Computational Fluid Dynamics Modelling of Aerosol Dispersion and - - PowerPoint PPT Presentation

Computational Fluid Dynamics Modelling of Aerosol Dispersion and Processes within Urban Street Canyons

Bee Kiat Tay1 Martin Gallagher1, Gordon McFiggans1 & Paul Watkins2

University of Manchester, UK

• 1. School of Earth Atmospheric and Environmental Sciences, Centre of

Atmospheric Science

• 2. Mechanical, Aerospace and Civil Engineering

31st March 2009

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Contents

• Background of Project
• Code Description & Design of Computational

Domain

• Research Themes

– Extent of heterogeneities of flow properties – Investigation of flux from street canyons – Urban canyon aerosol processes

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Background of Project

• Vehicle emission is a major source of aerosols in the urban environment

– Adverse health effects are attributed to exposure to aerosols (fine/ ultrafine)

• Nature of urban street canyon dispersion is influenced by both

micrometeorological factors & urban geometry

• Urban aerosol distribution

– Size spans across several orders of magnitude – Multi-modal in nature/ Lognormal – Modified by aerosol processes ( e.g. condensation, coagulation)

• Research Questions

– Structure of aerosol dispersion within street canyons – Sensitivity of aerosol flux to various meteorological factors – Importance of in-canyon aerosol processes

Coupling aerosol dynamical model with Computational Fluid Dynamics to model urban aerosol transport in street canyon

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

• In-house 2-Dimensional CFD code (Finite Volume Method)1

– Features

• Incompressible flow (constant density)
• Buoyancy represented by Boussinesq approximation
• Turbulent flow represented by standard k- turbulence model

– High Reynolds Number wall function to characterise shear at the walls

• SIMPLE (Semi Implicit Method for Pressure Linked Equation) algorithm to obtain

velocity vectors

– Dispersion patterns validated with wind tunnel dispersion data2

• Euler-Euler multiphase approach
• Discrete phase represented by moments, integral property of size distribution
• Modal Method: 3 moments (0th,3rd,6th) transported to facilitate lognormal closure of

distribution

• Low Volumetric Loading:

– Discrete phase carried by flow field of the continuous phase – One-way coupling (aerosol phase has insignificant influence on continuous phase)

Code Description

1.

• D. P. Jones, A. P. Watkins; Spray Impingement Model based on Method of Moments; 22nd European Conference on Liquid Atomisation and Spray Systems

2.

• M. Ketzel, P. Louka, P. Sahm, E. Guilloteau, J.-F.; Sini The use of computational fluid dynamics in modelling air quality in street canyons. Chapter 2.2 of the TRAPOS

summary report

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science 5H 10H 2H H W Wall (Smooth) Outflow Inflow Symmetry Velocity and Turbulence Intensity* Emission (0.3m above street)

Schematic Diagram of Computational Domain

*ratio of the root-mean-square of the velocity fluctuations, u', to the mean flow velocity

H/W= 1, 2

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Underlying Assumptions

• Steady state
• 2 D (perpendicular wind), ignoring lateral effects

– Assuming a long canyon

• Vehicular turbulence may be neglected

– For wind speeds >2.5 m/s

• Aerosols treated as inert scalar (advection and turbulent diffusion)
• Constant concentration of aerosols at “emission point”, approximately 1m from

the tailpipe – Valid for fast traffic movement

• Assumed distribution at emission point:

Standard Deviation Geometric Mean Diameter (nm) Number Concentration (particle/ m3)

Lognormal Parameters Lognormal Parameters

1.6 1.5

g

150 15

DGN

1x1010 1x1011

No

Accummulation mode Aitken mode

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Investigation of the Vertical Structure

Prashant Kumar, Andrew Garmory, Matthias Ketzel, Ruwim Berkowicz, Rex Britter; Comparative study of measured and modelled number concentrations of nanoparticles in an urban street canyon; Atmospheric Environment, Volume 43, Issue 4, February 2009, Pages 949-958

• Normalised Concentration

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

X/W Z/H

0.5 1 0.2 0.4 0.6 0.8 1

Frame 001  26 Mar 2009  continuity

1x10 8 5 x 1

8

1 x 1

9

1.5x10 9

2x10 9

X/W Z/H

0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1

Frame 001  26 Mar 2009  continuity

1x10 8 5 x 1

8

1 x 1

9

1.5x10 9 2 x 1 9

Aerosol Dispersion

Particle Number Concentration ( p m-3)

Aspect Ratio: 1 Aspect Ratio:2 Leeward Windward

Wind

Sharper concentration gradient

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

1 2 3 4 5 6 7 8 9 10 11 0.00E+00 4.00E+08 8.00E+08 1.20E+09 1.60E+09 2.00E+09

Particle Concentration (p/m3) Height of Canyon (m)

10 m/s Low TI 10 m/s Mid TI 10 m/s High TI 2.5 m/s Low TI 2.5 m/s Mid TI 2.5 m/s High TI

Leeward Vertical Profile

Comparable with previous measurement results

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

1 2 3 4 5 6 7 8 9 10 11 0.00E+00 1.00E+08 2.00E+08 3.00E+08 4.00E+08 5.00E+08 6.00E+08

Particle Concentration (p/m3) Height of Canyon (m)

10 m/s Low TI 10 m/s Mid TI 10 m/s High TI 2.5 m/s Low TI 2.5 m/s Mid TI 2.5 m/s High TI

Windward Vertical Profile

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Turbulent Shear Layer

Increasing Concentration with height

Decreasing concentration with height. Gradient proportional to turbulent viscosity. Concentration increasing with height at the lower part of canyon, then decreasing with height.

Wind Bulk entrainment of air into canyon Venting from canyon (AR=1) Leeward Windward

(Sharper negative concentration gradient)

Results Summary

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

___

( ) ( ) ( )

NET Turbulent advective t m L c L L

F F dx C cw CW dx K CW dx z β β β = + = + ∂ = + = − + ∂

• x

z

NET Turbulent advective

β β β = +

is horizontally integrated flux F is flux C is concentration of aerosol Kc is turbulent diffusivity

Transport and dispersion

• f aerosols due to

advection and turbulent diffusion

Inflow Condition:

Wind speed: 2.5- 10 ms-1 & Turbulence Intensity (TI): 0.05 to 0.26)

Flux from Street Canyon

Aspect Ratio: 1- 2

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

2 3 4 5 6 7 8 9 U 0.05 0.1 0.15 0.2 0.25 TI 1e+08 1e+08 2e+08 2e+08 3e+08 3e+08 4e+08 4e+08 5e+08 5e+08 6e+08 6e+08 Flux Flux flux1.xls : (1)Sheet1, X , Y , Z

Rank 4 Eqn 151232673 lnz=a+blnx+cy

r^2=0.99861805 DF Adj r^2=0.99778888 FitStdErr=6006567.1 Fstat=2167.842 a=16.960305 b=0.92517512 c=3.7288871

ASPECT RATIO: 1

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

2 3 4 5 6 7 8 9 U 0.05 0.1 0.15 0.2 0.25 TI 5e+07 5e+07 1e+08 1e+08 1.5e+08 1.5e+08 2e+08 2e+08 2.5e+08 2.5e+08 3e+08 3e+08 3.5e+08 3.5e+08 Flux Flux flux3.xls : (1)Sheet1, X , Y , Z

Rank 4 Eqn 151232673 lnz=a+blnx+cy

r^2=0.99897237 DF Adj r^2=0.99835579 FitStdErr=3319762.7 Fstat=2916.3289 a=16.880619 b=0.95533697 c=1.9685648

ASPECT RATIO: 2

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Discussion

• Turbulent flux dominates the ventilation process at the roof level

– Turbulent flux is of higher magnitude than advective flux – Advective flux is negative (reintroduction of aerosols into the canyon) and turbulent flux is positive (vents aerosols). – Net flux is positive

• Exponential relationship between TI and net flux
• A parameterisation is proposed for a perpendicularly blowing wind

and when buoyancy effects are unimportant

exp( ln )

N in

a b U cTI β = + +

Exponential Relation Positive Relation

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Relating parameterizations to measurements

Coupling between the measured flux at the roof of a tall building and flux at neighbouring street canyons. Measured flux is an aggregation

• f the fluxes from the different

street canyons in the vicinity. Applicability of the parameterisation, using measurements at Manchester as a test case. ASSUMPTIONS

• Effects of horizontal advection

may be ignored;

• Well mixed area between the

street canyon and measuring location;

• Aerosol transformation

processes within the urban canopy layer are complete or unimportant

• Sources and sinks within the

urban canopy are either absent,

• r unimportant.
• 8

exp(9.81 4.54 x10 ln 2.69 )

N in

U TI β = + +

This further suggests coupling of the dynamics between surface emissions and the urban canopy.

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

• Photochemical oxidation of primary emitted volatile
• rganic compounds (VOC)
• Secondary aerosol formation

– Gas-liquid condensation of semi-volatiles due to super-saturation ( partial pressure (Pi) > saturation vapour pressure (Psat))

Formation of SOA in Urban Canyon

VOC Oxidized VOC

OH/ O3

Intermediate VOC Semi-volatile VOC

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Primary Organic Aerosols condensation

svoc (l)

*

2 ( )

p i i i p aerosol

dv D c c D dt π ρ = −

Formation of SOA in Urban Canyon

Ci*=0.001 mg/m3

(Psat: approx 8x10-11 atm)

svoc (g)

Ci*: Saturation concentration Psat: saturation vapour pressure Condensing on the Aitken mode:

• Number Concentration (No): 1x1011 p / m3
• Geometric Mean Diameter (DGN): 15 nm
• Standard Deviation(g): 1.5

Csvoc=0.2 mg/m3

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

N0( p/m3) N0: Number Concentration

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

DGN :Geometric Mean Diameter

(µm)

DGN

Misguided measurement strategies will lead to erroneous conclusions!

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

g: Standard Deviation

g

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Primary Organic Aerosols

svoc (g)

condensation

voc (g) svoc (l)

[ ]

• 1

1

• c

C (mg.min )

• c

dC k dt = −

• xidation

evaporation

Formation of SOA in Urban Canyon

k1=1x10-4 min-1 Ci*=0.0032 mg/m3 Cvoc=50 mg/m3

Ci*: Saturation concentration Partitioning taking place at the Aitken mode:

• Number Concentration (No): 1x1011 p / m3
• Geometric Mean Diameter (DGN): 15 nm
• Standard Deviation(g): 1.5

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

N0( p/m3) N0: Number Concentration

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

DGN :Geometric Mean Diameter (µm) DGN

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

g: Standard Deviation

g

Modelling Aerosol Dispersion and Processes in Urban Street Canyons

B K Tay

Centre of Atmospheric Science

Conclusions

• Using a simplified set-up (2 D flows at steady

state conditions), heterogeneities of flow patterns within a canyon was demonstrated due to the varying extents of advection and turbulent diffusion

• Coupling between street level flux and urban

canopy fluxes

• The numerical model is able to represent

heterogeneities and aerosol processes within a canyon and thus inform and complement measurement campaigns