1127-1159
1255-1311
1324-1338
_
-
0853-0908
0908-0945
1034-1102
1114-1132
1240-1313
REMOTE SENSING
NUMBER
OF
EVENTS
4
9
7
3
3
2
5
7
-
6
4
4
10
5
3
_
-
6
7
11
5
5
AVERAGE S02 MASS FLUX (MT/D)
COSPEC
II
41.0
32.8
41.8
56.8
• . -
-
-
-
-
-
-
53.3
72.5
41.6
*
_
-
_
_
-
-
-
COSPEC
III
46.0
39.6
53.6
69.0
40.2
29.1
44.6
20.3
-
-
-
-
-
-
-
-
-
64.9
63.6
57.2
41.0
78.8
COSPEC
IV
45.5
35.2
53.1
51.3
41.1
37.6
39.4
23.8

40.0
36.1
59.1
76.3
49.4
45.8
-
-
68.6
70.1
57.6
39.4
74.1
STACK SAMPLING
TIME
(MDT)
_
0900-0920
0945-1005
-
1125-1145
1212-1232
-
1350-1410
1430-1450
0835-0855
0915-0935
1035-1055
1115-1135
1240-1300
1320-1340
1445-1505
1525-1545
0830-0850
0910-0930
1030-1050
1110-1130
1235-1255
RUN
NO.
_
20
21
-
22
23
-
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
ADJUSTED S02
MASS FLUX
(MT/D)
—
48.1
54.5
-
47.6
51.4
-
52.7
48.7
55.7
60.8
57.0
58.8
50.9
53.7
52.3
59.6
56.5
39.3
50.6
50.9
42.7
o
 I
    *Insufficient data for averaging.

-------
                                                TABLE  7


                                         COMPARISON OF RESULTS


                                   REMOTE SENSING AND STACK SAMPLING


                                         "20-MINUTE AVERAGES"

DATE

MAY

1975
16

19






20





11 Ml.

(MIH )

1313-1348
1416-1447
1132-1137
1156-1221
1221-1242
1320-1334
-
1507-1537
1537-1612
.
_
.

REMOTE SENSING
NUMBER

OF

EVENTS
13
17
2
8
4
1
-
6
4
_
_
_

AVERAGE S02 MASS FLUX (MT/D)


COSPEC

II

-
75.6
35.8
20.8
*
-
57.1
69.6
_
_
_


COSPEC

III
73.8
33.9
69.2
41.7
16.3
*
-
66.3
72.0
_
_
_


COSPEC

IV
71.2
31.6
_
-
-
*
-
-
-
_
_
-

STACK SAMPLING

TIME

(MDT)

1315-1335
1415-1435
1116-1136
1155-1215
-
1315-1335
1355-1415
1505-1525
1545-1605
0820-0840
0900-0920
1015-1035
1055-1115

RUN

NO.

39
40
41
42
-
43
44
45
46
47
48
49
50
ADJUSTED S02

MASS FLUX

(MT/D)
43.8
48.7
58.3
53.1
-
59.7
52.3
54.8
60.1
67.1
64.4
57.7
49.2
o
I
     *Insu£ficient data for averaging.

-------
                                                TABLE 7


                                         COMPARISON  OF  RESULTS

                                   REMOTE SENSING  AND STACK SAMPLING


                                         "60-MINUTE AVERAGES"

DATE

MAY

1975
8

9

12

13


14


15





TIME

(MDT)

0911-1001
1318-1457
0940-1026
1427-1524
1500-1542
1555-1644
0859-0950
1105-1213

0902-1018
1134-1245
1252-1428
0842-0934
1030-1159
1255-1338

REMOTE SENSING
NUMBER

OF

EVENTS
9
9
5
4
7
7
7
5

16
5
12
10
14
8

AVERAGE S02 MASS FLUX (MT/D)


COSPEC

II
62.4
56.9
72.9
40.7
41.9
33.7
64.9
59.0

36.7
-
-
-
67.0
*


COSPEC

III
59.5
65.7
65.0
44.6
40.6
35.0
73.4
41.8 .

45.7
35.8
31.3
-
-
-


COSPEC

IV

-
_
-
36.4
*
78.6
48.4

43.6
39.7
30.3
38.4
71.4
48.1

STACK SA.MPl.-fNG

TIME

(MDT)


-
_
-
_

0850-0910
1104-1210
1330-1350
0900-1005
1125-1232
1350-1450
0835-0935
1035-1135
1240-1340
1445-1505

RUN

NO.


-
_
-
_

15
17,18
19
20,21
22,23
24,25
26,27
28,29
30,31
32,33
\MJMSTED S02
AVERAGE
MASS FLUX

(MT/D)

-
_
-.
_

*
31.6
*
51.3
49.5
50.7
58.3
57.9
52.3
56.0
o
I
     *Insu£ficient  data  for  averaging.

-------
                                               TABLE   8



                                        COMPARISON OF  RESULTS


                                   REMOTE SENSING AND STACK  SAMPLING



                                        "60-MINUTE AVERAGES"
DATE
MAY
1975
16



19


20

TIME
(MDT)
0853-0945
1034-1132
1240-1348
1416-1447
1132-1221
1320-1334
1507-1612
_

REMOTE SENSING
NUMBER
OF
EVENTS
13
16
18
17
10
1
10
_

AVERAGE S02 MASS FLUX (MT/D)
COSPEC
II
.
-
-
-
43.8
*
, 62.1
_

COSPEC
III
64.2
52.1
75.2
*
47. 2
*
68.6
_

COSPEC
IV
69.4
51.9
72.0
*
_
*
-
-

STACK SAMPLING
TIME
(MDT)
0830-0930
1030-1130
1235-1335
1415-1435
1116-1215
1315-1415
1505-1605
0820-0920
1015-1115
RUN
NO.
34,35
36,37
38,39
40
41,42
43,44
45,46
47,48
49,50
ADJUSTED SO 2
AVERAGE
MASS FLUX
(MT/D)
47.9
50.8
43.3
*
55.7
56.0
57.5
65.8
53.5
o
 I
    *Insufficient data for averaging.

-------
oo
                                                TABLE 9




                                         COMPARISON OF RESULTS



                                   REMOTE SENSING AND STACK SAMPLING



                                          "DAILY AVERAGES"
DATE
MAY
1975
8
9
12
13
14
15
16
19
20
TIME
(MDT)
0911-1457
0940-1524
1030-1644
0859-1213
0830-1428
0842-1338
0853-1447
1132-1612

REMOTE SENSING
NUMBER
OF
EVENTS
18
9
17
12
40
32
64
25

AVERAGE S02 MASS FLUX (MT/D)*
COSPEC
II
57.5
59.8
45.7
59.9
43.1
55.8
-
51.8

COSPEC
III
60.8
56.5
48.8
54.5
42.8
-
59.0
53.1

COSPEC
IV
-
-
47.6
65.3
40.9
51.1
58.9
-

STACK SAMP I. INC,
TIME
(MDT)
-
-
-
0850-1350
0900-1450
0835-1545
0830-1435
1116-1605
0820-1115
RUN
NO.
-
-
-
4
6
8
7
6
4
ADJUSTED S02
AVERAGE
MASS FLUX
(MT/D)
-
-
-
41.6
50.5
56.1
47.5
56.4.
59.6

-------

-------
S02 MASS FLUX (MT/D)

   50        100
                                    150
 x
 CO
 co
 CM
 o
 CO
X
CO
CO
 CN
o
co
CO
CO
CN
o
CO
               50         100

           S02 MASS FLUX (MT/D)

                          •"0-10
150
100

50

0
150
100

50

0
150
100
50


n
•
o
0.
co
o
o
•
.
r = +.896
. • -"
COSPEC 1 1 1

• •
o
0-
co
O
0 ' .
m .. '" r = +.933
COSPEC IV

•
•
0
LU
Q.
CO
. 8
/ *
.'" ' r = +.951
COSPEC IV
50
       INSTRUMENT COMPARISONS



            Figure n

            COSPEC II
               vs
            COSPEC III

      "20-Minute  Averages"

         (Data from Table  7)
            Figure  12

            COSPEC  II
               vs
            COSPEC  IV

      "20-Minute Averages"

        (Data from  Table 7)
            Figure  13

            COSPEC III
               vs
            COSPEC IV

      '20-Minute Averages"

        (Data from Table  7)

-------
CD
    REMOTE/IN-STACK COMPARISONS

    The stack sampling mass flux values are a combination of
    Method 6  (concentration) and Method 2 (flow rate)  data,  just as
    the remote sensing mass flux values are a combination of COSPEC
    (concentration) and pibal (wind speed)  data.   For  clarity the
    comparison between the remote sensing and stack sampling results
    is labelled "COSPEC-Pibal versus Method 6-Method 2".

    Graphic Comparisons.   Comparison of the remote sensor and stack
    sampling S02 emission rates are shown graphically  in  Figures 14,
    15 and 16.  These are scatter plots of the "20-Minute Averages",
    "60-Minute Averages"  and "Daily Averages" for all  COSPEC II, III
    and IV readings for which there are Method 6  data  over the same
    time period (see Tables 7, 8 and 9).

    Generally, as the averaging time increases from "20-Minute" to
    "Daily" the scatter of the data is reduced.   However, for the
    "Daily Averages" there appears to be one set  of anomalous data:
    that of 13 May.  The   three   points above and to the  left of
    the main cluster of points suggests that either there was some-
    thing unusual about the power plant operation on that day or
    the in-stack results  were erroneously low.  The subcontractor
    has observed that the 13 May results are low  (see  Appendix A).
    Because the Method 6-Method 2 data for 13 May show significantly
    different results they were discarded from further analysis.
                                 10-11

-------
X
=3
CO
CO
 CN
o
CO
x
CO
CO
CM
o
CO
X
CO
CO
 CM
o
CO
    100
     50
      0


    100
     50
     0


    100
     50
          m
 o
 UJ
 CL
 CO
 O
 0
          o
          LU
          D.
          CO
         CD
O
UJ
CL
CO
O
             S02  MASS FLUX  (MT/D)

                      50
                METHOD 6-METHOD 2
                METHOD 6-METHOD 2
                 METHOD 6-METHOD 2
                     50

             S02 MASS  FLUX (MT/D)


                          10-12
                           100
                           100
REMOTE/IN-STACK COMPARISONS

     (THREE-COSPEC DATA)



       Figure  14

  COSPECs  II,  III,  IV
            vs
   METHOD  6-METHOD 2

  "20-Minute Averages"

     (Data  from Table 7)
                                       Figure  15

                                 COSPECs  II,  III,  IV
                                           vs
                                   METHOD 6-METHOD  2

                                 "60-Minute Averages"

                                  (Data from Table  -8)
       Figure  16

  COSPECs II,  III,  IV
            vs
   METHOD 6-METHOD 2

    "Daily Averages"

   (Data from  Table 9)

-------
CD
   Comparison of Means.  To determine whether the means .of the
   COSPEC-Pibal results were the same as the means of the Method
   6-Method 2 results a statistical analysis was performed on the
   paired data for 14, 15, 16 and 19 May 1975.  The 20-minute average
   remote sensor data were used (Table 7) because the Method 6
   runs were 20 minutes long.  (The 20 minute samples are consid-
   dered the fundamental measurement for both methods for this and
   succeeding analyses.)  Confidence limits were calculated for the
   four days of data using the Student-t distribution.

   The confidence limit analysis determined the following:  how the
   mean of the remote S02 mass flux differs with respect to the
   mean of the in-stack S02 mass flux.  The test was performed for
   each COSPEC using the 20-minute samples; three days of concur-
   rent measurements occured for each instrument.

   These confidence limits were calculated from the following for-
   mula:
        M(L,U)

        where:
 M

(L

(U

 z
               TA,V

                 <>z
                  A

                  v

                 ' n
z±(TA,v * (Tz)

confidence limit (at level A)

lower limit)

upper limit)

percent difference between remote and
in-stack fluxes

average percent difference between remote
and in-stack, in \ of Method 6-Method 2
flux value.

Student-t at level A, with v degrees of
freedom (from statistical tables)
          Standard  deviation  of percent  differences,z

          Confidence  level  (901)

          Degrees of  freedom  (n-1)

          Number of samples
                              10-13

-------
The 901 confidence limits are tabulated in Table 10.  The limits,
expressed in percent of the EPA reference method, are given for
each COSPEC.  The number of samples associated with each result
is shown, also.
                            Table 10


                       CONFIDENCE LIMITS


       Confidence Intervals at the 901 Confidence Level

 on the Percent Difference between COSPEC-Pibal Mean S02 Flux

              and Method 6/Method 2 Mean SO  Flux

                        14-19 May 1975


Confidence
Limits (%)
Number of
Samples (n)
COSPEC II
vs
In-Stack
+10.31
-19.1$
9
COSPEC III
vs
In-Stack
+23.5%
-12.11
16
COSPEC IV
vs
In-Stack
+16.8%
-15.31
18
Based upon the data generated in this study, the upper and lower
limits of the 90% confidence interval for the percent difference
between the COSPEC-Pibal SOo mass flux (for a given instrument)
and the Method 6/Method 2 562 mass flux are given by the "plus"
and "minus" percentage figures in the table.  Using COSPEC II as
an example, the true percent difference between the remote mass
flux and the in-stack mass flux lies between +10.3% and -19.1%,
unless an event of probability 10% or less has occurred.

The confidence limits for all three COSPECs are similar even
though the number of 20-minute samples (n) ranges from 9 to 18.
They fall within inclusive limits of  +24%.  Thus the conclusion
is reached that at the 90% confidence level, the actual percent
difference, in the very long-term average, between results obtained
using the remote sensing technique utilized in this program and
the "true" (i.e., as determined by Method 6) mass flux from a
source is within ±24%.
                            10-14

-------
Comparison of Dispersions.  A further analysis of the paired 20-
minute samples of remote and in-stack flux results was made to
compare the dispersion or spread of the two types of data.  A
separate analysis was done for each of the three COSPECs tested.
The instruments were used in the moving measurment mode over dif-
ferent periods so that unique pairings of COSPEC/Method 6 data
(from 9 to 18) result.  Thus, only 20-minute samples were used
for which there were comparative measurements.

The means and standard deviations of the COSPEC-Pibal and Method 6/
Method 2 results are given in Table 11.  The means for both methods
range from 50.6 to 55.1 MT/D, indicating uniformity of results.
The standard deviations (0"), on the other hand, differ between
methods even though they are consistent for each method.  The
remote sensing (P is 30.71-33.8% compared to 7.11-11.21 for the
in-stack
-------
It should be noted that the variability in the values of 
-------
CD
  WIND SPEED VARIATIONS

  The impact of wind speed variations on the  COSPEC SO,.,  mass  flux
  calculations is easily illustrated.  The mass  flux is  directly
  proportional to the wind speed (See Section 7);  a 10%  error in
  determining the speed of the wind at the plume  elevation  causes
  a 10% error in the calculated flux.  When available, pibal  results
  were averaged to determine a representative wind speed during a
  COSPEC plume measurement;  often only a single  pibal was tracked
  during as many as eight events.   Thus, any  fluctuations in  wind
  speed before and after the pibal were unknown.

  An example of a set of flux calculations based  on a single  pibal
  observation is presented in Table 12.  The  fluxes are  extracted
  from Table 5 (Section 8);  they are reproduced  here to  illustrate
  that the lack of sufficient winds-aloft data clearly affected
  the COSPEC results.

                             Table 12

               COSPEC-Pibal  S02 Mass Flux Variations

                            19 May 1975

                        COSPEC II S/N 5922
Event
No.
17
18
19
20
21
22
23
24 '
Time
(MDT)
1212-1214
1214-1216
1216-1219
1219-1221
1221-1230
1230-1237
1237-1239
1238-1242
S07 Mass Flux
L (MT/D)*
8.3
27.6
55.2
30.9
19.3
14.9
33.1
16.0
       *at 2.1  M/S  wind  speed
                               10-17

-------
(D
  There was one pibal that was tracked during these 30-minutes
  of COSPEC-Pibal plume measurements.   Knowing that the nominal
  Method 6-Method 2 S0'2 mass flux was  about 50.MT/D,  the time of
  the pibal observation can be deduced from'the tabulated fluxes.
  (The COSPEC flux(es)  closest to the  nominal flux would be near- o
  est the time of upper air measurements).   In fact,  the pibal
  used to calculate these eight fluxes was  released at 1219 MDT--
  as could be estimated from the time  of Event 19; this flux was cal-
  culated at 55.2 MT/D.  This means that the 2.1 m/s  wind speed was
  "correct" for Event 19 but was low for most of the  remaining
  events.  If more pibals had been run during this period the
  remaining seven fluxes would probably change upward.  The surface
  winds (see Appendix B) showed a t60% fluctuation about a mean of
  5 m/s; the pibal" measured wind speed (2.1 m/s) was  not repre-
  sentative for this measurement period.

  REMOTE SENSOR PRECISION

  The previous analysis implys that the COSPEC remote S02 flux
  measurements differ in the long term average by no  more than
  24% from the Method 6 results, at the 901 confidence level.
  However,  what is the  precision of the remote sensing technique?
  The precision is essentially the accuracy with which an estimate
  can be made of the true mean value of the COSPEC-determined fluxes,
  using only a finite number (n) of actual  COSPEC plume measurements.

  To develop the relationship between  the error interval and the
  required number of measurements requires  the following assump-
  tion:   In the long-term average the  percent difference between
  COSPEC and Method 6 flow estimates is zero or, if not identically
  zero,  within a few percent of zero.   The  preceeding analysis,
  while not supporting  this assumption directly, gives no reason
  to discard it.  Further, it is certainly  true that  adding more
  data to that gathered under this project  would permit the narrow-
  ing of the confidence intervals given in  Table 10.   Even if this
  should ultimately show that there is some small percentage off-
  set between the COSPEC and Method 6  results, this offset would then
  be accurately known and could, therefore, be accounted for in pro-
  cessing COSPEC flux calculations. The assumption of zero percent
  difference is one primarily of convenience and is not crucial to
  the following discussion of precision.

  In general, if a statistical population has a standard deviation
  ((T) , then the standard deviation of  the sample mean is (T/^fn1 where
  n is the number of samples.  (Thus,  for example, to double the
  precision of an estimate it is necessary  to quadruple the sample
  size).  For the purpose of determining precision it is helpful
  to assume that the sample mean follows a  Gaussian distribution
  (the Central Limit Theorem insures this as the sample size increa-
  ses) ;  then it is possible to specify the  actual probability of
                               10-18

-------
the sample average deviating from the true population mean by
more than a specified amount.  This well known relationship is
briefly summarized as follows:

          The probability of a Gaussian variate (where
          standard deviation is 
-------
(D
                             Table 13

                      COSPEC-PIBAL PRECISION

                    Number of 20-Minute Samples

            Required to Achieve a Given Error Interval
Error
Interval

±25%


±20%

Confidence
Level
90S
95%
991
901
95%
99%
COSPEC II
S/N 5922
3
4
6
4
6
9
COSPEC III
S/N 5932
7
11
16
11
17
26
COSPEC IV
S/N 5956
7
10
15
10
16
24
  From the table come the following  results:   to  achieve ±25%  error
  in remote sensing S02 Mass  Flux data 95%  of  the time, 4,11  or  10
  samples, each 20-minutes long,  should be  gathered.   Eleven  20-min-
  ute samples represents a one-half  day (4-hour)  measurement  period
  or about 55 plume crossings.   To raise the confidence level  to
  99 %  the required number of samples  increases  to 16  maximum, or
  5 hours of measurements (80 plume  crossings).

  A significant finding here  is  the  fewer number  of required  20-
  minute samples for COSPEC II  contrasted to COSPEC III and  IV.
  Given the higher noise level  of the  COSPEC II  (and hence lower
  sensitivity)  the opposite would be expected.   However, this  dif-
  ference may be caused by narrower  statistical  spread of  in-stack
  results on days when the COSPEC II was operational and wider spread
  on COSPEC III and IV days (as  previously  noted,  see  Table  11)  and
  not by remote sensing variations.

  Table 13 provides a means of  determining  how much COSPEC measuring
  is needed to achieve an accuracy comparable  to  other stack  monitor-
  ing techniques.  Using the  worst case  figures  (COSPEC III  and  IV),
  ±20% accuracy can be obtained  from a single  day (6 hours,  85 plume
  crossings) of data gathering  for the 95%  confidence  level.   The most
  favorable result (COSPEC II)  indicates only  two  hours of data are
  needed or about 30 plume crossings.   In summary, the COSPEC-Pibal
  measurement system when used  for two-to-six  hours produces  SO
  mass flux results with ±20% accuracy.   Longer  measurement  periods
  would improve the accuracy  and/or  improve the  confidence in  the
  remote sensing data.
                               10-20

-------
             APPENDIX  A
     AIR POLLUTION  EMISSION TEST
               REPORT
               (FINAL)


       COAL FIRED POWER  PLANT
            PREPARED FOR
  ENVIRONMENTAL MEASUREMENTS,  INC.
          PES JOB  NO.  088
               8/12/75
PACIFIC ENVIRONMENTAL  SERVICES, IMC,
          1930  14th  Street
  ' Santa Monica,  California  90404

  Robert J.  Bryan,'Project Manager
             A-l

-------
 I,    INTRODUCTION

      This test was conducted on a pulverized coal fired power plant
 in the southwestern United States.  The testing took place over the
 period of May 6-20, 1975.  The sampling location was in the stack
 following the collection device and blowers.
      Sampling was conducted for sulfur dioxide using EPA Method 6.
 Velocity traverses were conducted using both EPA Method 2 and a modi-
 fied EPA Method 2.  A moisture determination was also performed using
 EPA Method 4.
      The source test team consisted of Joseph  Boyd, John Stevenson
 and Robert Norton, all of whom are employed by Pacific Environmental
 Services, Inc.
      The purpose of the test was to supply extractive stack test
 data on SO  for comparison with remote sensing data obtained with
 Barringer correlation spectrometer.  The remote sensing was performed
 by Environmental 'Measurement Inc.

 II,    SULFUR BALANCE

      An attempt has been made to perform a sulfur balance based upon
 information supplied by the operator of the power plant and stack
 sulfur emission data.   In a rigorous approach it would be necessary to
 know the exact quantity of sulfur entering the combustion process
 through fuel, to identify all exit pathways, and to determine the
 quantity of the sulfur leaving the process.  In this case information
was supplied by the operator on coal firing rates and to a limited
 extent on coal composition.  Coal firing rate data were obtained from
 a computer print-out showing hourly and cumulative quantities of coal
 fired for each day.  Composition data were supplied on a shift basis
 for sulfur, ash and moisture content on a weight basis.  Samples were
 obtained from the coal pulverizers.  These data have some shortcomings
                                A-2

-------
in that (1) only total sulfur is shown whereas it may be present in
organic, pyritic, or  sulfate content.  Thus exact combustion reactions
cannot be written.  Furthermore, some small amount of sulfur originally
present in non-sulfate form may be converted to SCL which is not
determined in the Method No. 6 Procecure.
      Two approaches were used to examine the sulfur balance in view
of the above limitations.  The first approach involved the direct
comparison of the average daily charge of total sulfur to the boiler
(as obtained from the operator) with the quantity measured leaving
through the stack.  Figure 1 illustrates a possible set of in and out
pathways.   The only actual data available are that for sulfur entering
in the coal and that leaving in the stack gases as SO .   No specific
data are available on the combustible sulfur portion of the total
sulfur charged, the amount present in the ash, the amount present as
non-SO. sulfur in the stack gases and the amount of sulfur in leaks.
      Because no ultimate analysis was available on the coal, analyses
of six other coals originating from the same general area were examined.
From these analyses an ultimate analysis was- hypothesized for the coal
average determined over a three week period from May 1 to May 21, 1975.
Ash and moisture content were similarly selected.  The following assumed
analysis was used in further calculations.

                Constituent                   Weight %
                     C                         59.07
                     H                          4
                     0                         10
                     S                          0.93*
                     N                          1
                    Ash                        20.
                  Moisture                      5
                  TOTAL                       100
*
 Represents average S content from 5/1 to 5/21/75.  Non combustible
portion unknown, but analyses on other local coals show about 15% of
the sulfur in sulfate form.

                                 A-3

-------
                                                        Stack gases (SO-,  S0_,  S in fly ash)
        Coal fired
Total
  S
Organic S
Pyritic S
Sulfate
                                   Boiler
                                                                       Leakage (S02>
                                         Ash (unburned S, S in ash)
                         Figure 1:  Schematic Flow Chart for Sulfur

-------
      Based upon the weekly fuel analysis reports from the operator
an average of 31.56 metric tons per day of sulfur entered the boiler
through the coal burned.  If we assume that 85% is combustible, the
combustible sulfur entering expressed in terms of equivalent SO. is:
                        64
         31.56 x 0.85 x -^ = 53.65 metric tons SO  per day.
      Based upon stack gas analyses and flow rates determined by PES
the amount of SO  leaving through the stack averaged over runs 19-50
was 45.48 metric tons per day.  (Note: only runs 15&17-50 are considered
valid.  Runs 17&18 were lower than usual.  If runs 15&17-50 are included
the measured average S09 discharge rate would be 44.25 metric tons
per day.)  Using this approach to a sulfur balance, approximately 15%
of the entering sulfur is unaccounted for.
      The second approach used was to calculate the theoretical sulfur
dioxide concentration expected based upon the hypothesized coal com-
position.  Using this approach it was necessary to assume that the
measured oxygen concentration in the stack gases was correct in order
to allow for the proper quantity of excess air.  Both the oxygen con-
tent determined by PES and that given by the continuous analyzer
installed near the test location were in good agreement.  Therefore,
an oxygen content of 6.5% was assumed for the stack gas.
      The following calculations are used to determine the expected
SO  concentration:
Assume 100 Ib. coal as fired containing 75 Ib. dry coal, 5 Ib. water,
and 15 Ib. ash.
         Constituent           Wt. (Ibs.)            Lb. Mols.
              C                   59                   4.9
              H                    4                   2
              0                   10                   0.32
              S                    1                   0.031
              N                    1                   0.036
             H20                   5                   0.28
                              A-5

-------
Combustion Reaction;
    (4.9C + 2H2 +  .3102 +  .0315 +  .036^) +  (8.1102
                4.9CO  + 2.28H 0 +  .031SO  + 2.490  + 30.55N
               Number mols. combustion products  (dry basis) = 37.97
Concentration SO  by mol = C    (vol.)  _  .GJJ.     in- _ - ,
	/—	SO-	  - 	  x 10  - olo ppm
                                          j /. y /

      If S cone, in coal is .actually 0.93% and only 85% is
               available, then

          C'so  = 816 x .93 x .85 = 645 ppm  (dry basis)


The average SO- concentration determined by PES by testing = 708 ppm (dry basis)

Summary and Conclusions
      The calculated SO- concentration was approximately 9% lower than that
actually determined by testing.  The calculated concentration required
assumptions on actual coal composition, on combustible sulfur, on excess air
and on the degree to which the shift analysis for sulfur content of coal
actually represented the content during firing.   Measured SO- concentrations
depend upon an efficient collection of SO-, and accurate analysis and a
correct determination of sample volume.  Considering the assumptions neces-
sary to calculate a theoretical concentration and the inherent sampling and
analysis errors, this comparison seems well within the expected limits.
      On the other hand, the mass emission rate determined by testing was
less by about 15% than sulfur feed rate.  There are several possible
explantions for this:
                              A-6

-------
      1)  Measured SO  concentration lower  than  actually  exists
      2)  Measured stack gas volume lower than actually exists
      3)  Leakage of stack gases occurring  so  that  all volume
          not being measured.
    .  4)  Amount of sulfur contained In  ash.was  greater than
          estimated. •
      From earlier discussions the first reason  does not  seem  too
likely.  Either one or all  of the latter three  explanations are
possible.  Even though velocity traverses were quite consistent  and
were similar to those measured in earlier testing at the  plant,  some
error could be present.  Also only three of the  four sampling  ports
could be used.  However, because of the  distance downstream from any
disturbances (over 8 pipe diameters) an  uneven flow pattern would not
be expected.  With regard to the third point, it  is  known  that  some
leakage into the annular space between the  inner refractory stack and
the outer concrete support stack takes place.  PES  personnel were
required to wear gas masks to enter this space during probe position
changes.  The cross-sectional area is about the  same as the inner
stack.  No way of quantifying these leaks is possible at  this  time.
      Thus it would seem most credible to assume that the measured
concentrations were quite accurate, that the volumetric stack  gas flow
rate might be slightly higher than determined  during the  tests,  and
that some leakage from the inner stack into the  annular space  surrounding
the inner stack took place.  Should this be true, the mass S02 flux
leaving both the inner stack and the and the annular space at  the top
of the stack could be higher by as much  as  15% over that  reported
leaving the inner stack.

Ill ,   PROCESS  DESCRIPTION
      The process consisted of a pulverized coal-fired boiler  used to
generate steam for electrical power.
                             A-7

-------
               TABLE OF RESULTS
RUN NO.
   CONCENTRATION
MT/DAY          PPM
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
39.5
Results
30.7
22.9
47.9
40.8
46.2
40.3
43.6
44.7
41.3
47.2
51.5
48.3
49.8
43.1
45.5
44.3
50.5
47.9
33.3
42.9
43.1
36.2
37.1
41.3
49.4
45.0
50.6
44.3
46.4
50.9
56.9
54.6
48.9
41.7
568
Discarded
396
408
792
822
689
652
729
785
690
661
691
774
790
699
806
841
812
849
746
775
736
748
762
704
792
771
715
666
739
699
791
774
786
761
                    A-8

-------
 IV,   LOCATION  OF SAMPLING PORTS
      Sampling ports were located approximately  270  feet above ground
level and 170-200 feet above-the breeching.   The stack had an inner
stack of refractory material (approximately  8" thick) and an outer
concrete stack (approximately  one foot thick).   The  spacing between
stacks was 5 feet.  The outer  diameter of  the inside stack at the
sampling level was 23' 4" (inside diameter 22').
      Two velocity traverses were performed  with six points on a
diameter.  These traverses could only be performed utilizing three
ports.  Six points were used on the N-S axis and three points on the
E-W axis.  The east port was inaccessible  to sampling and traversing
because of the angle necessary for probe entry into  the port.
      A third velocity traverse was performed on the three available
ports.  The probe was inserted into the stack to its full effective
length (7 feet).  Velocities were recorded at one-foot intervals.
      The final traverse consisted on using  two  ports 90  apart-.  A
specially made 12-foot pitot tube was used.   The probe was inserted
into, the center of the stack and velocity  measurements were taken at
one-foot intervals.

V,   SAMPLING  AND ANALYTICAL  PROCEDURES

A.  SUMMARY
      Method No. 6, FR36, Dec. 23, 1971 was  used to  determine the concen-
tration of sulfur dioxide.  Method 4 was used to determine moisture
content.  Method 1 was used to determine the number  of sampling points
for the velocity traverses. Modified Method 2 was used to determine
stack gas velocity.  (One modification was to use the points determined
by Method 2 but to only three  ports, the fourth  being inaccessible.  The
other modification was to use  two ports at 90 and to take velocity
measurements from the center at one-foot intervals.   This method was
used to determine the extent of stratification across the stack.)
                            A-9

-------
 B.  EQUIPMENT
      The sampling probe for this test consisted of an eight-foot pitot
 tube  and heated glass-lined sampling probe combination manufactured by
 Michrochemical Specialties Company  (Misco).  The pitot tube used to
 reach the center of the stack was a twelve-foot homemade model.  Both
 pitot tubes were calbirated on return.
      The probe was connected through glass ball joint connections to
 the impingers.  The impingers used for "the sampling train were as
 follows:
      Impinger #1)  Midget impinger modified with a fritted
                    glass bubbler containing 15 ml of 80%
                    iso-propyl alcohol.
      Impinger #2)  Midget impinger containing 15 ml of
                    3% hydrogen peroxide.
      Impinger #3)  Midget impinger containing 15 ml of
                    3% hydrogen peroxide.
      Impinger #4)  Midget impinger empty (used as splash trap).
                    A silica gel drying tube, for pump protection,
                    followed the last impinger.
      The first pump to be used was a Cast Model 0531 rotary vane pump.
This was later found to have a severe leak on the positive side.  This
resulted in the discarding of the first 14 test runs as invalid.  For
this reason only the data for runs 15-50 were reported.  The rotary
vane pump was replaced with a Cast Diaphragm Pump.  This pump was
used for runs 15-50.
      The pump flow was regulated by use of a rotameter and needle valve.
The sample volume was measured with Sprague Model 1755 Dry Gas Meter.
Meter temperature was measured with a Weston model dial thermometer.
                               A-10

-------
      The stack temperature was monitored using an iron-constantan
thermocouple and portable potentiometer (Thermo Electric "Minimite"
Model 31101).   The thermocouple parted on the second to the ,last day
and an average stack temperature was generated from the previous day's
runs.
      Grab samples for CO- and 0  analysis using Fyrite detectors were
collected.  0. concentration data were also available from an in-stack
monitor.

C.  FIELD PROCEDURES
      On the first day of sampling, the test period was set at one hour.
After analysis this was determined to be much more than necessary.
Sampling time was then shortened to twenty minutes.  This was sufficient
for analytical purposes.
      On Tuesday, May 13, 1975, five 20-minute runs were completed.
Sampling had to be stopped early because the boiler was in an upset
condition.  To ease the sample handling, runs were performed in groups of
two with the purge time of 20 minutes separating the runs.  Each group
of two runs were performed at a different port.
      On Wednesday, May 14, 1975, six twenty-minute runs were performed.
A problem with carry-over from one impinger to another occurred.  For
these runs the entire impinger contents were saved.  A special analytical
run was performed on the isopropyl alcohol to determine the extent of the
interference.   The interference was less than 1% so these samples were
deemed valid.
      On Thursday, May 15, 1975, eight twenty-minute runs were performed.
On Friday, May 16, 1975, seven twenty-minute runs were completed and a
pitot traverse was also completed.
                                A-ll

-------
      On Monday, May 19, 1975, the boiler was in an upset condition in
the morning and testing could not be started until 11 o'clock.   Six
twenty-minute runs were then performed.
      On Tuesday, May 20, 1975 four twenty-minute samples were  received,
A pitot traverse was also performed.

D.  LABORATORY ANALYSES
      All laboratory titrations were performed on site by PES personnel.
Impinger contents were titrated with barium perchlorate using Thorin as
an indicator as specified by EPA Method  6.
E.  SAMPLE CALCULATIONS
      Shown on the following pages are the equations used and sample
calculations for determining these stack gas parameters:
      I.  Moisture content of stack gas
      2.  Dry molecular weight of stack  gas
      3.  Molecular weight of stack gas  on a wet basis
      4.  Stack gas velocity
      5.  Stack gas flow rate
      6.  SC>2 concentration
      7.  SO  metric tons per day
                                A-12

-------
DETERMINATION OF MOISTURE CONTENT OF STACK GAS




     V     = 0.0474  ft3  Vin
      w _,           —r—   10
       std           ml
where:
     V     = Volume of water vapor in the gas sample (standard
      w   ,                                       •
       std                   ~

             conditions) , ft



     V     .= Total volume of liquid collected in impingers, ml
     V     = (0.0474) (1)

      Wstd            -

           = 0.0474 ft
     V_    = 17.71 °R      /V    P
        _j         3 — 77
       std         in.Hg
m	m

  T
   m
where:


                                                                   3
     V     = Volume of gas through meter at standard conditions, ft

       std


                                                3
     V     = Volume of gas measured at meter, ft



     P     = Barometric pressure at meter, in.Hg



     T     = Temperature at meter,  F
      m



     V     = (17.71)  /(0.99) (23.98)\     7Q   3

      mstd            (	(529)	1=0. 79 ft



     B     = V
      wo      w  ,
               std
             V     + V

              mstd    wstd
where:
     B   = Moisture content of stack gas, dimensionless



     B   = 0.0474
      wo
           0.79 + 0.0474



         = 0.0566



         = 5.66%
                              A-13

-------
DETERMINATION OF DRY MOLECULAR WEIGHT OF STACK GAS
M, = 0.44 (% CO.) + 0.32 (% 00) + 0.28 (% N. + %CO)
 a             L             /             i


Where:



     M, = Stack gas dry molecular weight



  % CO  = Percent CO  in stack gas from daily Fyrite test
      £.             Z                       /


  % 0-  = Percent oxygen in stack gas from daily Fyrite test



  % N_  = Percent nitrogen in stack gas



  % CO  = Percent carbon monoxide in stack gas
      TUESDAY MAY 13 RUNS 15-19
     M, = 0.44 (12)  + 0.32 (6.5)  + 0.28 (81.5)
      d


        = 5.28 + 2.08 + 22.82 = 30.18 Ib/lb-mole
                             A-14

-------
MOLECULAR WEIGHT OF STACK GAS ON A WET BASIS
     M  =  M. (1-B  ) + 18 (B  )
      s     d     wo         wo
where:
     M  = Molecular weight of stack gas on a wet basis,  Ib/lb-mole
      S


     M, = Molecular weight of stack gas on a dry basis,  Ib/lb-mole



    B   = Moisture content of stack gas, dimensionless
     wo                             6  '
      TUESDAY MAY 13 RUNS 15-19
     M  = 30.18 (1-0.057)   f 18 (0.057)
      S


        = 28.46 + 1.03 = 29.49 Ib/lb-mole
                              A-15

-------
DETERMINATION OF STACK GAS VELOCITY
     V  = K  C       p  avg   T
      s    p  p   \v  • /     [  s
                              P  M
                               s  s,
where:
     V  = Stack gas velocity, ft/sec.
      S


     K  = 85.48 ft/sec  -  - —     **
            .
      p                \ Ib-mole °

     C  = Pitot tube coefficient
      P


    N/5p~ = Average velocity head of stack gas, in. H^O



     T  = Stack gas temperature,  R
      S


     P  = Absolute stack gas pressure, in.Hg
      S


     M  = Molecular weight of stack gas (wet basis), Ib/lb-mole
      S
RUN //15
     V  = 85.48 (0.77) (0.59)1* / 737.5

      s                       I 24.48(29.49)



        = 51.10 ft/sec
                              A-16

-------
DETERMINATION OF STACK GAS FLOW RATE
     Q  = 3600 (1-B  ) V  A/T   ,
     xs            wo   s   /  std

                             T
                              s avg
where:
                                                                  3
     Q  = Volumetric flow rate, dry basis, standard conditions, ft /hr
      S

                                        2.
     A  = Stack cross-sectional area, ft



    T   = Temperature at standard conditions, 530  F



   T    = Average stack temperature,  R



     P  = Stack pressure, in.Hg
      S


    P   = Pressure at standard conditions, 29.92 in.Hg
RUN //15
      s                               [ „„, , I    (29.92/
Q  = 3600 (1-0. 057) (51.10.) (380)  /530

                                 l?37.5
                                 »      i
        = 38.800.000 ft3/hr
                              A-17

-------
CALCULATION OF S00 CONG.  IN LBS/CF
or,
     C    =  7.05 x 10~5 lb-

      S°2
^(Vt-Vtb)  (W  Vsoln\

-V          '  \     IS-/
                                        (^Yfin.

                                             T
                                              m
where:
     C    - Concentration of sulfur dioxide at standard conditions, Ib/cf.

      S°2

     V    = Volume of barium pechlorate titrant used for sample, ml



     V ,   = Volume of barium perchlorate titrant used for blank, ml



     N    = Normality of barium titrant, g-eq/1



     V  , = Total solution volume of sulfur dioxide, 50 ml
      soln                                         '


     V    = Volume of sample aliquot titrated, ml
      3.


     V    = Volume of gas sampled at standard conditions, ft

      mstd
                                               3
     V    = Volume of gas measured at meter, ft
      m


     T    = Meter temperature,  R
      m


     P    = Barometric pressure at meter,  in.Hg
      m
RUN #15
            1771  R
      m ..     :U     x 2.27CF x 2A.52 in.Hg x      ,Op = 1.909 CF
       std    in.Hg                              ilo..) K





     CSO   =(7'05 x 10~5 lb-l\  x  52.05 ml x 0.00973 N x 5 x_J^__

        2   \            g-ml/                                1.909 CF
             9.35 x 10~5 Ib/CF
                               A-18

-------
 DETERMINATION-OF METRIC TONS/DAY
     Sat - (CS02)  (V
                2204.62



 where:



     C   = Concentration of S0n, metric tons/day
      mt                      L


     24  = Hours/day



2204.62  = Lb/metric ton





 RUN #15



     C   = (9.35 x 10~5)(388  x 1Q5) (24)

      mt            2204762	



         - 39.5  mt/day
                                 A-19

-------
0>
                           APPENDIX B

                       METEOROLOGICAL DATA
                         8-19 MAY  1975
                       • SYNOPTIC SUMMARY
                       • WINDS
                       • PIBAL RESULTS
                       • ANEMOMETER  RECORD
                           B-l

-------
                                        4PM
                      METEROLOGICAL DATA

                          8 MAY 1975

SYNOPTIC SUMMARY  The influence of the front that had passed
through the area on the 5th of May had given way to a high pres-
sure system with fair  and warm weather.  The temperatures
during the testing period were from 45° to 64° F.

WINDS  In the morning the winds were
0 to 3 mps below 700 meters with in-
creasing speed above (to 5 m/s). Winds
were from 80  to 115  in the lower
layers and 290° to 310° in the upper
layers.  Between 1100 and 1300 the
winds below 700 meters were light and
variable.  During the 1324 pibal the
lower winds had picked up to 5-10 mps,
from 270° to 310°.   This layer gradu-
ally deepened thru 700 meters by 1420.
           PIBAL RESULTS
          (212-316 meters)

Time        Direction          Speed
(MDT)       (degrees)          (mps)

0915            83              1.9
0933            83              2.5
1100           120              0.4
1109            73              0.8
1120           197              0.8
1315           306              1.8
1324           271              8.9
1334           269              7.0
1420           310.              7.7
1430           318              8.2
1445           290              2.8
                          B-2
                                       Wind Speed  Wind Direction
                                       (0-50 mph)     (0-540U)

                                         ANEMOMETER RECORD

-------
(D
METEROLOGICAL DATA

    9 MAY 1975
   SYNOPTIC  SUMMARY  Early in the day the area was still under the
   influence of  the high pressure zone but the advance of a weak
   upper  level trough  in the afternoon caused cloudiness of 90%
   cirrus  and cirocumulus by 1700.  The maximum temperature during
   the  day was 70°F while the minimum during the test period was
   49°F early in the morning.

   WINDS   The winds were from 90° to 135°
   through 800 m at 4-8 mps until 1200.
   After  1200 the winds were 1 to 4 mps
   and  the direction shifted to 135° to
   180°.   Later  in the afternoon the winds
   were less than 2 mps and variable in
   direction.

             PIBAL RESULTS
          (212-316meters)

   Time     . Direction           Speed
   (MDT)      (degrees)           (mps)

   0852           97              7.4
   0916      .    105              8.2
   0937          112              4.4
   1006          104              6.1
   1015          100              3.7
   1025          108              4.2
   1200          112              6.7
   1209          102              2.3
   1215          183              1.7
   1236          159              3.4
   1245          142              3.7
   1255          143              4.5
   1408          118              1.2
   1416          131              1.2
   1424          031              0.1
   1522          152              1.2
   1559          280              1.9
   1612          314              1.8
IS 4P*
{
|
2PV^
1
NOO/
IOAK
Wind
v
f
\~
>
-.>
v-
d
j
\
-u
?
f
L_
*5
AT
1
,\
,)
1
!
f
>
jl
-

'
>





-








-
-

-

-



-
-



-

-

-
-



-



-
-



-

-

-
-


H<

-
-

— -
£


9
\u.
Speed Win<
1
.
a
..
-
-
x
	
-,
~
-

,
5


•
-

'

<

i
*
«
•
'\
•
«
i
^
,


|
i
Dirt
t
^
Jr"
t-
>
>j
»
>


1
^
S
"!w^— l-
J*^
~\
1 r
I
\\
\
^
4
~4 -.-
V
(
1

—


'
1
— t-
i
, — "*-


rf
5Cti
i
i
•
—]
- —
i
i
1 i
; 1
1
i
I
-+--
T
or
                                        (0-50 mph)
                             B-3
                              (0-540U)

                   ANEMOMETER RECORD

-------
-•
s|
s
i
<
^
>
5


v
4
C
<
1
J
<
t
>
>
^
)
).
s
^
>

}
3


»
»•
r
•
»
'


,
>





-


'-

[










Speed




D/





	 ,
l¥
W

1^
in
mps.

— -_
i
— —
l$7
i






•i





!










'

—

•!



1

-

'
f\
r~
\
i
).
\
> .
>


In
> ,
L
<
\
1
-1
<
i
}
S
f

1




.
d Directi
-
'
_
-
c
.
>n
                                       (0-50 mph)
                          B-4
                                          ANEMOMETER RECORD

-------
                      METEROLOGICAL DATA

                         13 MAY 1975

SYNOPTIC SUMMARY  The upper level trough and associated un-
stable air behind the front that had passed through 24 hours
previously caused gusty westerly winds and scattered cumulus
clouds throughout the day.   During the test period the ground
level temperatures were from 52° to 70° F.
WINDS The winds were from 300° to
"340° at 2-7 mps up to 700 meters until
0940. The speed picked up to 8-12 mps
(with gusts to 18 mps) and shifted to
270°-315° by 11:00. Winds were under 4PW
10 mps after 1300.

PIBAL RESULTS
(21Z-316 meters)
Time Direction Speed
(MDT) (degrees) (mps)

0904 302 3.3 2Pv
0913 324 3.4
0928 323 4.6
0937 . 314 7.1
1100 312 10.2
1109 285 5.3
1119 297 8.2
1134 293 14.1
1155 300 8.5
1203 304 8.6 Noo
1210 270 2.3
1316 312 6.5
1329 288 4.9
1346 313 6.7
1352 325 6.9
1400 303 6.2
1435 321 8.8
\OM


•

8AA


-





















— «
>


~>

_<
s
f

1
\

-








'

^




C
<
<
_
••>
f
\


j
.5
.^
*

\
.\
I













<
x
<
4
^
S
<
i
•*••
k
•X
_f
C.
«
1
rf*
5-
^^
s
1

*

r!
_v



1
i
k
'.
k





\.


}






















_
i

i
r •


»






















-






-





-






















-






...
~'
„
-























-








-





















-





-
-






-


-

._













i»H
in






-
-
-






-



-


-









T







-
__j






















13














-















,
























i

i|_
1. 1
1 1

.


^













L



u
17















-_L



c
J


























































..j
_ul
















i






























i






»•
•H




^*

^





«•
p
>_.
«



>


•

^c
•=1


V «
]i



1
i
4
i








__.



i



lil


f. |
L-J
t~\ i

_
i i
fl.
Jl





"""1

r


T
M-
(2
B:

.•
• •
«:




/
•!
j,
e
-------
                      METEROLOGICAL DATA

                         14 MAY 1975

SYNOPTIC SUMMARY  The test area was under the influence of an
upper level high pressure system on the 14th of May.  There
were no clouds and the temperature  was as high as 77°.  In
the early morning hours the power plant plume was trapped
beneath a  low level inversion.
WINDS  In the morning winds were from
90° to 135° and of moderate strength
(as high as 10 mps).  This condition
continued throughout the day in the
upper levels with the speed gradually
diminishing to 2-4 mps.  At the lower
levels the winds shifted to 180° and
then 215° after 1430, with speeds of
1 to 4 mps.
          PIBAL RESULTS
        (212-316meters)

Time       Direction
(MDT)      (degrees)

0908           104
0924           100
0945           100
1003           108
1129           123
1215           207
1241           141
1358           131
1417           181
1440           236
Speed
(mps)

 6.4
 5.5
 5.5
10.1
 2.0
 1.6
 1.5
 1.2
 2.1
 1.7
                                       8AH
                                      Wind Speed  Wind Direction
                                      (0-50 mph)     (0-5400)
                        B-6
                                         ANEMOMETER RECORD

-------
0>
METEROLOGICAL DATA

   15 MAY 1975
   SYNOPTIC SUMMARY   The  advance of a trough in the upper levels
   caused a few  cirrus  and  cumulus clouds in the afternoon on
   this  generally  fair  day.  Temperatures ranged between 63 and
   80°  F during  the  test  period.

   WINDS  The  winds  were  variable, 3 to
   10 mps,  from  80°  to  110°  in th~ 1~"~
   below 700 meters  until 1300.
   and  variable  conditions  existed until
   the  1345 and  1421 pibals  which showed
   winds from  290° at speeds of 2 to 4
   mps.   After 1430  the winds were
   between 180°  and  270°  at  speeds in the
   range of 2-6  mps.

              PIBAL RESULTS
             (212-316meters)

   Time          Direction
   (MDT)         (degrees)

   0842            97
   0856            99
   0914            98
   0927            92
   0938            95
   1022            102
   1045            112
   1054            117
   1120            99
   1130            110
   1155            105
   1248            94
   1312            218
   1321            156
   1339            354
   1345            289
   1421            288
   1446            239
   1505            200
   1527            265
   1542            247
levels
ght
until
howed 4PM
o 4
in the
Speed ZPV
(mps)
7.0
8.9
6. 2
7.3
8.1
4.9
8.0 N0°
W . \J
8.4
7.5
6.1
4.5
4.7
.8
1.3
1.0
3 4 IOAH
5.1
3.4
3.7
6.0
6.2
8A^

y
P

«
C
-J
<•
1
iL.






V
f
?
S>
— >
1
9
s
1
f'
f
1
I
— V
(
,->



t





i
^'
: '
1

-









-



















—
-








-



M;


-


-
•-


i









15





-

-,

J



-
n
-



17



-
r
-



5





-






.


-


<
a
i

i




s:
<







-
•
>
)»
^






"i
V-
1
'!


} '
	


•
'
4-
T
                                        Wind Speed  Wind Direction
                                        (0-50 mph)      (0-540U)
                          B-7
                                            ANEMOMETER RECORD

-------
(D
METEROLOGICAL DATA

   16 MAY 1975
  SYNOPTIC SUMMARY  The advance  of a  frontal system from the west
  caused increasing clouds throughout  the day.  Early morning
  cirrus thickened to cirrus  stratus  and altostratus in the after-
  noon.
WINDS During the first pibals the ' u
wind below 6UU meters was from 8U° to ^,»
100° at speeds of 4-8 mps. By 1030
the winds had shifted to 160° and by ~^
1347 to 200°. The speeds during this
L-LJHc Vd.iJ.cU CUIlb J.U.C i dU J. y WJ.LJ1 a I dllgc *
of 4 to 16 mps. During the rest of
the afternoon the winds were between
170° and 200° with speeds still fluc-
tuating between 3 and 16 mps. _J
PIBAL RESULTS !
(212-316 meters) ^~
Time Direction Speed
(MDT) [degrees) (mps)
0842 91 5.0
0913 90 , 7.3
1 n 1 1 1 A n C. 7 N°°N-4-4
1032 140 5.7
I <
1110 173 4.4
*
1236 176 14.2 ^J
1301 181 6.7
|
1320 193 13.7
10AN -— _,
1347 195 15.8 1
<
1415 168 10.2 ;
1433 202 4.5
8Mv — _

V.
>
~——=
L::
!>
j >
s^
'$'
J>
j
^J
1 — r^ — ! —
,-=»
^
c
> . .
>
r
r
s
•^ *
~
I
^ T
r
.
>
r
.
k
i':-:
I.


..._-
. - -


\\t







-
• .
i
i
i
i
- 4^
i
111
¥-16(^7 i^
- 4+j-
r
i
- -W.
: f.


i \ L
. . . ^ . . •
'iff:-
V_, 	 rJ
•it- -rrT
\\
P|
1
! ! '
i)
1 _*_J— J— rl
j
1
~fe
' : !
l ; i
(\ • ' i
> L.-L-
tt
^
H _J-
" 'ity ' ' '
SI i
/ '
/ _j—*-
"It^
. ; ' i'
.
i
"TnTTTT
                                       Wind Speed  Wind Direction
                                       (0-50 mph)     (0-540U)
                            B-8
                                          ANEMOMETER RECORD

-------

-------

-------

-------
0)
  NOISE -
Spectrometer response due to spurious,  unwanted
electronic signals;  usually a few ppmM  depend-
ing on gas measured  and available light
  OVERHEAD
  BURDEN
See Burden
  PASSIVE MODE -•
Remote sensing measurements using scattered
natural light, usually with instrument viewing
vertically upwards;  contrasts with active mode
  PIBAL -
Pilot balloon, used to measure wind speed and
wind direction at the elevation of the plume
  PLUME -


  POINT MONITOR
Dispersing stack emissions


Instrument capable of monitoring ambient air
sample drawn into it through a sampling train
from a fixed nearby point
  PPB -


  PPM-M  -
Parts per billion, concentration measurement
Parts per million-meters,  concentration-path
length measurement
  RAYLEIGH
  SCATTERED -
Light scattering process produced by spherical
particles whose radii are smaller than one-tenth
the wavelength of the scattered radiation
  REMOTE SENSOR -
  SEGMENT -
  STANDARD
  DEVIATION
Instrument capable of monitoring a phenomenon
by electro-optical means across an intervening
distance; as COSPEC measures S0_/N0  in ambient air
                               2   2

Portion of traverse route for which incremental
mass flux calcuations are made



Positive square root of mean of squares of devia-
tion from mean of population; measure of disper-
sion of data points
                    L< X Ls il -*-.*. V Jll Jll^r V* « A vy -i-  ^J
                    sion of  data points

                                G-3

-------

-------
                                   TECHNICAL REPORT DATA
                            (Please read Instructions on the reverse before completing)
 1. REPORT NO.
   EPA-600/2-75-077
                                                           3. RECIPIENT'S ACCESSION NO.
 4. TITLE AND SUBTITLE
  EVALUATION OF  THE CORRELATION SPECTROMETER AS AN
  AREA S02 MONITOR
             5. REPORT DATE
              October  1975
             6. PERFORMING ORGANIZATION CODE
 7. AUTHOR(S)
  R. B. Sperling
                                                            8. PERFORMING ORGANIZATION REPORT NO.
 9. PERFORMING ORGANIZATION NAME AND ADDRESS
  Environmental  Measurements, Inc.
  215 Leidesdorff  Street
  San Francisco, California  94111
                                                            10. PROGRAM ELEMENT NO.
                1AA010
              11. CONTRACT/GRANT NO.
                68-02-1773
 12. SPONSORING AGENCY NAME ANP ADDRESS
  Environmental  Sciences Research Laboratory
  Office of Research  and Development
  U.  S. Environmental  Protection Agency
  Research Triangle Park, North Carolina   27711
                                                            13. TYPE OF REPORT AND PERIOD COVERED
                Final
              14. SPONSORING AGENCY CODE

                EPA-ORD
 15. SUPPLEMENTARY NOTES
 16. ABSTRACT
       The Barringer COSPEC II instrument,  an  S02 remote sensor,  was compared to
  the manual in-stack S02 and velocity compliance tests for emission measurements.
  The correlation  for short term (one hour  or  less) comparison was  poor.  Higher
  correlations for S02 emission rates on  a  daily basis were found.   In addition
  to the COSPEC  II,  a COSPEC III and COSPEC IV were used in the  study.  Correlations
  among the three  instruments were good  (90-95%).  Main source of error in the
  remote measurements was the wind velocity determinations.  For  a  short time span
  of less than an  hour,  wind velocity may vary 100% and only averages can be obtained
  for the measurements.
 7.
                                KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                              b.IDENTIFIERS/OPEN ENDED TERMS
                           c.  COSATI l-ield/Group
  *Evaluation
   Comparison
  *Remote sensing
  *Ultraviolet spectrometers
   Air pollution
  *Sulfur dioxide
   Mobile equipment
Barringer COSPEC
instruments
14G
14B
13B
07B
15B
 8. DISTRIBUTION STATEMENT .
 RELEASE  TO PUBLIC
                                              19. SECURITY CLASS (This Report)

                                              UNCLASSIFIED
                           21. NO. Oc PAGES

                               115
20. SECURITY CLASS (Tills page)
UNCLASSIFIED
                                                                         22. PRICE
EPA Form 2220-1 (9-73)
                                          G-5

-------