You are on page 1of 26

2003 Firenze University Press 244

Proceedings of PHYSMOD2003:
International Workshop on Physical Modelling of Flow and Dispersion Phenomena
3-5 September 2003, Prato, Italy
PHYSICAL MODELLING OF DISPERSION OF POLLUTANT EMITTED AT THE PORTALS OF
ROAD TUNNELS


Daniele Contini
Istituto di Scienze dellAtmosfera e del Clima
Sezione di Lecce
Consiglio Nazionale delle Ricerche
Str. Prv. Lecce-Monteroni km 1,2
73100 Lecce, Italy
Lorenzo Procino
CRIACIV
Boundary-Layer Wind Tunnel
Piazza Ciardi 25
59100 Prato, Italy

Michela Massini
Dipartimento di Energetica Sergio Stecco
Universit di Firenze
Via Santa Marta 3
50139, Firenze, Italy
Giampaolo Manfrida
Dipartimento di Energetica Sergio Stecco
Universit di Firenze
Via Santa Marta 3
50139, Firenze, Italy


ABSTRACT
The exit portals of road tunnels can be considered to any
extent as sources of pollution placed at ground level. Gaseous
contaminants, potentially dangerous (like, for example,
nitrogen oxides, carbon monoxide and hydrocarbons) are
emitted by vehicles passing through road tunnels and they are
collectively emitted at the exit portals. After release into the
atmospheric boundary layer the contaminants will mix with
ambient air and dispersion will take place in the immediate
surrounding of the road tunnel itself generating zones with
relevant pollutant concentrations. This topic is important
especially because sometimes road tunnels are placed in urban
environment or in proximity of areas of particular ecological
interest. Contaminants emitted by the road tunnels, basically at
ground level, will disperse in a high turbulence area in a way
that is different from the typical plumes emitted by elevated
sources like stacks and chimneys.
In this paper some experimental results obtained in the
CRIACIV (Centro di Ricerca Interuniversitario di
Aerodinamica delle Costruzioni ed Ingegneria del Vento) wind
tunnel on a physical model of a road tunnel built under a hill
are presented. In particular it will be described the model
characteristics and performances and the ground level
concentration field for different wind conditions. The results
obtained put in evidence the main aspects of the diffusion
including recirculation of pollutant from one side of the road
tunnel into the other side and the transport above the hill of
tracer released against the wind.

NOMENCLATURE

Symbol Description
C Tracer concentration
U
ref
Reference mean flow speed above
the boundary-layer
U(z) Mean flow speed at height z
u
*
Friction velocity
W Exit speed of the emissions at the
portals of the model.
z
ref
Boundary-layer height
z
o
Roughness length

Exponent of the exponential law for
the velocity profile


INTRODUCTION
Nowadays pollution produced by exhaust of combustion
engines, used in both private and commercial vehicles, is one of
the major sources of air contaminants in towns as well as in
some extra-urban areas located near the major road networks.
As matter of fact a source of gaseous contaminants, potentially
dangerous (like, for example, nitrogen oxides, carbon
monoxide and hydrocarbon), is represented by the emissions at
the exit portals of road tunnels. After release the contaminants
will mix with ambient air and dispersion will take place in the
immediate surroundings of the road tunnel itself generating
zones with relevant pollution concentrations. Even if the time
of stay inside the road tunnel is small it is important to take into
account the distribution of pollutant nearby the exit portals of
the tunnel itself where inhabited areas can be located or areas of
particular environmental interest. The details of the plume
dispersal are strongly related with the road tunnel ventilation
characteristics and to the local climate and are therefore
dominated by the interaction among the jet at the portal and the
ambient wind [Oettl et al, 2002]. Contaminants emitted by the
road tunnels, basically at ground level, will disperse in a way
that is different from the typical plumes emitted by elevated
sources like stacks and chimneys. This latter kind of source has
been widely studied with different typologies of mathematical
and experimental models but the configuration of source
associated with road tunnels emissions has been analysed to a
2003 Firenze University Press 245
much lower extent and in specific cases in both full scale and
laboratory [Carr et al, 1996], [Ide et al, 1987], [Nadel et al,
1996], [Lepade et al, 1996], [Contini et al, 2002]. Therefore it
is necessary to develop reliable simulation codes that can be
used in environmental impact assessment and in monitoring the
existing road tunnels. Validation of these codes could be
performed, at least partially, with the accurate development of
experimental methodologies on small scale models tested in
controlled conditions by using the wind tunnels capabilities.
In this paper an experimental analysis has been conducted
on a small scale model of a road tunnel placed under a hill in
order to investigate the principal characteristics of the
dispersion phenomena related to road tunnel emissions. A first
part of the paper is a characterisation of the boundary-layer and
of the model performances that are improved with respect to a
previous model [Contini et al, 2002]. In the second part of the
paper an analysis of the concentration fields, especially at
ground level, has been carried out in order to investigate the
phenomena that are peculiar to this kind of dispersion. In
particular it has been put in evidence the areas mainly
interested in pollution dispersion and the effect of ambient wind
on the flow inside the pipes of the model. It has been put in
evidence the recirculation effect that influence directly the
emissions because part of the pollutant travelling in one of the
pipe of the road tunnel can be sucked into the other pipe and
travel under the hill.
The results obtained show the complexity of the
phenomena that have to be faced in modelling the dispersion
from this kind of sources that are sometimes placed near urban
areas or near zones of particular environmental interest.

WIND TUNNEL SET-UP AND MODEL DESCRIPTION
A neutral boundary-layer has been artificially generated at
the CRIACIV (Centro di Ricerca Interuniversitario di
Aerodinamica delle Costruzioni ed Ingegneria del Vento) wind
tunnel facility. The wind tunnel facility is about 22.8 m long
(test section length about 11m) with a test section 2.4m wide
and 1.6m tall. The boundary-layer is generated by using large
vortex generators (spires and quarter of ellipse shaped
Counihan as shown in figure 1) placed at the beginning of the
flow development zone and a distribution of roughness placed
in the floor of the wind tunnel of height variable between 10
and 20 mm all over the wind tunnel floor. Largest roughness
are positioned at the beginning of the test section.


Fig. 1) Large vortex generators and surface roughness used
in the wind tunnel set-up for the generation of the neutral
boundary layer.
The characteristics of the boundary-layer have been
measured by using single hot-wire anemometer at different
distances from the position of the downwind portals of the road
tunnel model (the source of the emissions) in absence of the
model. In figure 2(a) the mean velocity profiles are reported for
a reference flow speed, measured above the boundary-layer,
U
ref
=2.8 m/s. The corresponding longitudinal turbulence
profiles are reported in figure 2(b). The boundary-layer has also
been measured at different values of U
ref
obtaining very similar
results. The set-up used allows to generate a boundary-layer
about 0.7m tall (about 3.9 times the maximum height of the
model). The mean velocity profiles is characterised by a power-
law behaviour: ( )

ref ref
z / z U ) z ( U with ranging from
0.15 to 0.17 in the different positions. Friction velocity u* has
been evaluated ranging from 5.2% to 5.4% of the reference
flow speed above the boundary-layer and the values of z
o

obtained by a two-parameters fit is ranging from 0.25 to 0.5
mm.
The two-ways road tunnel is a 1:200 model (82 cm in
length) built with a longitudinal ventilation generated by two
small DC axial fans placed in the middle of the model. Filters,
grid and a distribution of obstacles are placed inside the model
at different distances from the fans in order to artificially
generate a mean velocity profile inside the road tunnel able to
simulate the real profiles present in full scale road tunnel as a
consequence mainly of piston effect and eventual longitudinal
ventilation [Chen et al, 1998]. The tracer injection is made
around the centre of the model just before the axial fan and
separately for each pipe of the model. The total flow-rate of the
tracer is controlled by an electronic mass-flow regulator and the
split into the two pipes is obtained and checked with rotameter
flow-meters. The tracer is therefore well mixed when it reaches
the portals and concentration is quite homogeneous. Therefore
the model is able to reproduce an almost uniform, within a few
percent, distribution of pollutant at the portals that is
representative of full scale scenarios in which the vehicles
emissions are well-mixed inside the road tunnel by the effects
due to the movements of the vehicles themselves.
The mean velocity W profile at the exit portal of the road
tunnel model measured by using a single hot-wire anemometer
in the centre of the exit portal is reported in figure 3. The shape
of the profile is in good agreement with the ones measured by
Chen et al [1998] as consequence of the piston effect. The main
results obtained by Chen et al [1998] indicate that the
distribution of velocity due to the piston effects into the road
tunnel is not strongly influenced by the details of the traffic
inside the tunnel. The maximum speed is obtained nearby the
vehicles themselves. However the piston effect is not confined
nearby the vehicles because a significant speed is usually
measured well above the vehicles near the top of the road
tunnel. The horizontal homogeneity of the exit flow at the
portals is around 6% relatively to the flow at the centre of the
portal. Long term acquisition of W and of the concentration at
the centre of the portal show that the model is stable and
measured variations are well within experimental errors.
A description of the model and of its dimensions is
reported in figure 4.
The model is placed in the wind tunnel under a hill with
one pipe of the road tunnel releasing upwind of the hill against
the wind and the other releasing downwind of the hill in the
same direction as the ambient wind.
2003 Firenze University Press 246

The hill placed above the model has been realised by using
laser-cut wood according to the following EPA equation:













Where x indicate the longitudinal direction corresponding (in
the present measurements) to the wind tunnel axis and z
indicates the vertical direction. Both co-ordinates are expressed
in terms of the parameter with a . In our model a=50 cm
(half of the longitudinal length of the hill) and H=18 cm is the
maximum height of the hill.
0
10
20
30
40
50
60
70
80
90
0.55 0.65 0.75 0.85 0.95 1.05
U(z)/Uref
Z

[
c
m
]
Profile1 (X=0)
Profile2 (X=85 cm)
Profile3 (X=190 cm)

0
10
20
30
40
50
60
70
80
90
0 0.05 0.1 0.15 0.2
Turbulence intensity (%)
Z

[
c
m
]
Profile1 (X=0)
Profile2 (X=85 cm)
Profile3 (X=190 cm)

Figure 2. (a) mean wind vertical profiles at different
distances X from the downwind portal of the model. (b)
longitudinal turbulence profiles for the same values of X.

0
0.5
1
1.5
2
2.5
3
3.5
0 5 10 15 20 25 30
Z (mm)
W

(
m
/
s
)

Figure 3) Exit velocity W (z) vertical profile at the centre of
the portal.

The hill is placed inside the tunnel occupying almost all the
wind tunnel width and it is covered with the same roughness
present on the wind tunnel floor in order to avoid sudden
variations in the boundary-layer. This means that the speed-up
is due to the rapid convergence of the streamlines without
contributions due to ground level sudden changes in flow speed
induced by changes in surface roughness [Panofsky & Dutton,
1984] .
Concentration measurements have been made with a
system equipped with two identical FID (Flame Ionisation
Detector) calibrated with the tracer gas (ethylene). This tracer is
injected in the middle of the road tunnel model and it is mixed
inside the model itself with the environmental air moving
through the road tunnel pipes so that the effective ethylene
concentration at the sources is usually ranging from 400 to 700
ppm in volume according to the physical conditions the are
simulated. Each FID is connected to a motorized samplivalve
with 12 sampling lines placed inside the wind tunnel. Two of
the sampling lines were permanently used for background
concentration monitoring about 2m upstream of the model and
two lines were used to sample concentration in the middle of
the two portals of the road tunnel model (i.e. the sources). This
leaves 20 sampling lines to be used in each measurement.
Background concentrations have been measured every hour in
order to correct each measure with the background effectively
present during sampling. Ground level concentrations have
been measured at 1 cm above the ground corresponding to
about 2m at full scale and every measurement is the average
over 180s that is a period long enough to have a stable value of
the mean concentration.
The stability of concentration measured at the centre of the
portals, representing the source concentrations, has been
evaluated during 10 separate runs carried out in two different
days and the results are reported in figure 5 for the two portals
of the road tunnel model. The exit speed W was 2.9 m/s at the
centre of the portals and U
ref
was 3 m/s. Tracer flow-rate was
nominally the same in each pipe of the road tunnel model.
Results show a reasonable stability within the experimental
errors.
During concentration measurements the exit speed
evaluated at the centre of the portals has been kept constant at
2.9 m/s and equal for both pipes of the model. Because of the
different interaction of the two carriageways with the ambient
wind and of the changes of the mentioned interactions with the
( )1
1
]
1

+
+
2 2 2 2
2
a m
a
1
2
1
x
2 1
2
1
a
H
a
H
m
1
1
]
1

,
_

+
( )
( )1
1
]
1

+

2 2 2 2
2
2 1
2 2
a m
a
1 a m
2
1
z
(a)
(b)
2003 Firenze University Press 247
strength of the ambient wind it is necessary to change the
voltage at the axial fans inside the model in order to keep W
constant according to the different values of U
ref
. The behaviour
of W at different voltage and different U
ref
is reported in figure
6. These kind of measurements allowed to evaluate the voltage
to be applied in order to keep W=2.9 m/s in all conditions. This
of course does not mean that the concentration at the two
portals is constant, because the effective concentration is
influenced by the dilution with fresh and/or polluted air as
consequence of recirculation from one pipe into the other and
to the pushing of air into the pipe releasing downwind of the
hill. Concentrations at the portals have therefore to be used
carefully if a reference concentration for normalisation is
needed


CONCENTRATION FIELD RESULTS
In figure 8(a) longitudinal profiles measured along the road
tunnel pipe axis are reported for different values of U
ref

downwind of the hill. The exit speed W was equal to 2.9 m/s
for both portals. In figure 8(b) the results of measurements
along the axis of the pipe releasing against the wind upwind of
the hill are reported for the same physical situations.
Downwind of the hill it is present a sudden change in the
profiles due to the mixing of the jet emitted from the portal
with the flow coming from above the hill. This is more or less
pronounced as function of the wind velocity. The general area
of dispersion is quite small especially at high speed as a
consequence of the increased dilution. Upwind of the hill the
penetration of the pollutant against wind is quite small and this
generates an area of high concentration just nearby the exit
portal. The upwind jet seems to be folded back and part of the
pollutant is transported downwind passing above the hill. The
transport of pollutant above the hill is also confirmed by the
vertical concentration profiles reported in figure 7 where it is
clear that at only 7 cm from the portals tracer is present up to
about 30 cm from the ground (1.7 times the height of the hill).
In figures 9 and 10 are reported the results of an
investigation carried out with W=2.9 m/s and U
ref
=3 m/s to put
in evidence quantitatively the recirculation effect from one side
of the model into the other side. In figure 9(a) longitudinal
concentration profiles are reported downwind of the hill, and in
figure 9(b) upwind of the hill, for the three different situations:

1) standard situation (both pipes with ventilation on and
tracer injection);
2) the injection in the pipe releasing downwind off but
with ventilation on;
3) the injection and ventilation in the pipe releasing
downwind off.

In figure 10 similar results are reported for the three different
situation in which the changes in tracer injection and ventilation
have been made on the pipe emitting against the wind upwind
of the hill..
Figure 9(a) shows that at the portal releasing downwind it
is still present a certain concentration even if the tracer
injection in this pipe is turned off. This means that part of the
tracer emitted against the wind from the upwind portal is
actually re-circulated downwind inside the model itself. If the
ventilation into the pipe releasing downwind is turned off it is
still present a certain concentration although much lower than
the previous mentioned case. This means that the ambient wind
alone is actually able to push part of the tracer emitted against
the wind inside the model. Figure 9(a) shows that changes in
the working regime in one side of the tunnel influence the other
side because of the different efficiency of the recirculation
phenomena.
Figure 10 illustrates the situations when the changes are
applied to the pipe releasing upwind putting in evidence a
similar effect of the recirculation but the pushing effect of the
ambient wind clearly disappears.



Figure 4) Schematic of the internal set-up of the model. All dimensions are in millimeters.


Tracer injection
Tracer injection Tracer injection
Hill

2003 Firenze University Press 248
490
510
530
550
570
590
610
1 2 3 4 5 6 7 8 9 10
Run number
C

a
t

t
h
e

p
o
r
t
a
l

(
p
p
m
)
Downwind portal
Upwind portal
Downwind portal average
Upwind portal average
Figure 5) Concentration at the centre of the portals in
different runs with W=2.9 m/s and U
ref
=3 m/s. Tracer flow-
rate was nominally constant.



Figure 11 shows normalised concentration maps upwind of the
hill and downwind putting in evidence the area interested in the
dispersion of pollutant and it also illustrates the asymmetry
generated by the presence of the two-ways road tunnel with the
recirculation phenomena. Each map has been normalised with
the concentration at the releasing portals.


1.3
1.5
1.7
1.9
2.1
2.3
2.5
2.7
2.9
3.1
3.3
10 11 12 13 14 15 16 17 18 19 20 21 22 23
Axial Fans voltage (V)
W

(
m
/
s
)
Upwind portal; Uref=0
Downwind portal; Uref=0
Upwind portal; Uref= 5m/s
Downwind portal; Uref=5 m/s
Figure 6) Exit velocity at the centre of the portals as
function of the fan voltage for two different wind speed U
ref

in the wind tunnel.
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
40.0
45.0
0 100 200 300 400 500
Concentration (ppm)
Z

(
c
m
) Upwind
Downwind
Measurements at 7 cm from the portals

Figure 7) Vertical concentration profiles measured at the
two exit portals. W=2.9 m/s and U
ref
=3 m/s.
2003 Firenze University Press 249

0
100
200
300
400
500
600
0 25 50 75 100 125 150
Distance from the portal X (cm)
C

(
p
p
m
)
2 m/s
3 m/s
4.5 m/s
6 m/s
Downwind of the hill (a)

0
100
200
300
400
500
600
700
-100 -75 -50 -25 0
Distance from the portal X (cm)
C

(
p
p
m
)
2 m/s
3m/s
4.5 m/s
6 m/s
Upwind of the hill
(b)

Figure 8) Longitudinal concentration profiles measured at ground level along the model axis for different values of U
ref
. (a)
downwind of the hill and (b) upwind of the hill. W=2.9 m/s.

0
100
200
300
400
500
600
0 25 50 75 100 125 150
Distance from the portal X (cm)
C

(
p
p
m
)
Standard
Downwind tracer off ventilation on
Downwind tracer and ventilation off
O
(a)

0
100
200
300
400
500
600
-50 -37.5 -25 -12.5 0
Distance from the portal X (cm)
C

(
p
p
m
)
Standard
Downwind tracer off ventilation on
Downwind tracer and ventilation off
O
(b)

Figure 9) Ground level concentration measured along the longitudinal axis of the model for different working conditions in the
pipe releasing downwind of the hill. (a) Downwind of the hill. (b) Upwind of the hill.

0
100
200
300
400
500
600
0 25 50 75 100 125 150
Distance from the portal X (cm)
C

(
p
p
m
)
Standard
Upwind tracer off ventilation on
Upwind tracer and ventilation off
(a)

0
100
200
300
400
500
600
-50 -37.5 -25 -12.5 0
Distance from the portal X (cm)
C

(
p
p
m
)
Standard
Upwind tracer off ventilation on
Upwind tracer and ventilation off
(b)

Figure 10) Ground level concentration measured along the longitudinal axis of the model for different working conditions in
the pipe releasing against the wind upwind of the hill. (a) Downwind of the hill. (b) Upwind of the hill.


2003 Firenze University Press 250
10 20 30 40 50 60 70 80
X (cm)
-30
-20
-10
0
10
20
30
Y

(
c
m
)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
0.75
0.80
0.85
-80 -70 -60 -50 -40 -30 -20 -10
X (cm)
-30
-20
-10
0
10
20
30
Y

(
c
m
)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
0.60
0.65
0.70
(a)
(b)

Fig. 11) Normalized concentration maps. (a) downwind of the hill; (b) upwind of the hill. U
ref
=3 m/s and W=2.9 m/s.


CONCLUSIONS
The results obtained in the experiments described show
that the model built can reproduce with sufficient accuracy the
known details of the flow inside a road due to the piston effect
or the combined contribution of piston effect and internal
longitudinal ventilation.
The wind external to the road tunnel influences the internal
flow, especially in the aligned configuration analysed in this
paper.
Results show that a recirculation effect is present that
drives part of the pollutant released at one portal into the other
pipe. This pollutant is therefore moved from one side of the
road tunnel into the other side passing under the hill. As
consequence of the recirculation effect added to the influence
2003 Firenze University Press 251
of the external wind, the concentration at the portals placed at
the two sides of the road tunnel are different even if the exit
speed W at the portals of the model and the pollutant emission
inside the model are kept constant and equal in the two pipes of
the road tunnel model.
The emissions released against the wind are rapidly
stopped and folded back generating high level concentration
around the exit portal of the road tunnel. Part of the pollutant is
successively injected into the road tunnel itself and part is
transported downwind passing above the hill and mixing with
the emissions released at the portal downwind.


REFERENCES

Carr E.L., Ireson R.G., Biltoft C., 1996, Analysis of
Dispersion Characteristics and Induced Turbolent Flows Near
Roadway Intersections from Mobile Sources, Presentation at
the 89
th
Annual Meeting & Exhibition, june 23-28, 96-
WP86.03.

Chen T.Y., Lee Y.T., Hsu C.C., 1998, Investigations of piston-
effect and jet fan-effect in model vehicle tunnels, Journal of
Wind Engineering and Industrial Aerodynamics 73, pp. 99-110.

Contini D., Pasqualetti C., Massini M., Manfrida G., Corti A.,
Bartoli G., Procino L., 2002, Studio della diffusione di
contaminanti gassosi emessi in gallerie stradali mediante un
modello fisico in scala ridotta, 7 Convegno Nazionale di
Ingegneria del vento - IN-VENTO-2002.

Ide Y., Ueyama S., Kobayashi K., 1987, "Wind tunnel
modeling of gas diffusion from a road tunnel outlet", The
Science of the Total Environment, 59, 211-222.

Lepage M.F., Vanderheyden M.D., Davies A.E., Nadel C.,
1996, Simulating Vehicle Emissions at the Exit of a Vehicle
Tunnel, Presentation at the 89
th
Annual Meeting & Exhibition
Nashville, Tennessee June 23-28.

Nadel C., Vanderheyden M., Lepage M., Davies A., Wan P.,
Ginzburg H., Schattanek G., 1996, "Physical modelling of
dispersion of a tunnel portal exhaust plume", Presentation at the
89
th
Annual Meeting & Exhibition Nashville, Tennessee June
23-28.

Oettl D., Sturm P. J., Bacher M., Pretterhofer G., Almbauer R.
A., 2002, A simple model for the dispersion of pollutants from
a road tunnel portal, Atm. Env. 36, pp. 2943-2953.

Panofsky H.A., Dutton J.A., 1984, "Atmospheric Turbulence.
Models and Methods for Engineering Applications", John
Wiley & sons.







2003 Firenze University Press 252
Proceedings of PHYSMOD2003:
International Workshop on Physical Modelling of Flow and Dispersion Phenomena
3-5 September 2003, Prato, Italy
WIND TUNNEL MODELLING OF PEDESTRIAN WIND COMFORT AT A NEW
BUILDING COMPLEX IN HAMBURG


B. Leitl
Meteorological Institute at Hamburg University
Bundesstrasse 55, 20146 Hamburg, Germany
M. Schatzmann
Meteorological Institute at Hamburg University
Bundesstrasse 55, 20146 Hamburg, Germany



MOTIVATION
The process of urban planning and development requires a
number of environmental factors to be considered which are
critical with respect to living comfort and quality of live. Air
pollution, ventilation of built-up urban areas as well as aspects
of shading, thermal comfort and pedestrian wind comfort may
have a significant impact on how people will accept for
example a new building complex. Also the "economical value"
of a building complex is affected by local environmental
factors. Wind tunnel modelling can provide a very efficient
way for predicting winds at pedestrian levels. Even within very
complex urban structures the pedestrian wind comfort can be
assessed safely using a combined model approach consisting of
systematic flow visualization experiments and high resolution
flow measurements.

Within the scope of the structural design of a new building
complex in the city center of Hamburg, a 'typical' pedestrian
wind comfort study was carried out in the BLASIUS wind
tunnel at Hamburg University. Major aims of the project were
the identification of possible high wind areas at pedestrian
levels as well as an improvement of the aerodynamic design of
structural details.


WIND TUNNEL BOUNDARY LAYER MODELLING
The tests were carried out in the BLASIUS wind tunnel at
the Meteorological Institute of Hamburg University. Figure 1
shows a sketch of the 16 m long conventional type boundary
layer wind tunnel. The facility consists of an air intake nozzle
with a contraction ratio of approx. 4:1, screens and modified
Standen spires (Standen, 1972) at the entrance of the 11 m long
test section, a turn table and an adjustable double-ceiling along
the entire test section. The cross section of the tunnel is 1.5 m
wide and 1 m high. A radial-blow fan at the end of the test
section drives the tunnel and the wind speed can be precisely
controlled between 0.2 m/s and about 12 m/s. Placing the wind
tunnel drive at the end of the wind tunnel test section avoids the
flow disturbances generated by the fan affecting the flow in the
test section. The working section of the tunnel is equipped with
a 3-axis stepper-motor controlled probe positioning system
which enables extensive sets of measurements to be carried out
effectively. Many different kinds of flow and dispersion
measurement probes available in the laboratory can be
automatically positioned in the test section with an accuracy of
better than 0.1 mm.

For reference wind speed measurements a Pitot-tube
connected to a differential pressure transducer is used. The
pressure transducer is calibrated against a fine pressure balance
before and after each measurement campaign in order to
eliminate any possible drift of the reference wind speed
measurements. For flow measurements a 2D Laser-Doppler-
Anemometer (LDA) with a BSA burst processing unit from
Dantec

was used. The spatial resolution of the fiber-optic
probe applied is better than 1 mm, and two components of the
local wind vector can be measured simultaneously with a data
rate of at least several hundred Hertz. Although an LDA
system does not need to be calibrated under normal operating
conditions, the accuracy of the system was checked using a
scatter-disc calibrator.

In order to apply the wind tunnel results with confidence to
full scale conditions, the atmospheric boundary layer needs to
be replicated properly at model scale in the test section of the
wind tunnel. The reference profile to be modelled in the wind
tunnel was derived from a field measurement station operated
by the Meteorological Institute at the Hamburg radio
transmission tower (see Pascheke et al, 2003) as well as from
the guideline VDI 3783/12. For boundary layer modelling a
specific spires / floor roughness configuration was developed
by measuring the boundary layer flow in the test section,
comparing the measured flow with the reference flow and
iteratively improving the shape and the arrangement of spires
and floor roughness elements. Figure 2 shows the test section
of the tunnel with modified Standen spires and floor roughness
for modelling the essential wind and turbulence characteristics
at a model scale of 1:500. In Figure 3, the vertical mean wind
profile, turbulence intensity profile and the measured shear
stress profile are plotted. After carefully adjusting the ceiling
of the tunnel in order to minimize the longitudinal pressure
gradient in the test section, an almost ideal constant shear layer
could be established in the test section of the tunnel. In
addition, the spectral characteristics of the modelled turbulent
boundary layer were checked and found to be in good
agreement with full scale conditions for a model scale of 1:500.
2003 Firenze University Press 253




Figure 1: BLASIUS wind tunnel at the Meteorological Institute at Hamburg University.





Figure 2: Test section of the wind tunnel with optimized spires and floor roughness arrangement.

2003 Firenze University Press 254
Figure 3: Measured boundary layer flow in the test section
of the wind tunnel (squares: mean wind profile, empty
circles: turbulence intensity profile, filled diamond: shear
stress profile).

The physical model (scale 1:500) of the city district
included all buildings and topographical structure with a radius
of 600 m around the new building complex. The size of the
model was defined carefully to avoid discrepancies between
model wind flow and full scale conditions by leaving out
essential structures around the area of interest. An upstream
fetch of at least 25H (H: average building height) could be
ensured for all wind directions. Figure 4 shows the complete
model consisting of the turn table part (1.5 m diameter) and the
extended model part (diameter 2.4 m).

VISUALIZATION OF GROUND LEVEL WINDS
Pedestrian level wind fields can be visualized in a very
efficient way by means of so-called sand erosion tests. In a
sand erosion experiment, the surface of the model is covered
with a thin layer of washed silica sand with a narrow grain size
distribution. Typically, the thickness of the sand layer is less
than one millimeter to avoid excessive accumulation of sand in
low wind zones of the model which might affect the model
results. After carefully applying the sand the model was then
exposed to different wind speeds in the test section of the
boundary layer wind tunnel. Whereas no sand erosion may be
observed at low wind speeds, the sand is blown away at higher
wind velocities of the approaching boundary layer flow.
Erosion usually starts in areas where winds are amplified due to
the arrangement of building structures, for instance at the
upwind edges of large buildings. Increasing the wind speed
gradually leads to a further enlargement of areas were sand
erosion occurs as a result of high wind speeds and gustiness of
the wind at pedestrian levels. Taking pictures of the erosion
patterns observed for each wind speed and combining them
digitally, isolines of possible discomfort can be drawn easily.
However it must be stated clearly that the sand erosion tests are
supposed to deliver a qualitative information only on where
critical wind speeds might occur. The actual wind speed values
for instance in relation the above roof wind speed must be
measured directly using hot wire anemometry or LDA.


WIND SPEED MEASUREMENTS AT PEDESTRIAN
LEVELS
After the most critical zones concerning wind speeds close
to the ground were identified for a total of 36 wind directions
(0 to 360 in steps of 10), wind speed measurements were
carried out at 4 mm height above ground in the model
corresponding to 2 m height above ground at full scale. The
spatial resolution of the wind speed measurements was about
0.3 m (full scale) and the temporal resolution was better than 1
Hz for full scale conditions. The high spatial and temporal
resolution enables the local winds as well as the gustiness of the
winds to be assessed safely. Sufficiently long time series of the
horizontal wind vector have been recorded at the most sensitive
points for all wind directions found to be critical. The time
series were processed off-line with respect to mean wind speed,
local wind direction as well as gustiness and top wind speeds
for different averaging times at full scale conditions. The
results were documented using the above roof wind speed as a
reference. In a final step of processing, the wind tunnel
measurements were transferred to full scale conditions using
the annual wind statistics of Hamburg. As a result, the annual
exceedance of critical wind speeds could be calculated and a
quantitative assessment of pedestrian wind was made. Thus
some advice for the improvement of pedestrian wind comfort
was given for the most critical zones of the building complex.

RESULTS
Figure 5 shows an exemplary result of one sand erosion
experiment. Zones which are most critical with respect to
pedestrian level winds are marked by circles. The gradually
increasing size of 'discomfort zones' can clearly be detected.
All 36 wind directions investigated were assessed 'manually' by
visual analysis of the sand erosion pictures. Two distinct wind
direction sectors could be found, identifying winds from 90
(east) to 270 (west) as potentially critical wind directions and
identifying winds from about 315 (north-west) to be more
favourable regarding wind comfort. This basic outcome is
combined with the annual wind statistics of Hamburg in Figure
6. Unfortunately, the critical wind sectors cover almost all of
the most frequent wind directions in Hamburg. Consequently,
the result of the sand erosion experiments must be called
critical and qualitative wind speed measurements at pedestrian
levels had to be carried out. Table 1 gives an example of the
results of wind speed measurements.
U in m/s
Tu in %
u'w' in m
2
/s
2
z
m
o
d
e
l
l
[
m
m
]
1 2 3 4 5 6
0 25 50 75 100
-0.2 -0.1 0 0.1 0.2 0.3
50
100
150
200
250
300
350
400
450
500
U [m/s]
Tu [%]
u'w' [m
2
/s
2
]
2003 Firenze University Press 255





Figure 4: Detailed physical model of the city district (red roof: new building complex, inner circle: core
model, outer circle: extended fetch).


Figure 5: Typical result of a sand erosion experiment for visualization of wind-critical zones (circles: areas
identified as critical with respect to pedestrian wind comfort).
2003 Firenze University Press 256

At the same point wind speed measurements were carried out
with and without noise protection screens located above the
measurement point. The effect of the noise screen increasing
pedestrian winds, can clearly be detected. Without the
unfavourably designed noise protection mounted between the
building blocks the mean wind speed as well as the gustiness of
winds at pedestrian levels was reduced significantly. In order
to improve the aerodynamic behaviour of the noise screens,
several slightly modified noise screen structures were designed,
tested in the model and suggested as replacements for the
original screens.

A final assessment of pedestrian wind comfort was given
based on a number of wind comfort criteria which can be found
in literature. For instance in Baumller et al. (1993) a list of
threshold values for winds speeds can be found. As listed in
Table 2, critical wind speeds should be related to some
maximum allowable exceedance per year as well as to typical
zones of specific human activity. In the present study it was
found that severe wind problems must be expected if the
building structure would be built as tested.

CONCLUSIONS
Pedestrian wind comfort in a complex urban area was
visualized and measured by means of physical modeling in a
boundary layer wind tunnel. Building a detailed physical
model of the city district and carefully modeling a site-specific
boundary layer flow in the test section ensure a safe and
reliable transfer of the model results to full scale conditions. A
sand erosion technique was used to identify areas being
potentially critical with respect to pedestrian wind comfort. In
a second step of testing, a quantitative analysis of ground-level
winds was carried out based on high resolution flow
measurements using a 2D LDA system. Combining the results
of flow measurements with the annual wind statistics of the site
enabled a reliable assessment of pedestrian winds in and around
the investigated structure to be given. In addition, a number of
aerodynamic corrections of the building complex was
suggested in order to assist the architects improving the overall
value and environmental comfort in an around the new
structure.


Figure 6: Wind statistics and critical wind directions identified.
2003 Firenze University Press 257

U
i
/ U
ref
[ - ]
U
ref
= 5 m/s 8 m/s 12 m/s
Position A with noise protection screens
peak wind speed 1.97 9.86 15.78 23.67
minimum wind speed 0.04 0.19 0.3 0.45
mean wind speed 0.39 1.93 3.1 4.64
standard deviation of wind speed 0.35 1.76 2.82 4.22
gust wind speed 1.44 7.21 11.54 17.31

Position A without noise protection screens
peak wind speed 0.08 4.01 6.42 9.64
minimum wind speed 0.01 0.06 0.1 0.15
mean wind speed 0.24 1.2 1.92 2.89
standard deviation of wind speed 0.15 0.76 1.22 1.83
gust wind speed 0.7 3.49 5.59 8.39

Table 1: Exemplary results of wind speed measurements at 2m height above ground (full scale, wind
direction 160, gust defined as 10 seconds average)
gust wind speed allowable exceedances
in percent of time
assessment situation
< 6 m/s no problems related to wind comfort
> 6 m/s max. 5 % acceptable to pedestrian walk ways, shopping
areas, street cafes and play grounds
> 8 m/s max. 1 % acceptable to waiting / sitting areas
> 6 m/s
> 15 m/s
max. 20 %
max. 0.05 %
acceptable to areas quickly crossed by passers-
by (relaxed criteria)
> 10 m/s max. 1 % acceptable to areas quickly crossed by passers-
by (strong criteria)
> 13 m/s max. 1 % can be accepted at building corners
> 13 m/s > 1 % unpleasant / troublesome
WIND PROTECTION REQUIRED!
> 18 m/s potentially dangerous
WIND PROTECTION REQUIRED!

Table 2: Assessment scheme adopted from Baumller et. al (1993).
2003 Firenze University Press 258
REFERENCES

Standen, N.M. (1972): A spire array for generating thick
turbulent shear layers for natural wind simulation in wind
tunnels. Tech. Rept. LTR-LA-94, National Aeronautical
Establishment, Ottawa, Canada

Pascheke, F.; Leitl, B.; Schatzmann, M. (2003): Results
from Recent Observations in an Urban Boundary Layer.
Workshop on urban boundary layer parameterizations
(Extended abstracts), COST Action 715, Office for official
publications of the European Communities, ISBN 92-894-
4143-7.

VDI 3783/12 (2000): Physical modeling of flow and
dispersion processes in the atmospheric boundary layer
Application of wind tunnels. Part 12, Verein Deutscher
Ingenieure Clean Air Handbook, Vol. 1b

Baumller, J.; Hoffmann, U.; Reuter, U. (1993):
Stdtebauliche Klimafibel: Hinweise fr die Bauleitung, Folge
2, Hrsg. Witschaftsministerium Baden-Wrttemberg, 1993.



2003 Firenze University Press 259
Proceedings of PHYSMOD2003:
International Workshop on Physical Modelling of Flow and Dispersion Phenomena
3-5 September 2003, Prato, Italy
WIND TUNNEL EXPERIMENTS WITHIN THE SCOPE OF THE
OKLAHOMA CITY TRACER EXPERIMENTS


B. Leitl
Meteorological Institute at Hamburg University
Bundesstrasse 55, 20146 Hamburg, Germany
F. Pascheke
Meteorological Institute at Hamburg University
Bundesstrasse 55, 20146 Hamburg, Germany


M. Schatzmann
Meteorological Institute at Hamburg University
Bundesstrasse 55, 20146 Hamburg, Germany
P. Kastner-Klein
School of Meteorology, University of Oklahoma
100 E. Boyd, Norman, OK 73019, USA



ABSTRACT
The papers describes the basic concept of systematic wind
tunnel experiments carried out in support of an extensive field
measurement campaign, studying instantaneous dispersion in a
complex urban area. Wind tunnel results are used for an
efficient and reliable planning of the measurements as well as
in support of the interpretation of the results from a limited
amount of field data. The basic goals of accompanying wind
tunnel modelling are to increase the efficiency of time
consuming and expensive field measurements as well as to gain
further information on instantaneous flow and dispersion
phenomena in complex urban areas.


INTRODUCTION
Although the emissions from urban air pollution sources
have been significantly reduced in most industrialized
countries, urban air quality problems are still a matter of
concern. While dispersion from distributed urban pollution
sources can be modelled numerically in many cases with
reasonable accuracy, the short-term behaviour of instantaneous
emissions from ground level point sources within the urban
canopy is not yet completely understood. For the improvement
of numerical codes and model validation, field data as well as
results of systematic laboratory experiments are needed. In
order to generate a validation data set for dispersion modelling
in complex urban terrain, an extensive set of tracer gas
experiments will be carried in Oklahoma City (USA) in the
summer of 2003. Within the scope of the 'Joint Urban 2003'
experiments, releases from different point sources will be
simulated using a passive tracer and the resulting plume will be
measured by means of a variety of tracer gas sampling systems.
In addition, an extensive set of meteorological measurements
will be carried out simultaneously in order to quantify the
boundary conditions during the tracer release periods.


MOTIVATION
The emergence of increasingly powerful computers
stimulated the development of complex micro-scale flow and
transport model. Numerical modelling plays an increasingly
important role in many practical environmental applications
such as urban air quality prediction or modelling accidental
releases of hazardous gases. The models are now commonly
applied to predict pollutant dispersion in complex structured
urban canopy layers and use of such models is made, for
instance in the licensing of new industrial plants and in safety
analysis studies.
The model used vary in their level of complexity, ranging
from simple obstacle accommodating or so-called diagnostic
models to prognostic model which employ not only mass
conservation but also the equations for momentum and energy.
While a number of models are already commercially available,
others are still treated as research tools. Independent of the
type of model they all contain a substantial amount of empirical
input, for example in the turbulent closure scheme applied.
The increasing use of numerical models is accompanied by
a growing awareness that most of the model have never been
tested in a procedure of careful and rigorous evaluation for
complex urban type dispersion problems. Nevertheless, these
models are used as a basis for making decisions with profound
economic and social consequences.
Within the scope of the Oklahoma City field campaign, a
high-quality validation data set is generated, which should be
suitable for testing the 'fitness for purpose' of numerical models
in a unified and comprehensive way.
Validation data for development and testing of numerical
models are not just any experimental data. Validation data
must fulfil a solid set of requirements in order to be called
complete, representative, sufficiently accurate and completely
documented (Leitl/Schatzmann, 1999, Leitl 2000). If these
requirements are not met, the freedom in the setup of numerical
model runs is likely to distort the results of a model test. It is
well known that a wide variety of numerical results can be
generated within the limits of reasonable assumptions regarding
2003 Firenze University Press 260
input data. Consequently, a solid conclusion concerning the
capabilities of a model cannot be achieved as long as the degree
of freedom in setting up model test runs is minimized.
Field data are often designated as the one and only source
of validation data. However, it must be considered that field
results as well as other experimental data have their specific
limitations. Results from field campaigns usually represent
unique situations with a complex set of boundary conditions.
Appropriate recording of all essential boundary conditions is
often impractical because of the limitations in instrumentation.
Changing boundary conditions like the constantly changing
weather due to the diurnal circle can cause a large variation






Figure 1: The Large Boundary Layer Wind Tunnel "WOTAN" at Hamburg University.

Figure 2: Oklahoma City Model mounted in the test section of the wind tunnel upwind view with
spires/roughness in front of the model.
2003 Firenze University Press 261







Figure 3: Oklahoma City model in the test section of the wind tunnel (downwind view, floor roughness
in front of the model, traverse system).

Figure 4: Exemplary result of Laser light sheet visualizations
(horizontal light sheet, wind flow from the upper left corner
to the lower right corner of the image).
2003 Firenze University Press 262
even if the results are time averaged. As could be shown in
laboratory experiments and long-term field data sets, the
remaining scatter in typical half-hour averages of measured
flow and dispersion quantities can vary by one order of
magnitude. Consequently, the limited representativeness of
field data limits their usefulness for comparison with the steady
state results delivered by most of the numerical dispersion
models.
Another source of validation data is made available
through laboratory experiments in boundary layer wind tunnels.
Although the wind tunnel modelling is limited in most cases to
neutral stratification and the results of physical modelling
incorporate some simplification and abstraction from physical
reality, results of systematic wind tunnel experiments have
some major advantages over field data. Different levels of
complexity can be generated, ranging from single obstacle
situations to complex urban configurations containing all
geometric details. The boundary conditions can be precisely
controlled and kept constant over a long period of time. Thus,
it is possible to simulate the same steady-state situation as
calculated by most of the micro-scale flow and dispersion
models. The accuracy of laboratory grade instrumentation and
the spatial and temporal resolution of up-to-date wind tunnel
instrumentation enable measurements to be carried out with at
least the same full-scale resolution as in field experiments.
Another important advantage of laboratory data is of course,
that most of the physical boundary conditions of a test can be
controlled at will and that all of them can be measured with
high accuracy.
To overcome a number of limitations from both validation
data sources it is wise to combine field and laboratory data.
Supporting field measurement campaigns by corresponding
wind tunnel experiments can add significant value to field data
because gaps of information inherent in any field data set can
be filled by laboratory data. In addition, systematic wind
tunnel testing can help to plan extensive field campaigns more
safely. If a physical model of a complex urban structure is
available, results from wind tunnel simulations can be used for
example to define what source locations will deliver the most
valuable results depending on wind direction and release
condition. Wind tunnel measurements carried out in advance
can also used for planning the layout of instrumentation and the
timing of releases because all instantaneous behaviour of the
dispersion process due to turbulence is modelled directly.


OUTLINE OF THE WIND TUNNEL EXPERIMENTS
The experiments are to be carried out in the new Large
Boundary Layer Wind Tunnel 'WOTAN' at Hamburg
University (Figure 1). The 25 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 is 4 m wide
and 2.75 to 3.25 m high (variable ceiling). For probe
positioning and automated measurements, the tunnel has a
computer controlled traverse system with a positioning
accuracy of better than 0.1mm on all three axes for all types of
probes used in the tunnel. An extensive custom made software
package has been developed for automated and semi-automated
measurements, probe calibration and positioning, online data
visualization, data reduction and data validation.
In order to achieve a high accuracy of the wind tunnel data,
all measurement systems as well as the precision mass flow
controllers for emission sources are calibrated against
independent certified standards available in the laboratory.
In the first step of the project, the atmospheric boundary
layer flow across Oklahoma City was replicated at a scale of
1:300 in the test section of the boundary layer wind tunnel. As
a reference the results from local wind measurements in
Oklahoma City as well as from the University of Oklahoma
(School of Meteorology) were applied. The boundary layer
flow approaching the Central Business District was classified
as type B/C according to ANSI/ASCE 7-95. The large cross
section of the tunnel enables a large city area to be modelled.
For a model area 1.8 km x 1.8 km all significant buildings and
obstacles have been replicated in the model enabling a fetch of
about 1 km to be considered in the wind tunnel model for each
wind direction. For the given model scale, the typical
geometrical resolution of velocity measurements is about 0.6
mm in the wind tunnel which corresponds to about 18 cm at
full scale conditions. For concentration measurements, the
sampling area is about 0.2 mm in diameter in the wind tunnel,
which corresponds to about 5 cm in diameter in full scale.
For boundary layer modelling, a conventional
spires/roughness set up is used. In an iterative process the
shape and arrangement of the spires and the floor roughness
were gradually optimised until agreement with the required
full-scale conditions was reached. The proper scale of the wind
tunnel boundary layer was documented by high-resolution flow
measurements. The mean wind profiles as well as integral
length scales and spectra of turbulence were found to be in
good agreement with full-scale conditions. In addition it can be
shown, that a careful adjustment of the boundary layer enables
even large scale wind fluctuations up to a time scale of about 1
hour are replicated by the model boundary layer flow. The
complete documentation of the modelled boundary layer flow
consists of:

documentation of the spatial and temporal homogeneity
and stability of the approaching flow in the test section
representative vertical and lateral mean flow profiles (all
three components of the wind vector)
representative vertical and lateral profile of turbulence
intensities (all three components of the wind vector)
representative vertical and lateral profiles of turbulent
momentum fluxes of the approaching flow
representative integral length scales for different heights
above ground
representative power spectral density for all three
components of the wind vector
complete temporal and angular analysis of the wind
direction fluctuations inherent in the modelled boundary
layer flow.

Once the geometric scale of the model boundary layer was
defined, a detailed aerodynamic model of the city district of the
area to be covered in both field measurements and numerical
simulations was constructed. Figure 2 and Figure 3 shows the
model mounted in the test section of the wind tunnel. The
extended fetch around the core model allows an assessment of
the effects on calculated results due to the limited size of the
area modelled for example in a numerical model. Suitable
2003 Firenze University Press 263
point source were located in the model at all locations possibly
used for tracer gas releases. The point source design enables
smoke releases to be used for flow visualization experiments.
A first measurement campaign in the wind tunnel was
conducted concerning mainly with flow visualization
experiments. The results from Laser light sheet experiments
are assisting the set up of the field experiments. The overall
behaviour of the plumes formed at different release locations
was documented by taking still images as well as recording
video sequences. The latter was of special interest to the field
measurement groups because the highly instantaneous
behaviour of the plumes as well as the big scatter of the
emissions due to turbulent fluctuations will lead to a significant
scatter in field data even for similar and stationary boundary
conditions. Figure 4 shows a exemplary result of the Laser
light sheet experiments
In a further step, the instantaneous behaviour of the tracer
plumes resulting from different sources was measured for
different wind directions. In order to quantify the concentration
fluctuations to be expected during full scale experiments long
time series of concentration fluctuations were recorded at
several measurement points all across the different plumes. An
off-line analysis of the time series enabled all statistical
parameters as well as the intermittency and the mean time of
the presence of the plume at different measurement locations to
be documented. The results show that the resulting plumes are
clearly affected by the high-rise building structures and the axis
of the plume was not found to be parallel with the mean wind
direction. Measurement points with a high intermittency factor
could be found all across the model area. From this, a big
scatter in the field data thus is expected.
Further experiments are planned to simulate the Joint
Urban 2003 field experiments in order to prove the ability of
the wind tunnel model to replicate all essential features of the
dispersion processes. The main part of the project will be an
extensive set of high-resolution flow and dispersion
measurements. The intension is to measure flow and
concentration patterns in several horizontal and vertical
measurement planes throughout the entire core model. The
resulting flow and dispersion maps will enable a qualitative and
quantitative comparison of global measurements with
corresponding results from numerical modelling.

ACKNOWLEDGMENTS
The support by staff of the Department of Geosciences at
the University of Oklahoma is gratefully acknowledged.


REFERENCES

Leitl, B.; Schatzmann, M. (1999): Generation of High
Resolution Reference Data for the Validation of Micro-Scale
models. published in: Recent Developments in Measurement
and Assessment of Air Pollution, VDI-Komission Reinhaltung
der Luft, Report 1443, ISBN 3-18-091443-2, pp. 647-656

Leitl, B. (2000): Validation Data for Microscale Dispersion
Modelling. EUROTRAC Newsletter, 22. pp. 28-32




2003 Firenze University Press 264
Proceedings of PHYSMOD2003:
International Workshop on Physical Modelling of Flow and Dispersion Phenomena
3-5 September 2003, Prato, Italy
DEVELOPMENT OF VERTICAL TURBULENT FLUXES OVER AN IDEALISED URBAN
ROUGHNESS


Schultz M.
Div. of Technical Meteorology
Meteorological Institute, University of Hamburg
Dr. Leitl B.
Div. of Technical Meteorology
Meteorological Institute, University of Hamburg


Schatzmann M.
Div. of Technical Meteorology
Meteorological Institute, University of Hamburg




ABSTRACT
Systematic measurements above a regularly arranged array
of cubes with h = 25 mm height were performed in the wind
tunnel BLASIUS of Hamburg University.
In a first series of measurements the spacing of the cubes
was changed from 0.5h over 1h to 2h. The experiments were
carried out without using vortex generators in the wind tunnel.
It was intended to study solely the influence of the bottom
roughness on the flow development. During the second
measurement campaign the cube distance was 2h all the time,
but the length of the array varied from 38 rows over 45 rows
and 65 rows up to 85 rows. Again no spires were used. The
experimental programme was rounded up by a third series of
measurements in which the cubes were approached by a
boundary layer generated in the common way by means of
vortex generators and a ground roughness. Cube spacing here
was 2h, the number of rows was 38.
Subsequently the results will be presented and compared
with each other.


NOMENCLATURE

d
0
: = displacement height
h : = height of roughness elements
: = von Karman constant
u, v, w : = components of wind speed along x, y, and
z axis, respectively.
u
*
: = friction velocity
x, y, z : = coordinates of an rectangular Cartesian
system with x-axis defined in direction of
mean wind.
z
0
: = roughness length



INTRODUCTION
In the last decades more and more studies in the field as
well as in the wind tunnel deal with the examination of the
urban boundary layer. But in contrast to the rural boundary
layer that is comparatively well understood, urban boundary
layer concepts are not yet settled. A detailed review of results
from field measurements carried out within and above the
urban canopy layer can be found in the paper by Roth (2000).
According to him, the boundary layer over an urban structure
can be subdivided into two layers, the inertial sublayer and the
roughness sublayer. Inside the inertial sublayer (hereafter
denoted by IS), the turbulent fluxes are assumed to be constant
with height, and under neutral conditions the usual logarithmic
law applies:

,
_

0
0 *
ln
z
d z u
u

(1)


The roughness sublayer (hereafter denoted by RS) lies
beneath the IS. Inside the RS turbulent fluxes and all other
boundary layer properties are assumed to be influenced by
individual roughness elements. Therefore, the turbulent fluxes
are neither spatially homogeneous nor is the logarithmic law
applicable. The height of the RS is still subject of debate but it
is in the range of 2 to 5 times the average building height
(Raupach 1991, Roth 2000). Rotach (1999) states that the RS
over a very rough surface can occupy a substantial portion of
the surface layer.
After a roughness change an internal boundary layer
develops that is influenced by the new roughness. This internal
boundary layer is growing with fetch. Inside the internal
boundary an equilibrium layer is forming consisting of the RS
and IS. Above the internal boundary layer lies a transition
region, which is covered by the outer boundary layer. Cheng
and Castro (2002) found in their study that a growing
equilibrium layer behind a step change in roughness first has to
modify the RS before a new equilibrium IS can develop.

Task of this study was to investigate up to which height it
is possible to model a boundary layer that is representative for
2003 Firenze University Press 265
the underlying roughness. Second task was to investigate the
characteristics of this boundary layer for different types of cube
arrays, e.g. the heights of the RS and IS etc..


EXPERIMENTAL SET-UP
All measurements were carried out in the small boundary
layer wind tunnel Blasius of the Meteorological Institute.
This wind tunnel has a test section, which is 1.5 m wide, 1m
high. The usable length of the flow establishment and test
section is 17.5 m. The wind tunnel is equipped with an
adjustable double ceiling (see fig. 1). During the two
measurement campaigns an array of regularly arranged sharp
edged wooden cubes with cube height of h = 25 mm was put in
the test section of the wind tunnel. Above the cube array
profiles were measured with a LDA fiber probe of Dantec.
The probe had a focal length of 50 mm and a measuring
volume of d
x
= 0.121 mm, d
y
= 0.122 mm and d
z
= 1.151 mm.
The measured wind components were u and w for all profiles.
16 m
intake
honeycombs
doubleceiling
adjustable
blower
flowdirection
screens boundarylayer development section
(7.5mlong)
DCmotor
variablespeed
test section
1.5mwide, 1mhigh, 4mlong
turntable

Figure 1. Schematic view of the small wind tunnel of the
Meteorological Institute of Hamburg University.

The first measurement campaign was performed from
December 2002 to January 2003. To represent an idealized
urban roughness a regularly arranged array of cubes with size h
= 25 mm was positioned in the test section of the wind tunnel.
During this campaign the array of cubes consisted of 38 rows.
After 32 rows a small array in the centre of the tunnel with size
3 x 3 cubes was painted black in order to minimize reflections
during the LDA measurements. The last three rows were
placeholders to guarantee undisturbed measurements inside the
black array. Three different configurations were measured with
a a spacing between the cubes varying from 0.5h, over 1h to 2h.
Figure 2 shows the set-up for the 0.5h-configuration.
For all three configurations the following measurements
were carried out:

1. A profile was measured upstream of the cube array
to determine the characteristics of the approach
flow.
2. Along the centre line of the whole cube array
profiles were measured between 30 and 390 mm
height with a vertical resolution of 40 mm and a
horizontal distance of 50 mm. This should give a
fair representation of the development of the flow
above the cube array with increasing fetch.
3. Above the black array the density of measurement
points was increased to 21 spatially distributed
profiles each with 29 Points between 26 mm and
390 mm height.



Figure 2. Set-up for the configuration 0.5h





Figure 3. Set-up for the cube distance 2h and the extended
array of 40 rows.

During all measurements no spires were used in the wind
tunnel. The flow development was solely governed by the
roughness elements. The Reynolds number criterion for fully
2003 Firenze University Press 266
rough flows given in Snyder and Castro (1998) was well
fulfilled in all experiments.

In the second campaign, only experiments with the cube
spacing 2h were carried out, again without spires as in the first
campaign. Task in this part of the campaign was to extend the
number of cube rows behind the black array from 3 to 10 in
order to surely exclude possible upstream effects from change
in roughness at the end of the array. Subsequently, the fetch
was extended by adding 20 rows and then 40 rows in front of
the (black) intensive measurement array (figure 3). Again
profiles above the centreline were measured, but compared to
the first campaign the vertical resolution was reduced to 10 mm
on the expense of the horizontal distance that varied now
between 150 and 300 mm. Above the black cube array only
four spatially distributed profiles were measured, but with the
same resolution and position as in the first campaign.


Finally, in a third series of experiments the short array of
cubes consisting of 32 rows in front of the black field and a
cube spacing of 2 h was used again, but now the approach flow
was fully turbulent. Spires were installed at the entrance of the
flow establishment section and staggered arranged cubes with
distance 6h served as roughness elements. Again Profiles above
the centreline were measured between the heights of 30 and
390 mm with a vertical resolution of 10 mm. The horizontal
distance between these profiles was 300 mm. In the black
intensive measurement field 4 profiles at the same positions as
in the first part were measured.


RESULTS AND DISCUSSION
Only results from the first measurement campaign are
given here. The data from the second and third series of
experiments are presently being processed. They will be
presented at the conference. For further details see Schultz
(2003).
A proper equilibrium boundary layer should be
characterised by fully developed mean velocity and turbulence
intensity profiles. Since after a roughness step the height up to
which equilibrium is to be expected should grow with fetch, it
was worthwhile to document the development of flow
properties with downstream distance for selected heights above
the elements. Figs. 5a, b and c show that at the example of
turbulence intensities measured along the centreline of the test
section for the three different cube arrays.
As can be seen in Fig. 5a, for the cubes spaced by 0.5h a
constant turbulence intensity is found in 70 mm above ground
at a distance of X= 1000 mm whereas at height levels z = 110
mm and z= 150 mm the turbulence intensities are still
increasing. For the less dense roughness arrays (1h and 2h) the
fetch needed to achieve equilibrium 70 mm above ground
increases to 1200mm and 2000 mm, respectively (Figs 5b and
c). Again, at higher elevations equilibrium was not yet
obtained.





Figure 4. Set-up for the experiments with a fully turbulent
approach flow. Cube spacing was 2 h, spires were used in
the flow establishment section.


Results from turbulent flux measurements above the (black
painted) intensive measurement section are presented in Fig. 6.
Shown are numerous vertical uw profiles taken at different
horizontal positions relative to the cubes. Adopting the
procedure applied in Cheng and Castro (2002), an average
mean profile (white points) was calculated which was then used
to determine the roughness sublayer and the inertial sublayer.
Cheng and Castro defined the upper limit of the RS as the
height at which all flux profiles are converging. The vertical
extent of the IS is defined as the region within which the
vertical variation of the spatially averaged profile is below 5%.
Other possible ways to determine the IS like fitting the
logarithmic law to the spatially averaged profile are less
stringent but give similar results (see Schultz 2003). The grey
dashed lines indicate the scatter.
For all three cube arrays the RS extends up to a height of
38 to 40 mm which corresponds to 1.5 h. The upper edge of the
IS for the cube arrays with spacing 0.5h and 1h reaches 60 mm,
whereas for the low density array (spacing 2h) the IS height is
only 48 mm.

2003 Firenze University Press 267
a)
X [ mm]
u
r
m
s
/
u
0 500 1000 1500 2000
0
0.05
0.1
0.15
0.2
0.25
Z = 70 mm
Z = 110 mm
Z = 150 mm
height
0.5h array of cubes

b)
X [ mm]
u
r
m
s
/
u
0 500 1000 1500 2000 2500
0.01
0.03
0.05
0.07
0.09
0.11
0.13
0.15
0.17
0.19
0.21
0.23
0.25
Z = 70 mm
Z = 110 mm
Z = 150 mm
1h array of cubes
height

c)
X [ mm]
u
r
m
s
/
u
0 1000 2000 3000
0.01
0.03
0.05
0.07
0.09
0.11
0.13
0.15
0.17
0.19
0.21
0.23
0.25
Z = 70 mm
Z = 110 mm
Z = 150 mm
2h array of cubes
height

Figure 5a)-c). Development of turbulence intensities at
levels 70mm, 110mm and 150 mm above ground for
roughness arrays with cube spacing of 0.5h (a), 1h (b) and
2h(c). The error bars indicate the experimental scatter.


COMPARISON OF SPATIALLY AVERAGED PROFILES
FOR DIFFERENT CUBE DISTANCES
Fig. 7 shows spatially averaged profiles for the three cube
arrays for other flow characteristics, i.e. the horizontal
component of mean velocity u (Fig. 7 a) and the vertical
component of mean velocity w (Fig. 7 b). Fig. 7 c replicates
Fig.6 but shows only mean turbulent fluxes - ' ' w u as a function
of roughness spacing and normalized height z/h.
The profiles of the mean wind speed u for the cube arrays
0.5h and 1h have the same shape, but the profile of 1h is
slightly shifted to lower wind speeds. The gradient for these
two profiles is the same. The profile of the cube distance 2h has
a different gradient in wind speed up to the level z= 7h. Above
this level all three profiles lie within the scatter band. This
indicates, that free stream conditions have been reached at this
level.
The shape of the three w-profiles for the 0.5h, 1h and 2h
arrays are similar with the tendency of higher values for less
densely packed cube arrays. However, the measured values are
near to detection limit of the LDA instrument and lie more or
less within the scatter band.
The turbulent flux profiles (Fig 7 c) for 1h and 2h fall
nearly on top of each other whereas those for roughness
spacing 0.5 h are smaller, above all near to the roughness
elements. It appears that approach flow conditions are found
above z = 6 h.

SUMMARIZING AND CONCLUSION
For the investigated cube arrays a fully developed
equilibrium layer was found up to a level of 70 mm. The
inertial sublayer was found to be more or less independent of
the cube spacing. The cube arrays 0.5h and 1h had a very
similar influence on the mean velocity components. These
rather densely packed roughnesses behaved more like a sand
roughness whereas the cube array with 2h spacing showed
features of individual obstacles. With respect to turbulence
there were more similarities between the cube arrays 1h and 2h.
Both arrays generate more turbulence through the individual
obstacles than was found in the experiments with the array
0.5h. Also, the turbulence generated by cube array 2h reaches
to higher elevations than the others.
However, before final conclusions can be made, the
findings should be consolidated by experiments with one or two
additional roughness arrays.


2003 Firenze University Press 268
a)
- u'w'/ u ref
Z
[
m
m
]
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007
0
10
20
30
40
50
60
70
80
90
100
Roughness sublayer
Inertial sublayer
cube spacing 0.5h

b)
- u'w' / u ref
Z
[
m
m
]
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007
0
10
20
30
40
50
60
70
80
90
100
Roughness sublayer
Inertial sublayer
cube spacing 1h

c)
- u'w' / u ref
Z
[
m
m
]
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007
0
10
20
30
40
50
60
70
80
90
100
Roughness sublayer
Inertial sublayer
cube spacing 2h


Figure 6a)-c). Normalized turbulent flux profiles above
cube arrays with cube spacing 0.5h (a), 1h (b) and 2h (c).
For further explanations see text.


a)
u / u ref
Z
/
h
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
1
2
3
4
5
6
7
8
9
0.5h
1h
2h
cube spacing
b)
w/ u ref
Z
/
h
-0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05
0
1
2
3
4
5
6
7
8
9
0.5h
1h
2h
cube spacing
c)
u'w' / uref
Z
/
h
0 0.0025 0.005 0.0075
0
1
2
3
4
5
6
7
8
9
0.5h
1h
2h
cube spacing
Figure 7a)-c). Comparisons of spatially averaged Profiles
over cube arrays with cube spacing 0.5h (a), 1h (b) and 2h
(c). Error bars indicate the scatter band.
2003 Firenze University Press 269
REFERENCES

Cheng,H. , Castro,I. 2002: Near wall flow development
after a step change in surface roughness, Boundary Layer
Meteorology 105, 411-432
Cheng,H. , Castro,I. 2002: Near wall flow over urban-
like roughness, Boundary Layer Meteorology 104, 229-259
Feigenwinter,C. , Vogt,R. , Parlow,E. 1999: Vertical
Structure of Selected Turbulence Characteristics above an
Urban Canopy, Theoretical and Applied Climatology 62, 51-63
Hgstrm,U. , Bergstrm,H. , Alexandersson, H. 1982:
Turbulence characteristics in a near neutrally stratified urban
atmosphere, Boundary Layer Meteorology 23, 449-472
Kastner-Klein,P 2001: Overview of near-surface
turbulence parametrisations, COST Action 715 workshop on
urban boundary layer parametrisations, extended abstracts,
Zurich, Switzerland, 31-40
Macdonald R.W. 2000: Modelling the mean velocity
profile in the urban canopy layer, Boundary Layer Meteorology
97, 25-45
Macdonald,R.W. , Carter,S. , Slawson, P.R. 2002:
Physical modeling of urban roughness using arrays of regular
roughness elements, Water, Air and Soil Pollution: Focus 2,
541-554
Macdonald,R.W. , Griffiths,R.F. , Hall,D.J. 1998: A
comparison of results from scaled field and wind tunnel
modeling of dispersion in arrays of obstacles, Atmospheric
Environment, 32, 3845-3862
Oikawa,S. , Meng,Y. 1995: Turbulence characteristics
and organized motion in a suburban roughness layer, Boundary
Layer Meteorology 74, 289-312
Raupach,M.R. , Thom,A.S. , Edwards,I. 1979: A wind
tunnel study of turbulent flow close to regularly arrayed rough
surfaces, Boundary Layer Meteorology, 18, 373-397
Rotach, M.W. 1993a: Turbulence close to a rough urban
surface part I: Reynolds stress, Boundary Layer Meteorology,
65, 1-28
Rotach, M.W. 1993b: Turbulence close to a rough urban
surface part II: Variances and Gradients, Boundary Layer
Meteorology, 66, 75-92
Rotach, M.W. 1995: Profiles of turbulence statistics in
and above an urban street canyon, Atmospheric Environment,
29, 1473-1486
Rotach, M.W. 1999: On the influence of the urban
roughness sublayer on turbulence and dispersion, Atmospheric
Environment, 33, 4001-4008
Roth,M. 2000: Review of atmospheric turbulence over
cities, Quart. J. Roy. Meteorol. Soc., 146, 941-990
Roth,M. ,Oke,T.R. 1993: Turbulent transfer relationships
over an urban surface. I: Spectral characteristics, Quart. J. Roy.
Meteorol. Soc., 119, 1071-1104
Schultz, M. 2003: Development of turbulent fluxes over an
idealized urban roughness. Diploma thesis at the
Meteorological Institute at the University of Hamburg.
Snyder, H.S., and Castro, I.P. (1998) Surface roughness in
laboratory modelling - how rough is rough ? Proc. 4
th
UK
Conference on Wind Engineering, pp. 83-88.

You might also like