Professional Documents
Culture Documents
REVIEWS Further
Quick links to online content
Ann. Rev. Fluid Meeh. 1981. 13:131-56
Copyright <!l> 1981 by Annual Reviews 1nc. All rights reserved
IN TURBULENCE
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
MEASUREMENT
R. A. Antonia
Department of Mechanical Engineering, University of Newcastle, N.S.W., 2308
Australia
. 1 N
R(x,6x,7j)= N-+oo
hm
N
� c(x,t;)!(x, dx,tj+7j) (1)
i= 1
and
N N
in= lim �(1-I)1/�(1-I). (4)
N_oo i=1 ;=1
M
<f(x,'r.;)= � f(x, tm+1j)/M (5)
m=l
where tm are the points in time when detection occurs and M is the
maximum number of events detected. When turbulent and nonturbulent
regions are the events under study, the values of <1>' obtained when 1m
are the positions of the fronts and backs of turbulent regions, are
denoted here by <fr> and <fb> and represent the point averages
mentioned previously.
The overall and rather simplified scheme of Figure 1 is sufficiently
general to include most, if not all, conditional sampling and averap.ing
procedures used in practice. Of course, such generality means that it is
of little use in guiding us in individual cases, especially with respect to
the choice of parameters, threshold C, and operations required to form
d. A commonly used parameter is 'TH' a smoothing or hold-time which
should not be much larger than the smallest time scale of interest for the
event under study ( 'rH is sometimes applied after comparison with C).
An option (not included in Figure 1) available to the experimenter is to
adjust one or several of the parameters and/or C after comparing c with
the turbulence signal(s) in the detection scheme. Note also that c can be
used in a more practical sense than suggested by Figure 1 (it can for
example, be applied to flow visualization).
2 TURBULENT-NONTURBULENT INTERFACE
DETECTION AND RESULTS
(a) The physical difficulty that the interface is highly re-entrant (smoke
photographs of Fiedler & Head 1966 and Falco 1977 reveal the
highly contorted three-dimensional behavior of the interface);
(b) The mathematical difficulty that the criterion function, supposed
to be positive within the turbulence, can have occasional zeroes;
Access provided by Universite Lille 1 Sciences et Technologies on 05/03/17. For personal use only.
(e) The operational difficulty that this criterion function may respond
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
0-4
Access provided by Universite Lille 1 Sciences et Technologies on 05/03/17. For personal use only.
>
I
:::;
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
iii
«
�0-1
a.
o -2-0 I
a
_. •
20 4-I0
9/e•
p/=yp]+(I-y)it. (6)
region have confirmed the expectation that (uv) t =uv/r, with negligible
contribution provided by the irrotational flow. The previously noted
quantitative agreement between different measurements of <Vr> and
<Vb> does not extend to point-averaged streamwise velocities, perhaps
as a result of the less pronounced changes in U, in comparison to V or T
at the back of the bulge. The bulk of the available data seems to be
consistent with the observation that <Ur> is generally slightly larger
Access provided by Universite Lille 1 Sciences et Technologies on 05/03/17. For personal use only.
than <Ub)'
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
tional profiles, a result that seems consistent with the good similarity
exhibited by y profiles obtained at different streamwise stations in the
flow. Equivalent observations have been made by LaRue & Libby
( l974a, b) in a wakel and Subramanian & Antonia (1979) who measured
turbulent zone-averaged mean-temperature profiles in a boundary layer
at different Reynolds numbers. Jenkins & Goldschmidt's (1974) point
averaged velocity and temperature profiles, measured relative to the
interface at different values of y and different streamwise stations, also
exhibited good similarity. LaRue & Libby (1978) reported that similar
ity within the turbulent fluid exists in even greater detail than the
previous measurements suggest. They found, using the range
conditioning technique, that the mean and rms temperature distribu
tions within bulges (or "structures" in their terminology) of essentially
the same duration exhibit similarity in the central intermittent region of
the wake. Bulges of relatively large duration (lc/2Uw where Ic is an
appropriate characteristic length) have an exponential ramplike be
havior with the maximum temperature occurring near the back of the
bulge. LaRue & Libby identified three regions for these bulges: a
central region where gradients of the temperature moments are rela
tively small and two interfacial regions, typically of thickness 1011 (11 is
the Kolmogorov microscale), associated with large gradients. Recently,
Sreenivasan et al (1979) used this range-conditioning technique in the
intermittent region of a round jet for bulges of relatively large duration
(these bulges were identified, with reservation, with signatures of the
large structure of the flow). The temperature distribution bears close
similarity to that of LaRue & Libby (1978), except of course that the
maximum temperature now occurs at the front instead of the back of
the bulge. The front and back of the bulge have positive and negative
normal velocities, emphasizing the large-scale vortical motion within the
3 CONDITIONAL-SAMPLING QUADRANT
ANALYSES
the Reynolds shear stress during the cycle of events observed in the wall
region of turbulent-boundary-layer and pipe flows has often been used
in both wall-bounded and free turbulent shear flows. The cycle of
events includes the burst [a three-stage process described by Kim et al
(1971), by the formation of a low-speed streak, its liftup and oscillation,
and finally its breakup] and the sweep (Corino & Brodkey 1969, also
Offen & Kline 1974). Essentially, the technique sorts contributions to Uv
Access provided by Universite Lille 1 Sciences et Technologies on 05/03/17. For personal use only.
[
in ( I) is, in the notation of Lu & Willmarth (1973),
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
� (H, li)=
[
lent ejections and sweeps. Specifically, hi was replaced by Sj defined as
than the values determined by Lu & Willmarth. The period te was found
to be correspondingly smaller than that obtained by Lu & Wt11marth
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
been used in the core region of a pipe flow by Sabot & Comte-Bellot
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
(1976) to identify ejections from opposite halves of the flow. It has also
been used in a slightly heated circular jet (Sreenivasan & Antonia 1979)
where the quadrants (u> O, v> O) and (u<O, v<O) were identified with
ejections of high-momentum fluid and sweeps of relatively low
momentum fluid. At a radial position where the production of turbulent
energy is maximum, quadrant occupancy times and contributions from
ejections and sweeps are consistent with results of Brodkey et al (1974)
near y + = 15. Measurements of p( u, v) and p( v, fJ) in the turbulent
region only of the flow indicated that the departure from Gaussianity is
less pronounced for the outward than for the inward radial motion
(Antonia & Sreenivasan 1977).
Sweeps and ejections have also been studied by Zaric (1974, 1975)
with a different conditional-averaging technique, the conditioning func
tion being based on the product of u and au/at. Little usage has been
made of conditional sampling for the study of the Reynolds-number
dependent, intermittent fine-structure regions of turbulence. Kuo &
Corrsin (1972) attempted to identify the geometrical character of such
regions, dispersed in nearly isotropic grid turbulence, using two-point
coincidence functions of an intermittency function (see Kuo & Corrsin
1971 for details) set equal to 1 and 0 depending on when the hot wire is
in a fine-structure region and when it is not. The regions were found to
be more nearly like rods than blobs or slabs. Coincidence measure
ments, with a spatial array of wires, may, with appropriate conditional
averaging, yield interesting information about the geometry of the fine
structure in a turbulent shear flow and its possible interaction with the
large structure.
4 INTERACTING LAYERS
Internal Layers
Johnson (1959) first observed the temperature intermittency of the
internal layer that grows within a boundary layer subjected to a step
change in surface temperature. Antonia et al (1977) examined the
response of a turbulent boundary layer to a sudden increase in surface
heat flux (zero heat flux upstream) and presented statistics of the
thermal interface and conditional measurements with respect to this
interface. Near the step, the hot-zone-averaged Reynolds shear stress and
heat flux are considerably larger than conventional averages (when the
thermal intermittency factor is small) as temperature acts as an effective
marker of fluid ejected from the wall region. Zone- and point-averaged
measurements, with respect to both inner and outer thermal interfaces,
were made by Charnay et al (1979) in a turbulent boundary layer at one
station downstream of a sudden decreastl in wall temperature.
increase, respectively, with distance from the grid. The standard devia
tion (and frequency) of the thermal interface increased as the free-stream
turbulence level increased, a result supported by photographs (e.g.
Bradshaw 1977b) of a smoke-filled boundary layer. The importance of
the external-turbulence length scale was demonstrated by Hancock
(1 978) who also used temperature-conditioned sampling in a heated
boundary layer subjected to an external turbulent free stream with a
Access provided by Universite Lille 1 Sciences et Technologies on 05/03/17. For personal use only.
scale of the event under study). Reference times for each burst were
obtained whenever the VITA variance signal exceeded a certain
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
threshold value. Note that this type of detection (and the use of
threshold and hold times) is not very different from that used in Section
2. Once the reference times were identified, conditional sampling was
performed following the procedure outlined in Section 1. Using a rake
of 1 0 hot wires, Blackwelder & Kaplan obtained conditionally averaged
velocity profiles in the region 0 <y+ < 100 with the detection criterion
applied to the signal aty+ = 1 5. An inflexional profile (observed earlier
by Kim et al 1 97 1 and Corino & Brodkey 1969) was evident im
mediately before detection. The conditionally averaged Reynolds shear
stress was approximately an order of magnitude greater than its conven
tionally averaged value and decayed slowly downstream after correcting
for the random-convection velocity. Using the criterion that the filtered
velocity at two side-by-side points (y+!:::: 1 6) should be simultaneously
low and decreasing, Willmarth & Lu (1972) identified individual con
tributions to uv as large as 62uv. Recently, Blackwelder & Eckelmann
( 1979) used quadrant-probability analysis and conditional-sampling
techniques in the vicinity of the wall to infer that pairs of counter
rotating streamwise vortices are responsible for the ejection and forma
tion of streaks of low-speed fluid. The VITA technique had been earlier
applied to signals from an array of hot wires, mounted very near the
surface in a spanwise direction, by Gupta et al (1971). It was established
that the characteristic wavelength A of the streaky sublayer structure is
given by A + 100, consistent with the results of Kline et al (1967). For
==
y + < 1 00, Blackwelder & Kaplan (1976) found that their detection
frequency was in reasonable agreement with the "pulse" frequency of
Rao et al (197 1) and scaled on outer-flow variables. This detection
frequency was also found (Blackwelder & Woo 1974) to be independent
of the frequency of a periodic pressure perturbation applied in the
irrotational region outside the turbulent-nonturbulent interface.
Wallace et al ( 1977) deVeloped a scheme to detect and ensemble
average patterns that occur repeatedly in the u signal and are char
acterized by a relatively weak deceleration followed by a strong acceler
ation. In the region 10 <y + < 30, this pattern occurred over 65% of the
CONDITIONAL SAMPUNG 147
duction. The v signal was on average 1800 out-of-phase with the u signal
(as noted by Blackwelder & Kaplan 1976) and the ejection was found to
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
2A similar observation was found by van de Yen (reported by Beljaars 1979) who used
an analogous detection method.
148 ANTONIA
detection criterion and only a few iterations are necessary for conver
gence. This type of procedure has been used, with minor variations, by
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
of u near the core and near the low-speed side of the mixing layer of a
circular jet to identify the passage of the structure. Browand & Weid
man (1976) used combined flow visualization and hot-film measure
ments to obtain < u ) , < v ) , < uv ) , and isovorticity contours, during and
after pairing. Significant contributions to uv occurred during pairing.
Rajagopalan & Antonia (1979b) obtained <u ) , <v>, <fJ> , <uv > , and
< vfJ ) using a detection criterion based on fJ in the mixing layer of a
Access provided by Universite Lille 1 Sciences et Technologies on 05/03/17. For personal use only.
6 CONCLUDING REMARKS
ated with the events may also be possible provided a suitable model
equation for the conditioning function can be written.
ACKNOWLEDGMENTS
Literature Cited
Alcaraz, E., Charnay, G., Mathieu, J. 1 977. Antonia, R . A., Atkinson, J . D. 1973. High
Measurements in a wall jet over a con order moments of Reynolds shear stress
vex surface. Phys. Fluids 20:203-210 fluctuations in a turbulent boundary
Ali, S. F., Kovasznay, L. S. G. 1975. Struc layer. J. Fluid Mech . 58:581-93
ture of the turbulence in the plane wake Antonia, R. A., Atkinson, J. D. 1974. Use
behind a heated flat plate. Tech. Rep. of a pseudo-turbulent signal to calibrate
75-2. The Johns Hopkins University, an intermittency measuring circuit. J.
Baltimore, Md. Fluid Mech . 64:679-99
Anderson, P., LaRue, J. C., Libby, P. A. Antonia, R. A., Bradshaw, P. 197 1 . Condi
1979. Preferential entrainment in a two tional sampling of turbulent shear flows.
dimensional turbulent jet in a moving I.e. Aero Rep. 71-04. Imperial College,
stream. Phys. Fluids 22: 1857-61 London
Andreopoulos, J. 1979. Turbulent near Antonia, R. A., Danh, H. Q., Prabhu, A.
wake of a thin aerofoil. Proc. Symp. 1 977. Response of a turbulent boundary
Turbulent Shear Flows, 2 nd, London , pp. layer to a step change in surface heat
2. 1-2.5 flux. J. Fluid Mech. 80: 153-77
Antonia, R. A. 1972. Conditionally sam Antonia, R. A., Prabhu, A., Stephenson, S.
pled measurements near the outer edge E. 1975. Conditionally sampled measure
of a turbulent boundary layer. J. Fluid ments in a heated turbulent jet. J. Fluid
Mech . 56: 1 - 1 8 Mech . 72:455-80
1 52 ANTONIA
turbulent flow. J. Fluid Mech . 37: 1 - 30 boundary layer. J. Fluid Mech . 25 :719-
Corrsin, S. 1943. Investigation of flow in an 35
axially symmetrical heated jet of air. Grass, A. J. 197 1 . Structural features of
Annu. Rev. Fluid Mech. 1981.13:131-156. Downloaded from www.annualreviews.org
G. K. Patterson, 1. L. Zakin, pp. 274- 83. laRue, J. C., Libby, P. A. 1978. Detailed
Univ. Missouri- Rolla similarity in the turbulent wake of a
Kaplan, R. E., Laufer, J. 1 968. The inter heated cylinder. Phys. Fluids 2 1 :891 -97
mittently turbulent region of the Lau, J. C., Fisher, M. J. 1 975. The vortex
boundary layer. Rep. USC AE 110. Univ. street structure of turbulent jets. Part I.
S. Calif., Los Angeles J. Fluid Mech. 67:299-337
KawaU, J. G., Keffer, J. F. 1979. Interface Laufer, J. 1975. New trends in experimen
statistics of a uniformly distorted heated tal turbulence research. Ann. Rev. Fluid
turbulent wake. Phys. Fluids 22:31-9 Mech . 7: 307-26
Keffer, J . F., Olsen, G . J., Kawali, J . G. Libby, P. A. 1975. On the prediction of
Access provided by Universite Lille 1 Sciences et Technologies on 05/03/17. For personal use only.
Kibens, V., Kovasznay, L. S. G., Oswald, Libby, P. A. 1976. Prediction of the inter
L. J. 1974. Turbulent-non-turbulent in mittent turbulent wake of a heated cyl
terface detector. Rev. Sci. Instrum . inder. Phys. Fluids 19:494- 501
45 : 1 138-44 Lu, S. S., Willmarth, W. W. 1973. Measure
Kim, H. T., Kline, S. J., Reynolds, W. C. ments of the structure of the Reynolds
1971. The production of turbulence near stress in a turbulent boundary layer. J.
a smooth wall in a turbulent boundary Fluid Mech . 60:48 1 - 5 12
layer. J. Fluid Mech . 50:133-60 Mown, J. N., Kawall, J. G., Keffer, J. F.
Kline, S. J., Reynolds, W. C., Schraub, F. 1 979. Structural features of the plane
A., Runstadler, P. W. 1967. The struc turbulent jet. Phys. Fluids 22: 1 240-44
ture of turbulent boundary layers. J. Mulej, D. J., Goldschmidt, V. W. 1 975.
Fluid Mech. 30:741 -73 Velocity and foldover of the turbulent
Kovasznay, L. S. G. 1979. Measurement in non-turbulent interface in a plane jet.
intermittent and periodic flow. See Rep. HL 75-32. Purdue Univ., Lafayette,
Blackwelder 1979, pp. 133-59 Ind.
Kovasznay, L. S. G., Ali, S. F. 1974. Struc Murlis, J. 1 975. The structure of a turbulent
ture of the turbulence in the wake of a boundary layer at low Reynolds number.
heated flat plate. Proc. Int. Heat Transfer PhD thesis. Imperial College, London
Conf., 5th, Tokyo, pp. 99 - 103 Nakagawa, H., Nezu, I. 1977. Prediction of
Kovasznay, L. S. G., Kibens, V., Black the contributions to the Reynolds stress
welder, R. F. 1970. Large-scale motion from bursting events in open-channel
in the intermittent region of a turbulent flows. J. Fluid Mech . 80:99- 128
boundary layer. J. Fluid Mech. 41 :283- Nychas, S. G., Hershey, H. C., Brodkey, R.
325 S. 1973. A visual study of turbulent shear
Kuo, A. Y-S., Corrsin, S. 1971. Experi flow. J. Fluid Mech . 6 1 : 5 13-40
ments on internal intermittency and fine Offen, G. R., Kline, S. J. 1973. A compari
structure distribution functions in fully son and analysis of detection methods
turbulent fluid. J. Fluid Mech. 50:285- for the measurement of production in a
320 boundary layer. See Kaplan 1973, pp.
Kuo, A. Y-S., Corrsin, S. 1972. Experiment 289- 3 1 8
on the geometry of the fine-structure re Offen, G . R., Kline, S . J . 1974. Combined
gions in fully turbulent fluid. J. Fluid. dye-streak and hydrogen bubble visual
Mech. 56:447-79 observations of a turbulent boundary
laRue, J. C. 1974. Detection of the turbu layer. J. Fluid Mech . 62:223-39
lent-nonturbulent interface in slightly Oswald, L. J., Kibens, V. 197 1 . Turbulent
heated turbulent shear flows. Phys. Fluids flow in the wake of a disk. Tech . Rep.
17: 1 5 13- 17 002820. Univ. Mich., Ann Arbor, Mich.
laRue, J. C., Libby, P. A. 1974a. Tempera Paizis, S. T., Schwarz, W. H. 1 974. An
ture and intermittency in the turbulent investigation of the topography and mo
wake of a heated cylinder. Phys. Fluids tion of the turbulent interface. J. Fluid
17:873-78 Mech. 63 : 3 15-43
laRue, J. C., Libby, P. A. 1974b. Tempera Paizis, S. T., Schwarz, W. H. 1975. Entrain
ture fluctuations in the plane turbulent ment rates in turbulent shear flows. J.
wake. Phys. Fluids 17: 1956-67 Fluid Mech . 68:297-308
laRue, J. c., Libby, P. A. 1 976. Statistical Perry, A. E., Hoffman, P. H. 1976. An
properties of the interface in the turbu experimental study of turbulent convec
lent wake of a heated cylinder. Phys. tive heat transfer from a flat plate. J.
Fluids 19: 1 864-75 Fluid Mech . 77:355-68
CONDITIONAL SAMPUNG 155