EPA-600/2-75-012
July 1975
Environmental Protection Technology Series
CONTINUOUS MEASUREMENT
OF GAS COMPOSITION
FROM STATIONARY SOURCES
U.S. Envii i Protection Ayency
Office of Research and Development
Washinqtim. 0. C. 20460
-------
EPA-600/2-75-012
CONTINUOUS MEASUREMENT
OF GAS COMPOSITION
FROM STATIONARY SOURCES
by
E. F. Brooks, C. A. Flegal, L. N. Harriett,
M. A. Kolpin, D. J. Luciani, and R. L. Williams
TRW, Systems Group
One Space Park
Redondo Beach, California 90278
Contract No. 68-02-0636
ROAP No. 21ACX-132
Program Element No. LAB013
EPA Project Officer: William B . Kuykendal
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, North Carolina 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
WASHINGTON, D.C. 20460
July 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These series are:
1. ENVIRONMENTAL HEALTH EFFECTS RESEARCH
2. ENVIRONMENTAL PROTECTION TECHNOLOGY
3. ECOLOGICAL RESEARCH
4. ENVIRONMENTAL MONITORING
5. SOCIOECONOMIC ENVIRONMENTAL STUDIES
6. SCIENTIFIC AND TECHNICAL ASSESSMENT REPORTS
9. MISCELLANEOUS
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 for sale through the National
Technical Information Service, Springfield, Virginia 22161.
Publication No. EPA-600/2-75-012
11
-------
TRW Report 23060-6021-RU-00
ABSTRACT
The program objective was to develop and evaluate methods for the
continuous measurement of gaseous emissions from stationary sources,
specifically in large or complex ducts where total flow processing
techniques are not practical. This report is concerned with the measure-
ment of mean gas concentrations in rectangular ducts. Work consisted
of a review of related programs, development of a computer program to
assess stratification levels and evaluate sampling techniques, formulation
and evaluation of point sampling methods for continuous monitoring,
development of a multi-port continuous gas sampling probe, and field
demonstration of hardware and techniques. Results showed that emissions
can be accurately monitored using as few as one flow sensor and one
sampling probe, even in the presence of significant velocity and compo-
sitional stratification, although stratification levels were too high
for single point samples to be acceptable. It was shown for all data
examined that good accuracy can be attained by taking a spatial con-
centration average -- flow proportional sampling is not required. The
field demonstration verified the acceptability of the proposed methodology.
This report was submitted in fulfillment of Contract No. 68-02-0636,
by TRW Systems Group under the sponsorship of the Industrial and Environmental
Research Laboratory-RTP, of the Environmental Protection Agency. Work was
completed as of March 1975.
111
-------
CONTENTS
Page
Abstract iii
List of Figures vi
List of Tables ix
Acknowledgements x
Sections
I Conclusions 1
II Recommendations 2
III Introduction 3
IV Task I - Data Review 5
4.1 Review of Stratified Gas Program Results 5
4.2 Analytical Considerations 7
4.3 Desired System Accuracy 17
V Task II - Flow Analysis Program 20
5.1 Program Development 20
5.2 Program Description 21
5.3 Typical Program Results 23
VI Task III - Mapping Technique Evaluation 27
6.1 Exxon Data 27
6.2 TVA Data 43
6.3 Preliminary Sampling Technique Conclusions 50
VII Task IV - Sample Probe Development 51
7.1 Prototype Probe Design 51
7.2 Suggestions for Improved Probe Design 51
iv
-------
CONTENTS (Cont'd)
Page
VIII Task V - Field Demonstrations 57
8.1 Facility Description 57
8.2 Test Conduct 60
8.3 Flow Data Correlation 66
8.4 Test Results 66
8.5 Summary of Results 92
IX Discussion of Results 95
X References gg
XI Glossary 100
XII Appendix 101
-------
FIGURES
No. Page
1. Flow Measurement Control Volume 8
2. Duct Analogy for Transformation of Test Velocity to
Standard Velocity 11
3. Histogram of Deviations from Mean Value for NOV Traverse 16
A
4. Row Average Analysis for C02 Concentration Map —
Exxon Run 1 25
5. Normalized C02 Concentration Distribution —
Exxon Run 1 26
6. Location of Sampling Plane for Exxon data -- TVA Widows
Creek Unit 7 28
7. Duct Shapes for Exxon and TVA Tests 29
8. Normalized C0£ Concentration Distribution --
Exxon Run 1 30
9. Normalized C02 Concentration Distribution --
Exxon Run 2 31
10. Normalized SO? Concentration Distribution —
Exxon Run 1 32
11. Normalized SOo Concentration Distribution --
Exxon Run 2 33
12. Normalized NO Concentration Distribution --
Exxon Run 1 34
13. Normalized NO Concentration Distribution --
Exxon Run 2 35
14. Normalized 02 Concentration Distribution —
Exxon Run 1 36
15. Normalized 0? Concentration Distribution --
Exxon Run 2 37
16. Row Average Analysis for Gas Concentration Maps —
Exxon Runs 1 and 2 42
vi
-------
FIGURES (Cont'd.)
No. Page
17. Normalized S02 Concentration Distribution -- TVA Run 1 44
18. Normalized S02 Concentration Distribution -- TVA Run 2 45
19. Normalized S02 Concentration Distribution -- TVA Run 3 46
20. Normalized S02 Concentration Distribution -- TVA Run 4 47
21. Row Average Analysis for S0? Concentration Maps —
TVA Runs 1-4 49
22. Multi-Port Sampling Probe Basic Design 52
23. Suggestions for Continuous Gas Sample Probe Modifications
to Prevent Particulate Entry Into the Probe 55
24. Schematic of Moapa Power Plant 58
25. Duct 1 Geometry 59
26. Point Sampling Probe 61
27. Multi-Hole Sampling Probe for Field Demonstration 63
28. Schematic of Gas Sampling Train for Field Demonstration 64
29. Normalized C02 Concentration Distribution - Duct 1, 1974 73
30. Normalized 02 Concentration Distribution - Duct 1, 1974 74
31. Normalized NO Concentration Distribution - Duct 1, 1974,
Run 1 x 75
32. Normalized NO Concentration Distribution - Duct 1, 1974,
Run 2 . * 76
33. Normalized C02 Concentration Distribution - Duct 1, 1975 77
34. Normalized 02 Concentration Distribution - Duct 1, 1975 78
35. Normalized S02 Concentration Distribution - Duct 1, 1975 79
36. Normalized NO Concentration Distribution - Duct 1, 1975,
Run 1 80
37. Normalized NO Concentration Distribution - Duct 1, 1975,
Run 2 x 81
38. Normalized C02 Concentration Distribution - Duct 2, 1975 82
vii
-------
FIGURES (Cont'd.)
No. Page
39. Normalized 02 Concentration Distribution - Duct 2, 1975 83
40. Normalized S02 Concentration Distribution - Duct 2, 1975 84
41. Normalized NO Concentration Distribution - Duct 2, 1975,
Run 1 x 85
42. Normalized NO Concentration Distribution - Duct 2, 1975,
Run 2 86
43. Row Average Analysis for C02, 0£, and S02 Concentration
Maps -- Field Demonstration, Duct 1 87
44. Row Average Analysis for NO^ Concentration Maps --
Field Demonstrations, Duct 1 88
45. Row Average Analysis for C02> 02, S02, and NOX
Concentration Maps -- Field Demonstrations, Duct 2 89
viii
-------
TABLES
Np_. Page
1. Comparison of Deviations for Actual and Normal Distributions 15
2. Typical Analysis Program Results 24
3. Mean Values and Stratification Levels for Exxon Data 39
4. C02 Tracer Method Analysis of Exxon Data 41
5. Mean Values and Stratification Levels for TVA S02
Concentration Data 48
6. Mean Values and Stratification Levels for 1974 Field
Demonstration 67
7. Mean Values and Stratification Levels for 1975 Field
. Demonstration 69
8. Computation of Total Flow from Coal Analysis and Measured
C02 Concentration, 1974 70
9. Average Daily Flow Through Ducts, 1975 71
10. Multi-Hole Sampling Probe Overnight Sampling Results, 1975 90
11. C02 Tracer Method Analysis of Field Demonstration Data 91
12. Short Term Stationary Point Gas Sample Data, 1974, Duct 1 93
13. Overnight NOX Gas Sample Data, 1974 94
IX
-------
ACKNOWLEDGEMENTS
Program personnel are grateful to the Nevada Power Company for their
assistance and use of facilities for the field demonstrations. Ue also
wish to thank Exxon Research and Engineering Company for use of their
data obtained during EPA contract 68-02-1722, "Determination of the
Magnitude of S02» NO, C02> and Og Stratification in the Ducting of
Fossil Fuel Fired Power Boilers," and the Thermo Electron Corporation for
the loan of an SOp analyzer during the second field demonstration.
-------
SECTION I
CONCLUSIONS
• Single point gas sampling is considered unacceptable from an accuracy
standpoint due to the high stratification levels and poor temporal
repeatability of the cases examined.
• For gas sample traverses, spatial concentration averages are as
accurate as flow proportional concentration averages in rectangular
ducts for most engineering purposes. This justifies the development
of continuous sampling systems which do not need to take local flow
velocity into account.
• Measurement techniques developed during the program can be used
to reliably measure total mass flow, gas species concentration,
and gas species mass emission to accuracies of 5-10 percent, 2-6
percent, and 11-14 percent, respectively, in rectangular ducts on
a continuous basis. This can be accomplished through the use of
as few as one flow sensor and one sampling probe.
• The row average technique, developed originally for velocity measure-
ment during the first phase of the program, is also acceptable
for continuous gas sampling. Rows should be taken in the direction
of maximum compositional stratification whenever possible.
• The stratification level of gases for the data examined varied
from 8 percent to about 40 percent of the mean concentration value,
and was generally in the 10-20 percent range.
• Poor correlation was observed for the C02 Tracer Method.
• Use of continuous gas analyzers is much more compatible with spatial
gas sampling than with flow proportional sampling.
-------
SECTION II
RECOMMENDATIONS
• The data base for comparison of spatial and flow proportional
averages should be expanded to fully verify the adequacy of
spatial sampling.
• Applicability of developed methods to sampling in circular ducts
needs to be assessed.
• Development of multi-point continuous gas sampling probes should
be pursued.
• Agreement between results of traverses involving as few as 24
sampling points and the expanded computer matrix of 525 points
suggests that work should be done to determine a reasonable
maximum number of sample points for traverses. Reducing the
number of points would also reduce errors due to temporal varia-
tion during the traverse.
• Extensive work should be performed at a representative site,
such as an EPA demonstration plant, to fully assess the extent
of temporal concentration variations in a process stream and
their effect on the accuracy of both manual traverses and con-
tinuous measurements.
• A standard quantitative definition of the term "stratification
level" needs to be adopted. It is recommended that the 2a
designation used in the program be adopted for this purpose,
where 2o is twice the standard deviation of traverse points from
the mean concentration.
-------
SECTION III
INTRODUCTION
The objective of the program is to develop methods to measure total
gaseous mass emissions and individual gas species emissions from stationary
sources on a continuous basis. Continuous monitoring is important for
both process control and compliance with emission standards. The primary
methodological difficulty involved in continuous monitoring is due to
stratification of both velocity and composition in the stream. The
sampling methods developed during the program apply in general to any
stationary source.
The program is concerned with point sampling methods in large ducts,
that is ones where monitoring of the total flow is impractical. The first
phase of the program was concerned with the measurement of total mass
emission (total volumetric flow rate) in rectangular and circular ducts,
and a report (Reference 1) has been issued documenting the results. This
report is concerned with measurement of gas concentrations, which are
coupled with flow data to determine constituent gaseous emissions. Work
during this part of the program was concerned with measurements in rectangu-
lar ducts only due to shortage of data for circular ducts. It is felt
that measurements in rectangular ducts represent the more difficult problem,
and that the methods developed for that case can be easily adapted to
circular ducts. Examination of existing data showed single point monitor-
ing to be unacceptable. It was also recognized that continuous monitoring
systems using a large number of sampling points would be too expensive and
complex. The general approach used was to consider methods which extract
samples from a number of discrete points in the process stream, then mix
and analyze the samples. Emphasis was placed on minimizing the number of
sampling points and specifying optimum locations.
The program began in October 1972. The gas sampling phase started
in July 1974, and the technical effort was concluded in March 1975. The
task breakdown was as follows:
-------
Task I Data Review. The effort was begun with a review of. an EPA
program on stratified gas measurement conducted by Wai den Research.
Baseline work was then performed to determine a viable method for specifi-
cation of stratification levels, and to determine desired system accuracies.
Task II Flow Analysis Program. A computer program was developed to
evaluate stratification data and determine the applicability of proposed
sampling techniques.
Task III Happing Technique Evaluation. The analysis program developed
in Task II was used to analyze sample traverse data obtained from the
Wai den Research report and from a stratification assessment program con-
ducted by Exxon Research and Engineering Company. Results were used to
select methods for field evaluation in Task V.
Task IV Sample Probe Development. A multi-port sampling probe was
designed and prototype units were fabricated for field evaluation. The
probe was designed to obtain samples for the Row Average sampling method,
which was initially developed and successfully demonstrated for velocity
measurement.
Task V Field Demonstrations. Assessment of stratification levels and
demonstration of continuous sampling techniques were performed during
two field demonstrations at the Nevada Power Company station at Moapa,
Nevada.
The most important program results were the discovery that the average
stratification level was on the order of 10-15%, which resulted in the
conclusion that single point sampling is not generally acceptable, and
the discovery that spatial average concentrations agreed with flow
proportional average concentrations to within an average of 2% for all
data considered in spite of the stratification levels. The latter re-
sult means that concentration and flow data can be taken independently
and still be used to accurately determine species mass emissions. This
greatly simplifies hardware for continuous monitoring systems. As a
result of this discovery, emphasis was placed on development of spatial
averaging techniques. A monitoring system employing a single flow probe
and a single sampling probe showed accuracies on the order of 2-3% for
total and constituent mass emissions during the second field demonstration.
-------
SECTION IV
TASK I - DATA REVIEW
The objective of this phase of the program was to determine suitable
methods for continuous gas sampling, and to combine these methods with
those developed under the first phase of the program for measurement of
total volumetric flow. At the time Phase II of this program started, EPA
program 68-02-1306, "Procedure for Measurement in Stratified Gases," per-
formed by Walden Research Division of Abcor, Inc., had recently been
completed. By design, the results of the Walden program served as the
basis for Phase II gas sampling work. The first task was begun with
analysis of the Walden report (Ref. 2, EPA Report EPA-650/2-74-086-a) and
a review in their facility. This was followed by a survey of requirements
and analytical techniques used in gas sampling. Results were used to
determine the type and extent of work required for subsequent Phase II
tasks.
4.1 REVIEW OF STRATIFIED GAS PROGRAM RESULTS
The following excerpts are from pages 1 and 2 of Reference 2:
CONCLUSIONS
Results from the literature and field surveys indicate
that gas stratification exists, but it is likely to be
less general and less severe than particulate stratifi-
cation. For a given gas stream, it is necessary to make
a preliminary gas concentration survey to determine the
existence of spatial stratification.
Where stratification exists, we have concluded as a
result of this program, that there are two methods of
obtaining representative gas samples. Where conditions
permit, we recommend a system of monitoring the ratio
of pollutants such as SOo. NO , etc. to C02 from a
single location. Then from the measured fuel flow and
-------
chemistry of the process, the mass flow of CCL is the
mass flow of the pollutant. Where conditions do not
permit such a system, we recommend a schedule of manual
surveys and installation of a multi-element proportional
sampler and gas velocity array.
RECOMMENDATIONS
In the course of this project, areas requiring further
development were identified. These are as follows:
A. A program is needed to develop an automatic instru-
mentation system for extracting continuous representative
gas samples from stratified gas streams, for example, a
multi-probe automatic proportional gas sampler which
would be practical in terms of cost and adaptability to
various process streams.
B. A program is needed to develop techniques for determ-
ining the total gas flow profile or velocity vector in
process streams. It is likely that in practice a signifi-
cant fraction of errors in emission measurements are
attributable to errors in gas velocity/flow determination.
C. Further documentation of the extent and frequency of
gas stratification in process streams together with
statistical analyses of data would be helpful in refining
sampling methodologies.
The first conclusion and third recommendation were given first con-
sideration. It is clear that the extent of stratification present in a
process stream is a major factor in determining required gas sampling
techniques. In their field work, Walden found no evidence of significant
gas stratification in tests in oil-fired power plants in Weynouth,
Massachusetts and Boston (Boston Edison Mystic Station) or in a coal-fired
power plant in Bow, New Hampshire. Typical maximum deviations from the
mean concentration for C02 and S02 were +15%, with average deviations from
the mean on the order of +4%.
-------
For a continuous monitoring system, it would seem reasonable that the
number of sampling points required would increase with increasing composi-
tional stratification. Hardware cost would in turn be proportional to the
number of sampling points, which is clearly not the case for manual
sampling. Assessment of stratification levels to be expected under normal
circumstances is then very important in terms of eventual hardware costs
and system complexity.
The review of the Wai den report made it clear that the following work
would need to be performed:
t Develop analysis techniques to determine stratification
levels in a simple, uniform manner.
t Expand the data base for compositional stratification to
determine the severity of the problem.
• Develop a methodology for continuous gas sampling and/or
verify the acceptability of the Maiden COp tracer method.
4.2 ANALYTICAL CONSIDERATIONS
4.2.1 Proportional Sampling Requirements
The first step in formulating the required analytical techniques is
to start with the basic flow equations. Consider the duct shown in
Figure 1. The four sides and the entry and exit planes form a control
volume. By definition, the sides are solid, so that all fluid must enter
through the left plane and leave through the right plane. For simplicity
(which does not compromise accuracy), assume that the flow rate into the
control volume is always exactly the same as the flow rate out of the
control volume. This relation becomes true as the size of the control
volume approaches zero. The flow through the control volume may then be
given as the flow through the exit plane of the control volume:
J J
where A
iti = mass flow rate, gm/sec
p = local fluid density, gm/cm (gas phase only)
7
-------
Mean Flow
Direction
Flow enters from the left and exits
to the right.
Velocity at a point in the exit plane is given
by
I = u + v
"j + w~fc
where
T, J, "£ are unit vectors in the directions
shown, forming an orthogonal coordinate system
and
u, v, w are the scalar components of "u in
the ~t, ~j and "£ directions, respectively
The vector ~n is the unit vector normal to the
exit plane, so that
and the net flow component out of the duct at the point
shown is
"tT . ~n = u
Figure 1. Flow measurement control volume
8
-------
u = local velocity (vector), m/sec
n" = unit vector normal to exit plane, dimensionless
2
A = exit plane area, m
u = U* • n, scalar velocity component normal to exit
plane, m/sec
This equation represents the total 'instantaneous gas flow through the
control volume. In gas sampling work, it is desirable to know the instan-
taneous mass flow of a specific constituent, such as 02 or S02, for
control and/or regulatory purposes. For this case, we have
ih1 =//pi ""' n'dA (2)
where
( ).. denotes the property relative to species i, e.g.
m.. = mass flow of species i, gm/sec
pi = mass density of species i, gm/cm
This can be put in terms of normally measured parameters as follows:
P1 = c.M. = cy.M1 (3)
where c = global concentration, moles/cm
3
c^ = concentration of species i, moles of i/cm
M. = molecular weight of species i, grams of i/
mole of i
P.J = mole fraction of species i, moles of i/mole
c - KL (4)
where
2
r. . -i j. j. gm-m
R = universal gas constant, — a - —
mole-sec
T = absolute temperature, °K.
p = absolute pressure, torr
Substituting in equation (2), we get
-------
or
TH,U
m. = R M.A -i- (6)
where the average is defined by correct evaluation of the integral in
equation (5). Equation (6) is now in terms of the basic engineering
parameters commonly measured in a gas flow. The mole fraction y. is com-
monly given in units of parts per million for gases such as SO and NO ,
f\ J\
and in mole percent for N2, 02, H^O and C02<
It is common practice to evaluate equation (5) by means of a gas
sample traverse. The integration is then approximated by the summation
N T
VWlLp^1>n VA,
n=l
for N area segments. Usually the segments are of equal area, so that
n=l
This is the practical form of equation (5). It shows that for accurate
gas sampling measurements, the local velocity component normal to the
sampling plane and the local static pressure and temperature must be
measured as well as the local concentration (Note: from this point on,
the term yi will be referred to as "concentration" rather than "mole
fraction" in accordance with common practice). Temperature and pressure
effects are most easily handled through introduction of the concept of
"velocity at standard conditions." This is illustrated in Figure 2. For
this control volume, the gas enters at arbitrary pressure and temperature.
It then undergoes whatever changes are required in order to emerge uniformly
with a static temperature of 20°C and at an absolute static pressure of
760 torr, which have been defined as standard conditions for this program.
Also by definition, the gas is considered to be chemically frozen and
there are no phase changes. This is important from a practical measurement
standpoint. It means that the mass flow rate being considered is that of
10
-------
X
X
X
TEMPERATURE
PRESSURE
DENSITY
VELOCITY
AREA
X
X
X
STATION
1
(TEST)
Pi
Pi
Hi
A
X
X
X
X
STATION
2
(STANDARD)
FROM CONSERVATION OF MASS,
SO
AND
Pl
A - ps ns A
—
ps
Figure 2. Duct analogy for transformation of test
velocity to standard velocity
11
-------
the gaseous components only, since common velocity sensors for operation
in a gas stream respond to gas flow, and not to liquids or solids which
may be entrained in the flow. Therefore defining a gas flow in terms of
standard rather than actual conditions implies changes in pressure and
temperature only — not composition and phase. Thus for any one area
segment where the velocity and density are considered to be uniform,
- u = — u (9)
p p s ^ '
where
( ) = value at standard atmospheric conditions
so that
AT N
VNp: RMiZ>iWn
n=l
or
AT.
Thus for the term "proportional sampling" to be truly correct, it should
be interpreted as "sampling in proportion to the local velocity at standard
conditions" so that temperature and pressure effects are not ignored.
The proportional gas sampling requirement inherent in equation (11)
is the source of significant practical difficulties in the field, since
the sampling rate must be changed for each point when standard wet chemistry
methods are used. Common use of continuous gas analyzers may be expected
to change this approach. For that technique, it is easier to monitor the
local value of p.. with the analyzer, then multiply by the local velocity
to obtain the local mass flow rate. This was the approach taken for all
field work during the program, and is also the approach one would expect
to take in designing a continuous multi-point sampling system. The
alternative is somewhat less attractice — it would require obtaining a
volume of sample from each point, where the volume is proportional to the
local velocity. These samples would then have to be mixed thoroughly in
some type of mixing chamber and delivered to the analyzer. It is felt
12
-------
that the hardware requirements for this approach would be more expensive
and difficult to maintain than the software requirements for the first
method.
There is a third alternative which became apparent as a result of
the fairly low stratification levels reported by Walden: simple spatial
sampling rather than proportional sampling. Proportional sampling is
represented by the term y7u^~ in equation (11), which denotes that the
true constituent mass flow rate is given by the average of the products
of local concentration and local velocity. This is mathematically different
from the product of the average concentration and the average velocity,
i.e. in general,
yius * ^ • \TS (12)
The only times actual equivalence is guaranteed is if either concentration
or velocity is constant in the sampling plane. If gas stratification
levels are sufficiently low, and there is no direct correlation between
regions of high and low concentration and high and low velocity, then for
the accuracies desired, we may have
^•^"^ 03)
If this approximation were generally valid, large savings on systems
hardware requirements would be realized. It would mean that proportional
sampling would not be required, which would allow for a simple constant
flow rate sampling network. Further it would mean that gas sampling and
volumetric flow, as designated by u$, could be determined independently.
Since Phase I results (Reference 1) showed that volumetric flow can be
determined in complex ducts with as few as one velocity sensor, independent
determination offers even greater advantages. As will be shown in subse-
quent sections of this report, the approximation represented by equation
(13) proved to be very good for the data examined.
4.2.2 Stratification Assessment
In order to determine stratification levels from various sources in
a consistent manner, a uniform method to determine measures of data spread
13
-------
is required. Perhaps the most common and easy to work with measure of
spread is the standard deviation for a normal distribution. The potential
problem with using standard deviation as a measure of spread is that it
is normally applied to a number of measurements of the same parameter, such
as repeated measurements of the length of a table top, or repeated measure-
ments of gas composition in a fixed volume with no chemical reaction. For
the present case, we wish to apply it to a number of individual measurements
of concentration at separate points in a non-uniform gas stream. For this
case, there is no a priori reason to believe that the readings can all be
considered to belong to a normal distribution. Gas traverse data obtained
during the Phase I field test (Section VIII) were examined to see if they
approximated a normal distribution. Data consisted of 49 point traverses
for 02, C02, and NOX in a coal fired power plant. Actual data are pre-
sented in Section VIII. For each traverse, the mean value and standard
deviation of the gas concentration was determined as follows:
n=l
where for the case N - 49.
(15)
where
a = standard deviation of y^
It is common, particularly for small values of N, to use (N-l) instead of
N in the denominator of equation (15). For the purposes of this program,
one or at most two significant figures are desired for a, so that use of
N instead of (N-l) has no significant effect on desired accuracy. Results
of the calculations are shown in Table I. Results for all three cases are
very close to what would be expected for a normal distribution. This is
further illustrated in Figure 3 for the NO traverse. Figure 3 shows a
A
histogram of deviations from the mean NO value. Superimposed on this is
14
-------
Table 1. COMPARISON OF DEVIATIONS FOR ACTUAL AND
NORMAL DISTRIBUTIONS
DEVIATION
a
2a
3a
NORMAL
DISTRIBUTION
68.3
95.4
99.7
ACTUAL DISTRIBUTIONS
NOX
65.3
100
100
co2
61.2
100
100
°2
71.4
89.8
100
Data for actual distributions taken from TRW field test.
See Section VI.
Tabular values are percentage of data points which differ
from the mean value by less than the specified deviation.
15
-------
NOV DISTRIBUTION:
/\
MEAN VALUE
734 PPM
a = 25.8 PPM
12
NORMAL
ERROR
DISTRIBUTIONS
NUMBER INDICATES
NUMBER OF DATA POINTS
WHICH DIFFER FROM MEAN
VALUE BY SPECIFIED
AMOUNT (0 to -a/2
FOR THIS BAND)
DEVIATION FROM MEAN VALUE
Figure 3. Histogram of deviations from mean value
for NO traverse
A
16
-------
the normal distribution curve. The conclusion reached from Table I and
Figure 3 is that the standard deviation used is a reasonably accurate
reflection of the actual data distribution.
Given the validity of the standard deviation as a measure of spread,
we can now define a parameter to indicate the degree of stratification in
a stream. For most engineering applications, a 95% level of certainty
with regard to the accuracy of a measurement is acceptable. This corres-
ponds to a band of + 2o about the mean value. Consider the NO traverse
~™ /\
as an example. The mean concentration was 734 ppm with a standard
deviation of +26 ppm, or 3.5%. By definition, then, we will say that
the stratification level was +7% (i.e. 2a) from the mean, which means
that the probability is 95.4% that a point measurement taken anywhere
in the sample plane will deviate from the mean concentration by less than
+_7%. Stratification level is being defined as +2o from the mean concentra-
tion value in the sampling plane.
4.3 DESIRED SYSTEM ACCURACY
The proposed relation which we wish to consider for determing mass
flow rates of gaseous constituents is
AT
*1 ~ Np ^i ^-P ("? (16)
The uncertainty in the measurement may be given by (see Reference 3 and/or
Appendix A of Reference 1):
0 3m.. 2 2 3m.. 2 „ 3m, 2 2
rr . H
A
"1 yi us s
where
a = standard deviation of parameter x.
The terms N T,, p., R, and M. are considered to be known constants and
,55 I
as such are not error sources. Since each of the three variables is to be
determined independently of the others, equation (17) can be written
a2m. aA2 ffVt a^s
"i-i A2 v i uc
\ is
17
-------
The following assumptions may be made for normal field measurements:
• Area: OA = .01A
This is based on the assumption of a 2o uncertainty of 1% in the duct linear
dimensions.
• Velocity: a—= .04u_
This is based on an assumed 2a uncertainty of 8% for the average velocity
as determined by techniques developed during Phase I of the program, using
currently available hardware. Substituting, we get
9 2
2. a _
am- y.
~2L= -0017 + ^- (19)
Itl .• y .j
There are two main contributions to ov^: The sampling method error and
the error associated with the sampling hardware itself. Typical continuous
gas analyzers have published accuracies of about +1% of full scale. If
we arbitrarily assume that averages readings will occur at about 50% of
full scale, then the 2o uncertainty for the instrument would be +2% of
the reading. If we can obtain a 2a uncertainty of +5% for the sampling
method error, we would then have
2m
am. 99
£• = .0017 + [(.Oir + (.025H = .002425 (20)
so that
am.
— = +.049 (21)
m.
and
2am = .098 m. (22)
which gives about a +.10% uncertainty in the constituent mass flow rate.
It must be noted that this uncertainty does not include errors due to the
non-proportional sampling assumption underlying equation (16). If the
18
-------
2o error due to this assumption is +3%, then we get a +10.3% uncertainty
for the measurement of m..; if the 2a error is +_6%, then we get a +_11.5%
uncertainty for m^.
The conclusion is that the greatest uncertainty lies in the measure-
ment of the average velocity. While the +8% uncertainty for this
measurement can be reduced through careful selection of sampling locations
(rf a good one exists in the stream of interest), extensive in-place
calibration, and/or very expensive equipment, the quoted figure represents
a valid achievable level at reasonable cost. Preliminary determination
of stratification levels suggests that it is reasonable to expect 2a
accuracies on the order of +6% or better for the measurement of average
concentration. This combines with the velocity uncertainty to give a
basic constituent mass flow rate uncertainty of +10%. An additional
uncertainty due to non-proportional sampling errors would perhaps raise
the emission uncertainty to between +10.3% and +11.5%. This accuracy
level would appear to be reasonable for a continuous monitoring system.
One additional point should be made, and that is that the accuracy of
the total gas emission rate$ m, will be better than the constituent mass
flow rate mi since the uncertainty in m is only slightly larger than that
for ug. This is discussed in Reference 1. If 2o- = +8% of U , then we
would expect 20^ = +8.4% of m. S
19
-------
SECTION V
TASK II FLOW ANALYSIS PROGRAM
The purpose of this task was to formulate a computer program to
evaluate stratification data and determine the applicability of various
sampling techniques. Expected program inputs would be arrays of gas
concentration and velocity data. In Task I, the desirability of using
the average spatial concentration as opposed to the proportional sampling
average concentration was discussed, so it would be desirable for the
program to investigate this area. Since Phase I work for velocity
determination resulted in development of the Row Average technique, it
was decided that the program should assess the applicability of this
technique to gas sampling. The finalized program performed the desired
analyses by using the basic input data to predict the local velocity and
concentration at any desired point in the sampling plane.
5.1 PROGRAM DEVELOPMENT
When the program was conceptualized, there was little certainty
regarding the best sampling techniques to be evaluated. Possible approaches
included:
a. Wai den's C02 tracer method
b. Multi-point proportional sampling
c. Multi-point spatial sampling
d. Single point sampling
e. Row Average sampling
f. Use of sampling data to predict a better sampling site further
downstream.
It was decided that the basic program should at a minimum compute propor-
tional and spatial sampling mean values, and have the capability to
determine local values at any point in the sampling plane. This eventually
resulted in the capacity to evaluate approaches b-e. The C02 trace method
was evaluated separately, and is discussed in Section VI. Approach f
20
-------
would involve mixing calculations to predict lower stratification levels
at a given distance downstream. The extent of stratification in the data
examined was low enough that this approach was not felt to be justified.
The final program configuration incorporates the following features:
• Expansion of input data matrix into a 525 point
(21 x 25) matrix
• Calculation of average velocity, average spatial concen-
tration, and proportional average concentration
• Calculation of 21 row averages across the duct
• Computation of profiles of constant concentration in
intervals of 5% of the mean value
The program as it exists could easily be modified to evaluate the
performance of any proposed sampling array for either spatial or propor-
tional sampling. Due to the acceptability of the results for the Row
Average technique, other methods have not been investigated during the
program.
5.2 PROGRAM DESCRIPTION
The program itself is presented in Appendix A. The following is a
description of the program and how it performs. Major steps in the
program are:
1. Data Input - Arrays of velocity and concentration data
are entered.
2. Array Expansion - The input arrays are expanded to
25x 21 arrays using a TRW developed subprogram to
interpolate between input points and determine values
for the larger arrays. The subroutine is TRW BVIC
(Bivariate Interpolation Subroutine), which uses
polynomial interpolation to calculate the local values
from the input values. The program assumes that the
concentrations and velocities can be described by
continuous single valued functions everywhere in the
plane, which is physically realistic. For the input
21
-------
arrays, there were no data taken at the duct walls, so
no boundary conditions (i.e. conditions at the wall)
were input. It is of course recognized that the velocity
must always be zero at the wall. It has also be recog-
nized that for the data considered, the flows were very
non-developed, and consequently thin boundary layers
would be present since there was no opportunity for
them to become significantly large. It is felt that
ignoring boundary layer development (other than that
which showed up in the input data itself) by not speci-
fying a no-slip condition at the wall leads to smaller
errors than would be obtained with that boundary
condition. It is believed that imposition of a no-slip
condition would lead to inaccurately low predicted
velocities near the wall.
3. Calculation of Averages - Using the 25 x 21 arrays, the
program calculates three sets of 21 row averages:
velocity, flow proportional concentration, and spatial
concentration. In each case, the 25 points along each
line are averaged using Simpson's rule. The 21 row
averages are then averaged in the same way to produce
the three mean values.
4. Calculation of Profiles - The program takes the mean
flow proportional concentration and velocity and calcu-
lates values in 5% intervals on either side of the mean.
It then uses the BVIC subroutine described above to
calculate the points in the sampling plane corresponding
to the locus of each specified interval. The purpose
of this is to provide data which can be presented in
the form of a concentration map, showing profiles in 5%
increments from the mean. This allows for quick visual
inspection of the sampling data, including stratification
level, distribution of concentrations in the duct, and
comparisons of data for various constituents and temporal
variations of any one gas.
22
-------
5.3 TYPICAL PROGRAM RESULTS
Results for CC^ sampling by an Exxon team are shown in Table 2 and
Figures 4 and 5. The test is discussed in more detail in Section VI.
The duct measured 3.182m x 8.269m, and a 6 x 8 sampling array was used.
The first part of Table 2 shows results of calculations based on the 48
sampling points. The COp stratification level, defined by 2o, was +11%
of the mean C02 value. The second part of the table shows results from
the computer program. As would be expected, the computer results agreed
very closely with the sample point results. Of particular note is the
fact that the spatial mean values agreed with the flow proportional
mean values to within 1%. Profiles for the run are plotted in Figure 5.
The maximum deviations from the mean are about +15% and -25% of the mean
value. Row averages, where a row is being defined as a line parallel to
the short sides of the duct (vertical in Figure 5) are shown in Figure 4.
The plot shows that row averages taken between 5% and 85% of the distance
from the left wall deviate from the mean flow proportional concentration
by 3% or less. All row averages plotted are spatial averages.
23
-------
Table 2. TYPICAL ANALYSIS PROGRAM RESULTS
Exxon Run 1 - C02 Data
Direct calculations from field measurements:
48 point array (6 x 8)
SPATIAL C02 MEAN VALUE
STRATIFICATION LEVEL, 20
FLOW PROPORTIONAL AVERAGE
13.83% of dry gas
+11% of mean value
13.71% of dry gas
Computer Calculations:
273 point array (13 x 21)
SPATIAL C02 MEAN VALUE
FLOW PROPORTIONAL AVERAGE
13.82% of dry gas
13.69% of dry gas
24
-------
RATIO OF ROW AVERAGE CONCENTRATION
TO PROPORTIONAL AVERAGE CONCENTRATION
1.10
1.08
1.06
1.04
1.02
1.00
.98
.96
.94
.92
.90
.60
DISTANCE FROM DUCT WALL TO ROW
LOCATION, PERCENT OF DUCT WIDTH
100
Figure 4. Row Average analysis for C02
concentration map—Exxon run 1
25
-------
ro
DUCT DIMENSIONS 8.27m
3.18m
ROWS VERTICAL
—' \
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 13.69% (DRY GAS)
Figure 5. Normalized C02 concentration distribution—Exxon run 1
-------
SECTION VI
TASK III MAPPING TECHNIQUE EVALUATION
This task involved analytical evaluation of continuous gas sampling
techniques. Two separate groups of data were involved: one was obtained
through the courtesy of Mr. Michael Gregory of Exxon Research and Engi-
neering Company and Dr. Mike Barnes of EPA. The data were obtained at
TVA's Widows Creek Unit No. 7, under EPA Contract No. 68-02-1722, which
is concerned with stratification assessment in fossil fueled power plants.
The examined data consisted of two traverses for velocity and for C02> S02»
NO, and Op concentrations. The data are hereafter referred to as the
Exxon data. The second data group were obtained from the Wai den stratified
gas final report, and consist of four sets of velocity and S02 concentra-
tion data obtained at a TVA power plant in Shawnee, Kentucky. These data
are referred to as the TVA data.
The data were analyzed through use of the computer program described
in the last section. In addition, the Exxon data were used to evaluate
the Wai den C02 tracer technique. Results are presented and discussed
bel ow.
6.1 EXXON DATA
The sample plane location is shown in Figure 6. It was located
approximately 1.6 effective diameters downstream of an elbow. The duct
cross-section is shown in Figure 7. The raw velocity data were changed
to velocity at standard conditions, and the raw gas composition data
were used directly.
6.1.1 Concentration Maps
Concentration data are presented in Figures 8-15 (4 gases - 2 runs
each). Maps are shown as normalized concentration distribution, with
lines representing deviation from the mean flow proportional concentra-
tion. The SOp and NO maps, Figures 10-13, did not show recognizable
repeatable patterns. The COp and 0« maps, Figures 8, 9, 14, and 15,
did show a predominance of excess air on the right side of the duct, which
27
-------
~v
Sample plane located at A-A.
Short side of duct in plane of paper.
Figure 6. Location of sampling plane for Exxon data
TVA Widows Creek Unit 7
28
-------
8 PORTS THIS SIDE
8.27m
3.18m
A. Duct for Exxon Tests
6 PORTS THIS SIDE
7.32m
14.02m
B. Duct for TVA Tests
Figure 7. Duct shapes for Exxon and TVA tests
29
-------
co
o
DUCT DIMENSIONS
3..18m
8.27m
ROWS VERTICAL
PROFILES ARE VARIATION IN 52 INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 13.69% (DRY GAS)
Figure 8. Normalized C02 concentration distribution—Exxon run 1
-------
DIMENSIONS
ROWS VERTICAL
CO
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 12.77% (DRY GAS)
Figure 9, Normalized C02 concentration distribution—Exxon run 2
-------
DUCT DIMENSIONS
8.27m
ROWS VERTICAL
3.1
8m
CO
ro
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 2706 PPM
Figure 10. Normalized S02 concentration distribution—Exxon run 1
-------
CO
CO
DUCT DIMENSIONS
3.18m
8.27m
ROWS VERTICAL
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEW CONCENTRATION
MEASURED MEAN CONCENTRATION = 2042 PPM
Figure 11. Normalized S02 concentration distribution—Exxon run 2
-------
DUCT DIMENSIONS 8.27m
3.18m
ROWS VERTICAL
CO
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 305.3 PPM
Figure 12. Normalized NO concentration distribution—Exxon run 1
-------
DUCT DIMENSIONS
3.18m
8.27m
ROWS VERTICAL
CO
01
PROFILES ARE VARIATION IN 5X INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 322.5 ppm
Figure 13. Normalized NO concentration distribution—Exxon run 2
-------
DUCT DIMENSIONS
3.18m
8.27m
ROWS VERTICAL
CO
I
PROFILES ARE VARWION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 4.36% (DRY GAS)
Figure 14. Normalized 02 concentration distribution—Exxon run 1
-------
DUCT DIMENSIONS
8.27m
ROWS VERTICAL
3.1
CO
8m
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 4.56% CDRY GAS)
Figure 15. Normalized 02 concentration distribution—Exxon run 2
-------
showed up as high 02 values and low C02 values. Plant geometry would
appear to suggest air leakage ahead of the sample plane as the cause
for this stratification. There is no obvious explanation for the high
S02 and NO levels toward the right side for the first run. The high 0?
levels would seem to suggest that the S02 and NO levels should be low, as
they were for the second run.
The maps clearly show that 02 was the most stratified of the gases.
Since 02 and C02 together make up about 20% of the dry gas composition,
and since the mean C02 concentration is made higher than the 02 concen-
tration, it would be expected that the oxygen stratification would be
greater than that for C02. The maps suggest similar stratification levels
for C02, NO and S02,
6.1.2 Mean Values and Stratification Levels
Mean values of gas concentration and stratification levels are shown
in Table 3. Calculations from actual data points agreed with computer
calculations within 1% for all cases. Of particular note is the fact
that the spatial and flow proportional averages were in very close agree-
ment -- the only variations larger than 1.1% were for oxygen, where the
variation was less than 4%. This shows almost negligible coupling between
velocity and concentration and leads to the preliminary conclusion that
proportional sampling is not required for accurate measurements.
Stratification levels, as defined in Section IV, were on the order
of +10% of the mean value for C02, S02, and NO and were about +34% for
02. The higher 02 stratification level was expected, as discussed above.
Velocity stratification was notably higher than gas stratification: the
level was above +65% of the mean velocity. For all gases, the stratifi-
cation level was sufficiently high to discourage suggesting that simple
point monitoring would be adequate, especially since the concentration
maps did not show good repeatability.
6.1.3 Evaluation of COp Tracer Method
For each of the two runs, the ratios of S02 and NO concentrations to
C02 concentration were calculated at each of the 48 traverse points. The
mean ratio and standard deviation were then calculated. The deviations
38
-------
Table 3, MEAN VALUES AND STRATIFICATION LEVELS FOR EXXON DATA
to
• ;
MEAN VALUES
FROM DATA POINTS
Spatial
Flow Proportional
FROM COMPUTER
Spatial
Flow Proportional
STRATIFICATION
LEVEL, 2a,
% of mean value
Velocity
Run 1
4.957
4.912
+65.7
, m/sec
Run 2
4.400
4.364
+70.2
CO
Diim 1
Run 1
13.83
13.71
13.82
13.69
+11.0
2> %
Dun 9
Kun c.
12.91
12.79
12.91
12.77
+14.6
S02,
Dun 1
KUll 1
2712
2705
2711
2706
+11.8
ppm
Run 9
KUll C.
2063
2039
2063
2042
+12.4
NO, ppm
Run 1 Run ?
307.1
305.2
307.0
305.3
+10.6
324.4
323.3
324.2
322.5
+10.8
02,
Run 1
4.217
4.355
4.235
4.364
+36.4
%
Run 2
4.379
4.548
4.390
4.560
+32.2
Maximum variation between spatial and flow proportional average - 3.9%
Average variation between spatial and flow proportional average =1.4%
-------
in percent of the mean value are shown in Table 4. For each case, the 2a
deviation shown is about 50% higher than the stratification level of the
gases, which suggests that uniformity of the ratios is not good.
6.1.4 Row Average Evaluation
For this analysis, rows were selected parallel to the short sides
of the duct. This was done for several reasons. The first is that this
is the same direction in which velocity row averages would be taken (see
Reference 1). Other reasons are strictly practical: a row average could
be acquired through use of a single continuous sampling probe, as described
in Section VII. Existing ports in the duct are in the long side, which
suggests that one of these ports could be used to install the probe, or
at the least that there would be access if a new port were needed. Also,
it would be much easier to handle a three meterprobe than an eight meter
probe.
Row average data are shown in Figure 16. Results are not quite as good
as had been hoped. As discussed in Reference 1, the Row Average technique
works well for velocity measurement when the flow is conditioned to give
fairly constant row averages across the duct. For velocity, this occurs
immediately downstream of an elbow. For gas sampling, there is no common
mechanism which conditions the flow in an analogous manner. The COp row
averages for run 1 show a desirable pattern; the row average deviates
from the overall average by less than 3% over most of the duct. Conclu-
sions to be drawn are that repeatability between runs 1 and 2 was not
good, as noted previously, and that more uniform row averages would be
obtained if the rows were taken horizontal instead of vertical in Figures
8-15. The latter is clear because there is obviously greater stratifica-
tion from right to left than from top to bottom in those figures. The
first rule established for selection of row directions for velocity
measurement was that the rows should be parallel to the direction of
greatest stratification. It is unfortunately true for these runs that
the direction of highest gas stratification is perpendicular to the
direction of highest velocity stratification.
The problem of poor temporal repeatability could not be reasonably
assessed through analysis of only two runs, so the following TVA data
were examined primarily for that purpose.
40
-------
Table 4. C02 TRACER METHOD ANALYSIS OF EXXON DATA
RUN 1
so2/co2
N0/C02
RUN 2
so2/co2
N0/C02
2o, % of mean value
+15.2
£16.0
+16.6
^15.2
Explanation of tabular values:
2a = +15.2% of mean value for S02/C02 ratio means
that there is a 95.4% probability that the ratio
of the S02 reading to the C02 reading at any
point in the duct will agree with the average S02/
C02 ratio within +15.25S
41
-------
RATIO OF ROW AVERAGE CONCENTRATION
TO PROPORTIONAL AVERAGE CONCENTRATION
1.10
1.08
\°Z I
JLj ./^ . /
.90
1.10
1.08
1.06
10 20 30 40 50 60 70
DISTANCE FROM DUCT WALL TO ROW
LOCATION, PERCENT OF DUCT WIDTH
0
80
feO Ij
/ /
RUN
2
RUN
1
0 10 20 30 40 50 60 70 80 90 100
Figure 16. Row average analysis for gas
concentration maps—Exxon runs 1 and 2
42
-------
6.2 TVA DATA
No drawings were available showing location of the sample plane. The
duct cross-sectional shape is shown in Figure 7. A total of four pairs
of S02 and velocity maps was examined. The velocity data were not obtained
simultaneously with the S02 data, but it is felt that this does not compro-
mise the results to be discussed.
6.2.1 Concentration Maps
Concentration data are presented in Figures 17-20, using the same
format as for the Exxon data. There are again no immediately recognizable
stratification patterns among the runs. Stratification does appear to
be greater in the vertical than in the horizontal direction.
6.2.2 Mean Values and Stratification Levels
These values are shown in Table 5. Agreement between actual data
point calculations and computer calculations was not quite so good as for
the Exxon data. This is presumably due to the fact that the TVA traverses
used only 24 points (4x6), whereas the Exxon traverses involved 48 points
(6 x 8). The 1.75% average variation between spatial and flow proportional
mean concentrations is compatible with the Exxon results, and continues
to demonstrate the adequacy of spatial sampling. It would appear that the
variation is proportional to the stratification level in the gas -- both
the variation and the stratification level for the TVA data are higher
than for the Exxon COp, NO, and SOp data, but lower than for the Exxon
Op data.
6.2.3 Row Average Evaluation
Row Average data are presented in Figure 21.
The first three runs do not exhibit as great a uniformity as would
be desired, but do show fairly good repeatability. The change in shape
on the left side of the duct in Run 4 is clearly undesirable. A row
average taken at the center of the duct would show good correlation for
all four runs — much better than would be attained for the Exxon data.
43
-------
DUCT DIMENSIONS
14.02m-
RftWS VFRTTTfll
/ J '
+ 10 - •*'
x' / 'x^-rz-x ^NNXX - *-• -^^<--c:
' ,' ''''"'---^i^^^*^^ I--~'^'"N.X ^ "~ ^ ^ ^ '^.^ •^i'^'^''^'''^'
f i/S* S * ^*~ —- ~-«^«^>»X. XS vX ^^ ^^ ^ ^^SS^^C.Z.'Z
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 2011 PPM
Figure 17. Normalized S02 concentration distribution—TVA run 1
-------
DUCT DIMENSIONS 14.02m
7.32m
ROWS VERTICAL
en
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 2149 PPM
Figure 18. Normalized S02 concentration distribution—TVA run 2
-------
DUCT DIMENSIONS
7.32m
14.02m
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 2100 PPM
Figure 19. Normalized S02 concentration distribution—TVA run 3
-------
DUCT DIMENSIONS
7.32m
14.02m-
ROWS VERTICAL
-V-Y \ v
, . . . /'//'+'
-------
Table 5. MEAN VALUES AND STRATIFICATION
LEVELS FOR TVA S00 CONCENTRATION DATA
MEAN VALUES,
PPM OF S02
FROM DATA POINTS
Spatial
Flow Proportional
FROM COMPUTER
Spatial
Flow Proportional
STRATIFICATION
LEVEL, 2a,
% of mean value
RUN
1
2059.5
2056.9
1982.5
2011.2
+20.4
2
2204.4
2207.2
2085.0
2148.7
+26.5
3
2099.7
2109.2
2071.9
2100.3
+17.1
4
2036.5
2047.5
2029.5
2054.5
+19.7
Average variation between spatial and flow proportional
average = 1.75%
48
-------
RATIO OF ROW AVERAGE CONCENTRATION
TO PROPORTIONAL AVERAGE CONCENTRATION
.90
DISTANCE FROM DUCT WALL TO ROW
LOCATION, PERCENT OF DUCT WIDTH
Figure 21. Row average analysis for S02
concentration maps—TVA runs 1-4
49
-------
6.3 PRELIMINARY SAMPLING TECHNIQUE CONCLUSIONS
For any given gas concentration distribution, it is a trivial matter
to select a sampling array which would obtain a representative sample.
Problems arise when more than one gas species must be sampled, and when
the gas distributions change as a function of time. The greatest difficulty
with regard to continuous gas sampling is to effectively handle the
temporal changes, since they cannot be predicted easily if at all. Tem-
poral changes in velocity distributions are not such a problem. Velocity
distribution is primarily a function of local duct geometry, which is
normally fixed. Changes in gas distribution are much more a function of
combustion characteristics and air leakage.
Two major conclusions are apparent from the data examined. The first
is that stratification levels were too high to justify single point sampling
for a continuous monitoring system. The stratification level itself is a
measure of how accurate a single point sample is likely to be. For the
twelve runs examined, the mean stratification level, 2a» was + 18.6%,
which is far from the proposed desired accuracy of +_ e% for average
concentration. Single point sampling is also intuitively more prone to
temporal variation errors than any multi point sampling system. As a
general rule, single point sampling should only be considered acceptable
in situations where the stratification level is not worse than the desired
accuracy of the concentration measurement.
The other conclusion is that spatial gas sampling is as accurate in
determining effluent rates as flow proportional sampling, within about
2%. This justifies development of continuous monitoring systems
which acquire spatial samples, rather than more difficult (therefore
expensive) flow proportional samples. The spatial sampling method examined
has been the Row Average method, which showed reasonably good results for
the TVA data, but poorer results for the Exxon data. One reason for the
latter was high stratification levels normal to the adopted row direction.
Agreement between concentration levels of S02 and NO with respect to C02
was rather poor for the Exxon data. It was decided at this point to
select the Row Average technique for field evaluation, due to its simplicity
and easy adaption to hardware, as well as due to the promising results
from the TVA data analysis.
50
-------
SECTION VII
TASK IV SAMPLE PROBE DEVELOPMENT
The Row Average technique for gas sampling was discussed in the
previous section. The next objective was to develop a sampling probe to
satisfy the requirements of Row Average sampling. In this section, probe
design requirements are formulated mathematically and fabrication of a
prototype probe for the field demonstration is discussed. Also included
are suggestions for improved probe designs.
7.1 PROTOTYPE PROBE DESIGN
One of the continuous sampling techniques considered in Section VI
was the Row Average method. Since this method is probably the easiest
one to develop hardware for,it was decided to produce prototype sampling
probes for the Phase II field demonstration described in Section VIII.
The basic design criteria for such a probe are quite simple:
• The probe should draw samples at a number of evenly
spaced points along a line across the duct.
• The same sample rate should be maintained at each
sampling point.
t The samples from each point should be mixed and drawn
out through a single sampling line.
The probe design is illustrated in Figure 22. A small sampling hole
was located at each of eight preselected locations corresponding to an
even distribution along the probe. It was decided that control of flow
distribution among the holes would be accomplished by making the holes
small enough that the pressure drop across each hole would be large
compared to pressure changes along either the inside or outside of the
probe. Mixing of the acquired samples from each port would clearly be
accomplished within the probe body. The most direct alternative to the
technique used would be to have a separate sample line from each point.
Each line would then need a control value and flowmeter to insure equal
51
-------
01
ro
MIXING VOLUME
DUCT
WALL
PROBE BODY
FLOW
OUT
\
SAMPLE PORT
(TYPICAL)
\
\
K
FLOW
DIRECTION
Figure 22, Multi-port sampling probe basic design
-------
flow from each point. The lines would then feed into a suitable plenum
chamber and be mixed for delivery to the analyzers.
Once the number and location of sampling holes have been selected,
the only major analytical task is to select the hole sizes. This pro-
ceeded as follows:
• Determination of worst case pressure variation
Outside the probe: The Phase II field demonstration
was to be conducted at the same power plant as the
Phase I field test (see Section VIII). The maximum
observed velocity during the Phase I test was about
11 m/sec, which corresponded to a differential pressure
Of p .p^ = Ap = .75 torr. A sample flow rate of .5 I/sec
M CFM) was selected, along with a probe size of 2.54 cm
(1 in.). This corresponded to a maximum flow velocity
within the probe of about 1.2 m/sec. Since this was con-
siderably below the expected variation in freestream
conditions outside the probe, the external conditions
determined the maximum expected pressure variation.
• Desired pressure drop across orifices
It was arbitrarily decided that the duct pressure
variation should not be more than 5% of the pressure
variation across the sampling ports. This meant that
the latter should be at best 20x .75 torr = 15 torr.
• Determination of hole diameter
The sample volumetric flow is given by
V-n7(.61A) (23)
where V = volumetric flow
n = number of sampling ports
v = average velocity of gas through sample ports
A = sample port area
.61 = sample port discharge coefficient
53
-------
In terms of the pressure drop in the orifices, this becomes
71 2 J 5 T (24)
"oo
where d = hole diameter
R = gas constant
T = absolute temperature
p = absolute pressure ahead of orifice
p = static pressure in orifice
Solving for d, we get
1/2
V
If V = .5 I/sec
n = 8
T = 450° K
oo
pQ = 715 torr
p - 715-15 = 700 torr
1 CO
then d = 1.295 mm (.051 inches)
The nearest drill size was a #55 drill, 1.32 mm (.052 inches) diameter,
so this size was used. See Section VIII for discussion of actual probe
use.
7.2 SUGGESTIONS FOR IMPROVED PROBE DESIGN
The above design showed its capability to avoid clogging of the
orifices with particulate, as discussed in Section VIII. The problem
which did show up, however, was a buildup of particulate within the probe.
In an operational continuous monitoring system, such problems would be
unacceptable. It would appear that the only reasonable way to handle
the particulate problem would be to filter out the particulate before it
can enter the probe. Probe inlets to accomplish this are shown in
Figure 23. In each case, the key element is the filter. The operating
54
-------
A. At each specified sampling point, a porous filter insert is installed,
Uniformity of pressure drop, hence flow rate, is assured through
control of insert porosity.
PROBE
BODY
POROUS
INSERT
in
en
B. A combination filter/liner is used to obtain a sample along the entire
probe length. This assumes that a true Row Average sample is acquired.
The filter and liner form a one piece bonded unit.
PROBE
BODY
LINER
FILTER
C
Figure 23. Suggestions for continuous gas sample probe modifications
to prevent particulate entry into the probe
-------
principle is exactly the same as for the fabricated probe: the pressure
drop must be identical through each section and must be large compared
with external pressure variations. This requires uniformity of the filter.
So long as the particulate is reasonably dry, a slow rate of buildup
would be expected on the filters since a self-cleaning effect was noted on
the prototype unit. To clean the probe, an automatic intermittent purge
capability should be provided as part of the instrumentation system. The
purge intervals would be determined by the application, and the accumulated
dust would be blown back into the main stream. This type of purge system
was successfully demonstrated in a different application as part of the
Phase II field test (see Reference 1).
Recent TRW experience with Kapton liners for sampling probes and
our knowledge of properties of various high temperature polymers suggests
that such materials may be quite applicable in the production of filters
for the application in Figure 23. The desirable properties of polymers
for this application include chemical inertness and physical integrity
for temperature as high as 500°C. They are also castable into any
desirable shape, and it is believed that very accurate control of pore sizes
can be obtained. The latter is the property which controls pressure drop
across the filter.
56
-------
SECTION VIII
TASK V - FIELD DEMONSTRATIONS
The task objective was to demonstrate the applicability of techniques
and hardware for continuous gas sampling. Two field demonstrations were
performed at the Reid-Gardner Station of the Nevada Power Company, located
at Moapa, Nevada. The plant is about 72 km. northeast of Las Vegas. Each
test lasted about eighteen days. The first was begun in September, 1974,
and the second was begun in February, 1975. The primary purpose of the
first test was to demonstrate hardware and techniques for volumetric flow
measurement which had been successfully tested in the laboratory. Gas
sampling data were taken in support of the volumetric flow measurements
and to assess stratification levels. The primary purpose of the second
test was to demonstrate techniques and hardware for continuous gas
sampling. Volumetric flow measurements have been fully documented in
Reference 1. Only flow data pertinent to this report are discussed below.
8.1 FACILITY DESCRIPTION
The plant presently consists of two Foster-Wheeler 120 megawatt
boilers, with a third under construction. All work was performed on the
#2 unit, shown schematically in Figure 24. Flow from the boiler is
separated into two streams which pass through Lunjstrom rotary air pre-
heaters. The ducts at the two preheater outlets go through a shape
transition and then rejoin upon entering a mechanical dust collector. A
row of test ports is located on each duct just ahead of the dust collector.
This is shown in Figure 25. All rectangular duct mapping was done at
these locations.
When the plant was first constructed, the dust collector exhausted
directly into the stack. Subsequently, a venturi scrubber and separator
were added by Combustion Equipment Associates. When the scrubber is on,
the flue gas is diverted at the dust collector exit and routed to the
scrubber, where it is processed and fed into the stacks. If the scrubber
is not on, the flue gas goes directly into the stack after leaving the
dust collector. The inlet to the stack is about 13 m above ground level.
57
-------
en
oo
Plane
A-A
Scrubber
Flow routed to and from
scrubber at plane A-A
A = Annubar
S = Sampling Probe
Stack
Location 2
Dust
Collector
V
Location
1
S A
Duct 1
Preheaters
A S
Boiler
Location 1
Duct 2
Location 2
Figure 24. Schematic of Moapa power plant
-------
Duct 2 is a mirror image of duct 1
FLOW^
DIRECTION
SAMPLE PORTS (ROW NUMBER)
AMNUBAR LOCATION
SAMPLING
PLANE
S/WPLE POINTS
1
TRAVERSE MAP -
49 POINTS
Figure 25. Duct 1 geometry
59
-------
Sample ports are located in the stack at the 31.4 m level, or about four
stack diameters above the inlet. All circular duct testing was performed
at this location in the stack.
8.2 TEST CONDUCT
8.2.1 Gas Sampling Hardware
Velocity instrumentation is described in Reference 1. The continuous
flow monitors used were Ellison Annubars. Two were used in the first
test — one in duct 1 and one in the stack, as shown in Figure 24. A
third was added for the second test and installed in duct 2. Gas analyzers
used during the first test included a Beckman NDIR S02 analyzer, a Carle
thermal conductivity gas chromatograph to measure CO, C02> N2, and 02
concentrations, and a Thermo Electron chemiluminescent N0¥ analyzer.
A
Post test data examination revealed that the S02 analyzer had a very slow
time response (^20 minutes) which was not evident during the test, and
which resulted in incorrect instrument calibration. The result was that
the S02 data for the first test had to be discarded. Gas sampling for
the first test involved point sampling only.
For the second test, a new Carle gas chromatograph was purchased.
The new unit had an automatic sampling capability so that it could
operate unattended. The older model required manual operation. A
pulsed fluorescent S02 analyzer was loaned to us for the second test
through the courtesy of the Thermo Electron Corporation and their
Western Regional office manager, Mr. Jim Nelsen. The NO analyzer was
^
the same one used during the first test. The gas chromatograph had its
own integrator and digital paper tape printer. Outputs of the S02 and
NOX analyzers were fed through a scanner to a digital voltmeter and
paper tape printer. Data output intervals varied from ten seconds to
one hour, with a nominal ten minute interval. Data reduction was accom-
plished by use of a digital computer upon completion of field work.
The point sampling probe used during the second test is shown in
Figure 26. The sample line was heat traced up to the condenser used to
remove water vapor. Four continuous sampling probes were fabricated
for the second test. Design is described in Section VII, and shown in
60
-------
Nozzle
Thimble Filter
Assembly
Probe
Body
Heat Traced
Sample Tube
I
Heat Traced
Sample Line
Figure 26. Point sampling probe
-------
Figure 27. The probes were installed at the locations shown in Figure 24.
Heat traced lines were used as far as the condenser.
A schematic of the sampling trains is shown in Figure 28. For the
second test, two large plexiglas water knockouts (condensers) were fabri-
cated to allow for continuous sampling (one for the duct area and one
for the stack). Two pumps from Aerotherm "Super Sniffer" sampling trains
were used to provide the desired flow rates. The second pump was used as
a backup in case of failure.
8.2.2 Extent of Testing
Sample traverses were performed in duct 1 during the first test to
assess stratification levels and provide composition data for flow
calculations. In addition, time variation of concentrations at a fixed
point was also investigated. Traverses were performed in both ducts 1
and 2 during the second test in addition to the data obtained from the
multi-point probes.
A total of four basic plant conditions occurred during the first test,
each about equally often. The unit load was held at maximum KllO-120 MW)
during the PM hours and at about half that during the AM hours. The
scrubber was on about 40% of the time. The unit was down about five days
during the test. Full load was maintained during the second test, the
variation being from about 100 to 123 MW. The scrubber was on full time
except for one five hour period. There was no plant down time during the
second test.
8.2.3 Problems
The only notable gas sampling problem during the first test was mal-
function of the SOp analyzer. The absolute value of the NO data for
™" A
the test seemed unusually high. The analyzer was factory serviced betweer
tests and gave more reasonable data for the second tests. Since the
objective during the first test was to determine variations rather than
absolute values, the data were considered acceptable for that purpose.
Many problems occurred with the velocity measurements during the first
test, and these are documented in Reference 1.
62
-------
Heat Traced
Sample Tube
Purge
Plug
Coarse
Parti culate
Fi1ter
Probe
Body
733333:13
co
Duct
1
y
T
7 L
/
Heat Traced
Sample Tube
Mall
r
* • *
/
y
Approximate
Location of
Filter
• • •
Sampl e '
Port
Duct
Mall
• •
Figure 27. Multi-hole sampling probe for field demonstration
-------
SAMPLE
PROBE
I^^MM
CONDENSER
A.
PUMP
1 1 1
N0x
ANALYZER
so2
ANALYZER
GAS
CHROMATO-
GRAPH
SAMPLE
PROBE
CONDENSER
configuration prior to pump leakage
problem, 1975 test
1 1 1
M0x
ANALYZER
so2
ANALYZER
GAS
CHROMATO-
GRAPH
PUMP
B. Configuration after pump leakage problem discovered,
1975 test
Figure 28. Schematic of gas sampling train for field demonstrations
-------
The situation reversed for the second test. There were negligible
problems with the velocity measurements, but many with the gas sampling
work. In retrospect, the primary cause of difficulties was an unfortunately
tight test schedule prior to the actual field work. Work on the program
came to a virtual standstill during December of 1974 due to the fact that
data from various sources were showing lower gas stratification levels
than had been expected when the program was conceptualized. It was
eventually recognized that although the stratification levels were lower
than expected, the problem of obtaining a representative sample was far
from trivial. Approval to proceed as originally planned was received
early in January. For several reasons, it was decided that the second
field demonstration should be conducted at the same place as the first.
A severe time constraint became apparent when we were told that the
unit would be down for an extended period, starting around the first of
March, for annual maintenance. This gave us a month to develop the
specialized hardware needed to perform the test, and still have a
reasonable length of time to conduct the test. The major difficulties
were with procurement: the new gas chromatograph did not arrive in time
to be thoroughly checked out in the laboratory. This resulted in much
down time for the instrument in the field. The S02 analyzer did not
arrive until the sixth day of the test. A faulty integrated circuit
(replacement cost 59$) invalidated the output for several subsequent
days until the problem could be located through consultation with the
factory, and a replacement flown in from Los Angeles.
Other problems during the test included accidental damage to the
water knockout units, and accidental ingestion of a large slug of water
in the sample line, which rendered the S02 analyzer inoperable during
the last two days of testing. A significant problem was the development
of a leak in the Aerotherm pump midway through the test. The pump was
Initially placed upstream of the analyzers to provide an acceptably high
inlet pressure to the Instruments. The backup pump was also found to
have too high a leak rate. The pump then had to be placed downstream of
the analyzers, as shown in Figure 28. Since the analyzers could not
maintain an adequate flow rate due to the low line pressure ahead of the
pumps, it was necessary to manually throttle down the pump at each
65
-------
sampling point to allow the line pressure to rise to an acceptable level.
Data were not compromised using this procedure, but it precluded any
further continuous sampling. As a result, the only continuous monitoring
data obtained were in duct 1, since the data had been taken prior to
the leakage problem.
It is felt that the gas sampling data obtained during the second test
were adequate to demonstrate the chosen techniques. The problems which
occurred simply reduced the amount of data which could be taken.
8.3 FLOW DATA CORRELATION
The gas analyzers were calibrated with bottled calibration gases.
This was normally done twice a day, or before and after each traverse.
Coal usage per day and coal analysis data were coupled with test composi-
tion measurements of C02 in the ducts to make independent flow calculations.
Coal data were supplied by plant personnel.
8.4 TEST RESULTS
8.4.1 Composition Measurements and_ Plant Reference Data
Foster-Wheeler specifications for full load nominal bulk composition
at the duct sampling planes are as follows:
Gas % Met Gas % Dry Gas
N2 76.13 80.14
C02 13.24 13.94
02 5.63 5.92
H20 5.00
Results of the 1974 sampling measurements in duct 1 are summarized in
Table 6 for full load conditions. No full traverses were obtained at half
load due to the plant schedule. We did obtain two "half traverses" on
different days which were used to calculate nominal composition at half
load. Averages for the 1974 test were:
66
-------
Table 6. MEAN VALUES AND STRATIFICATION LEVELS
FOR 1974 FIELD DEMONSTRATION
GAS
co2
% of dry
gas
NOX,
ppm
o2,
% of dry
gas
DUCT
1
1
1
RUN
-
1
2
-
MEAN VALUES
Spatial
15.84
734
800
4.32
Flow
Proportional
15.85
735
805
4.30
STRATIFICATION
LEVEL, 2a,
% of Mean Value
+7.4
17.0
+16.0
+23.2
67
-------
Full Load Half Load
Gas % Wet Gas % Dry Gas % Wet Gas % Dry Gas
N2 76.65 79.85 79.96 82.43
C02 15.22 15.85 8.81 9.08
02 4.13 4.30 8.23 8.49
H2° 4.00 - 3.00
Results for the 1975 test are summarized in Table 7. Recall that the
unit was only at full load during the 1975 test. Bulk composition for the
1975 test was as follows:
Gas % Wet Gas % Dry Gas
N2 75.40 80.39
C02 12.99 13.85
02 5.40 5.76
H£0 6.21
Water vapor content was calculated from ambient humidity and plant measure-
ments of water content in the coal. In 1974 testing at half load, coal
usage was half that at full load while the total flowrate was about 70%
of the full load flow rate. Thus quite a bit more excess air was used
at half load than at full load. During the 1975 test the unit was always
at full load and the gas composition was very close to the nominal
specifications.
Good correlation was obtained for total mass flow as well as for gas
composition. Total flow measurements from the Annubars were compared to
flow calculations based on daily coal usage and average C02 concentrations
in the ducts. Results are presented in Tables 8 and 9, which are taken
from Reference 1. Table 8 shows the calculation technique and results
for two days of testing in 1974. Table 9 shows similar results for 1975
testing. The "insufficient data" designation in Table 9 means that the
pressure transducer used to monitor the duct 2 Annubar was being used for
other purposes on the days noted. Both Tables 8 and 9 show total flow
correlations on the order of 2% between test and reference measurements.
68
-------
Table 7. MEAN VALUES AND STRATIFICATION LEVELS
FOR 1975 FIELD DEMONSTRATION
GAS
co2,
% of dry
gas
so2,
ppm
N0y,
A
o2,
% of dry
gas
DUCT
1
2
1
2
1
2
1
2
RUN
-
-
-
-
1
2
1
2
-
-
MEAN VALUES
Spatial
13.16
14.54
277
344
373
479
449
416
5.79
5.75
Flow
Proportional
13.14
14.55
278
345
373
480
451
416
5.76
5.75
STRATIFICATION
LEVEL, 2a,
% of Mean Value
+9.0
+9.0
+20.6
+19.7
+8.6
+6.9
+_14.3
+13.5
+13.4
+8.6
69
-------
Table 8 . COMPUTATION OF TOTAL FLOW FROM COAL ANALYSIS AND
MEASURED C02 CONCENTRATION, 1974
Date
10-2
10-3
Coal Used
kg
3,251,380
3,571,480
Carbon Used
kg gm-mol es
2,752,618 4.729xl07
3,023,614 5.195xl07
Average C09
% *
11.89
12.74
Total Moles
of Dry Gas
3.977xl08
4.078xl08
Total Moles
of Wet Gas
4.122xl08
4.226xl08
Date
10-2
10-3
Annubar
Average Flow
Rate, SCMS
104.5
109.4
Annubar
Total Flow
SCM
9. 026x1 O6
9.451xl06
Total Wet Gas
Volume, SCM
9.234xl06
9. 466x1 O6
Difference
%
+ 2.3
+ 1.6
-------
Table 9. AVERAGE DAILY FLOW THROUGH DUCTS, 1975
Date
2-15
2-16
2-17
2-18
2-19
2-20
2-21
2-22
2-23
2-24
2-25
2-26
2-27
2-28
Ave
Average Flow And Standard
Deviation, Test
SCM/SEC
Duct 1
63.4 ± 1.7
63.5 ± 2.8
64.7 ± 2.3
67.3 ± 2.2
65.8 ± 2.3
65.3 ± 1.6
65.1 ± 0.9
63.2 ± 2.6
64.3 ± 5.2
60.1 ± 8.5
62.5 ± 1.7
62.7 ± 1.8
60.8 ± 2.1
60.3 ± 1.6
63.5
Duct 2
X
X
64.6 ± 1.9
64.5 ± 2.4
65.2 ± 3.0
66.7 ± 2.9
64.3 ± 3.9
67.3 ± 8.2
X
67.3 ± 3.0
67.2 ± 0.9
66.7 t 2.1
X
X
66.0
Total
X
X
129.3
131.8
131.0
132.0
129.4
130.5
X
127.4
129.7
129.4
X
X
129.5
Average Flow In Ducts
From Plant Data
SCM/SEC
132.0
132.6
134.3
135.1
132.8
134.7
133.7
132.9
128.1
130.6
130.1
132.7
130.2
126.3
131.9
^Indicates insufficient data
71
-------
8.4.2 Assessment of Stratification
Stratification levels for the various gases are shown in Tables 6 and
7. They are generally of the same order of magnitude as in the Exxon data
with the exception of relatively low (^ stratification. In all cases,
excellent agreement was obtained between spatial and flow proportional
averages, further strengthening the validity of spatial gas sampling.
Concentration maps are presented in Figures 29-42. The only general
comment to be made is that stratification is generally greater in the
direction parallel to the long sides of the duct, as was the case for
the Exxon data. This may be in part due to the fact that the rotary air
preheaters just upstream of the test area have a vertical interface with
the ducting, so any leakage at that point would tend to result in hori-
zontal rather than vertical stratification. Thus results are as expected
if we assume some leakage at the preheater area.
8.4.3 Technique Evaluation
Row Average analyses are presented in Figures 43-45. In Figure 43,
the C02 curves show the most uniform row averages while the S02 data show
the least desirable results. The NOX data in Figure 44 show good results
for three of the four runs, and fairly good agreement was attained for all
gases in Figure 45.
Due to the problems described in Section 8.2.3, the only multi-hole
probe sampling data were obtained in duct 1. Results for a twelve hour
sampling period are shown in Table 10. The average C02 and Op concen-
trations for the multi-hole probe agree with the traverse averages for
duct 1 in Table 7 within 3% and 2%, respectively. The row location was
35.7% of the duct width from the left wall. As can be seen in Figure 43,
this location showed good agreement between row averages calculated from
traverse data and the mean flow proportional concentration.
Evaluation of the Wai den C02 tracer method is shown in Table 11.
Correlation was about the same as for the Exxon data.
8.4.4 Assessment of Temporal Variations
Short term and overnight point sampling data were taken to determine
changes in concentration at a point as a function of time. Data in
72
-------
DUCT DIMENSIONS
5.48m-
ROWS VERTICAL
1.65m
u>
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 15.85% OF DRY GAS
Figure 29. Normalized C02 concentration distribution-duct 1, 1974
-------
DIMENSIONS
ROWS VERTICAL
1 ' / / ' / / // J \
I /////////J,
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 4.30% OF DRY GAS
Figure 30. Nromalized 02 concentration distribution— duct 1, 1974
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
C71
1.65m
«
/
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 735 PPM
Figure 31. Normalized N0¥ concentration distribution—duct 1, 1974, run 1
/\
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
1.65m
\
O I III A \\\\
/ / I I \ \ \ °
\\\J
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 805 PPM
Figure 32. Normalized NOX concentration distribution—duct 2, 1974, run 2
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
1.65m
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 13.14% OF DRY GAS
Figure 33. Normalized COp concentration distribution-duct 1, 1975
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
1.65m
I
oo
PROFILES ARE VARIATION IN 53! INCREMENTS FROM MEAN CONCENTRATION MEASURED
MEAN CONCENTRATION = 5.76% OF DRY GAS
Figure 34. Normalized 02 concentration distribution—duct 1, 1975
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
10
1.65m
1
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 278 PPM
Figure 35. Normalized S09 concentration distribution—duct 1, 1975
-------
DUCT DIMENSIONS 5.48m *-
IO
ROWS VERTICAL
»^^
~~" — -.
\
\
\
1
1
1
/
— -/"--/
1
1
1
1
1
1
1
1
/
J 0
1
I I
1
/
X
X
X
/x
r
' **.
v V-*
s-*» x
\ >
V
\
S—^ \
s
^^
^
\
•
\
1
f
^ -^^ 1
•** 1
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 278 PPM
Figure 35. Normalized S02 concentration distribution—duct 1, 1975
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
1.65m
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 373 PPM
Figure 36. Normalized NOX concentration distribution—duct 1, 7975, run 1
-------
DUCT DIMENSIONS
5.48m
1.65m
1
co
ROWS VERTICAL
PROFILES ARE VARIATION IN %5 INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 480 PPM
Figure 37. Normalized NOX concentration distribution—duct 1, 1975, run 2
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
1.65T
00
ro
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 14.55% OF DRY GAS
Figure 38. Normalized C0£ concentration distribution-duct 2, 1975
-------
DUCT DIMENSIONS
5.48m
00
ROWS VERTICAL
1.65m
-s
/\
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 5.75% OF DRY GAS
Figure 39. Normalized 02 concentration distribution—duct 2, 1975
Y
\
-------
CD
-pa.
DUCT DIMENSIONS
1.65i
5.48m
ROWS VERTICAL
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 345 PPM
Figure 40. Normalized S02 concentration distribution—duct 2, 1975
-------
DUCT DIMENSIONS
5.48m
ROWS VERTICAL
00
in
PROFILES ARE VARIATION IN 5% INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION = 451 PPM
Figure 41. Normalized NOX concentration distribution—duct 2, 1975,
run 1
-------
DUCT DIMENSIONS
5,48m
ROWS VERTICAL
1.65i
PROFILES ARE VARIATION IN 5X INCREMENTS FROM MEAN CONCENTRATION
MEASURED MEAN CONCENTRATION =416 PPM
Figure 42. Normalized NO concentration distribution—duct 2, 1975, run 2
-------
RATIO OF ROW AVERAGE CONCENTRATION
TO PROPORTIONAL AVERAGE CONCENTRATION
1.10
1.08
1.06
1.04
1.02
1.00
.98
.96
.94
.92
.90
0
20
4CJ
60
80
100
DISTANCE FROM DUCT WALL TO ROW
LOCATION, PERCENT OF DUCT WIDTH
Figure 43. Row average analysis for C02, 02, and S02
concentration maps—field demonstrations,
duct 1
87
-------
RATIO OF ROW AVERAGE CONCENTRATION
TO PROPORTIONAL AVERAGE CONCENTRATION
1.10
1.08 -
1.06
1.04
1.02
1.00
.98
.96
.94
.92
,90
20
40
60
80
100
DISTANCE FROM DUCT WALL TO ROW LOCATION,
PERCENT OF DUCT WIDTH
Figure 44. Row average analysis for NOV concentration
J\
maps—field demonstrations, duct 1
-------
RATIO OF ROW AVERAGE CONCENTRATION
TO PROPORTIONAL AVERAGE CONCENTRATION
DISTANCE FROM DUCT WALL TO ROW
LOCATION, PERCENT OF DUCT WIDTH
Figure 45. Row average analysis for C02, 02> S02>
and NOX concentration maps—field
demonstrations, duct 2
89
-------
Table 10. MULTI-HOLE SAMPLING PROBE OVERNIGHT
SAMPLING RESULTS, 1975
Start: 7 PM, 2-19
End: 7:10 AM, 2-20
Number of data points:
26 (taken on the hour and ten minutes
after the hour)
GAS
co2
°2
N2
MEAN VALUE,
% OF DRY GAS
13.55
5.88
80.57
20 DEVIATION,
% OF MEAN VALUE
+17.8
+4.2
+2.8
Probe position: Duct 1, Port 5
90
-------
Table 11. C02 TRACER METHOD ANALYSIS
OF FIELD DEMONSTRATION DATA
i ft "7 n TCCT
1974 TEST
Duct 1
NOX/C02
1975 TEST
Duct 1
NOX /C02
so2/co2
Duct 2
NOX/C02
so2/co2
2o , % OF MEAN VALUE
+8.9
+23.1
+8.8
+12.1
+15.7
Explanation of tabular values:
2a = +8.9% of mean value for N0x/C02 ratio means that
there is a 95.4% probability that the ratio of the
NO reading to the C0? reading at any point in the
A t*
duct will agree with the average NOX/C02 ratio within
+8.9%.
91
-------
Table 12 showed that at a given point, concentrations were very constant
for periods on the order of half an hour. Variations between points
were clearly much larger. Overnight data, as shown in Table 13, showed
significantly greater variations. Note that the standard deviation of
23 ppm is larger than most of the differences between successive
readings. This demonstrates a "time constant" for variations that is
more on the order of hours than minutes.
8.5 SUMMARY OF RESULTS
Average concentration determinations from gas sample traverses agreed
well with nominal plant specifications. Correlation of total mass
emission measurements with plant coal usage data was excellent. This
agrees with the statement made in Section IV that better accuracy should
be expected for total mass emission than for mass emission of any single
gas. The Row Average data suggest possible measurement accuracies for
individual species of about +4%, although the one twelve hour period of
continuous sampling showed 2-3% agreement. It would appear likely that
better correlation will be obtained for long sampling intervals rather
than short ones. The rate of concentration variations at a single point
suggests that sample traverses should be performed in as short a time as
possible to avoid errors due to temporal variation. A traverse time of
two hours or less would appear reasonable for this power plant. The C0?
tracer technique showed fairly poor results, as was the case with the
ixxon data. The test result which should have the greatest implication
for future work was the continued excellent agreement between spatial and
flow proportional mean values. The average stratification level, 2o, was
t!3%, which is lower than the levels of the Exxon and TVA data, but still
too high to justify single point sampling.
92
-------
Table 12. SHORT TERM STATIONARY POINT GAS
SAMPLE DATA, 1974, DUCT 1
AVERAGE
LOAD,
MW
109
108
100
88
80
TIME
DURATION,
MINUTES
26
25
32
26
15
READINGS
7
6
8
6
5
POSITION,
ROM-
POINT
2-4
3-4
6-4
7-4
5-4
GAS
N2
°2
co2
N2
°2
co2
N,
°2
co2
N,
°2
co2
N2
°2
co2
MEAN VALUE,
% OF DRY GAS
79.67
5.63
14.67
79.77
5.61
14.63
78.92
5.29
15.80
77.67
7.27
15.05
77.43
7.09
15.43
20.
% OF MEAN
VALUE
+ .22
+ .24
+ .24
+1.06
+ .86
1.38
1.30
1.38
1.14
1.38
1.38
1.30
1-44
1.38
1.20
93
-------
Table 13. OVERNIGHT N0¥ GAS SAMPLE DATA, 1974
n
Date
10-4
10-5
Time
19:50
20:50
21:50
23:40
0:30
1:00
3:00
4:00
5:00
6:00
7:00
8:00
9:00
9:30
Load
MW
53
54
52
53
53
53
55
54
54
54
55
55
53
54
NO Concentration, PPM
J\
261
308
314
282
273
268
250
256
234
243
236
252
273
279
AVE = 266 + 23
2a =
+17.6% of mean value
94
-------
SECTION IX
DISCUSSION OF RESULTS
9.1 GENERAL
Current sampling techniques in large ducts tend to fall into two
extremes: careful manual sampling is done according to Federal Register
Method 1 (Reference 4), and often involves in excess of forty traverse
points. Continuous monitoring is more often done at a single point for
reasons of cost and simplicity. Data examined during this program have
shown stratification levels on the order of 10-15% and poor temporal
repeatability. On this basis, we have concluded that single point
sampling is generally not acceptable if good accuracy is desired.
At the other extreme, we recognize the impracticability of large
sampling arrays. Thus, our purpose has been to find an economically
feasible and technically acceptable (from the standpoint of accuracy)
middle ground, both for measurement of total volumetric flow and gaseous
emissions.
The accurate measurement of constituent gaseous emissions in process
streams is inherently more difficult than the measurement of total gaseous
emissions simply because the latter is required in order to determine
the former. Methods to continuously monitor total gaseous emissions were
developed and evaluated during the first phase of the program. Results
documented in Reference 1 showed that good accuracy could be attained
through proper use of as few as one flow sensor in large ducts with very
undeveloped flow profiles. We therefore approached the gas composition
measurement phase of the program with the attitude that the simplest
hardware system for continous monitoring would probabily involve inde-
pendent flow and composition subsystems. On this basis, we were extremely
pleased to discover the excellent correlation between spatial and flow
proportional mean concentrations. This agreement justified the separation
of flow and composition measurements on the basis of accuracy.
There is an intuitive relationship between gas stratification levels
and the complexity of hardware required to obtain a representative sample.
95
-------
One program objective was to establish a standard technique to specify
the magnitude of stratification in a usable manner. The most obvious
parameter for this purpose would be the standard deviation for a normal
distribution (Reference 3, Chapters 8-9), since this parameter is the
most commonly used in the evaluation of propagation of errors. The key
question was whether concentration data followed a normal distribution
in the sampling plane. Data from several runs showed a reasonably
normal distribution, so the validity of the approach was accepted. It is
suggested that consideration be given to adopting the "2o" definition of
stratification levels for purposes of standardization.
9.2 METHOD DEVELOPMENT AND EVALUATION
The correlation between spatial and flow proportional mean values
was the basis for method development work. The primary proposed method
is the Row Average technique developed originally for velocity measure-
ment. This technique showed reasonable results for gas sampling,
especially during the second field demonstration. The main problem
with its use is a lack of predictability. The technique definitely
works best when the row is in the direction of highest stratification.
This direction can usually be predicted for velocity stratification by
examining the local duct geometry. There is no analogous situation for
gas sampling, so preliminary survey traversing must be considered a
requirement for selection of an optimum gas sampling system. For the
data examined during the program, the best Row Average correlation was
obtained for the TVA data, since maximum stratification was in the row
direction, which was not the case for most of the other data. Recall
that this correlation was obtained even though the TVA data had higher
stratification levels than did most of the other data.
The recommended procedure for application of the row average technique
is as follows:
Perform a preliminary traverse for the gases of
interest at each desired flow condition
96
-------
• Examine data to determine direction of highest stratifi-
cation. Couple this with considerations of duct dimensions,
access, and other physical constraints to determine the
most desirable row direction
• Perform a Row Average analysis of the data to determine
the optimum probe location
t Acquire and install sampling hardware
• Perform an in-place system calibration
The row average technique is being recommended for reasons of accuracy
and hardware simplicity.
The Maiden C02 Tracer method is as simple from a hardware standpoint
as the Row Average method, but the accuracy does not appear to be as good
for the data considered: a uniformly poor correlation was obtained for
the ratios of N0x and S02 to C02 concentrations. A good correlation may
be expected for the S02/C02 ratio as long as there is little sulfur
stratification in the fuel, but such correlation was not observed. The
formation of NOV and CO- take place through quite different mechanisms,
X £
so we see no obvious reason why the concentration ratio should be constant.
9.3 HARDWARE DEVELOPMENT
Development of gas sampling hardware was not strictly within the
scope of the program. Consequently, work in this area was done at a
very low level for the sole purpose of providing a way to demonstrate
the Row Average technique in the field. The concept behind the probe
design is valid for obtaining a representative line average sample.
Particulate control was the major problem with the prototype units pro-
duced. We feel that validity of the Row Average method has been
demonstrated sufficiently to justify serious development of a multi-point
linear gas sampling probe for continuous use.
9.4 FIELD DEMONSTRATIONS
A most common problem in field testing is that of obtaining adequate
traverse data in a short amount of time. During the field demonstrations,
97
-------
a 49 point velocity traverse with a pitot-static probe took less than
an hour. A full gas sample traverse took eight hours, primarily due to
the fact that the gas chromatograph required about five minutes between
sample injections. The most common guideline for selection of the
number of sampling points is EPA Federal Register Method 1 (Reference 4).
On the basis of the data examined, we feel that the number of sample
points specified for most situations in Method 1 is overly conservative
for gas sampling, and that work should be done to establish a proper
minimum number of points for gas sample traverses.
9.5 FINAL COMMENTS
Total mass emissions can be accurately monitored on a continuous
basis using hardware and techniques evaluated during the program. We
also believe that the proposed Row Average method is acceptable for the
continuous monitoring of gas concentration. However, there is no currently
available sampling probe hardware for this application. Work should be
performed to develop sampling hardware and to properly integrate the flow
and sampling subsystems to produce complete real time continuous monitoring
systems.
98
-------
SECTION X
REFERENCES
1. "Continuous Measurement of Total Gas Flowrate from Stationary
Sources", EPA-650/2-75-020, Brooks, et al. February, 1975.
2. "Procedures for Measurement in Stratified Gases," EPA-650/2-
74-086-a and -b, Zakak, et al, September, 1974.
3. The Analysis of Physical Measurements. E. Pugh and G. Winslow;
Addison-Wesley, 1966
4. "Standards of Performance for New Stationary Sources", Environmental
Protection Agency; Federal Register, Vol. 36, No. 159, Part II,
9-17-71.
99
-------
SECTION XI
GLOSSARY
SYMBOL
A
c
d
it)
M
ff
n
N
P
R
R
T
u
TT
•
V
v
y
P
a
TT
USAGE
area
concentration
diameter
unit vectors
mass flow rate
molecular weight
unit normal vector
number
number
absolute pressure
universal gas constant
gas constant
absolute temperature
axial velocity component
total velocity vector
volumetric flow rate
velocity
mole fraction
density
standard deviation
property relative to
species i
property at standard
conditions
mean value
DIMENSIONS
m2
moles/cm
m
dimensionless
gin/sec
gm/mole
dimensionless
dimensionless
dimensionless
torr 2
gm-m
mole-sec -°K
m2
sec -°K
°K
m/sec
m/sec
3
m /sec
m/sec
dimensionless
gm/cm
100
-------
SECTION XII
APPENDIX
FLOW ANALYSIS PROGRAM
A typical analysis program run is presented. The run shown is the
analysis of S02 data taken in duct 2 during the 1975 field demonstra-
tion. Program development is presented in Section V. Notation shows
various features of the program.
101
-------
o
ro
00100
00110
00120
00130
00140
00150
00160
00170
00180
00190
00200 C
00210
00220
00230
00240
PROGRAM C ONTOUiR (INPUT, OUTPUT, TA°E5= INPUT, TA°E 7,
1 TAPE6I
REAL LX,LY,KS02
DIMENSION Xim,Yl(9l,X2<9),YZ(l),IXK9l,IYl(9l
1 ,1X2191 ,IY2f 91, XXK25) ,YY1(25)
2 ,GG1(25,251,GG2(25,25>,FX1C25),FY1C25),FF(251
3 ,ZT1(9«9) ,ZZ1(69)«Z12(9,9),ZZ2(69)
4 ,FX2(25),FY2(25»,FFY<25)
5 , XU<21> ,610(101 ,G2C(101 ,£ST(3) ,SOL(3) ,SOL*<3 ) ,SOLI < 3 )
COMMON /AU/ B1,«32,B3,GG
EQUIVALENCE (ZT1, ZZ1) ,( ZT2,ZZ2)
NAMELIST /NAM/ LX, LY , A , IM.JM ,NC,IG1 , NSTOP, IOP, FACTOR
2 ,G1MAX,G2MAX,G1RANGE ,G2*ANGE ,XDIV, IFLAG1 , IFLAG2 ,IFL AG
NAMELIST /NflMt/ ZZ1, ZZ2, XI, Y 1, X2, Y2, IX 1, IY1 ,1 X2, IY2,KS02
ZZH1)- 6790. ,5260. ,5608., 5983. ,6759. ,6<*33. ,6867. ,6790.
6184. ,4 9C 3. ,<«957.,5322. ,6 070., 5 665. ,6210. ,618-*.
62 55., 492 5., 5183., 5 763. ,6302., 6 114. ,6194., 6255.
6301. ,4 =71., 5480 .,5748. ,6*74. ,6423. ,6450., 6 301.
63 24., 53 3 7., 5496., 5 750. ,6 565., 6 10 2. ,6449. ,6324.
6473. ,5146., 5600., 5 362. ,6711. ,6583. ,6748. ,6473.
66 39., 5 240., 5 663., 60 03. ,6 748., 662 «*. ,6730. ,6639.
6790. ,5 260., 5638., 598 3. ,6 759., 6 433. ,6867. ,6790.
ZZ2( 1)= 8. 35 1,7.660, 7. 1C d, 10. 17 3, 9. 443, 1.087,7.907, 8.351,
1.106,1.428,7.019,7.830,6.96,4. 478,4.495,1.106,
6.783,7.400, 5.365,7.942,9.309,8.978,8.693,6.783,
8. **4 4, 8. 180, 6. 26 2, 7. 6 97 ,7.153,6.760,7.478,8.444,
5. 99^,8. 4 €0 ,6.649,8.1*43 ,6.769,6.022,6.625,5.994,
5.749,6 .196, 6. 92 1,3. 8 57, 7. 32 7, 6. 9 16, 6. 7 82, 5. 7^+9,
5.818,7.05,6.722,13.759,3.889,6.513,7.464,5.818,
8. 351,7.660, 7.108,10.178, 9.4*3, 8. 087,7. 90 7, 8. 351 ,
KS02=. 05725
RAW DATA
INPUT
-------
o
CO
C 03UO
0 Q3<*2
00350
00352
00360
00362
OC370
00372
00360
00390
00409
OG410
OG<*20
00430
00440
00450
00460
00470
00460
00490
00500
00510
0 0520
OC53C
OC540
OC550
00560
00570
00560
0 0590
00592
00600 (
00610
00620
00623
DAT A (XI (11,1 = 1, 8) / 11.143,. 857, 2. 571, 4. 286,6. ,
1 7.714,9.429,11.143/
OATA(Yl(I) ,1=1,8) /18.571,1.429,4.286,7.143,10.,12.857,
1 15.714»18.5717
OATA(X2(I) ,1 = 1,8) 711.143,.857,2.571,4.286,6. ,
1 7.714,9.429,11.1437
DATMY2 (I ),! = !, 81 718 .571,1. <»29,4. 286 ,7 .143,10. ,12. 857 ,
1 15.714,18.5717
DATA(IX1(I),I=1,7) 78,0,1,4,1,-1,87
DATAdYK ]),I=1,5) 78,1,1,4,17
OATA(IX2(I),I=1,8) 73,0,1,4,1,-1,87
OATA(IY2(I),I=1,5) 78,1,1,4,17
A=240.
XOIV=12.
MAXITsia
EPSI=l.E-3
G2MAX=1JO.
G1RANGE=1000.
G2RANG£=100.
IFLAG2=0
IFLAG=C
FACTOR*. 05
IOP=fl
REMIMO 7
READ(7tNAMl)
00 « I»l,6<.
-------
H0625
00630
006<»0
00650
00660
00670
00680
00690
00700
00710
00720
00730
OG7<*0
00750
00760
OC770
00780
00790
OG800
00810
00820
00830
008/2
60 XX1(1I = 0.
00 200 1=2, IM
XX1(D = XX1 (I-1)*HI
200 CONTINUE
YY1(1)=0.
00 213 1=2, J*
YY1(I»=YY1CI-1)*HJ
90 FORMAT UX, 15, E 12. <»)
210 CONTINUE
Q
IFdFLAGl .E3. 11 GO TO 33C
00 323 J=1,JM
Y=YY1(J)
00 30iJ I=ltl*
X=XX1 (I)
CALL 8VICCX,Y,Z1,X1,Y1,ZT1,IX1,IY1)
CALL 8VIC
Z12=Z1*Z2
-------
o
en
OC970
0 0980
00990
01000
01010
01020
01030
010
IO = XDIV+1
HXM=LX/XOIV
XM (1) = C .
00 3<»0 I=?,I3
XM(I) sXMI-1
3*»0 CONTINUE
IFCIOP .£Q.
GO TO 360
-------
o
Ol
01260
01270
01280
01290
01300
01302 C
01310
01320
01330
01350
01360
01370
01380
01390
01400
01410
01420
01430
01440
01450
01460
01470
01480
0 14 80
0150G
01502 C
0151o
01520
01530
01540
01550
01560
01570
01580
01590
1
1
350
360
370
380
500
460
1320
470
1030
00 350 I=ltNC
G1C(I)=G1MAX-1*GHANGE/NC
G2C(I>=G2MAX-I*G2SANG£/NC
CONTINUE
GO TO 380
OIF is FACTO R*GB 12
OIF2-FACTCR*G82
GlC(l)=GBl2-(NC-l)/2
G2C(i) = GB2-(NC-l) /2
DO 370 1*2, NC
*OIF1
DIF2
CONTINUE
JJ=JHO
DO -530 IX = 1,ID
X=XM(IX>
DO 530 IY=1,JM
Y=YY1(IY)
CALL BVIC(X,Y,Z1 «X1« Yl , ZT1,IX1 , 1 YD
IFCZ1 .LT. 0) Zl=0.
GG1(IY,IX)=Z1
CALL BVIC(X,Y,Z2,X2, Y2»ZT2t 1X2,1 YZ)
IFCZ2 .LT. 01 Z2=0.
GG2(IY,IX)=Z2
CONTINUE
DO 513 IG=IG1,2
GO TO (**60,U70) IG
WRITE<6,102C>
FORMA T(/6X,*Y»,lflX,*SP*,9X,*Y*,10X,*SP»,
1 qx,*Y*,10X,*SP*/)
GO TO 430
WRITc (6tl030)
FORHAT(/6X,*Y*,9X,*V£L*,9X,*Y»,9X,*VEL*,
9X,»Y*,9X,*YEL*/)
-------
01600
01610
01620
01630
016<*0
0 1650
01660
01670
01680
C1690
01700
0 1710
01720
01730
017VO
01750
01760
01770
01780
01790
018 00
01810
01820
01830
01840
01859
01860
01870
01880
01890
01900
01910
01920
0 1930
01940
480
1340
1050
485
490
510
C
277
52C
540
56 G
610
620
no 490 IX = 1,ID
X=XM(IX)
WRITE(6»104Q) X
FORMAT(2X,*X=*,F5.2)
IF(IG ,N£.l) GO TO 485
WPIT£(6,105u) (YYKIY) ,GG1(IY,IXJ ,IY=1,JM)
FORMAT(2X,3(tl0.2,El2.4M
GO TO 4<=0
WRITE(6,lii5C> (YYKIYI ,GG2(IY,IX) »IY=1,JM)
CONTINUE
CONTINUE
00 800 IG=IG1,2
WRITE(6,277)
FORMAT(///)
00 300 1=1,NC
IF(IG .NE. 1) GO TO 520
GG=G1C(I)
GO TO 540
GG=G2C( I)
00 7-+0 J = 1,IO
X=XM(J)
00 610 IJ = 1,JH
Y=YY1(IJ)
IFCIG .NE. 1) GO TO 560
FX1(IJ»=GG1(IJ,J)
GO TO 610
FX1(IJ)=GG2(IJ»J)
CONTINUE
00 720 K=1,JJ
YRIGHT^YYl
IF(GG .LF.
IF(GG .GE.
FXKK)
FXl(K)
.AND.
.AND.
GS
GG
.G£.
.LE.
FXUK + 1
GO
GO
TO
TO
71C
710
-------
01950
0 1960
01970
01980
0 1990
02000
02010
02020
02030
02040
02050
02060
02070
02080
02090
02100
02110
02120
02130
02140
02150
02160
02200
02210
02220
02230
02240
02250
02260
02270
02280
02290
02300
02310
02320
G2330
710
715
1010
1000
720
975
7*0
93C
800
C
2
3
4
5
GO TO 720
ZD1=FX1(K+1)-FX1CK)
YD2=YY1(K+1)-YY1(K)
YY= (G-i-FXKK) I *Y02/ZD i+YYKK)
IF(IG .EQ. 2) GO TO 715
WRITt(6tlOOO) IGtI,K,GG,X,YY
GO TO 720
WRITE<6,1010) IGtltKt GGtXtYY
FORMAT(2X,*IG=*,I2t2X,*I-*,I3,2X,*K=*,I2,2X,
*V£L=*»FS,l,2Xt*X=*,F4.i,2X,*Y=*,F5.2)
FORMAT<2X,*IG=*tI2t2X,*I=*tI3,2X,*K=*,I2t2X,
CONTINUE
WRITE (6,975) K,X
FORMAT(2X,*K=*tI2,2X,*X=*,F5.1)
CONTINUE
FORMAT(2X, *J=*,I3 ,2X, *X = * ,F5 . 1)
CONTINUE
GO TO 10
END
SUBROUTINE SIMP1{YtNtOtLtL,M,ANS)
DIMENSION Y(ll
SUM1=0
00 2 1=1,L
11=2*1
SUM1 = SUMH-Y(I1)
SUM2=3
IF(M) 5,5,3
00 4 J=1,M
SUM2=SUM2*-Y(I2)
ANS = .333333333333*DEL*(Y(l»*Y(N)*i».*SUMl+2.*SUM2)
RETURN
END
-------
JNAM
LX
LY
A
IM
JM
NC
IG1
NSTOP
I OP
FACTOR
S1MAX
&2MAX
G1RANGE
G2RANGE
XDIV
I FLAG 1
IFLAG2
I FLAG
$£NC
=
s
~
s
=
s
3
—
-
S
=
=
s
-
-
s
2
=
1.
2.
2.
25
21
11
1,
Ot
Ct
5.
2.
1.
1.
1.
1.
Ot
0*
0«
2t«-
OE +
^fE*
t
t
t
OE-
3E*
0£ +
OE +
OE*
2E*
Git
Olt
02,
02t
03,
02 1
03 t
02t
Olt
CONSTAJ\|TS
-------
J=
J=
J=
J=
J=
J=
J=
J=
J=
J=
J=
J=
J=
J=
J=
1
2
3
5
6
7
8
9
10
11
12
13
15
16
17
18
19
20
21
INTG2X=
INTG2X =
INTG2X =
INTG2X=
INTG2X=
INTG2X=
INTG2<=
INTG2X=
INTG2X=
INTG2X=
INTG2X=
INTG2X=
INTG2X =
INTG2X=
INTG2X =
INTG2X=
INTG2X=
INTG2X=
INTG2X=
INTG2X =
INTG2X=
VELOCITY
3.792^E«-«31
8.Q8<*9£«-01
R-C'J PROPORTIONAL AVE.
7.8115E+31
8.4794E+01
9.6083£f 01
1.0038E«-02
9.523*»E+01
3.0006E+01
8.1941E+01
3.<*051E+D1
8.3560E«-i)l
6.9iH8E«-01
6.C551E*31
INTG12<=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X*
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
INTG12X=
2.
2.6816E-I-C'*
51*1 7 E*Q<»
SPATIAL /.V£.
2.
INTG1X =
INTG1X:
IMTG1X=
INTG1X =
INTG1X:
INTG1X-
INTG1X-
IMTG1X:
INTG1X:
INTG1X=
INTG1X =
INTG1X:
INTG1X:
INTG1X:
2.9395E+02
3.0D92E+02
3.0836E*02
3.2386E*C2
3.332QE+02
3.<*683E*-02
3.6295E*02
3.7308E+02
3.7C99E*C2
3.6301E+02
3.5939E*02
3.7073E*C2
3.7386E+-02
3.73*»2E*C2
3.7D53E+02
3.6518E*-02
3.5738£f 02
ROW
AVERAGES
A2= 1.70218'fE
G2BAR= 7.
A12= 5.876611E+Q5
G1BAR=
VALUES
-------
SP
SP
X= 0.00
c.
3.00E+00
6. ODE* 00
9.00E+OQ
1. 20E+Q1
1.50E+01
X= 1.00
0.
3.09E+00
6.00E + 00
1.50E«-01
X= 2.00
0.
3.0QE+30
6.00E+OD
9.00E«-00
1.50E+J1
1.80E + 31
X= 3.00
0.
3.00E + QG
6. CGE + OG
9.COE+00
i.20F+3i
1.50E + G1
l.GOE+CO
3.
-------
8.9230E+QC
Y VEL
0- 0.
3.OOE+00
6.0QE+00
9.0QE+00
1.2QE+01
1.50E+01
1.80E+01 o.
X= 1.00
0. 6.
3.00E+00 5.1307E+00
6.00E+CG
9.JOE«-00
1. 20E+-01
- 1.50E«-01
w 1.8QE4-01
X= 2.00
0.
3. OOE + OG
b.OOEfOO
9. COE<-OQ
1.20E+01
7.7122E+00
5.67JOE*OC
2.6572E*00
7.1645£fOO
6.9801E+00
8.3356E+-OG
1.80E4-01
X= 3.00
0.
3.00E+00
6.00E4-00
9.00E4-00
8.1234E*QO
6.0272E+00
1.
1.
1
1
OOE+
30 E*
6fl£*
90E +
01
01
Cl
01
VEL
0.
8.17&3E+OQ
4.Q990E+00
0.
1.98a5£-Ql
a.
1.1152E*00
VSL
7.GOE+00
1.00E+C1
1.30£*01
7.8397E*00
7.3279E*00
5.0&24E*00
9.5323E-01
1.00E+-00 6.27Q5E*00
1.00 Ef01
1.30Ef01
1.60E+01
1.90E+-01
7.8532E*00
9.
7.7398E+00
1.50E*01
1.11.3flc*01 1.0fl£*QO
7.00E+00
1.00E*ai
1.60E*01
1.90E+31
6.7557E+00
8.7928E+00
8.9527E+00
8.8607E*00
7.9Q10£«-aC
8.7831£*GO
5.4511E+JO
7.7896E*00
9.0462E+00
8.8556E*00
8.6919E+00
7. 306**£* 00
2.00E*-uC
5.00£*00
1.7QE+U1
2.0GE*01
5.00£»OC
6.0GE+GO
1.70E+01
2.COE*OC
5.GO£*OG
8.0G£*00
1.1GE+01
1.<»OE4-01
1.70£*G1
2.COE+01
2.00£*CC
5.DCE+00
1.10E+01
1.70£*01
2.0CF+01
G.
0.
0.
7.3M5E + 00
7.8660E*00
6.59J8E-»-00
5.0880E*OC
3.9617£fOQ
0.
5.7031E*00
6.0519E*00
8.866CE*00
8.1660E«-OG
7.0200E*00
3.2225E*00
7.C61GE+-00
5.7155E*00
8.3751E*00
9.0773E*00
8.8673E+00
8.3629£*00
6.5789E4-OC
EXPANDED
ilATRIX
VELOCITY
(CONTINUES)
-------
IG= 1 1= 6 K = 15 SP= 345.2 X= 0.0 Y=14.73
IG = 1 1= 6 K=19 SP = 345.2 X= 0.0 Y=18.95
K=20 X= Q.Q
IG= 11= 6 K= 9 SP= 345.2 X = 1.0 Y = 8.67
IG= 1 1= 6 K=li SP= 345.2 X = 1.0 Y=1D.54
IG= 1 I- 6 K=15 SP= 345.2 X= 1.0 Y=14.71
IG= 1 1= 6 K=20 SP= 345.2 X= 1.0 Y=19.83
K=20 X= 1.0
IG= 11= 6 K= 9 SP= 345.2 X= 2.0 Y = 3.43
IG= 1 1= 6 K=13 3P= 345.2 X= 2.0 Y=12.28
IG= 1 1= 6 K=14 SP= 345.2 X= 2.0 Y=13.95
K=20 X= 2.0
IG= 11= 6 K= 9 SP= 345.2 X= 3.0 Y= 8.33
K=20 X= 3.0
IG= 11= 6 K= 9 3P= 345.2 X= 4.0 Y= 8.26
K=20 X= *.0
IG= 11= 6 K= 9 SP= 345.2 X= 5.0 Y= 8.19
K=20 X= 5.0
IG= 1 1= 6 K= 9 SP= 345.2 X= 6.0 Y= 3.10
K=20 X= 6.0
IG= 11= 6 K= 8 SP= 345.2 X= 7.0 Y= 7.91
K=20 X= 7.0
IG= 1 I* 6 KS 8 SP= 345.2 X= 8.0 Y= 7.64
K=20 X= 8.0
IG= 11= 6 <= 6 SP= 345.2 X= 9.0 Ys 7.31
K=20 X= 9.0
IG= 1 1= 6 K= 8 SP= 345.2 X=10.0 Y= 7.17
K=20 X= 1C.C
IG= 11= 6 K= 8 SP= 345.2 X=11.0 Y= 7.26
K=20 X= 11.0
IG= 1 I* 6 K= 8 SP= 345.2 X=12.0 Y= 7.51
K=20 X= 12.0
LOCUS OF MEAN CONCENTRATION
PROFILES
(CONTINUES FOR OTHER
CONCENTRATIONS)
-------
IG= 2 1 =
IG= 2 1 =
K=20 X =
IG= 2 1=
IG= 2
K=20
IG= 2
IG= 2
IG= 2
K=20
IG= 2
IG= 2
IG= 2
K=20
IG= 2
IG= 2
r=
X =
1 =
1 =
1=
x=
1=
1=
1=
6 <= 4 VEL=
6 K= 8 V£L=
0.0
6
6
1.0
6
6
6
2.C
6
6
6
K= 5
K = ll
K= 1
K = 7
<=17
K= 2
<= 7
K=2C
7.1
7.1
VEL= 7.1
\/EL =
VEL =
VEL =
V£L =
VEL =
7
7
7
7
.1
.1
.1
.1
7.1
VtL= 7.1
VEL =
7.1
X= 0.0
x= a.o
Y= 3.67
Y= 7.68
X= 1.0 Y=
-------
TECHNICAL REPORT DATA
(Tlrosc read luslructiunf on llic reverse before completing)
'• REPORTNO.
EPA-600/2-75-012
HE:
' TITLE AND SUBTITLE
Continuous Measurement of Gas Composition from
Stationary Sources
5. REPORT DATE
July 1975
3. RECIPIENT'S ACCESSION NO.
6. PERFORMING ORGANIZATION CODE
F BrookS) c A Flegal, L.N. Harnett,
M.A. Kolpin, D.J. Luciani, and R. L. Williams
8. PERFORMING ORGANIZATION REPORT NO
PERFORMING ORQANIZATION NAME AND ADDRESS
PRW Systems Group
One Space Park
Redondo Beach, CA 90278
10. PROGRAM ELEMENT NO.
1AB013; ROAP 21ACX-132
11. CONTRACT/GRANT NO.
68-02-0636
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 7/74 - 3/75
4. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
Hi. ABSTRACT " ~
The program objective was to develop and evaluate methods for the contin-
uous measurement of gaseous emissions from stationary sources, specifically in large
or complex ducts where total flow processing techniques are not practical. This
report is concerned with the measurement of mean gas concentrations in rectangular
ducts. Work consisted of a review of related programs, development of a computer
program to assess stratification levels and evaluate sampling techniques, formulation
and evaluation of point sampling methods for continuous monitoring, development of a
multi-port continuous gas sampling probe, and field demonstration of hardware and
techniques. Results showed that emissions can be accurately monitored using as few
as one flow sensor and one sampling probe, even in the presence of significant vel-
ocity and compositional stratification, although stratification levels were too high
for single point samples to be acceptable. It was shown for all data examined that
good accuracy can be attained by taking a spatial concentration average -- flow pro-
Portional sampling is not required. The field demonstration verified the acceptabil-
" of the proposed methodology.
17.
--—
a.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Pollution
Gases
Chemical Analysis
Continuous Sampling
Measurement
Stratification
b.IDENTIFIERS/OPEN ENDED TERMS
Air Pollution Control
Stationary Sources
c. COSATI Field/Group
13B
21B
07D
14 B
IBUTION STATEMENT
Uni
19. SECURITY CLASS (This Report)
Unclassified
125
imited
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
F°rm
2220-1 (9-73)
115
------- |