Modeling Air Quality in Houston, Texas

1
Modeling Air Quality in Houston, Texas

• Byeong-Uk Kim
• November 17, 2004
• Department of Environmental Sciences and
  Engineering
• University of North Carolina at Chapel Hill

2
Content

• Background
• Uniqueness of Houston
• Air Quality Modeling for Houston
• Dissertation

3
Background

• Is ozone bad or good?
• Ozone formation science
• Non-attainment area and State Implementation Plan
  (SIP)
• Difficulties in developing control strategy
• Non-linearity
• Ozone transport
• Photochemical Air Quality Modeling (PAQM) system

4
Is ozone bad or good?
http//www.epa.gov/oar/oaqps/gooduphigh/
5
Ozone formation science
Jeffries, 1993
6
Ozone formation science

• Where do we get the precursors?
• NOx (NONO2)
• Automobiles
• Power plants
• VOCs (Volatile Organic Compounds)
• Automobiles
• Refineries
• Trees
• Other area sources

7
Non-attainment and SIP

• Non-attainment area
• NAAQS for ozone daily maximum one hour average
  concentration lt 0.12 ppm
• fourth highest reading in 3 years gt 125 ppb is a
  violation at that monitor
• highest violation in 3 years is the design
  value for control
• State Implementation Plan (SIP)
• Control strategy development
• Attainment demonstration

NAAQS National Ambient Air Quality Standards
8
Difficulties in developing control strategy
Non-linearity
Philadelphia
Chicago
Blanchard and Reynolds, 2002
9
Difficulties in developing control strategy
Ozone transport
100km Travel in 6 hrs 100km between New York
City- Philadelphia
10
Formulation of PAQM
Jeffries, 1993
11
Operation of PAQM
12
Operation of PAQM
13
Operation of PAQM
TCEQ developed 9114 new emission point speciation
profiles
From Byun, UH
14
Reported EI

• Normal Non-EGU VOC EI with Special EI additions
• hg_02km.tx_negu_si4a Total Point Source CB-IV HC
  Emissions, 08/25/2000
• Base inventory of 240 T/D
• Nearly constant emissions

TCEQ Cantu, 2002
15
SIP modeling
Base case explain current ozone with current
inventory
Future case predict ozone with growth and
existing regulations.
Future controlled case predict ozone with
growth, existing regulations, and newly proposed
regulations.
All three cases use the same episodic meteorology.
From Byun, UH
16
Uniqueness of Houston

• Transient High Ozone Events (THOEs)
• Large variability of VOC emissions
• Different Modeling Approach Needed
• Complex meteorological conditions

17
Transient High Ozone Event in Houston
Jeffries, 2004
18
gt10,000 lbs/hr ethylene release at La Porte,
(6700 lbs between 1100 AM and 1125 AM)
3/27/2002
19
Short term ozone enhancements of up to 100 ppb
20
Large Variability of VOC Emissions

• Permitted emission components
• Nearly Constant
• Routinely Variable
• Allowable Episodic
• Off-permit event emission releases
• Start-up, shut-down, maintenance operations,
  accidents, emergencies
• Subject to RQ value

Webster, 2003
21
Different Modeling Approach Needed

• Typical conceptual model

• New conceptual model

Jeffries, 2003
22
Complex Meteorological Conditions
http//www.islandnet.com/see/weather/elements/sea
brz.htm
23
Complex Meteorological Conditions
GOES Geostationary Operational Environmental
Satellites
24
Complex Meteorological Conditions
August 31, 2000 Estimated Mixing Height by Kv
profile
1000 AM
0200 PM
0600 PM
Land/sea breeze does not happen clearly all the
time. The synoptic weather pattern plays a key
role.
25
Air Quality Modeling for Houston

• Modeling Domain
• General Description
• Base case results
• Future case results

26
Air Quality Modeling for Houston
27
Air Quality Modeling for Houston

• UNC Beowulf servers
• More than 150 nodes, MPI support
• CAMx (Comprehensive Air quality Model with
  eXtensions Ver. 4.x)
• Serial/OpenMP
• Modeling period
• August 22, 2000 September 6, 2000
• Run statistics
• About 4 hours clock time for an episodic day on
  an Intel 2.8GHz CPU node
• About 2 GB size of outputs per episodic day

28
Base case results
Imputed base
Not imputed (max 112 ppb) Event
29
Base case results
30
Future case results
Without Event Imputed future
With Event Imputed future
31
Dissertation

• Does this model show or have all necessary
  components to produce the phenomena that we can
  expect from the current best perceptual/conceptual
  model?
• Can this model distinguish what precursor to
  control for ozone reduction?
• Does it estimate the control requirement
  unambiguously?
• What are the possible biases in the prediction
  and its impact on the policy choice?
• The answers to these questions may vary
  spatially over the modeling domain and temporally
  during the modeling period.

32
Why do we care those questions?

• The current SIP framework requires the use of
  PAQM for the control strategy developments.
• Once developed, the control strategies will
  influence our daily life. Some of them are also
  very expensive to implement.
• Highway speed limit change
• Restriction of the construction equipments
• Installation of NOx control device
• It may consume all important material for the
  device that is available in the USA to make the
  demanded amount of devices for a region.
• What if these are not the primary reason for the
  ozone violation in your area?

33
Lots of analysis and tools

• Model performance analysis
• Potential reliability of the modeling results for
  the policy use
• Area of influence analysis
• Source-receptor relationships
• Uncertainty/Sensitivity analysis
• Identification of the important
  process/parameter/input
• Process analysis
• Models internal ozone concentration by process
  contributions