Title: Modeling and Simulation of Turbulent Transport of Active and Passive Scalars above Urban Heat Island in Stably Stratified Environment
1Modeling and Simulation of Turbulent Transport of
Active and Passive Scalars above Urban Heat
Island in Stably Stratified Environment
- Albert F. Kurbatskii
- Institute of Theoretical and Applied Mechanics of
Russian Academy of Sciences, Siberian Branch - Russia, Novosibirsk
2 Introduction
- ?For stratified atmospheric flows the LES models
and third-order closure models should be
considered as fundamental research tools because
of their large computer demands. - ?A growing need for detailed simulations of
turbulent structures of stably stratified flows
motivates the development and verification of
computationally less expensive closure models for
applied research in order to reduce computational
demands to a minimum.
3 Objectives
- ?The principal aim of this investigation is the
development of turbulent transport model for the
simulation of the urban-heat-island structure and
pollutant dispersion in the stably stratified
environment.
4Objectives
- ?The algebraic modeling techniques can be used in
order to obtain for buoyant flows the fully
explicit algebraic model for turbulent fluxes of
the momentum, heat and mass. - ? The principal object of this work is the
development of three-four-parametric -
-
- turbulence model minimizes difficulties in
simulating the turbulent transport in stably
stratified environment and reduces efforts needed
for the numerical implementation of model.
5 The mathematical model of the urban heat-island
- ? The penetrative turbulent convection is induced
by the constant heat flux H0 from the surface of
a plate with diameter D. It simulates a prototype
of an urban heat island with the low-aspect-ratio
plume (zi / D 1) under near calm conditions and
stably stratified atmosphere.
6 The mathematical model of the urban heat-island
- ?The mixing height, zi , is defined as a height
where the maximum negative difference between the
temperature in the center of the plume and the
ambient temperature Ta is achieved.
7 The mathematical model of the urban heat-island
?The problem of the development and evolution
turbulent circulation above a heat island is
assumed to be axisymmetric. ? At the initial
moment the medium is at rest and it is stably
stratified.
8 Limitations of Laboratory Measurements for
Full-scale Simulation
- There are important limitations utilized in the
laboratory experiment and simulation of the real
urban heat-island in the nighttime atmosphere - ?Very large heat fluxes from the heater surfaces
- ?Very strong temperature gradients that required
to obtain the low aspect ratios (zi/D) and small
Froude numbers.
9Governing Equations
- Governing equations describing the
turbulent circulation above a heat island can be
written in the hydrostatic approximation at
absence of the Coriolis force and radiation with
use a Boussinesq approximation.
10Governing Equations in RANS-approach
11- Transport Equations
- for heat and mass fluxes
-
12Explicit Algebraic Expressions for Turbulent
Fluxes
- ?The explicit algebraic models for the
turbulent heat flux vector and turbulent mass
vector were derived by truncation of the closed
transport equations for turbulent fluxes of heat
and concentration by assuming weak equilibrium,
but retaining all major flux production terms. - ? For turbulent stresses we applied eddy
viscosity expression.
13- CLOSURE full explicit turbulent fluxes models
for active (heat) and passive (mass) scalars
14Three-Equation ModelE - ?- ??2?
- ?The closure of expressions for the turbulent
stresses and heat flux vector is achieved by
solving the equations for turbulent kinetic
energy, its dissipation rate and temperature
variance, resulting in three-equation model
E-?-??2? ?
15- CLOSURE three-equation model
- for active (heat) scalar
field
16Four-equation modelE - ?- ??2? - ?c??
- ?For the closure of expression for turbulent flux
vector of a passive scalar is involved the
equation for covariance concentration
temperature. - Thus, for the description of a concentration
field is formulated the four-equation model
E-?-??2?-?c?? ? -
-
-
17- CLOSURE four-equation model
- for passive scalar
field
18Boundary Conditions
- The domain of integration is a cylinder of a
given height H. The heated circular disc of
diameter D is located at the center of the
cylinder bottom.
19HEAT TRANSFER BOUNDARY CONDITIONS
- Domain of integration
- is a cylinder
z
- ?At the plume axis and at its outer boundary
- symmetric conditions
- (??/?r) (??/?r) (??/?r) (???2?/?r) 0
are prescribed. - (Ur0 at r 0 and at r 1.8R)
- ?At the top boundary
- the zero-flux condition
- ?V/?z ??/?z ??/?z
- ???2?/?z 0 is enforced.
Top boundary
Heat Flux, H0
r
s o u r c e
20HEAT TRANSFER BOUNDARY CONDITIONS
Z
- ?The surface heat source is placed on the bottom
(z 0) has the size 0 ? r / D ? 0.5. - ?Boundary conditions at the bottom are specified
as - ?heat flux H0 is prescribed
- ?values of E, ? and ??2? at
- the first level above surface are chosen
according to Kurbatskii - (JAM, 2001, vol.40, No.10)
-
- Domain of integration
- is a cylinder
Top boundary
Heat Flux, H0
0
r
s o u r c e
21MASS TRANSFER BOUNDARY CONDITIONS
- ?At the plume axis and at outer boundary,
- (?C/?r) (??c??/?r) 0.
- ?At the top,
-
- ? Constant flux of mass,
-
- is prescribed inside a source.
- ?At the bottom and outside of a source
Z
- Domain of integration
- is a cylinder
-
Top boundary
L 0.5 D
r
mass source
22MASS TRANSFER BOUNDARY CONDITIONS
Z
Domain of integration is a cylinder
Top boundary
?The same boundary conditions are used for source
of small length located at the center of a heat
island.
L0.1D
r
mass source
23MASS TRANSFER BOUNDARY CONDITIONS
Z
Domain of integration is a cylinder
Top boundary
?and the same boundary conditions are used for
source of small length located at the periphery
of a heat island.
L 0.1D
r
mass source
24Numerical Method
Fr , Fz turbulent fluxes of momentum, heat and
mass
Semi-implicit alternating direction scheme
25Numerical Procedure
- ?The numerical method uses a staggered mesh.
- ?The difference equations are solved by the
three-diagonal-matrix algorithm.
- Staggered mesh
- ? ? ?
- ? ? ?
- ? ? ?
-
-
z
?r
?r/2
?z
?z/2
r
0
Ur
Uz
?E, ?, T, lt?2gt, C, ltc?gt
26Main Results of Simulation
- ?The results of simulation correspond to a
quasi-steady state of circulation over an area
heat source in stable stratified environment. - ? Figure (c) shadowgraph picture at t 240 sec
when the full circulation is established.
27(No Transcript)
28Calculation of Normal Turbulent Stresses
- In this problem a simple gradient transport
model preserves certain anisotropy of the normal
turbulent stresses
is turbulent viscosity.
29(No Transcript)
30(No Transcript)
31RESULTS Temperature profiles
- ?Calculated temperature profiles inside the
plume have characteristic swelling - the temperature inside the plume is lower
than the temperature outside at the same height
creating an area of negative buoyancy due to the
overshooting of the plume at the center. - ?This behavior indicates that the plume has a
dome-shaped upper part in the form of a hat.
32(No Transcript)
33(No Transcript)
34Pollutant Dispersion
- Investigation of buoyancy effects on
distribution of mean concentration in mixing and
inversion layers of urban heat island was the
main goal in modeling and simulation of pollutant
dispersion from a continuous surface source.
35Pollutant Dispersion
- ?Experimental measurements were not available
for the quantitative validation of simulation
data. - ?Instead, we present some preliminary results
that illustrate interesting properties of
pollutant dispersion from a continuous source
located inside the urban heat island and on its
periphery.
36(No Transcript)
37(No Transcript)
38(No Transcript)
39Pollutant Dispersion
- ?One characteristic feature is observed in all
three cases. The contaminant penetrates not only
into the inversion layer but even higher beyond
its boundary. - ?This behavior was recently reproduced in the
laboratory experiment of Snyder et al. (BLM,2002,
vol.102, 335-413.). These experimental data
clearly show penetration of the continuous
buoyant plumes into inversion above the
convective boundary layer.
40 Difference in Turbulent Diffusion Between
Active and Passive Scalars 1
- ?Sometimes assumed that in the stratified
atmospheric boundary layer the eddy diffusivity
of heat (KH) is equal to the eddy diffusivity of
contaminant (KC). However, the stratification
causes a larger difference in the eddy
diffusivities between active heat and passive
mass. - ?Indeed, for the ratio of the vertical eddy
diffusivities of heat and mass can be written the
following expression ?
41Difference in Turbulent Diffusion Between Active
and Passive Scalars 2
- 1-D1bg ltcqgt (?C/?z)-1
1-D2 bg
ltq2gt (?Q/?z)-1 - ?If both mass and heat are passive additives
(the buoyancy terms in this expression are
negligible) then it is evident from this
expression that KCKH. - ? It appears that cases for which the largest
deflection of KC/ KH from unit will occur are
when either T or C is acting as a passive
additive. - ? In our case the mass is acting as a passive
additive. -
42Turbulent Fluxes of Active and Passive Scalars
43Ratio of Eddy Diffusivities of Passive Mass to
that Active Heat
44CONCLUSIONS 1
- ?The three-equation model of turbulent transport
of heat reproduces structural features of the
penetrative turbulent convection over the heat
island in a stably stratified environment. - ?This model minimizes difficulties in describing
the non-homogeneous turbulence in a stably
stratified environment and reduces computational
resources required for the numerical simulation.
45CONCLUSIONS 2
- ?The four-equation model for the description of
pollutant dispersion in the stable stratified
atmospheric boundary layer is formulated. - ?Favorable comparison the numerical results of
pollution dispersion from the continuous surface
source above the urban heat island with
laboratory measurements in the convective
boundary layer showing penetration of the
continuous buoyant plumes into inversion above
the convective boundary layer is found.
46The friction velocity u?(r) / wD
- The friction velocity on the underlying
surface can be obtained on the calculated mean
velocity as u? (r) ? ( ?Ur / ?z ).
47Turbulent Velocity Scale
- Turbulent velocity scale Uf was estimated as 1/30
of a mean wind velocity velocity scale wD
of the mean inflow velocity Uf ? 1/30?wD. This
value was used as characteristic scale of the
turbulent velocity for boundary conditions for
E1 and e1 at the first level of a grid above an
underlying surface.
48Numerical Procedure
- ?It took about 2.8104 time steps to drive the
numerical solution to a quasi-steady state. - ? Computations were performed on a mesh with 25
(and 50) points in radial direction. - ? In vertical direction 116 (and 232) mesh points
were used. - ? The time step was chosen so that the numerical
accuracy was preserved.