0 Up votes0 Down votes

34 views48 pagesDispersion Modeling

Jan 01, 2013

© Attribution Non-Commercial (BY-NC)

PDF, TXT or read online from Scribd

Dispersion Modeling

Attribution Non-Commercial (BY-NC)

34 views

Dispersion Modeling

Attribution Non-Commercial (BY-NC)

- Environmentalists' Comments on EPA's Proposed Oil and Gas Emission Rules
- Qt-i(Prob Dist III)(6)
- Combustion
- Lecture 1
- UBICC IKE07 Performance Analysis for Skewed Data 191 191
- mstatsample (1)
- - Fuel Combustion
- Sample Mean Distribution
- six sigmaaa
- pmc
- eode-air-quality-impact.pdf
- Chapter 19 Probability Distribution
- Stigler stat Ch. 2`
- 34
- Prob-8
- TR-Exl
- W6INSE6220.pdf
- Unfinished
- The Standard Normal Distribution.docx
- probabilidad

You are on page 1of 48

Inverse Problem: Estimating Emissions

Numerical Results

Mathematical Modelling of

Atmospheric Contaminant Dispersion

John Stockie

Department of Mathematics

Simon Fraser University

http://www.math.sfu.ca/stockie

June 22, 2010

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 1/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Acknowledgments

Enkeleida Lushi

MSc student intern

Funded jointly by MITACS and Teck-Cominco

Evgeniy Lebed

Summer undergraduate research assistant

Funded by NSERC

Ed Kniel

Supt, Environmental Management

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 2/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Motivation

Teck-Cominco operates one of the worlds largest integrated

lead-zinc smelting operations in Trail BC on the Columbia

River.

Its a major driver of BC

economy but also one of

Canadas biggest polluters

(lead, zinc, arsenic, cadmium,

mercury, SO

2

, . . . ).

Annual emissions reporting is

required under the National

Pollutant Release Inventory

(http://www.ec.gc.ca).

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 3/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Historical Sidebar

Trail was founded in 1890s as a

mine supply point. A small smelter

was built.

In 1906, Cominco was formed.

In 1941, WA state was awarded

damages from Cominco for

trans-border pollution one of the

most-cited international law cases.

From 1917 to 1940, emissions of

SO

2

were 100700 T/day.

Currently down to 22 T/yr.

Today Teck-Cominco prides itself

on its clean Trail operations.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 4/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

The Problem

Tecks Trail operation is unable to directly measure stack

emissions in a reliable or cost-eective manner.

Teck-Comincos reporting has so far relied on engineering

estimates based on various chemical processes.

Measurements of the following particulates have been taken

using dustfall jars during 20012002:

zinc (Zn), sulphates (x-SO

4

) and strontium (Sr).

Key Questions:

Is there a robust method that will provide reliable estimates of

stack emission rates based on deposition measurements?

And what is reliable? . . . errors of 25-50% in estimates are

considered acceptable, as long as they are overestimates!

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 5/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Outline

1

Background: Atmospheric Dispersion

2

The Gaussian Plume Solution

3

Inverse Problem: Estimating Emissions

4

Numerical Results

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 6/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Atmospheric Dispersion

Atmospheric disperion refers to transport of contaminants

via two processes:

1

advection by the wind, and

2

turbulent diusion.

Reduces to solving the advection-diusion equation

C

t

+u C = (KC) + Q R

where

C(x, t) = concentration (or density) of the contaminant (kg/m

3

)

u(x, t) = given wind velocity (m/s)

K = turbulent eddy diusivity (m

2

/s)

Q = emission source term (kg/m

3

s)

R = sink term from reactions, etc. (kg/m

3

s)

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 7/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Atmospheric Dispersion 2

A variety of approaches have been used for the solving the

advection-diusion equation, including

analytical: asymptotics, Greens functions.

computational: nite dierence, nite volume, spectral.

Scales of interest range from

10 m 100 km and minutes months,

. . . making many methods impractical, particularly in 3D.

Most industry-standard software is based on Gaussian plume

solutions (ref: epa.gov, recommended models).

Most work has focused on solving the forward problem:

Given a set of source emission rates, calculate deposition

as opposed to the inverse problem:

Given a set of deposition values, calculate emission rates

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 8/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Simplifying Assumptions

Typical assumptions for a single, isolated stack:

Stack is a point source located at (0, 0, H).

Constant emission rate Q (kg/m

3

s).

Constant wind velocity U (m/s) in x-direction.

Particles settle due to gravity at speed W

set

(m/s).

Deposition occurs at the ground at speed W

dep

(m/s).

Turbulent eddy diusivities K

x

, K

y

, K

z

(m

2

/s) are constant.

Turbulent eddy diusivities K

x

, K

y

, K

z

(m

2

/s) are constant.

Diusion downwind is negligible compared to convection (K

x

= 0).

Chemical reactions are negligible once particles are released.

Variations in topography are ignored (rectangular domain).

No inversions (z ) and no connement (x, y ).

Steady state. Steady state well relax this later.

Only short range transport is of interest. Only short range transport

is of interest bodes well for accuracy!

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 9/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Governing Equations

The advection-diusion equation reduces to:

U

C

x

. .

wind

W

set

C

z

. .

settling

=

y

_

K

y

C

y

_

+

z

_

K

z

C

z

_

. .

cross-wind and vertical diusion

+Q (x) (y) (z H)

. .

point source

Boundary

conditions:

K

z

C

z

+ W

set

C = W

dep

C at z = 0 (deposition)

C 0 as x, y and z

y

wind

effective

height

speed

plume centerline

U

z=H

z=0

x

z

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 10/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Outline

1

Background: Atmospheric Dispersion

2

The Gaussian Plume Solution

3

Inverse Problem: Estimating Emissions

4

Numerical Results

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 11/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Derivation

Assume no settling or deposition (W

set

= W

dep

= 0)

= Flux BC reduces to

C

z

(x, y, 0) = 0

Separable solution: C(x, y, z) = A(x, y) B(x, z).

Rescale variables: X =

1

UH

2

_

x

0

K(x

) dx

, Y =

y

H

, Z =

z

H

.

Yields two 2D diusion problems:

A

X

=

2

A

Y

2

A(0, Y) = C

o

(Y)

A(X, ) = 0

B

X

=

2

B

Z

2

B(0, Z) = C

o

(Z 1)

B

Z

(X, 0) = 0, B(X, ) = 0

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 12/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Gaussian Plume Solution

Use Laplace transform to obtain Gaussian plume (GP) solution:

C(x, y, z) =

Q

2U

y

z

exp

_

y

2

2

2

y

_

_

exp

_

(z H)

2

2

2

z

_

+ exp

_

(z + H)

2

2

2

z

_

_

Note: Eddy diusivities are re-

placed by standard deviations of

concentration:

2

y,z

(x) = 2xK

y,z

/U.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 13/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Typical Solution Contours

Ground!level concentration

x

y

0 500 1000 1500 2000

!250

!200

!150

!100

!50

0

50

100

150

200

0.5

1

1.5

2

x 10

!6

1e!06

3.16e!06

1e!05

3.16e!05

0.0001

Concentration on plume centerline

x

z

0 100 200 300 400

0

10

20

30

40

50

Notice peak in ground-level concentration downwind of source.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 14/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Practical Limitations

1

Choice of :

In practice, = ax

b

with a and b t to observations

(Brookhaven or Briggs formulas).

Only b =

1

2

is consistent with GP solution since

K =

2

U/2x = constant.

Other exponents lead to errors in mass conservation, but these

errors are only signicant at long range (Winges, 1990).

2

Errors at short and long range:

GP solution is singular as x 0 (limits accuracy near source).

Errors grow at large x, but our domain is small (only 2 km

2

).

3

Calm winds:

Common misconception: GP breaks down as U 0 Wrong!

In fact, C

Q

2x

K

y

K

z

+O(U) heat equation solution.

Real problem is an increase in plume rise (H) as U 0.

Rule of thumb: calm winds are dened as U = 0.5 m/s

(Hanna et al., 1982).

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 15/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Practical Limitations 2

4

Steady assumption:

GP is only strictly valid at steady state, but is a good estimate

of average concentrations over long enough time intervals.

A 10 min. averaging time is consistent with observations

(Hanna et al., 1982).

Also suggests that time-dependent simulations are feasible.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 16/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Solution with Deposition and Settling

Ermak (1977) derived a modied GP solution:

Replaces the zero vertical ux BC at z = 0 with

K

z

C

z

+ W

set

C = W

dep

C (deposition ux, kg/m

2

s)

Transform as before, but eliminate extra vertical convection term

2wB/Z using the substitution

B(X, Z) = B(X, Z) exp

_

w(Z 1) w

2

X

with w = HW

set

/2K

z

.

Yields modied diusion problems:

A

X

=

2

A

Y

2

A(0, Y) = C

o

(Y)

A(X, ) = 0

B

X

=

2

B

Z

2

B(0, Z) = C

o

(Z 1)

B

Z

(X, 0) = B(X, 0), B(X, ) = 0

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 17/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Solution with Deposition and Settling

Using Laplace transforms, the solution is:

C(x, y, z) =

Q

2U

y

z

exp

y

2

2

2

y

exp

W

set

(z H)

2K

z

W

2

set

2

z

8K

2

z

"

exp

(z H)

2

2

2

z

+ exp

(z + H)

2

2

2

z

2

W

o

z

K

z

exp

W

o

(z + H)

K

z

+

W

2

o

2

z

2K

2

z

erfc

W

o

z

2 K

z

+

z + H

2

z

#

where W

o

= W

dep

1

2

W

set

and W

set

= gd

2

/18 (Stokes law)

Key: Both concentration and deposition ux (W

dep

C) are linear in Q!!

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 18/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Other Generalizations

This analytical solution can be modied for a wide range of other

situations:

Line and area sources (Chrysikopoulos et al., 1992).

Instantaneous or Gaussian pu releases (basis for EPAs

CALPUFF code).

Inversion layers introduce a reecting BC at z = H

inv

> H,

leading to a series solution.

Include vertical dependence, U(z) and (z), owing to

boundary layer structure (Lin & Hildemann, 1996).

Take K

x

> 0 to handle U = 0 introduces integral terms

(Llewelyn, 1983).

. . . this is a real special function bonanza!!

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 19/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Other Applications

Stack emissions arent the only application of plume models:

Ash from volcanic eruptions (Turner & Hurst, 2001).

Release of radionucleotides from atomic power plants and

weapons blasts (Jeong et al., 1995).

Biological contaminants: e.g., anthrax release from Sverdlovsk

in 1979 (Meselson et al., 1994); terrorist attacks in urban

settings with complex geometries.

Seed dispersal (Levin et al., 2003).

Insect infestations: locusts, mountain pine beetles.

Odor propagation from livestock (Chastain & Wolak, 1999).

Dust and exhaust from automobiles roads are line sources.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 20/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Whats left?

So far, weve considered a single source in a constant wind.

We still need to include:

Multiple sources (still with constant emission rate).

Time-dependent wind velocity, not aligned with x-axis.

Multiple contaminants.

Ultimately . . . solve the inverse problem.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 21/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Time-dependent Wind

For source with location

, and wind

with speed U(t) and direction (t):

Shift and rotate coordinates

using x

= R

(x

), where

R is a rotation matrix.

Deposition ux is W

dep

C(x

).

Take wind measurements

{U

n

,

n

} at times t

n

= nt.

U

x

x

y

!

y

(x,y)

2 1

(" , " )

Rewrite deposition ux as W

dep

Q p(x;

, U

n

,

n

).

Total mass deposited within a small cell of area A, centered on x,

over total time Nt, is

D(x)

N

n=1

(W

dep

QAt) p(x;

, U

n

,

n

).

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 22/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Multiple Sources

Four sources (Sn) and nine dustfall jars or receptors (Rn):

400 metres

Scale:

100 200 0 300

R8

R9

S3

S4

R6

R5

R4

R3

R1 S1

R2

S2

R7

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 23/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Multiple Sources 2

Sources Q

s

have locations

s

for s = 1, 2, 3, 4.

At any location x, the deposition from all 4 sources is

D

tot

(x) =

4

s=1

N

n=1

(W

dep

Q

s

At) p(x;

s

, U

n

,

n

)

Example: Four sources with dierent emission rates:

0.05

0

.0

5

0.05

0.1

0

.1

0.1

0.5

0.5 1

x (m)

y

(

m

)

Contours of Zn concentration (g/m

2

)

S1

S2

S3

S4

0 500 1000

!500

0

500

1000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

+

0

.0

5

0.05

0.1

0.1

0.5

0.5

x (m)

y

(

m

)

Contours of Zn concentration (g/m

2

)

S1

S2

S3

S4

0 500 1000

!500

0

500

1000

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

+

0.05

0.05

0.1 0.1

x (m)

y

(

m

)

Contours of Zn concentration (g/m

2

)

S1

S2

S3

S4

0 500 1000

!500

0

500

1000

0.05

0.06

0.07

0.08

0.09

0.1

+

0.05

0

.0

5 0.1

0.1

x (m)

y

(

m

)

Contours of Zn concentration (g/m

2

)

S1

S2

S3

S4

0 500 1000

!500

0

500

1000

0.05

0.06

0.07

0.08

0.09

0.1

=

0.05

0.05

0.050.05

0.05

0

.0

5

0.05

0.1

0

.1

0.1

0.5

0.5

0.5

0.1

0.1

1

x (m)

y

(

m

)

Contours of Zn concentration (g/m

2

)

S1

S2

S3

S4

0 500 1000

!500

0

500

1000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 24/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Sample Output

One month cumulative zinc deposition using actual wind data:

0

.

0

5

0

.

0

5

0

.

0

5

0

.

0

5

0

.

0

5

0

.0

5

0.05

0

.0

5

0.1

0

.

1

0

.

1

0

.

1

0

.1

0

.

1

0

.

1

0

.1

0.1

0

.1

0

.5

0

.5

0

.

5

0

.

5

0.5

1

1 1

x (m)

y

(

m

)

Contours of Zn concentration (g/m

2

)

S1

S2

S3

S4

R1

R2

R3

R4

R5

R6

R7

R8

R9

0 500 1000

!500

0

500

1000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Wind histogram

500

1000

30

210

60

240

90

270

120

300

150

330

180 0

Due to Columbia River

valley, wind is

unidirectional!

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 25/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Outline

1

Background: Atmospheric Dispersion

2

The Gaussian Plume Solution

3

Inverse Problem: Estimating Emissions

4

Numerical Results

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 26/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Inverse Problem

At each receptor location

r

, for r = 1, 2, . . . , 9, measure zinc

deposition D

Zn

r

over one month:

D

Zn

r

=

4

s=1

N

n=1

(W

dep

Q

Zn

s

At) p(

r

;

s

, U

n

,

n

)

Yields 9 linear equations in 4 unknown Q

Zn

s

values.

Obtain a similar system for each q = Zn, Sr, SO

4

:

P

q

Q

q

=

D

q

where each P

q

is 9 4,

Q

q

is 4 1,

D

q

is 9 1.

Structure: a block diag. system of 27 eqns in 12 unknowns:

_

_

P

Zn

0 0

0 P

Sr

0

0 0 P

SO

4

_

_

_

Q

Zn

Q

Sr

Q

SO

4

_

_

=

_

D

Zn

D

Sr

D

SO

4

_

_

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 27/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Constraints

A number of other physical considerations impose equality and

inequality constraints:

Each emission rate is positive (12 inequalities).

Molar ratio of Zn relative to SO

4

generated at S1 (1).

Molar ratio of Zn relative to Sr from cooling towers at S3/S4

(2).

Sr is only emitted from cooling towers (2).

Two cooling tower sources are identical (3).

= An overdetermined linear system with linear constraints

(8 equality and 12 inequality)

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 28/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Ill-conditioning

The ill-conditioned nature of the unconstrained inverse

problem is well-studied in a series of papers by Enting &

Newsam (1988-2002).

When measurements are taken from long distances downwind,

or at high altitudes, then the ill-conditioning is severe.

The relatively mild degree of ill-posedness in the surface

source-deduction problem makes the numerical inversions

feasible.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 29/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Other Approaches

Brown (1993): used a nite dierence approximation for the

forward problem.

Mulholland & Seinfeld (1995): used a Kalman ltering

approach to estimate time-varying sources.

Bagtzoglou & Baun (2005): solved backwards

advection-diusion equation, using an equivalent beam

equation that is well-posed.

Hogan et al. (2005): calculated emission rate and source

location (4 variables) using 4 deposition measurements an

idealized case with synthetic data (exact solution!).

Jeong et al. (1995): used least squares to determine a single

emission rate from 51 measurements (very accurate).

MacKay et al. (2006): nonlinear least squares method for

estimating K and W

dep

for synthetic deposition data and

known emission rate (idealized).

Compare to our situation: time-varying wind, highly

over-constrained, real data.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 30/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Outline

1

Background: Atmospheric Dispersion

2

The Gaussian Plume Solution

3

Inverse Problem: Estimating Emissions

4

Numerical Results

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 31/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Numerical Simulations

Use Matlabs constrained linear least squares solver lsqlin.

Each run requires approx. 30 sec. on a Mac laptop fast!

Physical parameters were taken from the published literature.

Sensitivity study: results are most sensitive to

stack and receptor heights,

atmospheric stability class = determines

y,z

(x),

and (surprisingly) not so sensitive to noise (even up to 20%).

Use engineering estimates of zinc emissions as a guide:

Q

Zn

[35, 80, 5, 5] tons/yr

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 32/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Typical Deposition Measurements

R1 R2 R3 R4 R5 R6 R7 R8 R9

0

10

20

30

40

50

60

70

Jun 3!Jul 2, 2002

Receptor

A

m

o

u

n

t

d

e

p

o

s

i

t

e

d

Zn!mg

S!mg

Sr!ng

Most Zn measurements are consistent from month to month.

Certain Zn measurements exhibit large deviations at R3.

More variation in SO

4

and Sr data (but less sensitive).

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 33/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Results

Measured depositions Computed emission rates

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 34/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Results 2

Measured depositions Computed emission rates

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 35/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Conclusions

The method does a reasonable job of capturing total Zn

emissions.

Individual Zn source estimates still vary considerably.

Assuming near-constant emissions, wed expect all deposition

measurements to be similar

= suggests that discrepancies in the inverse solution can be

attributed to measurement errors (particularly at R3)

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 36/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Results Without R3

Estimated emissions without R3 are much better (except S3/S4):

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 37/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Summary

Used convection-diusion equation with ux boundary

conditions to model contaminant transport.

Solved exactly using Laplace transforms.

Linearity (in Q) allowed superposition of sources.

Given deposition data leads to overdetermined linear system.

Emissions rates obtained using linear least squares method.

Ill-posedness limits accuracy of results.

Total emissions are still reasonable . . . which is all thats

needed from a regulatory standpoint!

These results have appeared as

E. Lushi & JMS, Atmospheric Environment,

44(8):10971107, 2010.

and a second paper is soon to be submitted to SIAM Reviews

Education column.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 38/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Future Work

Investigate accuracy of R3 deposition measurements with

Teck-Cominco engineers.

Further validate the algorithm using other deposition

measurements for months with missing wind data.

Relate features of inverse solution to eigenvectors of P matrix

(Jackson, 1972).

Teck-Cominco is currently undertaking another round of

deposition measurements (Feb.Nov. 2010) . . .

Sudeshna Ghosh

PhD student

Internship funded jointly by

MITACS and Teck-Cominco

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 39/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

On-going Work

We aim to validate some of these results with direct simulations of

the advection-diusion equation:

C

t

+u C = (KC) + Q

Use CLAWPACKs high resolution schemes for advective

transport.

Handle diusion using the Peaceman-Rachford ADI scheme.

Approximate the delta functions in source terms using

(x) =

(x

2

+

2

)

Eddy diusivities: K

x

= K

y

= 2 3 m

2

/s, and K

z

taken from

Lettau & Dabberdt (1970):

K

z

(z) = 0.6033 + 0.0185z 0.000108z

2

m

2

/s

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 40/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

Thank-you!

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 41/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

References I

Amvrossios C. Bagtzoglou and Sandrine A. Baun.

Near real-time atmospheric contamination source identication by an

optimization-based inverse method.

Inverse Problems in Science and Engineering, 13(3):241259, 2005.

Margaret Brown.

Deduction of emissions of source gases using an objective inversion

algorithm and a chemical transport model.

J. Geophys. Res., 98(D7):1263912660, 1993.

John P. Chastain and Francis J. Wolak.

Application of a Gaussian plume model of odor dispersion to select a site

for livestock facilities.

Unpublished report, 1999.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 42/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

References II

Constantinos V. Chrysikopoulos, Lynn M. Hildemann, and Paul V.

Roberts.

A three-dimensional steady-state atmospheric dispersion-deposition model

for emissions from a ground-level area source.

Atmos. Environ., 26A(5):747757, 1992.

I. G. Enting and G. N. Newsam.

Atmospheric constitutent inversion problems: Implications for baseline

monitoring.

J. Atmos. Chem., 11:6987, 1990.

Ian G. Enting.

Inverse Problems in Atmospheric Constituent Transport.

Cambridge Atmospheric and Space Science Series. Cambridge University

Press, 2002.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 43/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

References III

Donald L. Ermak.

An analytical model for air pollutant transport and deposition from a

point source.

Atmos. Environ., 11(3):231237, 1977.

Steven R. Hanna, Gary A. Briggs, and Rayford P. Hosker Jr.

Handbook on atmospheric diusion.

Technical Report DOE/TIC-11223, Technical Information Center, U.S.

Department of Energy, 1982.

William R. Hogan, G. F. Cooper, M. M. Wagner, and G. L. Wallstrom.

An inverted Gaussian plume model for estimating the location and amount

of release of airborne agents from downwind atmospheric concentrations.

RODS technical report, Realtime Outbreak and Disease Surveillance

Laboratory, University of Pittsburgh, Pittsburgh, PA, 2005.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 44/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

References IV

D. D. Jackson.

Interpretation of inaccurate, insucient and inconsistent data.

Geophys. J. Roy. Astron. Soc., 28:97109, 1972.

Hyo-Joon Jeong, Eun-Han Kim, Kyung-Suk Suh, Won-Tae Hwang,

Moon-Hee Han, and Hong-Keun Lee.

Determination of the source rate released into the environment from a

nuclear power plant.

Rad. Prot. Dos., 113(3):308313, 2005.

Heinz H. Lettau and Walter F. Dabberdt.

Variangular wind spirals.

Bound. Layer Meteorol., 1:6479, 1970.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 45/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

References V

Simon A. Levin, Helene C. Muller-Landau, Ran Nathan, and Jerome

Chave.

The ecology and evoluation of seed dispersal: A theoretical perspective.

Annual Review of Ecology, Evolution, and Systematics, 34:575604, 2003.

Jin-Sheng Lin and Lynn M. Hildemann.

Analytical solutions of the atmospheric diusion equation with multiple

sources and height-dependent wind speed and eddy diusivities.

Atmos. Environ., 30(2):239254, 1996.

Richard P. Llewelyn.

An analytical model for the transport, dispersion and elimination of air

pollutants emitted from a point source.

Atmos. Environ., 17(2):249256, 1983.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 46/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

References VI

Cameron MacKay, Sean McKee, and Anthony J. Mulholland.

Diusion and convection of gaseous and ne particulate from a chimney.

IMA J. Appl. Math., 71:670691, 2006.

Matthew Meselson, Jeanne Guillemin, Martin-Hugh-Jones, Alexander

Langmuir, Ilona Popova, Alexis Shelokov, and Olga Yampolskaya.

The Sverdlovsk anthrax outbreak of 1979.

Science, 266:12021208, 1994.

Michael Mulholland and John H. Seinfeld.

Inverse air pollution modelling of urban-scale carbon monoxide emissions.

Atmos. Environ., 29(4):497516, 1995.

G. N. Newsam and I. G. Enting.

Inverse problems in atmospheric constituent studies: I. Determination of

surface sources under a diusive transport approximation.

Inv. Prob., 4:10371054, 1988.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 47/48

Background: Atmospheric Dispersion

The Gaussian Plume Solution

Inverse Problem: Estimating Emissions

Numerical Results

References VII

Richard Turner and Tony Hurst.

Factors inuencing volcanic ash dispersal from the 1995 and 1996

eruptions of Mount Ruapehu, New Zealand.

J. Appl. Meteorol., 40:5669, 2001.

U.S. Environmental Protection Agency.

Guideline on Air Quality Models, Appendix W to Part 51, Title 40:

Protection of the Environment, Code of Federal Regulations, 2010.

Source: http://ecfr.gpoaccess.gov.

Kirk D. Winges.

Users guide for the fugitive dust model.

Report EPA-910/9-88-202R, TRC Environmental Consultants Inc.,

Mountlake Terrace, WA, May 1990.

Mathematical Modelling of Atmospheric Dispersion John Stockie SFU 48/48

- Environmentalists' Comments on EPA's Proposed Oil and Gas Emission RulesUploaded byjendlouhy
- Qt-i(Prob Dist III)(6)Uploaded byPartha Boyal
- CombustionUploaded byapi-3721883
- Lecture 1Uploaded by0776
- UBICC IKE07 Performance Analysis for Skewed Data 191 191Uploaded byUbiquitous Computing and Communication Journal
- mstatsample (1)Uploaded bysap_chacks
- - Fuel CombustionUploaded bySivashankar Dhanaraj
- Sample Mean DistributionUploaded byVarun Gogia
- six sigmaaaUploaded byarifmukhtar
- pmcUploaded byVíctor Rivasplata
- eode-air-quality-impact.pdfUploaded byBalan Palaniappan
- Chapter 19 Probability DistributionUploaded byfazz_zalnia
- Stigler stat Ch. 2`Uploaded byplxnospam
- 34Uploaded byArthur Magalhães
- Prob-8Uploaded byAbdul Saboor Khan
- TR-ExlUploaded byWesternDigi
- W6INSE6220.pdfUploaded bypicala
- UnfinishedUploaded bywasdwasdwasdwasdwasd wasdwasdwasd
- The Standard Normal Distribution.docxUploaded byjanice lopez
- probabilidadUploaded bygmendozap
- Analisis de Planeacion de Proyectos Usando Cpm y PertUploaded byBrayan Huertas
- Oiml Bulletin Oct 2009Uploaded byAntmavr
- Air Quality Assessment for Smart Transportation System in GurugramUploaded byAdvanced Research Publications
- _24_._09_._2018_Economic DUploaded byMarius Ndong
- AHP survey 3.8.2019.pdfUploaded byPhạm Tiến Đạt
- Tonnage Calc-4 (1)Uploaded byEmba Madras
- Assignment 1Uploaded byKibrom Haftu
- lecture 6Uploaded byHamna abid
- Awardees Speech (1)Uploaded byDijah Mie
- sdasdsadsadsadsadsaUploaded byJorge

- YZ_11 Intumescent Coating ModellingUploaded byLTE002
- ZH_09 Steel ConnectionUploaded byLTE002
- YYJ_stiff_2003 Caternary Action -Steel BeamUploaded byLTE002
- YZ_10 Intumescent CoatingUploaded byLTE002
- Toxicity of BiodieselUploaded byLTE002
- Oxygen RequirementsUploaded byLTE002
- ZH_2005 RCUploaded byLTE002
- Bridges_Asia-22-25.02.10Uploaded byLTE002
- 833 Anaerobic Digestion ParametersUploaded byLTE002
- Bio Gas Burner 1Uploaded bysaadullah_siddiqui6076
- Building Digest 20Uploaded byLTE002
- Whitepaper - Affinity LawsUploaded bybitconcepts9781
- Soil WashingUploaded byLTE002
- Gas ChromatographUploaded byLTE002
- PH MeasurementsUploaded byLTE002
- MethaneUploaded byLTE002
- Centrifugal PumpUploaded byrumabiswas853
- Fire SprinklersUploaded byLTE002
- About ElevatorsUploaded byJosé
- 39261 High PerformanceUploaded byLTE002
- 39551 Cal Methodology - Energy Saving- Electrical -HouseholdUploaded byLTE002
- NRL ReportUploaded byWinson Yew
- 26157 Biomass to Ethanol ProcessUploaded byLTE002
- Simulation of High-Speed FillingUploaded byLTE002
- Trigger Sprayer Dynamic Systems ModelUploaded byLTE002
- Heat Exchanger VibrationUploaded byTim Ku
- Pipeline VAPIPE2Uploaded byLTE002
- EarthingUploaded bymamoun_hammad7917
- ETTV - BCAUploaded byHo Chee Yong

- Jackson ElectrodynamicsUploaded bynilayan mukherjee
- 350082467-Solucionario-Pollack.pdfUploaded byJoel Alayo Onton
- P _ C.pdfUploaded bysow
- Levy FLights in FractalsUploaded byParag Mahajani
- Parsons's _Arithmetic and Categories_ NotesUploaded byHüseyin Güngör
- u1-a7 skillsUploaded byapi-271298524
- Upheaval BucklingUploaded bykjnyman
- lesson plan - distance formula and eauations of circlesUploaded byapi-352809637
- warwick linear algebra innaUploaded byRushil Gholkar
- 20111007-Fernandez-Torres（revised）Uploaded bygfernand34
- L08 - Benchmarking Techniques (ENG)Uploaded bykillerkiss
- Lesson Plan Two BracketsUploaded byJonathan Robinson
- MIT16_30F10_lec01Uploaded byNizar Tayem
- Louis Leithold - The Calculus, With Analytic Geometry (Chapter 8)Uploaded bySiir Cy
- sol01Uploaded byJay Shah
- A Page on Eisenstein Series in Ramanujan's Lost NotebookUploaded byapi-26401608
- C5.pdfUploaded byRahul Sharma
- MATH39001 - Combinatorics and Graph Theory - Exam - Jan-2011Uploaded byscribd6289
- DIFFERENTIAL ANALYSIS CFDUploaded byotonbara
- lesson plan-valdez-8th-math week 15 5Uploaded byapi-431340065
- fl-who are the folk-alan dundes.pdfUploaded byAmrita Amrita
- Tridiagonal Matrix Algorithm - WikipediaUploaded byWindy H Mitrakusuma
- Advanced-Math-Evaluation Masters.pdfUploaded byAries
- Example 5 _ Plane Areas in Rectangular Coordinates _ Integral Calculus ReviewUploaded byRomeo Victor
- (Lecture Notes in Computer Science 7907) Rūsiņš Freivalds (Auth.), Marie-Pierre Béal, Olivier Carton (Eds.)-Developments in Language Theory_ 17th International Conference, DLT 2013, Marne-la-Vallée, FUploaded byMARCO
- Standard LaTeX Math ModeUploaded bykivava
- MTY5PKBS22008PAPER1_2Uploaded byAnonymous l2Fve4PpD
- Calculator Techniques PERCDCUploaded byHarold Silao
- 025 - Pr 01 - Drude Model and Poisson DistributionUploaded byBradley Nartowt
- Keys to Infinity - Clifford a. PickoverUploaded byChidranveshi

## Much more than documents.

Discover everything Scribd has to offer, including books and audiobooks from major publishers.

Cancel anytime.