Skip to main content

Full text of "NASA Technical Reports Server (NTRS) 19750009309: An investigation of rooftop STOL port aerodynamics"

See other formats

NASA CR-1 32570 


Jniv. ) 32 p HC 44.75 C3CL C 12 


03/v5 12191 


By Jeffrey N. Blanton and Hermon M. Parker 

Distribution of this report is provided in the interest of 
information exchange. Responsibility for the contents 
resides in the author or organization that prepared it. 

Prepared under Grant No. NGR-47-005-146 by 
Charlottesville, Virginia 




An Investigation into aerodynamic problems associated with large building 
rooftop STOLports has been performed. Initially, a qualitative flow visualiza- 
tion study indicated two essential problems: (l) the establishment of smooth, 

steady, attached flow over the rooftop, and (2) the generation of acceptable 
crosswind profile once (I) has been achieved. This study indicated that (1) 
could be achieved by attaching circular-arc rounded edge extensions to the 
upper edges of the building and that crosswind profiles could be modified by 
the addition of porous vertical fences to the lateral edges of the rooftop. 
Important fence parameters associated with crosswind alteration were found to 
be solidity, fence element number and spacing. Large scale building induced 
velocity fluctuations were discovered for most configurations tested and a 
possible explanation for their occurrence was postulated. Finally, a simple 
equation relating fence solidity to the resulting velocity profile was developed 
and tested for non-u.iiform single element fences with 30 percent maximum solidity. 


Jeffery N. Blanton and Hermon M. Parker 
University of Virginia 


An investigation into aerodynamic problems associated with large building 
rooftop STOLports has been performed. A qualitative flow visualization program 
carried out at the Langley Research Center indicated two essential problems: 

(1) the establishment of smooth, steady, attache flow over the rooftop; and, 

(2) the generation of an acceptable crosswind profile once (I) has been 
achieved. It was found that the configuration of the building itself determines 
the gross nature of the flow over the rooftop and that relatively smooth 
attached flow could be achieved by the addition of circular-arc rounded edge 
extensions to the upper edges of the building. It was further determined that 
crosswind profiles could be altered by the addition of vertical fences along 
the lateral edges of the building rooftop. 

A more quantitative program was undertaken at the University of Virginia. 
Effects of various fence parameters such as screen wire and mesh size, number 
of elements, etc., were investigated. Also a brief study of building induced 
turbulence was carried out and a qualitative description of the generation of 
the turbulence was hypothesized. 

In an effort to isolate fence effects from building effects, a study of 
simple one element fences in the absence of a building model showed that 
downstream flow velocity was reduced over a height equal to that of the 
fence. In contrast to this, a fence installed on a building model affects 
the downstream flow up to approximately two fence heights. Finally, a simple 
equation relating the velocity at any height behind a one element fence to 
the fence solidity (ratio of projected wire area to projected total area) 
was derived and validated for low solidity fences (up to 0.30 area blockage). 


During recent years, the rapid increase in the number of air passenger 
miles traveled has imposed considerable hardships on contemporary air transpor- 
tation systems. Both airports and airspace are congested to the extent that 
some immediate relief is needed. This is particularly true of short-haul 
operations where times required for ground transportation to and from the 
terminal areas often compare in magnitude to flight time. With the development 
of STOL aircraft has come the possibility of supplying some needed relief. 

Many variations of possible STOL transportation systems have been proposed, and 
many points of view have been expressed concerning the advantages and disadvan- 
tages of various components of these systems. One of the few issues upon which 
there is agreement is that the STOL passenger would definitely be subjected to 
conditions quite different from those to which he has been accustomed in 
traveling between major cities. The uncertainty concerning the acceptability 

of these factors to the traveling public has undoubtedly been an important 
factor contributing to the hesitation which has been evident in making a 
commitment to implement a STOL system. 

In order to help alleviate some of these uncertainties, a program was 
initiated at the University of Virginia, through its Center for the Application 
of Science and Engineering to Public Affairs, to develop modeling techniques to 
predict human acceptance of proposed STOL systems and to develop acceptability 
criteria for the many variables and their trade-off possibilities. This 
requires a thorough understanding of the nature of the variables and a quanti- 
tative formulation of response to them. 

The work reported herein represents one facet of this overall program 
which arose *rom the following considerations. STOLcraft can be accommodated 
in relatively small terminal areas; it is, therefore, possible that STOLports 
can be placed nearer the population centers of cities. Such an arrangement 
would have two major advantages. First, it would improve short-haul air trans- 
portation directly by decreasing ground travel time and consequent total trip 
time. Second, it would aid long-naul air transportation by relieving congestion 
at conventional airports currently caused by short-haul operations. 

One means of placing STOLports near city centers is to utilize the roofs 
of very large downtown buildings as landing and takeoff areas. This would 
minimize disruption of normal downtown economic activities since the building 
itself could be used for varied revenue producing activities, as well as for 
the STOL terminal. An artist’s conception of such a STOLport is shown in 
figure 1. There are, however, many types of problems associated with the 
development of such an unconventional airport, and the present work is directed 
toward those related to the quality of the airflow over such a structure. There 
are two important issues involved. One is the nature of the local flow near the 
landing surface induced by the design of the structure. This will have a direct 
influence on the motion experienced by the aircraft (and its passengers) in the 
landing and takeoff maneuvers - as well as in the safety of these activities. 

The other arises from the fact that the need to restrict the STOLport to a single 
runway will generally give rise to crosswind problems which could effect the 
quality and safety of the landings and takeoffs and the reliability of the 
system operation. Thus, the possibility of alleviating crosswind problems and 
generally improving flow quality through the use of wind screens and proper 
building design was deemed worthy of study. 



arc length of rounded edge extensions, cm (In.) 
building model width, cm (In.) 

screen drag coefficient based on average Interstitial velocity and 
blocked area (see eq. A-4) 

building model height, cm (In.) 

fence height, cm (in.) 

height of measurement above model, cm (in.) 

local drag coefficient of a variable solidity screen 

total pressure drop across screen or total pressure drop from far 
upstream to far downstream, N/m 2 (lb/ft 2 ) 

2 2 

static pressure far upstream of a screen, N/m (lb/ft ) 

static pressure far downstream of a screen, N/m 2 (lb/ft 2 ) 

2 2 

dynamic pressure, N/m (lb/ft ) 
pv cw b 

Reynolds number, — ^ — 

distance between upstream and downstream fences, cm (In.) 
local velocity in flow field, m/sec (ft/sec) 
free-stream velocity, m/sec (ft/sec) 

aircraft air speed, kts 

V cw crosswind velocity, m/sec/kts (ft/sec) 

V local crosswind velocity* m/sec (ft/sec) 


V ru maximum crosswind velocity, m/sec (ft/sec) 

c max 

Vg aircraft ground speed, kts 

x horizontal distance measured from fence, cm (in.) 

y vertical distance measured from bottom of screen, cm (in.) 




vertical distance measured on the screen, cm (in.) 
density of air, gm/cm (slugs/ft ) 
viscosity of air. Nsec/m* (slug/ft-sec) 
local turbulence. 

rms-veloci ty fluctuation 
mean velocity 

£ local solidity of screen * projected blocked area/projected total area 


The initial study into the feasibility of utilizing large building roof 
tops as elevated STOLports was begun in November 1969, in a joint effort by 
the University of Virginia, Department of Aerospace Engineering and 
Engineering Physics, and the Langley Research Center, Low Speed Vehicles 
Branch. Initial studies were of a qualitative nature and were performed in 
the 17-foot test section of the 300-mph 7-by 10-foot wind tunnel at the 
Langley Research Center. The purpose of this program was the qualitative 
observation of the overall flow field associated with a rooftop STOLport and 
the influence of the building conf igurat ion changes on that flow. 


Test Set Up 

The primary experimental technique utilized in this initial investigation 
was flow visualization achieved by the use of smoke and tufts. Figure 2 is a 
photograph of one of the configurations tested, and figure 3 shows cross 
sections of six of the configurations tested. The first configuration in 
figure 3 is the solid rectangular parallelepiped building itself. Next is a 
series of four configurations, each having an elevated rooftop (or raised deck). 
The first in this series is the basic raised deck. The second has side ramps, 
which were tried because they offered a method of lateral containment for land- 
ing STOlcraft. A circular arc side overhang was also tried. (See third 
configuration in the series.) Finally, porous side fences were added to the 
raised deck with curved overhang in an attempt to modify the crosswind velocity 
profiles; each side fence was constructed of three rather closely-spaced pieces 
of stainless steel screening of different heights. Figure 2 is photograph 
of this fourth elevated-rooftop configuration with its curved overhang and 
porous side fences. The last configuration in figure 3 differs from the 
preceding one in that the space under the raised deck is closed in order to 
determine whether the raised deck is essential in achieving the desired flow 

The building models are 1.2 meters by Ij. meters (4 ft by 8 ft) by either 
4) cm or SI cm (16 in or 20 in) high, depending on the configuration. If the 
full-scale building is 1 52 meters (500 ft) wide, then the IS cm (6 in) height 
of the model fences corresponds to an actual fence height of 19 meters (62. 5 ft) 
and the full-scale building would be SI or 63 meters high (167 or 208 ft high). 
The building model was mounted so that it could be placed at any yaw angle to 
the tunnel flow; in figure 2, the tunnel flow is parallel to the model runway. 

Qualitative Test Results 

A motion-picture film supplement showing smoke flow over the various 
building models has been prepared. * Six frames from this film are presented 
as figures 4 through 9. Figures 4 through 7 show the flow over the four con- 
figurations in the elevated-rooftop series. Figure 8 shows the flow over the 
solid rectangular para) lelepiped building alone, and figure 9 is of the solid 
building with rounded edges but not fences. In each case, the flow is from 
left to right and the tunnel flow is at 90* to the runway (pure crosswind). 

Figure 4 shows the flow over the model with the basic deck. (See the 
first schematic in the elevated-rooftop series in figure 3.) The f’ow is not 
good. It is separated at the leading edge, and reattachment does not occur. 
Over the entire rooftop, the flow is highly turbulent and unsteady. 

“This film supplement (16 mu, 23 min, color, narrated) may be obtained or 
loan by requesting film serial number L-1114 entitled "Roof-Top STOL/Port 
Flow Visualization" from NASA Langley Research Center, Attn: Photographic 

Branch, Mail Stop 171, Hampton, Virginia 23665. 


Flow over the second configuration in the elevated-rooftop series is 
shown in figure 5. The addition of side ramps to the basic raised deck is 
seen to be detrimental. The separation angle is larger. In the center region, 
the flow is even more turbulent and unsteady than that over the basic rai-ed 
deck. The unsteady flow region extends very high above the model. This con- 
figuration has the worst flow of any of those tested. 

Figure 6 shows the flow over the third configuration in the elevated-roof- 
top series. The effect of adding rounded edges to the raised deck is dramatic. 
Some turbulence can be seen, but the flow pattern Is considerably better than 
for the first two elevated-rooftop configurations. 

Once success in maintaining attached flow had been achieved by the addition 
of rounded edges, fences were added to determine whether the flow would remain 
smooth and whether the crosswinJ profile above the deck would change. The flow 
over the resulting configuration (fourth in the elevated-rooftop series in fig. 
3) is shown in figure 7. One must not confuse the spreading oi the smoke due 
to turbulence and unsteadiness with the spreading due to low velocities. The 
lazy drifting action of the smoke in this figure indicates low velocities. 

Figure 8 shows that the flow over the solid parallelepipeo building alone 
is separated and turbulent. In fact, it seems to be slightly worse than the 
flow over the basic-raised-deck configuration (fig. 4). 

Again, the effect of adding rounded edges is dramatic. Figure 9 shows 
that over the solid building with rounded edges the flow is attached and fairly 
smooth and steady everywhere; it seems to be as smooth as the flow over the 
raised-deck configuration with rounded edges (fig. 6). At this stage of the 
study, it was concluded that the raised deck is not necessary to obtain good 
attached flow but that the rounded edges are the important factor. However, 
later investigations showed that the elevated deck may be desirable. This 
will be discussed in a later section. Although not illustrated, it should be 
noted that the addition of fences to the solid building with rounded edges 
produced favorable results. The flow remained smooth and the crosswind veloci- 
ties near the building rooftop were greatly reduced. Most of the subsequent 
work has been without raised decks on the models. 

This first qual : t3tive phase of the research effort gave encouragement to 
the expectation that smooth, attached flow over an elevated STOLport could be 
achieved and that the crosswind profile could be modified in some desirable 


The next phase of the study was a souk / hat more detailed experimental in- 
vestigation carried out at the University of Virginia. Results concerning the 
influence of fence structure and geometry on changes in crosswind velocity are 



Figure 10 shows schematically the type of effect one expects fences on 
either side of a STOLport runway to have on the crosswind velocity profile. 

The left-hand sketch shows the tunnel flow with the tunnel boundary layer. In 
the right-hand sketch, the dashed line indicates the profile on the STOLport 
model at the centerline with no fences. With fences, the profile is modified 
as shown by the solid curve. The curves in this figure are just Illustrative. 

The effects of a crosswind velocity on an aircraft attitude and airspeed 
can be seen in figure 11, which is a schematic diagram of a STOLcraft, designed 
to land at an airspeed of 60 knots, landing at a STOLport between fences in a 
30-knot crosswind. The right-hand side of the figure indicates the situation 
initially in the full 30-knot crosswind. The airspeed is 67 knots; the ground 
speed is 60 knots; and the crab angle is 27 . The left-hand side indicates the 
situation at a pi ace where the fences have reduced the crosswind speed to IS 
knots, with the assumption of perfect decrabbing and a constant ground speed of 
60 knots. The airspeed is now 62 knots, and the crab angle is 14 . The effect 
of crosswind angles of other than 90 could be examined in a similar fashion. 
Some concern had been expressed about the potential loss of airspeed as the 
crosswind component is removed. It can be seen from figure II that even when 
the relatively large crosswind of 30 knots (50 percent of ground speed) is 
decreased by 50 percent, the airspeed is only decreased by about 7 5 percent. 
Consequently, it seems that loss of airspeed is probably not a sufficient reason 
for not seeking to control crosswind profiles. 

It seems obvious that if a pilot had to land such a STOLcraft an an 
elevated STOLport in a crosswind, he would be concerned about such factors as 
the following: Mt what height does the crosswind component begin to change? 

How rapidly does it change? At what height above the runway does the crosswind 
become negligible? 

It should be noted that the results to be presented herein are not given 
as a "solution" to the crosswind problem. With a given, unfenced STOLport 
building and a given flow over it, there will be a certain crosswind velocity 
profile provided by nature. If this profile is acceptable to all concerned, 
there is no problem. If it is not acceptable, then some mechanism of making the 
crosswind velocity profile acceptable becomes of interest. The question to 
which the present work is addressed is simply this: What are the possibilities 

of crosswind profile modification by putting fences on either side of the STOL- 
port building? 

Crosswind Investigation 

Figure 12 shows the test setup for a building model that Is 76 cm (30 in) 
wide and 15 cm (6 in) high. The rounded edges of the model are circular and of 
7 cm (2.75 in). If the 76 -cm (30-in) model width corresponds to a 152-meter 
(500-ft) full-scale building width, then the 15-cm (6-in) model height corres- 
ponds to a 30.5-meter (100-ft) building height. Also, the 7-cm (2.75-*n) model 
overhang radius corresponds to a 14-meter (46-ft) building overhang radius, and 
the 7.6-cm (3-in) rr,odel fence height corresponds to a 15-meter (50-ft) full- 
scale fence height. The fences were constructed of one, two or three wire 


Figures 13 and 14 show the measured profiles at the model centerline for 
different fence configurations. The dashed line indicates the top of the fences. 
The model is 76 cm (30 in) wide, and all the fence configurations are 7.6 cm 
(3 in) high. It is obvious that increasing the number of screens causes the 
centerline velocities near the deck to decrease and causes the height to which 
the fence effect extends to become larger. It is apparent that fence structure 
has a large influence on centerline crosswind velocity profiles. 

Figure 15 shows results of efforts to generate varied velocity profiles by 
proper fence design. The straight line profile is the result of a separate 
experiment conducted at Langley (ref. 1) where a pair of three screen fences 
were modified in a trial and error manner until a linear profile was produced. 

The other profile was generated at the University of Virginia in an effort to 
show that unusual profiles could be generated. These results indicate that 
centerline profiles are adjustable over a considerable range and suggest that 
one might learn how to predict profiles for given fence structures and 

It is also interesting to examine crosswind velocity profiles at positions 
other than the centerline. The investigations at the University of Virginia 
included experiments to determine the effects of measurement position. Figures 16 
through 21 compare profiles taken 11.4 cm (4.5 in) behind the leading-edge fence 
to profiles taken above the centerline. In general, there is a shift to the 
left and upwards with increasing distance. This indicates a slowing and diffu- 
sion of the flow with distance behind the fence. Figures 22 through 26 show 
variation of velocity profile with distance for different ratios of S/h^. The 

trends shown by these graphs are similar to those seen in figures 16 through 
21. Figure 22 corresponds to the maximum S/h^ tested. It can be seen that 

there is less difference between the 1/2 S and the 3/4 $ profiles than between 
the 1/4 S and the 1/2 S profiles. This seems to indicate that the flow is 
approaching an equilibrium condition. 

In an effort to determine the effect of screen mesh and wire size, similar 
fence configurations were constructed from two different mesh and wire size 
screens. The solidity or ratio of projected »Jire area to projected total are? 
was constant at about 0.30 for each of the screens. Results for similar fence 
configurations are shown on figures 27 through 32, The data taken above the 
centerline show negligible variation with type of screen for the one screen and 
two screen profiles. However, tMt taken at 11.4 cm (4.5 In) behind the screen 
shows appreciable differences. This is in agreement with other similar tests 
performed at the University of Virginia (ref. 2) which indicated that the equi- 
librium profile behind a screen depends only on the solidity of the screen, but 
that the distance to equilibrium depends on the wire and mesh size. The graphs 
also hint that equilibrium distance increases as the effective solidity of the 
fence increases by the addition of more screens. In other words, comparison 
of the three fence configurations tested at the centerline indicates small 
but increasing differences in velocity profiles with increasing number of 


An additional experiment showed that the centerline profile was essential- 
ly the same with and without the downstream fence. Thus, one tentatively 
concludes that the profiles in the runway area are fixed by the upstream fence 
and are nearly independent of the downstream fence. Some additional evidence 
on this point is provided in figure 33. A single two-screen fence was placed 
approximately at the centerline position, and profiles were measured 5 cm (2 in) 
upstream and 5 cm (2 in) downstream of the fence (that is, 0.07b upstream and 
downstream of the fence). The other curve (through the square symbols) is the 
no-fence profile at the fence position. The data clearly show that a fence 
alters the downstream flow velocity very much more than it does the upstream 
velocity. These results provide some insight into the mechanism by which 
fences alter the adjacent upstream and downstream flow fields. 

Velocity Fluctuation Investigation 

Throughout the quantitative test program, Hrge-scale velocity or dynamic 
pressure fluctuations were observed above >. te STOLport deck. These fluctua- 
tions appeared and were similar for all configurations tested. They usually 
died out on the order of 0.5 building heights above the deck. The predominate 
frequency associated with them was approximately one hertz, but slower and 
faster frequencies were superimposed on this. Unfortunately, no spectral 
analysis was carried out due to the lack of proper instrumentation. For most 
configurations, the gross appearance of the fluctuation was invariant over the 
small Reynolds number range investigated (Re = 100,000 to 750,000). Figure 3^ 
shows a typical variation of the fluctuations with height above the building. 
These measurements were obtained using a pitot-static tube placed at different 
heights above the centerline of the model. Approximately ±50 percent fluctua- 
tion in dynamic pressure (±25 percent velocity fluctuation) can be seen in the 
extreme case near the deck. It is realized that because of their relatively 
slow response times pitot-static tubes are not the ideal instruments to use in 
measuring flow unsteadiness. In this application, however, the low frequency 
oscillations are of primary interest since they would have the greatest effect 
on landing aircraft. The response time of the pitot-static tube is sufficient- 
ly fast to observe these low frequency oscillations. 

In an attempt, to determine scale effects, a single series of tests were 
performed on a larger model (one that had been used in the earlier flow visual- 
ization tests) in the Langley full-scale tunnel. The same instrumentation that 
had been used at the University of Virginia was used in this investigation. A 

Reynolds number one full order of magnitude higher (5 x 10^) was achieved. 

Mean velocity profiles and velocity fluctuations were similar to those previous- 
ly obtained at Re - 5 x 10^. Extrapolation of these results to a full-scale 


building with a Reynolds number of 5 x 10 is not possible. Hence the full- 
scale existence of this phenomena has not been verified. However, It can not 
be discounted on the basis of tests completed to date. 

Qualitative description of velocity f l uctuat ion . -One possible explanation 
of the flow fluctuations is the existence of an unstable vortex located in 
front of the building model. This is a f 1 ov pattern similar to that shown In 


figure 35. The flow field is divided into two regions separated by a stagnation 
streamline which is attached to the leading-edge overhang. Adjacent to the deck 
and elsewhere in the upper region the speed is high. The speed below the stag- 
nation streamline, however, is low. Associated with this difference in speed 
is a pressure gradient between regions 1 and 2 in figure 35. Under equilibrium 
conditions, this gradient is just balanced by the curvature of the stagnation 
streamline. If the pressure gradient is raised above this equilibrium con- 
dition, the stagnation streamline would detach as shown in figure 36. This 
would allow a "bubble" of turbulence to escape up over the rounded edge. When 
the bubble is released, the excess pressure gradient is relieved and the stag- 
nation streamline reattaches. This allows the pressure gradient to rise again 
and the cycle is repeated. This heuristic explanation has been substantiated 
by the use of flow visual izat ion. Smoke injected under the rounded edge in 
front of the building escapes in distinct bubbles. It was further hypothesized 
that if the above explanation is true, the velocity fluctuation could be 
eliminated simply by continuously relieving the pressure gradient. With this 
in mind, a building configuration such as that shown in figure 37 was installed 
in the tunnel. It was found that if the second deck is raised high enough, and 
if proper rounded edges are used, the velocity fluctuations can be virtually 
eliminated. Another interesting means of alleviating the fluctuation (although 
it may not be practical in actuality) is shown in figure 38, The wedge in 
front of the building can be adjusted so that the stagnation streamline has a 
stagnation point at A and another at S. With this situation, the stagnation 
streamline is stabilized and the vo*tex is stable. Consequently, the flow 
over the deck is smooth. 

Building induced turbulence measurements . -Since the magnitude and frequency 
of the velocity fluctuations are such that it would be potentially hazardous 
(depending o.. scaling, of course), it was decided that the phenomena should be 
more carefully investigated. Consequently, an experimental program was initia- 
ted to determine important parameters associated with the velocity fluctuation. 
The test set up was similar to that used earlier at the University of Virginia, 
the only difference being the use of hot wire anemometry for velocity 
measurements. Both mean velocity and RMS turbulence were measured with height 
above the deck for various building configurations. This program was not as 
complete as anticipated largely because of difficulties in using hot wires in 
the University of Virginia low-speed tunnel. The primary difficulty being the 
inability to adequately temperature compensate for the large scale temperature 
variations which occurred in this closed tunnel. Turbulence measurements, 
however, are insensitive to this problem, and consequently these data are 

Figures 39 and 41 show the height versus turbulence variation at stations 
above the deck for various rounded edges. A basic building configuration (no 
second elevated deck) was used for these tests. Each of the figures shows 
greater differences between the leading edge and centerline data than between 
the centerline and trailing edge data. This seems to indicate that some equili- 
brium is being approached. Figures 42 and 44 show differences due to 
different rounded edges. With increasing distance from the leading ed^e, data 
due to the 0.75b. Dia and 0.916h. Dia rounded edges become more similar. 

D D 


Next an elevated deck was installed above the basic building deck. The 
spacing between the basic building and the elevated deck was 7.6 cm (3 in) or 1/2 
building height. Tests were made with various rounded edge extensions on both 
the basic building and the deck. Figures 45 through 50 show the variation of 
turbulence with height for a given model configuration at different tunnel 
speeds, in each case there is a general decrease in turbulence with increasing 
wind velocity. The shape of the profile, however, seems to remain the same for 
a given configuration. There is no indication that * constant turbulence pro- 
file, independent of Reynolds number, is being appro^-hed. It should be 
remembered , however, that the maximum tested Re is two orders of magnitude 
smaller than that for an actual building so no real conclusions about scaling 
can be drawn. Figures 51 through 53 show turbulence profiles obtained by plac- 
ing different rounded edges on the building edge while keeping a given rounded 
edge on the upper deck. The profiles on each graph are similar. This indicates 
a relative insensitivity to building rounded edges for a given deck edge. 

Figures 54 through 56 show results for different deck edges and a constant build- 
ing edge. In each of these figures, the profiles fall into two groups. In each 
case, the similar profiles have similar arc length to radius ratios (a/r). 

These graphs indicate, therefore, that the rounded edge on the upper deck has a 
greater effect on the turbulence than the rounded edge on the lower section of 
the building. They also indicate the quantity a/r is important and should be 
investigated. Unfortunately, this could not be done in this program. The 
spacing of the elevated deck above the building is another important parameter 
that was not, but should be, investigated. It would also be desirable to per- 
form frequency and spectral -energy analyses on the turbulence above the deck. 

This investigation has slvwn that the building configuration has a marked 
effect on the turbulence above the building. 

Basic Fer-v Research 

During the eK ited deck turbulence program, it was noticed that mean 
velocity profiles took different shapes than corresponding profiles obtained 
with similar fence configurations on the basic building model. This indicated 
that the building configuration itself, or possibly the building-fence inter- 
action, has an effect on the velocity profile. In an attempt to isolate this 
effect and gain knowledge into the basic mechanism by which fences alone affect 
velocity profiles, a basic fence investigation was initiated. 

Experimental program . -The experimental part of this investigation 
utilized the test set-up shown in figure 57- In this arrangement, the fence 
was placed on a ground board above the tunnel boundary layer. Reasonably 
uniform flow was therefore incident on the fence as long as the fence was fairly 
close to the leading edge of the ground board. Figure 58 shows velocity profiles 
taken at various distances behind a simple one element fence. Near the screen 
the velocity profile is very irregular. This is because effects of individual 
wires are being seen. As the distance behind the fence increases, these 
differences are smoothed and a uniform profile is established. Figure 59 shows 
results for a more complicated two element fence configuration. Data were taken for 
this configuration up to 91 cm (36 in) (12 fence heights) downstream. As was 
the case with the simple one element fence, an equilibrium profile is established 
downstream and maintained until the boundary layer growing from the leading edge 


of the ground board becomes thick enough to affect the profile. At 91 cm 
(36 In) downstream, a large boundary-layer effect can be seen in these data up 
to about 1/2 fence height. Figure 60 shows velocity profiles in front of the 
fence for this 2 element configuration. It can be seen that there is very 
little upstream effects of the fence at 23 cm (9 in) (3 fence heights) in front 
of the fence. This is similar to results obtained in front of the fence on 
the building model. 

The striking difference between these results with fences on a ground 
board and those obtained with similar fences on building models is that the 
height to which the fence affects the flow is much less in the absence of the 
building. This could have important practical ramifications since it is 
structurally and economically desirable to build fences as low as possible. 
Corresponding profiles obtained with fences on the building model and on the 
rasied ground board also vary somewhat in appearance because of the different 
characteristics of the boundary layer associated with each case. 

Analytical program . - Analytical investigations of incompressible flow 
through wire gauze have been performed by Taylor and Batchelor in 1949 (ref. 3) 
and by Owen and Zienkiewicz in 1957 (ref. 4). The Taylor-Batchelor work deals 
with the effect of woven wire gauze on small disturbances to a uniform stream. 
In this work, the drag coefficient of a uniform screen is defined by 



p i - 

l/2cU 1 

( 1 ) 

Owen and 2ienkiewicz's analytical work on screens was performed as part of 
an attempt to produce uniform shear flow in a wind tunnel. They were able to 
predict the solidity distribution necessary to generate constant shear. 

The objective of the present work is to predict the fence solidity distri- 
bution necessary to generate a predetermined nonlinear velocity profile. In 
addition to differing in profile shape, this work differs from Owen and 
Zienkiewicz's in the boundary conditions on the fence. There is essentially 
no upper boundary on the fence in the current work. Owen and Zienkiewicz's 
grid spanned the entire test section of the wind tunnel. 

Qualitatively, the effect of a fence on a flow field can be described as 
follows. As a streamline passes through the fence, fluid momentum is lost. 

This corresponds to a decrease in stagnation pressure at the fence. Conserva- 
tion of mass requires that fluid velocity through the fence be continuous, 
therefore, the change in stagnation pressure must appear as a static pressure 
drop across the fence. Farther downstream an equilibrium condition is reached 
where static pressure does not vary vertically. At the equilibrium condition 
an exchange between static pressure and velocity has occurred so that the 
momentum (or stagnation pressure) decrease appears as a decrease in velocity. 

1 1 

Bernoulli's equation written for a stream) Ine passing through a fence is 

P- i + 1 °«_ 2 - p„ - i p° 2 (t) - Pj - p 2 

( 2 ) 

where p and p are the static pressures far upstream and far downstream, 
r r ®2 

respectively, and u(y) is the local equilibrium velocity downstream. In the 
absence of an upper boundary on the fence, p and D are equal. This allows 

r ®2 r <©2 

equation (2) to be rearranged as 

p i : i 





In a manner analogous to equation (I), a local drag coefficient can be defined 
by the left hand side of equation (3) so that 



It has been demonstrated that for practical solidities, streamlines passing 
through the fence are "essentially" straight so that y Q a y. 

For any arbitrary, single-element screen or parallel-rod grid in an in- 
compressible flow, it can be shown that the drag coefficient, K(y Q ) is related 

to the solidity, by 

«*„> C D 

U - C(y 0 >) 2 


where Cp is a drag coefficient based on the blocked area of the grid and the 

average interstitial velocity (see Appendix I). Experimental results have 
shown that in the absence of an upper boundary Cp * 0.5 for uniform grids of 

30 percent and 15 percent solidity. When the solidity is increased to 60 per- 
cent, however, the Cp decreases to approximately 0.14. This indicates that Cp 

is approximately constant at 0.5 for solidities less than approximately 30 per- 
cent. For higher solidities, however, it becomes a function of fence solidity. 
The following results are limited to maximum solidities of 0.3. Using Cp • 0.5. 

and combining equations (4) and (5) leads to 

£(y_) „ 2 

Ti"- »V') ] ; ’ 2 0 - (§-) 1 «> 

O 00 


Figure 61 shows a nonlinear velocity profile measured behind a parallel- 
rod grid installed in the University of Virginia Low Turbulence Wind Tunnel. 

The ratio of grid height to tunnel test section height was approximately 0.14. 
The experimental values of u/U^ from figure 61 were used in equation 
(7) to compute the solidity distribution. The computed solidity is compared 
to the actual measured solidity in figure 62. Figures 63 and 64 show similar 
results for another fence of approximately the same height, but constructed 
from a variable solidity screen. The average error is approximately 5 percent 
for the parallel-rod arrangement and approximately 15 percent for the screen. 
Host of this error can be attributed to wind tunnel blockage effects that 
prevent and from being exactly equal. Results are also sensitive to 
1 2 

accurate measurement of U . It should also be pointed out that these results 


cannot be extended to higher Reynolds number cases until Reynolds number effects 
on Cp can be determined. 


The problem of rooftop STOLport aerodynamics is a large and complicated 
one. Consequently, the initial investigation has generated more questions than 
it has answers. There are, however, some relatively firm conclusions which 
may be stated as follows. 

1) At the scale of these experiments, it Is possible to modify STOLport 
building models so that in a steady (wind tunnel) crosswind there is 
smooth attached flow over the models. This is achieved by attaching 
rounded-edge extensions to the upper lateral edges of the model. 

2) At the scale of these experiments, it Is possible to modify the 
steady crosswind profile above the building model, by attaching 
vertical porous fences to the upper edges of the model. By varying 
the fence geometry, a wide range of profiles can be obtained. 

3) By a not we 1 1 -understood mechanism, some model configurations produce 
large levels of turbulence at low frequencies (I Hz at this scale) 
above the STOLport deck. 

4) A simple theoretical analysis of the effect of a single 
element fence on a uniform velocity profile has been 
validated for low solidity fences (up to 0.30 area 

5) There are several important questions that have been raised, but left 
unanswered because of the limited scope of this research. The most 
important of these pertain to both fences and buildings and are: 

a) The determination of scale effects 


b) The determination of effects derived from the nature) 
unsteadiness in actual atmospheric flow. 

c) The determination of effects derived from the unique 
character of the earth's boundary layer. 

The above questions can only be answered by large scale atmospheric 



Relationship Between Solidity and Drag Coefficient 

-r he solidity of a uniform screen or parallel-rod grid is defined as the 
ratio of blocked area to total area. If a wire screen has grid dimensions of 
i by h where t is the grid length and h is the grid height and the diameter of 
the wire is d, the following can be written for each grid: 

Blocked Area = hd + id - d* 

Total Area = hi 


Open Area = h£ - (hd + id - d ) 

Consequently, the solidity, £, is given by 

, dCh + i - d) 

* ht 

or if the grid is square so that h -l 

, . d(2t - d) 

For the case of a parallel-rod grid, the solidity can be written as 

* “ S A3 



where d is the rod diameter and S is the spacing between successive rod 
centerlines. These ^'nations are valid local solidity expressions for a 
non-uniform screen (or parallel-rod grid with variable spacing) if h and £ 

(or S) are givf"’ neir local values. 

The relationship between the drag coefficient and solidity for a screen 
can be developed by considering the aerodynamic force on one grid as follows: 

Force » K(-j pU J) hi * C D (y pU aye 2 )(hd + td - d 2 ) A4 

where K is the drag coefficient based on total grid area and the freestream 
velocity, J , and C_ is the drag coefficient based on the blocked area of the 

oo y 

grid an d the average interstitial velocity, U aye - The average interstitial 
velocity can be found in terms of by considering conservation of mass in a 


rectangular (h x t) stream tube intersecting the screen at the grid where K 
is to be determined. For incompressible flows, this gives 

U hi ■ U [hi - (hd + Id - d 2 )] AS 

« eve 

which can be combined with equation A4 to give 

C D [(hd + Id - d 2 /(hl)] 

K “ [1 - (hd + Id - d*)/(hl)) 2 A6 

Similarly, it can be shown for a paral let-rod grid that, 

C D d/S 

* “ [1 - d/S] 2 


When written in terms of solidity, either of these equations reduces to the 
general result 


“ a"To 2 

As mentioned in the text, ■ 0.5 for solidities less then 30 percent (for the 
Reynolds numbers tested). “Equation A8 can therefore be used to determine 
the resistance, K, of a screen or parallel-rod grid as a function of solidity. 



1. Parker, Hermon M. ; Blanton, Jeffrey N. ; and Grunwald, Kalman, J.: Some 

Aspects of the Aerodynamics of STOl-ports. Proceedings of a Conference 
on NASA Aircraft Safety and Operating Problems, Vol. 1, NASA Sp-270, 
1971, pp. 263-276. 

2. Hwang, Men Shiuh: Experimental Investigation of Turbulent Shear Flow. 

Ph.D. Dissertation, University of Virginia, 1971. 

3. Taylor, G.I.; and Batchelor, G.K. : The Effect of Wire Gauze on Small 

Disturbances in a Uniform Stream. Quart. Journ. Mech. and Applied 
Hath., Vol. II, Pt. 1, I9*»9. pp. 1-29. 

k. Owen, P.R.; and Zienkiewicz, H.K. : The Production of Uniform Shear Flow 

in a Wind Tunnel. J. Fluid Hech., Vol. 2, 1957, PP- 521-531. 


m W» QtUUTtf 

Figure T 


Figure 2 










1 > 





LT) I 



tt ^ 

CL ~ 

£ W 








o o 
2! — 

O c uO 

2 x 

— ce ^ 

O u 

H? > O u_» 

3 ° a O 

® ^ oz 
o Q> ^ 

E > x 

__j CC *— 

° => > 
i/-> o > 

o O 

> o 
o 4/1 
o a. 


> X 
a: ►— 

o ^ 



r. — , 






> — 
O O 


> -Q 

cc Cg 

X ° 

<-> o’ 














(if 1 f*oCR 

Figure 3 


L flgr 




^we 8 


Of ICO* QlL. Ura 


Figure 10 


V *60 Knots V ■ 60 knots 

Figure 11 

Vcw moi 

Fig. 16- Effect of distance from upstream fence on cross wind profile 
3-Element fence. Each element construe ted from 
009 m wire , 0 4 96 * grid screen 

Fig. 20^ Effect of distonce from upstreom fence on cross wind profile 
2-Element fence ; constructed from 

0 02" wire , 0 12 5 H grid screen 

Fig. 22- E f fee t of distance from upstream fence on cross wind profile 

(Distance between upstream and downstream fences ,S) /(max fence height ,hf) = 867 

V CW max 

.Fig. 27- Effect of screen geometry 3-efemenf fences 


Top of fence 

* - w 

; O 0.009" wire ,0 0496” gr id screen 
TT $ 0 02” wire, 00125” grid screen 

i .2 4 .6 .8 1.0 

V C Wj 

v cw m o* 

Fig. 31 - Effect of screen geometry 2-element fences 


Figure 33 


o 5 10 15 20 25 30 

Time (sec) 

Fig. 34-Effect of height above building on turbulence 

Region of low pressure 

Fig. 36- Flow over building model when stagnation streamline has detached. 

Oi l i l_ i r, i, r i iti.i.i.i . 

.01 0.1 10 

Fig. 39- E f feet o f distonce from upstream building edge on turbulence profile 
building rounded edges 0.916 h^ diom. and 0.5ht,ore length R e =6.3xl0 5 

Fig. 41- Effect of distance from uf 
building rounded edges 05 

■01 01 1.0 10 


Fig. 43- Effect of building rounded edge on turbulence profile ot centerline 
R e -6 3x I0 5 

S 2323 SCI 


Measurements mode 
of I.Ob 



leat—a wi 



O 0-9l6h b diom ,05bb ore length 

A 0.75h|j diom ,0.543hbare length 
□ 0-5hh diom. ,0.292hb ere length 

Fig. 44- Effect of building rounded edge on turbulence profileat troilmgedge 
R e = 6 3 x I0 5 

Ele voted deckt 


Measurements made 

— — Mil 

B«aB8BB — » 


Reynolds Number 

-0 0.8x105 r -- 

' □ I.84xI0 5 “ 

"O 2.7x105 TJj 
'A 4.0x105 
- 0 5.2x 105 — 

: o 6.3 x 105 


Fig. 45- Effect of Reynolds number on turbulence P r ofiie above building 
center Ime. £ievoted deck configuration. Building rounded edge ■, 
0.916 bedlam and 0.5hbore length. Elevated deck rounded edge , 
0.75 hb diam ond 0.543 hjj are length 





Measurements made 
above centerlme 

Elevated deck 







Reynolds Number 



Fig. 46- Effect of Reynolds number on turbulence profile obove building centerline 
elevated deck configuration Building rounded edge diom and 0.543 h^ 

ore length. Elevated deck rounded edge,Q. 7 5h^ diom and 0 543h^ore length 


Measurements made 
above centerline 

Elevated deck 



I* — \ 

Reynolds Number 


08xl0 5 


I.84xl0 5 


2.75 xIO 5- 

~ A 


1 0 

52xl0 5 

1 o 

6.3xl0 5 



Fig. 47- Effect of Reynolds number on turbulence profile obove building centerline 
elevoted deck configuration Building rounded edge , 0.916 hb dtom ond 
0.5 h^ are length Elevated deck rounded - t 0.583 h^diam ond 0 438hbore 

8- Effect of Reynolds number on turbulence profile obove build ng centerline 
e leva ted deck configuration. Build ing rounded edge; 0.583 h^d'orn and 
0.438hbure length.elevated deck rounded edge ; 0 75 d mm and 
0 543 hb ore I ength 

Fig. 49- Effect of Reynolds number on turbulence profile obove building centerline 
elevoted deck configurotion.Building rounded edge;0.5h b diam and 
0 292 h b are length. Elevoted deck rounded edge; 0 916 h b diam 
and 05h b are length 

Meosurements made 
above centerline 

■■■■B MWWM iWW ijW WWiMiiiilil 





Reynolds Number; j 
O 0.8x105 pi 

n 1 .84xl0 5 ft 

!0 2.75 x I0 5 

A 4.0 x10 s 14- 

. 0 5.2 xio 5 Lj-j 

C 6.3 x 1 0 5 : : 


Fig. 50- Effect of Reynolds number onturbulence profile obove building centerline 
elevated deck conf igura lion. Bui Iding rounded edge ; 0.583 h^ diom and 
0.483 h^ ore length E le voted deck rounded edge; 0.916 h^ diom ond 
0 5 hb are length 

Fig. 51- Effect of geometry of building rounded edge on turbulence profile obove 
the centerline Elevated deck rounded edge; 05bb diom ,0292 ore length 

Re -'6 3xl0 5 



Measurements at <£ 

Elevoted deck 








Building rounded edge 
(> 0 9l6htj diom ond 0.5htjore length 

O 0.5 h^ diam ond 0 292 h^ ore length 
□ 0583hb diom ond 0 483hb ore length 
a 0.75h b diem and 0543hb ore length 





Fig. 52- Effect of geometry of building rounded edge on turbulence profile above 
the centerline Elevoted deck rounded edge ; 0.75 hfc diom ond 0 543 are length 

Re = 6. 3x105 


Measurements at <£ 

Elevofed deck 








■I in 

Building rounded edge 

0 0.9l6hb diom. and 0.5 hbore length 
O 0.5 hfcd iom. ond 0.292 are iength 
□ 0-583 hfc diom. ond 0.483 ore length 
A 0-75hx diam ond 0543 hbore lengi. 


Fig. 53- Effect of geometry of building rounded edge on turbulence profile obove 

the centerline. Ele vo ted deck rounded edge ; 0 91 6 hfc diom ond 0 5ht>ore length 

Re = 6 3xl0 5 

Fig. 54- Effect of geometry of elevoted deck rounded edge on turbulence prof ile obove 
the centerline. Building rounded edge ;09l6hbdiom.and0.5hb are length 

R e =63x I0 5 

Fig. 55- Effect of geometry ofelevated deck rounded edge on turbulence profile above 
the center I ine Bu i Iding rounded edge ; 0.583 h^d iom ond 0.483 h^ ore le ng th 

R e =6.3xl0 5 

Fig. 56- Ef fee t of geometry of ele voted deck rounded edge on turbu lence prof ne above 
the center ne. Bui Id ing rounded edge,0.5hb diam and 0 292 h^ ore length 

R e = 63xl05 

Wind tunnel test section 



Fig. 57- Experimental test set-up 




Fig. 62- Computed solidity compared to actual solidity for parallel rod grid