Sei sulla pagina 1di 15

Comparative Study of Flare Dispersion

Modeling Methodologies

Extended Abstract No 2012-A-298-AWMA

Prepared By:

Whitney L Boger Senior Consultant


Arun Kanchan Principal Consultant, International Services

TRINITY CONSULTANTS
12770 Merit Drive
Suite 900
Dallas, TX 75251
(972) 661 - 8100
trinityconsultants.com

June 19, 2012

Environmental solutions delivered uncommonly well


ABSTRACT
A comparative case study of nine methodologies for representing flare stack parameters
for use in air quality dispersion modeling was conducted. As part of this case study,
modeling was also conducted to examine the effects of using these methodologies. The
methodologies reviewed include flare modeling guidelines from the following:

1. Alberta Department of Environment and Water (Alberta)1


2. United States Environmental Protection Agency (U.S. EPA)2
3. Texas Commission on Environmental Quality (TCEQ)3
4. Oklahoma Department of Environmental Quality (Oklahoma DEQ)4
5. Ohio Environmental Protection Agency (Ohio EPA)5
6. Alabama Department of Environmental Management (ADEM)6
7. New Jersey Department of Environmental Protection (NJDEP)6
8. Iowa Department of Natural Resources (Iowa DNR)7
9. The approach noted in the Practical Guide to Atmospheric Dispersion Modeling by
Bruce Turner (Turner)8

The paper presents comparisons of ground-level modeled concentrations (GLCs) based on


the different modeling methodologies, as well as sensitivity of various related parameters.
The study should be used to assess the validity and accuracy of the associated methods and
lend to discussions regarding necessary modifications. The case study was based on an
upstream energy facility in the Middle East, which has some of the largest flares in the
world with considerable regulatory emphasis and compliance onus. The scope of the
paper and case study is limited by the equations provided in the various guidance
documents. Background documentation on the equations used in each methodology was
out of the scope of this present case study. This should, however, be a possible topic of
interest for future research.

2
INTRODUCTION

Flares can be represented in air quality dispersion modeling studies as follows: 1) as a


standard point source with no modified stack parameters (actual parameters for stack
diameter and stack height are used in this instance); 2) as a point source with modified
stack parameters (actual stack parameters are modified to better represent conditions
while flaring); or 3) as a pseudo-point source (flare option in the EPA beta version of
AERMOD).7 For this case study, only the point source with modified stack parameters
method was applied.

A case study was performed to assess the differences in modeled ground-level


concentrations based on various available flare modeling methodologies in the United
States. In order to perform the case study, the nine methodologies were reviewed and the
stack parameters calculated accordingly. The U.S. EPA preferred air dispersion model
AERMOD v09292 was utilized as the selected model for the air dispersion modeling runs. 1
The same heat release rate and emission rates were used throughout each of the modeling
runs. In addition, the stack exit velocity is assumed to be 20 meters per second (m/s) and
the stack exit temperature is assumed to be 1,273 degrees Kelvin (K), as indicated in each
set of modeling guidelines. The pollutant, SO2, was modeled for the case study for both 1-
hour and 24-hour averaging periods. In addition, modeling runs were performed with two
years of actual meteorological data that represent the worst case meteorological years over
a ten-year period for the study area. Finally, the modeling runs were completed for
different assumptions of operating hours for the flare for the 24-hour averaging period - for
example, air dispersion modeling runs were run as if the flare were operated at 1 hour per
day (10 am 11 am), 8 hours per day (6 am 2 pm), 12 hours per day (4 am 4 pm), 16
hours per day (2 am 6 pm), and 24 hours per day. These different operating scenarios are
identified as BRZFL1 for the one hour operating hour, BRZFL8 for 8 hours of operation, and
so on. No deposition or building downwash was included in the modeling runs.

Each methodology explored has different equations to account for the effective stack
diameter. Some of the methodologies also have equations to calculate an effective stack
height and buoyancy flux. For those methodologies where an effective stack height is not
calculated, the actual physical stack height was used. In addition, the effect of radiation
heat loss of the plume was also assessed. In the United States, it is typical to assume that
the steam-assisted flares will have a radiation heat loss of approximately 55%, based on
SCREEN3 defaults and Leahey and Davies observations.6, 11 This value is oftentimes the
most conservative estimate found for radiation heat loss within literature review.11 The
case study used in this analysis was performed for operations in the Middle East, where
steam assist is not commonly used. A lower radiation heat loss of 40%, based on
engineering judgment, was assumed for this case study. Some of the methodologies allow
for explicit adjustment of the radiation heat loss parameter, while others assume the
SCREEN3 default value of 55%. Therefore, for the methodologies that do not allow for
explicit adjustment of the radiation heat loss (i.e. Oklahoma DEQ, Ohio EPA, ADEM, and

1
AERMOD v09292 was the latest version at the time of the air dispersion modeling analysis was conducted.

3
NJDEP), a ratio technique is applied to the heat release rate in the associated equations to
account for the reduced radiation heat loss that should be used for this particular case
study. Details of each of the methodologies and associated equations are provided in the
Methodology Review section provided below.

At the completion of the modeling runs, the high second high (H2H) modeling result
concentration was determined and compared with the other methodologies. The resulting
concentrations for all modeling runs were normalized to the EPA methodology, where the
EPA methodology modeling results are represented with the value of 1. Therefore, a
methodology resulting in a lower concentration than the EPA method is represented by a
value less than 1, and a methodology resulting in a concentration higher than that resulting
from the use of the EPA method is represented by a value greater than 1.

Methodology Review

The following section compares the different equations that were used for each of the
methodologies. Only three of the nine methodologies account for buoyancy flux, as
presented in Table 1. It is within the Albert Environment and Turner buoyancy flux
equations in which the radiation heat loss can be explicitly altered. The buoyancy equation
for the EPA method is taken from the Briggs equation.9 Please note some of the notations
have been altered to make the equations more comparable.

Table 1. Buoyancy Flux Equations

Methodology Bouyancy Flux Equations


Alberta Environment1 (Eq. 1)
radiation heat loss
Fb = 0.000037 QH 1
100


EPA2 (Eq. 2) 5
Fb = 1.66 10 H t
Ohio EPA5 (Eq. 3) Fb = 1.162 Q
Iowa DNR7 (Eq. 4) For the flare,
F = ( g H r ) /(TC p ) = 2.59 (0.001) H r / T
And for the stack release,
F = g Vs (rs ) 2 (Ts T ) / T
gQH (1 radiation heat loss
Turner8 (Eq. 5)
)
Fb = 100
C pepa a T

4
Where,
Fb = buoyancy flux (m4/s3)
QH = source heat release (cal/s)
Q = heat release (MMBtu/hr) [conversion factor of 1 MMBtu/hr = 69998.82249
cal/s]
Ht = total heat release rate (cal/s)
Hr = net heat release rate (J/s) = (1-F)H (for a single gas)
H = total heat (sensible + radiated) release rate (J/s)
g = acceleration due to gravity = =9.81 (m/s2)
Vs = exit velocity (m/s)
rs = stack inner radius (m)
Ts = stack exit temperature (K)
= density of air = 1.2 (kg/m3)
T = air temperature (K)
Cp = specific heat of dry air constant = 1004 (J/(kg K)
Cpepa = specific heat of air at a constant pressure (cal/g-K) [assumed to be 0.24
cal/g-K]
a = air density (g/m3) [assumed to be 1204 g/m3]

Only some of the nine methodologies allow for the calculation of an effective stack height.
Those that do not account for an effective stack height use the actual stack height of the
flare. The methodologies that allow for effective stack height calculations are presented in
Table 2 below. The EPA and Iowa DNR methodology allow for the adjustment of the
radiation heat loss within the effective stack height equation.

Table 2. Effective Stack Height Equations

Methodology Effective Stack Height Equations


Alberta Environment1 (Eq. 6) H effective = H actual + 0.00456 QH
0.478

Oklahoma DEQ4 (Eq. 7) H effective = H actual + 0.00128 Qc


0.478

Ohio EPA5 (Eq. 8) H effective = H actual + 0.944 Q 0.478


NJDEP6 (Eq. 9) H effective = H actual + 0.944 Q 0.478
Iowa DNR7 (Eq. 10) H effective = H actual + 0.00456 ( H r / 4.1868) 0.478

Where,
Heffective = effective stack height (m)
Hactual = actual stack height (m)
QH = source heat release (cal/s)
Qc = flared gas heat release (Btu/hr) [conversion factor of 1cal/s =
14.28595Btu/hr]
Q = heat release in MMBtu/hr [conversion factor of 1 MMBtu/hr = 69998.82249
cal/s]
Hr = net heat release rate (J/s) = (1-F)H (for a single gas)

5
Each of the nine methodologies has an equation to calculate the effective stack diameter, as
presented in Table 3.

Table 3. Effective Stack Diameter Equations

Methodology Diameter Equations


Alberta Environment1 (Eq. 11) 4 FbTo
d=
gvT
EPA2 (Eq. 12) d = 9.88 10 4 Qh
0.5

Where,

radiation heat loss
Q H = H 1
100


TCEQ3 (Eq. 13)
d = 10 6 Qh
Where,
(
Qh = q 1 0.048 MW )
Oklahoma DEQ4 (Eq. 14) d = 1.752 10 4
Qc
Ohio EPA5 (Eq. 15) d = 0.1755 Q 0.5
ADEM6 (Eq. 16) d = 0.000988 (0.45 Qh ) 0.5
NJDEP6 (Eq. 17) d = 0.1755 (Q ) 0.5
Iowa DNR7 (Eq. 18) d = 0.1066 [(To / T (To T )) ( H r / Vs )]0.5
Bruce Turner8 (Eq. 19) 4 FbTo
d=
gvT

Where,
d = diameter (m)
g = acceleration of gravity (m/s2)
To = stack gas temperature (K) [assumed to be 1273K]
T = stack gas temperature (K) air temperature (K)
v = stack gas exit velocity (m/s) [assumed to be 20 m/s]
Fb = buoyancy flux (m4/s3)
Qh = source heat release (cal/s)
Qc = flared gas heat release (Btu/hr) [conversion factor of 1cal/s =
14.28595Btu/hr]
Q = heat release (MMBtu/hr)
T = air temperature (K)
Hr = net heat release rate (J/s) = (1-F)H (for a single gas)
F = radiative loss factor (%)

6
Vs = exit velocity (m/s)
H = total heat release rate (cal/s)
q = gross heat release (cal/s)
MW = weighted average molecular weight of the compound being flared

The Alberta method has a calculation to account for the stack exit velocity, as shown in
Equation 20.1 For all other methods, the stack exit velocity was set to 20 m/s.

4Q T
v= ...............(Eq. 20)
d 2 Ts

Where,
v = stack gas exit velocity (m/s)
Q = total gas flow rate at reference temperature (m3/s)
d = diameter (m)
T = ambient temperature (K) [assumed to be 300K]
Ts = standard temperature (K) [assumed to be 293K]

As noted previously, some of the methodologies allow for explicit alteration of the radiation
heat loss factor (Alberta, EPA, Iowa DNR, and Bruce Turner). In the original TCEQ method
development, the net heat released from the flare is calculated based on the mean
molecular weight of the stream. This results in 25 to 50 % of the gross heat released lost to
the atmosphere and 50 to 75% of the heat released from the flare to contribute to plume
rise.10 Based on the TCEQ interoffice memorandum, the molecular weight of the gas flared
determines the radiation heat loss. For this particular case study, the radiation heat loss
was calculated to be 29%; therefore, the heat released from the flare contributing to plume
rise is equal to 71%. Since this methodology differs significantly from the others and the
basis of the molecular weight is unclear in this determination, the ratio technique is not
applied. If it were applied, the modeling results would become even more conservative
than currently presented in the results, since one would be decreasing the amount of heat
release contributing to the plume rise. For the methodologies that do not allow for explicit
adjustment of the radiation heat loss (i.e. Oklahoma DEQ, Ohio EPA, ADEM, and NJDEP), a
ratio technique is applied to the heat release rate in the associated equations since the
calculations are either based on a radiation heat loss of 55% or are considered consistent
with the SCREEN3 methodology, which uses a radiation heat loss of 55%. In order to adjust
the radiation heat loss to 40%, a ratio of (0.60/0.45) is multiplied by the flared gas heat
release rate in the effective stack diameter and effective stack height equations.

Results and Discussion

After each of the runs was completed and the results of the high second high compiled, the
concentrations were normalized in accordance with the EPA method as 1 since this is the
federal standard in the United States. The results of the case study are presented in Tables
1 through 4.

7
Table 1. Comparison of Calculated Stack Parameters

Effective/Actual
Buoyancy Flux Diameter
Methodology Stack Height
(m4/s3) (m)
(m)

Alberta with EPA


1873 7.07 173
Constants

Alberta 1873 7.05 173

Turner 1823 6.97 145

EPA 1401 7.03 145

TCEQ -- 7.72 145

Oklahoma DEQ -- 6.09 173

Oklahoma DEQ
-- 7.03 177
(with ratio
technique)
Ohio EPA 1401 6.09 173

Ohio EPA (with


1867 7.04 177
ratio technique)
Iowa DNR -- 6.98 167

NJDEP -- 6.09 173

NJDEP (with ratio


-- 7.04 178
technique)
ADEM -- 6.09 145

ADEM (with ratio


-- 7.03 145
technique)

Tables 2 through 4 show the relative differences of the modeling methodology results
compared with the EPA method. Some of the modeling methodologies result in
concentrations higher than those predicted by the EPA method and some result in
concentrations lower than those predicted by the EPA method.

8
Table 2. Comparison of 1-Hour Averaging Period Results of Comparative Test Results

Modeled Modeled
Methodology Year Year
2005 2008

Alberta with EPA Constants 0.956 0.838


Alberta 0.957 0.838
Turner 1.006 1.008
EPA 1 1
TCEQ 0.942 0.914
Oklahoma DEQ 1.021 0.960

Oklahoma DEQ (with ratio


0.959 0.821
technique)
Ohio EPA 1.021 0.960

Ohio EPA (with ratio


0.958 0.820
technique)
Iowa DNR 0.962 0.881
NJDEP 1.021 0.960

NJDEP (with ratio technique) 0.958 0.814


ADEM 1.180 1.134

ADEM (with ratio technique) 1 1

Table 3. Comparison of 2005 24-Hour Averaging Period Results of Comparative Test


Results

Methodology ALL1 ALL8 ALL12 ALL16 ALL24


Alberta with EPA Constants 0.862 0.883 0.859 0.851 0.851
Alberta 0.862 0.884 0.859 0.851 0.851
Turner 1.011 1.009 1.008 1.008 1.008
EPA 1 1 1 1 1
TCEQ 0.888 0.922 0.916 0.915 0.915
Oklahoma DEQ 1.038 1.026 0.978 0.970 0.970

Oklahoma DEQ (with ratio 0.853 0.873 0.845 0.836 0.836


technique)
Ohio EPA 1.038 1.026 0.978 0.970 0.970

Ohio EPA (with ratio 0.852 0.871 0.844 0.835 0.835


technique)
Iowa DNR 0.898 0.919 0.897 0.891 0.891
NJDEP 1.038 1.026 0.978 0.970 0.970

0.848 0.868 0.839 0.830 0.830


NJDEP (with ratio technique)
ADEM 1.187 1.153 1.135 1.135 1.136

ADEM (with ratio technique) 1 1 1 1 1

9
Table 4. Comparison of 2008 24-Hour Averaging Period Results of Comparative Test
Results

Methodology ALL1 ALL8 ALL12 ALL16 ALL24


Alberta with EPA Constants 0.820 0.847 0.827 0.826 0.824
Alberta 0.820 0.848 0.828 0.826 0.825
Turner 1.012 1.012 1.007 1.007 1.007
EPA 1 1 1 1 1
TCEQ 0.877 0.922 0.925 0.923 0.923
Oklahoma DEQ 0.989 1.028 0.954 0.943 0.942
0.810 0.833 0.809 0.809 0.808
Oklahoma DEQ (with ratio
technique)
Ohio EPA 0.989 1.028 0.954 0.943 0.942

Ohio EPA (with ratio 0.809 0.832 0.808 0.808 0.807


technique)
Iowa DNR 0.857 0.892 0.871 0.868 0.866
NJDEP 0.989 1.028 0.954 0.943 0.942

0.805 0.827 0.802 0.803 0.802


NJDEP (with ratio technique)
ADEM 1.201 1.189 1.122 1.123 1.125

1 1 1 1 1
ADEM (with ratio technique)

The methodologies were also ranked according to the high second high maximum
modeling concentration results. Due to the confidentiality of the results, a point system
was applied for the ranking in order to determine the overall ranking of the methodologies
across the different modeling scenarios. If the methodology resulted in the lowest
maximum concentration from the other methodologies, it was given a value of 1 for that
particular run. The methodology with the highest maximum concentration from the other
methodologies was given the highest value. This was done for each averaging period and
meteorological year. The generalized results over all modeling scenarios and for both
meteorological years for this case study are presented in Table 5.

10
Table 5. Modeled Concentrations Results Ranking

Methodology Ranking Value


ADEM 132
Turner 115
Oklahoma DEQ 106
Ohio EPA 106
NJDEP 106
EPA 103
ADEM with Ratio Technique 103
Iowa DNR 75
TCEQ 67
Alberta 60
Alberta with EPA Constants 48
Oklahoma DEQ with Ratio Technique 41
Ohio EPA with Ratio Technique 29
NJDEP with Ratio Technique 16

As shown in all the results tables above, the methodology that produced the highest
maximum modeling concentrations results was the ADEM method, followed closely behind
by the Turner approach. The methodology producing the lowest maximum modeling
concentration results was the NJDEP methodology when the ratio technique was applied to
the effective stack diameter and effective stack height equations to account for the change
in radiation heat loss. The Ohio EPA and Oklahoma DEQ methodologies also provided very
similar results when the ratio technique was applied. The major differences between these
methodologies are the calculated effective stack diameter and effective stack height. There
is a difference of 0.95 meters in stack diameter between the NJDEP calculated effective
stack diameter and the ADEM effective stack diameter (although the largest effective stack
diameter is calculated based on the TCEQ method). In addition, the effective stack height
using the NJDEP methodology with the ratio technique applied was the tallest, at 177
meters. All other parameters remained the same between the methodologies. It is not
expected that various meteorological conditions will change the expected results since two
years of meteorological data were used for the case study. Additional air dispersion
modeling runs can be performed as part of an extended study.

As can be seen from the results in Tables 2 through 4, the Ohio EPA, Oklahoma DEQ, and
NJDEP methodologies produce the same results. Although the equations are unique, the
differences appear to be attributable to insignificant variations in unit conversions and
constants. In fact, the only difference between the NJDEP methodology and the Ohio EPA
methodology is that there is an equation to account for buoyancy flux within the Ohio EPA
methodology. The results are only slightly different from one another when the ratio
technique is applied.

When the ratio technique is applied to the ADEM methodology, the results are the same as
those derived from the EPA method. This is evidenced in the equations because the only
difference between the two equivalent diameter equations in these methodologies is that
the ADEM method multiplies the total heat release rate by 0.45. This 0.45 represents the

11
55% heat radiation loss, with 45% contributing to plume rise. The EPA method accounts
for the radiation heat loss in the source heat release equation. The models do not even
need to be run to see these similarities, because the stack parameters are identical.
The TCEQ method varies the greatest from all other methods since the net heat released
from the flare is based on the molecular weight of the stream being flared. The gross heat
released from the flare that contributes to the plume rise varies between 50% and 75%,
which is higher by at least a minimum of 5% than the EPA method and standard SCREEN3
default of 45%. This difference can be seen in the effective stack diameter, which, based on
this case study, was higher than all other methodologies. If a ratio technique were applied,
the modeling results would become even more conservative than currently presented in
the results, since one would be decreasing the amount of heat release contributing to the
plume rise.

Therefore, there are truly only six methodologies that differ from one another since the
Ohio EPA, Oklahoma DEQ, and NJDEP methodologies can be grouped together and the EPA
and ADEM methodologies can be grouped together. The difference in results derives
primarily from the difference in effective stack diameter and whether or not an effective
stack height or the actual stack height of the flare is used. In keeping with general air
dispersion modeling concepts and assuming the equivalent stack diameter is
approximately the same, the methodologies that account for an effective stack height tend
to result in lower modeling concentrations than those that use the actual stack height since
the effective stack height is higher than the actual stack height. These observations are
provided in Table 6, where a ranking of 1 represents the lowest value for stack diameter
and stack height but the highest modeling concentration (i.e. from the results ranking).

Table 6. Comparison of Stack Parameters and Results Ranking

Methodology Stack Stack Height Results


Diameter Ranking
ADEM 1/2/3/4 Actual (1/2/3/4/5) 1
Turner 5 Actual (1/2/3/4/5) 2
Oklahoma DEQ 1/2/3/4 7/8/9/10/11 3/4/5
Ohio EPA 1/2/3/4 7/8/9/10/11 3/4/5
NJDEP 1/2/3/4 7/8/9/10/11 3/4/5
EPA 7/8/9 Actual (1/2/3/4/5) 6/7
ADEM with Ratio Technique 7/8/9 Actual (1/2/3/4/5) 6/7
Iowa DNR 6 6 8
TCEQ 14 Actual (1/2/3/4/5) 9
Alberta 12 7/8/9/10/11 10
Alberta with EPA Constants 13 7/8/9/10/11 11
Oklahoma DEQ with Ratio Technique 7/8/9 12/13 12
Ohio EPA with Ratio Technique 10/11 12/13 13
NJDEP with Ratio Technique 10/11 14 14
*Actual represents the actual physical stack height.
**A value of 1 represents the lowest value for the stack diameter and stack height but the highest resulting
modeled concentration.

12
The radiation heat loss factor also proves to have significant impacts on the results. The
lower the amount of radiation heat loss, the higher the contribution of heat to the plume
rise. Therefore, if one can refine the radiation heat loss value to be lower than the
SCREEN3 default value of 55%, improved modeling results can occur.

The delta between the lowest resulting modeling concentration and the highest resulting
modeling concentration ranges from 20 to 32% of each scenarios maximum resulting
concentrations (i.e. minimum resulting modeling concentration/maximum modeling
concentration). Therefore, the methodology chosen can significantly impact the modeling
results.

SUMMARY
In this paper a case study was performed to compare the different methodologies available
for flare dispersion modeling when treating the source as a modified point source. The
case study was based on an upstream energy facility in the Middle East, which has some of
the largest flares in the world with considerable regulatory emphasis and compliance onus.
Consequently, any Environmental Impact Assessment (EIA) for major upstream
developments that envision addition of new gas processing trains or significant process
modifications, there is a substantial regulatory requirement to perform flare modeling. The
project proponent air quality dispersion modeling portfolio encompasses a gamut of flaring
scenarios that include unit startup, planned or unplanned unit shutdowns etc. In the
absence of a clear flare modeling directive from the regulators, most practitioners are left
to their experience in formulating a flare modeling solution predicated by earlier project
experience or precedence. Further, the diversity of guidance available in the public domain
for modeling industrial flares (as discussed in this paper) serves to further compound the
selection of an appropriate flare modeling solution.

As determined through the case study, some of the modeling methodologies result in
concentrations higher than those predicted by the EPA method and some result in
concentrations lower than those predicted by the EPA method. The methodologies
consistently resulting in higher predicted maximum concentrations include the Turner
approach and the ADEM approach. In addition, the Oklahoma DEQ, Ohio EPA, and NJDEP
methods, which are all very similar, will also provide higher predicted maximum
concentrations depending on the averaging period and hours of operation of the flare. The
ADEM methodology with the ratio technique applied results in the exact same predicted
maximum concentrations as the EPA method. The modeling methodologies that result in
lower predicted maximum concentrations than the EPA method include the Alberta
method, TCEQ, and all modeling methodologies for which the ratio technique was applied,
with the exception of the ADEM method. The NJDEP methodology with the ratio technique
applied generally tends to produce the lowest maximum modeled concentration results out
of all of the methodologies reviewed.

The stack height and effective stack diameter are the main differences between the
methodologies and the cause of the different modeling results. Based on the modeling

13
results, the effective stack diameter seems to have a larger impact on the modeling results
than the stack height (effective or actual). The radiation heat loss factor also proves to
have significant impacts on the results. The lower the amount of radiation heat loss, the
higher the contribution of heat to the plume rise. Therefore, if one can refine the radiation
heat loss value to be lower than the default value of 55%, lower resulting modeling
concentrations can be realized. Based on a literature review, the default value of 55% in
SCREEN3 appears to very conservative and actual radiation heat loss values may be only
half of this value.6,7,8,11,12 Therefore, this parameter should be studied and reviewed closely
to see if the methodologies can be further refined to be more accurate. The authors hope
that besides the obvious usefulness of compiling the available flare modeling guidance in
this paper, there are other utilitarian dividends as well. Notwithstanding the reasons for
choice of a case study located in the Middle East with the importance of assessing impacts
of some of the largest flares in the world to the local airshed, the delineation of the
respective modeling approaches and their potential impacts (albeit limited to the case
study conditions) is a useful exercise. The next phase of the study will extend the study to a
larger meteorological data set and examine the impact of radiant heat loss and the
interaction of effective stack diameter over effective stack height to have a more extended
narrative on flare modeling and its unique focus in the energy industry.

14
ACKNOWLEDGMENTS

George Schewe, Trinity Consultants, Covington, KY


Jacquie Hui, Trinity Consultants, Dallas, TX
Dr. Weiping Dai, Trinity Consultants, Dallas, TX

REFERENCES

1. A. Idriss, and F. Spurrel. Air Quality Model Guideline. Climate Change, Air and Land
Policy Branch, Alberta Environment, May 2009.
2. SCREEN3 Model User's Guide. U.S. Environmental Protection Agency (c), 1995. Office
of Air Quality Planning and Standards. Emissions, Monitoring and Analysis Division.
Research Triangle Park, NC. EPA-454/B-95-004
3. NSR Emission Calculations for Vapor Oxidizers. Texas Commission on Environmental
Quality, Air Permits Division.
(http://www.tceq.texas.gov/assets/public/permitting/air/Guidance/NewSourceRe
view/emiss_calc_vaporox.pdf)
4. Air Dispersion Modeling Guidelines for Oklahoma Air Quality Permits, Engineering
Section of the Permitting Unit, Air Quality Division, Oklahoma Department of
Environmental Quality, December 2006.
5. Engineering Guide #69, Air Dispersion Modeling Guidance, Air Quality Modeling
Planning Section, Division of Air Pollution Control, Ohio EPA, 2003.
6. Modeling Flares in BEEST. (http://www.beeline-
software.com/LinkClick.aspx?fileticket=YC1Ay0ftkXQ%3D&tabid=403&mid=851).
7. Modeling Flares. Iowa Department of Natural Resources.
(http://www.iowadnr.gov/portals/idnr/uploads/air/insidednr/dispmodel/flares.p
df)
8. Turner, Dr. Bruce, and Richard Schulze. Practical Guide to Air Dispersion Modeling.
Trinity Consultants, Inc. and Air & Waste Management Association, 2007.
9. Plume Rise, G.A. Briggs, 1969.
10. Texas Commission on Environmental Quality, Interoffice Memorandum on
Technical Basis for Flare Parameters, September 10, 2004.
11. Selma E. Guigard, Warren B. Kindzierski, and Nicola Harper. Heat Radiation from
Flares. Department of Civil and Environmental Engineering, University of Alberta,
for Alberta Environment, May 2000.
12. EPA/OAQPS Control Cost Manual Chapter 7. Flares. United States Environmental
Protection Agency, December 1995.

KEYWORDS

Flare, dispersion, modeling, radiation heat loss

15

Potrebbero piacerti anche