Environmental Modelling
542
Mark Lay
[Link]/wp-content/ November/[Link]
Environmental modelling
• Models mathematical
– Understanding pollution causes and
consequences
– Predict impact of new or modified pollution
sources
– Explore regulatory and control effects before
they are implemented
Modelling approaches
• Statistical
– Use probability and statistics for analysis
– Used for near term predictions
• How often will ozone levels exceed standards?
• Given certain weather conditions what are the expected
pollution levels
– Do not need more than a superficial knowledge of
chemistry and physics of pollution in air, soil, or water
– Large data requirements
– Incapable of predicting events under significantly
changed conditions
Modelling approaches
• Deterministic
– Based on material balance
– Describes chain of events that link source emissions
to observed concentrations away from source
– Requires information on water, soil or air conditions,
and transport and transformation processes
• Need sound understanding of chemical, biological and
physical processes
– Air pollution models can be broken down into
Gaussian plume and transport and transformation
models
Modelling approaches
• Gaussian plume model
– Predict relationship between source
emissions and downwind concentrations
– Easy to use
– Requires a leap of faith in determining model
from first principles
– Uses approximate analytical solutions
– Limited in complex terrain
– Cannot accommodate any chemistry beyond
first order decay
Modelling approaches
• Transport and transformation models
– Predict species concentration taking into
account all important emission sources
– Uses numerical methods to solve equations
– Widely used for understanding and controlling
urban and regional photochemical smog
problems
– Produce time dependent predictions for
periods of a few days
Modelling approaches
• Transport and transformation
– Two types of model: trajectory and airshed
• Trajectory
– Simulates pollutant concentrations in a parcel
of air as it is pushed by wind through a basin
or region
– Requires good wind data to predict movement
of the parcel
– Pollution emissions into parcel are calculated
from all sources
Modelling approaches
• Airshed
– A region is discretized into subregions
– Mass balance equations are solved for each
subregion
– Predicts pollutant concentrations as a function of
position and time
– May include
• Emission point sources
• Up to 100 or more chemical reactions
• Solar simulator to predict photolysis
• Windfield generator
Modelling approaches
• Airshed
– May include (continued)
• Transport simulator that accounts for wind, atmospheric
stability, dispersion and topography
• Pollution deposition
• Algorithm that combines these elements into mathematical
equations that solved numerically
– Computationally intensive
• Divided up into >10,000 subdomains
• Tracks up to 20-50 chemical species in each subdomain
requiring separate mass balance equations
• May be done over 50,000 time steps
• Requires >2.5x1010 calculation steps
Gaussian plume modelling
• Most effluent discharged vertically into
atmosphere through a stack or duct
u (m/s)
Wind vs (m/s)
E (g/s) plume
Ts(K) Ta(K) P (kPa)
h (m)
and factory
d (m)
Gaussian plume modelling
• Plume will disperse outwards downwind,
concentration forming a Gaussian curve in
the vertical and horizontal planes
perpendicular to the wind flow
Gaussian plume modelling
• The plume height affected by gas stream’s
upward inertia and buoyancy
– inertia related to exit gas velocity and mass
– buoyancy related to exit gas mass relative to
surrounding air
H (m)
Gaussian plume modelling
• Effective stack height
v sd 2 Ts Ta
H 1.5 2.68 10 P d
u T s
H
H (m) = H + h
h (m)
Gaussian plume modelling
• Wind velocity at effective stack height may
be different to measured wind velocity
p
H
u2 u1 Valid up to elevations of several hundred metres
h
u2
u1
Gaussian plume modelling
Exponent p values for rural and urban areas
Stability Class Rural Urban
A (most unstable) 0.07 0.15
B 0.07 0.15
C 0.1 0.2
D 0.15 0.25
E 0.35 0.3
F (most stable) 0.55 0.3
Gaussian plume modelling
Key to stability classes
Surface wind Day Night
speed Incoming Solar Radiation
(at 10 m)
(m/s) Strong Moderate Slight Thinly No cloud to
overcast to thinly
cloudly overcast
<2 A A-B B
2-3 A-B B C E F
3-5 B B-C C D E
5-6 C C-D D D D
>6 C D D D D
Incoming solar radiation drops with increasing cloud cover.
Strong is clear skies with solar altitude greater that 60 degrees, slight is clear
skies with solar altitude between 15 to 35 degrees.
Night is one hour before sunset to one hour after sunrise.
Gaussian plume modelling
• Plume dispersion downwind is dependent
on stability, distance from emission source
and windspeed A
F
Gaussian plume modelling
• Downwind plume concentration is given by
2
2
E 1 y 1z H
C x, y , z, H exp exp
s y sz u
2 sy 2 sz
Davis and Masten 2004, Principles of Environmental Engineering and Science
Wind direction, E = emission rate of pollutant (g/s)
z(m)
velocity u (m/s)
sy, sz = plume standard deviations (m)
H x(km)
sy a x 0.894
sz c x d f
y(m)
Horizontal dispersion
coefficients. Ref: Davis
and Masten 2004,
Principles of
Environmental
Engineering and
Science
Vertical dispersion
coefficients. Ref: Davis
and Masten 2004,
Principles of
Environmental
Engineering and
Science
Gaussian plume modelling
Values of a, c, d and f for calculating sy and sz
x < 1km x 1km
Stability
class a c d f c d f
A 213 440.8 1.941 9.27 459.7 2.094 -9.6
B 156 100.6 1.149 3.3 108.2 1.098 2
C 104 61 0.911 0 61 0.911 0
D 68 33.2 0.725 -1.7 44.5 0.516 -13
E 50.5 22.8 0.678 1.3 55.4 0.305 -34
F 34 14.35 0.74 -0.35 62.6 0.18 -48.6
Gaussian plume modelling
• Does not take into account inversions
• Inversions occur when a layer of cold air
near surface is trapped by a layer of warm
air
– Can occur during night time or early mornings
– Stops convection, air becomes still
– Responsible for fog, bad pollution problems in
Los Angles, Shanghai, etc
Inversions
Smoke rising in
Lochcarron is
stopped by an
overlying layer of
warmer air. Ref:
Wikipedia
Inversions
Inversion over
Shanghai city
Ref:
Wikipedia
Gaussian plume modelling
• What if an inversion is present in the
atmosphere and the plume hits it?
Use standard equation Use another
xL 2xL
Inversion
L(m)
Gaussian plume modelling
• Need to find xL
– Find sz at point where plume hits inversion
sz 0.47L H Only use to find XL, do not use to
calculate concentration
– Find xL by rearranging 1
sz f d
sz c x d f x xL
c
– To calculate centre-line concentration at and
beyond 2xL use
E
C
2
0. 5
sy u L
Gaussian plume modelling
• Limitations
– Not good for when reactions are occurring in
plume
– Assumes windspeed is constant with height,
width and length of plume
– Assumes turbulence is also constant
– Not good over complex terrain (hills or
buildings)
– Does not account for thermals, crosswinds
and complex climatological conditions
Non Gaussian plumes
• Looping plume, occurs when there are
strong convective currents
Non Gaussian plumes
• Lofting plume, high buoyancy
Non Gaussian plumes
• Fanning and fumigation, occurs during
inversions, overcast days
6 mile long toxic plume of Doe Run's Herculaneum Smelter (Robyn Warden, Ref:
[Link]