Reactivity Scales via 3-DPhotochemical Modeling

Robert Harley

Dept. of Civil & Environmental Engineering

University of California at Berkeley

Acknowledgments

• Phil Martien (UC Berkeley)

• Jana Milford (CU Boulder)

• Amir Hakami & Ted Russell (GA Tech)

• California Air Resources Board

Introduction

• California considers mass and reactivity ofVOC in regulating some sources

• Relies on concept of maximum incrementalreactivity (“MIR”), based on work of Carterwith 0-D box model

• 3-D grid-based models provide more realisticrepresentation of atmosphere

0 2 4 6 8 10 12

MethaneEthane

n-Butane224-TMP

ethenepropeneisoprenebenzene

toluenem-xylene124-TMB

HCHOCCHO

MEKacetylene

ethanolMTBE

Reactivi ty (g O3 / g VOC)

MIR

MOIR

Incremental Reactivity Scales (Carter, 2000)

Objectives

• Use 3-D air quality models with onlinesensitivity analysis to assess reactivityof individual VOC

• Compare results from 3-D models toCarter’s MIR scale

• Conduct formal uncertainty analysis

Approach

• SAPRC-99 mechanism with 30 individualVOC represented explicit ly

• DDM-3D (Yang et al., 1997) used to calculatesensitivity of ozone to emissions of VOC

• Monte Carlo analysis to propagate inputuncertainties through modeling system

Model Application

• South Coast Air Basin

– 24-25 June 1987 (SCAQS)

• Central California

– 2-6 August 1990 (SJVAQS/AUSPEX)

• Use previously defined model inputs formeteorology and emissions

Calculating Reactivity

• Sensitivity Coefficient from DDM-3D:

• Absolute Incremental Reactivity:

[ ] [ ]j

33* OO

EEs j

jij ∂

∂=

∂∂

=ε

jj

ijj MWE

sAIR

*

=

Relative Reactivity

• AIR from 3-D model varies by location

– Coastal/mid-basin sites not affectedby increases in downwind emissions

• Define relative incremental reactivity:

∑=

kkk

jj AIRw

AIRRIR

Reactivity Rankings

• Sort compounds from highest to lowestbased on Carter MIR

• Plot RIR from 3-D modeling for eachcompound

• Expect RIR to decrease monotonically

Uncertainty Analysis

• Treat 33 input parameters as uncertain:

– Chemical rate coefficients

– Oxidation product yields

– Emissions of CO, VOC & NOx

– Deposition velocities for O3 and NO2

• Use trajectory model and MonteCarlo/LHS to get output uncertainties

Absolute Incremental Reactivities

0.0

0.4

0.8

1.2

1.6

2.0

CO x 10

ETOH

HCHOM

EKN-C

4

PRPE22

4P

XYLMBASE

Anaheim Azusa Claremont Riverside

Central California

• Used different 3-D model (MAQSIP)

• Considered many reactivity metrics:

– 1 hr vs. 8 hr ozone

– MIR vs. MOIR conditions

– Population exposure

• Long modeling period (2-6 Aug 1990)for larger region incl. Bay Area & SJV

0 2 4 6 8 10

2-me-2-Butene1,3-Butadiene

PropeneIsoprene

m-XyleneEthene

FormaldehydePropionaldehyde

Lumped OLE11,2,4-TMB

Acetaldehydea-Pinenep-XyleneToluene

me-CyclopentaneEthanol

relative reactivity

Exposure

MOIR-3D-8h

MIR-3D-8h

MOIR-3D

MIR-3D

Comparison of 3-D Metrics

Comparison of 3-D Metrics

0 0.5 1 1.5 2

Isopentane

n-PentaneMEK

2,2,4-tm-

n-ButaneAcetylene

n-Butyl Acetate

BenzeneMTBE

Methanol

IsopropanolAcetone

Ethane

C OMethane

Benzaldehyde

relative reactivity

Exposure

MOIR-3D-8h

MIR-3D-8h

MOIR-3D

MIR-3D

Summary

• Reactivity more robust on relative ratherthan absolute basis

– less site-to-site variability

– typical RIR uncertainty 20-35%

• Reactivity metrics derived from 3-Dmodeling similar to Carter MIR scale

• Spatial distribution of emissions can beimportant (lg. point sources, biogenics)

Summary (continued)

• HOx radical initiators (esp. C4-C5

alkenes and HCHO) have variable RIR

• Acetaldehyde shows negative reactivityat upwind boundaries because PANformation competes with NO 2 photolysis

• RIR increases downwind for some low-reactivity compounds