Vous êtes sur la page 1sur 3

15 Nunner, W.

, "Warmeubergang und Druckabfall in rauhen


= 0.2 Rohren," VDI-Forschungsheft 465, Vol. 22, 1956, pp. 5-39
R' (Translated: A E R E lib./Trans. 786, 1958).
16 Mobius, H., "Experimental Investigation of Resistance
for at least 0.03 < e/8 < 0.08. From the locus of the merge and Velocity Distribution in Tubes with Regularly Arranged
points in the mean velocity and turbulence distributions it is Roughnesses in Turbulent Flow," Phijsik Zeitschrift, Vol. 41,
concluded t h a t the spread of turbulent kinetic energy and Reyn- 1940, pp. 202-255.
olds shear stress change precede the spread of momentum change. 17 Hanjalic, K., and Launder, B. E., "Fully-Developed
Asymmetric Flow in a Plane Channel," Journal of Fluid Me-
6 In the region near the wall, termed the lower layer, whose chanics, Vol. 51, 1972, pp. 301-335.
boundary is defined by the locus of the lower knee points, the 18 Makita, H., "Response of a Turbulent Boundary Layer
flow assumes a character influenced by the rough wall. The to a Sudden Change in Surface Roughness," Unpublished M.
growth of this layer was found to correlate with the growth of the Eng. thesis, University of Tokyo, 1968.
19 Counihan, J., Hunt, J. C. R., and Jackson, P. S., "Wakes
wake behind the initial element, i.e., Behind Two-dimensional Surface Obstacles in Turbulent Bound-
ary Layers," Journal of Fluid Mechanics, Vol. 64, 1974, pp.
l(x) 529-563.
KX-
e
where K can be viewed as a turbulent Reynolds number for the
element based on the element height and the mean velocity of the
fully-developed flow t h a t precedes the element at the location of
the crest of the element. DISCUSSION
7 T h e flow in the wake behind an isolated element was found 2
W. H. Schofield
to be approximately self-preserving.
This paper competently treats the response of a turbulent
Acknowledgments shear flow to a change in surface roughness in a flow situation of
practical importance. The outstanding feature of the work is
The authors appreciate the helpful comments of Professor D . the turbulence data around the roughness elements. These data
M. McEligot, Department of Aerospace and Mechanical Engi- illustrate the periodic flow regime close to the rough wall for the
neering, University of Arizona. first time. Complete data of this sort will eventually lead to a
detailed knowledge of the turbulence structure around a rough-
References ness array.
1 Jacobs, W., "Umformung eines turbulenten Geschwindig- The most disappointing aspect of the paper is t h a t although a
keits-Profils," Zeitschrift fur angewandt Maihematik und Mecha- large number of roughness geometries were tested there is very
nik, Vol. 19, 1939, pp. 87-100 (Translated: NACA T M No. 951.) little comparison of the results for different geometries. The great
2 Carper, H. J., Heilhecker, J. K., and Logan, E., " T h e majority of the results presented refer to a single geometry—
Effect of a Change in Wall Roughness on Turbulence Intensity
and Shear Stress Distribution in a Two-dimensional Channel configuration A, which in the usual roughness terminology, is a
Flow," Developments in Mechanics, Vol. 4, Proceedings, Eighth "k" type roughness pattern (Perry, et al. [15]). Recently Woob
Midwestern Mechanics Conference, Elmsford, N . Y.; Pergamon, and Antonia [16] have shown that the turbulent structure over a
1965, pp. 16-25. "d" type rough wall (a close packed array similar to configura-
3 Islam, O., and Logan, E., "Channel Flow Over a Smooth- tion D of the authors which they termed "skimming type") is
to-Rough Surface Discontinuity with Zero Pressure Gradient,"
ASME Paper No. 76-FE-4, 1976. quite different from t h a t over a "k" type rough wall. A differ-
4 Tani, I., and Makita, PL, "Response of a Turbulent Shear ence in behaviour after a change in roughness could therefore be
Flow to a Step-wise Change in Wall Roughness," Zeitschrift fur expected for this roughness geometry (and is in fact indicated in
Flugwissenschaften, Vol. 19, 1971, pp. 335-339. Fig. 11). I t is to be hoped t h a t further detailed comparisons for
5 Clauser, F . H., "The Turbulent Boundary Layer," Ad-
vances in Applied Mechanics, Vol. 4, 1956, pp. 1-51. the various roughness types will be presented by these authors in a
6 Antonia, P. A., and Luxton, R. E., "The Response of a subsequent Technical Note.
Turbulent Boundary to an Upstanding Step Change in Surface An overshoot of wall shear (above its final value) immediately
Roughness," Journal of Basic Engineering, TRANS. ASME, after a change in roughness is a common and important feature
Series D, Vol. 93, 1971, pp. 22-34.
7 Antonia, P. A., and Luxton, R. E., "The Response of a of flows after a change in roughness. The pressure gFadient in
Turbulent Boundary Layer to a Step Change in Surface Rough- Fig. 1 does not appear to "assume its fully developed value im-
ness, P a r t 1, Smooth to Rough," Journal of Fluid Mechanics, mediately" as stated in the paper. In fact the pressure gradient
Vol. 48, 1971, pp. 721-761. shown implies an overshoot in T wa n with a magnitude of c^ 40
8 Antonia, P. A., and Luxton, R. E., "The Response of a
Turbulent Boundary Layer to a Step Change in Surface Rough- percent of the final value over the first one or two diameters after
ness, P a r t 2, Rough to Smooth," Journal of Fluid Mechanics, the change in roughness. However this value for the overshoot
Vol. 53, 1972, pp. 737-757. will be inaccurate if at the change in roughness there is also a
9 Mueller, T. J., and Robertson, J. M., "A Study of the change in effective tube diameter, which would also affect the
Mean Motion and Turbulence Downstream of a Roughness Ele-
m e n t , " Proceedings of the First Southeastern Conference on pressure gradient.
Theoretical and Applied Mechanics, Gatlinburg, Tenn., M a y 3-4, This overshoot is also well displayed by the slopes of the inner
1962. profile in Fig. 2, which is effectively a Clauser Chart (Clauser
10 Plate, E . J., Aerodynamic Characteristics of Atmospheric [21]) if the small variation in Ucis ignored. In addition the authors
Boundary Layers, (USAEC TID-25465), National Technical In-
formation Service, Springfield, 1971. note t h a t at X/D' = 0.425 the Reynolds stress overshoots the
11 Oka, S., and Kostic, Z., "Flow Field Past a Single Rough- equilibrium distribution near the wall, however it also appears
ness Element in Channel of Rectangular Cross-section," Heat and that this Reynolds stress distribution is heading for a wall shear
Mass Transfer in Boundary Layers, Vol. I, Pergamon, Oxford, t h a t is greater than the equilibrium value.
1972.
12 Logan, E., and Jones, J. B., "Flow in a Pipe Following In Fig. 2 the final profile (X/D' = 18) is labelled "fully de-
an Abrupt Increase in Surface Roughness," Journal of Basic veloped" meaning the profile shape is invariant with further
Engineering, TKANS. ASME, Series D, Vol. 85, 1963, pp. 35-46. axial development. However this has not been demonstrated by
13 Morris, H. M., "Design Method for Flow in Rough Con-
duits," Journal of the Hydraulics Division, Proceedings of the
ASCE, Vol. 85, 1959, pp. 43-62.
14 Schlichtmg, H., Boundary-layer Theory, McGraw-Hill, "Aeronautical Research Laboratories, Department of Defense, Victoria,
New York, 1968, p. 585. Australia.

Journal of Fluids Engineering SEPTEMBER 1977 / 553


Copyright © 1977 by ASME
Downloaded 09 Dec 2010 to 82.137.208.59. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
comparison of the X/D' = 18 profile with another profile further Additional References
downstream nor is there reference in the paper to such a profile 15 Perry. A. E., Schofield, W. H., and Joubert, P. N., "Hough
having been recorded. Fig. 9 shows the "lower knee point" has Wall Turbulent Boundary Layers," / . Fluid Mech., Vol. 37
not reached the pipe centerline at X/D' = 18 and hence it seems 1969, pp. 383-413. '
possible that this profile has not yet reached equilibrium. A 16 Wood, D. H., and Antonia, R. A., "Measurements in a
Turbulent Boundary Layer Over a 'd' Type Surface Roughness,"
simple test for equilibrium is to plot the X/D' = 18 profile on Journal of Appl. Mech., Vol. 42 TKANS. ASME, Vol. 97, Series
defect coordinates [(Uc — U)/u*, Y + e)/R'Y and compare it E, No. 3 Sept. 1975, p. 591.
with the Universal Velocity Defect Law for pipes. If the profiles 17 Moore, W. L., PhD thesis, State University of Iowa, 1951.
agree then the flow has completely adjusted to the new boundary 18 Antonia, R. A., and Wood, D. H., "Calculation of a Tur-
bulent Boundary Layer Downstream of a Small Step Change in
condition and has forgotten its early smooth wall history. It Surface Roughness," Aero. Quarterly, Aug. 1975, pp. 202-210.
seems preferable to use defect coordinates here as they expand the 19 Townsend, A. A., "Self Preserving Flow Inside a Turbu-
outer part of the profile where the last part of the flow adjust- lent Boundary Layer," / . Fluid Mech., Vol. 22, 1965, pp. 773-797.
ment occurs. If on this basis the X/D' = 18 profile is fully de- 20 Schofield, W. H., "Measurements in Adverse - Pressure -
Gradient Turbulent Boundary Layers with a Step Change in
veloped then pipe flow requires almost twice the distance (36 Surface Roughness," J. Fluid Mech., Vol. 70, 1975, pp. 573-593.
W) as a similar zero pressure gradient flow (20 5, Antonia and 21 Clauser, F. H., "Turbulent Boundary Layers in Adverse
Luxton, author's reference [6]) to become fully developed. Pressure Gradients," / . Aero. Sci., Vol. 21, 1954, p. 91.
The edge of the internal layer is that point where some meas-
ured quantity (say U/Uc) merges with the corresponding profile
of a reference layer that developed under the same conditions J. C. R. Hunt 5
but without the change in roughness. In pipe flow the profile
before the change in roughness can be used as a reference (as This is an interesting and worthwhile contribution to the study
was done in this paper) in determining the merge points. I am of flow over roughness elements. The authors make detailed
not aware of any physical basis for using "knee" points to deter- comparison with two of the findings of Counihan, Hunt, and
mine the edge of the internal layer or for assuming that they will Jackson (reference [19]):
necessarily coincide or correlate with merge points in this OT (i) that the thickness of the wake l(x) behind a single surface
any other flow with a sudden change in boundary conditions. obstacle grows approximately as x0-6, (much as a wake behind an
Knee points were introduced by Antonia and Luxton—author's obstacle in a uniform flow)
reference [6] (on the basis of a tentative half power velocity dis- (ii) that the velocity deficit Aw has a self preserving form in
tribution analysis) and later abandoned by them. There seems the wake behind the single element. (In the case of reference [19]
no reason to continue their use when merge points are simple to Auax-if(y/l(x)))-
determine and have physical meaning. In making this comparison the authors do not emphasize the
There appears to be no evidence presented in the paper to difference between the zero pressure gradient boundary layer
support the assumption that the upper knee points can be used studied in reference [19] and the pipe flow studied here. A simple
as an approximate measure of the extent of the shear layer pro- extension of the analysis of reference [19] to the wake behind a
duced by the first element. In fact Fig. 9 would suggest a dif- single roughness element in a pipe wall suggests (a) that l(x)
ference between the two that is variable and as large as 60 per- is proportional to x0-5 as before, (6) that when x > D and
cent (at X/D' = 0.425). Similarly no evidence is cited for taking when D%/(ke) >> x where K is a dimensionless eddy viscosity,
the lower knee point as a measure of the extent of the "reat- 6 is the element height and the pipe diameter is D, the perturbed
tachment wake" behind an element. flow outside the wake is affected by the confinement of the pipe
Conclusion 5 could be read as meaning that the internal layer walls, so that Auax~mf(y/l(x)). The authors observe in Fig. 13
grows as 0.2 [X/5]°* for any change in roughness provided 0.03 that Auax"112, as we suggest, but they do not comment on the
< e/S < 0.08. The important qualification that should be difference between their x~~m decay and the ar 1 decay of Ait in
stated is that the roughness geometry must also be upstanding. reference [19]. It would be interesting to know whether they
If it is downstanding the growth exponent will be 0.8 (author's measured the wall pressure behind the single element; if so, how
reference [7]). Again for flow over "d" type roughness (skimming slowly did the pressure gradient revert to its upstream value?
roughness in this paper) the growth constant is significantly
smaller (0.14) and the flow model given in the paper cannot be
applicable (see Wood and Antonia [16] and Antonia and Wood
[18] for flow details over this roughness type). Author's Closure
The internal layer growth rate is expressed in this paper (as in We are grateful to Drs. Schofield and Hunt for their helpful and
earlier papers) in terms of 6—an outer flow length scale. It is illuminating comments on this paper. Particularly useful are the
preferable to relate the internal layer growth rate to a length suggestions for future work, e.g., a detailed comparison of results
scale associated with the wall, as the flow modification originates for various roughness geometries and a new correlation using the
from the wall and is driven by the change in wall shear. The wall wall length scale Z.
length scale, z obtained by writing the universal law of the wall as We agree with Dr. Schofield that an overshoot of wall stress
does occur and is clearly indicated by the slopes of the velocity
u 1, / yu* \ 1
profiles in Fig. 2. Although no velocity defect profile has yet been
plotted using the mean velocity data at X/D' — 18 (Fig. 2),
confirmation of the fully-developed character of the profile was
was used in Townsend's [19] analysis of this type of flow. This accomplished by comparing it with the fully-developed profile
analysis related 8o/z to X/z and Schofield [6] subsequently of Nunner (reference [15]). Measured mean velocity profiles at
showed that these variables correlate results for three different X/D' = 16.425, 18.025, and 30.025 were compared with Nunner's
flows. It would be interesting to see if the present results agree profile at X/D' = 30 for the same roughness geometry and were
with this result. found to agree well. Further evidence of full development was
obtained by observing the similarity of the turbulence profiles
measured at X/D' = 30 and at X/D' = 18 (e.g., see Fig. 3).
'Numbers in brackets indicate Additional References at end of discussion.
'u* = friction velooity (Twall/p)1'2; e = error in origin of the mean velocity
profile, see Moore [3], Sohofield [6J. 6
University of Cambridge, Cambridge, England.

554 / SEPTEMBER 1977 Transactions of the ASME


Downloaded 09 Dec 2010 to 82.137.208.59. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
The term "knee" point should perhaps be abandoned as it is covered by the equation of conclusion 5. The authors were not
only an approximation of the merge point which has real physical aware of the growth constant found by Antonia and Wood for
significance. Knee points do have the advantage of being easily skimming flow but have independently verified the downstanding
determined. Shear layer thickness, as pointed out by Dr. Scho- exponent (0.8) using the data of reference [12].
field, is correctly based on the upper merge points, since a larger We appreciate Dr. Hunt's extension of his theory to include the
error exists at small X/D' if knee points are used. Similarly the single element in pipe flow. Our data agree with his theoretical
lower merge points should be used in lieu of lower knee points in prediction, i.e., the velocity deficit is proportional to X -1 ' 2 .
determining the wake depth behind the element. Unfortunately, we did not obtain the detailed pressure distribu-
The slumming flow and downstanding element cases are not tion for the single element.

Readers of the Journal of Fluids Engineering may be interested in:

Condensation in High-Speed Flows


y
A collection of papers presented at the 1977 Joint Meeting of the Fluids Engineering, Applied Mechanics and
Bioengineering Divisions of The American Society of Mechanical Engineers, New Haven, Connecticut, June 15-17,
1977.

Editor: A. A. Pouring

On the Growth of Steam Droplets Formed in a Laval Nozzle Using Both Static Pressure and Light Scattering
Measurements C. A. Moses and G. D. Stein

Optical Studies of Nucleation in High-Pressure Expanding Steam I. J. Valha and D. J. Ryley

The Direct Measurement of Nucleation Rate Constants R. A. Dobbins, T. I. Eklund and R. Tjoa

Analysis of Nitrogen Condensation in an Expanding Nozzle Flow F. C. Wang

Analysis of Condensation in the Centered Expansion Wave in a Shock Tube B. J. C. Wu

Condensation of H 2 0 and D 2 0 in Argon in the Centered Expansion Wave in a Shock Tube C. F. Lee

1977. Book No. G00119, 98pgs. $16.00, ASME members $8.00

A S M E Order Department — 3 4 5 East 4 7 t h Street — N e w York, N.Y. 1 0 0 1 7

Journal of Fluids Engineering SEPTEMBER 1977 / 555

Downloaded 09 Dec 2010 to 82.137.208.59. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm

Vous aimerez peut-être aussi