- Email: [email protected]
Wind tunnel investigation of dispersion of pollutants due to wind flow around a small building
Pergamon
Armorphmic Environmenr Vol. 28. No. II. PP. lL719-1826, 1994 Eloevier Scicoa Ltd Printed in Great Brilain
WIND TUNNEL INVESTIGATION OF DISPERSION OF POLLUTANTS DUE TO WIND FLOW AROUND A SMALL BUILDING M. H. 21 Douglas
Road,
J. Aeronautics
Department,
Imperial
MIRZAI
Hornchurch, K.
College,
Essex
RMl
I IAN,
U.K.
HARVEY
Prince
Consort
Road,
London
SW7
ZBY,
U.K.
and C. D. Chemical
and Biological
Defence
JONES
Establishment,
Porton
Down,
Salisbury
SP4 OJQ, U.K.
Abstract-Experimental investigations have been conducted in a wind tunnel on flow and dispersion of pollutants around an isolated building. A neutrally stable atmosphere at l/75 scale was simulated. A substantial quantity of experimental data was collected to produce pictures showing the mean time pollution concentration at a predetermined plane behind the building. The normalized results are shown in a three-dimensional format. The tests described here have been done for deviations in the wind direction of -10, -5.0, +Sand +lo”. Key
word
index:
Concentration,
pollution,
simulation,
main building height = 63.2 mm measured percentage helium concentration zero plane displacement normalized helium concentration = lo-* cCJL*/Q length scale, the distance between the source and the building windward (5 BH) an experimentally determined constant volumetric flow rate of the helium source wind speed at 1 BH
mean velocity at height Z velocity
vertical coordinate reference height horizontal deviation in the wind direction (degree).
INTRODUCTION
Understanding
the
details
of the
dispersion
probe,
visualization.
build up in the near wake, even if only approximately. It would be important also to know the concentration dosages around the building if this were ventilated. If the release is transient, the time necessary for the concentrations to build up and then decrease to a safe level, once the contaminant source had ceased, is also valuable information. Thus the problems in wind engineering and dispersion of pollutants due to wind flow especially in the wakes of building are of increasing concern. Flow visualization methods and direct concentration measurements have been widely used for scaled down building models, and to a lesser extent for full scale buildings. Notable among recent full-scale field works are those conducted by Drivas and Shair (1974), Ogawa and Oikawa (1982) Davis (1982) and Jones and Griffiths (1984). However, field tests are very time consuming and expensive, and the test conditions (such as ambient temperature, wind velocity, wind direction, speed advection, turbulence intensity) are dictated by nature and are hence uncontrollable. Wind-tunnel simulations offer cost effectiveness and convenience. However, we then have the limitation of these experiments being performed at smaller scale and thus at reduced Reynolds number. As public awareness of the environment has increased in the last few years, the simulation of the
NOMENCLATURE
reference
aspirating
of pollu-
tion in the near wake of buildings is very important for estimating the concentrations or dosages of dangerous contaminants in the event of accidental releases and the consequential effect on personnel in or near the buildings. Particular cases would be leaks from nuclear reactors or chemical plants or other industrial scenarios. In such cases it would be very important to know the concentration levels or doses that could 1819
M. H.
1820
MIRZAI
atmospheric boundary layer in a wind-tunnel has been developed, and has become an important tool to study the dispersion and transport of pollution using small scale models. Among these works are those of Wilson (1971), Vincent (1977) Wedding er al. (1977), Huber and Snyder (1982), Fackrell (1984), Boreham (1984), Bachlin and Plate (1988) and others. During field trials, shifts in the wind direction are experienced and the times when the wind blows perpendicular to a face of the building are therefore only occasionally observed. The tests described here have thus been conducted for deviations in the wind direction (0) of from - IO to + IO” for the case of the building aligned lengthways with the flow, to enable a better understanding of the flow and dispersion fields to be obtained.
8
0.8 -
i 2 'c)
. 0.6 -
5,
04-
cr 01
EXPERIMENTAL
The subject present work
APPROACH
AND
TECHNIQUE
which was used throughout the a scaled down (l/75) representative small
of the test was
isolated building. It had a simple rectangular plane form with a flat roof, on which there were two penthouses and had a small annexe on one end. The penthouses and the annexe were interchangeable. All the tests carried out in this work were conducted using the complete shape as shown in Fig. 4. Simulated boundary layer The first stage of the present work was to simulate an atmospheric boundary layer in a wind tunnel. To reduce the complexity of the problem a neutrally stable atmosphere was selected to be simulated (this restriction eliminates the effects of buoyancy forces as well as the turbulence associated with the thermal free convection due to solar heating). The experiments were carried out in a low speed wind tunnel with a working section of 9 m x 3 m x I.5 m length,
02-
00
" 0.0
0.2
04
0.6
0.8
I.0
I 1
UNref Fig.
I. Comparison
between the model atmosphere mean velocity law profiles (exponents 0. I85 and 0.240).
profile
and the power
1.0
08
8 'e 8 k 0
0.6
5
0.4
02
00 0
2
Fig. 2. Model
4
atmosphere
6
longitudinal
6
turbulent
10
intensity
12
profile.
14
Wind
Row around
a small
1821
building Clgarette /
Hotfllm
Fig. 3. The modified
aspirating
probe
width and height, respectively (the IO’ x 5’ Honda Tunnel situated at Imperial College, London). One-third of the lower region of the atmospheric boundary layer in a neutrally stable atmosphere at l/75 scale was modelled. This approach has been described in more detail by Cook (1973) and by Boreham (1984). An accelerating technique for growing a thick boundary layer in the wind tunnel within a short distance, similar to that proposed by Counihan (1969). has been adopted. This method involved installing a castellated barrier on the wind tunnel floor at the upstream end of the working section. The drag produced by this barrier was of the same order as the momentum loss of the required boundary layer velocity profile. This was followed by five “vorticity generators”* placed evenly across the working section. The purpose of these generators was to produce large-scale vorticity of the opposite sign which accelerates the upward diffusion of the high intensity turbulence which is produced at floor level by the roughness elements. The whole wind tunnel working section floor was covered with 34 mm high block-like roughness elements. These elements were made in arrays from 1.5 mm thick hard polystyrene sheets using a vacuum moulding technique. The roughness density (defined by the total plan area of the roughness elements divided by the total floor area of the wind-tunnel working section) was 20%. Velocity and turbulence intensity profiles were obtained using hot-wire anemometery techniques. An exponential power law I Z-d n v -= I/ rcl
( Z,,,-d
>
was used to correlate the mean velocity data. In this expression Li is the mean velocity at height Z; Urcl and Zrel are the reference velocity and reference height, respectively; d is the zero plane displacement and n is an experimentallydetermined constant. Plotting O/U,,, vs (Z-d/Z,,,-d) and applying a least-squares fitting method produced a value for I/n of 0.185. With the mean longitudinal velocity of about 5.5 m s- ’ at the main building height, this gives Reynolds numbers of 22,OQO upwards based on the main building height. Typical velocity and turbulence profiles are presented in Figs I and 2.
/
for a better
signal-to-noise
ratio.
diameter containing within it a hot-film mounted upstream of a sonic nozzle, connected 10 stable vacuum source. The sonic nozzle was intended to remove velocity variations in the flow past the hot-film. However, in practice the probe seemed sensitive to turbulence or flow direction and overestimated the RMS concentration values. This noise could not be safely and reliably removed by either digital or analogue filters. It was discovered, however, that this noise could be substantially reduced by modifying the tube which carries the sampled flow 10 the probe sensor so as to minimize the turbulence in the sampled flow. The modification consisted of a funnel-shaped glass contraction, with a total volume of less than 0.2 cm3, which constrained the flow to become nearly laminar in the same way as the contraction
and honeycomb filters in a wind tunnel (Fig. 3). The hot-film was also replaced by a hot-wire which reduced the projected hot sensor surface by nearly 40 times. These modifications improved the signal to noise ratio by a factor of 5. Introducing a filter in the glass contraction and replacement of the hot-film by a hot-wire, also led to a longer lasting calibration (due to reduction of settled airborne particles on the hot-wire as a result of filter and smaller surface). Helium gas was chosen 10 serve as the contaminant source. The point source was a small (IOmm high by 13 mm diameter) cylindrical shape aerator, similar to those found in fish tanks. In each test the location of source was kept fixed with respect to the building in order to simulate the change in the wind direction. The concentration measurements were made by releasing helium gas at a rate ofO.001 m3 s-l from a fixed point ahead of the model, and sampling with the aspirating probe for the sufficient time for a reliable average value to be recorded. Because the wind tunnel was of the closed returned type, during the process the background helium concentration will slightly but steadily increase. To minimize this effect each reading in the survey was preceded by one method in which no helium was released. This reading provided a datum. The concentration patterns were mapped out using an automatic probe traversing system linked to a computer. The release initiation and shut-off of the helium flow was also under the control of the computer.
RESULTS
Concentration
Filter
AND
DISCUSSION
nwasuremmt
Contaminant concentration measurements have been made around the building model using a modified “T.S.I.” aspirating probe. An aspirating probe measures composition (or temperature) by controlling the fluid velocity and sensing only the composition (or temperature). The original T.S.I. aspirating probe consisted of a tube of about 2 mm outside *These have also been referred as “vortex generators”. Huber and Snyder (1982). Counihan (1969).
e.g.
Flow visualization has been widely used to study the flow structure in the wake of buildings or obstacles and especially to investigate the transport of pollutant materials to and from the wake. The concentration is very time-dependent and each flow visualization picture shows only the instantaneous components in a highly turbulent flow. Hence, many such pictures need to be taken and analysed to build up an understanding
M. H. MIRZAI
1822
of the wake flow. Thus pictures showing the timeaveraged value of the pollutant concentration continue to be very valuable and complementary to the flow visualization. In the tests described here, the building model has been aligned lengthways at small angles, 0, to the main free-stream vector (with the L-shaped penthouse at the downwind end). The helium source was positioned at about 5 EH (main building height) upwind away from the model, and kept aligned with the building centre line for each set-up during the scanning. Each scanned picture comprised of 319 (array of II x 19) points (stations). Each point represents the mean time concentration of 2500 samples taken at a frequency of 500 Hz. The Y-Z plane I2 mm downwind, as defined in Fig. 4, was surveyed for five fixed angles (0 = 0, - 5, - 10, + 5 and + IO”). The I2 mm gap between the Y-Z plane and the model was necessary to allow the model to be rotated either way without interfering with the Y-Z plane. The interpolated results which are in three-dimensional format are shown in Figs 5-9 as contour plots. The horizontal axis, Y, and the
ef a/
vertical axis, Z, are width and height of the building, respectively. The third variable (dimension), k is the normalized helium concentration defined by k = 10-l
CUL2 __.
Q
For the case of wind blowing normal to the building (O=O), a high concentration of helium gas is observed, restricted to a relatively small area close to the ground (Fig. 5). Although the source was set on the centre-line of the building the asymmetry in its shape due to the penthouses causes the helium plume to flow around the opposite side of the building. In addition to the main peak a secondary one just below roof height is also seen due perhaps to vortices shed along the roof edge. Significant spreading of helium throughout the wake can be observed. When wind yawed (0 #O), the first noticeable feature of the surveys is the position of the maximum concentration of helium gas. As could be expected, this moves in sympathy with the position of the source. In Figs 6
Penthouses
Fig. 4. The
origin
and detail
of the model
building
with
all the extras
attached
Wind
Fig.
5. Interpolated
Fig. 6. Interpolated
helium
helium
concentration
flow
around
a small
data measured in the Y-Z plane for varying upwind and in-line with the origin.
concentration BH
data measured in the Y-Z upwind and in-line with
1823
building
plane for varying the origin.
angle of yaw (0 =O”);
angle
of yaw (0=
source
5 BH
+ 5”); source
5
1824
M.
H.
MIRZAI
ef al
Wmd
-5
-4
-3
-2
-1
0
2
1
4
3
5
Y W-U Fig. 7. Interpolated
-5
helium
-4
concentration
data measured in the Y-Z plane for varying 5 BH upwind and in-line with the origin.
-3
-2
-1
0
1
2
angle of yaw (0 = + 10”): source
3
4
5
Y KW Fig. 8. Interpolated
helium
concentration BH
data measured in the Y-Z plane for varying upwind and in-line with the origin.
angle of yaw (0 = - 5”): source
5
Wind flow around a small building
1825
Fig. 9. Interpolated helium concentration data measured in the Y-Z plane for varying angle of yaw (0 = - loo); source 5 Bff upwind and in-line with the origin.
and 7, where wind has yawed anticlockwise (0= +5 and + IO”, respectively), the source is biased to the
right hence the hot spots have formed on the right of the building. In Figs 8 and 9, the wind has yawed clockwise (6 = - 5 and - lo”, respectively),the source is biased to the left, and hence the hot spots have formed on the left of the building. Although
the helium
gas is very buoyant
with a
density only l/7 that of air, in all the casesthe peak concentration was found close to the ground, i.e. the buoyancy forces had a negligible effect* and the helium marker gas was swept around the side of the building, from the left (asseenfacing upwind) in Figs 5, 8, and 9 and from the right in Figs 6 and 7. When
the wind
has yawed
either
clockwise
or
anticlockwise, it can be seen that the main helium plume which forms the hot spots has been split into two unequal parts. One part has formed very close to the downwind comer of the building and the other part away from it. The relative concentration in each part varies with the angle of yaw. For the present setup (fI= +_5and + lo”), the part which is closer to the building is significantly stronger. It is thought that the helium plume has entrained into the horse-shoe vortex that wraps around the front of the building, and has then been passed into a secondary vortex shed from
+ This is correct for the present experimental set-up which had an average wind speed of about 5.5 ms- ’ at 1 BH. The buoyancy effects of the source, will vary considerably with wind speed. AE 28:11-B
the roof and finally transported downwind alongside the building relatively undispersed. It is anticipated that a frequent small change in the wind direction is likely to change this trend. Thus in atmospheric conditions where significant wind shift is common a more uniform distribution of marker gas in the wake is likely. Further surprising behaviour can be observed by comparing Figs 6 and 7 with Figs 8 and 9. The only difference in experimental set-up and configuration between these two setsof figures is the wind direction. In Figs 6 and 7 the wind yawed anticlockwise, while in Figs 8 and 9 the wind yawed clockwise. It can be seen that the peak concentration where wind yawed clockwise is much higher. A possible explanation for this behaviour is that the position of the penthouses are rather less obstructive when the wind yawed anticlockwise. Hence, the helium plume could passby less dispersed. The mean concentration results inside the near wake for those setsof data which wind blew at zero or a small angle is rather uniform. There is a significant transverse transfer of material within the base cavity flow. The pictures also shows that there is a tendency for transportation of pollutant materials across the wake towards the taller part of the building. Figure 10 shows the summary of data presented in the Figs 5-9. In this plot maximum concentration is plotted vs changes in the wind direction. The concentration has reached to a maximum when 0x - 5” and dropped to a minimum when 0 z + 5”. This behaviour was caused by the building asymmetry.
1826
M. H. MIRZN et al. 9.0
1.0
7.0
(I.0
5.0 -10
0
Aqlr
5
10
of wind (deg.)
Fig. 10. Maximum concentration data for varying angle of yaw (0 from - 10” to + 10”).
CONCLUSION Pollutant concentrations are affected by the building wake flow which is highly dependent on the building shape and orientation, and also on the nature and scale of the approaching boundary layer flow. The mean concentration results which have been produced with the source close to the building show clearly the path of the helium plume. In all cases this passes around the side of the building and close to the ground. There is significant transverse transfer of material within the base cavity flow. The pictures also showed that there is a tendency for transportation of pollutant materials across the wake towards the taller part of the building. Although the helium gas is very buoyant with a density only l/7 that of air, the fact that the hot spots occurred at ground level prove that the turbulent mixing within the flow is the dominant mechanism in dispersing the helium gas swamping any buoyancy effects for this particular wind speed. REFERENCES Bachlin W. and Plate E. J. (1988) Wind tunnel simulation of accidental releases in chemical plants (edited by Grefen K. and L&e1 J.). En&. Met. 291-303. Boreham B. W. (1984) Preliminary investigation of dispersion of pollutants in model building wake Bows using bipolar space charge. IC Aero Report 84-04, Imperial College of Science and Technology, Department of Aeronautics, August 1984.
Cook N. J. (1973) On simulation of the lower third of the urban adiabatic boundary layer in a wind tunnel. Atmospheric Environment 7, 691-705. Counihan J. (1969) An improved method of simulating an atmospheric boundary layer in a wind tunnel. Atmospheric Environment 3, 197-214. Davis R. D. (1982) Investigations of transport in complex atmospheric flow systems. Ph.D. thesis, California Institute of Technology. Drivas P. J. and Shair J. H. (1974) Probing the air flow within the wake downwind of a building by means of a tracer technique. Atmospheric Enuironmenl 8, 1165-1175. Fackrell J. E. (1984) Parameters characterising dispersion in the near wake of buildings. J. Wind Engng ind. Aerodyn. 16, 97-118. Huber A. H. and Snyder W. H. (1982) Wind tunnel investigation of a rectangular-shaped building on dispersion of effluents from short adjacent stacks. Atmospheric Enoironmerit 16, 2837-2848. Jones C. D. and Griffiths R. F. (1984) Full-scale experiments on dispersion around an isolated building using an ionised air tracer technique with very short averaging time. Atmospheric Environment 18, 903-916. Ogawa Y. and Oikawa S. (1982) A field investigation of the Bow and diffusion around a model cube. Atmospheric Environment 16, 207-222. Vincent J. H. (1977) Model experiments on the nature of air pollution transport near buildings. Atmospheric Enuironmerit 11, 765-774. Wedding J. B., Lombardi J. and Cermak J. E. (1977) A wind tunnel study of gaseous pollutants in city street canyons. Colorado State University, June 1977. Wilson D. J. (1971) Turbulent dispersion in atmospheric shear flow and its wind tunnel simulation. Von Karman Institute for fluid dynamics, Technical note 76, August 1971.
Armorphmic Environmenr Vol. 28. No. II. PP. lL719-1826, 1994 Eloevier Scicoa Ltd Printed in Great Brilain
WIND TUNNEL INVESTIGATION OF DISPERSION OF POLLUTANTS DUE TO WIND FLOW AROUND A SMALL BUILDING M. H. 21 Douglas
Road,
J. Aeronautics
Department,
Imperial
MIRZAI
Hornchurch, K.
College,
Essex
RMl
I IAN,
U.K.
HARVEY
Prince
Consort
Road,
London
SW7
ZBY,
U.K.
and C. D. Chemical
and Biological
Defence
JONES
Establishment,
Porton
Down,
Salisbury
SP4 OJQ, U.K.
Abstract-Experimental investigations have been conducted in a wind tunnel on flow and dispersion of pollutants around an isolated building. A neutrally stable atmosphere at l/75 scale was simulated. A substantial quantity of experimental data was collected to produce pictures showing the mean time pollution concentration at a predetermined plane behind the building. The normalized results are shown in a three-dimensional format. The tests described here have been done for deviations in the wind direction of -10, -5.0, +Sand +lo”. Key
word
index:
Concentration,
pollution,
simulation,
main building height = 63.2 mm measured percentage helium concentration zero plane displacement normalized helium concentration = lo-* cCJL*/Q length scale, the distance between the source and the building windward (5 BH) an experimentally determined constant volumetric flow rate of the helium source wind speed at 1 BH
mean velocity at height Z velocity
vertical coordinate reference height horizontal deviation in the wind direction (degree).
INTRODUCTION
Understanding
the
details
of the
dispersion
probe,
visualization.
build up in the near wake, even if only approximately. It would be important also to know the concentration dosages around the building if this were ventilated. If the release is transient, the time necessary for the concentrations to build up and then decrease to a safe level, once the contaminant source had ceased, is also valuable information. Thus the problems in wind engineering and dispersion of pollutants due to wind flow especially in the wakes of building are of increasing concern. Flow visualization methods and direct concentration measurements have been widely used for scaled down building models, and to a lesser extent for full scale buildings. Notable among recent full-scale field works are those conducted by Drivas and Shair (1974), Ogawa and Oikawa (1982) Davis (1982) and Jones and Griffiths (1984). However, field tests are very time consuming and expensive, and the test conditions (such as ambient temperature, wind velocity, wind direction, speed advection, turbulence intensity) are dictated by nature and are hence uncontrollable. Wind-tunnel simulations offer cost effectiveness and convenience. However, we then have the limitation of these experiments being performed at smaller scale and thus at reduced Reynolds number. As public awareness of the environment has increased in the last few years, the simulation of the
NOMENCLATURE
reference
aspirating
of pollu-
tion in the near wake of buildings is very important for estimating the concentrations or dosages of dangerous contaminants in the event of accidental releases and the consequential effect on personnel in or near the buildings. Particular cases would be leaks from nuclear reactors or chemical plants or other industrial scenarios. In such cases it would be very important to know the concentration levels or doses that could 1819
M. H.
1820
MIRZAI
atmospheric boundary layer in a wind-tunnel has been developed, and has become an important tool to study the dispersion and transport of pollution using small scale models. Among these works are those of Wilson (1971), Vincent (1977) Wedding er al. (1977), Huber and Snyder (1982), Fackrell (1984), Boreham (1984), Bachlin and Plate (1988) and others. During field trials, shifts in the wind direction are experienced and the times when the wind blows perpendicular to a face of the building are therefore only occasionally observed. The tests described here have thus been conducted for deviations in the wind direction (0) of from - IO to + IO” for the case of the building aligned lengthways with the flow, to enable a better understanding of the flow and dispersion fields to be obtained.
8
0.8 -
i 2 'c)
. 0.6 -
5,
04-
cr 01
EXPERIMENTAL
The subject present work
APPROACH
AND
TECHNIQUE
which was used throughout the a scaled down (l/75) representative small
of the test was
isolated building. It had a simple rectangular plane form with a flat roof, on which there were two penthouses and had a small annexe on one end. The penthouses and the annexe were interchangeable. All the tests carried out in this work were conducted using the complete shape as shown in Fig. 4. Simulated boundary layer The first stage of the present work was to simulate an atmospheric boundary layer in a wind tunnel. To reduce the complexity of the problem a neutrally stable atmosphere was selected to be simulated (this restriction eliminates the effects of buoyancy forces as well as the turbulence associated with the thermal free convection due to solar heating). The experiments were carried out in a low speed wind tunnel with a working section of 9 m x 3 m x I.5 m length,
02-
00
" 0.0
0.2
04
0.6
0.8
I.0
I 1
UNref Fig.
I. Comparison
between the model atmosphere mean velocity law profiles (exponents 0. I85 and 0.240).
profile
and the power
1.0
08
8 'e 8 k 0
0.6
5
0.4
02
00 0
2
Fig. 2. Model
4
atmosphere
6
longitudinal
6
turbulent
10
intensity
12
profile.
14
Wind
Row around
a small
1821
building Clgarette /
Hotfllm
Fig. 3. The modified
aspirating
probe
width and height, respectively (the IO’ x 5’ Honda Tunnel situated at Imperial College, London). One-third of the lower region of the atmospheric boundary layer in a neutrally stable atmosphere at l/75 scale was modelled. This approach has been described in more detail by Cook (1973) and by Boreham (1984). An accelerating technique for growing a thick boundary layer in the wind tunnel within a short distance, similar to that proposed by Counihan (1969). has been adopted. This method involved installing a castellated barrier on the wind tunnel floor at the upstream end of the working section. The drag produced by this barrier was of the same order as the momentum loss of the required boundary layer velocity profile. This was followed by five “vorticity generators”* placed evenly across the working section. The purpose of these generators was to produce large-scale vorticity of the opposite sign which accelerates the upward diffusion of the high intensity turbulence which is produced at floor level by the roughness elements. The whole wind tunnel working section floor was covered with 34 mm high block-like roughness elements. These elements were made in arrays from 1.5 mm thick hard polystyrene sheets using a vacuum moulding technique. The roughness density (defined by the total plan area of the roughness elements divided by the total floor area of the wind-tunnel working section) was 20%. Velocity and turbulence intensity profiles were obtained using hot-wire anemometery techniques. An exponential power law I Z-d n v -= I/ rcl
( Z,,,-d
>
was used to correlate the mean velocity data. In this expression Li is the mean velocity at height Z; Urcl and Zrel are the reference velocity and reference height, respectively; d is the zero plane displacement and n is an experimentallydetermined constant. Plotting O/U,,, vs (Z-d/Z,,,-d) and applying a least-squares fitting method produced a value for I/n of 0.185. With the mean longitudinal velocity of about 5.5 m s- ’ at the main building height, this gives Reynolds numbers of 22,OQO upwards based on the main building height. Typical velocity and turbulence profiles are presented in Figs I and 2.
/
for a better
signal-to-noise
ratio.
diameter containing within it a hot-film mounted upstream of a sonic nozzle, connected 10 stable vacuum source. The sonic nozzle was intended to remove velocity variations in the flow past the hot-film. However, in practice the probe seemed sensitive to turbulence or flow direction and overestimated the RMS concentration values. This noise could not be safely and reliably removed by either digital or analogue filters. It was discovered, however, that this noise could be substantially reduced by modifying the tube which carries the sampled flow 10 the probe sensor so as to minimize the turbulence in the sampled flow. The modification consisted of a funnel-shaped glass contraction, with a total volume of less than 0.2 cm3, which constrained the flow to become nearly laminar in the same way as the contraction
and honeycomb filters in a wind tunnel (Fig. 3). The hot-film was also replaced by a hot-wire which reduced the projected hot sensor surface by nearly 40 times. These modifications improved the signal to noise ratio by a factor of 5. Introducing a filter in the glass contraction and replacement of the hot-film by a hot-wire, also led to a longer lasting calibration (due to reduction of settled airborne particles on the hot-wire as a result of filter and smaller surface). Helium gas was chosen 10 serve as the contaminant source. The point source was a small (IOmm high by 13 mm diameter) cylindrical shape aerator, similar to those found in fish tanks. In each test the location of source was kept fixed with respect to the building in order to simulate the change in the wind direction. The concentration measurements were made by releasing helium gas at a rate ofO.001 m3 s-l from a fixed point ahead of the model, and sampling with the aspirating probe for the sufficient time for a reliable average value to be recorded. Because the wind tunnel was of the closed returned type, during the process the background helium concentration will slightly but steadily increase. To minimize this effect each reading in the survey was preceded by one method in which no helium was released. This reading provided a datum. The concentration patterns were mapped out using an automatic probe traversing system linked to a computer. The release initiation and shut-off of the helium flow was also under the control of the computer.
RESULTS
Concentration
Filter
AND
DISCUSSION
nwasuremmt
Contaminant concentration measurements have been made around the building model using a modified “T.S.I.” aspirating probe. An aspirating probe measures composition (or temperature) by controlling the fluid velocity and sensing only the composition (or temperature). The original T.S.I. aspirating probe consisted of a tube of about 2 mm outside *These have also been referred as “vortex generators”. Huber and Snyder (1982). Counihan (1969).
e.g.
Flow visualization has been widely used to study the flow structure in the wake of buildings or obstacles and especially to investigate the transport of pollutant materials to and from the wake. The concentration is very time-dependent and each flow visualization picture shows only the instantaneous components in a highly turbulent flow. Hence, many such pictures need to be taken and analysed to build up an understanding
M. H. MIRZAI
1822
of the wake flow. Thus pictures showing the timeaveraged value of the pollutant concentration continue to be very valuable and complementary to the flow visualization. In the tests described here, the building model has been aligned lengthways at small angles, 0, to the main free-stream vector (with the L-shaped penthouse at the downwind end). The helium source was positioned at about 5 EH (main building height) upwind away from the model, and kept aligned with the building centre line for each set-up during the scanning. Each scanned picture comprised of 319 (array of II x 19) points (stations). Each point represents the mean time concentration of 2500 samples taken at a frequency of 500 Hz. The Y-Z plane I2 mm downwind, as defined in Fig. 4, was surveyed for five fixed angles (0 = 0, - 5, - 10, + 5 and + IO”). The I2 mm gap between the Y-Z plane and the model was necessary to allow the model to be rotated either way without interfering with the Y-Z plane. The interpolated results which are in three-dimensional format are shown in Figs 5-9 as contour plots. The horizontal axis, Y, and the
ef a/
vertical axis, Z, are width and height of the building, respectively. The third variable (dimension), k is the normalized helium concentration defined by k = 10-l
CUL2 __.
Q
For the case of wind blowing normal to the building (O=O), a high concentration of helium gas is observed, restricted to a relatively small area close to the ground (Fig. 5). Although the source was set on the centre-line of the building the asymmetry in its shape due to the penthouses causes the helium plume to flow around the opposite side of the building. In addition to the main peak a secondary one just below roof height is also seen due perhaps to vortices shed along the roof edge. Significant spreading of helium throughout the wake can be observed. When wind yawed (0 #O), the first noticeable feature of the surveys is the position of the maximum concentration of helium gas. As could be expected, this moves in sympathy with the position of the source. In Figs 6
Penthouses
Fig. 4. The
origin
and detail
of the model
building
with
all the extras
attached
Wind
Fig.
5. Interpolated
Fig. 6. Interpolated
helium
helium
concentration
flow
around
a small
data measured in the Y-Z plane for varying upwind and in-line with the origin.
concentration BH
data measured in the Y-Z upwind and in-line with
1823
building
plane for varying the origin.
angle of yaw (0 =O”);
angle
of yaw (0=
source
5 BH
+ 5”); source
5
1824
M.
H.
MIRZAI
ef al
Wmd
-5
-4
-3
-2
-1
0
2
1
4
3
5
Y W-U Fig. 7. Interpolated
-5
helium
-4
concentration
data measured in the Y-Z plane for varying 5 BH upwind and in-line with the origin.
-3
-2
-1
0
1
2
angle of yaw (0 = + 10”): source
3
4
5
Y KW Fig. 8. Interpolated
helium
concentration BH
data measured in the Y-Z plane for varying upwind and in-line with the origin.
angle of yaw (0 = - 5”): source
5
Wind flow around a small building
1825
Fig. 9. Interpolated helium concentration data measured in the Y-Z plane for varying angle of yaw (0 = - loo); source 5 Bff upwind and in-line with the origin.
and 7, where wind has yawed anticlockwise (0= +5 and + IO”, respectively), the source is biased to the
right hence the hot spots have formed on the right of the building. In Figs 8 and 9, the wind has yawed clockwise (6 = - 5 and - lo”, respectively),the source is biased to the left, and hence the hot spots have formed on the left of the building. Although
the helium
gas is very buoyant
with a
density only l/7 that of air, in all the casesthe peak concentration was found close to the ground, i.e. the buoyancy forces had a negligible effect* and the helium marker gas was swept around the side of the building, from the left (asseenfacing upwind) in Figs 5, 8, and 9 and from the right in Figs 6 and 7. When
the wind
has yawed
either
clockwise
or
anticlockwise, it can be seen that the main helium plume which forms the hot spots has been split into two unequal parts. One part has formed very close to the downwind comer of the building and the other part away from it. The relative concentration in each part varies with the angle of yaw. For the present setup (fI= +_5and + lo”), the part which is closer to the building is significantly stronger. It is thought that the helium plume has entrained into the horse-shoe vortex that wraps around the front of the building, and has then been passed into a secondary vortex shed from
+ This is correct for the present experimental set-up which had an average wind speed of about 5.5 ms- ’ at 1 BH. The buoyancy effects of the source, will vary considerably with wind speed. AE 28:11-B
the roof and finally transported downwind alongside the building relatively undispersed. It is anticipated that a frequent small change in the wind direction is likely to change this trend. Thus in atmospheric conditions where significant wind shift is common a more uniform distribution of marker gas in the wake is likely. Further surprising behaviour can be observed by comparing Figs 6 and 7 with Figs 8 and 9. The only difference in experimental set-up and configuration between these two setsof figures is the wind direction. In Figs 6 and 7 the wind yawed anticlockwise, while in Figs 8 and 9 the wind yawed clockwise. It can be seen that the peak concentration where wind yawed clockwise is much higher. A possible explanation for this behaviour is that the position of the penthouses are rather less obstructive when the wind yawed anticlockwise. Hence, the helium plume could passby less dispersed. The mean concentration results inside the near wake for those setsof data which wind blew at zero or a small angle is rather uniform. There is a significant transverse transfer of material within the base cavity flow. The pictures also shows that there is a tendency for transportation of pollutant materials across the wake towards the taller part of the building. Figure 10 shows the summary of data presented in the Figs 5-9. In this plot maximum concentration is plotted vs changes in the wind direction. The concentration has reached to a maximum when 0x - 5” and dropped to a minimum when 0 z + 5”. This behaviour was caused by the building asymmetry.
1826
M. H. MIRZN et al. 9.0
1.0
7.0
(I.0
5.0 -10
0
Aqlr
5
10
of wind (deg.)
Fig. 10. Maximum concentration data for varying angle of yaw (0 from - 10” to + 10”).
CONCLUSION Pollutant concentrations are affected by the building wake flow which is highly dependent on the building shape and orientation, and also on the nature and scale of the approaching boundary layer flow. The mean concentration results which have been produced with the source close to the building show clearly the path of the helium plume. In all cases this passes around the side of the building and close to the ground. There is significant transverse transfer of material within the base cavity flow. The pictures also showed that there is a tendency for transportation of pollutant materials across the wake towards the taller part of the building. Although the helium gas is very buoyant with a density only l/7 that of air, the fact that the hot spots occurred at ground level prove that the turbulent mixing within the flow is the dominant mechanism in dispersing the helium gas swamping any buoyancy effects for this particular wind speed. REFERENCES Bachlin W. and Plate E. J. (1988) Wind tunnel simulation of accidental releases in chemical plants (edited by Grefen K. and L&e1 J.). En&. Met. 291-303. Boreham B. W. (1984) Preliminary investigation of dispersion of pollutants in model building wake Bows using bipolar space charge. IC Aero Report 84-04, Imperial College of Science and Technology, Department of Aeronautics, August 1984.
Cook N. J. (1973) On simulation of the lower third of the urban adiabatic boundary layer in a wind tunnel. Atmospheric Environment 7, 691-705. Counihan J. (1969) An improved method of simulating an atmospheric boundary layer in a wind tunnel. Atmospheric Environment 3, 197-214. Davis R. D. (1982) Investigations of transport in complex atmospheric flow systems. Ph.D. thesis, California Institute of Technology. Drivas P. J. and Shair J. H. (1974) Probing the air flow within the wake downwind of a building by means of a tracer technique. Atmospheric Enuironmenl 8, 1165-1175. Fackrell J. E. (1984) Parameters characterising dispersion in the near wake of buildings. J. Wind Engng ind. Aerodyn. 16, 97-118. Huber A. H. and Snyder W. H. (1982) Wind tunnel investigation of a rectangular-shaped building on dispersion of effluents from short adjacent stacks. Atmospheric Enoironmerit 16, 2837-2848. Jones C. D. and Griffiths R. F. (1984) Full-scale experiments on dispersion around an isolated building using an ionised air tracer technique with very short averaging time. Atmospheric Environment 18, 903-916. Ogawa Y. and Oikawa S. (1982) A field investigation of the Bow and diffusion around a model cube. Atmospheric Environment 16, 207-222. Vincent J. H. (1977) Model experiments on the nature of air pollution transport near buildings. Atmospheric Enuironmerit 11, 765-774. Wedding J. B., Lombardi J. and Cermak J. E. (1977) A wind tunnel study of gaseous pollutants in city street canyons. Colorado State University, June 1977. Wilson D. J. (1971) Turbulent dispersion in atmospheric shear flow and its wind tunnel simulation. Von Karman Institute for fluid dynamics, Technical note 76, August 1971.