SOURCE APPORTIONMENT MODELS FOR AIR QUALITY MANAGEMENT

1
SOURCE APPORTIONMENT MODELS FOR AIR QUALITY
MANAGEMENT

• Arpa Wangkiat
• Rangsit University
• presented at the
• Source Apportionment for Particulate Matter
  WorkshopPollution Control Department,
  ThailandMarch 4, 2008

2

• State the fundamental structure and assumptions
  of receptor and source models.
• Identify the usefulness and limitations of
  receptor and source models.

3
Basic Requirements to Assess Air Pollution
Source/Receptor Relationships

• Understand the chemical and physical
  characteristics of air pollutants.
• Understand transport from sources to receptors.
• Estimate the contribution of each source to air
  pollutant concentrations measured at receptors.

4
Definitions

• Model
• Based on a scientific understanding of physical
  interactions.
• A set of mathematical relationships between
  variables.
• Values are provided to some variables to
  calculate others.
• Input values are obtained from measurements.
• Each model is an imperfect representation of
  reality.

5
Definitions (continued)

• Measurement Process
• Observables measured.
• Range of values of observables.
• Frequency and duration of sampling.
• Spatial density of samples.
• Validity, precision, and accuracy of
  measurements.
• Each measurement process is an imperfect
  representation of reality.

6
Source Apportionment Models

• Conceptual model. Describes the relevant
  physical and chemical processes in an area. No
  mathematics
• Emissions model. Estimates temporal and spatial
  emission rates based on activity level, emission
  rate per unit of activity, and meteorology
• Meteorological model. Describes transport,
  dispersion, vertical mixing, and moisture in time
  and space

7
Source Apportionment Models

• Air quality model. Estimates concentrations at
  receptors based on emissions, transport, and
  transformation
• Chemical model. Describes transformation of
  gases to particles and equilibrium between gas
  and particle phases
• Chemical mass balance receptor model. Infers
  source contributions from chemical fingerprints
  of source emissions and receptor concentrations
• Multivariate receptor models. Infer source
  profiles from ambient data

8
(No Transcript)
9
Source Models vs. Receptor Models
The source model uses source emissions as inputs
and calculates ambient concentrations. The
receptor model uses ambient concentrations as
inputs and calculates source contributions.
(From Watson, 1979.)
10
Source and Receptor Models
Source Model Independent variables are measured
and dependent variables are calculated. Y
f (X1 Xn)
(Calculated) (Measured) Receptor
Model Dependent variables and some independent
variables are measured and the remaining
independent variables are calculated.
Y f (X1 Xi , Xi1
Xn) (Measured) (Calculated)
(Measured)
11
Model Equations
Source Model Receptor Model where Ci Concen
tration of chemical species i Djk D(u(t),
Xj ) dt where u(t) winds and Xj
source/receptor distance Ejk Emission
rate Fij Fractional quantity of species i in
source j Sjk Contribution from the jth source
to kth receptor sample
12
Chemical Mass Balance

• Equation
• Input
• Ambient concentrations (Cj)and uncertainties
  (sCj),source composition (Fij),and
  uncertainties (sFij).
• Output
• Source contributions (Sj)and uncertainties
  (sSj).
• Measurements
• Size-classified mass, elements, ions, and carbon
  concentrations on both ambient and source samples.

13
Receptor models
Source models
Advantages They are based on observed data, and
hence independence of source data
DisadvantagesModel results are highly dependent
on source emission inventory
Source data
Garivat,1999
14
Receptor models
Source models
Advantages The use of meteorological data is
optional
DisadvantagesThe use of meteorological data is
necessary and affects the result considerably
Meteorological data
15
Receptor models
Source models
Advantages Do not affected the result
DisadvantagesThey are not accurate when
inventories are fitted with pollution control
devices
Pollution control devices
16
Receptor models
Source models
Advantages They take into account unidentified
and fugitive emissions
DisadvantagesThey cannot identify fugitive
emissions (unknown sources)
Fugitive Emissions
17
Receptor models
Source models
Advantages They are cost effective and very
useful for developing local control strategies
DisadvantagesCollection of meteorological and
source data is complex and costly control
strategies
Utility
18
Applicability
Receptor models
Source models
Disadvantages They are site specific
AdvantagesThey are universal and can be applied
to many areas
19
Observation data
Receptor models
Source models
Disadvantages Require large data sets on physical
and chemical analysis
AdvantagesObservation data needed only for
validation of models
20
Receptor models
Source models
Disadvantages They can not identify any point
source as well as can not deal with source
co-linearity very efficiently
AdvantagesThey are very suitable for specific
sources. Source specific data do not pose the
problems of co-linearity
Specific source
21
Source Model Limitations

• Emission rates and transport are hard to
  estimate.
• Deposition, boundary conditions, and initial
  conditions must be estimated.
• Changes occur from gas to particle.

22
Receptor Model Limitations

• Many emitters have similar profiles.
• Profiles change between source and receptor.
• Cannot predict consequences of emissions
  reductions.

23
Protocol for Reconciling Differences Among
Receptor and Dispersion Models

• 1. Compare CMB and DM results.
• 2. Verify input data in both models.
• 3. Recompare results.
• 4. Refine CMB model inputs.
• 5. Recompare results.
• 6. Refine dispersion model inputs.
• 7. Recompare.
• 8. if it is clearly evident that the
  dispersion model is not valid, the CMB estimates
  should be used as the basis for control strategy
  development. However, if the disparity is not
  clearly attributable to either model alone, the
  dispersion model should be used for control
  strategy development.

24
Source model Computer simulation of atmosphere
processes
Simulated PM ambient concentration
PM ambient data
Source data
Receptor model Statistical analysis of data
Apportionment of source categories
EPA SIP major features of source and receptor
model,1999
25
Conclusions About Air Quality Models

• Every model is a simplification of reality.
• Model results can be no better than the input
  data supplied to the model.
• Concurrent sampling of gas- and particle-phase
  organic compounds is needed for carbon
  apportionment
• The application of a single model will supply an
  answer. The application of several models will
  supply doubt.