Critical Review of the Building Downwash Algorithms in AERMOD

www.cppwind.comwww.cppwind.com
Critical Review of the
Building Downwash
Algorithms in AERMOD
Guideline on Air Quality Models: The New Path
Chapel Hill, NC
April 12-14, 2016
Ron Petersen, PhD, CCM, FASHRAE
Sergio Guerra, PhD
CPP, Inc.
2400 Midpoint Drive, Suite 190
Fort Collins, CO 80525
www.cppwind.com @CPPWindExperts

Background
• AERMOD with PRIME became the preferred model in 2006.
• Primary documentation on PRIME is in 2000 A&WMA Journal
article by Schulman, Strimaitis and Scire. Not all the details
included.
• PRIME code is in Fortran and is well documented. Details are
there.
• Overprediction and underprediction problems have been
found for various building configurations.
• CPP funded research into the PRIME theory to find the cause
of these problems.

Examples Problems -Overprediction
From 10th Modeling Conference
Schulman, 2012, Wide/Long Building Issue
• Wide Buildings: Concentration
increased by factors of 3 to 14
when Width > 4 x Height
• Long Buildings: Concentration
increased by factors of 4 to 10
when Length > 4 x height for
GEP stack.
• Field Observations at ALCOA TN
wide/long facility: Model
overpredicts by factor of ~10.
Hs/Hb = 2.5

An Assessment of the AERMOD by
IDEM
Keith Baugues, Assistant
Commissioner
AERMOD not working!!!!
• Q:Q: Model Overpredicts by
Factor of 2 or More
• Paired: Very Poor
Agreement
Model (ppm)
Model (ppm)
Observed(ppm)
Observed(ppm)

AECOM Field Study at Mirant Power
Station (Shea et al., 2012)
2Shea, D., O. Kostrova, A. MacNutt, R. Paine, D. Cramer, L. Labrie, “A Model Evaluation Study of AERMOD Using Wind Tunnel
and Ambient Measurements at Elevated Locations,” 100th Annual AWMA Conference, Pittsburgh, PA, June 2007.
• Model overpredicted by factor
of 10 on residential tower
• Best agreement with no
buildings, still overpredicted by
factor of 2.
• In reality, plume is not affected
by building downwash.
AERMOD
with BPIP
AERMOD
No Buildings

Theoretical Problems that may
Cause the Overpredictions

AERMOD Building Wake Problems –
AERMOD Overestimates Downwash (BPIP & PRIME Problem)
Hb = 20 m
Problem even worse for longer buildings
• Wake height
overestimated: need
higher plumes to
avoid downwash.
• Start of maximum
building downwash
farther downwind than
in reality

Turbulence Calculations in Wake Flawed
- Constant downwash enhancement up to wake height (Fix?)
Starting Relation
J.C. Weil, A New Dispersion Model for Stack Sources in Building Wakes,
9th Joint Conference on Air Pollution Meteorology with A&WMA, 1996.
Bad assumptions
𝑖 𝑧 = 𝑖 𝑜
1 +
Δ𝜎 𝑤𝑜
𝜎 𝑤𝑜
𝜉
𝑅
−
2
3
1 + Δ𝑈 𝑜/𝑈 𝑜
𝜉
𝑅
−
2
3

AERMOD/PPRIME
Overestimates
Downwash
Reality
Illustration of PRIME Building Downwash Problem:
Height of Building Downwash Zone Overestimated

Another PRIME Problem
Downwash (turbulence) enhanced by factor of ~10
under stable conditions: not documented (Fix?).
Not supported by theory. No evidence
supporting this is provided!!
ξ = R @ X = 0, the lee wall
Is PRIME really enhancing
turbulence like this? Yes
𝑖 𝑧(𝜉, 𝑧) = 𝑖 𝑧𝑜(𝑧) 1 +
1.7𝑖 𝑧𝑁
𝑖 𝑧𝑜(𝑧)
− 1 +
∆𝑈 𝑜
𝑈 𝑜
ξ
𝑅
2
3
−
∆𝑈 𝑜
𝑈 𝑜
Wake
Turbulent
=
Approach
Turbulence
Function
(ξ, Stability)*

W/H
2.4
2.4
2.4
8.4
H (cm)
3.1
6.5
8.3
6.4
AERMOD
Turbulence decay in x direction and with building
height flawed: NASA Database
Affects wake turbulence calculations
Short/narrow
building
Tall/wide
building
Turbulence
Increase
Factor
Downwind Distance Divided by Building Height

W/H
2.4
2.4
2.4
8.4
H (cm)
3.1
6.5
8.3
6.4
AERMOD
Velocity deficit in x direction and with building
height flawed: NASA Database
Affects plume rise and turbulence calculations
Velocity
Decrease
Factor

Refinery Structures Upwind
- Horizontal flow
Solid BPIP Structure Upwind
No Structures
Streamlines for Lattice Structures
Should be horizontal (Fix?)
Affects Plume Rise Calculations

Realistic Equation
ξ = R @ X = 0, the lee wall
𝑖 𝑧(𝜉, 𝑧) = 𝑖 𝑧𝑜(𝑧)
1 +
Δ𝜎𝑧𝑜
𝜎𝑧𝑜
(𝑧, 𝑖 𝑧𝑜, 𝑠𝑡𝑟𝑢𝑐𝑡𝑖𝑜𝑛 𝑡𝑦𝑝𝑒)
𝜉
𝑅
−
2
3
1 + Δ𝑈 𝑜/𝑈 𝑜(𝑧, 𝑖 𝑧𝑜, 𝑠𝑡𝑟𝑢𝑐𝑡𝑖𝑜𝑛 𝑡𝑦𝑝𝑒)
𝜉
𝑅
−
2
3
𝑖 𝑧(𝜉, 𝑧) = 𝑖 𝑧𝑜(𝑧) 1 +
1.7𝑖 𝑧𝑁
𝑖 𝑧𝑜(𝑧)
− 1 +
∆𝑈 𝑜
𝑈 𝑜
ξ
𝑅
2
3
−
∆𝑈 𝑜
𝑈 𝑜
Existing
Realistic
Turbulence increase and velocity deficit vary based on x, z,
approach turbulence and structure type
Turbulence increase factor only varies with x and L (stability)

CPP’s Limited Research
Velocity Mapping for 1:1:2 Building

Findings from CPP’s Limited Research
- Wind tunnel measurements show little enhancement
above building height (Fix?)
Building
Distance Turbulence Increase Factor
ξ/Hb AERMOD Observed
1 5.7 1.0 to 5.7
2 4.4 1.0 to 5.2
3 2.9 1.0 to 2.2
Turbulence Increase Factor

FDS LES Simulation for 1:1:2 Building
Very little downwash enhancement above the building

Other
Problems

Streamline Calculation Comparison
Flawed (Bug?)
Figure 6. Prime predicted and observed streamlines from Schulman1
Given:
• H=W=L=R
PRIME Logic
• If L> 0.9R (= 0.9L)
reattachment occurs, and
Hr = H
For this case,
• L>0.9R = 0.9L, therefore
• Hr = H
That means all streamlines
should be horizontal and they
are not in example.
What is PRIME really doing?

Another Streamline Calculation
Problem (Bug?)

More Difficult Fixes:
AERMOD Underprediction Cases
Require Major Research and New Theory
• Factor of two:
Corner Vortex
Rhinelander Corner Vortex
AERMOD ~ Factor of 2 Low at Monitor
Upwind Terrain Wakes Not
Treated in AERMOD
• Factor of 2-6:
Upwind
Terrain

Solutions and Next Generation
• Short Term Fix: Use Equivalent Building Dimensions
• EBDs are the dimensions (height, width, length
and location) that are input into AERMOD in place
of BPIP dimensions to more accurately predict
building wake effects
• Not a complete fix because of problems with the
theory
• Determined using wind tunnel modeling
• Next Generation: Improved PRIME (AERMOD
and SCICHEM) and BPIP
• Collaboration between EPA and Industry

BPIP Diagnostic Tool
http://bit.do/cppwind-BPIPDiagnostic
Likely Overprediction Factor for each Flow Vector
Source 1

Complianc
e
CPP’s EBD
BPIP Diagnostic
ToolBuilding Geometry
Meteorological Data
Terrain Data
AERMET
AERMAP
Operating Parameters
AERMOD
OtherInputs
Building
Inputs
BPIP Diagnostic Tool

• Equivalent Building Dimensions (EBDs) are the dimensions (height, width, length
and location) that are input into AERMOD in place of BPIP dimensions to more
accurately predict building wake effects
• Guidance originally developed when ISC was the preferred model –
– EPA, 1994. Wind Tunnel Modeling Demonstration to Determine Equivalent
Building Dimensions for the Cape Industries Facility, Wilmington, North
Carolina. Joseph A. Tikvart Memorandum, dated July 25, 1994. U.S.
Environmental Protection Agency, Research Triangle Park, NC
• Determined using wind tunnel modeling
EBD Method

Why EBD helps but
doesn’t solve
problem.
EBD is smallest
building that
replicates the site
dispersion
AERMOD works best
for smaller buildings
AERMOD/PPRIME
Overestimates
Downwash
EBD
~ reality
Reality

AERMOD Results With Wind Tunnel EBD
Very wide/narrow building
Stack height: 47 m
Building height: 31 m
Property line in Red
Emission rate: 1 g/s
AERMOD RESULTS
Five years of met data
AERMOD Building Dimension
Inputs 1-hour 24-hour annual
BPIP 15.19 8.20 0.89
Wind Tunnel EBD 3.99 1.88 0.18
Reduction Factor 3.80 4.37 4.93
AERMOD Maximum predicted

Long Buildings with Wind
at an Angle
Figure created in BREEZE® Downwash Analyst
BREEZE is a registered trademark of Trinity Consultants, Inc.

Downwash Based on EBD and BPIP
Figures created in BREEZE® Downwash Analyst
BREEZE is a registered trademark of Trinity Consultants, Inc.

Stack height: 45 m
Structure height: 61 m
Emission rate: 1 g/s
Five years of met data
Stack
AERMOD Result
Lattice Structure
Building
Input 1-hour 24-hour annual
BPIP 23.21 5.51 0.37
Wind Tunnel EBD 7.72 2.36 0.11
Reduction Factor 3.01 2.33 3.51
Maximum Concentration Results

FACTOR of 4 to 8
reduction when EBD used
Short building with a large foot print
FACTOR of 2 to 4
Hyperbolic cooling towers
Typical AERMOD Overprediction Factors When
Using BPIP Inputs and Current Theory

Typical AERMOD Overprediction Factors When
Using BPIP Inputs and Current Theory
FACTOR of 2 to 5
Very Wide/Narrow Buildings
FACTOR of 2 to 3.5
Lattice Structures

Summary
The Next Generation Downwash Model
• Correct all the bugs and fix the known problems in
the theory
• Incorporate the current state of science
• Advance the current state of the science
• Expand the types of structures that can be accurately
handled
• Well documented and verified model formulation
document and code for PRIME
• Add section to Appendix W that outlines a method
that all can use (versus just EPA) to update models
based on current research.

First Steps
• EPA policy that encourages industrial research
and freely and openly collaborates during the
execution of that research → theme of this
conference.
• PRIME2 Advisory Committee has an initial
kickoff meeting to start this cooperative
process after lunch. A subcommittee of APM.
• Meeting is open to all interested in improving
building downwash.

Ron Petersen, PhD, CCM
rpetersen@cppwind.com
Direct: + 970 498 2366
Sergio Guerra, PhD
sguerra@cppwind.com
CPP, Inc.
2400 Midpoint Drive, Suite 190
Fort Collins, CO 80525
+ 970 221 3371
www.cppwind.com @CPPWindExperts
Questions?

Critical Review of the Building Downwash Algorithms in AERMOD

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Was ist angesagt?

Was ist angesagt? (20)

Andere mochten auch

Andere mochten auch (16)

Ähnlich wie Critical Review of the Building Downwash Algorithms in AERMOD

Ähnlich wie Critical Review of the Building Downwash Algorithms in AERMOD (20)

Kürzlich hochgeladen

Kürzlich hochgeladen (20)

Critical Review of the Building Downwash Algorithms in AERMOD