372 AIAA JOURNAL VOL. 27, NO.

3
Ueda, T., Mizomoto, M., Matsubatashi, Y., and Ikai, S., ^Tur- nonuniformity is equal to the maximum tolerable level for a
bulent Properties of a Flat Plate Boundary Layer Diffusion Flame," given application (typically less than 1% variation in mean
AIAA Paper, Jan. 1983.
4
stream wise velocity outside the boundary layers).
Ueda, T., Mizomoto, M., and Ikai, S., "Thermal Structure of a Several papers have been published on the design or choice
Flat Plate Turbulent Boundary Layer Diffusion Flame," Bulletin of of contraction wall shapes using a variety of analytical and
the JSME, Vol. 26, March 1983, pp. 399-405. numerical techniques (see Ref. 3 for a review). Most recent
5
Senda, M., Suzuki, K., and Sato, T., "Turbulent Structure
Related to the Heat Transfer in a Turbulent Boundary Layer with In-
studies have involved the calculation of the wall pressure
jection," Second Symposium on Turbulent Shear Flows, Imperial distributions, using some potential flow numerical scheme,
College, London, England, July 1979, pp. 9.17-9.22 and then the application of a boundary-layer separation crite-
6
Kent, J. H., "Turbulent Jet Diffusion Flame," Ph.D. Thesis, rion based on a critical value of the pressure coefficient. The
Univ. of Sydney, Sydney, Australia, 1972. most popular separation criterion used is that due to Strat-
7
Yam, C., "A Study of a Turbulent Boundary Layer Diffusion ford4 for turbulent boundary-layer separation. We propose in
Flame," M.S. Thesis, Univ. of California, Davis, Davis, CA, 1986. this Note that a boundary layer in a small, low-speed contrac-
8
Cebeci, T. and Smith, A. M. O., Analysis of Turbulent Boundary tion is more likely to start in a laminar state and remain so, for
Layers, Academic, New York, 1974. the most part, in passage through it. The normally applied
9
Wooldridge, C. E and Muzzy, R. J., "Boundary-Layer Turbu-
lence Measurements with Mass Addition and Combustion," AIAA
Stratford's criterion for turbulent boundary-layer separation
Journal, VoL 4, Nov. 1966, pp. 1009-2016. therefore may be too liberal for these designs.
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on February 2, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.10122

Computational Approach
A three-dimensional potential flow code (VSAERO) was
used to compute the velocity distributions along the contrac-
tion walls.3 VSAERO uses a singularity panel method employ-
Boundary-Layer Predictions ing sources and doublets to solve the Laplace equation.
It was hypothesized in this study that for small, low-speed
for Small Low-Speed Contractions wind tunnels the boundary layers enter the contraction in a
laminar state. In most small wind tunnels, the flow entering
the contraction comes through a honeycomb and a series of
screens (usually at least three). The effect of a screen on a
James H. Bell* and Rabindra D. Mehtat turbulent boundary layer is to significantly reduce its thickness
Stanford University, Stanford, California and turbulence stress levels and scales, as shown by Mehta.5
The results from that investigation showed that a turbulent
boundary layer at moderate Reynolds numbers (Re0~1600)
Introduction was effectively relaminarized immediately downstream of the
screen. Note that the typical Ree encountered in small, low-
C ONTRACTION sections form an integral part of all
wind tunnels, whether designed for basic fluid flow re-
search or model testing. The main effects of a contraction are
speed settling chambers is likely to be lower by at least an
order of magnitude. However, "forced" transition may still
occur through either the effects of the Taylor-Gortler instabil-
to reduce both mean and fluctuating velocity variations to a ities in the regions of concave curvature or a separation bub-
smaller fraction of the average velocity and to increase the ble. In either case, the strong favorable pressure gradient,
flow mean velocity.1 The most important single parameter in encountered in contractions with reasonable area ratios
determining these effects is the contraction ratio c. Contrac- (c — 6-10), would invariably relaminarize the boundary layer
tion ratios of between 6 and 10 are found to be adequate for
most small, low-speed wind tunnels—defined here as tunnels
with a test section cross-sectional area of less than about 1.2 r
0.5 m2 and freestream velocities of less than about 40 m/s.
The wall shape design of a contraction of given area ratio
and cross section centers on the production of a uniform and
steady stream at its outlet. These conditions generally can be 1.0
met by making the contraction section sufficiently long. On CURVED WALL
the other hand, another desirable flow quality, namely a min- >
H
imum boundary-layer thickness (in a laminar state) at the CJ
contraction exit, suggests that the contraction length should be .8
minimized. However, the risk of boundary-layer separation
near the two ends of the contraction increases as the length is
reduced. In general, the boundary layer is less liable to sepa-
rate at the contraction exit, due to its reduced thickness caused .6
by passage through the strong favorable pressure gradient. Q
Also, the concave curvature at the contraction inlet has a LU
N
destabilizing effect on the boundary layer, in contrast to the
convex curvature near the exit that has a stabilizing effect.2 In
addition to unnecessary thickening of the boundary layer,
separation also generally leads to flow unsteadiness, which
cannot be easily eliminated from the test section flow. A
design satisfying all criteria will be such that separation is just
avoided (implying a minimum acceptable length), and the exit

Received March 21, 1988; revision received June 2, 1988. Copyright I

© American Institute of Aeronautics and Astronautics, Inc., 1988. -.5 .5 1.0 1.5
All rights reserved.
* Graduate Student, Department of Aeronautics and Astronautics. X/L
tResearch Associate, Department of Aeronautics and Astronautics.
Member AIAA. Fig. 1 Typical calculated wall velocities.
 MARCH 1989 TECHNICAL NOTES 373

Table 1 Details of contractions used for comparisons

Inlet Exit
height height Length-
2-D x x Contraction to- Exit Wall Splitter
Wind Contraction or width, width, length, height aspect contour plate
tunnel ratio 3-D cm cm cm ratio ratio shape Y/N

a) Shear layer tunnel 10 2-D 38 x 76 7.6 x 38 91 1.20 5.0 "Eye" Y

(NASA Ames) design
b) Mixing layer tunnel 7.7 2-D 91 x 137 91 x 18 244 0.89 5.14 Fifth-order Y
(NASA Ames) polynomial
c) Boundary layer 7.5 3-D 120 x 100 20 x 80 120 1.0 4.0 Fifth-order N
tunnel (NASA Ames) polynomial
d) Boundary layer 9 3-D 114x76 13x76 122 1.07 6.0 "Eye" N
tunnel (Imperial College) design
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on February 2, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.10122

Table 2 Comparison of momentum thickness results

Freestream
Wind velocity Measured Predicted
tunnel Test wall Ue, m/s 6, cm 0, cm Difference,^

a Splitter plate low-speed side 10 0.046 0.0425 -8

Splitter plate high-speed side 21 0.032 0.0288 -11
b Splitter plate low-speed side 9 0.0615 0.0621 1
Splitter plate high-speed side 15 0.0508 0.0483 -5
c Test surface 25 0.0325 0.0326 1
following curved wall
d Test surface 16 0.0350 0.0347 - 1
following curved wall

WIND TUNNEL
1.0
PDMINT

.8

.6

.4

.2

1.0 1.25 1.50 .2 .4 .8 1.0

X/L

Fig. 3 Wall contour shapes of contractions used for verification of

Fig. 2 Calculations of boundary-layer development through a typi- the computational scheme (L is the contraction length and /// the inlet
cal contraction. height).

soon after. Therefore, the assumption of a laminar boundary of the contraction wall and the plane of symmetry (represent-
layer originating from stagnation conditions at the contraction ing a splitter plate).
entrance and remaining laminar throughout was considered to
be an adequate approximation for all cases. Results and Discussion
The wall velocity data computed by VSAERO were used as A typical example of the wall velocity distributions as com-
boundary conditions for the viscous calculations with stagna- puted by VSAERO is given in Fig. 1. Note the presence of
tion conditions specified at the contraction entry. Two differ- regions of adverse pressure gradient on the curved wall near
ent two-dimensional boundary-layer codes were employed, the contraction inlet and outlet indicated by the decrease in
one a simple integral method solving the momentum integral wall velocity. The momentum thickness B and skin-friction
equation for laminar boundary layers (Thwaites method)6 and coefficient C/ distributions computed using both the
the other a finite-difference technique solving the boundary- boundary-layer programs along the curved wall of the same
layer equations (PDMINT).7 It is worth noting that neither of contraction section are shown in Fig. 2. Initially, the boundary
these methods accounted for the effects of longitudinal curva- layer grows rapidly in the inlet region where the effects of the
ture or lateral divergence on boundary-layer structure explic- first adverse pressure gradient are felt. This is soon over-
itly. The boundary layers were calculated along the center line whelmed by the effects of the strong favorable pressure gradi-
 374 AIAA JOURNAL VOL. 27, NO. 3

ent, resulting in a rapid reduction in 6, before the effects due References

to the adverse gradient near the outlet take over. The Cf 'Mehta, R. D. and Bradshaw, P., "Design Rules for Small Low
remains well positive throughout the contraction, thus indicat- Speed Wind Tunnels," AeronauticalJournal, Vol. 83, No. 827, Nov.
ing that separation is not predicted for this particular design. 1979, pp. 443-449.
2
The results from the two techniques are seen to compare Bradshaw, P., "Effects of Streamline Curvature on Turbulent
favorably, especially for the region near the contraction out- Flow," AGARDograph 169, Aug. 1973.
3
let. Since the program for Thwaites method is relatively easy Bell, J. H. and Mehta, R. D., "Contraction Design for Small
to obtain and use,6 all of the boundary-layer data presented Low-Speed Wind Tunnels," Dept. of Aeronautics and Astronautics,
below were computed using this method. Stanford Univ., Stanford, CA, JIAA Rept. TR-84, April 1988; also
In order to validate the proposed computational scheme, NASA-CR 177488, Aug. 1988.
4
Stratford, B. S., "The Prediction of Separation of the Turbulent
the boundary-layer properties in four contractions installed on Boundary Layer," Journal of Fluid Mechanics, Vol. 5, Pt. 1, Jan.
blower-driven wind tunnels (Table 1) were calculated and 1959, pp. 1-16.
compared with actual measurements. The wall shapes for the 5
Mehta, R. D., "Turbulent Boundary Layer Perturbed by a
four contraction sections are plotted in normalized coordi- Screen," AIAA Journal Vol. 23, Sept. 1985, pp. 1335-1342.
6
nates in Fig. 3. Two of the wall shapes were based on a Cebeci, T. and Bradshaw, P., Momentum Transfer in Boundary
fifth-order polynomial, while the other two were designed "by Layers, Hemisphere, New York, 1977.
7
eye" following the guidelines given in Ref. 1. Note that for all Murphy, J. D. and King, L. S., "Airfoil Flow-Field Calculations
four contractions investigated, the nondimensional geometric with Coupled Boundary-Layer Potential Codes," Proceedings 2nd
Conference on Numerical and Physical Aspects of Aerodynamic
Downloaded by UNIV OF CALIFORNIA SAN DIEGO on February 2, 2015 | http://arc.aiaa.org | DOI: 10.2514/3.10122

parameters are similar; the contraction ratio is about 8, the Flows, California State Univ., Long Beach, CA, 1983.
length to inlet height ratio is about 1, and the cross-sectional 8
Mokhtari, S. and Bradshaw, P., "Longitudinal Vortices in Wind
aspect ratio at the exit is about 5. Tunnel Wall Boundary Layers," Aeronautical Journal, Vol. 87,
For all four contraction shapes, the computations predicted June/July 1983, pp. 233-236.
an attached boundary layer along the entire length. The mea-
sured values of the boundary-layer momentum thickness at
the exit of the contractions are compared with those predicted
by Thwaites' method in Table 2. The comparisons are made
along the wind-tunnel centerline at a short distance (typically
less than 15 cm) downstream of the contraction exit. The
predicted values of the momentum thickness for all four cases
are within about 10% of the measured ones; the typical mea- Motion and Deformation of Very Large
surement repeatability was within 1%. The predicted values
generally are lower than the measured ones, presumably due
Space Structures
to the generation of weak secondary flows along the contrac-
tion walls,8 which tend to thicken the boundary layer along the Ramesh B. Malla*
tunnel centerline. Note that the maximum error between the
predictions and experiments occurs for wind tunnel A, which University of Connecticut, Storrs, Connecticut
also has the maximum three-dimensionality in the contraction and
geometry. The agreement between predictions and measure- William A. Nasht and Thomas J. Lardnert
ments for all the other boundary-layer properties was com- University of Massachusetts, Amherst, Massachusetts
parable and details are given in Ref. 3.

Conclusions Introduction
A scheme is proposed for the prediction of boundary-layer
development in small, low-speed contraction sections. The
T HE effects of various space environmental factors (such
as gravitational forces, radiation heating, radiation pres-
sure, space drag, and particle impact) on space structures
wall pressure distributions, and hence the wall velocity distri-
butions, are first calculated using a three-dimensional poten- have long been realized and studied by numerous investigators
tial flow method. Although a panel method was used in this in past for small-sized space structures (for example, see refer-
investigation, in principle, any potential flow solver should be ence list in Ref. 1). The deployment of very large flexible light-
acceptable. Once the wall velocities have been obtained, the weight structures in space, which have characteristic dimen-
boundary-layer behavior can be adequately calculated, rather sions of several kilometers operating at altitudes of
than relying on some separation criterion. For the family of 200-35,900 km above the Earth's surface, gives rise to a
contractions discussed in this Note, the assumption of a lami- number of new and significant problem areas that may or may
nar boundary layer originating at the contraction entrance and not have been associated with smaller systems.
remaining laminar in passage through it seems justified. The This Note presents a theoretical development of the equa-
measured boundary-layer momentum thicknesses at the exit of tions of motion of a very large axially flexible structure or-
four existing contractions, two of which were three-dimen- biting the Earth with planar motion in a general noncircular
sional, were found to lie within 10% of the predicted values, orbit. With the help of these equations of motion, it is antici-
with the predicted values generally lower. The present results pated that effects of many disturbances can be investigated on
indicate that the relatively simple Thwaites method is probably large space structures' orbital motion, attitude motion, and
adequate for most purposes. If the prediction accuracy of axial deformation (length). As an application of the equations
within 10% on 6 is acceptable, then the present results also of motion, the effects of the orbit eccentricity are studied on
suggest that an iterative process, accounting for the boundary- the coupled orbital, attitude, and axial motion of a large space
layer displacement thickness, is not necessary. structure under the influence of the Earth's gravitational
forces.

Acknowledgments
Received Oct. 10, 1987; revision received April 27, 1988. Copyright
This work was supported by the Fluid Dynamics Research © American Institute of Aeronautics and Astronautics, Inc., 1988.
Branch, NASA Ames Research Center under Grant NCC-2- All rights reserved.
294. We would like to thank our colleagues at NASA Ames * Visiting Assistant Professor, Department of Civil Engineering.
Research Center for many helpful comments. tProfessor, Department of Civil Engineering.