wind-Tunnel Measurements of Flow Fields in the Vicinity of Buildings
(U.S.) Environmental Protect ion Agency, Research Triangle Park, NC
17 Aug 93
V
U.S. DEPARTMENT OF COMMERCE
National Technical Information Service
-------
EPA/600/Ar-93/230
WINBSWHS
8/17/93
WIND-TUNNEL MEASUREMENTS OF FLOW FIELDS
IN THE VICINITY OF BUILDINGS
William H. Snyder*
and
Robert E. Lawson, Jr.*
Atmospheric Sciences Modeling Division
Air Resources Laboratory
National Oceanic and Atmospheric Administration
Research Triangle Park, NC 27711
1 jNTrODUCTION
In order to develop physically realistic models that predict me behavior of pollutants
released in the vicinity of buildings, an understanding of the flow field is essential. The main
features of such flow fields around isolated block-shaped buildings are reasonably well
understood (Hosker, 1984). Separation of the flow generally occurs at the leading edges of the
roofs and sides of the buildings and these separated layers move into the surrounding fluid. If
the building is sufficiently long, these separated layers may reattach onto the surface, so that
separation will occur again at the downwind edges of the roof and sides. Whether the building
is long or short, these separated layers will eventually curve inward toward the wake axis,
forming a rather imprecisely defined region called a "cavity". It is bounded upwind and above
by the separation streamline emanating from the roof edge, and downwind by a reattachment
streamline. Unlike two-dimensional flows, the separation streamline is not the same as the
reattachment streamline (see Hunt et al, 1978, as well as later discussion). The "cavity* is also
bounded laterally by the streamlines emanating from the corners. Within this roughly
ellipsoidal-shape cavity, the flow is of exceptionally high turbulence intensity and small mean
velocity, and frequently reverses direction.
Because of the shear in the approaching atmospheric boundary layer, a stagnation point
will appear well-above ground on the upstream building face, with upward flow on the surface
above the point and downward flow below it. An associated vortex will thus be formed at the
upwind base of the building; it is forced around the sides of the structure and trails off
downwind. Because of its shape as viewed from above, it is frequently referred to as a
horseshoe vortex.
Some of the gross features of the flow fields, such as the length of the cavity, have been
parameterized, primarily from flow visualization studies (see, for example, Hosker, 1984). With
the notable exceptions of works by Castro and Robins (1977) and Davies et al (1980), few
detailed flow-field measurements have been made, primarily because of the difficulty in dealing
* On assignment to the Atmospheric Research and Exposure Assessment Laboratory, U.S.
Environmental Protection Agency.
-------
with the reversing flow fields in the wakes. Much remains to be understood and quantified.
The present experiments take advantage of symmetry to examine the flow fields in vertical
centerplanes as the building dimensions are systematically varied. Although both mean velocity
and turbulent fluctuations were measured during these experiments, we concentrate in the present
report on the mean flow fields; we show how, with the wind perpendicular to a building face,
the mean streamline patterns change as the length, width, and height of a building are
systematically changed. A cubical building is also rotated 45® to examine changes in the
streamline pattern.
2. EXPERIMENTAL DETAILS
The experiments were conducted in the EPA Meteorological Wind Tunnel (Snyder,
1979). The "buildings" were rectangular-shaped blocks which were immersed in a simulated
atmospheric boundary layer that was generated using the Irwin (1981) system of "spires" and
roughness on the floor downwind. This combination produced a 2-m deep boundary layer with
a roughness length of 1 mm. The standard of reference was t cubical building with dimensions
of 200 mm on each side. Four series of measurements were made. In the first, the crosswind
dimension of the building was increased to 2, 4 and 10 times that of the cube. In the second
series, the flow fields were measured behind buildings with along-wind dimensions of 0.015,
0.5, 1, 2 and 4 times that of the cube. In the third series, the height of the building was
increased to 2 and 3 times that of the cube. Finally, the cube was rotated 45°.
A pulsed-wire anemometer (PWA) was used for the velocity measurements. This
instrument is superior to the hot-wire anemometer for use in flows of very high turbulence
intensities and, especially, in reversing flows (Bradbury and Castro, 1971). It is less suited for
low intensity flows, especially for measuring components perpendicular to the mean flow vector.
The basic principle of operation of the PWA is that of the measurement of the transit time of
a heal pulse from a central wire to either of two sensor wires, one located upstream, the other
downstream, of the central pulsed wire.
The PWA probe was oriented to measure the velocity components (one at a time) in the
longitudinal and vertical directions. Measurements were made at approximately 300 points in
the vertical centerplane both upwind and downwind of the buildings for each case.
3. RESULTS
Mean velocity and turbulence intensity profiles of the simulated atmospheric boundary
layer in the vicinity of the model buildings (but in their absence) were measured with hot-wire
anemometry and are shown in Figure 1. The boundary layer was approximately 1.8 m deep ano
may be characterized reasonably well by a power-law profile with an exponent of 0.16. The
roughness length z, and friction velocity uJUR were found to be 1 mm and 0.05, respectively.
At an assumed scale ratio of 200:1, these parameters correspond to a full-scale boundary layer
typical of rural terrain with shrubs and small trees.
Turbulence intensity profiles are also shown in Figure 1, where they are compared with
bounds suggested by ESDU (1972, 1974) with full-scale roughness lengths between 5 and 50
cm. Our data, corresponding to a full-scale roughness length of 20 cm, generally fit within the
-------
bounds suggested by ESDU.
Figure 2 shows the mean streamline patterns deduced from the mean-velocity
measurements in the centerplane for the first series of experiments, where the only parameter
varied was the crosswind width (W) of the building. These streamlines were generated using
a commercially available program TEC PLOT, where a predictor-corrector algorithm is used to
move a point in small steps in the direction of the local velocity field. (We have assumed that
the crosswind component of mean velocity is zero on this plane of symmetry.)
The main features of upstream stagnation point, separation and reattachment streamlines,
and. "cavity" are immediately apparent in Figure 2. The cavity size obviously increases as the
crosswind width of the building increases, but other aspects of the flow field change markedly
also. The location of the stagnation point cm the upwind face of the building appears to move
only slightly upwards from its cube height of approximately 2H/3, but the far upstream elevation
of the stagnation streamline changes continuously from about 2H/3 for the cube to essentially
ground level for the building with crosswind width of 10H. The streamlines upstream of the
buildings thus slope much more prominently upwards as the building width is increased. The
horseshoe vortex is barely perceptible upwind of the cube, but grows in size as the crosswind
width of the building is increased. At W = 10H, its diameter appears to be about H/2.
The implications of the above flow fields on plume behavior should be obvious. Low
plumes from sources located upwind of rather narrow buildings are quite likely to impinge
directly on the upwind building faces, whereas those from sources located upwind of wider
buildings are much more likely to be lifted over the top, with perhaps only the lower edges of
the plumes diffusing to the building surface.
That the flow separates from the upwind edge of the roof is apparent in all cases shown
in Figure 2. In the case of the cube, this separation streamline clearly reattaches to the roof,
as was also evidenced by Castro and Robins (1977). This reattachment is followed immediately
by a horizontal separation from the downwind roof edge, and the cavity height appears to be
constrained to be the same as the building height. For the wider buildings, however, whereas
the initial separation streamline appears to reattach to the roof, a horizontal separation r the
downwind roof edge does not exist. Instead, the cavity grows in height and the associated
upward velocities on the lee face of the building appear to predominate, with separation of the
flow progressing up the lee ouilding face. This is obviously associated with the much stronger
vertical velocities in the cases of the wider buildings. The cavity height grows from about H
in the case of the cube to about 3H/2 in the case W = 10H.
The length of the cavity (from the lee face of the building to the reattachment point)
varies from 1.4H for the cube to 5.6H when W = 10H. These values agree quite well (within
about 10%) with Hosker's (1984) equation for the cavity length where reattachment of the flow
on the roof w%s observed. Another point to note for the widest building is the formation of a
secondary vortex at the downwind base.
An important point to note here is that these streamline patterns differ qualitatively from
those described by Hunt et al (1978). They suggest that in the centerplane of a three-
dimensional flow, a streamline originates upstream and attaches to the surface downwind (see
Figure 3a); streamlines below this one, then, spiral into the node N, so that the flow is laterally
outward in the y-direction at N. Thus, N is a separation point. Our measurements suggest that
the flow is laterally inward in the y-direction at N, so that it is an attachment point. The flow
coming onto this centerplane at N, then, spirals outward, forming the attachment point at S on
the ground surface (see Figure 3b). The streamline attaching to the surface does not originate
-------
from upwind, but rather from the node in the centerplane. This topological structure appears
more consistent with the streamline patterns presented by Davies et at (1980), which were also
derived from pulsed-wire measurements for a tall building.
Figure 4 shows how the streamline patterns change as we vary the along-wind length (L)
of the building. The upstream patterns appear to be completely independent of L, The cavity
height is a maximum (of about 1.4H) when L = 0.015 (square flat plate), since reattachment
on the roof obviously cannot occur. When L = H/2, the cavity height is reduced to about
1.15H; for L s» H, reattachment occurs on the roof, horizontal separation follows at the
downwind roof edge, and the cavity height is constrained to be the same as the building height.
Correspondingly, the cavity length (measured from the rear building face) decreases from a
value of 2.3H for the to plate to 1.5H when L = H/2. For L a H, the cavity length is nearly
constant with a value near 1.3H. Finally, in the far wake of the short building (L < H), the
streamlines are observed to descend quite rapidly, whereas they are more nearly horizontal
downwind of the longer buildings.
Figure S shows how the streamline patterns change as we vary the building height. The
elevation of the stagnation point on the upwind face of the building remains at approximately
2H/3, and the streamlines upstream of about l.SW are essentially horizontal. According to
Corke and Nagib (1976), the height of this stagnation point is rather strongly dependent on the
exponent of the power law describing the wind profile, and our value appears to agree quite well
with their observations. The streamline pattern above the building is largely independent of
building height; in all cases, the flow reattaches to the building roof, then separates again at the
downwind edge of the roof. Perhaps surprisingly, the cavity length is independent of the
building height, but for a tall building, the distance to reattachment should obviously be more
closely linked to the width of the building rather than to its height. The streamline patterns
presented here display a fr?e stagnation point (denoted by S), as was also shown by Davies et
al (1980) through pulsed-wiie measurements in the wake of a building with height 6 times its
length and width.
The case of flow approaching the cube at 45" is shown in Figure 6. This pattern displays
the qualitative features described by Castro and Robins (1977) and others, namely, that the
horseshoe vortex is less prominent and that downwash is much stronger in the wake. Although
not evident from this centerplane pattern, this flow is dominated by the delta-w ing-type vortices
generated by the swept-back leading edges.
SUMMARY
A pulsed-wire anemometer was used to measure the flow fields in the vicinity of a variety
of rectangular-shaped model buildings immersed in the simulated atmospheric boundary layer
of a wind tunnel. The crosswind width, height, and along-wind length of the building were
systematically varied, and the longitudinal and vertical components of the velocity fields were
measured in the plane of symmetry (centerplane). Measurements were also made with the
cubical building rotated to 45°. These measurements were used to deduce the streamline patterns
and, hence, to identify and quantify important features of the flow fields.
The location of the stagnation point on the upwind building face was found to be 2H/3,
practically independent of the crosswind width or along-wind length of the building. The far-
upstream height of the stagnation streamline decreased with an increase in the crosswind width
-------
of the building; when W « 10H, its height was essentially ground level. A horseshoe vortex
appeared at the upwind base of the building; it became more prominent as the crosswind building
width was Increased. The flow always separated at the upwind edge of the roof. It reattached
on the roof when L ^ H. The "cavity" length and height grew as the crosswind width of the
building increased. The cavity length was observed to be independent of building height and,
for L ^ H, independent of L. Our measurements suggest that the structure of the streamline
patterns in the wake is qualitatively different from that described by Hunt et ol (1978). The flow
spirals out of a node located inside the "cavity" and reattaches to the surface.
This data set should prove useful to the mathematician attempting to develop physically
realistic models that predict down wash of pollutants released in the vicinity of buildings.
REFERENCES
Bradbury, L.I.S. & Castro, I.P. 1971 A Pulsed-wire Technique for Velocity Measurements in
Highly Turbulent Flows. J. Fluid Mech. 49, 657-691.
Castro, LP. & Robins, A.G. 1977 The Flow Around a Surface-Mounted Cube in Uniform and
Turbulent Streams. J. Fluid Mech., 79, 307-335.
Corke, T.C. & Nagib, H .M. 1976 Sensitivity of Flow Around and Pressures on a Building
Model to Changes in Simulated Atmospheric Surface Layer Characteristics. IIT Fluids & Heat
Trans. Rpt. R76-1, 111. Inst. Tech., Chicago, IL, 291p.
Davies, M.E., Quincey, V.G. & Tindall, S.I. 1980 The Near-Wake of a Tall Building Block
in Uniform and Turbulent Flows. Proc. 5th Int. Conf. Wind Engr,, Fort Collins, CO, July,
1979 (J.E. Cermak, ed.), v. 1, p. 289-298. Pergamon Press, NY, NY.
ESDU 1972 Characteristics of Wind Speed in the Lower Layers of the Atmosphere near the
Ground: Strong Winds (Neutral Atmosphere). Item No. 72026, Engineering Sciences Data Unit,
London, UK.
ESDU 1974 Characteristics of Atmospheric Turbulence near Ground. Part II: Single Point Data
for Strong Winds (Neutral Atmosphere). Item No. 74031, Engineering Sciences Data Unit,
London, UK.
Hosker, R.P. Jr. 1984 Flow and Diffusion Near Obstacles. Atmospheric Science & Power
Production, DOE/TIC-27601, D. Randerson, Ed., Ch. 7, 241-326. U.S. Dept. of Energy,
Wash., DC.
Hunt, J.C.R., Abell, C.I., Peterka, J.A. & Woo, H. 1978 Kinematical Studies of the Flows
Around Free or Surface-Mounted Obstacles: Applying Topology to Flow Visualization. J. Fluid
Mech., 86, 179-200.
Irwin, H.P.A.H. 1981 The Design of Spires for Wind Simulation. J. Wind Engr. Indus.
Aerodyn., 7, 361-366.
-------
Snyder, W.H. 1979 The EPA Meteorological Wind Tunnel: Its Design, Construction, and
Operating Characteristics. Rpt. No. EPA-600/4-79-051, Envir. Prot. Agcy., Res. Tri. Pk., NC,
78p.
DISCLAIMER: This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for presentation and
publication. Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
-------
A EPRRV.003 (2, 1) EPR, VERT,SP0-4»/S, TACH-412,BLDG-NONE,X-e0e2Mi,y.0,XUIRe,H354uu
04-13-92 leees'S FOR 60S,1000Hz FILTER,244e« Irwin spires,HONEYCOMB
~ EPRRV.002 <2,1> EPR, VERT,SPOMa/S, TACH"412, BLDG*NONE, X*9831aa,Y"0# XUIRE,H354uu
84-13-92 ieees/S FOR 60S,lB00Hz FILTER,244e» Irwin *pir««, HONEYCOMB
O EPRRV.801 <2,1) EPR,VERT,SPO*4«/S,TACH-412,BLDG-NONE.X-l1680m,Y.e,XWIRE,K354uu
04-13-92 1000S/S FOR 60S.1000H* FILTER,244cm Irwin splres,HO»CYCOnB
O EPRRV.007 (2,1) EPR, VERT, SPO-4^S,TACH-412,BLDG-NONE,X-11080m, V-0, XUIRE, H3S4uu
04-14-92 iMi/l FOR 60s, 1000Hz FILTER, 244c. Irwin spires,HONEYO*»,top of BL
V EPRRV.004 (2,1) EPR,VERT,SPO-4»^,TftCH-412,BLDG-NONE,X-15318m, Y-0,XWIRE,H354uu
04-13-92 10004/-S FOR 60S,1000Hz FILTER,244c* Irwin *Pir«s,HONEYCOMB
KM^U.001 (2,1) U. i ¦ ¦. (¦ ^agggnep.ff
0.4 0.8 1.2 1.6 2
0
2.4 2.8 3.2 3.6
4
4.4
U, m/%
(a) Mean velocity profiles
Figure 1. Simulated at*ospheric boundary layer structure.
-------
EPRRV.003 (10,1)
04-13-91
EPRRV.002 (10,1)
g4.43.f2
EPRRV.001 (10,1)
84-13-92
EPRRV.007 (10,1)
04-14-92
EPRRV.004 (10,1)
04-13-92
-ESOU.001 (2,1)
00-13-91
EPR, VERT, SP0-4a'S, TACH-412, BLOS-NONE, X«8082aa, Y«0, XUIRE, H3S4uy
1000s'S FOR 60S, 1000Hz FILTER,244ca Irwin spires,HONEYCOMB
EPR,VERT,SP0«4a^S,TACH-412,BLOG-NONE,X»9831aa,¥>0,XUIRE,K3S4uu
1000S/S FOR 60S,1000Hz FILTER,244c« Iruin spires,HONEYCOMB
EPR,VERT,SP0-4a/S,TACH-412,BLOG-NONE,X-11660aa,Y-0,XUIRE,H3S4uu
1000S/S FOR 60S,1000Hz FILTER,244ca Iruin spires,HONEYCOMB
EPR,VERT,SP0«4a/S,TACH«412,BL06-N0NE,X-11660aa,Y-0,XUIRE,H3S4uu
1000S/S FOR 80s,1000Hz FILTER,244ca Iruin spires,HONEYCOMB,top of
EPR, VERT, SPO"4«^S, TACH-412, &.D6-N0NE, X«15318aa.Y«0, XUIRE, H3S4uw
1000s^S FOR 60S,1000Hz FILTER,244ca Iruin spires,HC»€YCOfe
BL
ESOU turbulence intensity profiles - lower liait - zo ¦ 0.05 to 0.5 a
-ESOU.002 (2,1)
08"13"91 =¦&£
1.000,
iwm AAt SHMk
• V0O| tWAf »«V«
intensity profiles - upper 111
zo
0.5 a
f
N
1800 ¦
1600
1400
1200
1000
600
400 -
200
1 ""I" <
im
¦ 111
"t T"
I I I
m
ma*
Bounds suggested by ESOU
| I I I | I .J
-------
£ 4C0
e
^ 200
«§¦
500
0
400
800
600
x, ni"
Figure 2, Streamline patterns around buildings of various crosswind widths.
Number on building is W/H. L = H.
-------
WEPRSTOM.DRW
(a)
S
(W
S
Figure 3.
Sketches of basic flow structure as (a) viewed by Hunt et at (1978),
where the node N is a separation point, and (b) suggested by current
measurements, where the node N is an attachment point.
-------
400
^ 200
400
400
800
1200
500
0
1 BOG
x, mm
E
E
^ 200
0.5
600
400
400
0
800
1200
1600
x, mm
-600 -4C0 0 400 800 1200 1600
x, mm
^ 200
'600
800
400
0
SCO
x, mm
*500
800
200
SO
0
1
Figure 4. Streamline patterns around buildings of various along-wmd lengths.
Number on building is L/H. W = H.
-------
*> r\
£ WsJ
2 w w
Figure 5. Streamline patterns around buildings ol various heights.
Number on building is H/W. L = W.
-------
400
^ 200
400
400
800
1200
1600
0
x, mm
Figure 6. Streamline patterns around a cubical building at 90° Mid 45° to the wind.
-------
TECHNICAL REPORT DATA
(Ann rtad Inamctions on the rtvem before compl
1. REPORT HO. ft
EPA/600/A-93/2W
«. rrai amo mmttiu
ft REPORT MTt
1993
Wind-Tunnel Measurements of Flow Fields in the Vicinity of
Ruilriifiss
ft PERFOMMM OfttAMXAHON COOC
7. AUTHOR**)
William H. Snyder* and Robert E. Lawson, Jr.*
Environmental Protection Agency, Research Triangle Park, NC
& M9VF0MHNCI OMMM
CATION REPORT NO.
•l PCItfOMMNO OMAMBAHOM NAUt AMO AOONKSft
Atmospheric Research & Exposure Assessment Laboratory
Ml MOOIIAM EUMENT NO.
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
11. COMTHACT/MAMT NO.
12. IMMONM AOEMCV HAMC AMO AOOMCM
Atmospheric Research A Exposure Assessment Laboratory-RTP, NC
u twc of mum amo moo covered
Presentation and Preprint Volume
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
14 tPOMSOMMCI "TfT*
EPA/600/09
11 SUPPUMENTANV MOTE*
Preprint Vol: Eighth Joint Conf. on Appl. of Air Pollution Meteor., Jan. 1994, Nashville, TN.
*On assignment from the National Oceanic A Atmos. Admin., US Dept. of Commerce.
It. ABSTRACT
Pulsed-wire anemometer measurements have been made in the vicinity of rectangular shaped
model buildings immersed in the simulated atmospheric boundary layer of a wind tunnel.
The primary purpose of the measurements was to delineate the size and shape of the
"cavity* as a function of the building dimensions. The crosswind width, the height and
the along-wind length of the building were systematically varied, and the longitudinal
and vertical components of the velocity fields were measured in the plane of symmetry
(centerplane). For one case, all three velocity components were measured in the full
three-dimensional space surrounding the building. These measurement , were used to deduce
the streamline patterns and, hence, the height and downwind extent of the recirculation
zones or cavities and other important features of the fiow fields.
IT.
KEY WORM ANO DOCUMENT ANALYSIS
4. descriptors
ft IOENTIFIERS/OPEN ENDED TERMS
«. COSAT1 FtaM/Olw*
11 MTNMITIOII STATEMENT
RELEASE TO PUBLIC
It. SECURITY CUM (Tkit Jbfmt)
UNCLASSIFIED
SI. NO. Of
20. SECURITY CLASS (Thlt
UNCLASSIFIED
Sft PRICE
EPA Form 2220-1 (Rtv. 4-77) mmow rami ¦ onouii put
------- |