Академический Документы
Профессиональный Документы
Культура Документы
International Workshop
on Physical Modelling of Flow
and Dispersion Phenomena
ISBN 2-913454-32-1
EAN 9782913454323
Presses universitaires dOrlans
Crdits photos : - Universit dOrlans / Communication / JSL
- David Hall
Proceedings of
International Workshop on Physical Modelling of
Flow and Dispersion Phenomena
August 29-31, 2007. Orlans, France
Scientific committee:
Dr. Sandrine Aubrun. L.M.E., University of Orlans, France.
Dr. Daniele Contini. Institute of Atmospheric Sciences and Climate, CNR, italy.
Dr. David Hall. Envirobods, Ltd. UK.
Prof. Jerry Havens. University of Arkansas, USA.
Dr. David Heist. U. S. Environmental Protection Agency, N Carolina, USA.
Prof. Zbynek Janour. Academy of Sciences, Prag, Czech Republic.
Prof. Tamas Lajos. Budapest University of Technology and Economics, Hungary.
Dr. Bernd Leitl. University of Hamburg, Germany.
Prof. Masaaki Ohba. Tokyo Polytechnic University, Japan.
Prof. Hideharu Makita. Toyohashi University of Technology, Japan.
Dr. Robert Meroney. Colorado State University, USA.
Prof. Alan Robins. University of Surrey, UK.
Dr. Eric Savory. University of Western Ontario, London, Canada.
Prof. Michael Schatzmann. Universiy of Hamburg, Germany.
Prof. Ted Stathopoulos. Concordia University, Montreal, Canada.
Prof. Jeroen Van Beeck. Von Karman Institute, Rhode-St.-Gense, Belgium.
Editor :
Dr. Sandrine Aubrun. L.M.E., University of Orlans, France.
Preface
On behalf of the Scientific Committee, it is my great pleasure to present to you the
proceedings of the international workshop on Physical Modelling of Flow and Dispersion
Phenomena, PHYSMOD 2007, held at Orlans, France, on August 29-31, 2007.
The workshop had 46 submissions and each of these was reviewed by two members of
the Scientific Committee. We have selected 32 as oral presentations and 14 as poster
presentations. Related full-length papers are gathered in these proceedings. I thank the
members of the Scientific Committee for their important contributions to the workshop,
for providing major input to the formulation of the workshop goals and their careful
reviews of submissions.
The objective of PHYSMOD is to bring together the community active in physical
modelling of flow and dispersion in wind tunnels or water channels. It is intended to
discuss, assess and report on the state-of-the art of experimental work in this field,
define directions of future research and encourage wider collaboration between
research institutes. It is intended to provide a wide platform for information exchange
and knowledge transfer, and participating institutions and laboratories are encouraged
to also bring their undergraduate and postgraduate students to present their work and to
incorporate them into the active fluid modelling community.
PHYSMOD 2007 concentrates on the physical modelling of flow and dispersion in the
natural environment, referring mainly to the following topics:
Heat and mass transfer due to atmospheric dispersion in urban areas (also
including heat island problems)
Unsteady properties of the dispersion process
Building effects on the flow characteristics in urban areas
Validation of numerical and analytical modelling methods
Improvement and validation of atmospheric flow and dispersion modelling
techniques
Quality assurance in physical modelling
Use of boundary layer modelling for wind technology
CONTENTS
KEY-NOTE LECTURE
Compiling validation data sets from systematic wind tunnel measurements
requirements and pitfalls ........................................................................................................................ 1
B. Leitl, M. Schatzmann
DENSITY EFFECTS
Effect of Semi-circular Windbreak Array on the Heavy Gas Plume Dispersion in Urban Areas............ 7
B.S. Shiau, Y.C. Wu
Wind tunnel studies of LNG vapor dispersion from impoundments ..................................................... 15
J. Havens, T. Spicer, W. Sheppard
The Effect of Release Time on the Dispersion of a Fixed Inventory of Heavy Gas
A Wind Tunnel Model Study ................................................................................................................. 17
D.J. Hall, V, Kukadia, S. Walker, P. Tily, G.W. Marsland.
QUALITY ASSURANCE
Quality assurance of micro-scale meteorological models Action COST 732.................................... 23
M. Schatzmann, B. Leitl.
How comprehensive is comprehensive enough?
Model-specific reference data for the validation of micro-scale LES flow and dispersion models....... 27
F. Harms, B. Leitl, M. Schatzmann
How dense is dense enough?
Systematic evaluation of the spatial representativeness of flow measurements in urban areas......... 33
D. Repschies, B. Leitl, M. Schatzmann.
How close is close enough?
Sensitivity of wind tunnel results with respect to changing approach flow conditions.......................... 41
I. Herbst, B. Leitl, M. Schatzmann
WIND TECHNOLOGY
Properties of the far wake of a wind turbine in an atmospheric boundary layer .................................. 47
G. Espaa, S. Aubrun, P. Devinant, L. Laporte, E. Dupont.
Aerodynamic Design of the Princess Elizabeth Belgian Antarctic Research Station........................... 53
J. Sanz Rodrigo, C.Gorle, J. Van Beeck, P.Planquart.
Feasibility study of tests on scale models for the evaluation of the overpressures induced
by a passing train on adjacent structures............................................................................................. 59
L. Procino, G. Bartoli, C. Borri, A. Borsani.
Wind Tunnel Simulations of Pollution from Roadways......................................................................... 63
D. K. Heist, S. G. Perry, L. A. Brixey, G. E. Bowker.
URBAN FLOWS
How rough is rough?
Characterization of turbulent fluxes within and above an idealized urban roughness ......................... 69
M. Schultz, B. Leitl, M. Schatzmann.
Effect of roofshape on unsteady flow dynamics in street canyons....................................................... 75
J. Barlow and B. Leitl.
Study of Flow Fields in Asymmetric Step-Down Street Canyons......................................................... 79
B. Addepalli, E.R. Pardyjak
Spanwise variation of drag on roughness elements
in a nominally two-dimensional boundary layer.................................................................................... 87
P. Hayden, T. Mapurisa and A.G. Robins
CFD analyses of flow in stratified atmosphere ..................................................................................... 93
G. Kristf , N. Rcz,, M. Balogh
POSTERS
Use of detection of coherent flow structures
for better understanding of 3D flow fields in urban environment........................................................ 185
T. Rgert, I. Goricsn, M. Balcz, K. Czder, T. Lajos
Air-quality and spatial planning .......................................................................................................... 191
FL.H. Vanweert,. J.I.J.H. van Rooij
Charateristics of the lowspeed wind tunnel of the LMF, France ...................................................... 197
L. Perret
Evaluation of pollution dispersion prediction using RANS
with turbulence models available in FLUENT 6.3............................................................................... 201
R. Izarra-Garcia, J. Franke
Flow and dispersion around tall buildings .......................................................................................... 203
T. Lawton, A. Robins
Quantifying the temporal representativeness of flow
and dispersion measurement in a complex urban area ..................................................................... 207
M. Rix, M. Schtzmann, B. Leitl.
How often is sufficient? A program for the statistical analysis
of puff dispersion in urban environments ........................................................................................... 213
R. Fischer, M. Schtzmann, B. Leitl
i}i>}}-i
>i*6-i
i`>}>i>``
>iy\VVV>`i`VV
*6>`}-ii`*6
/}>V*6
>i>}}-i
->>`*>Vi>
>v>i>`
LVii
-v>Vi>}}
-i-i
>ii6iVi
*>iiivii
LAVISION GMBH
ANNA-VANDENHOECK-RING 19
37081 GOETTINGEN
TSI Europe
Call us:
TSI France
Europarc Bt D, 26 rue John Meynard Keynes, Technople de Chteau-Gombert, 13013 MARSEILLE
Tl : +33 (0)4.91.95.21.90 - tsifrance@tsi.com - Site internet : www.tsiinc.fr
Imaging Systems
- Global flow
measurements
- Combustion
diagnostics
- Particle size
measurements
Fluid Dynamics
Optimisation of product
design and combustion
Aerodynamics and
Solid Mechanics
Optimisation of
materials and
components
Indoor Climate
- Thermal Comfort
- Air Quality
- Ventilation Efficiency
For improved
quality of life
Center for Marine and Atmospheric Research, Environmental Wind Tunnel Laboratory
University of Hamburg,
Hamburg, Germany
bernd.leitl@zmaw.de
Abstract - Laboratory data are often used as a
reference for validating micro-scale flow and
dispersion models. In contrast to field data,
carefully generated test data sets compiled from
systematic wind tunnel tests provide a number of
advantages regarding the consistency and accuracy
of the test data, the completeness of a dataset or
concerning the documentation of the boundary
conditions of an experiment. Based on more than
10 years of experience in generating validation data
at the Environmental Wind Tunnel Laboratory
(EWTL), the paper is illustrating common pitfalls
and problems related to the compilation of
reference data sets.
From a wider, projectindependent perspective it is intended to define
minimum modeling quality requirements in order to
further promote boundary layer wind tunnels as a
source of reliable reference data for atmospheric
flow and dispersion modeling.
Key words physical modeling, atmospheric flow and
dispersion, quality assurance, modeling standard.
Introduction
Physical modeling of environmental flow and
dispersion phenomena is providing reliable and detailed
information on wind driven pollutant transport in built-up
areas for several decades. Whereas wind tunnel
modeling was used for almost any kind of environmental
flow and dispersion problem within the atmospheric
boundary layer in the past, nowadays a large fraction of
environmental flow and dispersion phenomena can be
modeled numerically with accuracy sufficient for many
practical applications. However, physical modeling in a
boundary layer is still the preferred choice if local-scale,
transient flow and dispersion phenomena in the range of
a few hundred meters and at time scales in the order of
minutes are a matter of particular interest. Based on
careful and diligent physical modeling and with the help
of state-of-the-art instrumentation, information and data
can be provided with a temporal and spatial resolution
which cannot yet be achieved by numerical modeling.
The high resolution of wind tunnel test data acquired
under controlled boundary conditions also enables
validation data to be compiled from systematic wind
tunnel tests. Based on systematic reference data, the
quality of different types of numerical models can be
evaluated and the 'fitness for purpose' for specific model
applications can be tested.
In general, the availability of test data and their
quality and accuracy significantly affect the outcome of
model validation procedures. In a strict sense, the
results of a model evaluation are as good or as
uncertain as the reference data used (Britter and
Schatzmann, 2006). Depending on the point of view
2 Methodical Requirements
2.1 General Remarks
References
Britter, R. and
Schatzmann, M. (Ed.) (2006).
"Background and justification document to support the
model evaluation guidance and protocol", COST
Action 732, Quality Assurance and Improvement of
Microscale Meteorological Models, ISBN 3-00-0183124.
Leitl, B.; Chauvet, C. and Schatzmann, M. (2001).
"Effects of Geometrical Simplification and Idealization
on the Accuracy of Microscale Dispersion Modeling"
Proc. 3rd Int. Conf. Urban Air Quality, March 19-23
2001, Loutraki, Greece
Snyder, W.H. (1981) "Guideline for Fluid Modeling of
Atmopsheric Diffusion" US EPA, Environmental
Sciences Research Laboratory, Research Triangle
Park, NC 27711
VDI 3783/12 (2000) "Environmental Meteorology
Physical modeling of flow and dispersion processes in
the atmospheric boundary layer, Application of wind
tunnels" Verein Deutsche Ingenieure VDI, Handbuch
Reinhaltung der Luft, Band 1B (in German & English)
Liedtke, J; Leitl, B.; Schatzmann, M. (1998) "Car
exhaust dispersion in a street canyon Wind tunnel
data for validating numerical dispersion models" Proc.
2nd East European Conference on Wind Engineering,
Prague, 7-11 September 1998, vol. 1, pp. 291-297
Schatzmann, M.; Bchlin, W.; Emeis, S.; Khlwein, J.;
Leitl, B.; Mller, W.J., Schfer, K.; Schlnzen, H.
(2006) "Development and validation of tools for the
implementation of european air quality policy in
Germany (Project VALIUM)" Atmos. Chem. Phys., vol.
6; pp. 3077-3083
Herbst, I.; Leitl, B.; Schatzmann, M. (2007) "How close
is close enough? Sensitivity of wind tunnel results
with respect to changing approach flow conditions"
Physmod2007, ibid.
Aubrun, S.; Leitl, B. (2004) "Development of an
improved physical modelling of a forest area in a wind
tunnel" Atmospheric Environment, vol 38, 2004, pp.
2797-2801
Meroney,R.N.;
Pavageau,
M.;
Rafailidis,
S.;
Schatzmann, M. (1996) "Study of line source
characteristics for 2-D physical modelling of pollutant
dispersion in street canyons". J. Wind Eng. Ind.
Aerodyn. 62 (1996), pp. 3756.
Pascheke, F.; Leitl, B.; Schatzmann, M. (2003)
"EVALUATION OF FIELD TRACER EXPERIMENTS
WITH RESPECT TO COMPLEX URBAN TYPE
BOUNDARY CONDITIONS". In Manfrida, Giampaolo
and
Contini,
Daniele,
Eds.
Proceedings
PHYSMOD2003: International Workshop on Physical
Modelling of Flow and Dispersion Phenomena, PratoItalia.
Yi-Ching Wu2
Department of Harbor and River Engineering, National Taiwan Ocean University, Keelung 202, Taiwan
E-mail: b0085@mail.ntou.edu.tw
2 Experimental Set-up
The experiments were carried out in the National
Taiwan Ocean Universitys Environmental Wind Tunnel.
The wind tunnel test section has a cross section of 2 m
wide by 1.4 m high, and 12.5 m long. The tunnel is an
open suction type and it contracts to the test section with
an area ratio of 4:1. The turbulence intensity of empty
tunnel in test section is less than 0.5 % at the mean
velocity of 5 m/s.
Four spires of 100 cm height and cubic elements (5 cm
x 5 cm x 5cm) are properly arranged as the roughness at
the entrance of test section to generate a thick turbulent
boundary layer which is used as the approaching flow.
An X-type hot-wire incorporating with the TSI IFA-300
constant temperature anemometer was employed to
measure the turbulent flow signals. Output of the analog
signals for turbulent flow was digitized at a rate of 4 K Hz
each channel through the 12 bit Analog-toDigital
converter. Since none of the analog signals containing
significant energy or noise above 1 K Hz, with the Nyquist
criteria, a digitizing rate of 2 K Hz was sufficient. The low
pass frequency for the analog signals is set as 1 K Hz in all
runs of the experiments.
Semi-circular windbreaks were arranged in array with
different opening ratios. The windbreak array was placed
four times of spill height (W=4hs) behind the spill source.
The spill height is the same as the windbreak height (hs=H).
Fig.1 is the schematic diagram of the arrangement of
windbreak array in the experiments.
Carbon dioxide (CO2) is chosen as tracer of heavy gas,
1 Introduction
Many kinds of chemical gases storage tanks exist in
the industrial park. The chemical gases are almost toxic
and they are heavier than the air. These kinds of gases are
usually called heavy gases or dense gases. Accidents of
continuous spill of heavy gas from the storage tanks will be
dangerous and have a strong impact on the near
environment of air quality.
Studies on the heavy gas, such as Schatzmann [1, 2],
he indicated that the dispersion of the heavy gas is quite
different from that of the airborne pollutants. Britter [3] had
made a review on the atmospheric dispersion of heavy gas.
Zhu et al. [4] investigated experimentally on the flow
structure within a Dense Gas Plume. Robin et al. [5]
conducted the wind tunnel study on the dense gas
dispersion in a neutral boundary layer over a rough
surface. Donat and Schatzmann [6] had conducted
experiments on the one single-phase heavy gas jets
released in the turbulent boundary layer. Nielsen et al. [7]
made the field tests on the dispersion of pressure liquefied
ammonia. Khan and Abbasi [8] developed the numerical
model to simulate the heavy gas dispersion on the basis of
F1{x, y , z , g , hs , Ds , H ,U s , U s , P s ,U a ( hs ), n, Z ref ,
U a , Pa , P}
(1)
F2 {
x y z Us
, , ,
, n, P, Fr, Re s , Re}
H H H Ua
(2)
CQ
U s Ds H
(3)
hs
( y y ) C ( x, y, z)dy
C ( x, y, z)dy
2
V 2y
(4)
yC ( x, y, z)dy
C ( x, y, z)dy
(5)
( z z) C ( x, y, z)dz
C ( x, y, z)dz
(6)
where y
V z2
C ( x, y, z)dy
zC ( x, y , z )dy
where z
(7)
U (Z )
U ref
Z n
)
Z ref
(8)
n=0.222
Zref=100cm
U(Z)/Uref=(Z/Zref)n
0.8
X=10m,Y=1m
n=0.222
0.8
0.6
Z/Zref
0.6
0.4
Z/Zref
0.2
0.4
0.2
0.004
u'w'/Uref2
0.006
0.008
0
0.4
0.6
U(z)/Uref
0.8
0.002
u rms
U
(9)
X/H=25
P=50%
P=70%
P=85%
Z/H 2
4
X/H=8.333
P=50%
P=70%
P=85%
Z/H 2
40
90
140
190
240
X/H=41.667
P=50%
P=70%
P=85%
3
0
40
80
120
160
200
240
Z/H 2
4
X/H=11.667
P=50%
P=70%
P=85%
Z/H 2
40
90
140
190
240
X/H=58.333
P=50%
P=70%
P=85%
3
0
40
90
140
190
240
Z/H 2
4
X/H=16.667
P=50%
P=70%
P=85%
Z/H 2
40
90
140
190
240
0
40
90
140
190
240
10
X/H=25
Fr=30
Fr=22
Fr=16
Z/H 2
4
X/H=8.333
Fr=30
Fr=22
Fr=16
0
0
100
Z/H 2
200
300
X/H=41.667
Fr=30
Fr=22
Fr=16
0
0
50
100
150
200
250
300
Z/H 2
X/H=11.667
Fr=30
Fr=22
Fr=16
0
0
100
Z/H 2
200
300
X/H=58.333
Fr=30
Fr=22
Fr=16
0
0
100
200
300
Z/H 2
X/H=16.667
Fr=30
Fr=22
Fr=16
0
0
Z/H 2
100
200
300
0
0
100
200
300
11
0.5
0.48
200
Fr=16
0.46
P=50%
P=70%
P=85%
160
sigmaZ/H
0.44
Kground 120
0.42
Fr=16
P=50%
P=70%
P=85%
0.4
80
10
20
30
40
X/H
50
60
40
0
10
20
30
X/H
40
50
60
2.8
300
2.6
P=50%
Fr=30
Fr=22
Fr=16
sigmaY/H
2.4
200
2.2
Kground
Fr=16
P=50%
P=70%
P=85%
100
10
20
30
X/H
40
50
60
0
0
10
20
30
X/H
40
50
60
5 Conclusion
12
6 References
[1] Schatzmann, M., (1993), Experiments with Heavy Gas
Jets in Laminar and Turbulent Cross-Flows,
Atmospheric Environment, Vol. 27A, pp. 1105-1116.
[2] Schatzmann, M., (1995), Accidental Release of Heavy
Gases in Urban Areas, Wind Climate in Cities,
Cermak, J.E., et al. (Eds.), pp. 555-574.
[3] Britter, R.E., (1989), Atmospheric Dispersion of Dense
Gases, Annual Review of Fluid Mechancis, Vol.21,
pp.317-344.
[4] Robins, A., Castro, I., Hayden, P., Steggel, N., Contini,
D., and Heist, D., (2001), A Wind Tunnel Study of
Dense Gas Dispersionin a Neutral Boundary Layer
over a Rough Surface, Atmospheric Environment,
Vol.35, pp. 2243-2252.
13
15
djhall@envirobods.co.uk ; kukadia
Introduction
djhall@envirobods.co.uk ; kukadiav@bre.co.uk
Introduction
Risk assessment studies for the dispersion of
accidental releases of toxic and flammable gas clouds
are based either on direct experimental data or, more
frequently, on numerical models derived from this sort
of data. Though there is a body of large scale
experimental field data of this sort, most experimental
data is obtained using small scale models in wind
tunnels and water channels, providing both generic
studies and specific site investigations. The majority
of these measurements cover two types of release,
instantaneous (see, for example, Spicer and
Havens(1984) and Meroney and Lohmeyer(1983))
and continuous (see, for example, KoningLanglo
and Schatzmann(1991)).
Instantaneous release
models mostly use the sudden exposure of a fixed
volume of gas, initially contained in a tent of some
sort, to the prevailing wind conditions. Though the
release is instantaneous, it is also passive in the
sense that it has no initial energy, following the
Thorney Island type of field trial (Mcquaid and
Roebuck(1983)). Continuous releases are those for
which the release time is long enough for the gas
plume to have reached a steady state condition with
no further development. In the case of heavier-thanair gas clouds this time can be considerable, the
equivalent of hours in large scale releases.
17
18
Acknowledgments
The work described here was jointly funded by the UK
Health and Safety Executive and the European
Commission as part of the FLADIS (CT90-0125)
project in the Major Technical Hazards Programme.
The experiments were carried out a the Warren
Spring Laboratory prior to its closure and the data
analysis and report completed at the Building
Research Establishment, to which two of the authors
and the dispersion wind tunnel transferred.
References
Hall D.J. Kukadia V., Walker S., Tily P, Marsland
G.W.(2006) The Effects of release time on the
dispersion of a fixed inventory of heavier-than-air
gas A wind tunnel model study. Building
Research Establishment, UK.
Report No.
CR149/96.
Hall D.J. and Waters R.A.(1989). Investigation of Two
Features of Continuously Released Heavy Gas
Plumes. Warren Spring Laboratory, UK. Report
No.LR707(PA).
Havens J.A., Spicer T.O.(1983). Gravity Spreading
and air Entrainment by Heavy Gases
Instantaneously Released Into a Calm
Atmosphere.
IUTAM
Symposium
on
Atmospheric Dispersion of Heavy Gases and
Small Particles, Delft, The Netherlands, August
29-September 2cnd. Springer-Verlag. ISBN 3540-13491-3.
Konig-Langlo G., and Schatzmann M.(1991). Wind
Tunnel Modelling of Heavy Gas Dispersion.
Atmospheric Environment, Vol. 25A, No. 7, pp.
1189-1198.
McQuaid J.D., Roebuck B.(1984). Large Scale Field
Trials on Dense Vapour Dispersion. Final Report
to Sponsors on Heavy gas Dispersion Trials at
Thorney Island, 1982-1984. Health and Safety
Executive, Sheffield, UK.
Meroney R.N., Lohmeyer A.(1983). Statistical
Characteristics of Instantaneous Dense Gas
Clouds Released in an Atmospheric Boundary
Boundary
Layer
Layer
Wind
Tunnel.
Meteorology. Vol. 28. pp. 1-22.
19
Gas
g /a
/a
U
m s-1
Air
0.97
0.5
Argon
1.38
0.38
Argon
1.38
BCF/Air
50/50
5
10
Ri
0.3
10
30
0.77
3.4
#
2.31
0.38
B
7.7
0.034
#
23.1
0.0037
B
77
0.0003
231
0.00004
0.97
0.77
4.7
2.31
0.52
7.7
0.047
23.1
0.0052
77
0.0005
231
0.00005
0.38
0.69
0.55
9.12#
1.64
1.03B
5.47
0.091
16.4
0.010
54.7
0.0009
164
0.00001
3.37
2.37
1.207
0.96
7.3
2.87
0.82
9.6
0.073
28.7
0.0082
96
0.0007
287
0.00008
BCF
5.74
4.74
1.08
0.86
15.5
2.57
0.1.74
8.57
0.15
25.7
0.017
85.7
0.0016
257
0.00017
BCF
5.74
4.74
0.77
0.61
31#
1.83
3.42
-
6.1
0.31
18.3
0.034
61
0.0031
183
0.0003
20
21
Figure 4. Contours of dimensionless distance (X/L) to 1% and 10% maximum (left) and mean (right) concentration
as a function of release Richardson number and release time.
Figure 5. Plots of maximum ground level gas cloud concentration against distance for four Richardson numbers
and all gas cloud release times.
22
Introduction
The emergence of increasingly powerful computers
enabled the development of tools that have the potential
to predict flow and transport processes within the urban
canopy layer. These new tools are micro-scale
meteorological models of prognostic or diagnostic type.
Prognostic models are based on the Reynolds-averaged
Navier-Stokes (RANS) equations, whereas diagnostic
models are less sophisticated and ensure only the
conservation of mass. These two model types are
presently supplemented by even simpler engineering
tools. It is to be expected, however, that the latter will
sooner or later be replaced by Computational Fluid
Dynamics (CFD) tools such as RANS codes or the even
more complex Large Eddy Simulation (LES) models.
Micro-scale meteorological models are special in so
far as they are tailored to the needs of meteorologists.
They are adjusted to domain sizes of the order of
several decametres to a few kilometres (street canyons,
city quarters). They usually use boundary conditions
based on surface characteristics like land use,
roughness and displacement thickness and they may
contain modules that have the potential to simulate
chemical transformations, aerosol formation or other
important atmospheric physico-chemical processes.
Typically these models contain a substantial amount of
empirical knowledge, not only in the turbulent closure
schemes but also in the use of wall functions and in
other parameterisations.
Models play an important and often dominant role in
environmental assessment and urban climate studies
that are undertaken to investigate and quantify the
effects of human activity on air quality and the local
climate. Their increasing use is paralleled by a growing
awareness that the most of these models have never
been subject to rigorous evaluation. Consequently there
is often a lack of confidence in the modelled results.
23
24
25
Future Work
Europe-wide discussion of the quality assurance
procedure, the use of specific data sets and the
recommendations given in the Best Practise Guideline
will lead to a harmonised and accepted approach. A
quality assurance activity will be launched and the
community of model developers and users will be invited
to apply the procedure to their models and to prepare
model evaluation protocols based on selected data sets.
This will be combined with a model inter-comparison
exercise within which several model developers and
users will simulate identical cases. Ideally this exercise
should comprise test cases for which the solution is not
known beforehand (blind testing). The intent is not to
pillory models that perform badly or to rank the models
in one way or the other. That only blocks the flow of
information and obstructs scientific exchange. The
differences in model results should be discussed and
the reasons for deviant model results should be
investigated. The strengths and weaknesses of
particular modules, parameterisations or closure
schemes will be determined. It is expected that
modellers will take this opportunity to test various
modules, develop common views about the most
appropriate set-up of micro-scale meteorological models
and, thereby, the quality standard of micro-scale
meteorological models and their application will
significantly improve. This leads to the expectation
expressed in the Introduction that the 'culture' within
which urban air pollution models are developed and
applied will be significantly improved.
References
Britter, R., and Schatzmann, M. (Eds.) (2007a):
Background and justification document to support
the model evaluation guidance and protocol
document. COST Office Brussels, ISBN 3-00018312-4.
Britter, R., and Schatzmann, M. (Eds.) (2007b): Model
evaluation guidance and protocol document. COST
Office Brussels, ISBN 3-00-018312-4.
Borrego, C., and Tchepel, O. (1999): General
Requirements for a Quality Assurance Project Plan.
Proceedings, 3rd SATURN Workshop, Aveiro,
Portugal.
Franke, J., Hellsten, A., Schlnzen, H., and Carissimo,
B. (Eds.) (2007a): Best Practice Guideline for the
CFD simulation of flows in the urban environment.
COST Office Brussels, ISBN 3-00-018312-4.
Hall, R.C. (Ed.) (1997): Evaluation of modelling
uncertainty - CFD modelling of nearfield
atmospheric dispersion. EU Project EV5V-CT940531, Final Report. WS Atkins Consultants Ltd.,
Woodcote Grove, Ashley Road, Epsom, Surrey
KT18 5BW, UK.
Irwin, J.S. (1998): Statistical Evaluation of Atmospheric
Dispersion Models. Proceedings, 5th Int. Conf. on
Harmonization within Atmospheric Dispersion
Modelling for Regulatory Purposes, Rhodes,
Greece.
Irwin, J.S. (1999): Effects of Concentration Fluctuations
on
Statistical
Evaluations
of
Centreline
Concentration
Estimates
by
Atmospheric
Dispersion Models. Proceedings, 6th Int. Conf. on
26
Introduction
The increasing possibility of an accidental or
deliberate release of chemical/biological/radiological
pollutants has caused detailed simulations of releases
and dispersion in complex geometries to attain a much
higher importance. For example, a credible and
validated computational fluid dynamics (CFD) based
plume prediction model of CT in urban areas can be
used in the licensing of new industrial plants, in safety
analysis studies for accidental releases of hazardous
materials in the chemical industry, or in the context of
crisis management after terrorist attacks in urban
environments.
Numerical simulations of dispersion in an urban
environment are of complex nature and need a
comprehensive computational effort. Hence simpler
models as, e.g., analytical (Gaussian) models,
diagnostic models (which use only the mass
conservation equation) or CFD-models with full
parameterization of turbulence, i.e., Reynolds-averaged
Navier-Stokes (RANS) codes were typically used in the
past for these complex tasks. Such models can
compute urban dispersion within a reasonable time, but
they are unable to capture the inherently unsteady
plume dynamics driven by urban geometry. Direct
numerical simulations (DNS) are simulations in
computational fluid dynamics in which the NavierStokes equations are numerically solved without any
turbulence model. These models are able to compute
transient flow dynamics, but they are prohibitively
expensive for most practical flows at moderate-to-high
Reynolds numbers, and especially so for urban CT
studies. LES constitutes an effective intermediate
approach between DNS and the RANS methods
27
2 Experimental setup
The wind tunnel measurements were carried out in
the large atmospheric boundary layer wind tunnel
facility (WOTAN) at the Meteorological Institute of
Hamburg University. The 26 m long facility provides an
18 m long test section equipped with two turn tables
and an adjustable ceiling. The cross section of the
tunnel measures 4 m in width and 2.75 to 3.25 m in
height, depending on the position of the adjustable
ceiling. For precise probe positioning, the test section of
the tunnel is equipped with a computer controlled
traverse system with a positioning accuracy of better
than 0.1 mm on all 3 axes. The 1:300 scaled model of
the central business district (CBD) of Oklahoma City
was covering an area of approximately 1.6 km x 1.6 km
around the city centre in full-scale. Figure 1 shows a
downwind view of the wind tunnel model mounted on
the second turn table in the test section.
28
29
200
c [ppm]
100
50
610
615
620
wt time [s]
625
150
630
30
300
ontime : wt 0.288s
ontime : wt 0.493s
ontime : wt 0.689s
ontime : wt 0.991s
ontime : wt 1.998s
250
200
150
100
1.5
50
1
0.5
0.5
1.5
wt time [s]
2.5
run Nr.
Finally, Figure 9 shows the results of a Reynoldsnumber or wind speed test. The puff measurements
were repeated at different approach wind speeds while
all other experimental parameters were kept constant.
The reference wind speed was measured in a height of
80 meters (full scale) upstream of the model area. The
test documented that the results of puff measurements
are scaleable also regarding to the mean wind speed.
EWTL@ ZMAW.DE University of Hamburg
200
180
dosage/ontime
160
140
120
100
1.4
1.2
1
0.8
60
40
20
0
0.6
0.4
0.2
0
80
50
100
150
31
4 Conclusions
The validation of LES based flow and dispersion
models requires well-defined data sets with information
adequate for defining and evaluating unsteady
simulations. For these datasets flow and dispersion
measurements with a sufficiently high resolution in time
and space are essential.
To meet the validation data requirements,
comprehensive wind tunnel measurements were carried
out in the large boundary wind tunnel (Wotan) of the
Meteorological Institute at the University of Hamburg.
Acknowledgements
The results presented above are compiled in a joint
research project of the Navy Research Laboratory and
the University of Hamburg, sponsored by DTRA.
References
Sagaut, P. (2004) Large Eddy Simulation for
Incompressible Flows, 2nd Edition, Springer.
Allwine, K. J., M. J. Leach, L. W. Stockham, J. S. Shinn,
R. P. Hosker, J. F. Bowers and J. C. Pace, (2004):
Overview of Joint Urban 2003 an atmospheric
dispersion study in Oklahoma City, Symp. on
Planning, Nowcasting and Forecasting in the Urban
Zone, January 11-15, Seattle, WA. Amer. Meteor.
Soc.. J7.1.
Herbst, I.; Leitl, B.; Schatzmann, M. (2007) How close is
close enough Sensitivity of wind tunnel results
with respect to changing approach flow conditions",
Physmod 2007, ibid.
Rix, M.; Schatzmann, M; Leitl, B. (2007) "How long is
long enough? - Quantifying the temporal
representativeness of flow and dispersion
measurements in a complex urban area", Physmod
2007, ibid.
Leitl, B. 2000 Validation Data for Microscale Dispersion
Modeling, EUROTRAC Newsletter, 22, 28-32
32
Abstract For estimating the spatial representativeness of locally measured wind velocities in an
urban roughness structure, wind tunnel experiments were carried out in a dense array of idealized
obstacles. In a subunit of the cube field horizontal
and vertical planes were selected. At a large number
of points located in these planes time series of the
wind velocity vector were measured and subsequently analyzed. For several parameters characterizing the flow areal averages based on all values in
a plane were determined. Subsequently the number
of data points included in the averaging procedure
was systematically reduced by taking only every
second, third (and so on) data point into account.
The deviations were determined and used to define
a metric which quantifies the spatial representativeness of measurements based on a limited number
of data points only.
Key words wind tunnel experiments, cube fields, velocity time series, measurement of planes, reduction in
number of measuring points, spatial representativeness
Introduction
Pollutant transport within the urban canopy layer is
strongly affected by the individual geometrical structure
of the building array. Both mean velocity and turbulence
vary from point to point, local differences can be large. A
full documentation of the flow field within a city would
require measurements taken at numerous positions. To
obtain a complete picture of the flow field by field measurements is more or less impracticable and could not
be afforded. In reality, measurements are taken at selected points only, and the question arises how representative point-wise measurements are to characterize
the flow field within urban quarters or individual streets.
To obtain such an estimate wind tunnel experiments
were carried out in Hamburg Universitys large boundary
1 Methodology
The two obstacle arrays which were used in our investigations are shown in figures 2 and 3. Clearly, these
are not yet really urban geometries. Main advantage of
these idealized structures is, however, that measurements taken in a small subunit of the two cube arrays
can be expected to be representative for other similarly
chosen subunits as well and thus for the complete array.
To create the needed data base, measurements with
high spatial resolution in horizontal and vertical planes
were carried out. The horizontal planes were situated at
different heights and the vertical planes were chosen at
different horizontal positions within a subunit to get information about the local variability.
For each of the planes the areal average for a variety of flow properties (mean wind velocity components
u , v , w , the standard deviations u rms , v rms , w rms and the
mean turbulent fluxes u' v' , u' w' ) was calculated based
on data from all positions in the plane at which measurements were taken. Subsequently the ensemble of
data points incorporated in the averaging process was
reduced by systematically increasing the distance between points from which data were taken into account.
To obtain a quantitative indicator for the spatial (or more
exact: areal) variability of a flow property a metric was
defined (chapter 3) which, in a specific way, compares
the average over all data points with that over a subset
of points only.
In order to smooth the results and since also in a
laboratory experiment the number of measurements
which can be performed is limited, a plot software (Table
Curve 3D ) was
utilized
to
interpolate
the
effectively
measured
data
points (figure 4) to
an
even
larger
1
number of points
(figure 5) before the
variability indicator
0
2,625m
y/h
y/h [-]
14.625m
umean [m/s]
2.6
2.2
1.8
1.4
1
0.6
0.2
-0.2
-0.6
-1
-1
-2
x/h
Figure 1: Cube field (configuration 1) in the Environmental Wind Tunnel Laboratory (EWTL) with the used
-3
coordinate system and the field of view.
-2
-1
0
max. windspeed 20 m/s
field of view
33
blower (130kW) 1
x/h [-]
y/h [-]
umean [m/s]
2.6
2.2
1.8
1.4
1
0.6
0.2
-0.2
-0.6
-1
-1
-2
-3
-2
-1
x/h [-]
34
2 Experimental Setup
The idealized obstacle area was accomplished by
cubes of the length h = 125mm which were placed at a
distance of 1h to each other (for configuration 2 the
cube height was 1h as well but roofs were mounted on
top of some of the cubes which increased the total
height of those obstacles to 1.5 h). The whole cube field
consisted of 59 rows with 11 cubes per row. The first 39
rows served to generate a turbulent boundary layer,
whereas the last 12 rows were installed to prohibit any
disturbing influences from the downstream end of the
array.
Figure 2 shows the cube field of configuration 1,
whereas in figure 3 the cube field with added roofs is
displayed.
The area in which the intensive observations were
carried out is in the center of the block cube array in
Figure 2 (rows 46 to 48). The exact position of the measurement planes may be taken from figures 6 to 9.
Information on the distribution of roof shape patterns is
also given in these figures, the symbols are explained:
Figure 10.
The coordinate system (Cartesian) consistently used
throughout the study is as follows: The x-axis shows into
the direction of the wind, y is the lateral and z the vertical coordinate.
Using a 2d-Laser-Doppler-Anemometer, time series
of the velocity were recorded in horizontal planes at the
heights of 0.25h, 1h, 1.5h for configuration 1 and 0.64h,
1h, 1.5h and 2h for configuration 2, respectively (h =
cube height without roofs). The shapes of the measurement areas differ for the two configurations as can be
seen in figures 6 and 7 (darkened areas).
The vertical planes were positioned either at the
edges or in the center between two cubes and oriented
either in the x-z or y-z direction.
y/h
0
-1
-2
-1
x/h
y/h
0
-1
-2
-3
-2
-1
x/h
y/h
without roof
pitched roof in
flow direction
flat roof
x/h
The whole experimental program comprised measurements in 18 planes at altogether 2080 individual locations.
3 Data handling
y/h
x/h
35
100(u r u a )
for u r > u a
| ur |
|delta u mean |=
100( u a u r )
for u r < u a . (2)
| ur |
(1)
60 number of points
60 number of points
40
30
20
1060
250
120
100
70
60
30
24
20
14
12
10
8
6
5
4
3
2
50
50
40
30
20
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
10
10
0
0
0
0.2
0.4
0.6
point-distance [h]
0.2
0.4
0.6
0.8
point-distance [h]
Figure 11: Results of a complete analysis of data points
at the example of u .
0.8
4 Results
Figure 12 showed that the area average for a selected flow property may be different for the same data
point density but a data point selection from a different
raster. To elucidate this fact even more, the total area of
the horizontal planes according to figure 6 is subdivided
into two sub-domains located in different parts of the
canopy (along-canyon and cross-canyon areas in figure
13).
36
0.6
point-distance [h]
0.8
0.6
point-distance [h]
0.8
0.6
point-distance [h]
0.8
60 number of points
40
30
20
50
50
60 number of points
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
30
20
10
10
0
0
0.2
0.4
0.6
point-distance [h]
0
0
0.8
0.2
0.4
0.6
point-distance [h]
0.8
40
30
20
60 number of points
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
50
50
0
0
40
30
20
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
10
10
37
40
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
0.2
0.4
0.6
point-distance [h]
0.8
0
0
0.2
0.4
0.6
point-distance [h]
0.8
Figure 16: Results for the plane at 1h in the area crosscanyon (left) and along-canyon (right) for
the wind velocity u (configuration 1).
30
20
10
40
30
20
0.2
0.4
0.6
point-distance [h]
0
0
0.8
0.2
0.4
0.6
point-distance [h]
40
30
20
50
50
10
30
20
30
20
10
0.2
0.4
0.6
point-distance [h]
0.8
0
0
0.2
0.4
0.6
point-distance [h]
0.8
1500
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
10
0
0
40
40
1500
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
0.8
60 number of points
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
20
0
0
30
50
10
10
0
0
40
60 number of points
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
40
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
50
60 number of points
1060
250
120
100
60
40
30
24
20
14
12
10
8
6
5
4
3
2
50
50
60 number of points
60 number of points
0.2
0.4
0.6
point-distance [h]
0.8
0.2
0.4
0.6
point-distance [h]
0.8
5 Conclusions
An extensive study has been carried out to investigate the local variability of flow properties within and
above the urban canopy layer. From the numerous results obtained (Repschies, 2006) only a small selection
could be presented here.
38
200
150
250
250
500
375
100
80
40
30
20
15
12
8
6
3
2
1
100
50
0
0
200
150
500
375
100
80
40
30
20
15
12
8
6
3
2
1
100
50
0.2
0.4
0.6
point-distance [h]
0.8
0
0
0.2
0.4
0.6
0.8
point-distance [h]
u rms
(30%)
v rms
(30%)
w rms
<u
(50%
)
<v
(300%)
<w
(1700%)
(30%)
Vertical planes:
u rms
(30%)
v rms
(30%)
w rms
<u
(160%)
<w
(490%)
(30%)
39
References
Repschies, D. (2006) "Untersuchungen zur Reprsentativitt lokaler Strmungsmessungen in einer
idealisierten
urbanen
Rauigkeitsstruktur"
(in
german), Univer-sity of Hamburg, Meteorological
Institute
Richtlinie 96/62/EG des Rates der EU vom 27.09.1996
ber die Beurteilung und Kontrolle der Luftqualitt.
Amtsblatt L296 vom 21.11.1996 (in german)
Schultz, M. (2006):
Fortschrittsbericht DFG 04-174: Systematische
Untersuchung der Grenzschichtentwicklung ber
Stdten, Hamburg (in german)
Schultz, M. (2007):
Systematic Investigation on the Urban Boundary
Layer project DFG 04-174, PhD-Thesis (under
preparation), Hamburg (in german)
VDI Richtlinie 3783, Blatt 12 (1999):
Umweltmeteorologie. Physikalische Modellierung
von Strmungs- und Ausbreitungsvorgngen in der
atmosphrischen Grenzschicht. VDI Handbuch
Reinhaltung Luft, Band 1b, Vorentwurf August 1999
(in german)
VDI Richtlinie 3783, Blatt 9 (2005):
Umweltmeteorologie. Prognostische mikroskalige
Windfeldmodelle - Evaluierung fr Gebude- und
Hindernisumstrmung. Available in English as VDI
Guideline on Environmental meteorology Prognostic microscale windfield models - Evaluation
for flow around buildings and obstacles.
Introduction
Reliable modeling of flow and dispersion processes
within complex urban structures is one of the focal
points of environmental research. Besides expensive
and time consuming field measurements, physical and
numerical models are applied to investigate transient
dispersion phenomena of locally released pollutants in
urban areas.
From a mathematical point of view, flow and
dispersion processes in the atmospheric boundary layer
can be treated as a boundary value and initial value
problem.
Concerning the modeling of flow and
dispersion phenomena this means, that the model
results will certainly depend on the boundary conditions
applied. Most often, it is an open question, how sensitive
results from a particular boundary layer wind tunnel
experiment are with respect to a change in the modeled
approach flow conditions and how close model results
can represent the conditions expected at full scale.
Moreover, it is often impossible to define physically
precise enough the boundary conditions to be modeled.
Even in ideal situations, when mean approach flow
conditions can be derived from representative field data,
an uncertainty is introduced by averaging over physically
different situations within a certain class of
meteorological conditions and further uncertainty is
caused by the technical realization of a certain approach
flow condition in a wind tunnel.
Depending on the type of model investigated and
depending on the complexity of the model geometry, the
uncertainty inherent in the modeled boundary layer will
affect the reliability and representativeness of the model
results to a certain degree. Often it is argued, that
41
reference BL
accelerated BL
decelerated BL
const. cross section
higher turbulence BL
no BL simulation
Zo [m]
0.20
0.35
0.43
0.31
1.53
0.03
0.18
0.19
0.20
0.18
0.26
0.13
300
250
accelerated BL
decelerated BL
constant cross section
higher turbulence BL
no BL simulation
reference BL
Z fs [m]
200
150
100
EWTL@ZMAW.DE University of Hamburg
300
50
accelerated BL
decelerated BL
constant cross section
higher turbulence BL
no BL simulation
reference BL
250
0
-0.008
-0.004
-0.002
200
Zfs [m]
150
100
50
-0.006
0.2
0.4
0.6
0.8
1.2
2 Results
In the subsequent part of the paper, selected results
for one exemplary measurement location are discussed.
In Figure 3, the measurement location in the center of a
narrow street canyon (Park Av.) is marked with the big
circle. The average building height at the measurement
42
300
accelerated BL
decelerated BL
constant cross section
higher turbulence BL
no BL simulation
reference BL
250
200
Zfs [m]
150
100
50
0.2
0.4
0.6
0.8
1.2
First, the mean wind profiles are compared with Uand W-component investigated seperately. The grey bar
attached to the vertical axis of the plots is indicating the
average building height around the measurement
location.
In general it can be stated that the location in Park
Avenue shows the lowest deviations from the reference
case of all measurement locations for the U-component
of the wind vector because of the high buildings
surrounding the measurement area (Figure 4).
For an accelerated/decelerated boundary layer flow
approaching the model, the difference in the measured
mean wind profile is in the order of 5% on average.
However, the highest deviations can be found right
above roof level, where differences in the mean wind
speed would have a major impact on pollutant transport
and dispersion. The wind profiles of the higher turbulent
flow and the flow without an explicitly modeled boundary
layer flow document a slightly higher similarity at this
location but local differences reach values of up to 12 %.
As documented in Figure 5, the vertical wind
component (W) is more sensitive with respect to
changes in the approach fllow than the horizontal
components. Nevertheless, the range of the deviations
from the reference case is similar to the deviations
observed in the approach flow and hardly influenced by
the building structure in this case. From Figure 5 it
becomes clear that the deviations from the reference
case are nearly the same for heights above 80 m for all
variations of the approaching boundary layer flow.
43
300
250
200
Zfs [m]
150
accelerated BL
decelerated BL
constant cross section
higher turbulence BL
no BL simulation
reference BL
100
50
-0.06
-0.03
0.03
0.06
0.09
300
250
Zfs [m]
200
150
300
accelerated BL
decelerated BL
higher turbulence BL
no BL simulation
reference BL
250
accelerated BL
decelerated BL
constant cross section
higher turbulence BL
no BL simulation
reference BL
100
50
200
Zfs [m]
0
-0.02
-0.016
150
-0.008
-0.004
100
0.002
0.004
0.006
0.008
0.01
300
250
accelerated BL
decelerated BL
more turbulence BL
no BL simulation
reference BL
200
Zfs [m]
50
0
-0.002
-0.012
150
100
50
0 1
10
102
Lux [m]
44
103
3 Conclusions
The detailed analysis of locally measured data
reveals that proper boundary layer modelling is essential
for obtaining reliable model results.
However, the accuracy required for modeling proper
approach flow conditions certainly depends on the type
of problem to be investigated. If a more global flow
pattern over and around a complex urban roughness,
represented by horizontal mean velocity measurements,
is of particular interest, improper boundary layer
modelling can cause a minimum uncertainty of the
measured velocity components in the order of 10 % to
20 %.
If the local flow and dispersion phenomena at a
certain location are of particular interest, much more
care must be taken regarding a qualified experimental
setup providing proper approach flow conditions. As
shown in this paper, a highly dense built-up area can
even intensify the influence of varying approach flow
conditions on the locally observed turbulent properties.
Uncertainties up to a factor of three were found for the
example given here.
It is obvious that no atmospheric boundary layer
simulation at all produces the biggest errors. Hence, an
experimental setup without explicit boundary layer
modeling should never be used for investigations of
atmospheric flow and dispersion processes in urban
areas. Approach flow conditions with higher turbulence
levels than expected at full scale will have a significant
impact on the momentum fluxes and integral length
scales measured even at measurement locations below
roof level in a street canyon oriented perpendicular to
the mean wind direction. Turbulence characteristics
must be modeled and replicated in a wind tunnel
carefully in order to ensure a reasonable quality of
model results.
The longitudinal pressure gradient along the test
section in the wind tunnel and resulting accelerated or
decelerated model boundary layers approaching a
model was found to be an important factor regarding
uncertainty of the model results. Consequently, at least
the longitudinal pressure gradient across the model
needs to be checked and minimized carefully before any
model test is carried out. The most sensitive criterion
regarding accelerated/decelerated approach flow is,
however, a direct measurement of the vertical
momentum flux profile. Documenting a sufficiently thick
constant shear layer at the right momentum levels turns
out to be a necessary prerequisite for any reliable flow
and dispersion modeling in a boundary layer wind
tunnel.
References
Allwine, K. J., M. J. Leach, L. W. Stockham, J. S. Shinn,
R. P. Hosker, J. F. Bowers and J. C. Pace, (2004):
Overview of Joint Urban 2003 an atmospheric
dispersion study in Oklahoma City, Symp. on
Planning, Nowcasting and Forecasting in the Urban
Zone, January 11-15, Seattle, WA. Amer. Meteor.
Soc.. J7.1.
Leitl, B.; Pascheke, F.; Schatzmann, M. (2003)
"Generation of Wind Tunnel Data Sets in Supprt of
the Joint Urban 2003 Atmospheric Dispersion Study
Final Reort, Phase I", Meteorological Institute,
University of Hamburg, Germany
Leitl, B.; Schatzmann, M. (2005) "Generation of Wind
Tunnel Data Sets in Supprt of the Joint Urban 2003
Atmospheric Dispersion Study Final Reort, Phase
45
L. Laporte, E. Dupont
1 Experimental set-up
Measurements were performed in the Malavard
wind tunnel of the Laboratoire de Mcanique et
dEnergtique (L.M.E), university of Orleans. It is a
close-circuit type with a test section of 2m high, 2m
wide and 5m long. However, as studies concerning
atmospheric flows (especially concerning the ABL)
require greater dimensions, we had to transform the
return section to get closer to atmospheric wind
tunnels conditions. Figure 2 underlines these
transformations. A sketch of the Malavard wind
tunnel is presented in figure 1 and a sketch of the
return section in figure 2. A convergent was placed
to control and homogenize the flow and a 16m long
and 5m wide wooden plate was constructed in the
return test section to model a flat terrain. It is
possible to see on this sketch that the return section
is divergent (angle 2.6) from the exit of the
convergent to approximately 12,5m. Beyond this
location, the section has a constant 5m width
(measurements were made to validate the
properties of the ABL in this constant part of the test
section).
Pieces of corner iron were placed on this flat terrain
to model the ground roughness, and turbulence
generators at the exit of the convergent allow the
establishment of the ABL. It was checked that
several ABL could be properly modelled at a
geometric scale of 1:400, from moderatly rough (low
grass, steppe) to very rough surfaces (centres of
European towns, suburbs).
Introduction
Wind farm efficiency naturally relies on
technological improvement of the wind turbine but it
also and especially relies on the resource
assessment of the upstream flow conditions. The
global resource assessment depends on the type of
terrain and on local meteorological conditions but it
could locally be modified when wind turbines
interact with each other. Indeed, the wind turbine
wake is characterised by a deficit of velocity and
production of turbulence. The estimation of flow field
modifications in the far wake of a wind turbine
located in the atmospheric boundary layer (ABL) is
not obvious. On one hand, field measurements are
rare, difficult to interpret and do not allow
parametrical study and on the other hand, numerical
simulations have still, in this topic, a high degree of
simplification. Physical modelling is therefore a
good tool to study this problem.
The goal of the present study is to model in a wind
tunnel, at a geometric scale of 1:400, a wind turbine
in a rough ABL. The first part of the project was to
modify the wind tunnel to allow the modelling of the
ABL. Then, a first set of measurements was needed
to validate the experimental device and for
comparisons with litterature concerning the wake
behind a single turbine. A third part focusses on a
parametric study and comparisons of the wakes of 2
wind turbines placed in 2 different turbulent flows.
Finally, in a future collaboration with CEREA (joint
laboratory between the French electricity provider
EDF and the Ecole Nationale des Ponts et
Chausses), comparisons will be made between
their numerical simulations and our physical
modelling to validate both methods. Thats why a
part of this paper is dedicated to the explanations of
the numerical method used by CEREA.
Return
section
Return
testtest
section
5 x 5 m
4u4m
Vmax = 22 m/s
Vmax = 12.5 m/s
47
2 Numerical set-up
Numerical simulations of a horizontal axis wind
turbine wake could be achieved using different
methods (Vermeer et al. [11]). The porous disc
concept has been chosen to model the rotor and
source terms were added in the Navier-Stokes
equations to model the thrust and the rotation of the
blades in three dimensions. This approach was
chosen for different reasons. Firstly, the main
interest is the far wake region. Thus, there is no
need to reproduce the region directly behind the
turbine blades with accuracy. Secondly, the cost of
the simulation is very low compared to simulations
where the entire rotor is meshed. Thirdly, this
porous disc approach is not very different from the
physical modelling. Finally, using source terms
allows us to simulate the rotation of the blades and
as a result to simulate the rotation of the wake.
Mercure_Saturne is a three-dimensional CFD model
adapted to atmospheric flow from Code_Saturne,
CEREA general-purpose computational fluid
dynamic code for laminar and turbulent flows in
complex geometries. Mercure_Saturne solves the
RANS equations and the transport equations for
turbulent kinetic energy and dissipation. The
turbulence closure model used in the simulations is
k-
with
the
standard
parameterisation
(Archambeau et al. [1]). The discretization in space
is based on the unstructured finite volume
framework with collocated arrangement for all
variables. The time-marching scheme is based on a
prediction of velocity followed by a pressure
correction step. In all the numerical simulations, the
inflow wind profile is modelled in neutral conditions
by a logarithmic law. The inflow turbulent energy
and dissipation profiles are computed following the
Monin-Obhukov theory. The initial conditions and
the inflow boundary conditions are the same.
In order to validate this method, CEREA used the
wake measurements of a small wind turbine
(Rutland WG503) tested in the ENSAM Laboratoire
de Mcanique des Fluides wind tunnel (Rignault,
[7]). To illustrate the results of 3D numerical
simulations, the turbine wake is shown in figure 3
and figure 4. The simulation is in good agreement
with the measurements (figure 3). As it can be seen
in figure 4 the annular structure of the wake is well
defined. At 0.1D downstream of the disc, the hub
has a strong influence. The speed range is 0-10m/s.
The wake is very inhomogeneous.
U disc
(where U disc is the mean velocity just
Uf
behind the rotor and U f the mean upstream
a 1
H hub )
, where U RMS is the root
H hub )
mean scare velocity and U mean the mean velocity,
Iu 0
U RMS ( z
U mean ( z
was Iu 0
16% .
48
Fi
2.5
x = 2D
x = 4D
x = 6D
x = 10D
x = - 1D
1.5
z/D
1
0.5
0
-0.5
-1
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
u / Uo
49
U d
in percents and
U
hub max
results represent
x = 2D
36.2%
x = 4D
18%
x = 6 / 7D
14%
51%
24%
12%
deficits defined as
-0.5
0.1
0.2
0.3
0.4
0.5
z/D
Iu Iu 0
U rms ( z )
and
U w ( z)
0.5
-1
-0.1
U hub
x = 2D
x = 4D
x = 6D
x = 10D
0.6
Ud / Uhub
50
0.5
z/D
z/D
x = 2D (D=80m)
x = 4D (D=80m)
x = 2D (D=40m)
x = 4D (D=40m)
x = 2D (D=80m)
x = 4D (D=80m)
x = 2D (D=40m)
x = 4D (D=40m)
1.5
0.5
-0.5
-0.5
-1
-0.1
0.1
0.2
0.3
-1
0.4
-0.1
-0.05
0.05
0.5
z/D
z/D
x = 6D (D=80m)
x = 10D (D=80m)
x = 6D (D=40m)
x = 10D (D=40m)
x = 6D (D=80m)
x = 10D (D=80m)
x = 6D (D=40m)
x = 10D (D=40m)
1.5
0.5
0
-0.5
-0.5
-1
-0.1
0.1
0.2
0.3
-1
0.4
-0.1
-0.05
0.05
x=2D
x=4D
x=6D
x=8D
D=40m
36.1%
17.9%
13.8%
6.4%
percents.
D=80m
33.8%
17.5%
7.2%
| 0%
U hub
x=2D
x=4D
x=6D
x=8D
at z / D = 0.5, in
D=40m
5.8%
4.8%
3.3%
2.5%
D=80m
4.8%
2.8%
1.1%
0.7%
51
x = 2D (D=80m)
x = 4D (D=80m)
x = 2D (D=40m)
x = 4D (D=40m)
22
20
18
Iu+ (%)
16
14
12
10
8
6
-2
-1
y/D
References
[1]
industrial
applications. International Journal on Finite
Volumes, 1.
[2]
[3]
[4]
[5]
[6]
[7]
L.Rignault
(2002)
Etude
numrique
et
exprimentale du sillage d'un rotor olien. Technical
report, EDF.
[8]
[9]
22
x = 6D (D=80m)
x = 10D (D=80m)
x = 6D (D=40m)
x = 10D (D=40m)
20
18
Iu+ (%)
16
14
12
10
8
6
-2
-1
y/D
5 Conclusions
The validity and the feasability of the experimental
set-up is proved. By playing with the porosity level
of the porous disc, it has been shown that any
velocity deficit could be reproduced (Aubrun [2]). In
this way, as to a porosity level corresponds a
precise induction factor linked to power and thrust
52
Introduction
The new Belgian Antarctic Research Station will be
constructed in the Sor Rondane Mountains (7157S
02320E, 1390m a.s.l.), during the austral summer
2007-2008. It is designed for optimal use by 12 people
with a surface area (living, technical, research, storage)
of 800 m. An extension, based on heated shelters, will
make it possible to accommodate another 8 to 18
people.
53
U ref
Ub
(1)
54
55
'p50| AWS
Cp AWS
1
2
UU 50
| AWS
2
(2)
Cp AWS
'p
1
2
U 0U AWS
2
(3)
56
Fi
1
2
UU AWS
Ai
2
(4)
57
4 CFD Modeling
A CFD model was also implemented to support the
detailed design phase. The comercial solver Fluent 6.3
was used with steady RANS k- turbulence model.
Modified wall functions were used to account for the
roughness of the snow. A validation of the pressure
coefficient at the tap positions was performed at wind
tunnel scale on the reference building. In this case the
boundary conditions are defined by the wind tunnel walls
and the inlet velocity and turbulence profiles as
measured. Figure 13 shows that the discrepancy in the
pressure coefficients from the CFD and the wind tunnel
tests was less than 25% of the stagnation pressure
coefficient (this corresponds to a margin of 0.2 in Cp as
indicated by the thin lines in Figure 13) for 75% of the
measurement points.
5 Conclusions
The design of an Antarctic building requires careful
consideration of the environmental conditions in which it
will be immersed. Wind and snow effects are important
parameters to be considered throughout the design
process.
In the case of the Princess Elizabeth station, the
positioning of the building on the top of a small ridge
requires carefull consideration of the flow under the
building in order to control the snow erosion and
accumulation and the wind loading.
The presented project is a good example of the
advantages of including wind engineering since the
beginning of the building design process, giving the
opportunity to consider the most effective use of the
environment.
References
Anno, Y. (1984). Requirements for Modeling of Snow
Drift, Cold Regions Science and Technology, Vol 8,
pp 241-252.
Belgian Science Policy and International Polar
Foundation (2007). Construction and Operation of
the New Belgian Research Station, Dronning Maud
Land,
Antarctica.
Final
Comprehensive
Environmental Evaluation (CEE).
Comite Europen de Normalisation (2005). Eurocode 1:
Actions sur les structures Partie 1-4: Actions
gnrales Actions du vent. EN 1991-1-4.
Uematsu Y. and Isymov N. (1999). Wind Pressures
Acting on Low-rise Buildings. Journal of Wind
Engineering and Industrial Aerodynamics Vol 82, pp
1-25.
58
A. Borsani
CRIACIV,
University of Florence,
Florence, Italy
lorenzo.procino@pin.unifi.it
gbartoli@dicea.unifi.it
cborri@dicea.unifi.it
1 Scale considerations
Model tests are often performed in order to
understand
aerodinamical
problems;
but
the
transferability of the results to full-scale is not always
certain.
In this study a 1:50 scale test simulates a real case
in which the train is moving at a velocity of
approximately 83 m/s. In order to maintain the Reynolds
number similarity the model train should be 4150 km/h
fast, which is not possible from the point of view of the
reliability of the test. Such a high velocity is not
desiderate also because it is necessary to remain under
the critical Mach number (Ma<0.3), which allows us to
consider the air as an uncompressible fluid.
The solution adopted is to reach a train velocity
which is high enough to induce a significant
overpressure. It should be also possible to measure the
chosen velocity with the available instruments. Changing
the shape of the train with a sort of technical roughness
allowed us to influence the flow structure to some
amount. The best experimental set-up was chosen by
comparing the obtained results with in situ
measurement.
Introduction
The purpose of this study is to determine the
overpressures on the roof and on the noise-barriers at
the new high-speed railway station of Reggio Emilia
(Italy), which are induced by the running through of a
train.
The following procedure has been accepted:
x gathering of available full-scale data,
x manufacturing of a model at the scale of 1:50
related to one of the real cases described in the
preceeding point (from now on called 'case 1') ,
x manufacturing of a lauching system which
simulates the moving of the train,
x calibration of the measuring system varying the
speed and the shape of the train,
59
cp
pbarrier p air
U air 2 vtrain 2
3.8 cm
6 cm
7 cm
3.8 cm
24 cm
Figure 5. Model 3
6 cm
4.8 cm
35 cm
Figure 6. Model 4
2.2 Results
3.8 cm
20 cm
3.8 cm
Figure 3. Model 1
Train passage
3.8 cm
3.8 cm
24 cm
Figure 4. Model 2
60
Cp max
0.16
cp [-]
0.14
real case
0.1
model n.1
model n.3
0.06
model n.4
0.04
0.02
0
40
60
80
v [km/h]
100
2
Noise barrier
3
4
model n. 2
0.08
0.18
0.12
Steel structure
glass
3.2 Results
120
3 Case of interest
Once the optimal experimental setup was
determined (case 1), the test for Reggio Emilia train
station were carried out.
3.2.1
Position 1a
test
cp_max1
cp_min1
v [m/s]
v [km/h]
a2
0.058
-0.071
11.308
40.708
a3
0.044
-0.038
18.208
65.549
a4
0.054
-0.022
20.082
72.295
a5
0.045
-0.029
17.133
61.678
g2
0.039
-0.036
18.000
Table 1. Cp measured in case 1a
64.800
0.015
0.01
0.005
-0.005
-0.01
-0.015
Train passage
1.4
1.5
1.6
1.7
1.8
1.9
2.1
3.2.2
test
d2
61
Position 1b
cp_max1 cp_min1
0.036
-0.034
v [m/s]
v [km/h]
19.955
71.837
e2
0.047
-0.030
18.329
Table 2. Cp measured in case 1b
65.985
0.3
0.25
0.2
0.15
0.1
0.03
0.05
0.02
0
0.01
-0.05
-0.1
Train passage
-0.15
-0.01
-0.2
0.5
1.5
2.5
3.5
4.5
-0.02
-0.03
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.1
3.2.3
Position 2 and 3
test
v [m/s]
v [km/h]
h2
0.034
-0.038
19.865
Table 3. Cp measured in case 2
71.514
cp_max1 cp_min1
0.03
0.02
0.01
-0.01
-0.02
-0.03
Train passage
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
3.2.4
test
Position 4
v [m/s]
v [km/h]
18.798
67.673
f3
0.315
-0.239
22.205
Table 4. Cp measured in case 4
79.940
f2
Train passage
cp_max1 cp_min1
0.293
-0.153
4 Conclusions
The obtained results show that the pressure effects
generated by the passing of trains are very low both on
the superior part of the steel structure and on the
glasses of the roof. In these positions the measured
values are near the value of the sensibility of the
pressure transducer. On the barrier instead the
overpressure is appreciable. The results of the adopted
measuring system are acceptable and it is possible to
validate these results with the results of full-scale
measurements. In order to have a good agreement with
the reference values it was necessary to modify the
shape of the train model. The maintenance of the
geometric staircase can cause errors because the
Reynolds number similarity was not obtained.
This study is to be considerd as a first attempt to
resolve this kind of problems. The aspects regarding the
correct scaling of the this phenomena are to be further
investigated.
Acknowledgments
All the test are realized in the Boundary Layer Wind
Tunnel from CRIACIV (University of Florence, Italy).
References
Niemann H. J. and Hlscher N., (2003) "Messung der
aerodynamischen
Einwirkungen
aus
der
Zugvorbeifahrt am Trogbauwerk Ittembach"
Macciacchera I. U. and Ruck B., Pressure Fluctuations
Induced by Road Vehicles in Ambient Air A Model
Study-.
Shaw C. T., Garry K.P. and Gress T., (2000) Using
singular system analysis to characterise the flow in
the wake of a model passenger vehicle. Journal of
Wind Engineering and Industrial Aerodynamics, 85
(2000) 1-30.
Kanda I., Uehara K., Yamao Y., Yoshikawa Y. and
Morikawa T., (2006) A wind-tunnel study on
exhaust gas dispersion from road vehicles- Part I:
Velocity and concentration fields behind single
vehicles. Journal of Wind Engineering and
Industrial Aerodynamics, 94 (2006) 639-658.
62
*Atmospheric Sciences Modeling Division, ARL/NOAA, MD-81, USEPA, RTP, NC 27711, USA
**Alion Science and Technology, P.O.Box 12313, RTP, NC 27709, USA
g
Atmospheric Modeling Division, NERL/EPA, MD-81, USEPA, RTP, NC 27711, USA
heist.david@epa.gov
Introduction
Studies have shown that long-term exposure to
traffic-related pollutants is an important risk factor for
certain respiratory problems and for mortality (e.g.,
Nitta et al., 1993, Finkelstein et al., 2004). In the U.S.
there is increasing concern specifically for the many
people that live, work and attend school in close
proximity to major roadways. Many applied dispersion
models in use currently were developed for simplified
roadway scenarios that do not include the complex
geometries often found surrounding urban highways.
Hosker et al. (2003) found that guidance for the
application of such models in these situation has not
been adequate. In particular, they note that problems
can be anticipated in applying Gaussian dispersion
models in low wind situations, in areas with complex
highway configurations, and at urban intersections.
While a number of wind tunnel studies (e.g.,
Hayden et al. 2002, Kastner-Klein and Plate, 1999)
have examined the effects of urban street canyons
and intersections, there is a need to examine the
influence of roadway configurations and nearby
structures where urban buildings are not dominating
the overall flow. Therefore, the current wind-tunnel
study has been designed to consider these influences
on the flow and dispersion of traffic-related pollutants
within a few hundred meters of the roadway. This
paper reports on the five selected configurations
shown in Figure 1 (all with a model six-lane, divided
highway at a 1:150 scale). They included flat terrain
with no structures, flat terrain with sound barriers and
depressed roadways with vertical or sloping walls.
Experiments involved smoke visualization and tracer
concentration measurements from a uniform line-type
source as well as flow characterization.
1 Experiment
Experiments were conducted in the US EPAs
Fluid Modeling Facility meteorological wind tunnel.
The test section is 370 cm wide, 210 cm high, and
1830 cm long. The air speed in the test section was
fixed at 4.7 m/s at a height of 165 cm.
To simulate the traffic along a six lane highway, a
280 cm long and 24 cm wide roadway was installed in
the wind-tunnel with the roadway perpendicular to the
wind direction. Six parallel line-source segments were
placed in the center (laterally) of the roadway oriented
parallel to the axis of the highway.
Each line
contained approximately 30
63
U
u*
2 Methods
After measuring the pattern of downwind
concentration from the finite line segment, the infinite
line source results can be constructed by
superimposing the results from the finite line source
employing a lateral shift in the source location, as
follows:
C x , z
C x, y iL , z
i f
fls
1 z d
ln
N z 0
64
10
10
-1
10
-2
10
-3
10
-4
-40
-20
Y/H
20
40
Flat terrain
Figure 4. Superposition of results from a finite linesegment source to create the effect of an infinite line
source. Open diamonds are measured results; filled
diamonds represent measured results shifted laterally
by multiples of the source width; red line is the
summation of all five profiles.
4
2
0
3 Results
65
-12
-6
Y/H
12
Z/H
4
2
0
-12
-6
Y/H
12
Z/H
Chi
50
20
10
5
2
1
0.5
0.2
0.1
Z/H
Non-dimensional concentration
10
2
0
-12
-6
Y/H
12
Figure 5. Concentration contours from finite linesegment sources measured at the downwind edge of
the roadway for three cases. The flow in these
figures is toward the viewer.
Flat terrain
Z/H
10
0
0
10
X/H
20
30
40
Z/H
10
0
0
10
X/H
20
30
40
Z/H
10
0
0
10
X/H
20
30
40
Chi
100
50
20
10
5
2
1
0.5
0.2
0.1
Chi
100
50
20
10
5
2
1
0.5
0.2
0.1
Chi
100
50
20
10
5
2
1
0.5
0.2
0.1
Figure 6. Concentration contours for an infinite line source. Flow is from left to right.
4 Discussion
The departures from flat terrain, though
significantly different from each other geometrically,
produce similar results, especially at ground level and
downwind of the immediate vicinity of the source.
This can be seen in the Figures 6 and 7, but also in
an analysis of the plume heights and the ground-level
concentrations downwind of the source.
The height of the plume can be characterized by
the plume centroid, z , defined:
f
zCdz
0
f
Cdz
0
66
67
5 Summary
These measurements demonstrate the use of a
finite, line-segment source to simulate the effect of an
infinite line source in modeling emissions from
various roadway configurations. The configurations
studied include flat terrain, flat terrain with sound
barriers upwind of the roadway, flat terrain with sound
barriers on both sides of the roadway, depressed
roadway with vertical walls and depressed roadways
with sloping walls.
Concentration measurements show that soundbarriers and depressed roadways produce remarkably
similar ground-level concentration patterns downwind,
which are significantly lower than flat terrain results.
The concentration patterns suggest that an effective
shift in the flat terrain concentration results a distance
of 7H upwind would describe the measurement results
well for the barrier and depressed roadways cases.
The utility of this approach needs to be investigated
with addition roadway configurations and wind
directions.
Acknowledgments
The authors would like to thank Richard Baldauf,
Vlad Isakov, and Tom Pierce for their advice and
support.
Disclaimer
The research presented here was performed
under the Memorandum of Understanding between
the US Environmental Protection Agency (EPA) and
the US Department of Commerces National Oceanic
and Atmospheric Administration (NOAA) and under
agreement number DW13921548. This work
constitutes a contribution to the NOAA Air Quality
Program. Although it has been reviewed by EPA and
NOAA and approved for publication, it does not
necessarily reflect their policies or views. Mention of
trade names or commercial products does not
constitute endorsement or recommendation for use.
68
References
Briggs, G.A., Britter, R.E., Hanna, S.G., Havens, J.A.,
Robins, A.G. and Snyder, W.H. (2001) Dense
gas vertical diffusion over rough surfaces: results
of
wind-tunnel
studies,
Atmospheric
Environment, vol. 35, pp. 2265-2284.
Finkelstein MM, Jerrett M, Sears MR. (2004) Traffic
related air pollution and mortality rate
advancement periods. Am J Epidemiol. Vol. 160,
no. 2, pp.173-177.
Hayden, R.E., Kirk, W.D., Succi, G.P., Witherow, T.
and Bouderba, I.
(2002) Modifications of
Highway Air Pollution Models for Complex
Geometries Volume II: Wind Tunnel Test
Program US Department of Transportation,
Federal Highway Administration Report, FHWARD-02-037.
Hosker, R.P, Jr., Rao, K.S., Gunter, R.L., Nappo, C.J.
and Meyers, T.P. (2003)
Issues affecting
dispersion near highways: light winds, intra-urban
dispersion, vehicle wakes, and the ROADWAY-2
dispersion model, US Department of Commerce,
NOAA Technical Memorandum OAR ARL-247.
Kastner-Klein, P. and Plate, E.J. (1999) Wind-tunnel
study of concentration fields in street canyons.
Atmospheric Environment 33, 3973-3979.
Nitta, H., Sato, T., Nakai, S., Maeda, K., Aoki, S. and
Ono, M., (1993) Respiratory health associated
with exposure to automobile exhaust I. Results of
cross-sectional studies in 1979, 1982, and 1983.
Archives of Environmental Health 48, pp. 5358.
Introduction
The urban surface layer possesses a pronounced
layer, the roughness sublayer (RS), which is highly influenced by the local canopy and characterized by a
strong heterogeneity of the flow field. E.g. Cheng and
Castro (2002) showed in a wind tunnel study, that especially above very rough surfaces the vertical extent of
the RS is not negligible. It covers a big portion of the
lower part of the surface layer, usually on the expanse
of the inertial sublayer (IS). Furthermore, for specific
roughness it can be questioned if there exists an IS at
all. The logarithmic wind profile is not valid within the
RS. This makes it difficult to determine for example the
roughness length z0 as a parameter describing surface
characteristics. To replicate the effect of the urban
roughness, e.g. in mesoscale numerical models,
roughness parameterizations are needed. However, to
improve an appropriate roughness parameterization
detailed information of the flow field within the canopy
and the overlying RS and IS is required.
1 Experimental Set up
This study presents first results of an extensive
measuring campaign which focused on the determination of vertical momentum fluxes in and directly above
an idealized urban roughness. LDA Flow Measurements were carried out at the large boundary layer wind
tunnel Wotan. The wind tunnel has a total length of 25
m with a test section which is 4 m wide and 2.75 m high
and contains a flow establishment section of about 18
m length. The wind tunnel is equipped with an adjustable ceiling allowing 0.5 m height extension of the test
section.
All together 59 rows of cubes were positioned in the
wind tunnel, covering the whole floor from the vortex
generators at the entrance into the flow establishment
section to the suction fan at the end of the testsection.
An intensive observation area where most of the measurements were made was located after row 43 in order
to ensure a sufficiently long upstream fetch. Figure 1
69
Figure 1: sketch of the experimental set up in the Wotan wind tunnel at the EWTL of Hamburg University.
Figure 2: Left hand side Pictures of the three investigated configurations in the wind tunnel. Right hand
side: Profile measurement locations for the three investigated types of roughness. Top: aligned without
roofs (config 1), centre: aligned with roofs (config 2) and bottom staggered with roofs (config 3). The
coordinate system for all three configurations is defined as follows: X=0 between rows 46 and 47, Y=0 at
the wind tunnel centerline and Z=0 at the ground floor, posive into the upward direction.
70
71
3.5
3
2.5
Z / hc
1.5
1
0.5
0
-0.25 -0.2 -0.15 -0.1 -0.05
U'W' / U ref2
0.05
0.1
0.05
0.1
3.5
3
2.5
Z / ha
1.5
1
0.5
0
-0.25 -0.2 -0.15 -0.1 -0.05
4
3.5
Z / ha
2.5
2
1.5
3.5
0.5
0.05
0.1
2.5
Z / hpeak
0
-0.25 -0.2 -0.15 -0.1 -0.05
1.5
The question remains, how to scale data accomplished over various surfaces (or cube fields) in order to
make them comparable to each other. Basis information
is needed which involves the characteristics of the individual roughness. Within this paper it is suggested to
use the peak height hpeak (which is roughly the height of
the surrounding dominating rooftops) as a basis of scaling the data. The reference wind speed Uref is scaled to
the wind speed at peak height hpeak. Adopting the proposed scaling to Figure 3 would lead to almost identical
expressions describing the extension of the RS for all
three investigated types of roughness, namely hRS = 1.5
hpeak (not shown as a figure). The vertical extension of
the IS is slightly higher for the homogenous roughness
(roughly 2.3 hpeak = 2.3 hc, compare above) than for
heterogeneous cube fields with roofs added. Figure 5
presents measured profiles scaled with hpeak behind the
cube (homogeneous array of configuration 1) and the
profiles measured behind a cube with flat roof of the
other two configurations. The profiles show surprisingly
similar shapes and collapse well especially between 1
hpeak and 1.5 hpeak, where they are directly influenced by
1
0.5
0
-0.25 -0.2 -0.15 -0.1 -0.05
72
0.05
0.1
4
3.5
Z / hpeak
Z / hpeak
2.5
2
4
3
1.5
2
0.5
0
-0.25 -0.2 -0.15 -0.1 -0.05
0.05
0
-0.15
0.1
Figure 5b
7
3 Conclusions
First results of a measurement campaign with three
different types of roughness were presented. For selected profiles measured behind similar types of obstacles only minor differences can be found between a
surface consisting of obstacles with equal height and a
surface consisting of obstacles with heterogeneous
height. Clearly, more measurement locations and spatially averaged profiles have to be considered to complete the investigation.
73
-0.05
0.05
0.1
0.05
0.1
Z / hpeak
-0.1
4
3
2
1
0
-0.15
-0.1
-0.05
Acknowledgments
The authors gratefully acknowledge financial support from the DFG (Deutsche Forschungs Gemeinschaft) in the frame of the project DFG 04-174 Systematic Investigation of the Urban Boundary Layer
References:
Cheng, H. & Castro, I. P. 2002. Near wall flow over
urban-like roughness. Boundary Layer Meteorology,
104, 229259.
Feigenwinter, C., Vogt, R. & Parlow, E. 1999. Vertical
structure of selected turbulence characteristics
above an urban canopy. Theor. Appl. Climatol., 62,
5163.
Kastner-Klein, P., Berkowicz, R. & Britter, R. 2004. The
influence of street architecture on flow and
dispersion in street canyons. Meteorology and
Atmospheric Physics, 87, 121131.
Louka, P., Belcher, S. E. & Harrison, R. G. 2000.
Coupling between air flow in streets and the welldeveloped boundary layer aloft. Atmospheric
Environment, 34, 26132621.
MacDonald, R. W., Carter Schofield, S. & Slawson, P.
R. 2002. Physical modelling of urban roughness
using arrays of regular roughness elements. Water,
Air and Soil Pollution: Focus, 2, 541554.
Oikawa, S. & Meng, Y. 1995. Turbulence characteristics
and organized Motion in a suburban roughness
sublayer. Boundary Layer Meteorology, 74, 289
312.
Rafailidis, S. 1997. Influence of building areal density
and roof shape on the wind characteristics above a
town. Boundary Layer Meteorology, 85, 255271.
Raupach, M. R., Antonia, R. A. & Rajagopalan, S. 1991.
Rough-wall turbulent boundary layers. Appl. Mech.
Rev., 44 (1), 125.
Rotach, M. W. 1993a. Turbulence close to a rough
urban surface part I: Reynolds stress. Boundary
Layer Meteorology, 65, 128.
74
B Leitl
Department of Meteorology,
University of Reading,
Reading, UK
j.f.barlow@reading.ac.uk
Meteorologisches Institut,
University of Hamburg,
Hamburg, Germany
bernd.leitl@zmaw.de
1 Experimental Set-up
Introduction
Ventilation of pollution from street level to the
boundary layer above is dependent on the flow dynamics
around buildings close to the source. The conceptual
model of street canyon flow has been formed from classic
experiments highlighting the mean flow patterns in a flat
roofed street canyon and their dependence on aspect
ratio. However, due to high turbulence intensities in the
roughness sublayer, the turbulent component of the flow is
just as important as the mean flow in dispersing pollution.
Several authors have noted the intermittent nature of
roughness sublayer turbulence, and have studied the role
of the shear layer at the top of the urban canopy in
generating intermittent, coherent turbulent structures
which penetrate downwards, causing mixing of air within
the streets (Christen, 2005; Coceal et al., 2007). It is
hypothesised here that shear layer dynamics are sensitive
to roof shape and that local turbulent flow statistics will be
significantly altered, as well as overall mean flow patterns.
75
Roof shape
Roof angle
(degrees)
Flat
0
Low pitch
26.6
High pitch
45
Table 1. Dimensions of
pitched roofs.
Height
(mm)
Lowest LDA
measurement
mm
(ratio to H)
63
33 (0.52H)
78.7
49 (0.62H)
94.5
67 (0.71H)
street canyons for different
2 Flow visualisation
Flow within the first street canyon in the row for each
setup is presented in Figure 1 a to d. Visualisation was
achieved by placing a line source of smoke on the ground
along the canyon axis, equidistant between the buildings.
Laser light was reflected from a cylindrical mirror located
above the canyon, causing a longitudinal cross section of
the flow to be visible at the canyon centre, i.e. the (x,z)
plane at y=0. Snapshots of the flow are shown, which
highlight more intermittent turbulent structures ventilating
smoke with the air above. Wind vectors measured by LDA
are overlain on the instantaneous flow visualisation in
each case. Note that the flow is from right to left, and the
upward tilt of the vectors on the right hand side are due to
the displacement of the flow over the first building row.
3 Flow measurements
As the flow visualisation suggested intermittent vertical
ventilation of smoke into the air above, differences in flow
statistics between street canyon geometries are now to be
investigated. Preliminary results presented here focus on
Skewness in the vertical wind component, Skw, as it is
indicative of intermittent vertical motions caused by
coherent turbulent structures. For reference, the results of
Christen (2005) showed that profiles measured to one
side of a full scale street showed negative values within
76
0
1.5
0.5
z=33mm 0.5H
z=40mm 0.6H
z=47mm 0.7H
z=54mm 0.8H
z=61mm 0.9H
z=73mm 1.1H
z=83mm 1.2H
z=98mm 1.5H
z=113mm 1.7H
0
Skw
-0.5
-1
z=92mm 1.0H
z=107mm 1.1H
-1
-1.5
0.5
-0.5
-1
-0.5
4 Preliminary Conclusions
z=87mm 0.9H
z=107mm 1.1H
-0.5
x'/H
0.5
z=97mm 1.0H
-0.5
z=122mm 1.3H
-1
z=77mm 0.8H
z=77mm 0.8H
-0.5
1
z=67mm 0.7H
z=67mm 0.7H
-1.5
-1.5
x'/H
1.5
0.5
It has been shown thus far that mean flow patterns for
street canyons of aspect ratios H/W = 0.6 and 1 with
either flat or high pitched roofs show some consistency
with previous results, viz.: the centre of recirculation is
moved upwards for the pitched roof case. Flow was still
observed at ground level, even for the H/W=1,high pitched
roof case, where it was seen to reverse weakly in the flow
visualisation, indicating a weak counter-rotating vortex.
This is different to Kastner-Klein et al (2004) who
observed no in-street vortex for the pitched roof case,
although their aspect ratio of H/W=1.3, with otherwise
-1
z=122mm 1.3H
-1.5
x'/H
77
-1
Skw
0.5
0
z=33mm 0.5H
z=40mm 0.6H
z=47mm 0.7H
z=54mm 0.9H
z=61mm 1.0H
z=73mm 1.2H
z=83mm 1.3H
z=98mm 1.6H
z=113mm 1.8H
-1.5
-1.5
x'/H
0.5
-1
-1
0.5
-0.5
-0.5
1.5
Skw
0.5
Acknowledgments
JB acknowledges the generous use of facilities at the
University of Hamburg granted by M. Schatzmann and
support and advice from B. Leitl.
References
Christen, A. (2005) Atmospheric turbulence and surface
energy exchange in urban environments, PhD thesis,
.
78
Eric R. Pardyjak2
1. Introduction
2. Experimental Method
2.1 Flow Conditions and Setup
The experiments were performed at the Environemental
Fluid Dynamics Laboratory (EFD) at the University of Utah
at an atmospheric pressure of 640mm of Hg, temperature of
21C in a 7.9 m long boundary layer wind tunnel facility
having a working cross section of 0.61m 0.91 m. Lego
sheets with circular dimples of height 2mm lined the floor of
the tunnel and were used to produce rough walled turbulent
flow. The experiments were run at a free stream velocity of ~
7.2 m/s with a corresponding boundary layer depth () and
power law exponent () at the measurement location of /H ~
2 and ~ 0.205.
79
2.3 Notation
The following notations will be used to address the
different canyon configurations and the various parameters
used to understand the flow physics. Some of these
parameters are shown in Figure 1.
H
Hd
Hu
L
PQ
S
UH
W
X
XP
XS
XV
ZS
ZSt1
ZSt2
ZV
Vortex
Core
Saddle
Point
Q
Hu
Ground
Shear Layer
ZV ZS
P ZSt1
Hd
XP
XV
XS
S
Figure 1: Figure showing the definition of the various
parameters used in the paper
80
Y
X
Wake Dominated
Regime
S
k
i
m
m
i
n
g
F
l
o
w
81
Hd / Hu= 0.08
Hd / Hu= 0.16
Hd / Hu= 0.61
Hd / Hu= 0.69
Hd / Hu= 0.27
Hd / Hu= 0.80
Hd / Hu= 0.35
Hd / Hu= 0.43
Hd / Hu= 0.53
Hd / Hu= 0.88
Hd / Hu= 0.96
Hd / Hu=1.00
Hd / Hu= 0.35
Hd / Hu= 0.43
Hd / Hu= 0.53
Hd / Hu= 0.88
Hd / Hu= 0.96
Hd / Hu= 1.00
Hd / Hu= 0.08
Hd / Hu= 0.16
Hd / Hu= 0.61
Hd / Hu= 0.69
Hd / Hu= 0.27
Hd / Hu= 0.80
Hd / Hu= 0.08
Hd / Hu= 0.16
Hd / Hu= 0.61
Hd / Hu= 0.69
Hd / Hu= 0.27
Hd / Hu= 0.80
Hd / Hu= 0.35
Hd / Hu= 0.43
Hd / Hu= 0.53
Hd / Hu= 0.88
Hd / Hu= 0.96
Hd / Hu= 1.00
82
Hd / Hu= 0.08
Hd / Hu= 0.16
Hd / Hu= 0.61
Hd / Hu= 0.69
Hd / Hu= 0.27
Hd / Hu= 0.80
Hd / Hu= 0.35
Hd / Hu= 0.43
Hd / Hu= 0.53
Hd / Hu= 0.88
Hd / Hu= 0.96
Hd / Hu= 1.00
83
-uw = T U
Z
Figure 14 shows the vertical gradient of the mean
streamwise velocity (Um) plotted at the center of the canyon.
It is evident from the figure that the changes in the sign of the
Reynolds stress track the changes in the sign of the velocity
gradient with a near zero value of the Reynolds stress where
the vertical shear is zero.
Figure 15 shows profiles of mean turbulent kinetic
energy (TKE) at the center of the canyon. The maximum TKE
occurs in the shear layer at the top of the canyon and
decreases to a local minimum where the velocity gradient
goes to zero. Below the saddle point, the TKE begins to
increase reaching a local maximum at Z / H ~ 0.3. The
second peak in the TKE profile is associated with strong
mean shear in the lower 40% of the canyon.
Figure 16 corresponds to the mean normalized vorticity
profile at the center of the canyon. The peak values of
vorticity occur at the top of the canyon (in the rooftop
seperation shear layer) and decrease into the canyon. At
84
4. Conclusions
For Step-down street canyons formed from square
footprint buildings with Hu/W = 3.75 and S/W 2.5, the flow
structure in the canyon is quite complex. In these situations
where the upwind building is relatively tall and slender, the
flow field is strong function of the ratio of downwind and
upwind building heights (Hd/Hu). It was observed that there
exist two distinct flow regimes in the canyon. For Hd / Hu
0.88, the flow patterns in the canyon come under the wake
dominated regime and for Hd / Hu > 0.88, the flow structure
is in the skimming flow regime.
Acknowledgments
This work was funded through an LDRD project by Dr.
Robert E. Ecke, CNLS Director - Los Alamos National
Laboratory. The financial support is gratefully acknowledged.
We would like to thank Dr. Xavier Tricoche, Research
Assistant Professor, Scientific Computing & Imaging Institute
University of Utah for helping us with the critical points
detection code. We would also like to thank Dr. Inanc
Senocak & Dr. Michael J. Brown at the Los Alamos National
Laboratory for their valuable inputs during the course of this
research.
References
Addepalli, B. and Pardyjak, E.R., (2005). 2D PIV
measurements between a pair of model buildings with
varying geometries, 58th Annual Meeting of the Division
of Fluid Dynamics, Chicago,IL, USA.
Baik, J.J., Park, R.S., Chun, H.Y. and Kim, J.J., (2000). A
laboratory model of urban street-canyon flows, Journal
of Applied Meteorology,vol 39, pp. 15921600.
Davies, M.E., Quincey, V.G. and Tindall, S.J., (1979). The
near wake of a tall building block in uniform and
turbulent flows, Journal of Wind Engineering
Proceedings of the Fifth International Conference, Fort
Collins,Colorado,USA, pp. 289 - 298.
Jiang, Y., Liu, H., Sang, J. and Zhang, B (2007). Numerical
and experimental studies on flow and pollutant
dispersion in urban street canyons, Advances in
Atmospheric Sciences, vol. 24, no. 1, pp. 111 - 125.
Ohba, M., (1998). Experimental Study of effects of
seperation distance of twin high-rise tower models on
gaseous diffusion behind the downwind tower model,
Journal of Wind Engineering and Industrial
Aerodynamics, vol. 77-78, pp. 555 - 566.
Santiago, J.L. and Martin, F., (2005). Modelling the air flow
in symmetric and asymmetric street canyonsl, Int J.
Environment and Pollution, vol. 25, no. 1/2/3/4, pp. 145
- 154.
Vortex
Core
Saddle
Point
Origin of the
Ground Shear Layer
Figure 13: Mean Reynolds stress profile
85
Introduction
Recent studies of boundary layer development over
rougness arrays, reported by Reynolds et al (2006),
indicate the presence of spanwise variations in the
streamwise velocity field having amplitudes greater than
5% in the mean and greater than 10% in the
turbulence quantities. The initial wavelength of these
spanwise variations was set by the width of the
repeating unit of the roughness, but as the boundary
layer developed downstream this wavelength was
observed to double. For the 10mm cube array this
doubling occured 3 times over the 3.5m fetch of
roughness that was studied, resulting in wavelengths of
8 times that of the repeating unit of roughness.
Figure 1 shows some results from this work at 0.23m
into an array of 20mm obstacles. Lateral profiles of the
streamwise velocity are plotted throughout the
boundary layer (here 56.4mm deep), showing that the
spanwise periodicity extended to all heights. The peakto-peak amplitude of the variation decayed with
increasing fetch, but was still clearly present at the
furthest downstream station studied, 3.2m. There is an
associated periodicity in the shear stress and surface
drag, which has implications for the specification of the
aerodynamic properties of the underlying surface.
The surface drag of an array of cubes can be
inferred from velocity profiles measured above the array
but a more direct measurement can be obtained from
pressure measurements on the individual cubes; e.g. as
described by Cheng and Castro, 2002. The assumption
was made that the pressure drag on each cube at a
given downstream location was approximatly the same.
However, the measurements discussed above suggest
that significant lateral variations are likely. Given that the
87
1 Experimental setup
The experiments were conducted in the EnFlo "A"
wind tunnel, which has a test section 4.5m long, 0.6m
high and 0.9m wide. The floor of the wind tunnel was
covered with 9mm baseboards onto which the 20mm
wooden cube array was fixed with a packing density of
25%. A section of the array is shown in Figure 2.
At the leading edge of the array a 175mm long by
9mm high full width ramp was used to bring the flow up
to the level of the base of the cubes from the tunnel
floor. The measurements presented in this paper were
made 150 cube heights downstream from the first row of
cubes (i.e. at x = 3.0m). To improve the quality of the
cube roughness array around the measurement
location, a section of 11 rows of wooden cubes was
replaced with aluminium cubes screwed to an
th
aluminimum base plate. The 9
row (looking
downstream) on the aluminium plate had holes bored
out underneath the central 13 cubes so that the
pressure tapped cube (shown in Figure 3) could be
inserted in any of these locations. Individual aluminum
cubes were used to fill the remaining holes.
88
Cp =
2
2 #U 25
Cd =
D
2
2 "U 25 A
" U 25 (y) % 2
Cd (y) = Cd (y = 0)#
&
$ U 25 (y = 0) '
89
90
3 Conclusions
A direct link was proved between the pressure drag
on elements in an array of cubes and the lateral
variation in the velocity field over them. The drag
coefficient based on the local flow speed was effectively
constant. This was only apparent once the array had
been carefully adjusted to be level. It is likely that the
effect would be masked in many arrays used in wind
tunnel research by small variations in the level of the
array elements. The sensitivity of the pressure drag
coefficient to element height perturbations was shown to
be large. It was also non-linear, in that the increase in
drag on tall elements was greater than the decrease on
the correspondingly short elements. In an overall sense,
the increased drag on a tall element is off-set to some
degree by the sheltering of donwind elements. The net
effect in the case illustrated was that something like a
half of the drag on the tall element was cancelled.
The span-wise flow and drag variations and the
sensitivity of local drag to element height variations
probably explain a significant part of the scatter in
compilations of the measured properties of roughness
arrays.
Acknowledgments
We are grateful to Allan Wells for his technical
support in this project and for consultation from Tom
Lawton. Particular thanks needs also go to Florent
Champet who worked during the summer of 2005 taking
prelimiary pressure drag measurments without the
benefit of a high precision array.
References
Reynolds, R.T. Hayden, P. Castro, I.P. and Robins A.G.
(2007) Spanwise variations in nominally twodimensional
rough-wall
boundary
layers.
Experiments in Fluids, 42(2):311320.
Cheng, H. and Castro, I.P. (2002) Near wall flow over
urban-like roughness. Bound. Lay. Met., 104:229
259.
91
Introduction
More and more detailed numerical description of the
atmospheric flows is possible due to the development of
the computer technology. There are several advantages
of the usage of CFD solvers for computing atmospheric
flows in micro and meso-scale. High flexibility in mesh
refinement and grid structure allow more accurate
description of the relief, which has an impact on forming
both the aboveground and the upper streams, therefore
the computed flow field could be highly detailed. Many
type of available turbulence models, application of
effective numerical techniques, easy implementation of
model refinements, and the possibility of parallel
computing are all useful features for researchers of
micro meteorology, climatology and environmental
technology. Most CFD solvers have compressible flow
option, which means, that the air density can be a
function of local pressure and temperature through the
ideal gas law. This option, however, does not allow
effective simulation of buoyancy driven atmospheric
93
Mathematical transformation
The mathematical relation between the absolute
physical quantities and the field variables used in the
CFD solver are described by a set of transformation.
The transformation formulae are based on the standard
ICAO [5] temperature and pressure profiles (Eq.1 and
2) and an approximate exponential function for the
density profile which allows the simplification of some of
the transformation expression (Eq.3).
T = T0 z
(1)
g
T z R
p = p0 0
T0
(2)
= 0 e z
(3)
= 0.65 C / 100 m ,
T0 = 288.15 K ,
p0 = 1.01325 10 5 Pa ,
0 = 1.225 kg / m 3 ,
g /(R ) = 5.2553 ,
= 10 4 m 1 .
~ 2
2
p
0
Sw = 0 g z T~ T0 02 1
2
T = T~ T0 + T
p=
~
~+p
p + p = e z p
0
= ~ 0 +
z=
w=
Ln (1 z~)
0 ~ ~ z
w =we
~ = 0 0 (T~ T0 )
(6)
(7)
(8)
(9)
w = ~ w~ 0w~
(11)
(12)
(13)
(5)
(10)
)(
2
~(1 z~)1
Sw = 0 g z T~ T0 (1 z~) 1 + p
(4)
dz = ~ dz~ 0 dz~
~ ( )
ST = c p ~ w
(14)
T
z
(15)
stable
~
T T
=
z z
(16)
Sk = g
S = C1 C3
t
( )
Pr
(17)
g t ( )
k
Pr
(18)
94
Preprocessing
As it is known simulation of boundary layer flows,
such as atmospheric phenomena in focus of our
investigation, are very sensitive for grid quality and for
grid resolution in vertical direction. Optimal computing
cost can be achieved only by applying structured or
layered type unstructured grids. Although these type of
grids can be generated by general purpose meshing
software, it is a rather complex and time consuming
work, therefore it was necessary to develop an
automated fast running preprocessor utility involving the
following functions:
1. Extraction of surface points of arbitrary terrain area
from the online USGS elevation database of 1 arc
second (~ 30 m) resolution.
2. Transformation of co-ordinates from spherical to a
local projection Cartesian system employing UTM
(Universal Transverse Mercator) and EOV (Uniform
National Projection system) projections (Bcsatyai,
2005). Deformation of the transformed domain is
negligible in case of horizontal extents smaller than
100 kilometers which is the case in simulation
problems of our interest.
3. Extension of natural relief by a ramp zone around
the boundaries, as shown from Fig.1., for easier
definition of inlet velocity profile.
95
96
Acknowledgement
Authors of this paper whish to express their thanks
for the support of OTKA T049573 project and to
Hungarian National Office for Research and Technology
for support of NKFP 3A/088/2004 project.
References
Conclusions
General purpose CFD software has been adapted to
atmospheric simulation problems involving stratification
and
compressibility
effects.
Modifications
are
formulated in a novel system of transformation together
97
F. de Paoli
Depto. Ingeniera,
Universidad del Nordeste,
Resistncia, Argentina
a_wittwer@yahoo.es
A. M. Loredo-Souza
E. B. Camano Schettini
Abstract The
concentration
fields
in
the
proximities of a local gas emission source are
experimentally analyzed in several combinations of
wind incidences and source emissions. The
concentration measurements were performed by an
aspirating probe, in a boundary layer wind tunnel.
The analysis has included the determination of the
concentrations mean values as well as the intensity
of the fluctuations. The source model represents a
local gas emission which is dispersed in a turbulent
boundary layer in neutral stability. Several different
conditions are determined by the plume buoyancy,
the emission velocity and incident flow wind speed.
The model scales were 1:200 and 1:400, according
to the simulated atmospheric boundary layer factor
scale. Pure helium as well as an air-helium mixture
are used for the emissions. The approaching flow
represents a neutral atmospheric boundary layer.
Three configurations were tested: an isolated
chimney in a homogeneous terrain, the same
chimney with one bluff body in close proximity, and
the chimney surrounded by a non homogeneous
urban terrain. The experimental mean concentration
values are compared with Gaussian profiles. The
plume vertical dispersion values are obtained and
compared with the Briggs curves. The profiles of
the
concentration
fluctuation
intensity are
compared with the experimental results obtained
from other authors. Finally, the dilution factor is
analyzed with regard to the empirical curves of the
minimum dilution.
Key words plume, gas emission, physical model,
wind tunnel, aspirating probe.
Introduction
The study of dispersion and pollutant concentration
levels discharged in the atmosphere has become a
fundamental issue due to the new environmental
demands. Nowadays numerous computational works
related with dispersion phenomena are being
developed. Usually these studies must be validated with
experimental results. The high costs of field
experimentation lead to laboratory reduced models
studies. In this context, the boundary layer wind tunnel
becomes an important tool. However, it is necessary
that the main characteristics of the atmospheric
boundary layer and of the dispersion processes be
reproduced. Up to now there are no publications
99
1 Experimental Design
The basic requirement of a wind tunnel dispersion
study is the physical simulation of the atmospheric flow.
600
Alta velocidade
Uref=0.96 m/s
450
Uref=3.44 m/s
Perfil 0.23
z [mm]
150
0
0,00
0,50
1,00
1,50
U/Uref
Alta velocidade
Uref=0.96 m/s
Uref=3.44 m/s
w0 [m/s]
U0 [m/s]
A
B
C
D
E
F
G
H
He
He
He
He - Ar
He
He
He
He - Ar
0,56
1,26
0,95
0,75
0,56
0,95
0,56
1,45
0,85
1,91
0,85
0,85
3,04
3,04
1,91
1,91
Condition
w0
U0
0 w0 2
aU 0 2
[( 0 a )gw0 D0 ]
A
B
C
D
E
F
G
H
0,66
0,66
1,11
0,88
0,18
0,31
0,29
0,76
0,060
0,060
0,171
0,278
0,005
0,013
0,012
0,145
-0,154
-0,031
-0,260
-0,154
-0,003
-0,006
-0,014
-0,031
Logartmica (Alta
velocidade)
y = -257,28Ln(x) + 827,38
z [mm]
Emission
600
450
Condition
300
150
0
0
10
20
30
100
aU 0 3
2 Results
Instantaneous concentrations were measured in
vertical profiles located in several distances x/H,
measured from the emission. The concentration mean
and rms values were obtained from Reynolds
decomposition, for each point.
Figure 4 presents the vertical profiles of the
concentration coefficient K and the IC for condition A
and configuration I (no building), related to the positions
x/H = 0,33, 0,66 and 1,00. Due to buoyancy effects, the
profiles present an asymmetry, tending to deviate the
plume upwards. Similar behaviour is observed in
Figure 5, related to condition B and to the same
configuration I, for the three leeward positions.
2
x/H=0.315 (a)
x/H=0.630 (a)
x/H=0.315 (b)
x/H=0.630 (b)
x/H=1.260 (b)
z/H
1.5
0.5
0.0
1.0
2.0
3.0
K=
CUHH2 ,
Q0
Ic = c ,
C
x/H=0,33
x/H=0,66
1.4
x/H=1,00
z/H
1.6
1.2
(1)
0.8
0.1
10
100
Ic
(2)
101
1.6
2
x/H=0.315 (a)
x/H=0.630 (a)
x/H=0.315 (b)
1.4
x/H=0.630 (b)
x/H=0.60
x/H=1.260 (b)
x/H=1.20
x/H=1.80
1.5
z/H
z/H
1.2
1
1
0.8
0.6
0.5
0.0
2.5
5.0
7.5
0.0
1.0
2.0
3.0
1.6
2
x/H=0.60
x/H=1.20
x/H=1.80
1.4
x/H=0.315 (a)
x/H=0.630 (a)
1.5
x/H=0.315 (b)
x/H=0.630 (b)
1.2
z/H
z/H
x/H=1.260 (b)
1
1
0.8
0.6
0.5
0.1
10
100
0.1
Ic
10
100
Ic
102
I c,
x/H=0.33 (inf)
x/H=0.33 (sup)
4
x/H=0.66 (inf)
x/H=0.540 (a)
x/H=0.66 (sup)
x/H=0.540 (b)
3
x/H=0.540 (c)
(z-z 0)/
1.6
x/H=1.00 (inf)
x/H=1.080 (a)
x/H=0.540 (d)
x/H=1.00 (sup)
Gauss
z/H
0
0.00
1.2
0.25
0.50
0.75
1.00
1.25
C/C0
5
x/H=0.60 (inf)
x/H=0.60 (sup)
4
x/H=1.20 (inf)
x/H=1.20 (sup)
0.8
x/H=1.80 (inf)
2.0
4.0
6.0
x/H=1.80 (sup)
(z-z 0 )/
0.0
K
2
Gauss
x/H=0.540 (a)
x/H=0.540 (b)
x/H=1.080 (a)
x/H=0.540 (c)
x/H=0.540 (d)
0
0.00
0.25
0.50
0.75
1.00
1.25
C/C0
1.6
z/H
1.2
0.8
0.1
10
100
Ic
103
10000
0.1
z/H
1000
Cond. A (inf)
Cond. A (sup)
0.01
Dn=C0/Cmax
Cond. B (inf)
Cond. B (sup)
Urbano D
D [Briggs]
Urban
Rural D [Briggs]
0.001
0.10
1.00
Cond. E
Cond. H
100
Cond. F
10.00
x/H
Configurao I I
Configuration
10
Configurao IIII
Configuration
Configurao IIIIII
Configuration
Configurao IVIV
Configuration
3 Conclusions
The dilution analysis held and the comparison with
the curves of minimum dilution, show good agreement
with the experimental values obtained by & Wilson
[1988] and Saathoff et al [1998], for small gaseous
emissions at the level of the top of the buildings. From
the analysis of concentration fluctuations profiles, the
general behaviour is similar to that obtained by other
authors [Li & Meroney, 1983, Fackrell & Robins, 1982],
noting similar tendencies, even if the results obtained
by Li & Meroney [1983] are smaller at the plume
boundaries.
On the other hand, it is important to note that there is
a major asymmetry between the upper and lower
regions, consequence of the buoyancy generated by
the emission of a light gas as the helium. In both
theoretical models, for the analysis of the standard
deviation of the fluctuations, and in the experimental
studies, this asymmetry is not considered.
The plume emission and buoyancy conditions do
influence the process, but this influence is relevant only
in the configurations corresponding to the single
building near the emission source.
The distributions of the mean concentrations and of
the concentration intensities are, in general, similar to
gaussian configurations. The values measured in this
work, are in good agreement with the results of
theoretical models as well as experimental values
obtained in other studies. However, the possible
asymmetry of the concentration profiles is seldom
studied in the literature. This behaviour requires a
probabilistic
analysis
of
the
instantaneous
concentrations which is under study, including
considerations regarding the inttermitence in the
dispersion process.
104
Dmin (M = 0.245)
Dmin (M = 0.414)
1
0,1
1,0
10,0
x/H
Acknowledgments
The authors are gratefull to CAPES and CNPq for
then financial support conceded to two first authors.
References
Blessmann, J. The Boundary Layer Wind Tunnel of the
UFRGS. Journal of Wind Engineering and
Industrial Aerodynamics, Amsterdam, vol.10, 1982.
pp. 231-248.
Chui, E., Wilson, D. [1988], Effect of varing wind
direction on exhaust gas dilution, Journal of Wind
Engineering and Industrial Aerodynamics, 31, 87104.
Fackrell, J. E, Robins, A. G. [1982], Concentration
fluctuations and fluxes in plumes from point
sources in a turbulent boundary layer, J. Fluid
Mech., vol. 17, pp. 1-26.
Hanna, S. R. [1984], Concentration fluctuations in a
smoke plume, Atmospheric Environment, vol. 18,
No. 6, pp. 1091-1106.
Li, W., Meroney, R. N. [1983], "Gas dispersion near a
cubical model Building. Part II: Concentration
fluctuation measurements", Journal of Wind
Engineering and Industrial Aerodynamics, 12, 3547.
Robins, A., Castro, I., Hayden, P., Steggel, N., Contini,
D., Heist, D. [2001], A wind tunnel study of dense
gas dispersion in a neutral boundary layer over a
rough surface, Atmospheric environment 35,
2243-2252.
Saathoff, P., Stathopoulos, T., Wu, H., [1998], The
influence of the turbulence en near field dilution of
exhaust from building vents, Journal of Wind
Engineering and Industrial Aerodynamics, 77&78,
741-752.
Zannetti, P. [1990], "Air pollution modeling: Theories,
computational methods and available software",
Comp. Mech. Publications, Van Nostrand Reinhold,
New York.
1 Introduction
This paper describes a wind tunnel study conducted
for the Mirant Potomac River Generating Station
(MPRGS) located as shown in Figure 1. The study was
commissioned by Mirant because a USEPA
recommended computer dispersion model, AERMOD
[1], predicted high impacts of plant emissions on a
nearby high-rise tower (Marina Towers) as shown in
Figure 2. The tower was built near the plant in the
1970s, without the benefit of site-specific modeling or a
wind tunnel study, long after the MPRGS was built in
1949. The heights of the stacks at MPRGS were
restricted due to the proximity of the power plant to
Reagan National Airport.
105
2 Similarity Requirements
To model plume trajectories for EBD determination,
the velocity ratio, R (Ve/Uh), and density ratio, (s /a)
were matched in model and full scale, where Uh = wind
velocity at stack top (m/s), Ve = stack gas exit velocity
3
(m/s), s = stack gas density (kg/m ), and a = ambient
3
air density (kg/m ). In addition, the stack gas flow in the
model was fully turbulent upon exit as it is in the full
scale.
To simulate the airflow and dispersion around the
buildings, the following criteria were met as
recommended by EPA [10]: 1) all structures within a
518-m radius of the stacks were modeled at a 1:300
scale reduction; 2) appropriate mean and turbulent
approach boundary layer was established; 3) building
Reynolds number independence was verified through
testing; 4) a neutral atmospheric boundary layer was
established simulating an approach surface roughness
of 0.79 m for wind directions of 175-360 (urbanized
sector) and 0.15 m for all other wind directions (water
and low roughness sector).
The above scaling parameters were used to
determine the model operating conditions. It should be
noted that the use of these scaling parameters is the
recommended method for determining GEP stack
heights by EPA [11] and have been used on past EBD
studies. The use of these scaling parameters does not
include an exact simulation of full buoyancy, and as a
result, full-scale plume rise is underestimated (i.e., a
conservative scaling approach). If one wants to compare
the wind tunnel results with AERMOD, the full scale
source parameters have to be back calculated from the
conditions set in the wind tunnel by using the
appropriate buoyancy and momentum scaling method.
[10] These full scale conditions are provided in Table 1.
106
107
5 Results
108
BS4
50.0
BS4
0.0
50.0
45.0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
310
-50.0
XBADJ (M)
40.0
BUILDHGT (m)
35.0
30.0
-100.0
-150.0
25.0
-200.0
20.0
-250.0
15.0
10.0
-300.0
Wind Direction (degrees)
5.0
EBD
0.0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
230 240
280 290
300
310
320
330 340
350 360
EBD
BPIP
BS4
BS4
60.0
40.0
180.0
160.0
20.0
YBADJ (M)
140.0
120.0
BUILDWID (M)
BPIP
0.0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
310
350 360
-20.0
100.0
-40.0
80.0
60.0
-60.0
40.0
-80.0
Wind Direction (degrees)
20.0
EBD
0.0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
310
320
BPIP
BS4
180.0
160.0
140.0
BUILDLEN (M)
120.0
100.0
80.0
60.0
40.0
20.0
0.0
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
310
320
BPIP
109
BPIP
6 Conclusions
This analysis has demonstrated that EBD and BPIP
determined building dimension inputs are significantly
different. The EBD building dimension inputs are based
on a characterization of the wake effects created by all
site structures. The BPIP determined building inputs are
based on logic algorithms that consider building tiers,
building spacings and building angles to the wind.
Unfortunatly, the BPIP inputs may or may not be
appropriate to characterize building wake effects for the
site under evaluation and may result in over- or
underpredicted concentrations.
For this particular study, the effect of the EBD values
on concentration predictions is discussed elsewhere.[9]
References
1. Cimorelli, A.J., S.G. Perry, A. Venkatram, J.C. Weil,
R.J. Paine, R.B. Wilson, R.F. Lee, W.D. Peters, and
R.W. Brode. AERMOD: A Dispersion model for
Industrial Source Applications. Part I: General
Model Formulation and Boundary Layer
Characterization. JAM, 44, 682-693. American
Meteorological Society, Boston, MA. (2005).
2. Petersen, R. and J. Reifschneider. Wind Tunnel
Modeling Evaluation for the Mirant Potomac Power
Generating Station, CPP, Inc. Wind Engineering
and Air Quality Consultants Report 05-3527, Fort
Collins, CO., August 2006.
3. Tikvart, J.A. , Chief, Source Receptor Analysis
Branch, United States Environmental Protection
Agency, Letter to Brenda Johnson, Regional
Modeling Contact, Region IV and Douglas Neeley,
Chief Air Programs Branch, Region IV, July 25,
1994.
4. Petersen, R. L., D. N. Blewitt, and J. A. Panek,
Lattice Type Structure Building Height
110
P. Saathoff
Centre for Building Studies
Concordia University
saathoff@bcee.concordia.ca
T. Stathopoulos
Centre for Building Studies
Concordia University
statho@bcee.concordia.ca
Abstract
A wind tunnel study was carried out to investigate the influence of a rooftop structure (RTS) on the dispersion
of exhaust from a downwind rooftop stack. Previous studies have shown that downwash produced by the RTS
can significantly increase roof level concentrations compared to those obtained with a flat-roofed building
[Saathoff et al. (2002, 2003), Gupta et al. (2005)]. These studies indicated that the severity of RTS downwash
depends on a number of factors such as building height, wind direction and crosswind width of the RTS. Of
primary importance is the location of the RTS with respect to the separated flow region on the roof of the
building. For downwash to occur, the RTS must be outside the separated flow region.
In the present study, the downwash effect of an RTS was investigated for different stack locations. The
experiments were performed on a square plan building with a height of 15 m and a width of 50 m. The RTS
height (h) and length (l) were 4 m and 8 m, respectively, and the width (w) varied from 10 m to 50 m. Figure 1
shows a plan view of the test building.
Wind tunnel experiments were carried out for five stack locations (S1-S5), as indicated in Figure 1. The distance
from the stack to the RTS (xc) varied from 0.5h to 2.5h. Results were obtained for stack heights of 0.25h to
1.75h and for wind directions () of 0 and 45 degrees. A key parameter is the exhaust momentum ratio, M,
defined as the ratio of exhaust speed (we) to wind speed at building height (UH). Values of M varied from 1 to 5,
which is a typical range for building emission sources. Experiments were carried out with and without the RTS
using SF6 as the tracer gas. Mean concentration measurements were obtained on the roof surface along the
plume center-line.
The study aims at proposing simple rules of thumb in order to determine the required stack height (hreq) to avoid
the downwash effect of an RTS. Figure 2 shows the variation of normalized hreq with exhaust momentum M for
stacks S1 to S5 for an RTS with w/h = 7.5. As expected, the value of hreq decreases significantly as M and xs
increase. Figure 2 shows that for a typical design M-value of 2.0, the required stack height is reduced from 2.5h
(10m) to approximately 1.5h (6m).
= 0
UH
= 45
UH
w/h = 7.5
x RTS
y
S1
S3
L = 50
S1
S2
S2
h s /h
xS
S4
S5
S3
S5
h
H = 15m
Stack
S1
S2
S3
S4
S5
S4
0
0
xs/h
0.5
1.0
1.5
2.0
2.5
Exhaust momentum
(M)
M
References
Saathoff, P., Gupta, A., Stathopoulos, T., Lazure L., 2003. Effect of roof top structures on the plume from a
nearby stack. Proceedings of 11th International Conference on Wind Engineering, Lubbock, Texas, USA, June 2-5.
Saathoff P., Stathopoulos T., Lazure L, Peperkamp H. (2002) The influence of rooftop structure on the dispersion of
exhaust from a rooftop stack ASHRAE Transactions, v.108, Pt.2
Gupta, A., Saathoff, P., Stathopoulos, T., (2005) Effect of building orientating on downwash effect of rooftop
structures, Proceedings of Physmod 2005, London, Ontario, Canada, August 24-26
111
Introduction
Urban pollutant concentrations are controlled by two
dispersion processes, which are acting on similar
time scales: horizontal advection through the urban canopy and turbulent exchange with cleaner air
above (ventilation). Within the urban canopy layer
local flow characteristics are determined by the surrounding buildings. Wake interactions result in highly
efficient mixing and diffusion processes for momentum and scalar properties. The scales of the larger
eddies involved in that process are determined by
the building dimensions. Therefore, a greater surface roughness caused by geometrical diversity and
variable building heights induce high turbulence levels which in turn intensify dispersion processes and
canopy ventilation. On the other hand, increased
momentum exchange from the flow to the surface
due to form drag causes a reduction of the advection
velocity through the canopy. Whether urban canopy
ventilation is increased or decreased as roughness
is increased thus depends on the relative magnitudes of these opposing effects (Britter and Hanna,
2003).
Flow parameterizations are often based on morphological descriptors such as plan area density
p and frontal area density f . Oke (1987) for instance described the flow in and above the canopy
as isolated, wake interference or skimming flow, with
thresholds based on values of the frontal index f .
However, based on systematic windtunnel experiments over a wide range of urban-like roughness arrays, Hall et al. (1996) pointed out the importance
of height variability in inhibiting the skimming flow
regime. Cheng and Castro (2002) performed windtunnel experiments over idealised urban-like rough
surfaces and found that that an array with nonuniform height roughness elements with the same
plan area density p and frontal area density f as
an uniform height array is more efficient in generating surface stress. Both studies indicate that certain
dynamical mechanisms acting in more realistic geometries may not be captured in flow parameterizations based on p and f .
Another important factor in a morphologically
based parametrization of urban canopy turbulence
may be the skewness of the building height distribution. This concept is supported by results from
a recent work of Heist et al. (2005). They studied
the effect of a single high-rise building on residence
times of pollutants in a regular array of cubical buildings and found that directly upwind of the tall building
characteristic time scales of the pollutant venting decrease significantly as the building height increases.
This finding suggests that ventilation is particularly
enhanced by rapid vertical dispersion in the wake of
tall buildings.
Urban pollution is caused by numerous (mobile)
point sources. However, for studies at neighbourhood scale it is helpful to define an area source.
The naphthalene sublimation technique provides a
means of directly simulating an area source and
113
also measuring the total area-averaged transport of the plan area density and frontal area density were
scalars out of any given geometry. The ventilation equal, i.e. p = f = 25%.
efficiency of the surface can be quantified by calculating a transfer coefficient (equivalent to a Stanton number, in engineering terms). A full description
of the technique and its use for the study of scalar
transfer from complex surfaces is reviewed by Goldstein and Cho (1995). Recently, Barlow and Belcher
(2002) and Barlow et al. (2004) applied the method
to study the ventilation of idealised 2-dimensional
street canyons.
A key objective of the present study was to investigate the effect of varying roof height on scalar
1: Sketch of 3D and plan view of one unit for array with uniform
dispersion within and above the urban canopy. The Figure
heights (C10S). All dimensions in mm, element heights are indicated on
rough surfaces chosen for the study were two of the plan view. Mean flow along x-axis.
those studied by Cheng and Castro (2002), having equal f and p but with respectively uniform
and non-uniform height buildings. The use of Flame
Ionisation Detectors (FIDs) for the measurement of
naphthalene concentrations was required. The development of this technique, which is entirely novel,
is described in Section 1. A careful study of the effect of Reynolds number on the measurements is
presented in Section 2. Finally, the transfer coefficients and concentration fields for the two surfaces
are presented and discussed in Sections 3 and 4.
Figure 2: Sketch of 3D and plan view of one unit for array with non-uniform
heights (RM10S). All dimensions in mm, element heights are indicated on
the plan view. Mean flow along x-axis.
Experimental Set-up
Amodel
0.02,
Awindtunnel
p 2
< 0.006.
x U2
1.3
To simulate a scalar area source, the naphthalene sublimation technique was used (Barlow and
Belcher, 2002; Goldstein and Cho, 1995). Naphthalene sublimes readily at room temperature, and
if applied to a surface, a thin layer of saturated naphthalene vapour is rapidly formed and maintained in
the immediate vicinity of the surface. Given that
the surface temperature is uniform and constant for
the time of an experimental run, the naphthalene
vapour pressure above the surface will also be uniform and constant, representing a constant concentration boundary condition. The area- and timeaveraged flux from the surface F is given by:
F = w T (S ) ,
(1)
where over-bars denote temporal averages and angled brackets denote spatial averages. w T is the
transfer velocity and (S ) is the difference between the mean naphthalene vapour density at the
active surface and the mean naphthalene vapour
density in the free stream. The naphthalene vapour
density in the free stream was negligibly small
( 5 108 kg m3 , whith typical surface vapour
1.2 Roughness surfaces
densities of S 5 104 kg m3 ). The flux density
The two urban-like roughnesses with different ge- F adjacent to the source is derived directly from the
ometries were composed of many repeated units, mass loss m of naphthalene from the coated area
each 80 80 mm in size and made of moulded A taking place during an experimental run of time t:
plastic material (Acrylonitrile Butadiene Styrene). A
single unit consisted of 16 uniform height cubes
m
F =
.
(2)
(C10S), or non-uniform height cuboids respectively
At
(RM10S), which were arranged in staggered arrays
as shown schematically in Figures 1 and 2. The
The ideal gas law gives the mean naphthalene
cube side length was 10 mm. For both geometries, vapour density at the surface for a given surface tem(X, Y, Z) are the stream-wise, lateral and vertical
coordinates respectively. The plane X = 0 is at the
beginning of the roughness, Y = 0 denotes the test
section centre line and the plane Z = 0 is at ground
level.
114
perature
S =
eS
RN TS
(3)
13.57 3729
TS
(4)
wT =
(5)
The relationship between transfer velocity and reference windspeed U is an independent function for
each surface geometry and source location:
wT = CT U
(6)
N = .
S
(7)
For a fixed emission rate source, the vapour density would be normalized by a given reference windspeed Uref and area emission rate QA to give
F =
Uref
.
QA
(8)
(9)
1.5
1.5.1
Measurement techniques
Temperature
1.5.2
Mass
1.5.3
Velocity
The mean boundary layer flow windspeed profiles were measured using a miniature hot wire
anemometer probe (55P11, Dantec). During all exThe objective was to measure fluxes and vapour periments, the reference free stream velocity was
densities due to an area source of one unit size continuously recorded with a Pitot-static tube which
within the roughness array, with leading edge at was mounted at (Xref , Yref , Zref ) = (2.7, 0, 0.45)m and
115
its value was used to close the tunnel speed control obstacle height H as a characteristic lengthscale of
loop to maintain velocity within 0.05ms1 .
the obstacles giving
1.5.4
ReH =
Concentrations
The naphthalene vapour concentrations were measured using a fast flame ionisation detector (HFR400
fast FID, Cambustion). As the measurement of the
naphthalene vapour (C10 H6 ) using a FID system is
entirely novel, the following procedures were developed to ensure accurate concentration measurements.
Due to the lack of available naphthalene calibration gas, the FID was calibrated against propane
calibration gas at concentrations of 151 ppmv and
749 ppmv. An automated recalibration every 30 minutes was included in the experimental procedure;
background concentrations were also measured every five minutes. Any calibration drifts, and the rising background concentration, were removed from
the data using linear de-trending. For a given surface temperature, the concentration of naphthalene
vapour at the surface, CS , can be expressed in units
of ppm by volume according to
CS = 106
eS
p
(10)
where p is the air pressure. A normalization of measured concentrations with the corresponding saturated source concentration CN = C/CS provides temperature independent results 1 .
Since the concentration sampling system was calibrated against propane but the source concentrations were calculated using Equation 3 and 10, a
methodology was required to convert the measured
propane-equivalent concentrations into naphthalene
concentrations. For this, the inner walls of a small
plastic cylinder were coated with molten naphthalene (cylinder volume: V = 5 cm3 , coated surface:
A 23.5 cm2 ). The cylinder was placed in a temperature controlled water bath. Assuming that the
vapour pressure inside the cylinder was saturated after a reasonably short duration, concentrations and
temperatures were measured at the cylinder centre
for a temperature range of 13 32 C. A clear linear
relationship between the measured and the theoretical concentrations in the studied temperature range
Cpropane = 2.27 Cnaphthalene 11.31
(11)
(12)
Uref
[ms1 ]
C10S
UH
[ms1 ]
ReH
RM10S
UH
[ms1 ]
Re H
24
16
8
5.3
3.5
1.8
3520
2333
1200
4.3
2.9
1.4
2880
1933
933
2.1
Flow profiles
2.2
UH H
Concentration
116
estimated to be approximately 0.05Uref and it was assumed that a similar outflow velocity, WS , would be
appropriate to ensure minimal effect on the flow field.
For each windspeed tested, the source strength was
adjusted to keep a constant ratio of WS /Uref = 0.04.
Figure 3a shows concentration measurements at
a number of locations within the canopy for both surfaces. Concentrations are given in dimensionless
form
C, ps =
C UH H2
Q
(13)
1.4 for both surfaces. Figure 4b shows the percentage deviation from the average concentration
in the windspeed range 20 24 ms1 . Generally,
there is a more systematic dependence of concentration on windspeed for the naphthalene area
source compared to the point source (Figure 3), and
only for U > 20 ms1 do variations appear to approach a constant limit, within the experimentally derived reproducibility of the measurements, marked
in grey (3%). In comparing measurements made
at Uref = 10 ms1 , Figure 4b shows percentage deviations from the high speed results of up to 40%
for C10S and 30% for RM10S in comparison to 15%
and 10% respectively for the point source. Again,
the magnitude of the deviation depends on measurement location with respect to recirculation zones
around the obstacles.
These tests indicated that concentration fields
measured within the canopy using the naphthalene
area source showed undetectable changes when the
reference windspeed was increased above 20 ms1 ,
and hence Uref = 20 ms1 was used throughout all
measurements of concentration fields presented in
Sections 4.1 and 4.2.
Transfer velocities
Figure 5 shows transfer velocity wT as a function of windspeed UH for each surface. Taking the
complete windspeed range into account, a linear relationship between transfer velocity and windspeed
was observed for all tested roughness geometries
with R2 0.99 (Figure 5a, and Table 2). Following
the discussion in Section 2 on the low windspeed
Figure 4: a) Effect of windspeed on concentrations from naphthalene area limitation of the naphthalene sublimation method, it
source at different locations below canopy height for C10S and RM10S
surfaces. b) Percentage deviation from average concentration in the was concluded that the non-zero offset was a result
windspeed range 20 24 ms1 (grey area indicates 3% reproducibility of mixed convection processes at low windspeeds.
range).
However, when data points for lower windspeeds
were excluded, the measurements were sufficiently
Figure 4a shows the results from concentration well described by zero offset linear relationships,
measurements over a range of windspeeds using with R2 0.92 (Figure 5b and Table 2). A repeatathe naphthalene area source described in Section bility of 1.5% was determined for the fit parameters
117
Table 2: Observed linear relationships between transfer velocity wT and wind speed Uref or UH respectively, for the complete windspeed range, and for
high windspeeds only.Dynamical properties of each surface included for comparison.
Dynamical
propertiesa
C10S
RM10S
CRM10S
a Taken
u
Uref
UH
Uref
0.058
0.063
-
0.22
0.18
-
High windspeeds
wT = d Uref ;
wT = e UH ;
All coefficients 103
d
e
R2
1.43
1.27
1.54
1.78
1.68
1.89
6.52
7.07
8.56
6.9
8.9
8.9
8.10
9.32
10.5
0.92
0.97
0.95
0.99
0.99
0.99
Concentration Mappings
Figure 6a shows the mean concentration distribution at Z = 0.3H for the C10S surface. The concentration distribution shows a very regular pattern with
increasing magnitude in the direction of the flow (left
to right). This is due to stream-wise advection of
naphthalene vapour. A regular pattern of concentrations is observed around individual roughness elements: lower concentrations ahead of the roughness
element, and higher concentrations behind. This is
consistent with LES simulations of flow for the same
experimental layout (Z. Xie, pers. comm.), which
show higher velocities in the impact zone in front
of an roughness element, and lower velocities and
strong updraughts in the wake of each roughness element. Hence, the vertical transport is controlled by
the repeated flow structures generated by individual
roughness elements.
Figure 6b shows that roughness element height
variation has a clear influence on the transport processes near the ground. The concentration pattern
is no longer regular and the stream-wise increase
118
tions in the wakes of roughness elements intersecting the measurement plane act as point-like sources.
The plume shape is thus strongly influenced by
the locations of these elevated sources. For the
measurement planes at Z = 1.8H and Z = 2.4H, the
plume shape becomes more consistently Gaussian
further downstream, indicating effective mixing.
Conclusions
Naphthalene coated surfaces were used to represent area sources at street level and thereby to
study ventilation from two different urban-like geometries at neighbourhood scale: one with uniform height roughness elements, the other with nonuniform heights. Area-averaged scalar fluxes were
measured and the results represented in terms of a
dimensionless transfer coefficient wT /Uref . Naphthalene concentrations were measured with high spatial
resolution in two horizontal planes above the active
source area to capture the scalar transport at street
scale. Coarse mesh concentration measurements
at four different heights in the wake of the active surface were also conducted to study differences in the
plume spread arising from the different geometrical
surface characteristics. A novel Fast Flame Ionisation Detection technique was developed to measure
naphthalene vapour concentrations.
For both of the geometries studied, the plan
area density as well as the frontal area density
was p = f = 25%. Two different height distributions were studied, one uniform and the other randomly (Gaussian) distributed but with the same
mean height. The difference in height distribution
had a significant effect on the ventilation efficiency of
the geometries considered. The transfer coefficient
was calculated using two different reference velocities. Using UH , which represents the flow at the top
of the canopy, resulted in an increased ventilation
119
Acknowledgments
The authors are very grateful for the outstanding
technical support from Paul Hayden, Tom Lawton and Allan Wells at the EnFlo laboratory. The
project was funded by the Engineering and Physical Sciences Research Council, grant number
GR/S71798/01
References
Barlow, J. F. and S. Belcher, 2002: A wind tunnel
model for quantifying fluxes in the urban boundary
layer. Boundary-Layer Meteorol., 104, 131150.
Barlow, J. F., I. N. Haman, and S. Belcher, 2004:
Scalar fluxes from urban street canyons. part i:
Laboratory simulations. Boundary-Layer Meteorol., 113, 369385.
Britter, R. E. and S. R. Hanna, 2003: Flow and dispersion in urban areas. Annu. Rev. Fliud Mech.,
35, 469496.
Cheng, H. and I. P. Castro, 2002: Near wall flow over
urban-like roughness. Boundary-Layer Meteorol.,
104, 229259.
C.R.C., 1993: Handbook of Chemistry and Physics.
CRC Press Inc. 74th edn.
Goldstein, R. J. and H. H. Cho, 1995: A review of
mass transfer measurements using naphthalene
sublimation. Exp. Thermal and Fluids Sci., 10,
461434.
Hall, D. J., R. Macdonald, S. Walker, and A. M. Spanton, 1996: Measurements of dispersion within
simulated urban arrays a small scale wind tunnel study. BRE Client Report 178/96, Build. Res.
Establ., Garston, Watford, UK.
Heist, D. K., L. A. Brixley, S. G. Perry, and G. E.
Bowker, 2005: Residence time measurements in
an array of buildings. In Proceedings of PHYSMOD 2005, Int. Workshop on Physical Modelling
of Flow and Dispersion Phenomena.
Oke, T. R., 1987: Boundary Layer Climates. Routledge, Taylor & Francis Group, 435 S, 2 edition.
Pascheke, F., J. F. Barlow, and A. Robins, 2007:
Wind tunnel modelling of dispersion from a scalar
area source in urban-like roughness. BoundaryLayer Meteorol., submitted.
Uehara, K., S. Wakamatsu, and R. Ooka, 2003:
Studies on critical reynolds number indices for
wind-tunnel experiments on flow within urban areas. Boundary-Layer Meteorol., 107, 353370.
120
1 Introduction
The emission of traffic exhausts is one of the major sources for airborne pollutants in urban areas.
In particular, narrow urban street canyons with
large traffic volume are subject to high pollutant
concentrations. Thus, it is important for the healthiness of inhabitants that sufficient natural ventilation is ensured which dilutes and removes the
traffic emissions. A large number of studies have
been devoted to pollutant dispersion processes in
urban street canyons in the past, however, all of
them were restricted to empty, obstacle-free street
canyons (Chang and Meroney, 2003; Gerdes and
Olivari, 1999; Kastner-Klein et al., 2001; Pavageau and Schatzmann, 1999). By contrast, the
central question of the present investigation is, to
what extent avenue-like tree plantings in urban
street canyons influence the natural ventilation.
y
line source
model tree
x
L = 180 m
u
roughness
elements
concentration
measurement
taps
u(z)
D = 0,30
2 Approach
z
A
W = 18 m
H = 18 m
121
aspect ratio L/H > 7, a distinct region with the canyon vortex being the solely dominating vortex
structure can be observed.
In Figure 3, the normalized pollutant concentrations at the inward canyon walls are shown. These
concentrations have been made dimensionless
according to the formula
Capture Detector (ECD) yielding mean concentrations. Laser-Doppler-Velocimetry (LDV) was used
to measure flow velocities in the street canyon
and to determine the rates of vertical air exchange
between street canyon and atmospheric flow at
roof top level.
The numerical computations have been performed by using the commercial CFD-code
FLOVENT. This software package offers a LVEL
k-H turbulence model to close the ReynoldsAveraged-Navier-Stokes (RANS) equations. In
near-wall regions, the turbulent viscosity Qt is determined by a blending of the turbulent viscosity
Qt,k-H, calculated by the classical k-H approach, and
the turbulent viscosity Qt,LVEL, calculated by an
algebraic LVEL approach using a characteristic
length and velocity scale. Furthermore, wall functions for the treatment of wall-adjacent cells are
included. The governing equations are numerically
solved on a structured, staggered grid, using the
finite volume method. Underlying discretization
schemes are first order upwind and second order
central differences for convection and diffusion
terms, respectively. The turbulent Schmidt number
Sct in the advection-diffusion equation is 1 and
can not be varied (FLOVENT 6.1, 2005).
(1)
z/H
Wall A
1
0.5
-5
-4
-3
-2
-1
z/H
y/H
Wall B
1
0.5
-5
-4
-3
-2
-1
0
y/H
u(z)
[ ]
4 Measurement Results
Wall A
c
Wall B
Canyon Vortex
Corner Eddy
122
At the leeward wall A, high increases in concentration are evident. Three pronounced areas of
increases are evident. One is in the center part of
the street canyon at y/H = 0 and the others at the
canyon outer parts at 2.5 < y/H < 4. The high
relative increases at the canyon outer parts are
easy to explain. Due to the tree crowns, the corner
eddies are effectively hindered in entering the
street canyon laterally. Thus, one vortex structure
which was serving for ventilation in the outer regions of the tree-free street canyon is missing or
at least significantly reduced in strength. Comparing the flow fields in front of wall A at y/H = 0.5 in
Figure 4 and Figure 6 reveals that the maximum
velocity w+ is only slightly smaller in the presence
of the tree planting. However, because of the limited passage width, the volume flow rate of the
remaining canyon vortex is reduced significantly.
Consequently, the upward streaming flow in front
of wall A contains higher concentrations of pollutants.
1.2
z/H
0.8
0.6
0.4
0.2
-0.4
-0.2
0.2
0.4
x/H
1.2
4.2 Continuous Tree Planting with impermeable Crown of rectangular Cross Section
The relative change in pollutant concentration at
the canyon walls in the presence of a continuous
tree planting with block shaped cross section of 9
m x 12 m when compared to the reference case
(Figure 3) is presented in Figure 5. This configuration represents an avenue-like tree planting with a
leaf-free stem height of 6 m and crowns interfering
with each other, which is common for the urban
environment. In this example, the crown permeability was neglected and Styrofoam was used to
model the trees.
0.8
z/H
impermeable crown
cross section:
9 m x 12 m
0.6
0.4
0.2
-0.4
-0.2
0.2
0.4
x/H
z/H
-5
At the windward wall B, decreases in concentration are visible (Figure 5). Considering the velocity plots of Figure 4 and Figure 6 shows modified flow fields at the roof top level. The upward
moving stream in front of wall A extends farther
into the skimming atmospheric flow in the case of
the street canyon with tree planting. This means
that the polluted air is discharged into higher layers of the above roof wind and is thus better diluted. In addition, analyses of the fluctuating part
of the vertical velocity (not shown here) show enhanced turbulence intensities in the presence of
the tree planting. The air which is entrained into
the canyon in front of wall B is less polluted, leading to the smaller concentrations observed.
Wall A
1
0.5
-4
-3
-2
-1
z/H
y/H
Wall B
1
0.5
-5
-4
-3
-2
-1
0
y/H
Figure 5. Relative change in pollutant concentration [%] for street canyon with impermeable, continuous crown of cross section 9 m x 12 m when
compared to reference case (Figure 3)
123
z/H
Now, the relative change in pollutant concentration for the same street canyon arrangement as
addressed before, but with a permeable instead
an impermeable crown is discussed. In order to
take the crown porosity and permeability into account, an open-pored foam material (foam ppi 10)
with a volume porosity of 97 % and a pressure
loss coefficient O = 250 Pa (Pa m)-1 was used to
model the trees. Figure 7 shows the relative
change in concentration when compared to the
reference case (Figure 3), (top) and when compared to the impermeable crown arrangement,
(bottom).
0.8
z/H
permeable crown
cross section:
9 m x 12 m
0.6
Wall A
1
0.5
-5
1.2
0.4
-4
-3
-2
-1
z/H
y/H
-5
0.2
Wall B
1
0.5
-4
-3
-2
-1
-0.4
z/H
y/H
-5
-4
-3
-2
-1
z/H
-5
-3
-2
-1
0.4
5 Numerical Results
The first step in numerical modeling was to generate the approaching flow and boundary conditions
of the wind tunnel experiments (Gromke and
Ruck, 2005). Figure 9 shows the computed and
measured vertical profiles of mean flow and turbulence intensity profiles u(z) and Ixyz(z) at the origin
of the co-ordinate system (Figure 1) for the empty
domain. The agreement between experimental
and numerical data is quite satisfactory.
Wall B
-4
0.2
y/H
1
0.5
0
x/H
Wall A
1
0.5
-0.2
y/H
Figure 7: Relative change in pollutant concentration [%] for street canyon with permeable, continuous crown of cross section 9 m x 12 m when
compared to reference case (Figure 3), (top)
and relative change in pollutant concentration [%]
when compared to impermeable crown, (bottom).
1.0
0.8
z [m]
u (CFD)
0.6
u (wind tunnel)
0.4
0.2
0.0
0.0
2.0
4.0
u [m/s]
6.0
8.0
1.0
z [m]
0.8
I_xyz (CFD)
0.6
0.4
0.2
0.0
0
10
20
30
Ixyz [%]
40
50
60
124
and windward edge of building A. In a first attempt, numerical computations were performed
without any special gird refinement (coarse grid)
near the building walls. The resulting flow fields at
the roof level above the windward edge of building
A and the street canyon are shown in Figure 11.
1.4
1.2
z/H
1
10
0.8
8
0.6
Reference Vectors
0
2
4
6
8
10
0.4
0.2
0
-2.4
-2.2
-2
-1.8
-1.6
-1.4
-1.2
-1
-0.8
-0.6
x/H
z/H
1.1
1
-0.5
-0.25
0.25
0.5
x/H
125
z/H
z/H
0.8
0.6
0.4
Wall A
1
0.5
-5
1.2
-4
-3
-2
-1
0.2
1
z/H
y/H
Wall B
1
0.5
-5
-4
-3
-2
-1
z/H
-5
-3
-2
-1
-4
-3
-2
-1
z/H
0.4
y/H
-5
0.2
5.2 Continuous Tree Planting with impermeable Crown of rectangular Cross Section
Wall B
1
0.5
Wall A
-4
-0.2
x/H
y/H
1
0.5
-0.4
y/H
z/H
The contour plot of the normalized vertical velocity component w+ inside the street canyon with
refined grid at the roof level of windward building
A is shown in Figure 14. In comparison to the LDV
measurements of the wind tunnel experiment
(Figure 4), the numerical computation results in
significantly lower flow velocities. A reason for
underestimating the velocity is the combination of
first order upwind convection scheme with the
present LVEL k-H model. This means that the canyon vortex strength is weaker and less air is exchanged between canyon flow and atmospheric
above-roof flow. Consequently, the pollutant concentrations, as predicted by the numerical computation, are higher. A comparison between the numerical results of the coarse and refined grid reveals similar vertical velocity fields, but slightly
smaller flow velocities in the case of the coarse
grid (not shown here). This is because of reduced
momentum exchange between the skimming flow
and the air masses below the roof top level, which
is driving the canyon vortex (Figure 11 and Figure
12).
Wall A
1
0.5
-5
-4
-3
-2
-1
z/H
y/H
Wall B
1
0.5
-5
-4
-3
-2
-1
0
y/H
Figure 15. Relative change in pollutant concentration [%] for street canyon with impermeable, continuous crown of cross section 9 m x 12 m when
compared to reference case (Figure 13 bottom)
Like in the experimental results (Figure 5), pronounced increases in concentration are visible at
wall A. Whereas the increases in concentrations
at the canyons outer part at 3.5 < y/H < 4.5
are of same size, both in numerical and experimental results, the relative change in concentration at the canyons middle part is significantly
126
6 Summary
Flow and pollutant dispersion processes inside a
typical urban street canyon with and without trees
have been investigated. Avenue-like tree plantings
consisting of crowns interfering with each other
and forming a rectangular cross-sectioned continuous block have been considered. Wind tunnel
experiments as well as numerical computations
have been performed for single-row tree arrangements placed along the street center axis. Tree
plantings with different crown permeabilities, covering 33 % of the street canyon volume have been
investigated.
For all tree planting arrangements investigated,
an increase in the overall pollutant concentration
inside the street canyon was found when compared to the empty street canyon without trees
(reference case). A more detailed analysis revealed increased pollutant concentrations at the
leeward canyon wall of the upwind building and
decreased pollutant concentrations at the windward canyon wall of the downwind building. High
relative increases were found at the street ends
towards the intersections. The entrainment conditions at the roof top level as well as at the sidewise end cross sections were considerable modified due to tree plantings.
Qualitatively, the CFD results are in agreement
with the wind tunnel experiments. However, the
quantitative agreement is limited. The numerical
computations result in higher pollutant concentration levels and lower flow velocities inside the
street canyon. Furthermore, the importance of
flow separation at the windward edge of building A
for the concentration field inside the street canyon
was demonstrated by the CFD-simulations. As a
result of the grid refinement study, the LVEL k-H
turbulence model together with the first-order upwind convection scheme predicts flow separation
only in the case of a very fine grid near the windward edge of building A.
In Table 1 the normalized wall average concentrations of all configurations investigated are
summarized. The values in the brackets denote
the relative change in concentration when compared to the reference case with refined grid.
1.2
0.8
z/H
impermeable crown
cross section:
9 m x 12 m
0.6
0.4
0.2
-0.4
-0.2
0.2
0.4
x/H
configuration
ref. case
(coarse grid)
ref. case
(refined grid)
exp. result
wall A wall B
19.6
impermeable
crown
(+71 %)
33.5
permeable
crown
(+66 %)
32.6
num. result
wall A wall B
37.3
15.9
26.1
15.0
5.4
3.6
(-33 %)
2.4
(-55 %)
67.1
9.7
(157%)
(-35 %)
Figure 17. Vector plot of mean flow in street canyon with impermeable crown at y/H = 0.5
127
References
Chang, C. and Meroney, R.N., (2003). Concentration and flow distributions in urban street
canyons: wind tunnel and computational data,
Journal of Wind Engineering and Industrial
Aerodynamics, vol. 91, pp. 1141 - 1154.
FLOVENT 6.1 (2005). Users Manuel, Flomerics
Limited, September 2005.
Gerdes, F. and Olivari, D., (1999). Analysis of
pollutant dispersion in an urban street canyon,
Journal of Wind Engineering and Industrial
Aerodynamics, vol. 82, pp 105 - 124.
Gromke, C. and Ruck, B., (2005). Die Simulation
atmosphrischer Grenzschichten in Windkanlen, Proc. 13. GALA Fachtagung "Lasermethoden in der Strmungsmesstechnik", Cottbus,
September 2005, pp. 51-1 - 51-8.
Gromke, C. and Ruck, B., (2006). Der Einfluss
von Bumen auf das Strmungs- und Konzentrationsfeld in Straenschluchten, Proc. 14.
GALA Fachtagung "Lasermethoden in der
Strmungsmesstechnik", Braunschweig, September 2006, pp. 59-1 - 59-10.
Gromke, C. and Ruck, B., (2007a). Influence of
trees on the dispersion of pollutants in an urban
street canyon - experimental investigation of
the flow and concentration field, Atmospheric
Environment, vol. 41, pp. 3387 - 3302.
LDV-Measurements
+
w [-]
0.3
reference case
0.2
impermeable crown
x/H [-]
0.1
-0.50
permeable crown
0.0
0.00
-0.1
-0.25
0.25
0.50
-0.2
-0.3
CFD-Computations
+
w [-]
0.3
0.2
x/H [-]
0.1
-0.50
-0.25
0.0
0.00
-0.1
0.25
0.50
http://dx.doi.org/10.1016/j.atmosenv.2006.12.043
Gromke, C. and Ruck, B., (2007b). Trees in urban street canyons and their impact on the dispersion of automobile exhausts, Proc. 6th International Conference on Urban Air Quality,
Cyprus, March 2007.
Gromke, C. and Ruck, B., (2007c). Flow and
dispersion phenomena in urban street canyons
in the presence of trees, Proc. 12th International Conference on Wind Engineering, Australia, July 2007.
Gromke, C. and Ruck, B., (2007d). Effects of
trees on the dilution of vehicle exhaust emissions in urban street canyons, Special Issue
on Urban Air Pollution in International Journal
of Environment and Waste Management
(IJEWM), paper accepted for publication, paper
submitted on invitation.
Hunter, L.J., Watson, I.D. and Johnson, G.T.,
(1990/91). Modelling air flow regimes in urban
canyons, Energy and Buildings, vol. 15, pp.
315 - 324.
Kastner-Klein, P., Fedorovich, E. and Rotach,
M.W., (2001). A wind tunnel study of organised
and turbulent air motions in urban street canyons, Journal of Wind Engineering and Industrial Aerodynamics, vol. 89, pp. 849 - 861.
Meroney, R.N., Pavageau, M., Rafailidis, S. and
Schatzmann, M., (1996). Study of line source
characteristics for 2-D physical modelling of
pollutant dispersion in street canyons, Journal
of Wind Engineering and Industrial Aerodynamics, vol. 62, pp. 37 - 56.
Pavageau, M. and Schatzmann, M., (1999). Wind
tunnel measurements of concentration fluctuations in an urban street canyon, Atmospheric
Environment, vol. 33, pp. 3961 - 3971.
-0.2
-0.3
exp. result
wall A wall B
0.082
-0.015
0.034
-0.020
(-57 %)
permeable
crown
(-59 %)
0.040
-0.023
0.049
-0.029
0.025
-0.006
-0.057
0.035
impermeable
crown
num. result
wall A wall B
(-74 %)
(-65 %)
(-49 %)
(-79 %)
Acknowledgement
The authors gratefully acknowledge the financial
support of the Deutsche Forschungsgemeinschaft
DFG, grant-no. Ru 345/28.
128
Introduction
Nowadays the pollutant diffusion is a big problem in
the urban build-up area. The character of the flow in the
rural boundary layer is well understood but in contrast
the flow inside the urban boundary layer needs more
exploration. The same situation hold for the diffusion of
the air pollution.
The task of this experiment is to study the flow in the
urban canopy layer - the layer of air in the urban canopy
beneath the mean height of the buildings and trees. Also
the pollutant diffusion from the point source placed
inside the urban area is investigated.
2 Experimental set-up
As the model of the urban build-up area the array of
cubes was used. Two model variants were studied.
Figure 2 shows the scheme of the arrangement of both
models.
4
4
3
2
2
y/H
y/H
1 Experimental Equipment
0
-1
-2
-2
-4
-3
-1
x/H
x/H
10
3 Results
3.1 Flow field
129
b)
3
measured values
average
10
U [m/s]
1
0
-1
nS ww(n)/u2* [-]
10-1
-2
-3
0
20
40
60
80
100
time [s]
10-2
measured z/H = 4
Kaimal
10-3
10-2
10-1
n = fz/U [-]
100
101
8
7
D1 = 0.27
z01= 0.87
D2 = 0.3
z02= 0.94
z/H
4
1.5
3
2
magnitude
0
1.5
2.5
u [m/s]
y/H
1
3.5
0.5
-0.5
-1
-1
-0.5
0.5
x/H
a)
1.5
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
z/H = 0.5
2.5
2
10
magnitude
1.2
1.1
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
1.5
y/H
nS uu(n)/u2* [-]
10-1
10
0.5
-0.5
-1
-2
-1.5
measured z/H = 4
Kaimal
Karman
Simiu&Scalan
10
-3
10
-2
10
-1
n = fz/U [-]
10
-1
x/H
10
130
y/H
magnitude
0.5
2.00
1.94
1.88
1.81
1.75
-0.5
c*
-0.5
0.5
x/H
1.5
z/H
0.5
0
-0.5
0.5
x/H
3
2.5
2
x/H = 2.5
x/H = 4.5
1.5
x/H = 6.5
x/H = 10.5
x/H = 14.5
0.5
0
-10
-5
10
y/H
b)
1.4
1.2
1.5
x/H = 2.5
0.8
x/H = 4.5
x/H = 6.5
c*
c*
-1
-1
b)
0.6
x/H = 10.5
0.4
z/H
1.5
mean (u'w'): -0.13 -0.12 -0.10 -0.09 -0.08 -0.07 -0.05 -0.04 -0.03 -0.01 -0.00 0.01 0.02 0.04 0.05
x/H = 14.5
0.2
-8
-6
-4
-2
10
y/H
0.5
0
-1.5
-1
-0.5
0.5
x/H
1.5
2.5
131
2
1.5
1
y/H
0.5
0
magnitude: 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
-0.5
-1
1.4
-1.5
-1
x/H
1.2
1
z/H
0.8
0.6
0.4
0.2
0
-1.5
-1
-0.5
0.5
x/H
1.5
2
y/H
magnitude: 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
z/H
1.5
-1
-1
x/H
0.5
0
-1.5
-1
-0.5
0.5
x/H
1.5
2.5
1.8
2
1.5
1
1.6
y/H
0.5
0
1.4
-0.5
magnitude: 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6
-1
-1.5
z/H
1.2
-1
x/H
1
0.8
0.6
0.4
0.2
0
-1.5
-1
-0.5
x/H
0.5
1.5
Figure 8 vertical cross-section of the flow field, model 1 - first and second plot, model 2 third plot
132
4 Conclusion
The flow field inside two different arrangement of the
model of the urban canopy layer was done. For both
model was found out that the obstacles dont affect the
horizontal direction of the flow at the height z/H = 1.5
above the ground. For model 1 was also discovered that
the center of the leeward eddy in the vertical profile is
moving towards up at the position behind the cube.
Measurement of the pollutant diffusion from the point
source was done for the model 1. The dependency of
the concentration field on the source placing (in the
middle of the cross-roads, in the street canyon) was
found.
Acknowledgments
This work was supported by Institutional Research
Plan AVOZ20760514 and by COST 732.
References
Bedn J., Zikmunda O. (1985): Fyzika mezn vrstvy
atmosfry, Academia, Praha
Stull R. B. (1988): An Introduction to Boundary Layer
Meteorology,
Kluwer
Academic
Publishers,
Dordrecht
Jaour Z. (2001): Modelovn mezn vrstvy atmosfry,
Karolinum, Praha
133
Yen-Shen Lin2
1 Introduction
Dispersion of emitted airborne pollutants in urban
environment is mainly affected by the buildings density
and wind attack angles on the buildings. Due to the
complexity of the buildings arrangements in the urban
region, it is difficult to predict precisely the dispersion of
pollutant by the numerical model. Field study can achieve
the goal in a more precision status. But works of the field
investigation [1] cost much. Wind tunnel experimental
simulation is therefore a feasible alternative. Hanna et al.
[2] had conducted the model simulations and observations
of mean flow and turbulence within simple obstacle arrays.
Mavroidis and Griffiths [3] had studied experimentally the
local characteristics of atmospheric dispersion within
building arrays. Mfula et al. [4] performed physical model
study in wind tunnel for urban building exposure to outdoor
pollution.
In this study, experiments on the dispersion of
pollution in urban environment of cubic building array
in-line configuration with different wind attack angles were
2 Experimental Works
Experiments were conducted in the environmental
wind tunnel. The test section of the wind tunnel had the
cross section of 2m by 1.4m and 12.6m long. The tunnel
was an open suction type and it contracted to the test
section with an area ratio of 4:1. The turbulence intensity
of empty tunnel is less than 0.5 % at the mean velocity of 5
m/s.
Four spires were placed at the entrance of the test
section and roughness elements succeeded to be
arranged 9 m long. The fully developed turbulent boundary
layer was generated as the approaching flow.
An X-type hot-wire incorporating with the TSI IFA-300
constant temperature anemometer was applied to
measure the turbulent flow signals. Output of the analog
signals for turbulent flow was digitized at a rate of 4 K Hz
each channel through the 12 bit Analog-toDigital
converter. Since none of the analog signals containing
significant energy or noise above 1 K Hz, with the Nyquist
criteria, a digitizing rate of 2 K Hz was sufficient. The low
pass frequency for the analog signals is set as 1 K Hz in all
runs of the experiments.
Methane was used as tracer gas and it mixed with the
standard gas. The mixed gas emitted from the stacks as
the sources in the experiments. The rake of sampling
tubes was placed at the sampled position. The rake is
composed of ten tubes. The cam system is adopted as the
pump function to suck the samples to the airbags through
tubes simultaneously. Each sample was taken for 5
minutes. The sampled tracer gas in airbag was analyzed
with FID (Flame Ionization Detector) to count the methane
concentration.
Three kinds of the cubic buildings arrangement are
made, and Fig.1 is the schematic diagram of one kind of
arrangement. In the figure, G is the gap between cubic
buildings, and H is the width and height of cubic building.
The source with a height of 0.5H is placed at 2H distance
in front of the first building of the building group. Wind
attack angles of 0o, 15o, 30o, and 45o are executed in the
experiments.
3 Analysis of Concentration
Dispersion Parameters
3.1 Scaling of the concentration
135
and
CU H H 2
Q
WIND
U (Z )
U ref
Z n
)
Z ref
(6)
(1)
G=H
Source
2H
H
H
Zref = 1m
n=0.21
0.8
( y y ) C ( x, y, z)dy
C ( x, y, z)dy
0.6
Z / Zref
0.4
V 2y
V z2
0.2
(2)
( z z ) 2 C ( x, y , z )dz
C ( x, y, z)dz
0.5
zC ( x, y, z)dy
C ( x, y, z)dy
0.8
0.9
(4)
n = 0.21
Zref = 1m
0.8
0.6
(5)
Z/Zref
0.7
U(z) / Uref
yC ( x, y, z)dy
C ( x, y, z)dy
0.6
(3)
0.4
4 Results
0.2
0
8
12
16
Iu (%)
20
136
24
(7)
2cfLux 2 6
U ( z )[1 (
) ]
U ( z)
TE
R (W )dW
0
1.4
1.2
velocity (Z/H=0.5)
open terrain
(8)
U(x) / U(hS)
Su ( f )
G/H=1
G/H=2
G/H=3
0.8
0.6
10
0.4
10
15
20
25
USu(f)/u'2Lx u
X/H
0.1
Z / Zref = 0.2
observational value
von karman
0.01
TIu
open terrain
50
G/H=1
0.001
0.001
0.01
0.1
fLxu/U
10
100
G/H=2
G/H=3
40
30
20
10
0
10
15
20
25
X/H
137
U(z) / U(H)
0.80.9 1 1.1 0.80.9 1 1.1 0.80.9 1 1.1
0.80.9 1 1.1
0.80.9 1 1.1
Z /H
4
3
2
1
0
0
10
12
14
16
18
20
22
24
26
X/H
(a)
U(z) / U (H)
0.60.70.80.9
(K)
4
3
2
1
0
9E-008
5
7E-008
10
12
14
16
18
20
22
24
Y/H
Z /H
26
X/H
(b)
5E-008
1
-1
4E-008
-3
3E-008
-5
U(z) / U(H)
0.50.60.70.80.50.60.70.80.50.60.70.8
0.50.60.70.8
0.50.60.70.8
10
12
14
16
18
20
22
24
X/H
2.5E-008
2.2E-008
(a)
4
3
2
1
0
Z /H
10
(K)
8
7.2E-008
10
12
14
16
18
20
22
24
26
Y/H
U(z) / U(H)
0.70.80.9 1
5.2E-008
(c)
0.6 0.8 1
6.2E-008
X/H
0.6 0.8 1
4.2E-008
0
3.2E-008
-2
0.6 0.8
3E-008
-4
Z /H
4
3
2
2.6E-008
-8
2.2E-008
-10
2.8E-008
-6
10
12
14
16
18
20
22
24
X/H
10
12
14
16
18
20
22
24
(b)
26
X/H
(K)
8
(d)
2.5E-007
4.3 Concentration
8E-008
4
7.2E-008
2
Y/H
6.2E-008
5.2E-008
4.2E-008
-2
3.2E-008
-4
3E-008
-6
2.5E-008
-8
2.2E-008
2
10
12
14
X/H
(c)
138
16
18
20
22
24
(K)
8E-008
6
5.2E-008
Y/H
G/H=2
X/H=1
street 1(X/H=4)
4.32E-008
street 2 (X/H=7)
-4
X/H=25
3E-008
-6
-8
X/H=13
6E-008
3.2E-008
-2
2.3E-008
2
10
12
14
16
18
20
22
X/H
4E-008
(d)
Fig. 8 Concentration contours for different building array
gap at Z/H=0.5, =0o; (a)open terrain, (b)G/H=3, (c)G/H=2,
(d)G/H=1
2E-008
-8
-4
Y/H
1.2E-007
(c)
Open terrain
X/H=1
8E-008
street 1 (X/H=4)
1E-007
X/H=13
G/H=1
X/H=1
X/H=25
street 1 (X/H=3.5)
street 2 (X/H=7)
street 2 (X/H=5.5)
8E-008
X/H=13.5
6E-008
X/H=23.5
6E-008
4E-008
4E-008
2E-008
-8
-4
Y/H
2E-008
(a)
-8
-4
Y/H
8E-008
(d)
G/H=3
X/H=1
street1 (x=22.5)
street2 (x=42.5)
X/H=12.5
6E-008
X/H=24.5
4E-008
2E-008
-8
-4
Y/H
(b)
139
sigma y / H
-5
-10
Y/H
sigma (G/H=3)
wind direction: 0deg
wind direction: 15deg
-15
concentreation line
G/H=1
wind direction:0 deg
-20
10
X/H
15
20
25
1.1
sigma z / H
1
Sigma z
open terrain
S/H=1
S/H=2
S/H=3
0.8
0.7
0.6
0
10
X/H
15
20
X/H
12
16
5 Conclusion
1.2
0.9
25
References
140
141
1. INTRODUCTION
143
3.
4.
5.
6.
7.
8.
9.
(1)
C
C
?
<
m
max m health / odor
(2)
144
145
7. Energy Issues
Several factors affect exhaust system energy consumption,
including: 1) the design and operation of the laboratory,
specifically the relative location of exhaust sources and air
intakes, the presence of nearby building elements such as
screen walls and penthouses, the exhaust volume flow rates
and exit velocities, and the chemical utilization within the
fume hoods; 2) the environment surrounding the laboratory,
involving the presence of nearby structures, air intakes, and
other critical receptor locations; and 3) the local
meteorology, specifically the distribution of local wind
speeds and wind directions.
Chemical utilization is the basic criterion used to judge
whether a specific exhaust/intake design is acceptable. An
overly conservative judgment regarding potential toxicity of
an exhaust stream may result in a high-energy-use exhaust
system as volume flow or exit velocity is increased
unnecessarily. A more accurate assessment of the intended
chemical use, with some consideration of future program,
will result in an exhaust system yielding acceptable air
quality while consuming a minimum amount of energy.
Local wind speeds may be used to set exit velocity
targets, as discussed previously. However, exhaust
momentum is the true parameter governing exhaust plume
146
10
147
11
12
13
14
15
16
148
Introduction
The need to understand the behaviour of emissions
from multiple stacks is a matter of considerable interest
in air pollution control studies. Building exhaust systems
often incorporate two or more stacks placed only on a
few diameters apart, for which the effect of factors such
as stack separation distance, stack array pattern and
effluent exit velocity ratio should be considered.
However, in standard design texts such as the ASHRAE
Handbooks, only single exhaust sources are considered
in any detail. ASHRAE [1,2] suggest that a cluster of
stacks may be considered to act as a single stack,
where a cluster is defined as a group of sources all of
which lie within a radius of two stack diameters from the
centre of the grouping. However, there is little physical
basis for this statement nor has is been confirmed by
existing data and analysis of underlying mechanisms.
The well-studied jet in a crossflow (JICF)
phenomenon from aerodynamics is also very relevant to
buoyant plumes where the density distribution, due to
the initial jet-to-crossflow differences and/or crossflow
stratification, plays a crucial role [3-6]. The dominant
149
1 Experimental details
1.1 Plume measurements
repeated with the dye first issuing from all the stacks
and then from each stack in turn. By examining the
different data sets it was possible to observe the spread
of each plume individually and assess its contribution to
the overall plume at each downstream location. The LIF
technique requires a careful calibration to convert image
intensity to concentration. Hence, for each crosssection, a calibration measurement was first carried out,
as described above. The results showed that a linear
relationship had been achieved for all the experiments.
The plume cross-section concentration fields were
obtained from the video images by applying the
calibration charts of pixel intensity against concentration.
Non-dimensionalized
plume rise (z/D)
3
2.5
2
1.5
1
0.5
0
Int e g ral mo d e l
Exp e rime nt
2 / 3 p o we r law
0
20
40
60
80
Non-dimensionalized downstream
distance (x/D)
150
6
5
4
3
2
Sigma y/D
Sigma z/D
1
0
0
20
40
60
80
31 3
151
1000
2/3 Law
1000
CFM model
1S Experim ents
S
z/lb
z/lb
1S Experiments
100
CFM m odel
Uc
2SBS Experiments
Uc
2/3 Law
100
10
10
100
1000
10
x/lb
10
100
1000
x/lb
2/3 Law
CFM model
1S Experiments
Uc
3SBS Experiments
z/lb
S
S
100
1000
2/3 Law
1S Experiments
2SIL Experiments
3SIL Experiments
2SBS Experiments
3SBS Experiments
10
100
1000
x/lb
z/lb
10
100
2 /3 L a w
CF M m o de l
Uc
10
2 SIL E x p e rim e n ts
S
z/lb
10
1 S E x p e rim en ts
100
100
x/lb
1000
10
10
100
1 00 0
x /lb
152
SINGLE
24
2SBS8.7L
21
21
18
18
SIDE-BY-SIDE
24
0 .0
0.14
2SBS8.7R
0.03
21
18
0.14
0. 25
0.03
0. 3
5
0 .1
0 .2
0. 25
14
0.
21
18
18
15
15
0.0 3
14
-0 .
-0 . 03
0.03
03
-0. 03
0.0 3
36
0.
-6
0.14
-0.03
-9
03
0. 47
0.
-3
Y/D
2SBS34.8A
0.
0 .2
0.03
0. 14
0 .0
-0. 03
0.03
12
5
-0 .26
-0 .3
-0 .1
.3
0. 25
-0
.2 5
Z/D
-0
9
6
-0 .0
21
-0.01
-0 .0
Z/D
Y/D
24
-0 .0
Y/D
12
0
1
0 . 03
-0 .
(C-Cmin)/Cs
0.01
03
Y/D
2SBS34.8L
0.02
(C-Cmin)/Cs
2SBS17.4L
0. 2
Y/D
2SBS17.4R
-1
6
3
0. 0
0.02
-2
2SBS17.4A
-3
0 .0
0.04
0.03
0. 14
0.36
0.03
0.03
Z/D
0. 03
0.1 4
12
407.3
0 .0
0.
0.03
36
Z/D
-0.01
-1
0.03
0.4
-2
12
0.36
-3
0. 03
0.
03. 1
0
15
0.
12
0.01
0 .1
0.02
15
0. 14
15
0 .2
0.
(C-Cmin)/Cs
24
0.03
0.04
TANDEM
2SBS8.7A
03
0.05
2SBS34.8R
0.01
0
-3
-2
-1
-0.01
Normalized lateral distance (y/D)
16
14
153
12
10
Z/D
8
6
SINGLE
TANDEM
OBLIQUE (AV.)
SIDE-BY-SIDE
4
2
0
10
12
14
16
18
20
X/D
Figure 10. The z-location of vorticity peaks for the
different jet nozzle arrangements.
15
Circulation
12
Acknowledgments
9
6
SINGLE
TANDEM
OBLIQUE (AV.)
SIDE-BY-SIDE
3
0
10
12
14
16
18
20
X/D
3 Conclusions
The results from the single stack buoyant plume
experiments are in good agreement with previous work.
In comparison with single stack experiments, there was
no plume rise enhancement but, rather, some
diminishment for the two stack side-by-side case.
However, some plume rise enhancement occurred for
three stack side-by-side case, indicating that both the
number of stacks and the stack separation distance
affect the plume rise. As expected, plume rise was
greatly enhanced for the in-line stack cases, because
the individual plumes can quickly mix and re-organize
into a single plume with more buoyancy flux. In all the
experiments, the maximum rise enhancements have
approached the theoretical limit of
13
References
1
154
13
14
15
16
17
18
19
20
21
155
A. G. Robins
Introduction
The accurate knowledge of the rise and spread of
plumes in a crossflow stably stratified is important in the
analysis of pollution dispersion in atmosphere. The
interest in the details of plumes development arises
because the understanding of the plume-crossflow
interaction is still incomplete despite the large
experimental and theoretical efforts of researchers.
What is more the study of plume-crossflow interaction
can be useful in several other applications like, for
example, VTOL aircraft, disposal of sewage into rivers
and estuaries and cooling of blades in turbine. The
frequent occurrence of plumes developing into a
crossflow has lead to numerous studies, in the last
several years, that face the problem from a theoretical,
numerical or experimental point of view (Bornoff &
Mokhtarzadeh-Dehghan 2001; Briggs 1984; Huq 1997;
McDonald et al 2002; Overcamp, 2007). In this paper
nineteen plumes in different physical conditions have
been simulated in the water towing tank at EnFlo
(University of Surrey) by using a stable cross-flow with a
linear stratification. The main characteristics of plume
development including trajectories and vertical spread
are discussed and compared with the predictions of
different models. The analysis represents an extension
of the study reported in Contini et al (2005) that was
limited to plumes developing in a neutral environment. In
particular it has been used the Briggs (1984) formula, an
integral model already described and used in Contini &
Robins (2005) and a new analytical model. The new
analytical model has been developed extending the one
reported in Slawson & Csanady (1971) in order to be
used in the general case of plumes in which both
momentum and buoyant fluxes are important. Results
indicate that the new model is able to predict the
oscillations of the plumes around their equilibrium height
but the frequency is underestimated as well as the
damping of the oscillations.
A fitting procedure has been used in order to
evaluate the entrainment coefficients that best fit
157
C(x, y, z)dy
I( x , z ) v
(1)
Z( x ) !
f
0 zIz, x )dz
f
0 Iz, x )dz
(2)
Vz
f z Z ! 2 Iz, x ) dz 2
0
Iz, x ) dz
(3)
d
ds
d
ds
d
ds
d
ds
Ur 2Us 2rUa Ve
Ur 2Us Us cosJ Ua Ur 2Us sinJ dUdza
Ur 2Us2 sinJ gr 2 Ua U
gr2Us U Ua Ua Us sinJ r 2Z2
(4)
(5)
(6)
(7)
ds
U
Us
J
2r
Plumecent reline
Plume element
x
dr
U dx Ve
dM F
dx U
M
dF
Z 2
U
dx
2
M Ur w
2 Ua U
F Ur g U
(8)
z( x )
3Fo
Zx
Zx
3Mo
sin
2 2 2 1 cos
2
Z E U a
U a ZE U a U a
where:
Fo
Mo
grs2 Ws
U a Us
Ua
Us 2 2
Ws rs
Ua
(9)
(10)
(11)
158
ti
1
ZMo (i)S
arctg
Z
Fo Z
i 1, 2, 3, ...
(12)
z( t )
1/ 3
3Fo
2
2
Z U a E
3
i
z( t i ) (1) 3Mo (1) i sin(Zt ) (1) i 1 sin(Zt ) R i
i
E
ZU a E 2
i
E
R i 1
R i E Z( t i 1 ) Z( t i )
for t i d t d t i 1 with i
0,1,2.....
(13)
t1
1
ZMo S
arctg
Z
Fo Z
(14)
159
Ve
D 'U t E 'U n
(15)
Ua
cm/s
4.6
wU / wz
4
kg/m
-300
D
mm
7
7
10
Q
l/min
1
1
2
Ua
kg/m3
1062
Us
kg/m3
1098
6a
8a
4.6
-284.1
1087
1200
9a
4.85
3
10.36
-75.7
7
2
1032
1127
5
19.75
7
9
4.85
-75.7
7
1
1032
1127
11
10.36
-78.2
7
1
1032
1129
13
19.75
15
1
20
-81.9
7
1031
1129
17
2
19
7.56
-81.9
7
2
1031
1129
21
6.23
-81.9
7
2.7
1031
1129
19.75
23
10.36
-35.1
7
2
1032
1125
25
4.85
27
29
4.85
-114.8
7
2
1036
1141
31
10.36
-117.4
7
2
1036
1141
33
19.75
Table 1. Summary of the different plumes simulated
during experiments.
The density gradient wU / wz used in the measurements
goes from slightly stable environment to a strong
stratification. There are three main differences between
the plumes developing in stable crossflow with respect
to the ones developing in neutral crossflow (Contini &
Robins, 2001 and 2005). The first difference is that
plumes rising in stable crossflow can reach an
equilibrium height ze and, in cases of strong density
gradient and important momentum flux at the source, ze
is reached through a series of damped oscillations in
which the plume reaches a maximum height zmax. In our
experiments evident oscillations are present, near the
source, in 9 cases out of 19; an example is reported in
Fig. 1
The second difference with respect to plumes
developing in neutral crossflow is that the vertical
spread rz of the plume is relevant up to zmax; afterwards
rz is more or less constant (or at least its growth is
strongly reduced) and the plume tends to spread in the
600
y = 0.8409x
500
R = 0.991
400
300
200
100
0
0
100
200
300
400
500
Measured position maximum xm ax (mm)
600
Z
Figure 1. Example of a plume final average image
showing the oscillations around ze. Reference system is
added to the figure.
R = 0.9712
10
11
12
13
350
40
20
0
0
50
100
150
200
<z> (mm)
250
300
350
3F
3M o
3Fo2
o
sin(Zt1)
2
2
2
3
2
ZU E
Z U a M oE
Z U aE
a
(16)
60
300
y = 0.9912x
250
R = 0.8933
200
150
100
50
0
0
50
100
150
200
250
Measured Zmax (mm)
300
350
rz (mm)
80
Zmax
y = 0.8843x
120
100
12
11
10
9
8
7
6
5
4
3
2
y = 0.9408x
250
R = 0.9652
200
150
100
50
0
0
50
100
150
200
Measured Ze (mm)
250
300
160
0.7
Integral model
0.65
Briggs model
0.6
90
RUN 6a
80
70
60
50
Measurements
Briggs model
Integral model
New model
40
30
20
10
0
0
50
100
150
200
250
Distance X (mm)
300
350
400
RUN 7
100
Measurements
Briggs model
Integral model
New model
80
60
40
New model
0.55
20
0.5
0.45
0
0
0.4
0.35
0.3
0
10
20
30
Z/Ua
40
E New model
E Integral
D Integral
Z/Ua>6
Z/Ua<6
0.45
(0.11)
0.44
(0.11)
0.49
(0.05)
0.01
(0.03)
0.39
(0.06)
0.37
(0.05)
0.48
(0.04)
0.58
(0.05)
0.56
(0.06)
0.5
(0.05)
0.026
(0.04)
~0
300
400
Distance X (mm)
500
600
700
350
300
E Briggs
All data
200
Parameter/mode
l
100
250
200
RUN 27
Measurements
Briggs model
Integral model
New model
150
100
50
0
0
161
100
200
300
400
Distance X (mm)
500
600
700
Hi
1 N
z meas x k z mod el x k 2
Nk 1
1 M 2
H
Mi 1 i
(17)
(18)
Parameter/mode
l
E Briggs
E New model
E Integral
D Integral
All data
Z/Ua>6
Z/Ua<6
0.39
0.38
0.47
0.005
0.36
0.35
0.42
0.014
0.53
0.53
0.51
~0
5 Conclusions
The behaviour of buoyant plumes developing in a
linearly stratified stable crossflow has been studied with
experiments in a water towing tank. Measured
trajectories have been compared with the analytical
Briggs model, with an integral model and with a new
analytical models that is presented in this work. The new
Acknowledgments
This research was funded by Ministero della Universit
e della Ricerca Scientifica e Tecnologica in
collaboration with the Conferenza dei Rettori delle
Universit Italiane (CRUI), by British Council and by the
European Community through the Large Scale Facilities
section of the Training and Mobility of Researcher
program, contract ERBFMGECT980117. The authors
wish to thank Dr Paul Hayden and Mr. Tom Lawton at
EnFlo, University of Surrey, for their help in setting up
the experiments and in developing software and
equipment.
162
References
Bornoff R. B., Mokhtarzadeh-Dehghan, M.R. (2001)
Numerical study of interacting buoyant coolingtower plumes, Atmospheric Environment, vol. 35,
pp. 589-598.
Briggs G. A. (1975) Plume Rise Prediction in Lectures
on Air Pollution and Environmental Impact
Analyses, Workshop Proceedings, pp. 59-111,
American Meteorological Society.
Briggs G. A. (1984) Plume rise and buoyancy effects,
in Atmospheric science and power production,
Randerson D. Eds., pp. 327-366, DOE/TIC
27601, USA Dept. of Energy.
Contini D. and Robins A. (2001) Water tank
measurements of buoyant plume rise and structure
in cross flow, Atmospheric Environment, vol. 35,
pp. 6105-6115.
Contini D., Robins A. (2004) Experiments on the rise
and mixing in neutral crossflow of plumes from two
identical sources for different wind directions,
Atmospheric Environment, vol. 38, pp. 3573-3583.
Contini D., Donateo A., Robins A. (2005) Experimental
and modelled plume rise in neutral environments
proceedings of the International Workshop
PHYSMOD 2005, pp. 2-3.
Davidson G. A., (1989). Simultaneous trajectory and
dilution predictions from a simple integral plume
model, Atmospheric Environment, vol. 23, pp. 341349.
Huq P, Stewart E.J. (1996) A laboratory study of
buoyant plumes in laminar and turbulent
crossflows, Atmospheric Environment, vol. 30, pp.
1125-1135.
Huq P. (1997) Observations of jets in density stratified
crossflows, Atmospheric Environment, vol. 31, pp.
2011-2022.
MacDonald R. W., Strom R. K., Slawson P. R. (2002)
Water flume study of the enhancement of buoyant
rise in pairs of merging plumes, Atmospheric
Environment, vol. 36, pp. 4603-4615.
Ooms G. and Mahieu A. P. (1981) A comparison
between a plume path model and a virtual point
source model for a stack plume, Applied Scientific
Research, vol. 36, pp. 339-356.
Overcamp T.J. (2007) Stable plume rise in a shear
layer, Journal Of The Air & Waste Management
Association, vol. 57, pp. 328-331.
Slawson P. R., Csanady G. T. (1971) The effect of
atmospheric conditions on plume rise, J. Fluid
Mech., vol. 47, pp. 33-49.
Weil J. C. (1988) Plume Rise in Lectures on air
pollution modelling, A. Venkatram, J. C. Wyngaard
Eds., 121-166, American Meteorological society.
163
165
Figure 1
References
D. S. Henn and R. I. Sykes, 1992, Atmos. Environ. 26A,
3145-3159.
P. J. Mason et al., 1990, Boundary-Layer Meteorol. 53,
117-162.
E. Fedorovich and J. Thater, 2002, Atoms. Environ. 36,
2245-3355.
1 Introduction
Variation of gas concentration in the field depends
on unsteady meteorological conditions of wind velocity,
wind direction and atmospheric stability. However,
conventional
wind
tunnel
experiments
and
Computational Fluid Dynamic (CFD) models cannot
simulate these unsteady phenomena, because they
assume steady state meteorological conditions.
Environmental
assessments need a long time
averaged concentration for 30 min., 1 hour or 1 year in
actual site. Some empirical formula for a meandering
factor have been used for these needs. However, the
meandering factor depends on a terrain and a stability
condition at each site.
We simulated a meandering effect of wind direction
fluctuation on horizontal gas diffusion over Mt. Tsukuba
near Tokyo, by our original technique using a rotating
turntable. In the wind tunnel experiment, and by the
mesoscale meterological model (RAMS/HYPACT).
2 Field experiments
167
Overlapping
method
under
the
corresponding
meteorological condition to the field experiment at Mt.
Tsukuba under 3 kinds of wind fluctuation variannce;
=4.5, 9.0 and 12.0 deg. (Hayashi et al.
2001) .Correlation of these results in Table 1 indicate
that wind fluctuation variance of 9.0 deg seems to be
good.
Table 1 Correlation coefficient of concentration data
Wind fluctuation
Correaltion
Regression
4.5 deg
0.82
0.59
9.0 deg
0.90
0.89
12.0 deg
0.89
0.98
Fig. 3 Overlapping system to simulate a meandering effect on gas diffusion in wind tunnel
168
Fig. 5 Ground level concentration distribution around Mt. Tsukuba without and with Overlapping
method, where broken curves represent field data (RUN 89-3) and solid ones wind tunnel
The results of conventional wind tunnel experiment
minutes sampling time, they are lower than they
and the Overlapping one were compared with field
value of =9.0 in correspondance with 30
data, as shown in Fig. 5. The Overlapping method
minutes sampling time.
simulates a wide lateral spread caused by
Fig. 7 Lateral plume spreads of wind tunnel
10000.00
`
a
b
c
d
e
1000.00
=
0deg
100.00
Field
Conventional W/T
Overlapping W/T
=
9.0deg
10.00
1000
10000
100000
@
w
i
169
100
withoutterrain
with terrain
3 2
60
1
`
5
10
10000
1
1
X(km )
10
Lateralplum e spread
0.1
withoutterrain
with terrain
1000
W ithoutterrain
100
W ithoutterrain
Overlapping case
10
Conventionalcase
10
100
1000
10000
Downwind disatance
1
0.1
1
X(km)
10
PG (1hr )
D u PG (3 min)
170
360
Observation
315
270
225
180
135
90
45
0
12km
15
4km mesh
21
15
21
15
21
9 15
Time
21
15
21
15
21
Observation
360
315
270
225
180
135
90
45
0
15
250m mesh
1km mesh
21
15
21
15
21
9 15
Time
21
15
21
15
21
Observation
360
W ind direction (deg.)
315
270
225
180
135
90
45
0
15
21
15
21
15
21
9 15
Time
21
15
21
15
21
20
W ind speed (m /s)
Observation
Sim ulation (RAM S)
15
10
5
0
15
21
15
21
15
21
15
21
15
21
15
21
Tim e
20
Observation
15
10
0
15
21
15
21
15
21
15
21
15
21
15
21
15
21
Tim e
20
Observation
15
10
0
15
21
15
21
15
21
9
15
Tim e
21
15
21
171
14
12
Observation
10
8
6
4
2
0
15
21
15
21
15
21
11/10
9 15
Tim e
21
15
21
15
21
11/16
14
12
Observation
10
8
6
4
2
0
15 21
15 21
15 21
11/10
9 15 21
Tim e
15 21
15 21
11/16
14
12
Observation
10
a) Field observation
6
4
2
0
15
11/10
21
15
21
15
21
9 15
Tim e
21
15
21
15
21
11/16
b) Simulation
172
a) Field observation
1000
750
b) Simulation
500
25000
250
0
0
20000
5000
15000
10000
10000
15000
5000
20000 0
1000
750
500
25000
250
0
0
20000
5000
15000
10000
10000
15000
RUN89-2
C/Q(*10-9)s/m 3
100000
1000
100
10
0
2
3
4
Downwind distance (km )
100000
C/Q(*10-9)s/m 3
5000
20000 0
Observation
Sim ulation(RAM S/HYPACT)
Plum e equation
10000
1000
100
10
0
2
3
4
Downwind distancek
173
dE
dt
w
wE
KE
P
wz
wz
dE
dt
w
wE
KE
P
wx
wx
Observation
Sim ulation (RAM S/HYPACT)
Plum e equation
10000
a) MY model
We are now developing a new assessment
system to estimate the pollutant concentration for one
year with RAMS and HYPACT codes, in stead of
conventional assessment scheme using the plume
model, as shown in Fig. 20.
Conventional
method
New method
Meteorological
observation for one
year
Meteorological
simulation for one
year
Gas diffusion
calculation by plume
model, discretely
Gas diffusion
simulation for one
year, continuously
b)Castelli model
Fig. 20 Flowchart of a conventional and a new
environmental assessments
Acknowledgements
Wind tunnel experiments were conducted under
the sponcership by Japan Atomic Energy Research
Institute (Japan Atomic Energy Agency at present).
References
5 Conclusions
174
Bert Holtslag
Harm Jonker
Introduction
TNO in Apeldoorn has been active in the field of
wind tunnel research for more than three decades.
Research activities have been carried out in the field of
wind nuisance, atmospheric dispersion in urban areas,
wind loading on constructions, heavy gas dispersion,
chemical reactive plumes etc.
The most recent wind tunnel in Apeldoorn is,
however ,already more than 20 years old. Furthermore,
due to a reorganisation focussed on centralizing most
environmental research in the Netherlands in and
around Utrecht, the TNO Environment Institute will be
transferred from Apeldoorn to Utrecht. This has led to
the decision to investigate the possibility of building a
new wind tunnel. The concept is to develop a wind
tunnel which is suited for applied as well as for more
fundamental research. TNO, in cooperation and
discussions with universities and other research
institutes, has made a first out-line for such a state-ofthe-art wind tunnel. The main new aspect is the
possibility to generate also thermally stratified flows over
a heated or cooled surface.
The wind tunnel should cover the following areas of
research:
x Air Quality and Safety
x Climate Change
x Built Environment
x Miscellaneous
An analysis of the requirements for these different
areas has led to a first technical lay-out of the wind
tunnel.
175
2 Climate Change
In particular, the description of stable boundary
layers in current numerical weather prediction models,
air quality models and climate models ,is one of the
weakest elements in these models. Although field and
laboratory experiments have been carried out, in fact
more in convective than in stable cases, there is a
general lack in detailed experimental data under
controlled conditions . In case such experimental data
would become available, this would enable the
validation of the current model descriptions more
accurately, including the validation of LES-models,
again especially under stable conditions, see Holtslag,
2006
Furthermore, the current descriptions are valid over
homogeneous surfaces, related to the grid-averaging in
the models. Sub-grid scale effects caused by
heterogeneously heated surfaces, and frictional
convergence driven by a change in roughness length,
are all aspects which could be studied in the new
windtunnel.
The impact of climate change on the processes
close to the surface is a general topic of research. This
3 Built Environment
4 Miscalleneous
The following items can be found in this category,
based on the projects carried out over the years:
x The layout of wind turbine farms, often off-shore,
and the investigations of wind turbine wakes
x Testing of instruments like sonic anemometers
x Force reduction for skaters and bikers, posture,
helmet, garments
In general, this type of research requires high wind
speeds in the tunnel.
Another research area might be the study of cloud
microphysical aspects. A pre-humidified air flow could
be led into a turbulent flow , further downwind seeded
with droplets of a particular size (say 1-10 Pm). The
basis research question is the interaction between
droplets and turbulence and the possible enhancement
of droplet collision coalescence.
Because the wind tunnel would be run in cooperation
with three universities, it will also be applied for
educational purposes.
References
Holtslag, A.A.M. (2006) GEWEX Atmospheric
Boundary-layer Study (GABLS) on stable boundary
layers. BoundaryLayer Meteorology 118, 243-246
Ohya, Y et al (1996) A thermally stratified wind tunnel
for environmental flow studies.Atm. Env. 30, 16,
2881-2887
176
We need well-controlled high Reynolds number turbulence fields to investigate characteristics of atmospheric turbulence fields in which turbulence Reynolds numbers, R, may be
more than 1000. Some large wind tunnels can realize such high Reynolds number flow fields.
However, there are few ones can easily control turbulence characteristics.
The multi-fan wind tunnel in Miyazaki University as shown in Fig. 1 has 99 fans
controlled independently with each other by a computer. We can form various velocity
profiles and add various fluctuations in the test section. In the present work, we tried to realize
quasi-isotropic and homogeneous turbulence fields with large R.
The fans were driven by a single basic wave starting from the different phase with each
other. We checked two basic waves, one was transformed from the Karman type spectrum
and the other was a low-pass filtered random noise. Though each fan induce only streamwise
velocity fluctuation less than 4 Hz, the local shear between neighboring fans can generate
lateral velocity fluctuations. We measured two components of fluctuating velocity by using an
X-probe and hotwire anemometers. Measured velocities were sampled at 4 kHz by a 16bit
digital recording unit and stored in a computer.
The resultant turbulence fields were found to have large turbulence intensity more than
0.1 of mean velocity even near the end of test section. Their homogeneity and isotropy
became better as they flowed downstream, and reached the levels of conventional grid
turbulence. The energy in the lower wavenumber region was excited and the energy gap was
observed in the energy spectra in X < 3 m. But, the spectra in the downstream region of the
test section have clear inertial subrange more than 2 order of bandwidth as shown in Fig. 2.
The turbulence Reynolds number was R=880. It is still less than the value of atmospheric
turbulence, we expect to achieve R>1000 by improving the driving mode.
177
K. OHBA
Introduction
Generally, the occurrence of internal gravity wave
keenly affects heat and momentum transfer and
turbulence production in natural strongly stably-stratified
flow fields [1],[2]. Buoyancy force, originating in the large
negative density gradient of the stably-stratified flow with
local Richardson number Ri0.25, induces internal
gravity waves in it. Its transition makes quite important
roles in the thermo-fluid dynamics of the transfer
problems. In our previous studies, we analyzed the
transition process of the spontaneously generated
internal gravity waves and their effects on heat and
momentum transfer in the same experimental situation
by thermo-anemometer single-point measurements [3][5]. The instantaneous information of local temperature
gradient is also important to understand the occurrence
of turbulence production and counter-gradient heat flux
in the downstream collapse process of the internal
gravity wave.
In the present experiment, a strongly stably-stratified
mixing layer was realized by giving a large positive
temperature gradient in a low turbulence wind tunnel
equipped with a thermal stratification generator. Multipoint simultaneous measurement on temperature and
velocity fluctuations was made using a ladder of 7 coldand hot-wires.
179
installed upstream of a contraction nozzle of the lowturbulence wind-tunnel. It can realize arbitrary
temperature profiles in the test section by independently
regulating the electrical power of each coil heater. A
temperature-controllable ceiling and adiabatic sidewalls
made of vacuum glass were employed for the test
section and thermal contamination disturbing inside
temperature profiles from the outside was almost
completely rejected.
Figure 1 illustrates the ladder probe used for the
spatial measurement. Each three set of I-I type probe is
arranged in the upper and the lower sides of the central
I-X type probe every 6mm. Each sensor is numbered
from Ch.1 to 7 from upper one as denoted in fig.1. They
gave the measuremental span of 36mm. Simultaneous
measurement must be made on the temperature
fluctuation component, , and the velocity fluctuation
components, u and w, in order to get higher order
correlation terms indispensable to minutely understand
the turbulent thermal transfer problems. We employed a
high precision thermo-anemometer system with S/N
ratio of 60dB and a frequency range of DC~5kHz. It is
equipped by a digital delay circuit which can equivalently
compensate the error due to the gap between the coldwire and the hot-wire, X=1mm, based on the frozen
pattern hypothesis, X=UT. Detailed analysis on the
characteristics of streamwise change in spatial structure
of the internal gravity wave and its effect on the heat
transfer and turbulent production are of the authors
special interest.
2 EXPERIMENTAL CONDITIONS
The initial flow conditions of the present experiment
are shown in table 1. A stable stepwise mixing layer with
20.0mm thickness was produced at the entrance of the
test section, X/D=0, where D=0.42m is the span of the
test section. Mean velocity was U0=3.0m/sec in the nonheated region and the maximum temperature difference
was max=26.8K. The maximum local temperature and
velocity gradients reached about 1200K/m and 9.7/sec
at the center of the mixing layer of X/D=0 as shown in fig.
2. The local Richardson number exceeded Ric=0.25 [6]
across the mixing layer where only the internal gravity
wave can be spontaneously generated in it.
'l=1mm
20.0
0.81
Ch.7
d
NB
g
,
(1)
dZ
where g is the gravitational acceleration and is the
coefficient of cubic expansion. Energy density level is
quite low and no peaks are clearly observed at X/D=0. It
remarkably increases downstream below the B-V
frequency. The temperature spectrum at X/D=5 clearly
shows three frequency components of 1.2, 1.8 and
3.0Hz satisfying the three-wave resonance condition,
f1+f2+f3=0 [7]. These facts show that the buoyancy
actively forces on the motion of fluid elements inside the
layer and the internal gravity wave was enhanced from
the smallest perturbations in the upstream region. At
X/D=9, the energy density level is observed to increase
drastically even in the frequency range above NB. It
suggests that the grown-up internal gravity wave begins
to collapse and the turbulence components were
produced through the nonlinear wave-wave interaction.
X/D=0
11
1200K/m
d0
Z/d0
-6
0
0
/max , U/Umax
3.6Hz
4.8Hz
5.4Hz
6.0Hz
10
X/D=11
Z/d0=1.72
10
-4
Prong (X type)
I=0.3mm
Hot-wire
I=2.5mm
l=1.0mm
Cold-wire
I=2.5mm
l=1.0mm
0.21
1200
Flow
Ch.4
26.8
Prong (I type)
I=0.2mm
6mm
Ch.1
3.0
1.8Hz
1.2Hz
X/D=9
Z/d0=1.51
10
3.0Hz
X/D=5
Z/d0=1.10
X/D=0
Z/d0=1.00
NB=6.3
10
20
Frequency [Hz]
180
X/D=5, Z/d0=1.10
23
12
15
0
-6
-12
-18
1
0
Ch.4-5
1
0
Ch.4-6
15
[K]
6
Z[mm]
Ch.4-3
12
-6
Ch.4-7
1
0
NB=5.2Hz
Ch.4-3
1
0
Ch.4-5
1
0
Ch.4-6
1
0
Ch.4-7
1
0
Ch.4-1
Ch.4-2
NB=3.2Hz
20
40
Frequency [Hz]
Frequency [Hz]
14
[K]
6
Z[mm]
40
12
10
0
-6
-12
-18
20
X/D=9, Z/d0=1.51
-12
-18
Coherence [-]
Ch.4-2
[K]
Z[mm]
Ch.4-1
5
Time [sec] (0.5/div)
181
Coh(f)
S (f) 2
4i
44 (f)Sii (f)
(2)
6
5
Z/d0
X/D=0
Ch.1
Ch.1
Ch.7
Ch.7
X/D=0
9
Ch.1
Ch.1
Ch.1
Ch.7
Ch.7
Ch.7
9
Ch.1
Ch.7
0
: w/U0 (0.001/div)
: / max (0.004/div)
Fig. 6 Streamwise change in vertical distributions of (a) heat flux and temperature gradient
and (b) turbulent intensities of w and .
: vertical heat flux (110-5/div),
: temperature gradient (200[K/m]/div)
internal gravity waves in the process of turbulence
production during the collapse of the wave front.
Figure 6(a) compares the distributions of the timeaveraged vertical heat flux, -w/U0max, and the local
temperature gradient, d/dZ. The intensities of the
vertical component of velocity fluctuation, w, and the
temperature fluctuation, , are shown in fig. 6 (b), in
which the arrows indicate the vertical position of each
sensor shown in the figs. 4 and 5. The positive and the
negative values of the heat flux correspond to the downand the counter-gradient heat transfer, respectively. The
initial turbulent intensity is so low that the heat transfer is
quite small at X/D=0 as shown in fig. 6 (a). The local
temperature gradient denoted by a solid line and the
vertical heat flux have similar one-peak distributions at
X/D=0. That is, the heat transfer is dominated by the
usual down-gradient mechanism proportional to the local
temperature gradient there.
At X/D=5, the heat flux and the local temperature
gradient became not to keep such similar profiles to
each other and the heat flux was apparently suppressed
in the region of Z/d0=0.7~1.6. In such a region, the
internal gravity wave is assured to grow up downstream,
as estimated from figs. 4 and 5. The heat flux across the
wave front was suppressed as the internal gravity wave
develops with strong correlation suggests that the
internal gravity wave exerts decisive influence on the
aspects of heat transfer in the stably stratified mixing
layer.
The counter-gradient heat flux is clearly observed in
the hatched region at Z/d0=0.6 and 2.4 of X/D=9, though
the local temperature gradient is still kept positive.
These regions correspond to the upper and the lower
sides of the ladder probe. There, the counter-gradient
heat flux apparently grew larger, as mentioned before.
These heights catched the wave crest of the internal
gravity wave when estimated by the isothermal lines and
the coherences. It must be noticed that the wcomponent of the velocity fluctuation was produced
where the counter-gradient heat flux occurred as shown
in fig. 6(b). With these results, the internal gravity wave
began to collapse from around the wave crests and the
counter-gradient heat flux induced the negative turbulent
Acknowledgments
This work was partially supported by the Japanese
Ministry
of
Education
through
grants-in-aid
(No.15560138).
References
[1] Turner J.S., (1973), Buoyancy Effects in Fluids,
Cambridge University Press.
[2] Lighthill J., (1978), Waves in Fluids, Cambridge
University Press.
[3] Makita H, Mori S, Yahagi A., (1994), Spontaneous
generation of internal gravity wave in a wind tunnel,
Stably Stratified Flows:Flow and Dispersion over
Topography, pp. 81-91.
[4] Makita H, Ohba K, Sekishita N., (2002),
Occurrence of internal gravity waves and countergradient heat flux in a strongly stably-stratified
mixing layer, Advances in Turbulence IX, pp. 605608.
[5] Makita H, Ohba K, Sekishita N., (2005),
Experimental Analysis on the Transition Process
of Internal Gravity Waves in a Strong StablyStratified Mixing Layer, Proceedings of iTi
Conference on Turbulence 2005, p. 89.
[6] Hazel P., (1972), Numerical studies of the stability
of inviscid stratified shear flows, Journal of Fluid
Mechanics, vol. 51, pp. 39-61.
[7] Stewart R.W., (1969). Turbulence and waves in a
stratified atmosphere, Radio Science, vol. 4, pp.
1269-1278.
[8] Thorpe S.A., (1966). On wave interactions in a
stratified fluid, Journal of Fluid Mechanics, vol. 24,
pp. 737-751.
182
POSTERS
Post-processing,
CFD,
coherent-
Introduction
The dispersion and transport of pollutants in urban
environment is a task of high significance in fluid
dynamics research. The flow past buildings can be
characterized by very high Reynolds number. The
critical point of the investigation of this type of flow is the
lack of accurate reference data on the real phenomenon
due to the reliability problems of field tests (Schafer et
al. 2005). The first phenomenon that is usually, however
not always (see f.e.g. Uehara et al. 2000) neglected is
the thermal stratification. The constant density
investigations restrict only to cases when atmospheric
wind is dominant. Due to the very complex geometry of
a city, mostly wind tunnels are the most appropriate
tools for obtaining answers on the required pollution
185
1
tr ( T ) tr (S S T )
2
1
w iu j w j ui
2
(1)
'p tot
r2
U u u rot u Qrotrot (u ) dr
(2)
r1
186
b
Figure 1. Investigated models by Miskam (a) and Fluent
(b) software
187
b
Figure 4. Iso-Q surfaces over the city model (a) and the
simplified matrix of buildings (b). Flow from the upper
left corner to the lower right for both figures.
188
5 Conclusion
In this paper different flow visualization methods were
shown that form a useful tool for understanding the flow
field around complicated geometries, like urban
environments. For the investigation of the flow field two
city models were applied and the flow was determined
via numerical modeling. The model that represented the
inner city of Budapest was used for numerical
simulations by software Miskam. On this model the wall
surfaces did not contain any information thus
characteristics regarding to the volume of the flow field
were extracted only. To show additional information
originating from the wall surfaces flow patterns an
additional simplified city model was applied over which
the flow was simulated by software ANSYS-Fluent.
Surface flow patterns are in strong relation with the
discomfort parameters of the pedestrians.
The flow field was visualized by different properties
deduced from the variables.
189
Acknowledgments
Authors of this paper wish to express their thanks for
NKFP 3A/088/2004 for providing the financial basis to
make these investigations possible.
References
Adrian, R. J. (2007) Hairpin vortex organization in wall
turbulence, Physics of Fluids, Vol. 19, Issue 4, pp.
041301-041301-16
Dezs-Weidinger, G., Stitou, A.; van Beeck, J.,
Riethmuller, M. L. (2003) Measurement of the
turbulent mass flux with PTV in a street canyon
Journal of Wind Engineering and Industrial
Aerodynamics, vol. 91, pp. 1117-1131
Franke, J., Hirsch, C., Jensen, A. G., Krs, H. W.,
Schatzmann, M., Westbury, P.S., Miles, S.D.,
Wisse,
J.A.,
Wright,
N.
G.
(2004)
190
191
4 Computational parameters
The prevailing concentrations of the pollutants are
determined by calculating:
- the background concentrations: the contribution
from an unspecified origin;
- the contribution from local traffic.
In the Netherlands, the concentrations of NO2 and
PM10 are most likely to exceed the limit values. The
chemical conversion of NO to NO2 is taken into account
on the basis of an empirical formula. The number of
days on which the 24-hour mean limit for PM10 is
exceeded is determined on the basis of the annual
mean
concentration.
The
above
mentioned
computational parameters are discussed in detail below.
20 0,53C
and:
'C a , jm >NO2 @ C 25m >NO2 @
a , jm*
20 0,53C
a , jm*
4500
, x ! 500 m
.x d 500m
'C a , jm
and:
'C a , jm
p.C 25 m
5000 x
, ( x ! 500m)
4500
where:
Ca, jm
: the adjustment for the background
concentration
C25m : the contribution from the local traffic route at
a distance of 25 m
p 0.08
Ca, jm*
:
the
unadjusted
background
concentration
x
: the distance from the local traffic route
192
Speed type
NOx
PM10
Private cars
A
0,448
0,367
0,527
0,589
E
0,648
5,916
7,447
9,129
9,783
12,952
Goods vehicles
8,143
10,986
13,192
14,050
18,600
Private cars
0,041
0,037
0,054
0,061
0,070
0,173
0,249
0,332
0,364
0,466
Goods vehicles
0,226
0,321
0,401
0,432
0,546
f NO 2 .Cb ,i >NOx @
where:
Cb,i [NO 2] :
PM10 is exceeded
193
2
ODPM10 = 0.10498.(Cjm[PM10 ]- 31.2) +3.1092.Cjm[PM10
]-31.2)+35
If Cjm [PM10] > 16 g/m3:
ODPM10 = 12
where:
Cjm[PM10]: annual mean concentration of suspended
particles (PM10), calculated using formula 1.1.
ODPM10: the number of days on which the 24-hour mean
concentration
of
PM10
exceeds
50 g/m3.
5 Practical examples
In recent years, the CFD method has been used in a
variety of air-quality surveys. CFD provides added value
compared with the traditional transfer calculations in
scenarios where:
5.1.2
The CFD model
A three-dimensional CFD model is created of both
the existing and the future scenario in the plan
area and its surroundings. Developments which
could be expected if the proposed plan was not
put into effect are extrapolated from the existing
scenario. We call this "autonomous development".
The CFD models of the autonomous development
and the scenario on completion of the plan are
shown in Figs. 4.1 and 4.2.
5.1.3
The results
5.2.2
Autonomous development
The model
194
on
Autonomous development
Autonomous development
The results
195
6 References
[1] MISKAM, Handbuch zu Version 4, Institut fr Physik
der Atmosphre, Dr Joachim Eichhorn, Johannes
Gutenberg University Mainz
[2] WinMISKAMMiskam for Windows Manual,
Ingenieurbro Lohmeyer GmbH & Co. KG
[3]
Concentratiekaarten
voor
grootschalige
luchtverontreinigingen in Nederland rapportage
2006, MNP Bilthoven Nederland.
[4]
Meeten
rekenvoorschrift
bevoegdheden
luchtkwaliteit VROM Nederland
[5] Handleiding CAR II, versie 5.1, TNO ApeldoornNederland
Autonomous development
196
Introduction
Laboratory investigations of flows over urban- or
rural-type roughness above and within the canopy
region are still mainly limited to the evaluation of one- or
two-point statistics that are accessible via hot-wire or
Laser Doppler Velocimetry measurements. Even if ideal
to determine turbulent statistical properties, these
techniques can only provide limited information about
the detailed time-dependent structural characteristics of
the flow and its coherent structures whereas the
important role of these structures on the dynamics of the
flow, on the energy transfers and on the transport
mechanisms is well-established (Finnigan, 2000,
Jimenez, 2004, Zhu, 2007). As pointed out by Castro et
al. in their conclusion (Castro, 2006), the most adequate
experimental technique to investigate the spatiotemporal organization of the flow within and above urban
or vegetation canopies appears now to be the Particle
Image Velocimetry (PIV), which can give access to
instantaneous bidimensional velocity fields with up to
three velocity components. The extensive use of this
technique in the study of the structure of the smoothwall boundary layer (among a wide variety of turbulent
flows) has proved its reliability and efficiency (see for
instance the work of Stanislas and coworkers in France
or Adrian and coworkers in the USA). With its ability to
provide spatially dense measurements, the PIV has
enabled to draw a rather precise picture of the spatiotemporal structure of boundary layer based on the
model of clusters of hairpin vortices (Adrian, 2007).
Recently, PIV has been applied by Zhu et al. (Zhu,
2007) to the study of the flow structure and turbulence in
a wind-tunnel model canopy.
Furthermore,
because
of
its
principle
of
measurement (which will be detailed below) based on
the cross-correlation of small areas of two consecutive
images of the studied flow, velocity fields obtained by
PIV are very close in their nature to those computed by
197
198
0.75 m
110
85
1.30
1.00 m
150
115
1.80
1.50 m
220
170
2.60
2.1
PIV principle
Lx (mm)
Ly (mm)
Resolution
(mm)
199
2.2
PIV setup
Conclusions
References
Adrian, R. J. (2007). "Hairpin vortex organization in wall
turbulence", Physics of Fluids, vol. 19, no 4.
Castro, I. P., Cheng, H. and Reynolds, R. (2006).
"Turbulence over urban-type roughness: deductions
from wind-tunnel measurements", Boundary-Layer
Meteorology, vol 115, pp. 109-131
Finnigan, J. (2000). "Turbulence in plant canopies",
Annual Review of Fluid Mechanics, vol. 32, pp. 519
- 571.
Jimenez, J. (2004) "Turbulent flows over rough walls",
Annual Review of Fluid Mechanics, vol. 36, pp. 173
- 196.
Raffel, M., Willert, C. E. and Kompenhans, J. (2000).
"Particle Image Velocimetry", Springer.
Zhu, W., van Hout R. and Katz J. (2007) "On the flow
structure and turbulence during sweep and ejection
events in a wind tunnel model canopy", Boundary
Layer Meteorology, published online.
200
References
[1] Pavageau M., Schatzmann M. (1999). Wind tunnel
measurements of concentration Fluctuations in an
urban street canyon, Atmospheric Environment,
33, 3961-3971.
[2] Fluent 6.3 Users Guide, (2006).
[3] Daly B. and Harlow F., (1970) Transport Equations
in Turbulence. The Physics of Fluids. Vol. 13,
Number 11.
[4] Kim J., Moin P., (1989) Transport of passive scalar
in a turbulent channel flows. Turbulent Shear
Flows, Vol.6 Springer, Berlin 85-96.
[5] Abe K., Suga K., (2001) Towards the development
of Reynolds-averaged algebraic turbulent scalarflux model. International Journal of Heat and Fluid
Flow, 22 19-29.
0.06
0.06
0.02
0
-0.06
Leeward
side
-0.04
0.04
Windward
side
-0.02
0.02
0.04
0.06
0.02
(b)
1
Leeward
side
0.8
0
-0.06
-0.04
Windward
side
-0.02
0.02
0.04
y/H
0.04
(a)
1.2
0.1
0.08
0.1
0.08
0.6
0.4
0.06
0.2
0
0
(c)
Exp
S-A
50
SKE
100
C*
KERE
RNG
150
KW
200
SST
LRR-IP
Streamlines details for (a) Realizable k-H and (b) standard k-. (c) Non-dimensional concentration C*=CHL/Qs in the
centre line of the canyon.
201
Introduction
That tall buildings are capable of dispersing roof level
emissions to the ground and ground level emissions to
roof height has been long known; Figure 1 illustrates such
behaviour for a building of rectangular plan and height 3.6
the characteristic cross-section side.
Clearly, the
assumption of well-mixed near-wake conditions isnt
satisfied here and a more complex balance prevails
between vertical dispersion within the near-wake and
local loss from it into the main wake, leading to a nonuniform vertical concentration distribution.
The ADMS building effects model (Robins et al, 2000;
2001) was developed from extensive wind tunnel
investigations of flow and dispersion around cubes and
cuboids, generally with either L/HB > 1 or W/HB > 1, where
L is the building length, W the width and HB the height.
(The same is broadly true of AERMOD-PRIME). ADMS
dzSL
w
dz z z (x) | |
=
= R B SL
dx
UH
dx zB zR (x) 4
(1)
1.00
Height,
z/HB
0.75
Roof level
emissions
Roof source, Zs/Hb = 1; 0
0.50
Zs/Hb = 1; 34
Zs/Hb = 1; 90
Ground source, Zs = 0; 0
0.25
Zs = 0; 34
Ground level
emissions
0.00
0
10
Concentration,
CUHHB2/Q
15
20
Figure 1. Vertical profiles of non-dimensional concentrations on the centre-line, 0.1HB downstream of the rear face of a
tall building for a ground and roof level source; building dimensions: HB/3, HB/4.4, HB.
203
zB zP (x S ) zB zR (x S )
(2)
(3)
2 Results
The effect of a tall building in increasing ground level
concentrations downstream is demonstrated by Figure 2
for the HB/L = 3 building at orientation 45. The figure
shows data for four source heights from zS/HB = 1.02 to
1.5 compared with equivalent results in the undisturbed
flow. The recirculation region of this building reaches
about x = 2L (x/HB = 0.67) downstream and high
concentrations can be seen there for the two lower source
heights, zS/HB = 1.02 and 1.07; indeed the maximum
ground level concentration lies in the recirculation region
in the former case. The maximum ground level
concentration for the source at height 1.02HB is
approximately the same as would be produced by a
source at height 0.5HB in the undisturbed flow.
The figure once again shows that material released
above roof level can be brought rapidly to the surface in
the near-wake, but to a degree that falls rapidly with
increasing source height. The processes at work here are
downward plume deflection in the roof vortex system and
entrainment into the recirculating flow. Downward
deflection continues downstream because the wake
structure enhances downward dispersion, and wake
decay and the persistence of the roof vortices both create
down-flow. The net result is that ground level
concentrations are significantly raised above the values
that would exist in the absence of the building for all the
source heights considered.
The high spatial resolution of the laser anemometry
data has allowed mean streamlines to be calculated, as
illustrated in Figure 3, where the recirculation region and
above-roof streamline deflections are visualised.
The roof vortex system is clearly observed in the LDA
data, as Figure 4a illustrates. This shows secondary flow
vectors in the y-z plane at x/L = 1 from the centre of the
HB/L = 3 building at orientation 45. A well organised
secondary flow is clear, with downward flow above the
roof centre and upward flow in the near-wake
recirculation. Further downstream, at x/L = 2 (Figure 4b),
the secondary flow is basically inflow, with a substantial
(4)
dzSL
w
dz z' z (x) | |
=
= R B SL
dx
UH
dx z'B zR (x) 4
(5a)
dzSL
w
=
=0
dx
UH
(5b)
z'B = H B + (zB H B )
D = 1m
z0 = 1.5mm
u*/Uref = 0.055
Uref = 2.5ms-1
H (L + W ) /2
B
= min1,
HB
dx
dx dx
(6)
1 Methodology
Experiments were and are being carried out in the
20x3.5x1.5m EnFlo wind tunnel at the University of
Surrey. The characteristics of the simulated neutral
204
Concentration
1.00
CUH2/Q
0.75
0.50
0.25
0.00
0.0
5.0
10.0x/H
Fetch
15.0
20.0
400
400
300
300
Height, mm
Height, mm
200
100
100
0
-200
200
-100
0
100
Lateral position, mm
0
-200
200
-100
0
100
Lateral position, mm
200
Figure 4. Secondary flow velocity vectors in the wake of a building of height H = 3L = 300mm at orientation 45. A vector
length of 100mm is equivalent to 1ms-1. 4(a) left, at x = 100mm; 4(b) right, at x = 200mm from centre of building
500
Height, mm
400
1000
100
10
300
200
100
0
-100
100
300
500
700
Distance downstream, mm
900
1100
1300
Figure 5. Velocity vectors and non-dimensional concentration contours downstream of a building of height H = 3L =
300mm at orientation 45. A vector length of 100mm is equivalent to 1ms-1. The source height is 320mm.
205
3 Discussion
The results illustrate how combined velocity and
concentration field measurements reveal the complexity
of wake and dispersion processes and the role of specific
effects such as streamline deflection. They show that the
approach adopted in ADMS for treating dispersion near
tall, narrow buildings is not unreasonable, at least for
mean streamline deflections. Further experimental work
will address a wider range of geometries so that the
details of the model revisions can be made more assured.
Attention will also be focussed on behaviour in the nearwake.
It is quite straightforward to write a mass balance
model for the near-wake in which vertical transport within
the wake is included; e.g. the horizontally averaged
concentration, CA(z), at any level z is given by:
wA A
dCA
dwc
+A
+ sun c = 0
dz
dz
(7)
wc =
References
dCA
dx
(8)
d 2CA
= sun c
dx 2
(9)
A = LRW B ; s = 2LR + W B
(10)
un c = U H CA ; = U H W B
(11)
206
2 Experimental setup
The experiment to acquire the data used for the
analyses was conducted at the Environmental
Windtunnel Laboratory (EWTL) of the Meteorological
Institute at the University of Hamburg in the large
boundary layer wind tunnel WOTAN. The
measurements were carried out using a detailed model
of the Central Business District of Oklahoma City, USA.
The model had a scale of 1:300 and the modeled area
was about 1.6 km x 1.6 km.
During the experiment several hundred systematic
flow and dispersion experiments were conducted, for all
measurements statistically representative long-term
time series were recorded and later analyzed for
different length of averaging intervals. During the whole
experiment the wind direction was kept constant (
180). For the dispersion measurements the tracer g as
used was Ethan ( C 2 H 6 ).
1 Introduction
The understanding and the prediction of the
characteristic flow and dispersion processes within
complex urban areas is a critical issue, as around the
world more and more people tend to live within urban
areas. Especially in the case of an emergency, for
example the release of toxic gases, a high number of
people is influenced and without a well functioning
warning and forecasting system the casualties are
expected to be high. Therefore models to simulate and
forecast such scenarios have been developed. To test
the liability and accuracy of such models they are
validated using data from numerous field campaigns
that are often very time-consuming and expensive. The
field data is expected to represent the reality and the
results from field measurements are commonly used as
a reference for numerical modeling or physical
modeling in boundary layer wind tunnels.
In the context of using field data for model validation
and further analysis little attention is paid to the
temporal representativeness of the measurements. In
general the results are averaged over time intervals of
15 or 30 minutes. It is questionable however whether
such short time intervals are sufficiently long to assure
that the average is representative.
The fact, that the flow and turbulence structure plays
an important role in determining the temporal
representativeness, is also often neglected. At
measurement locations with a uniform flow structure
averaging over a shorter interval may result in the same
statistical significance as the same parameter averaged
over a longer interval at a measurement location with
207
The statistical analysis was performed for the uand v-component of the wind, the turbulent fluxes u 'v'
and for the dimensional concentration c*.
4 Results
3 Data analysis
c* =
c u ref L2ref
Q
u ref
Lref
1
a = ai
n
i = 1,,n
1
ai a
n 1
skewness
1 n
ai a
n i =1
Sk =
s3
1
ai a
n
Fl = i =1 4
s
n
-0.2
and flatness
7.5 min
15 min
30 min
60 min
Long-term average
0.2
v [m/s]
standard deviation
0.4
-0.4
0
1000
2000
dt [s]
3000
4000
Vr = a max a min
208
0.4
0.6
0.4
0
-3
-2
-1
v [m/s]
0.4
0.2
0.6
0.4
v [m/s]
-2
-1
v [m/s]
-2
-1
800
t [min]
1000
1200
-3
1400
200
400
600
800
t [min]
1000
v [m/s]
v [m/s]
0.95
0.9
0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
-1
-2
-1
-2
200
400
600
800
t [min]
1000
1200
-1
1400
-3
v [m/s]
200
400
600
800
t [min]
1000
1200
0.4
0
-2
-1
v [m/s]
15 min
1
200
400
600
800
1000
t [min]
1200
-2
1400
200
400
600
800
1000
t [min]
1400
frequency [-]
60 min
0.95
0.9
0.85
0.8
0.75
0.7
0.65
0.6
0.55
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
-1
Figure 3: Short-term probability distribution for 30minute averages plotted against time for a normal
distribution (same as Figure 2)
1200
30 min
-1
-2
1400
-1
frequency [-]
60 min
0.6
-2
1400
30 min
-3
1200
v [m/s]
600
0.2
-1
-2
400
0.4
v [m/s]
v [m/s]
v [m/s]
-1
3600s
7.5 min
200
0.6
15 min
v [m/s]
7.5 min
-1
0
-2
0.8
0
-2
-3
v [m/s]
0.2
0
-3
0.4
1800s
0.2
-1
0.8
frequency [-]
frequency [-]
0.6
-2
3600s
0.8
0
-3
-1
1800s
0.8
0.6
0.2
0
-2
frequency [-]
v [m/s]
v [m/s]
0.4
v [m/s]
-1
0.6
200
400
600
800
1000
t [min]
1200
-2
1400
200
400
600
800
1000
t [min]
1200
1400
Figure 5: Short-term probability distribution for 30minute averages plotted against time for a bimodal
distribution (same as Figure 4)
360-390 Min
frequency [-]
0.8
0.6
0.4
0.2
0
-2
v [m/s]
frequency [-]
0.6
0.4
0.2
-1
0
-2
-1
v [m/s]
420-450 Min
0.8
450-480 Min
0.8
0.6
0.4
0.2
0
-2
390-420 Min
0.8
frequency [-]
-2
900s
0.8
0.2
0.2
0
-3
450s
0.8
frequency [-]
frequency [-]
frequency [-]
0.6
0.2
frequency [-]
900s
0.8
frequency [-]
450s
frequency [-]
0.8
0.6
0.4
0.2
-1
v [m/s]
0
-2
-1
v [m/s]
Figure 6: Probability distributions for succeeding 30minute averaging intervals for the v-component of the
wind in a street canyon
209
5 Conclusions
The study showed that the issue of temporal
representativeness of measurements within complex
urban areas should be treated with more care, because
210
211
Introduction
As a result of industrialisation, a large percentage
of the earths population lives in densely populated
cities. In a global view this process of urbanisation as
well as the industrialisation itself are still in progress.
As a result, the increasing likelihood of an accidental
or deliberate release of hazardous material in an
urban area is posing a threat to more and more
people. Consequently, reliable simulations of
accidental releases and the dispersion of puffs in
complex geometries attained a much higher
importance during the last years. A credible and
validated Computational Fluid Dynamics (CFD) based
plume prediction model of Contaminant Transport
(CT) in urban areas can be used in the licensing of
new industrial plants, in safety analysis studies for
accidental releases of hazardous materials in the
chemical industry, or in the context of a proper crisis
management after terrorist attacks in urban
environments.
The numerical simulations of cloud dispersion in
urban envirionments are very complex and they
require a comprehensive computational effort. In the
past it was common practice to use simpler models
as, e.g., analytical (Gaussian) models, diagnostic
models, based the mass conservation equation only
or CFD-models with full parameterization of
turbulence (RANS-models) for these complex tasks.
Such models are able to compute urban dispersion
within a reasonable time, but on the other hand, they
213
2 Experimental setup
214
3 Puff measurements
cef[ppm]
400
200
0
1414
1416
tt_abs[s]
1418
1420
pc
60
cef[ppm]
40
at
: arrival time
pt
: peak time
lt
: leaving time
pc : peak concentration
do : dosage
asct : ascent time
20
do
threshold
0
asct
0
at
dsct
4
pt
tt_puff[s]
lt
215
F ra m e 0 0 1 1 9 M a y 2 0 0 7
100
b)
80
c[ppm]
60
F ra m e 0 0 1 1 9 M a y 2 0 0 7
40
20
200
0
a)
5 5 7 .0 2
150
5 5 7 .0 4
5 5 7 .0 6
tt[s ]
5 5 7 .0 8
5 5 7 .1
5 5 7 .1 2
c[ppm]
F ra m e 0 0 1 1 9 M a y 2 0 0 7
100
c)
60
cef[ppm]
50
40
20
0
555
556
557
tt[s ]
558
559
0
5 5 6 .9 8
557
5 5 7 .0 2
5 5 7 .0 4
5 5 7 .0 6
tt[s ]
5 5 7 .0 8
5 5 7 .1
5 5 7 .1 2
Figure 4. Principle of puff removing: a) unfiltered puff signal with spike; b) spike in detail, showing a damped
harmonic oscillation; c) same section as in b) after spike removal and subtraction of background concentration
216
Frame 00 1 19 M ay 20 07
200
150
150
c[ppm]
cef[ppm]
200
100
100
50
50
0
545
0
550
555
560
tt[s]
565
570
575
580
545
550
555
560
tt[s]
565
570
575
580
Figure 5. Subsection of timeseries; left: unfiltered, with spikes; right: after spike removal and subtraction of
background concentration
217
60
frequency of occurence
frequency of occurence
50
40
30
20
10
0
0
3
Figure 6. Frequency distributions of the leaving time; left: for 215 puffs; right: for 10 puffs
probabilistic approach to the prediction of possible
extreme values for example from a limited number of
field trails. Furthermore, the frequency distributions of
observed puff parameters are calculated for different
ensemble sizes. Figure 6 shows a comparison
between frequency distributions of the leaving time
for 10 and 215 puffs. The frequency distribution
based on 215 puffs shows an almost symmetric nearGaussian. Similar distributions of the same ensemble
size have been obtained for peak time, descent time
and duration of the puff, whereas peak concentration,
dosage, arrival time and ascent time mostly show a
significantly positive skewness. However, for all puff
parameters calculated the frequency distributions
based on ten puffs only do not give any indication for
the shape of a statistically safe ensemble of, for
example, 215 puffs. Consequently, a very limited
amount of puff samples cannot be used to predict
puff exposure in a probabilistic way or in other words,
a limited number of releases do not give a realistic
impression of what the threat at a certain location
would be, if a cloud of hazardous material passes by.
In order to obtain information about the representativeness of results calculated from a limited
number of individual releases the detected puff
signals are subdivided into ensembles of different
sizes. Since all puff signals collected within a test
series were recorded with exactly the same mean
boundary conditions present in the wind tunnel, the
data is qualified for puff ensemble analysis with
systematically changing ensemble sizes. Typically,
the size of the averaged puff ensembles was
increased in steps of 5 ranging from 5 puffs per group
up to 215 puffs. The smallest ensemble size analyzed
is in the order of the number of puff releases during
the Joint Urban 2003 field campaign repeated for
similar ambient conditions. Only three to six puff
releases could be carried out under more or less
similar boundary conditions in the field.
Every ensemble is examined with respect to the
minimum, maximum and mean value of every puff
parameter. The extreme values of puff parameters
observed for a certain ensemble were also
determined as they are of importance for a
218
Figure 7. Scatterplot on the mean value of the dosage for several ensemblesizes (every dot is assigned to one
ensembele, and all ensemblesizes are represented equal )
6 Conclusions
For the validation of micro-scale LES based C&T
models, qualified experimental datasets are re-quired.
For an adequate description of puff disper-sion, puffs
must be characterized by well defined parameters
such as arrival time, leaving time, peak time, peak
concentration, puff duration or dosage, for example.
For statistical analysis and description of puff
dispersion phenomena, a large number of puff
releases under similar environmental is required.
Data qualified for a safe statistical and probabilistic
219
References
Britter, R.E., Disabatino S., Caton, F., Cooke, K.,
Simmonds, P., Nickless, G., (2002). Results
from three field tracer experiments at the
neighborhood scalein the city of Birmingham,
220
Ri*
g ' h / u *2 ,
g ' g (U p U ) / U ,
Introduction
It is well known that the thermal stratification as well as
density of pollutants influences turbulence properties of
the ABL, i.e. dispersion characteristics. Dispersion of
different pollutants, including the heavy gas dispersion
problems, came forward in eighties of 20th century with
the expansion of chemical industries. Since many new
instrument equipments and methods were developed from
those times, it is highly interesting to return to this topic
and conducted systematic study. The study of heavy
gases dispersion was firstly arouse on a basis of chemical
factory accidents, where very dangerous substances like
chlorine or ammonium were released. Nowadays this
topic is becoming actual again with regard to terrorist
attack with aim to harm citizens by chemical or biological
weapons (e.g. the Sarine accident in Tokyo underground).
Dispersion of toxic pollutants emitted accidentally or
intentionally from a ground level in an urban area under
various atmospheric thermal stratifications is scientifically
important. Obtained data will be used to design a risk
assessment guideline that considers thermal stability
effects and physical and chemical properties of released
gases.
The wind tunnel modelling offers the possibilities of
conducting experiment with high temporal and spatial
resolution. Well-equipped wind-tunnel laboratory of the
Wind Engineering Center of Tokyo Polytechnic University
has several wind tunnels including one allowing modelling
a thermal stratification of ABL. There is also available a
new method for simultaneous measurement of velocity,
concentration and temperature in one place using hot wire
and hot film probes together with fast flame ionization
detector. This set-up allows deriving turbulent fluxes
related to heat, momentum, and concentration, which are
of high interest for numerical modellers using advanced
Computational Fluid Dynamics (CFD) models, i.e. Large
Eddy Simulation (LES).
Ri
g' L / U 2 ,
g' g(U p U ) / U,
Ri 0
g 0 ' Q / D S u *3 ,
g 0 ' g (U 0 U ) / U ,
221
(3)
where n is the toxic load exponent equal to r/s from Eq.2
and the toxic load L is defined as L = cnt. In terms of the
toxic load L, Eq.3 can be rewritten as:
(4)
P q s ln L ,
To determine the proportion of a population
responding to a release, the toxic load L is calculated and
then Eq.5 is used to determine the probit value Pr. The
percentage fatalities is obtained from the Pr value and
Eq.1.
In animal experiments, the exposure is in controlled
conditions at a constant concentration for a set period of
time. In this case, there are no fluctuations in
concentration
and
the
instantaneous
exposure
concentration c is constant with time over the entire
exposure duration te. With instantaneous concentration c
equal to the mean concentration the toxic load Lmean is
Lmean = Cnmean te. This is the original definition of toxic
load. Note that Lmean is not the mean toxic mean load, but
rather is a representative toxic load based on the mean
concentration.
The toxic load of a fluctuating exposure concentration
could also be calculated with the mean concentration and
the exposure duration te. If n = 1, the effect of e
concentration is linear and this is a reasonable approach,
but it still does not take into account any uptake, recovery
or saturation processes. For most chemicals, where n > 1,
the mean concentration toxic load Lmean misses the
important non-linear effects of the mean concentration
fluctuations as well as any limitations on receptor
response.
In the risk assessment literature, the definition of toxic
load has been extended, without any toxicological
justification, to include time varying exposure
concentrations, see Ride (1984) and ten Berge et al.
(1986)
3 Probit method
The probit method is a way of linearizing a cumulative
normal distribution of population response to some toxic
exposure variable. One probit unit, P, is equal to one
standard deviation of the normal distribution. The median
or 50th percentile response was defined arbitrarily as P =
5.0. A probit value of P = 4.0 is one standard deviation
below the median at a cumulative probability of 16%. That
is, it is expected that 16% of the population responds to a
toxic load that produces a probit value of 4.0. Similarly,
84% of the population would be expected to respond to a
toxic load that produces a probit of P = 6.0, one standard
deviation above the mean. The fraction F of the
population responding to a toxic exposure can be
calculated from the probit value P using the following
relationship
1 P 5
efr
1,
2 2
te
c n dt ,
(5)
(1)
erf ( x)
q s ln c n t ,
e t dt .
222
4 Application
The wind tunnel modelling offer an opportunity to
measure concentration time series with high spatial and
temporal resolution. This concentration time series can be
used for better estimating of the personal risk
assessment.
Examples of concentration time series are shown in
Fig.1 and Fig.2, respectively. The time series in Fig.1 are
example of intermittent plume behavior. On the other
hands the time series in Fig.2 are concentration time
series taken in the middle of thr plume, where the passive
tracer is relatively well mixed within the plume. These time
series were measured in the wind tunnel within the
idealized urban roughness (for more details see
Bezpalcova, 2007). The wind-tunnel based dimensionless
data were adapted to the full scale situation under these
conditions:
x
x
x
time series 1
[s]
0.5
25
250
1000
2
1
0
0
41
38
33
30
time series 2
5 Conclusions
The risk assessment issue is quite new discipline and
it is based on predictions. In the case of sudden or
delivery release of toxic gas the prediction of dispersion is
needed. However, it is common practice to use simple
parametric models, which often failed with the urban or
industrial areas. Therefore the wind tunnel modelling
should be consider as a tool for obtaining more trustful
prediction of actual concentration. Most of the toxic gases
are heavier than air, therefore more advanced simulation
technique is required.
averaging
time
q = -15.6;
s = 1;
n = 2;
Acknowledgments
time series 1
time series 2
[s]
dose
dose
0.5
25
250
1000
30 000 000
23 000 000
15 000 000
11 000 000
References
Bezpalcova, K. Physical Modelling of Flow and Diffusion
in Urban Canopy, Ph.D. thesis, Charles University in
Prague, 2007.
Briggs, G. A., Britter, R. E., Hanna, S. R., Havens, J. A.,
Robins, A. G., and Snyder, W. H., (2001). Dense
gas vertical diffusion over rough surfaces: results of
wind-tunnel studies, Atmospheric Environment 35,
pp. 2265-2284.
Center for Chemical Process Safety (CCPS), Guidelines
for Chemical Process Quantitative Risk Analysis,
Wiley-AIChE, 1999.
Committee for the Prevention of Disasters (CPR),
Guidelines for quantitative risk assessment Purple
Book CPR18E, SDU, The Hague, 2005.
EPA, (1986). Emergency Planning and Community Right
to Know Program, 40 CFR 300. Washington, DC:
U.S. Environmental Protection Agency.
Finney, D.J., Probit Analysis, 3rd edition, Cambridge
University Press, 1971.
time series 1
time series 2
[s]
probit
probit
0.5
25
250
1000
1.61
1.37
0.91
0.61
3.4
3.31
3.17
3.08
223
Hall, D.J., Walker, S., (1997). Scaling rules for reducedscale field releases of hydrogen fluoride, Journal of
Hazardous Materials 54, pp. 89-111.
Hilderman, T.L., Hrudey, S.E., and Wilson, D.J., (1999).
A model for effective toxic load from fluctuating gas
concentration, Journal of Hazardous Materials A 64,
pp. 115-134.
Ride, D. J., (1984). An assessment of the effects of
fluctuations on the severity of poisoning by toxic
vapours, Journal of Hazardous Materials 9, pp. 235240.
Robins, A., Castro, I., Hayden, P., Steggel, N., Contini, D.,
and Heist, D., (2001). A wind tunnel study of dense
2000
c [mg/m3]
1500
1000
500
0
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
t [s]
Figure 1: An example of intermittent concentration time series. Different lines show diffrent averaging time of
measurements.
2500
c [mg/m3]
2000
1500
1000
500
0
2000
2100
2200
2300
2400
2500
2600
2700
2800
2900
3000
t [s]
Figure 2: An example of well mixed pasive tracer in the middle of the plume. Different lines show different
averaging time of measurements.
224
1 Introduction
In the surrounding of livestock buildings odours can
cause annoyances and ammonia can lead to damages
on plants and ecosystem. There exist German guidelines and standards for example minimum distances
between animal houses and residential area. To control
the environmental laws, measurements in and around
stables are necessary.
The ATB possess a large equipment of climate and
emissions technology. These are used for measure
dates of ammonia, odours and climate parameters. So it
is possible to get information about the high of disturbance and the over or under usage of critical value.
Furthermore in the boundary layer wind tunnel of the
institute there are realise systematically measurements
regarding to the dispersion of gases. The physical surrounding of the investigation area can be modelled and
analysis of emission can be carried out. The airflow can
be descripted by quantitative and qualitative measurement. One of the qualitative investigations represents
the laser-light-sheet technique, which is used to illuminate the flow in the working section of the wind tunnel. It
is possible to explain complex questions of the flow and
the dispersion of substances in a clear and visually way.
225
3 Results
In the surrounding of a big livestock building for pigs,
disturbances appear because of odour (figure 2.). The
task was to reduce these negative effects by structural
arrangements as well as inside and outside. In this
article only the possible outdoor changes are mentioned. For further details towards different variances of
the ventilation system see Krause, 2006.
wind
wind
Figure 4. Average picture of a stable with 3 stacks and a wall on the Lee site
226
160
1200 mm (360 m)
800 mm (240 m)
400 mm (120 m)
140
high [mm]
120
wind velocity in a high of 150 mm: 0.95 m/s
exit velocity: 0.47 m/s
100
80
60
40
20
0
0
500
1000
1500
2000
concentration [ppm]
2500
800 mm (240 m)
400 mm (120 m)
120
wind velocity in a high of 150 mm: 0.95 m/s
exit velocity: 0.47 m/s
high [mm]
100
80
60
40
20
0
0
200
400
600
800
1000
1200
concentration [ppm]
Figure 6. Gas measurement of a stable with 3 stacks and a wall on the Lee site
grams in figure 5 and 6 are results from model measurements in the scale of 1:300. But the velocity of the air
reaches 0.95 m/s in a high of 150 mm. As an outputsignal we used a mixture of an Ethan / Air volume
stream with a ratio of 1:10. The diagrams 5 and 6 show
the relations between high (mm) and concentration
(ppm) in different distances from the emission source
(1200 mm, 800 mm, 400 mm). Thereby the blue graph
is in the distance of 1200 mm, the red one is located in
800 mm and the green graph stands for the distance of
400 mm from the font. These curves represent 360 m,
240 m and 120 m in nature at a scale of 1:300. Figure 5
shows the dispersion situation of a stable without an
obstacle. The high is equivalent to our studies in the
wind tunnel and so the y-coordinate is determined in the
units of mm.
In a scale of 1:300 the maximum of the high reaches
50 m in nature.
Contrary to figure 5 represents image 6 the influence
of a wall on the Lee site as a disturbance of the dispersion process. As a commonality of both situations the
concentration is sinking with the high and the distance
227
to the emission font. But with a wall on the Lee site the
distinction of concentration is nearly the half. Without
the wall the concentration in 3 m high detracts from
1800 ppm in a distance of 120 m to 1100 ppm in a
space of 360 m. This responds 38 % minimization. In
comparison to the situation with a wall the reduction
reaches 80 % in 360 m distance toward the value in 120
m distance (figure 7).
Within the high from 3 to 30 m in space of 120 m to
the emission font in case of an obstacle, the rising of
reduce of concentration is the highest. In the distance of
240 and 360 m the decrease is much more less.
4 Conclusions
In the wind tunnel it is possible to create the surrounding of a special situation. Thereby complex queries toward the attitude of flow and dispersion of substances can be explained in an easy manner. It is simple to change the structural surrounding of an area and
find out an optimal set-up for a special problem. The
dispersion of emissions can research in a qualitative
120
1200 mm (360 m)
110
800 mm (240 m)
100
400 mm (120 m)
90
high [mm]
80
wind velocity in a high of 150 mm: 0.95 m/s
exit velocity: 0.47 m/s
70
60
50
40
30
20
10
0
30
40
50
60
70
80
90
100
reduction [%]
Figure 7. Comparison of concentration reduction [%] without and under the influence
way by the laser-light-sheet technique and through a
quantitative method by gas measurements. Both investigations are complementing each other. Obstacles at
the Lee site (Figure 4) visualise a better deflection of the
plumes in higher air levels and a better admixing with
fresh and waste air. The combined measurements (Figure 6) confirm this receiving by a high decrease of the
concentration mainly in the close-up range of 120 m
transferred to nature. Because of the similarity of the
wind profile inside the tunnel with a naturally one in a
suburban area a portability to practise gets possible. So
an obstacle like a wall can improving the ambient air of
a livestock building and brings more comfort to the environment and the people.
228
References
Krause, K.-H. , Linke, S., Mulick, M., Mller, H.-J.,
Weihs, C. 2006. A New Ventilation System for Reduction of Odour and Ammonia Emissions from Pig
Houses. In: VDI.-Berichte Nr.1958: World Congress
Agricultural Engineering for a Better World, VDI
Verlag GmbH, Dsseldorf, pp. 447 448.
Blenk, H., Trienes, H. 1956. Strmungstechnische Beitrge zum Windschutz./An article to the topic windbreak in fluid mechanics. Grundlagen der Landtechnik. Heft 8. VDI-Verlag GmbH, Dsseldorf.
VDI 3783, Part 12. 2000. Environmental meteorology.
Physical modelling of flow and dispersion processes in the atmospheric boundary layer. Application of wind tunnels. VDI-Verlag GmbH, Dsseldorf.
References
Burton, T., D. Sharpe, N. Jenkins, and E. Bossanyi,
2002: Wind Energy Handbook. John Wiley &
Sons, Ltd.
Vermeer, L., J. Sorensen, and A. Crespo, 2003:
Wind turbine wake aerodynamics. Progress in
Aerospace Sciences, 39, 467510.
Figure 1: Model set-up for the 5MW turbine array (topview). Flow from right
to left, shaded area indicates the probe traversing range.
229
Mohamed F. Yassin1,3
1
Institute of Termomechanics, Czech Academy Science, Prague, Czech Republic
Department of Meteorology and Environment Protection, Charles University, Prague, Czech Republic
3
Faculty of Engineering, Assiut University, Assiut, Egypt
radka.kellnerova@email.cz
Zbyn
ek Ja
nour1
Introduction
At rst, it had to be veried that internal roughness layer is fully developed above measuring position. We checked the sucient distance from the
model front that is the required fetch for ow inside
the roughness layer to be in equilibrium with the surface beneath.
Under neutrally stratied conditions, vertical proles of turbulent characteristics were measured in the
central and the symmetrically upstream and downstream intersection (see Figure 2).
Strong similarity of the proles suggests that inFlow measurements were carried out using twodimensional bre-optic laser Doppler anemometry ternal boundary layer above these positions has very
(LDA). In order to measure air-particle velocity it close turbulent characteristics.
Requirements for equilibrium of canopy sublayer,
was necessary to add the glycerine droplets in the
main ow , because the droplets can be traced by e.g. layer below the roof level, are weaker. Hence
The experiment was conducted in low-speed
open-circuit aerodynamic tunnel IT AS CR in Nov
y
Knn. Tunnel had 20.5m long generating section
with roughness elements and spires. With the air
owing over the section, fully turbulent boundary
layer can be developed. On the beginning of working
section the thickness of turbulent layer attained
more than 50 cm.
231
Figure 2: Vertical velocity proles in the intersections. The positions of prole are shown in the
sketch. U0 means the reference velocity.
2
2.1
Results
Velocity eld
The ow inside the canopy is strongly threedimensional. Velocity in canyons has both horizontal
and vertical components and spiral vortex is formed.
An almost symmetrical velocity eld with corner
vortexes and slow moving contra-ow in the street
canyon was found (Figure 6 - up). Within an intersection, area velocity drops due to divergence. Figure
6 - down shows symmetrical area of inverse values of
Reynolds stress < u v >. High values of horizontal
Reynolds stress occurs close to corner areas. Further
in the canyon, the sign of Reynolds stress reverses.
Figure 5:
3
233
2.2
Concentration eld
We have found the critical Reynolds building number ReB = uH H/, where is kinematic viscosity,
and veried the requirements for Towsend hypothesis. The experiment was done with Reynolds building
number ReB =15 200 that lies on the lower edge of
interval for valid Towsend hypothesis. Relevant free
stream velocity was 4 m.s1 . The vehicle trac was
simulated by continual double line source situated
on the oor, therefore the height of the source equals
zero. The dimensionless concentration was obtained
from formula 1 :
K =
CuH HLQ
.
Q
(1)
Conclusion
Acknowledgments
This project was supported by Institutional Research
Plan AVOZ20760514 and COST 732.
References
[1] Cheng H., Castro I.P.: (2002) Near Wall Flow
Over Urban-Like Roughness. Boundary-Layer
Meteorology 104: 229259.
[2] Britter R.E., Hanna S.R.: (2003) Flow and Dispersion in Urban Areas. Annu. Rev. Fluid Mech.
35:46996
[3] Rafailidis S.: (1997) Inuence of Building Areal
Density and Roof Shape on the Wind Characteristics Above a Town. Boundary-Layer Meteorology 85, 255-271.
The wind tunnel experiments described in this paper have provided signicant data sets about pollutant diusion in an urban intersection. The ow
inside the canopy has strong three-dimensionality, [4] Cheng H., Castro I. P.: (2002) Near-Wall Development After a Step Change in Surface Rough1 C means measured concentration, u
H means velocity at
ness. Boundar- Layer Meteorology 105, , 411the height of buildings H, LQ is length of the line source, Q is
432.
source ux.
5
235
1 Introduction
The major activity of the Fluid dynamics team
from the LME is focused on experimental aerodynamics.
Fine analysis of separated flows, flow control in order to
reduce drag on vehicles, aerodynamics of rotating
systems (helicopter rotors and horizontal-axis wind
turbines) are its priority issues. After several
contributions to the wind turbine aerodynamics and
regarding the needs expressed by industrial partners,
the LME focused on the wind turbine wake, and
particularly on the far-wake interactions with other wind
turbines in order to minimize it. Indeed, the wind
turbines arrangement in a wind farm should be well
predicted before its implantation. It is therefore of great
interest to correctly assess these interactions. Since the
field measurements are rare and difficult to interpret,
wind turbine wakes and their interactions are usually
treated with numerical models. Though numerical wake
models are still very simple and they cannot take
correctly into account complex terrain configurations. An
alternative is therefore the physical modelling in wind
tunnels. This issue motivates the LME to modify the
return test section of its wind tunnel to model
atmospheric boundary layers at a scale of around 1:400.
The presentation will focus on the validation of the
obtained boundary layers using standard litterature as
VDI (2000), ESDU (1985), Snyder (1981). The first
application of this set-up is the study of wind turbine
wake and that will be presented in the paper Properties
of the far wake of a wind turbine in an atmospheric
boundary layer, G. Espaa, S. Aubrun, P. Devinant, L.
Laporte-daube, E. Dupont.
2 Experimental set-up
The Lucien Malavard wind tunnel is a closed-circuit
wind tunnel (Fig. 1). It has two test sections V1 and V2.
The first one has a square section of 2m times 2m and
is 5m long. The maximum velocity is 60 m/s and the
turbulence intensity is lower than 0.4%. The second test
section V2 is located in the return circuit of the wind
tunnel. It has a square section of 5m times 5m and is
20m long. The maximum velocity is 10 m/s. The input
section can be reduced up to 3m times 3m with a
second convergent and a turbulence grid can be added
237
4 Mean profiles
To determine the properties of the modelled
boundary layer, velocity profiles were measured at 13m
from the entrance of the wind tunnel. Fig. 5 presents the
mean streamwise velocity in a linear and in a semilogarithmic scale, at a geometric scale of 1:400. Since it
was the first attempt to model an ABL in this facility, a lot
of care was taken to properly characterise the modelled
boundary layer .Consequently, the velocity profiles were
measured up to 360m F.S. from the ground.
These data enables us to determine the roughness
length and the power law exponent of the boundary
layer. Fig. 5 shows that the wind profile fits to a
logarithmic function from 12m d z d 100m full scale.
The extrapolation of the logarithmic law towards the
zero-value of velocity gives the associated roughness
Figure 3 : Transformations in the Malavard return
section
wp 1
2
G U uG d 0.05
wx 2
Where
wp
is the longitudinal pressure gradient, G is
wx
uG
4
3
zd
100
160
200
260
360
0.2
0.22
0.23
0.25
0.27
2
1
0
12
z0
length,
of a rough terrain.
13
14
15
16
238
400
10
a)
400
b)
350
c)
350
102
300
300
250
Z [m]
250
101
200
200
150
150
10
100
100
Z0 = 0.3m
50
0
0
U mean [m/s]
10
-1
50
U mean [m/s]
0
0
0.5
-u'w'/u2*
1.5
shear stresses - u cwc non-dimensioned with the friction velocity u * . Measured in wind tunnel at 13m (model scale)
from the entrance of the wind tunnel
Z [m]
400
400
a)
400
b)
350
350
350
300
300
300
250
250
250
200
200
200
150
150
150
100
100
100
50
50
50
0
0
10
20
Iu [%]
30
40
0
0
10
20
Iv [%]
30
40
0
0
c)
10
20
Iw [%]
30
40
Figure 6 : a) Longitudinal turbulence intensity profile. b) Lateral turbulence intensity profile. c) Vertical turbulence
intensity profile. Lines delimit the ranges authorized by VDI 3783/12 (2000) for the rough class. Measured in wind
tunnel at 13m (model scale) from the entrance of the wind tunnel
239
f Svv f
u*2
f S ww f
u*2
1 33n
17 n
1 9.5n
5/ 3
102
103
Lux [m]
10
-1
6 Conclusions
Sxx u f / u *
10
-2
10
-2
-1
10
10
f u z / Umean
10
10
Suu , Svv , S ww
Lux
Lux
Lux
Counihan (1975) :
Lu z C z0 z
with
z0
101
10 -3
10
of
40
20
2.1n
1 5.3n5/ 3
Wind tunnel
Counihan (1975)
60
5/3
of
80
102n
where
10
100
Z [m]
f Suu f
u*2
1 n z0
240
References
Counihan J. (1975). Adiabatic atmospheric boundary
layers: a review and analysis of data from the
period 1880-1972. Atmospheric Environment 9, pp.
871-905.
Kaimal, J.C., Finnigan, J. (1994). Atmospheric Boundary
Layer Flows. Oxford University Press.
Snyder, W.H. (1981). Guideline for fluid modelling of
atmospheric diffusion. US Environmental Protection
Agency. EPA-600/8-81-009, p. 185.
VDI-guideline 3783/12 (2000). Physical modelling of flow
and dispersion processes in the atmospheric
boundary layer Application of wind tunnels. Beuth
Verlag, Berlin
ESDU Engineering Sciences Data Unit, (1985).
Characteristics of atmospheric turbulence near the
ground. Part II: Single point data for strong winds
(neutral atmosphere). Item no 85020. ESDU
International, London.
Teunissen H.W. (1980) Structure of mean winds and
turbulence in the planetary boundaryr layer over
rural terrain. Boundary-Layer Meteorology 19, pp.
187-221.
241
Authors Index
Author
Pages
Author
Pages
Addepalli
79
Herbst
41
Aubrun
47, 237
Holtslag
175
Balcz
185
Izarra-Garcia
201
Balogh
93
Jaour
129, 231
Barlow
75, 113
Jonker
175
Bartoli
59
Kellnerov
231
Bezpalcov
221
Kristf
93
Borri
59
Kukadia
17
Borsani
59
Lajos
185
Bowker
63
Laporte-Daube
47
Brixey
63
Lawton
203
Builtjes
175
Leitl
Carter
105, 143
Lin
135
Cesari
157
Loredo-Souza
99
Contini
157
Makita
179
Czder
185
Mapurisa
87
De Paoli
99
Marsland
17
Denev
121
Miyagi
177
Devinant
47, 237
Moallem
241
Donateo
157
Mueller
225
Duan
149
Nagai
167
Dupont
47
Nakayama
165
Esfahanian
241
Ohba K.
179
Espaa
47, 237
Ohba M.
221
Fischer
213
Ohba R.
167
Franke
201
Ohta
165
Goricsn
185
Okabayashi
167
Gorle
53
Pardyjak
79
Gromke
121
Pascheke
113, 229
Gupta
111
Perret
197
Hall
17
Perry
63
Hancock
229
Petersen
105, 143
Hara
167
Planquart
53
Harms
27
Procino
59
Havens
15
Rcz
93
Hayashi
167
Rgert
185
Hayden
87
Repschies
33
Heist
63
Restorick
149
243
Author
Pages
Rix
207
Robins
Ruck
121
Saathoff
111
Sanz Rodrigo
53
Sassa
177
Savory
149
Schatzmann
Schettini
99
Schultz
69
Sedenkova
129
Sheppard
15
Shiau
7, 135
Spicer
15
Stathopoulos
111
Tamura
165
Tily
17
Van Beeck
53
Van Dop
175
Van Rooij
191
Vanweert
191
Von Bobrutzki
225
Walker
17
Wittwer
99
Wu
Yassin
231
Yoneda
167
244