TemplateTemplate
Toward Improved Basin-Level Oil and Gas Inventories and Reconciliation
with Measurements
Amnon Bar-Ilan, John Grant, Rajashi Parikh, Ralph Morris ENVIRON International Corporation
Garvin Heath, Viktor DiakovNational Renewable Energy Laboratory
Dan ZimmerleColorado State University
Tom MooreWRAP/WESTAR
EPA Emission Inventory ConferenceApril 16, 2015, San Diego
Overview
• Results from recent WRAP work in the Williston Basin
• Innovations implemented in the Williston inventory
• Reconciliation study with D-J Basin inventory• Upcoming NETL reconciliation study
2
Williston Basin
• Consists of a large area in North Dakota and Montana• Legacy gas production and recent booming oil production
– Centered on the Bakken oil shale formation– North Dakota now 2nd largest onshore oil production state in the
US– Significant areas of production
on tribal land - Fort Berthold Indian Reservation (FBIR) and limited production on Fort Peck Indian Reservation
0
200000
400000
600000
800000
1000000
1200000
1400000
FL NY PA VA WV IL IN KS KY MI
MO NE
ND
OH OK SD TN AL AR LA MS
NM TX CO MT
UT
WY AK AZ CA NV
Ons
hore
Oil
Prod
uctio
n by
Sta
te (t
hous
ands
of b
arre
ls)
3
Williston Basin Key Features
• 2011 – roughly 10,000 producing wells and over 175 million bbls of oil production– Over 200,000,000 MCF of gas produced of which the vast
majority is associated gas– Lack of infrastructure to capture and process associated gas
4
Williston Basin
5
Williston Basin Inventory
Mineral DesignationNOx VOC CO SOx PM
[tons/yr] [tons/yr] [tons/yr] [tons/yr] [tons/yr]Tribal 3,485 17,306 6,245 432 103Private/State 22,715 231,430 33,837 5,834 833BLM 1,738 27,981 3,445 293 72USFS 1,466 19,771 2,778 337 51Total 29,404 296,488 46,305 6,895 1,060
6
Williston Basin – Tribal MNSR
• Subpart OOOO requires reporting of minor O&G sources on tribal land– FBIR represented 14%, 8%, and 5% of oil production, gas
production, and active well count, respectively, in the Williston Basin in 2011
– 10 tons per year of carbon monoxide (CO), nitrogen oxides (NOx), sulfur dioxide (SO2), or particulate matter (PM), or 5 tons per year of volatile organic compounds (VOCs), or 2 tons per year of hydrogen sulfide (H2S)
– Midstream sources were separately identified in MNSR registrations and major source inventories provided by EPA Region 8
7
Williston Basin – Tribal MNSR
• Data mined tribal MNSR registrations for FBIR– Over 150 well site registrations randomly sampled– Input data for emission calculations
• Artificial lift engines• Casinghead gas• Wellhead compressors• Fugitives• Miscellaneous engines• Water tanks• Heaters• Oil Tanks• Truck loading of oil• Gas compositions
OIL TANKS
Parameter SurveyTribal MNSR Units
Representative Input Factors
% o
f Ta
nks Uncontrolled 10% 0% -
Flare 70% 0% -VRU 13% 0% -Enclosed Combustor 6% 99%
-
VOC Emission Factor 5.6 5.4 lb VOC/bbl68.2 65.9 SCF/bbl
VOC Mole Fraction 55% 79% -
Per Surrogate EmissionsVOC 0.97 0.11 lb/bbl
8
Inventory Reconciliation
• Reconciliation with top-down measurements of VOC or methane flux– Inventories underestimate emissions– Reasons unclear
• Examples of reconciliation studies– DJ Basin overflight inventories for Weld County and ground
tower-based measurements (Petron et al., NOAA)– Uinta Basin overflight and ground-based mobile lab
measurements (Karion et al., NOAA)– Barnett Shale suite of studies by EDF-funded team– Denver ozone modeling O&G source apportionment study
9
Background
• 2008 Denver ozone State Implementation Plan (SIP) used a June-July 2006 photochemical modeling database to demonstrate attainment of the 1997 8-hour ozone NAAQS (0.08 ppm) by 2010
• During June-July 2006, CDPHE/APCD collected VOC measurements on several days– Evaluation of the CAMx photochemical grid model using the
VOC measurements found that it underestimated the observed VOC concentrations
10
VOC and Ethane Underestimated at Weld Co Sites
Predicted vs. Observed VOC at Ft Lupton
0.00
200.00
400.00
600.00
800.00
1000.00
1200.00
6061
6
6061
9
6062
2
6062
5
6062
8
6070
1
6070
4
6070
7
6071
0
6071
3
6071
6
6071
9
6072
2
6072
5
6072
8
Date
VOC
[ppb
C]
Predicted Observed
Predicted vs. Observed VOC at Platteville
0.00
500.00
1000.00
1500.00
2000.00
2500.00
6061
6
6061
9
6062
2
6062
5
6062
8
6070
1
6070
4
6070
7
6071
0
6071
3
6071
6
6071
9
6072
2
6072
5
6072
8
Date
VOC
[ppb
C]
Predicted Observed
Predicted vs. Observed ETHA at Ft Lupton
0
50
100
150
200
250
300
6061
6
6061
9
6062
2
6062
5
6062
8
6070
1
6070
4
6070
7
6071
0
6071
3
6071
6
6071
9
6072
2
6072
5
6072
8
Date
ETH
A [p
pbC
]
Predicted Observed
Predicted vs. Observed ETHA at Platteville
0
100
200
300
400
500
600
700
6061
6
6061
9
6062
2
6062
5
6062
8
6070
1
6070
4
6070
7
6071
0
6071
3
6071
6
6071
9
6072
2
6072
5
6072
8
Date
ETH
A [p
pbC
]
Predicted Observed
11
CAMx VOC Sensitivity Test
• No or very small improvements (few ppb at most) in ozone performance on some days/sites
• Significant improvements in VOC at Weld County sites– At Platteville, factor of 6 average VOC under-prediction reduced to factor of 1.5
• Days when back trajectories have longer residence time over Weld County O&G sources VOC underestimation bias is the greatest– Especially for PAR and ETHA, source signatures for
O&G emissions12
Preliminary CMB and PMF VOC Source Apportionment
• Input VOC Source Profiles for CMB:– Compressed Natural Gas (CNG)– Geogenic Natural Gas (GNG)– Liquid Petroleum Gas (LPG)– Gas Evaporative (Gas Evap)– Vehicle Exhaust (Gasoline Combustion)– Biogenic Oil and gas sources include combination of CNG, GNG, LPG and
Gas Evap
• Preliminary PMF using 4 Factors
13
CMB and PMF Receptor Modeling• Very good match
between first three PMF factors and Oil and Gas, Vehicle Cold Start and Vehicle Running Exhaust VOC profiles
• Fourth factor compared with Biogenic+Solvent, but not a very good match.
0%
5%
10%
15%
20%
25%
30%
35%
40%
BZ12
3MBZ
124M
BZ13
5MPE
NTE
1PA
224M
BU22
DMPA
234M
BU23
DMPE
N23
MPE
N24
MH
EP2M
EH
EXA2
MPE
NA2
MH
EP3M
EH
EXA3
MPE
NA3
MAC
ETYL
BEN
ZEC2
BUTE
C2PE
NE
CYH
EXA
CPEN
TAET
HAN
EET
BZI_
BUTA
LBU
T1E
IPEN
TAI_
PREN
DETB
Z1M
ECYH
XM
CYPN
AM
_ETO
LM
P_XY
LN
_BU
TAN
_DEC
N_H
EPT
N_H
EXN
_NO
NN
_OCT
N_P
ENT
N_P
RBZ
N_U
NDE
O_E
TOL
O_X
YLP_
ETO
LN
_PRO
PST
YRTO
LUE
T2BU
TET2
PEN
E
Oil & Gas Emissions Profiles (CMB = CNG + GNG + LPG + GEVAP)
PMF
CMB
0%2%4%6%8%
10%12%14%16%18%
BZ12
3MBZ
124M
BZ13
5MPE
NTE
1PA
224M
BU22
DMPA
234M
BU23
DMPE
N23
MPE
N24
MH
EP2M
EH
EXA2
MPE
NA2
MH
EP3M
EH
EXA3
MPE
NA3
MAC
ETYL
BEN
ZEC2
BUTE
C2PE
NE
CYH
EXA
CPEN
TAET
HAN
EET
BZI_
BUTA
LBU
T1E
IPEN
TAI_
PREN
DETB
Z1M
ECYH
XM
CYPN
AM
_ETO
LM
P_XY
LN
_BU
TAN
_DEC
N_H
EPT
N_H
EXN
_NO
NN
_OCT
N_P
ENT
N_P
RBZ
N_U
NDE
O_E
TOL
O_X
YLP_
ETO
LN
_PRO
PST
YRTO
LUE
T2BU
TET2
PEN
E
Vehicle Cold Start Emissions Profiles
PMF
CMB
0%
5%
10%
15%
20%
25%
30%
35%
40%
BZ12
3MBZ
124M
BZ13
5MPE
NTE
1PA
224M
BU22
DMPA
234M
BU23
DMPE
N23
MPE
N24
MH
EP2M
EH
EXA2
MPE
NA2
MH
EP3M
EH
EXA3
MPE
NA3
MAC
ETYL
BEN
ZEC2
BUTE
C2PE
NE
CYH
EXA
CPEN
TAET
HAN
EET
BZI_
BUTA
LBU
T1E
IPEN
TAI_
PREN
DETB
Z1M
ECYH
XM
CYPN
AM
_ETO
LM
P_XY
LN
_BU
TAN
_DEC
N_H
EPT
N_H
EXN
_NO
NN
_OCT
N_P
ENT
N_P
RBZ
N_U
NDE
O_E
TOL
O_X
YLP_
ETO
LN
_PRO
PST
YRTO
LUE
T2BU
TET2
PEN
E
Vehicle Running Emissions Profiles
PMF
CMB
0%
10%
20%
30%
40%
50%
60%
BZ12
3MBZ
124M
BZ13
5MPE
NTE
1PA
224M
BU22
DMPA
234M
BU23
DMPE
N23
MPE
N24
MH
EP2M
EH
EXA2
MPE
NA2
MH
EP3M
EH
EXA3
MPE
NA3
MAC
ETYL
BEN
ZEC2
BUTE
C2PE
NE
CYH
EXA
CPEN
TAET
HAN
EET
BZI_
BUTA
LBU
T1E
IPEN
TAI_
PREN
DETB
Z1M
ECYH
XM
CYPN
AM
_ETO
LM
P_XY
LN
_BU
TAN
_DEC
N_H
EPT
N_H
EXN
_NO
NN
_OCT
N_P
ENT
N_P
RBZ
N_U
NDE
O_E
TOL
O_X
YLP_
ETO
LN
_PRO
PST
YRTO
LUE
T2BU
TET2
PEN
E
Remaining Sources Profile Comparison (CMB = Sovent + Biogenic)
PMF
CMB
14
Emissions-Based VOC Source Apportionment Modeling and Comparison to Receptor Modeling
• Use CAMx Ozone Source Apportionment Technology (OSAT) to track VOC emissions for major source categories:– Mobile Sources (on-road plus non-road)– Biogenic Sources– Oil and Gas Sources– Area/Point Sources
• Compare CAMx/OSAT VOC source apportionment with Revised CMB and PMF VOC Source Apportionment
15
Biogenics
6%
Mobile4%
Oil and Gas85%
Area+Point5%
Ft Lupton Revised PMF
Biogenics16%
Mobile15%
Oil and Gas51%
Area+Point18%
Fort Lupton CAMx OSAT
Biogenics9%
Mobile6%
Oil and Gas78%
Area+Point7%
Plateville CAMx OSAT
16
Conclusions: VOC Source Apportionment
• Comparison of monitor-based CMB/PMF and emissions-based OSAT VOC source apportionment inconclusive:– Is CAMx VOC underestimation bias due to missing VOCs or
differences between modeled volume average and surface point measurement
– VOC source categories in CMB, PMF and OSAT represent different sources
• Results consistent with O&G VOC emissions being understated– Work led to FLIR camera purchases and deployment in DJ
condensate tank thief hatch identified as key VOC category
17
Future Work – NETL Reconciliation Study• NETL funding a group of researchers (NOAA, NREL, CSU,
CSM) to study methane emissions from onshore gas development– Includes top-down measurements, bottom-up inventory
move to reconcile the two
• Improve bottom-up inventories– Separate episodic and routine sources through surveys of
operator activities– Time period of inventory aligned with measurement period– Align surveys with Subpart W reporting to access that activity
data– Use distribution-based EFs & Monte Carlo methods to generate
inventory uncertainty estimates18
Acknowledgements
Susan BassettBLM Montana/Dakotas Field Office
Billings, MT
Lee GriboviczAirstar Consulting
Ken Lloyd, Jerry DilleyRegional Air Quality Council (RAQC)
Denver, CO
19