Variable VOC Emissions from Point Sources in the HoustonGalveston Area and Their Impacts on Ozone Fo

1
Variable VOC Emissions from Point Sources in the
Houston-Galveston Area and Their Impacts on
Ozone Formation

• A Conceptual Model

Harvey Jeffries and David Allen University of
North Carolina (at UT) and University of Texas
http//airchem.sph.unc.edu/Research/Projects/Texas
/UT_UNC/
2
Conceptual Model for Point Source VOC Emissions

• Overview of the reported point source VOC
  inventory
• Observational evidence indicating variability in
  point source VOC emissions
• Process data indicating variability in point
  source VOC emissions
• Modeling variability in point source VOC
  emissions
• Implications for ozone formation and air quality
  modeling

3
Overview reported EI

• Most emissions are estimated, rather than measured

Neece, 2001
4
Overview reported EI

• Emissions for non-EGU point sources are generally
  assumed to be constant

Cantu, 2002
5
Overview reported EI

• Fugitive (dispersed) sources represent a large
  fraction of the inventory

Neece, 2001
6
Overview reported EI

• Total Harris Pt Src VOCs is 100 tons/day
• Highly reactive VOCs are 16 tons/day

Neece, 2001
7
Conceptual Model for Point Source VOC Emissions

• Overview of the point source VOC inventory
• Observational evidence indicating variability in
  point source VOC emissions
• Process data indicating variability in point
  source VOC emissions
• Modeling variability in point source VOC
  emissions
• Implications for ozone formation and air quality
  modeling

8
Observations indicating that point source VOC
emissions are highly variable

• Ground based monitors show significant hourly
  variability in hydrocarbon concentrations, much
  larger variations than can be explained by
  meteorology (100s to 1000 times)
• Variation is largest near large industrial
  sources, smallest in urban areas
• Ground based monitors show a long term mean
  composition that is consistent with reported VOC
  EI
• Short duration aircraft VOC and ozone formation
  measurements over large area show high spatial
  and temporal variability with most high values
  near SC
• Aircraft VOC obs show large areas of agreement
  with VOC EI in AQ model, but also show large
  disagreement at some locations and times.

9
Observations auto_GC
Sexton, Jeffries, 2001
10
Observations auto_GC
Sexton, Jeffries, 2001
11
Observations auto_GC

• Variability in hydrocarbon concentrations is much
  greater in industrial source regions than in
  residential regions

Sexton, Jeffries, 2001
12
Observations auto_GC
VOC hourly concentrations vary by factors of 100
to 1000s on a weekly basis, yet half of the time
values are quite low
Sexton, Jeffries, 2001
13
Observations indicating that point source VOC
emissions are highly variable

• Ground based monitors show significant hourly
  variability in hydrocarbon concentrations, much
  larger variations than can be explained by
  meteorology (100s to 1000 times)
• Variation is largest near large industrial
  sources, smallest in urban areas
• Ground based monitors show a long term mean
  composition that is consistent with reported VOC
  EI
• Short duration aircraft VOC and ozone formation
  measurements over large area show high spatial
  and temporal variability with most high values
  near SC
• Aircraft VOC obs show large areas of agreement
  with VOC EI in AQ model, but also show large
  disagreement at some locations and times.

14
Observations auto_GC
Average Composition HRVOCs Similar to reported
annual EIUrban Composition similar to US
39-cities
Sexton, Jeffries, 2001
15
Observations indicating that point source VOC
emissions are highly variable

• Ground based monitors show significant hourly
  variability in hydrocarbon concentrations, much
  larger variations than can be explained by
  meteorology (100s to 1000 times)
• Variation is largest near large industrial
  sources, smallest in urban areas
• Ground based monitors show a long term mean
  composition that is consistent with reported VOC
  EI
• Short duration aircraft VOC and ozone formation
  measurements over large area show high spatial
  and temporal variability with most high values
  near SC
• Aircraft VOC obs show large areas of agreement
  with VOC EI in AQ model, but also show large
  disagreement at some locations and times.

16
Observations aircraft VOCs
Aug 25
Aug 27
Aug 28
Aug 30

• Aircraft detect isolated high concentrations of
  HRVOCs and other compounds
• High concentrations are observed at different
  sites on different days

17
Observations aircraft VOCs

• Only 17 of 211 Rapid Ozone Formation aircraft
  measurements gt 40 ppb/h
• All high ROF events associated with high
  concentrations of 2-5 HRVOCs
• All high ROF events are near large industrial
  point sources of VOCs
• High ROF and HRVOCs are observed at different
  sites on different days and not at same site each
  day

Data of Klineman and Daum, 2002
Jeffries, Blanchard 2002
18
Observations aircraft VOCs

• Five factors explain 75 of the variance of the
  ROF/VOC/NOx data
• ROF, P(O3), correlated with HRVOC and precursors
  not correlated with NOx Not correlated with
  aromatics or C4 alkanes

Blanchard, Jeffries, 2002
19
Finding

• Observations from ground monitors and aircraft
  measurements indicate that emissions from
  industrial sources are variable and can lead to
  concentrations of highly reactive hydrocarbons
  (and other compounds) that exceed 100 ppbC.
• These high concentrations of highly reactive
  hydrocarbons occur, on average, on a weekly basis
  at individual monitoring sites in the industrial
  source regions, and can occur on nearly a daily
  basis if all sampling sites in the region are
  considered.
• VOC variability greatly exceeds NOx variability.

20
Conceptual Model for Point Source VOC Emissions

• Overview of the point source VOC inventory
• Observational evidence indicating variability in
  point source VOC emissions
• Process data indicating variability in point
  source VOC emissions
• Modeling variability in point source VOC
  emissions
• Implications for ozone formation and air quality
  modeling

21
Sources of variable industrial emissions what we
know

• Sources are ubiquitous (all volatile compounds,
  all locations, all times of day)
• Sources are intermittent
• Causes are different for different VOCs
• A relatively small number of major sources appear
  to be major contributors to variability
• Ozone concentrations only reach 250 ppb
• There are limits to ozone formation process in
  HGA air

22
Sources of HRVOCs

• Major sources include olefin manufacturing,
  olefin polymerization and refining

Deason, 2001
23
Flares as VOC Source

• 19 flares (roughly 4 of all flares) account for
  50 of total VOC emissions from flares

Jeffries, 2003
24
Hourly Timeseries for a Major Flare
One Year
MCCG, 2002
25
Types of Variance for a Major Flare
Symons, Webster, Pennington, 2003
26
Timeseries for a Cooling Tower

• Cooling tower emissions can exhibit the same
  general characteristics

MCCG, 2002
27
Different process units have different
variability
MCCG, 2002
28
Finding

• The overall magnitude and the variability in
  emissions of total point source VOCs and
  especially highly reactive volatile organic
  compounds are dominated by contributions from a
  small number of source accounts and a small
  number of process units at these accounts.
  These units are flares, cooling towers, various
  vents and, to a lesser extent, sources of
  fugitive emissions.

29
Conceptual Model for Point Source VOC Emissions

• Overview of the point source VOC inventory
• Observational evidence indicating variability in
  point source VOC emissions
• Process data indicating variability in point
  source VOC emissions
• Modeling variability in point source VOC
  emissions
• Implications for ozone formation and air quality
  modeling

30
Modeling variability in point source VOC
emissions

• Use a probability distribution function (PDF) to
  describe the variability in the emissions

Jeffries, 2002
31
Application of Statistical Mixture Theory Leads
to Multiple Overlapping PDFs

• Use statistical mixture theory to fit
  probability distribution functions (PDF) provide
  a description of the total source behavior over
  time.
• Use this in simulations.

Symons, Webster, Pennington, 2003
32
Application of Statistical Mixture Theory Leads
to Multiple Overlapping PDFs
Symons, Webster, Pennington, 2003
33
Finding

• The magnitude and variability in VOC and HRVOC
  emissions from point sources can be effectively
  characterized using statistical mixture theories
  and component probability distribution functions
  fitted to a variety of industrial process
  measurements.
• These can be used to vary the reported modeling
  point source emissions inventory, producing a
  large number of emissions snapshots.

34
Conceptual Model for Point Source VOC Emissions

• Overview of the point source VOC inventory
• Observational evidence indicating variability in
  point source VOC emissions
• Process data indicating variability in point
  source VOC emissions
• Modeling variability in point source VOC
  emissions
• Implications for ozone formation and air quality
  modeling

35
Conceptual issue
In most of US, industrial emissions are
relatively constant or are small enough that
meteorology is cause of worst conditions In
HGA, both meteorology and emissions are cause of
worst conditions
Jeffries, 2002
36
Meteorology and Ozone Exceedences
More than half the time, winds are conducive for
ozone exceedence. Only one day in eight of
these days actually has an exceedence.
Blanchard, Jeffries, 2002
37
Air Quality Modeling for SIP Development

• In HGA, historical episodes are merely emission
  snapshots and are not likely to be representative
  of future conditions.
• In doing AQ Modeling we need to separate
  emissions that change from episode to episode
  from those that remain nearly constant
• In evaluating SIP effectiveness, need to consider
  an album of many industrial emission snapshots

38
Use PDFs to create an album of emission snapshots
Allen, 2002
39
What photochemical modeling tools do we use?

• Currently impossible (due to resource and
  computational constraints) to consider enough
  emission snapshots with the full CAMx
  photochemical model
• Episodic emissions are most important in
  industrial source region
• Consider many emission snapshots using a closely
  coupled but simpler version of the full CAMx
  model that focuses on a smaller spatial area
  (industrial source regions)
• Examine most important snapshots with the full
  CAMx model

40
Planned Approach
41
Planned Approach
42
Planned Approach
Aggregate cells
43
Planned Approach
Use Process Analysis on CAMx basecase
Calculate conditions for Aggregate Cell Model
basecase
Jeffries, Kimura, Vizeute, 2003
44
A Simple Prototype Example
Say we have an aggregate cell model that is
process-related to the full 3-D air quality
model. That is, the operating conditions for the
aggregated cell model are the same as those
that occurred in a ozone conducive area in the
full 3-D air quality model. This is done using
Process Analysis on both models.
Allen, 2002
45
Preliminary Results
This is an example of the right hand side of
VEIMA Plan
Allen, 2002
46
Preliminary results
Allen, 2002
47
Finding

• If ozone maximum concentrations are to be kept
  below a threshold value, such as those set by the
  NAAQS, then concentrations of reactive
  hydrocarbons must also be kept below a threshold
  value. Furthermore, preliminary results suggest
  that the threshold concentration is only
  moderately influenced by the magnitude of oxides
  of nitrogen concentrations. This is to be
  explored in a more fully coupled modeling system.

48
Summary

• Point source VOC emissions are highly variable
• Variability is due to a ubiquitous group of
  source types probably flares, cooling towers,
  process vents and, to a lesser extent, fugitives
• Variability (not nearly constant emissions) leads
  to observations of HRVOC concentrations in excess
  of 100 ppbC on a weekly basis
• HRVOC concentrations in excess of 100 ppbC can,
  under commonly observed conditions, lead to
  extensive ozone formation
• Need to model an album of emission snapshots to
  adequately describe ozone formation

49
Summary

• Point source VOC emissions are highly variable
• Variability is due to a ubiquitous group of
  source types probably flares, cooling towers,
  process vents and, to a lesser extent, fugitives
• Variability (not nearly constant emissions) leads
  to observations of HRVOC concentrations in excess
  of 100 ppbC on a weekly basis
• HRVOC concentrations in excess of 100 ppbC can,
  under commonly observed conditions, lead to
  extensive ozone formation
• Need to model an album of emission snapshots to
  adequately describe ozone formation

50
Summary

• Point source VOC emissions are highly variable
• Variability is due to a ubiquitous group of
  source types probably flares, cooling towers,
  process vents and, to a lesser extent, fugitives
• Variability (not nearly constant emissions) leads
  to observations of HRVOC concentrations in excess
  of 100 ppbC on a weekly basis
• HRVOC concentrations in excess of 100 ppbC can,
  under commonly observed conditions, lead to
  extensive ozone formation
• Need to model an album of emission snapshots to
  adequately describe ozone formation

51
Summary

• Point source VOC emissions are highly variable
• Variability is due to a ubiquitous group of
  source types probably flares, cooling towers,
  process vents and, to a lesser extent, fugitives
• Variability (not nearly constant emissions) leads
  to observations of HRVOC concentrations in excess
  of 100 ppbC on a weekly basis
• HRVOC concentrations in excess of 100 ppbC can,
  under commonly observed conditions, lead to
  extensive ozone formation
• Need to model an album of emission snapshots to
  adequately describe ozone formation

52
Summary

• Point source VOC emissions are highly variable
• Variability is due to a ubiquitous group of
  source types probably flares, cooling towers,
  process vents and, to a lesser extent, fugitives
• Variability (not nearly constant emissions) leads
  to observations of HRVOC concentrations in excess
  of 100 ppbC on a weekly basis
• HRVOC concentrations in excess of 100 ppbC can,
  under commonly observed conditions, lead to
  extensive ozone formation
• Need to model an album of emission snapshots to
  adequately describe ozone formation

53
Remember

• Making policy decisions based on mathematical
  models is like marriage at some point you
  decide that you can live with certain flaws and
  trade-offs.
• students answer to a test in one of Prof.
  Jeffries Class