1. Ezra Wood, Scott Herndon, Luwi Oluwole, Simon Albo
6/21/2011
FLAIR workshop, Austin TX
Acknowledgments: TCEQ
Characterization of gas and particle emissions in the greater Houston area using the Aerodyne mobile laboratory
Ezra Wood, Scott Herndon, Luwi Oluwole, Simon Albo
T. Onasch1, E. Fortner1, J. Jayne1, J. Wormhoudt1, P. Massoli1, C. Kolb1, H. B. Lee2, M. Zavala3, L. T. Molina3, and W. B. Knighton4
Aerodyne Research, Inc., Harvard University, Molina Center for Energy & Environment,
Montana State University

2. Instrumentation CO, NO2, C2H4, HCHO (QC-TILDAS) O3, NO, CO2, SO2
aromatic VOCs, 1,3-butadiene,
OVOCs (PTR-MS/NO+MS)
O3, NO, CO2, SO2
PAN (GC), VOCs (canisters) (UH)
Instrumentation
Size and chemically-speciated PM
(Aerosol Mass Spectrometer) Particle number (CPC)
Black Carbon (MAAP), extinction
PM size distribution (SMPS)
Wind speed/direction
Actinic flux
GPS position

3. • quantification (pounds/hr) • location identification
Goal:
Conduct measurements that support
emissions characterization
• quantification (pounds/hr)
• location identification
Mobile sampling:
Texas City, Mont Belvieu, Ship Channel
Stationary sites:
Texas City
Mont Belvieu
U. Houston

4. C2H4 and HCHO mapping in Mont Belvieu
WIND

5. Quantification of emissions from “traditional” combustion source: fuel-based emission factors
Ship Channel
May 28, 2009

6. Carbon balance method Emission Inventory 58 g NOx/kg fuel
18 ppb NOx / ppm CO2
x total fuel consumption
Emission
Inventory
58 g NOx/kg fuel

7. HONO/NOx: 0.7 to 1.4% (similar to on-road diesel vehicles)
Ship emission factors
Plume time
Vessel name/type
NOx
g/kg fuel
dOx/
dNOx
HCHO
g/kg
CO g/kg
SO2
10:04
Izumo Princess 58
0.1
10:26
Vega Spring 61 21
10:59
Odfjell Seachem 54
0.06 12 25
11:04
UBC Bremen? 80
0.08
11:05
Eitzen Chemical 50
0.11
0.19 48 34
11:17
Leyte Spirit 89
0.07
0.17 10 37
11:24
tug/ferry 55
0.12
0.40
1.2
11:36 30
0.09 – 0.15
0.13
11:54
Petropavlovsk 44
0.16 17
Williams
MSDc
61.5
0.15
11.0
6.3
SSDb
79.6
11.8
27.8
This method works great with good winds. And when mobile, can usually drive around to make the winds work for you. Any decent research grade 1-second instrument can be incorporated into the mix. Daytime, nighttime both work. But still, wind has to be cooperative. Thus this is a good complement to the remote sensing methods that others will talk about.
HONO/NOx: 0.7 to 1.4% (similar to on-road diesel vehicles)
-based on comparison with UCLA iDOAS HONO/NO2 ratios

8. use carbon balance method …
Flares:
use carbon balance method …
…with a few complications
TCEQ’s Comprehensive Flare Study
September 2010:
Assumed DRE = 0.98, more like 0.3 to 0.7ish
HCHO/CO = 0.02 to 0.05 for propene flare
• Destruction Removal Efficiency (DRE) vs. fuel-based emission factors • Assumption that most C ends up as CO, CO not valid

9. Emissions observed with ARI mobile laboratory during FLAIR 2009:
1) Useful correlations between combustion tracers (CO, CO2) and VOCs
2) Obvious fugitive emissions
3) Unclear – no obvious correlation between combustion tracers
and VOCs, but can’t rule it out

10. 1. (Useful VOC-COx correlations) Flare Emission Capture from Mobile Laboratory
Mobile Lab Maneuvered Here
Prevailing Wind
Known Plant Flare P-200

11. Flare Emission Capture from Mobile Laboratory
Prevailing Wind
This is NOT self sampling.
The incoming wind speed is 4-7 mph.
VOC/CO ratio invariable, but CO2/CO changes – possibly from a clean CO2 source nearby.
C2H4/CO2 ratios higher than highest superemitting on-road vehicle
Known Plant Flare P-200 11

12. Flare Emission Capture from Mobile Laboratory
Prevailing Wind
There are different CO2 to CO ratios – there seems to be only one VOC to CO ratio & HCHO/CO is RIGHT IN LINE WITH TCEQ Comprehensive Flare Study!!
myriad other sources of CO2
Ethylene to formaldehyde ratio is high suggesting ethylene is vent gas (but don’t know for sure).
DRE def less than 98%.
Known Plant Flare P-200
Carbon balance methods with a guess about vent gas composition:
DRE = 94% (88% - 96%) 12

13. Large ethene leak, Winfree Rd
2) obvious fugitive emissions / non-combustion source
Large ethene leak, Winfree Rd
·Localized (<10 m)
· no CO/CO2/NOx

14. (5/19/2009, Mt. Belvieu) Unlit flare 2) obvious fugitive emissions
/ non-combustion source
(5/19/2009, Mt. Belvieu)

15. 3) No obvious correlation between combustion tracers and VOCs, but can’t rule out low DRE flare vs. leak

16. 3) No obvious correlation between VOCs and COx – low DRE flares?
(VOC to CO/CO2 ratio believable if DRE < 30%).
C2H4 of 200 ppb = 0.4 ppm of C. Low DRE flare could give just a ppm or 2 of CO2 – within the atmospheric noise.

17. The Aerodyne Inverse Modeling System (AIMS)
Driver
SCIPUFF
SCIPUFF TL/Adjoint
Minimization
algorithm
Obs. Data
(MET, Sensors)
# of sources,
Emission rates,
Locations,
Start and
End times.
Given knowledge of the wind history, determine emission source parameters that when applied in atmospheric dispersion model yield pollutant concentration profiles that are most consistent with observed profiles
Aerodyne Research, Inc.

18. Inversion model results
Texas City
15 pounds/hr benzene source
identified by inversion model
WIND

19. Stationary data: SO2 “upwind” from courthouse site (TC)
Stationary data was useful too. Data points projected 5 min back, linearly. Known regeneration facility nearby.

20. HCHO: same spatial signature filtered day/night

21. Consistent HCHO/SO2 ratio
Each marker (color) is a different transect (90 total). Occasionally there is HCHO w/ little SO2, but usually a consistent HCHO/SO2 ratio.

22. HCHO: Primary vs. secondary?
C2H4 + OH → → 1.43 HCHO
photochemical age (OH exposure):
What else can we learn from these downwind observations? WE can constrain the photochemical age (OH exposure) by examining the HCHO/C2H4 and HCHO/propene ratios… 22

23. Primary HCHO in Texas City?
slope implies [OH] = 2 ×107 to 4 × 107 molecules/cm3
→ evidence for primary HCHO

24. Primary HCHO from Chevron?
C2H4 (ppb)
HCHO (ppb)
Slope = 0.02
Slope and transit time imply [OH] = 1.33 × 106 molecules cm-3 at 07:20 CST
5/21/2009 → no evidence for primary HCHO

25. 1,3-butadiene mapping (Ship Channel)
Goodyear facility – known source of 1,3-butadiene. Were able to measure with Berk Knighton’s “special” 1-second PTR. Luwi will talk more about this facility.
WIND

26. 1,3-butadiene, styrene
1,3-butadiene and styrene emitted from different locations.

27. Summary • Mobile measurements useful for
locating and quantifying emission sources
• Rich dataset:
Marathon flare DRE
Ship emission factors
Winfree road Ethylene leak
Primary HCHO emissions from Texas City facililty
Ethylene, propylene emission from Chevron (Mont Belvieu)
1,3-butadiene, styrene from Goodyear

28. back-up slides

29. Photochemistry

30. P(OH) = L(OH)
Total OH loss rate = 47.3 s-1, and is dominated by reaction with C2H4.
This yields an OH concentation of 1.6 × 105 molecules/cm3. Since the HO2 + NO term is obviously not zero, this number should be considered a lower limit to the true OH concentration. This value is likely higher than the [OH] in non alkene plume air considering the time of day (06:12 local time). Further analysis will address the likely range of values for the HO2 + NO term.