+ All Categories
Transcript

« Simplified Urban Heat Island Modelling »

By Adil Rasheed

Some Questions

• How does a city development affect the microclimate ?• How important is the effect of environment on buildings ?

– How should building design respond to urban microclimate ?• Is it possible to minimise the energy consumption and improve

pedestrian comfort by designing/modifying certain parameters in a city?

• Is it possible to develop a strategy for city design ?

Overview of the presentation

• Brief description of Urban Heat Island • State of Research• State of research conducted by the student• Future plan of research

Urban Heat Island Effect

Temperature profile in a plain rural area

Urban Heat Island Effect

Temperature profile of a UHI

-0.5 to 7 C

Measured data: London

Measured data: London

Urban Heat Island Effect

• Causes of UHI:– Radiometric

• Albedo, emmisivity– Thermophysical

• Cp, K…– Geometric

• Change in drag and shear– Evaporative and evapotranspiration– Anthropogenic heat

Consequences of UHI

• Changes heating and cooling loads.

• Changes comfort level.• Influences pollutant dispersion.• Can severely affect the cloud

formation and rain– mean monthly rainfall rates within 30-

60 kilometers (18 to 36 miles) downwind of the cities are, on average, about 28% greater than the

upwind region. (research conducted by NASA) http://www.gsfc.nasa.gov/topstory/20020613urbanrain.html

Conclusion: Urban area changes the energy, momentum and humidity balance.

UHI Modelling

• Governing Equations:

Mass

Mom.

Energy

Humidity

TKE

• Comments on the governing equations:

– Highly non linear– Strongly coupled.– Mismatch between the

number of unknowns and number of equations (problem of closure)

UHI modelling

• All the existing models can be categorized into– Analytical Model– Physical Model– Numerical Model

All these models are governed by the same set of conservation equations

Analytical Modelling

• Solving the governing equation with assumptions which simplifies the equation.– Flow over a gaussian mountain– Plume from a constant area source

• Advantages: – Gives exact solution.– Reliable for validation.– Computationally inexpensive

• Disadvantages: – Doesn’t represent reality.– Very limited application.

Sample assumptions:

• No pressure gradients in 2 directions

• Free slip BC near the ground

• No consideration for urban geometry

• Radial Symmetry

Physical Modelling• Requirements:

– Similarity criteria– 3-D model of the city– Hi-tech instrumentation

• Advantages:– More close to reality.– Can handle the complexities of fluid flow.

• Disadvantages:– Difficulties in satisfying the similarity

criteria.– Measuring instruments can alter the flow.– Difficult to maintain the desired boundary

conditions– Effects of radiation, cloud formation etc.

can not be simulated

Re o o

o

U L

v

Pr o o

o

v

k

o o

o o o

v TFr

L g T

2o

o o

UEc

Cp T

3

2

( )sg T T LGr

v

Re o o

o

U L

v

Similarity Criteria

Physical Modelling

Coef. Of thermal expansion

Surface heat flux

LU et al. Dept. Of marine, earth and atmospheric sciences, North Carolina State Univ.

Numerical Modelling

• Involves solving the governing equations numerically.• Advantages:

– Takes into account the key parameters that can affect UHI with (potentially) good spatio-temporal resolution.

• Disadvantages:– Numerically very expensive.– The model is complex and needs experties to run the model.– Error analysis is not feasible at the moment.

Numerical Modelling: Scales

Microscale (buildings resolved)

Mesoscale (Buildings can’t be resolved)

Macroscale

•Microscale

•Grid size is much smaller than the dimension of buildings to be resolved.

•Mesoscale

•Grid is very coarse (a few km)

•Simulation should be run for sufficiently long duration so that the wind sweeps the whole domain

•Macroscale

Present Challanges: Urban and Turbulence parametrisation

Numerical Modelling: Urban Parametrization

City to be modelled.

Numerical Modelling: Urban Parametrization

Fine Grid: Can resolve the effects of buildings but is computationally intractable.

Numerical Modelling: Urban Parametrization

Very coarse Grid: Can’t resolve the effects of buildings explicitly but is computationally feasible. Implicit modelling required(parametrization of urabn effects)

Martilli’s Urban Parametrisation• Highlights:

– Impact of horizontal and vertical wall (drag and shear)

– Accounts for solar radiation– Accounts for building density,

urban forms and different landuse

– 1D heat conduction is solved in the soil, wall and roof to estimate the surface heat fluxes

Urban Grid

Mesoscale Grid

(Source)mesoscale = (urban fraction)*Σ(urban effect on urban grid)+(rural fraction)*(rural effect)+…..

Surface temperature: Computed by solving 1D heat transfer equation

Source

Numerical Modelling: Turbulence

Big whorls have little whorls, Which feed on their velocity;

And little whorls have lesser whorls, And so on to viscosity

-Lewis Richardson in 1922

Numerical Modelling: Turbulence

• Different approaches to make turbulence computationally tractable:

– DNS: Direct Numerical Simulation.

– RANS: Reynolds Averaged Navier Stokes (or time or ensemble)

– LES: Large Eddy Simulation (Spatially average )

DNS3D, unsteady

RANSSteady / unsteady

LES 3D, unsteady

Κ (wave number)

E(k)

Resolved

Modelled

Resolved Modelled

Filter

Turbulence: Direct Numerical Simulation (DNS)

• Resolves all the turbulent length and time scales

• Computationally expensive.

• Can be applied only for low reynolds number flow.

Turbulence: RANS

• Reynolds Averaged Navier Stokes – Predicts only the time averaged effects.– Based on the assumption that the instantaneous fluctuations are

much smaller than the fluctuations in the mean flow.– Since all the turbulent scales are modelled the result may deviate

significantly from reality.– Two additional equations are required to predict the length and

velocity scales.

Large Eddy Simulation (LES)

• Some applications need explicit computation of accurate unsteady fields.

– Bluff body aerodynamics– Aerodynamically generated noise (sound)– Fluid-structure interaction– Mixing– Combustion – …

LES - Rationales

• Large eddies: responsible for the transports of

momentum, energy, and other scalars.

anisotropic, subjected to history effects,

are strongly dependent on boundary conditions, which makes their modeling difficult.

k,f

Small eddies tend to be more isotropic and less flow-dependent (universal), mainly dissipative scales, which makes their modeling easier.

Eu,ET

RANS vs LES

RANS model1

RANS model 2

LES

Historic climate file for some (possibly

distant) locationBuilding

Current situation

Historic climate file for some

(possibly distant) location

RealityDue to large scale flow and topographical features, the climate bounding the city is

different

Urban albedo (sw, lw), evapotranspiration,

anthropogenic gains, momentum transfer

Building

Possible solutionHistoric climate files for some

(possibly distant) locations

Interpolation or macroscale flow

model

Mesoscale flow model with urban parameterisation

Local microscale (simple CFD) flow

modelBuilding

Pre-process

Unidirectional model nesting

Hypothesis

Approach

Microscale Model

Mesoscale ModelGlobal Model + Measured data

Building SimulationProgram

Research done by the candidate

• Co-development and Restructuring of the FVM code.

• Development of the Microscale Model.• Test of various Convective Schemes for

Atmospheric flow.

Co-development and restructuring of the FVM code

iz

j j

zj j

zj j

i

j i

j

j

j

j

jE

iu i u

E

Uk

x x

kx x

Ek

x x

p

x

U U

x

U

x

U

AUU

B

A B

AE

E

t

E

x

t

GBt

Advection

Turbulence

Diffusion

Pressure

Surface

This work was done along with Andrea Krpo from LPAS

0i

i

U

x

Microscale Model

• Features:– Based on SIMPLE algorithm. (Semi

Implicit Method for Pressure Linked Equations)

– Explicit resolution of buildings.– Solves for velocity and scalars in

3D• Problems:

– Based on cartesian grid so can’t be adapted to the terrain

Convective Schemes

• One dimensional fluid flow:

?

Closer to 100 C Closer to 0C

Conclusion: Interpolation schemes should be sensitised to magnitude as well as the direction of flow

100 C 0 C

where

Future plan of research

• Validation of the basic assumptions in Urban Parametrisation• Improvement in the Urban Parametrisation• Coupling with the Microscale Model• Application to real cities

Verification

1. Test the urban canopy model against experimental and LES results.

2. Finetune the RANS model to suite the common geometries occuring in the city by comparing the result against LES for isothermal cases.

3. Repeat the step 1 and 2 for non-isothermal cases.

Improvement of Urban Parametrization

1. Study the effects of changing the urban geometry.2. Introduce the simplified radiosity algorithm to compute

the radiative fluxes accurately.3. Further Verifications.

Numerical Modelling: Urban Parametrization

Can be approximated to a regular array of buildings

Numerical Modelling: Urban Parametrization

May not be well represented by a regular array of buildings.

Numerical Modelling: Urban Parametrization

Parametrization schemes can be defined which cover the entire set of urban characteristics

Coupling of Mesoscale and Microscale Models

Supply the boundary conditions from the Mesoscale Model to the Microsclae Model

Microscale Model

Mesoscale Model

Application of the Model to real cities

• Run the model on cities and investigate – Effects of variables (geometric, surface, sources,

sinks)– Effects of latitude

• Generate annual data sets.• Statistical Reduction of the annual data sets.• First contribution to urban planning guidelines to best

control UHI

Schedule

Thank You


Top Related