Air Quality Modeling

Dr. Wesam Al Madhoun

5/26/2018 1
 Overview

5/26/2018 2
 Overview

• Air Quality Models are mathematical formulations that

include parameters that affect pollutant concentrations.

• They are used to

– Evaluate compliance with NAAQS and other
regulatory requirements
– Determine extent of emission reductions required
– Evaluate sources in permit applications

5/26/2018 3
 Types of AQ Models

Meteorological
Model

Emission Chemical
Model Model
Temporal and spatial emission rates
Topography
Chemical Transformation
Pollutant Transport
Equilibrium between Particles and gases
Vertical Mixing
Source
Receptor
Dispersion
Model
Model
5/26/2018 4
• Emission 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
• Chemical Model
– Describes transformation of directly emitted
particles and gases to secondary particles and
gases; also estimates the equilibrium between gas
and particles for volatile species
5/26/2018 5
• Source Dispersion Model
– Uses the outputs from the previous models to
estimate concentrations measured at receptors;
includes mathematical simulations of transport,
dispersion, vertical mixing, deposition and chemical
models to represent transformation.

• Receptor Model
– Infers contributions from different primary source
emissions or precursors from multivariate
measurements taken at one ore more receptor sites.

5/26/2018 6
 Classifications of AQ Models
• Developed for a number of pollutant types
and time periods
– Short-term models – for a few hours to a few
days; worst case episode conditions
– Long-term models – to predict seasonal or annual
average concentrations; health effects due to
exposure

• Classified by
– Non-reactive models – pollutants such as SO2 and
CO
– Reactive models – pollutants such as O3, NO2, etc.
5/26/2018 7
 AQ Models
• Classified by coordinate system used
– Grid-based
• Region divided into an array of cells
• Used to determine compliance with NAAQS
– Trajectory
• Follow plume as it moves downwind
• Classified by level of sophistication
– Screening: simple estimation use preset, worst-
case meteorological conditions to provide
conservative estimates.
– Refined: more detailed treatment of physical and
chemical atmospheric processes; require more
detailed and precise input data.
5/26/2018 8
 Fixed-Box Models
• The city of interest is assumed to be rectangular.
• The goal is to compute the air pollutant concentration in this
city using the general material balance equation.

Fig. 6.1 De
Nevers

9
 Fixed-Box Models
Assumptions:
1. Rectangular city. W and L are the dimensions, with one side
parallel to the wind direction.

2. Complete mixing of pollutants up to the mixing height H. No

mixing above this height.

3. The pollutant concentration is uniform in the whole volume

of air over the city (concentrations at the upwind and
downwind edges of the city are the same).

4. The wind blows in the x direction with velocity u , which is

constant and independent of time, location, & elevation.
10
… Assumptions
5. The concentration of pollutant in the air entering the city is
constant and is equal to b (for background concentration).

6. The air pollutant emission rate of the city is Q (g/s). The

emission rate per unit area is q = Q/A (g/s.m2). A is the
area of the city (W x L). This emission rate is assumed
constant.

7. No destruction rate (pollutant is sufficiently long-lived)

11
Now, back to the general material balance eqn

Accumulation rate = (all flow rates in)

– (all flow rates out)
+ (creation rate)
– (destruction rate)

→Destruction rate = zero (from assumptions)

→Accumulation rate = zero (since flows are independent of time and

therefore steady state case since nothing is changing with time)

→ Q can be considered as a creation rate or as a flow into the

box through its lower face. Let’s say a flow through lower
face.
12
the general material balance eqn becomes:
0 = (all flow rates in) – (all flow rates out)
0=uWHb+qWL–uWHc
qL
c b
uH
Where c is the concentration in the entire city

• The equation indicates that the upwind concentration is added

to the concentrations produced by the city.

• To find the worst case, you will need to know the wind speed,
wind direction, mixing height, and upwind (background)
concentration that corresponds to this worst case. 13
 Example 6.1
A city has the following description: W = 5 km, L = 15 km, u
= 3 m/s, H = 1000 m. The upwind, or background,
concentration of CO is b = 5 μg/m3. The emission rate per
unit are is q = 4 x 10-6 g/s.m2. what is the concentration c
of CO over the city?
qL
c b
uH
 g 
 4 10
6
2 
15000 m
5 μg  s.m 
c 3 
m 3 m/s 1000 m
= 25 μg/m3
14
 Comments on the simple fixed-box model

• The fixed-box models does not distinguish between area

sources and point sources.

• Both sources are combined in the q value. We know that

raising the release point of the pollutant will decrease the
ground-level concentration.

• So far, the fixed-box model predicted concentrations for

only one specific meteorological condition. We know that
meteorological conditions vary over the year.

15
 Modifications to improve the fixed-box model
1) Hanna (1971) suggested a modification that allows one to
divide the city into subareas and apply a different value of q to
each. (since variation of q from place to place can be
obtained; q is low in suburbs and much higher in industrial
areas).

2) Changes in meteorological conditions can be taken into

account by

a. determine the frequency distribution of various values of

wind direction, u, and of H
b. Compute the concentration for each value using the fixed-
box model
16
 …Modifications to improve the fixed-box model

c. Multiply the concentrations obtained in step b by the

frequency and sum to find the annual average

 Annual   concentrat ion  frequency of 

    
 Average     for that  occurrence of that 
 Concentrat ion  over all meteorologies  meteorolog y  meteorolog y 
    

17
 Example 6.2

For the city in example 6.1, the meteorological conditions

described (u = 3 m/s, H = 1000 m) occur 40 percent of the
time. For the remaining 60 percent, the wind blows at right
angles to the direction shown in Fig. 6.1 at velocity 6 m/s
and the same mixing height. What is the annual average
concentration of carbon monoxide in this city?

First we need to compute the concentration resulting from each

meteorological condition and then compute the weighted
average.
For u = 3 m/s and H = 1000 m → c = 25 μg/m3

18
 Note that L is now
…example 6.2 cont. 5km, not 15km

 g 
 4 10
6
2 
5000 m 
5 μg  s.m 
For u = 6 m/s and H = 1000 m → c  
m3 6 m/s 1000 m 
g
c  8.33 3
m

 Annual   concentrat ion  frequency of 

    
 Average     for that  occurrence of that 
 Concentrat ion  over all meteorologies  meteorolog y  meteorolog y 
    
 Annual 
  μg μg μg
 Average   25 3  0.4  8.33 3  0.6  15 3
 Concentrat ion  m m m
 

19
 Gaussian Dispersion Models
• Most widely used , Lakes Environnemental
Software: http://www.weblakes.com

• Plume spread and shape vary in response to meteorological

conditions
Fig. 6.3 De
Nevers

20
 Z X

u Y

Fig 7.11
 Model Assumptions
• Gaussian dispersion modeling based on a number of
assumptions including
– Steady-state conditions (constant source emission
strength)
– Wind speed, direction and diffusion characteristics
of the plume are constant.
– Conservation of mass, i.e. no chemical
transformations take place
– Wind speeds are >1 m/sec.
– Limited to predicting concentrations > 50 m
downwind
5/26/2018 22
 The general equation to calculate the steady state concentration of an air
contaminant in the ambient air resulting from a point source is given by:

 1  y 2 z  H 2 
C  x, y , z  
Q
exp   2  
2 y z u 
  y   2 
2 z 
Where;
c(x,y,z) = mean concentration of diffusing substance at a point (x,y,z) [kg/m3]
x = downwind distance [m],
y = crosswind distance [m],
z = vertical distance above ground [m],
Q = contaminant emission rate [mass/s],
 y = lateral dispersion coefficient function [m],
 z = vertical dispersion coefficient function [m],
ῡ = mean wind velocity in downwind direction [m/s],
H = effective stack height [m].

5/26/2018 23
 What are the A to F categories?
• A to F are levels of atmospheric stability (table 6.1).

• Explanation:
– For a clear & hot summer morning with low wind speed, the sun heats
the ground and the ground heats the air near it. Therefore air rises and
mixes pollutants well.
►► Unstable atmosphere and large σy & σz values

– On a cloudless winter night, ground cools by radiation to outer space

and therefore cools the air near it. Hence, air forms an inversion layer.
►► Stable atmosphere and inhibiting the dispersion of pollutants
and therefore small σy & σz values

2) Diffusion Models 24
Stability Classes

• Table 3-1 Wark,

Warner & Davis
• Table 6-1 de Nevers
 Dispersion Coefficients: Horizontal

Fig 7.12

5/26/2018 26
 Dispersion Coefficients: Vertical

Fig 7.13

5/26/2018 27
 Gaussian Dispersion Equation

If the emission source is at ground level with no

effective plume rise then
  2 2 
C x, y, z  
Q 1 y z 
exp   2 
y  z u  2   y  z 
2

Plume Rise
• H is the sum of the physical stack height and
plume rise.
H  h plume rise  hactual stack
5/26/2018 28
 Plume Rise
h 
Vs D 
1.5  2.68 10 PD
3 Ts  Ta  

u  Ts 
Δh = plum rise in m
This equation is only correct for the Vs = stack exit velocity in m/s
D = stack diameter in m
dimensions shown.
u = wind speed in m/s
P = pressure in millibars
Ts = stack gas temperature in K
Ta = atmospheric temperate in K

Correction is needed for stability classes other than C:

→ For A and B classes: multiply the result by 1.1 or1.2
→ For D, E, and F classes: multiply the result by 0.8 or 0.9

2) Diffusion Models 29
 Example 6.3

Q = 20 g/s of SO2 at Height H

u = 3 m/s,

At a distance of 1 km, σy = 30 m, σz = 20 m (given)

Required: (at x = 1 km)

a) SO2 concentration at the center line of the plume

b) SO2 concentration at a point 60 m to the side of and 20 m

below the centerline

5/26/2018 30
 … solution of example 6.4
Q   y2 ( z  H ) 2 
c exp   2  
2 u y z   2 y 2 z 
2

a) At the centerline
y  0, z - H  0  e 0  1
Q 20 (g/s)
c   .00177 g/m 3  1770 g/m 3
2u y z 2 (3 m/s)(30 m)(20 m)
b) At a point 60m to the side and 20 m below the CL
y  60 m, z - H  - 20 m
20 (g/s)   60 2 (20) 2 
c exp    
2 (3 m/s)(30 m)(20 m)   2(30 ) 2
2( 20) 2

 (1770 g/m 3 )(0.0818)  145 g/m 3
5/26/2018 31
 Chemical Mass Balance Model
• A receptor model for assessing source apportionment using
ambient data and source profile data.
• Available at EPA Support Center for Regulatory Air Models -
http://www.epa.gov/scram001/tt23.htm
PM10 emissions from permitted sources in
Alachua County (tons) (ACQ,2002)

2000 Values
1. GRU Deerhaven 144.2
2. Florida Rock cement plant 34.35
9
3. Florida Power UF cogen. plant 3.19
11 10 1997 Values
4. VA Medical Center incinerator 0.2
1 8 5. UF Vet. School incinerator 0.2
14 13
2 6 6. GRU Kelly 1.9
7. Bear Archery 9.5
3,4,5,12
8. VE Whitehurst asphalt plant 4.9
7 9. White Construction asphalt plant 0.7
10. Hipp Construction asphalt plant 0.3
11. Driltech equipment manufacturing 0.2
Receptor Sites
12. University of Florida
13. Gainesville Regional Airport
5/26/2018 14. Gainesville Regional Utilities (MillHopper)
32
 Principles
• Mass at a receptor site is a linear combination of the
mass contributed from each of a number of individual
sources;
• Mass and chemical compositions of source emissions
are conserved from the time of emission to the time the
sample is taken.

Cij = Σ(aik×Skj)
• Cij is the concentration of species ith in the sample jth
measured at the receptor site:
• aik is the mass fraction of the species in the emission from
source kth, and
• Skj is the total mass contribution from source kth in the jth
sample at the receptor site.
5/26/2018 33
 Example
• Total Pb concentration (ng/m3) measured at the site: a linear
sum of contributions from independent source types such as
motor vehicles, incinerators, smelters, etc
PbT = Pbauto + Pb incin. + Pbsmelter +…
• Next consider further the concentration of airborne lead
contributed by a specific source. For example, from
automobiles in ng/m3, Pbauto, is the product of two cofactors:
the mass fraction (ng/mg) of lead in automotive particulate
emissions, aPb, auto, and the total mass concentration (mg/m3)
of automotive emission to the atmosphere, Sauto
• Pbauto = aauto (ng/mg) × Sauto (mg/m3air)

5/26/2018 34
 Assumptions
• Composition of source emissions is constant over
period of time,
• Chemicals do not react with each other,
• All sources have been identified and have had their
emission characterized, including linearly independent
of each other,
• The number of source category (j) is less than or equal
to the number of chemical species (i) for a unique
solution to these equations, and
• The measurement uncertainties are random,
uncorrelated, and normally distributed (EPA, 1990).

5/26/2018 35