An Atmospheric Chemistry Module for Modeling Toxic Industrial Chemicals
(TICs) in SCIPUFF
Douglas S Burns, Veeradej Chynwat, Jeffrey J Piotrowski, Kia Tavares, and Floyd Wiseman
ENSCO, Inc.
Science and Technology for Chem-Bio Information Systems(S&T CBIS)
October 28, 2005
BAA TYN 03-001Atmospheric Chemistry Module for
Toxic Industrial Chemicals
Tyndall AFB / DTRA
Michael HenleyAFRL/MLQ Tyndall AFB, FL
Martin BagleyDTRA / TDOC Alexandria, VA
Outline
• Project Goals• Methodology
– Integration in SCIPUFF– Chemistry of 1-butene– Derivation of keff, Xeff– Parameter Space
• Results– Model output
• Decay of TICs (1-Butene, Methylpropene)• Product Formation
• Summary
Project Goals• Develop initial atmospheric chemistry capability
– Develop Atmospheric Chemistry Algorithm• Algorithm MUST run rapidly.• Develop generic algorithm so that a detailed chemical kinetics approach
is not required.• Algorithm must account for all (most) modeling scenarios (e.g., CC, T,
ambient conditions).• Algorithm must be robust enough to account for diurnal changes to
degradation rates.• Algorithm should account for the potential generation of intermediate
toxic compounds.– Develop Chemical data for the Chemistry Algorithm
• Review existing chemistry data for nine alkenes (and H2S)• Develop mechanisms used to generate chemistry algorithm.
• Couple Algorithm to SCIPUFF– Work with Dr. Sykes to create interface with SCIPUFF
• Launch Chemistry Module from HPAC
Methodology: Minor Modification to SCIPUFF
SCIPUFF Dispersion
−= ( )(min)(max) ),sin(max AAA kkk φ=
∫=.),,,],[,( etchumidityccTambfkk effA φ==
Method: Create Degrade Dynamic Link LibraryDetails in the Software Development Plan
API
SCIPUFF Function Library Plottool
Interface Functions
FileManager
PlotManager
ProjectManager
Dispersion
MeteorologyGenerator
SWIFT
MC-
SCIPU
FF
FileReaders
UtilityFunctions
PlotGenerator
EffectsManager
EffectM
odule
EffectM
odule
DEGRADE DLL
• Algorithm is transparent to the User
• Code alwayscalls chemistry
Methodology: Chemistry of 1-butene
Determine ID Rxn’s, EA, k(T)
(w/ OH, NO3, H2O, O3, etc.)
Implement inDetailed
Mechanism
Run PBM as
f(met parm’s)
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
Methodology: Chemistry of 1-buteneCH2=CHCH2CH3 + OH ⎯⎯→ 0.94 C2H5CHO + Other products
CH2=CHCH2CH3 + NO3 ⎯⎯→ 0.12 C2H5CHO + 0.6 C3H5O-CH-O-NO2+ Other products
CH2=CHCH2CH3 + O3 ⎯⎯→ 0.35 C2H5CHO + 0.41 OH + Other products
Rate = -(kOH[OH] + kNO3[NO3] + kO3[O3]) [1-butene]
Rate = -keff [1-butene]
Methodology: Chemistry of 1-butene
Determine ID Rxn’s, EA, k(T)
(w/ OH, NO3, H2O, O3, etc.)
Implement inDetailed
Mechanism
Run PBM as
f(met parm’s)
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
krxn as f(T)
5
10
15
20
25
260 270 280 290 300 310 320
Temperature [K]
k NO
3 and
kO
3 [pp
m-1
min
-1]
40000
45000
50000
55000
60000
k OH [p
pm-1
min
-1]
kO3 * 10^3kNO3kOH
Methodology: Detailed Mechanism
Determine ID Rxn’s, EA, k(T)
(w/ OH, NO3, H2O, O3, etc.)
Implement inDetailed
Mechanism
Run PBM as
f(met parm’s)
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
• Carbon Bond Mechanism– Mass consistent atmospheric chemistry
mechanism.– EPA Model (Adelman, 1999).– Used to model the ambient conditions.– [OH], [NO3], [O3], NOX, VOCs, etc..
• Append chemistry for TIC– Data provided for 9 alkenes and H2S
Methodology: Run Detailed Chemistry
Implement inDetailed
Mechanism
Run PBM as
f(met parm’s)
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
or
NOX VOC’s
O3 CO
Location (lat,long)
ZA = f(Lat, Lon, DOY, Time of day) H2O
Temperature
keff is a function of solar elevation, cloud cover, air quality, temperature, humidity, etc
Methodology: Parameter Space
Parameter Units SCIPUFFSolar Zenith Angle 0 – 90 Deg X
Location (lat, lon) 0 – 70 Deg X
Time of Day 1440 min X
Day of Year 3/21, 6/20, 12/20 X
Photochemistry (Cloud Cover) 0 – 8 Eighths X
Temperature 230 – 310 K X
Water Concentration 100 – 40000 PPM
Moisture Mixing ratio X
Air Quality [NOX], VOC, O3, …
Land Use Urban, ocean, forest, … X
Methodology: Surrogate for Air Quality• Land Use 1=Developed
2=Dry Cropland & pasture3=Irrigated Cropland5=Cropland/Grassland6=Cropland/Woodland7=Grassland8=Shrubland9=Shrubland/Grassland10=Savanna11=Deciduous Broadleaf12=Deciduous Needleleaf13=Evergreen Broadleaf
14=Evergreen Needleleaf15=Mixed Forest16=Water17=Herbaceous Wetland18=Wooded Wetland19=Barren20=Herbaceous Tundra21=Wooded Tundra22=Mixed Tundra23=Bare Tundra24=Snow or Ice25=Partly Developed
1001=Urban Superclass1002=Grassland Superclass1003=Forest Superclass1004=Desert Superclass1005=Water Superclass
Methodology: Surrogate for Air QualityNOx vs VOC (vary by Latitude)
(Mar, Jun, Dec, 2000, T = 280K, CC = 0, Lat 0-60)
1.00E-06
1.00E-05
1.00E-04
1.00E-03
1.00E-02
1.00E-01
1.00E+00
0.001 0.01 0.1 1
Propene-Eq (ppm-C)
NO
x (p
pm)
Water
Forest
Urban
Grassland
Desert
Default
Methodology: Refined Parameter Space (T, H2O)
• Surface Stations Nov 2003 – Sep 2004.• Global 0.5 km LU Data Set
Temperature (K) [H2O] (x103) ppm)Min12.47.054.55
30 274 310 3.81 34.940 265 304 1.54 28.550 257 299 1.02 19.860 245 294 0.400 14.270 231 291 0.113 11.6
LatitudeMin Max Max
0 288288288
310 37.110 310 37.420 310 37.1
1. Extracted data using 3 hr interval instead of 30 sec data. (both day and night)
2. Removed extreme data points (i.e., T<-60 °C or T<Dew point).
3. Matched weather station data with LU data before analysis (5 categories).
Methodology: Run Detailed Chemistry
[ ][ ] [ ][ ] [ ][ ] [ ] .....33 33−−−−−=⎟
⎠⎞
⎜⎝⎛
∂∂
−= iiOiNOiOHChemistry
ii ckcOkcNOkcOHk
tcr
Implement inDetailed
Mechanism
Run PBM as
f(met parm’s)
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
[ ]ieffchemistry
i ckdtdc
=⎟⎠⎞
⎜⎝⎛−
[ ]cdtdc
keff
−=
Methodology: Obtain CTIC as f(t)[ ]
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0 360 720 1080 1440
20253035404550
Implement inDetailed
Mechanism
Run PBM as
f(met parm’s)
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
[ ]cdtdc
keff
−=
[Butene] as f(time)Latitude = 20 – 50º N
20º
50ºT = 290 K, Land Use = Urban
Time of Day (min from midnight)
[But
ene]
[PPM
]
Methodology: Obtain keff as f(met parms)
0.0000
0.0004
0.0008
0.0012
0.0016
0.0020
0 360 720 1080 1440
253035404550
Implement inDetailed
Mechanism
Run PBM as
f(met parm’s)
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
[ ]cdtdc
keff
−=
T = 291 K, Land Use = Urban
25º
50º
keff as f(time)
k eff
[min
-1]
Time of Day (min from midnight)
Methodology: Derive Empirical keff
• Generate keff for various combinations of meteorological parameters for each land use
• Transform data to center on all parameters• Perform statistical regression - correlation
– Review Equation– Review Statistical Parameters (e.g., r2)– Weigh fit vs number of parameters
• Derive an empirical keff = f(SE, T, lat, tod, CC, [H2O])• Compare the keff (empirical model) with the PBM derived keff.
Obtain cTIC(t) as
f(met parm’s)
Derive Empiricalkeff (met parm’s)
Populate SCIPUFF data tables w/
keff for TICs
Results: keff (polynomial) vs keff (PBM) for butene
Model: 7 Parametersr2 = 0.95
keff [min-1] from PBM
Land Use = Urbank e
ff[m
in-1
] fro
m P
olyn
omia
l
> 500,000 points
Results: keff (polynomial) vs keff (PBM) for butene
Land Use = Grassk e
ff[m
in-1
]
PBM
Polynomial
Lat 0°, Temp 300 K, Cloud Cover 0/8, [H2O] = 20000 ppm,
Time from Solar Noon [min]
1.50x10-3 min-1
1.43x10-3 min-1
Results: keff (polynomial) vs keff (PBM) for butene
Land Use = WaterLat 0°, Temp 300 K, Cloud Cover 0/8, [H2O] = 20000 ppm,
k eff
[min
-1]
PBM
Polynomial
Time from Solar Noon [min]
9.3x10-4 min-1
9.2x10-4 min-1
Methodology: Obtain Xeff
CH2=CHCH2CH3 + OH ⎯⎯→ 0.94 C2H5CHO + Other products
CH2=CHCH2CH3 + NO3 ⎯⎯→ 0.12 C2H5CHO + 0.6 C3H5O-CH-O-NO2+ Other products
CH2=CHCH2CH3 + O3 ⎯⎯→ 0.35 C2H5CHO + 0.41 OH + Other products
Rate = -(kOH[OH] + kNO3[NO3] + kO3[O3]) [1-butene]
Rate = -keff [1-butene]
Rate = +(0.94 kOH[OH] + 0.12 kNO3[NO3] + 0.35 kO3[O3])[butene]
Rate = +Xeff keff [1-butene]
Methodology: Obtain XeffT = 295 K, Land Use = Water
Stoichiometry for Propanal Formation
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 360 720 1080 1440
Time from Midnight [Min]
Stoi
chio
met
ric C
oeffi
cien
t [-]
OH
O3
NO3
Results: Nine Alkenes
• Priority I– 1-Butene
• Products (Propanal, Nitroxybutanone).– Ethene– Propene– Methylpropene– 1,3-Butadiene
• Priority II– Styrene
• Priority III– cis-2-Butene– trans-2-Butene– Isoprene
Why Chemistry is Important
in AT&D Modeling
Results: T&D Compared to T&D + Chemistry (Butene)
T&D Only T&D + Chemistry
SCIPUFF
Tracer 1-Butene
Results: Methylpropene8 hr continuous release starting at 8 am local time
No Chemistry
Chemistry
8 AM9 AM10 AM11 AM12 PM1 PM2 PM3 PM Local Time
Results: Calculated Plume is TIC Dependent
3-Methyl Propene
1,3-Butadiene
Ethene Propene
Results: TIC Decay and Product Formation
1-Butene Propanal
At 4 hrs and 8 hrs after release
2 hr continuous release starting at noon local time
Results: Test and Evaluate (Output)Comparison of Original SCIPUFF and “Degrade”
kmin / kmax
keff (met parms)
Diffusion
Summary & Future Work
• Developed Chemistry Model for 10 TICs– 9 Alkenes + H2S
• No slow down in SCIPUFF• Ability to model product formation• Future Work
– Site specific keff’s– keff’s for other TICs
• Complementary lab / theory development of fundamental kOH, kO3, kH2O, etc.
– Chamber Studies (Chemistry Validation)– Field Studies (Model Validation)
End of slides
Example: Crop Dusting Scenario – Bottom line
03 July 06:00 (Local) 13:00 Z 03 July 07:00 (Local) 14:00 Z 03 July 08:00 (Local) 15:00 Z 03 July 09:00 (Local) 16:00 Z 03 July 10:00 (Local) 17:00 Z 03 July 11:00 (Local) 18:00 Z 03 July 12:00 (Local) 19:00 Z 03 July 13:00 (Local) 20:00 Z 03 July 14:00 (Local) 21:00 Z 03 July 15:00 (Local) 22:00 Z 03 July 16:00 (Local) 23:00 Z 03 July 17:00 (Local) 00:00 Z 03 July 18:00 (Local) 01:00 Z 03 July 19:00 (Local) 02:00 Z 03 July 20:00 (Local) 03:00 Z 03 July 21:00 (Local) 04:00 Z 03 July 22:00 (Local) 05:00 Z
Chemistry vs T&D
Degrade Generation