Sie sind auf Seite 1von 10

Experiments in Fluids 4:97-106 (1986)

Expemnentsin Flui
Springer-Verlag 1986

The structure of the turbulent boundary layer over a mobile and deformable boundary *
Y. Papadimitrakis**, En Hsu, and R. Street
Environmental Fluid Mechanics Lab, Dept. of Civil Engineering, Stanford University, Stanford, CA 94305, USA

Abstract. The structure of the velocity field above a propagating water wave of fixed frequency was investigated in order to evaluate the transport of wind momentum to water waves and the influence of a mobile and deformable boundary on the bursting cycle. The vertical and horizontal velocities were measured in a transformed Eulerian wave-following frame of reference with the aid of a cross hot film, in a wind-wave research facility at Stanford University. The mean velocity profiles have a log-linear form with a wake free-stream characteristic. The wave-coherent motion in the freestream is irrotational; in the boundary layer, it has a strong shear behavior related to the wave-associated stress. The wave-induced velocity field and the wave-perturbed turbulence depend strongly on the ratio of the wave-speed to the mean free-stream velocity,
c/U&. '

The presence of the propagating waves affects the bursting cycle, making the contribution of sweeps and ejections almost equal and dependent on the ratio c/Uao. The magnitudes of the contribution of the bursting events are generally enhanced by the presence of water waves. The time interval between ejections or sweeps does not scale with either the inner and/or outer flow variables.

1 Introduction
The knowledge of the detailed structure of the velocity field above water waves is valuable for understanding a variety of different problems such as: 1. the generation and growth of water waves; 2. the gas and heat exchange processes at the air-water interface; and 3. the transport of pollutants in the atmospheric boundary layer and water enclosures. Although there exist several experimental and theoretical investigations of both the pressure and velocity field in the air boundary layer above water waves, their results show considerable disagreement. Among them, only a few have considered the Reynolds stress production mechanism in the water proximity. Considerable theoretical and experimental work, however, using both flow * This paper was presented at the Ninth Symposium on Turbulence, University of Missouri-Rolla, October 1-3, 1984 ** Now at the College of Marine Studies, University of Delaware, Lewes, Delaware 19958

visualization and either velocity or pressure sensors, has been done for wall flows. Experimental studies of the bursting phenomenon, a process responsible for most of the Reynolds stress production in these flows, generally concentrate on determining the average spatial and temporal scales of ejection and sweep events. Although the spatial configuration and scales of the wall structure are generally agreed upon by different investigators, there is considerable disagreement concerning the frequency of occurrence and the scaling of the bursting events in bounded shear flows. These differences have been attributed to the different detection schemes; Bogard and Tiederman (1981) have suggested that the probe measurements probably detect something different from the bursts identified using flow-visualization techniques. Previous field and laboratory measurements of the velocity and pressure fields over water waves, conducted by Kendall (1970), Takeuchi etal. (1977), Hsu eta!. (1981) and Hsu and Hsu (1983), have shown considerable alteration of the wind turbulence due to the presence of waves. Therefore, it will be of great importance to know the characteristics of the bursting cycle under the presence of a mobile and deformable boundary. In this study, the velocity field was measured with an "X" hot film mounted on a wave-following device operating in a transformed coordinate system similar to that used by Hsu et ai. (1981), and Hsu and Hsu (1983). This paper, being part of the doctoral work of the first author (Papadimitrak!s 1982), describes in detail the structures of the mean, waveinduced, and turbulent velocity fields in the air, and the influence of water waves on the turbulent Reynolds stress production mechanism and on the bursting-related phenomenon.

2 Experimental apparatus and data analysis


The experiments were conducted in the Stanford windwave research facility (Hsu 1965). The depth of the air flow H (= 260) measured from the mean water level

98 (MWL) to the channel roof was 1.07 m. The water depth d was 0.83 m. The 1 Hz mechanically-generated wave was in deep water, with a wave length L = 1.56m, a wave number k = 4.03 m -I and a phase speed c = 1.56 ms -l. The air flow above the propagating wave is considered to be two-dimensional. In the Cartesian coordinate system used, x is measured in the wave-propagation or mean air-flow direction and y is the vertical coordinate measured upward from the MWL. The coordinate transformation used contains only vertical translation; namely,
t=t*, f(y*)= x=x*, Y=Y*+f(Y*)O

Experiments in Fluids 4 (1986) transformation (FFT) to determine the amplitude and phase of each harmonic contained in ffi (Hsu et al. 1981). The bursting events were identified by classification of the measured Reynolds stress u v according to the signs of u (= ~ + u') and v (= ~r+ v'). Brodkey et al.'s (1974) terminology was used in this study to characterize the various events of the cycle. For comparison of our results with others, the stress uv was further classified according to the "hole" method introduced by Lu and Willmarth (1973).

(2.1a, b,c) (2.2)

3 Results 3.1 Mean statistics

sinh ( k H - k y* ) sinh (k H )

and has been described by Hsu et al. (1981); F/represents the sinusoidal water surface displacement from the MWL. Wave-height gauges were used to obtain the wave characteristics and drive the wave follower. They were o f capacitance type with an accuracy in static and dynamic calibration tests of __ 0.25 ram. The velocity field in the air was measured with a hot film. It was a TSI Model 1248-10 end-flow "X" probe, 0.0254 m m in diameter, 0.508 m m long, with a flat frequency response up to 40,000 Hz; it was operated in a constant-temperature mode. For the wind speeds examined in this investigation, its viscous length 1 was in the range 1.6 < / + < 5.0. The uncertainty for the mean velocity was 3%. However, for the measured turbulent quantities such as turbulent Reynolds stresses, it was higher and was found to be approximately 10%. The hot-film signals were zero-suppressed, amplified, and low-pass filtered at 250 Hz to fulfill the Nyquist criterion, as samples were taken every 0.002sec for 3 minutes and recorded on a tape for later analysis. The data taken correspond to seven different free-stream mean wind velocities in the range 140-400 cm/sec, with I Hz 2.54 cm nominal amplitude, mechanically-generated waves. They were collected at 20 or 21 elevations, ranging from 0.75 and 53.3 cm above the water surface. Since the instantaneous velocity field above the waves consists of a mean, a wave-induced perturbation, and a turbulent component, viz.
ui = Ui + t2g+ u}

The experimental data for the mean velocities were curvefitted, in a least-squares sense, to the wake log-linear expression: U~0- U H, 1 In )4 + 1 + cos (3.1)

to determine 5, u , , and ~(x); )4, 5, and ~(x) are the Von Karman constant, boundary-layer thickness, and the wake parameter, respectively. In wall coordinates, the above expression yields: u+ = 1 l n y + + C + )4 with u_ U., b/, 6 + = cSu, ," y + _ y * u , v V -2 n(x) - (3.3a, b, c, d) 1 - cos (3.2)

)4

\-U-j/

+_lln6 C = u~0

(i = 1, 2),

(2.3)

Table 1 lists the determined parameters u , , n (x), C, c5 +, and Fig. 1 shows the mean wind profiles in wall coordinates (u + vs. In y+). The lower part represents the logarithmic region of a typical turbulent-boundary-layer velocity profile, while the upper part shows the wake characteristic. For the higher wind speeds, the wake becomes more pronounced, the elevation where it begins to be felt, lower, and the lower portion of the profile close to the airwater interface deviates from the logarithmic law. There,
Table 1. Mean velocity profile characteristics

familiar time and phase averages were used to extract the wave-induced fluctuations from the total velocity signals. The vertical wave-induced velocity was corrected to account for the spurious component introduced by the wavefollower motion. Cross- and auto-spectral analysis for ~i, using ~/ as a reference, were performed by fast-Fourier

U~o (cm/sec) 141 179 200 231 346 402

u, (cm/sec) 4.25 5.52 6.20 7.24 11.91 15.57

7~(x) 0.53 0.34 0.30 0.43 0.31 0.26

C 12.76 12.35 11.87 10.77 7.48 4.12

~+ 1,453 1,886 1,922 2,425 2,725 3,556

1 Defined as l + = l" u,/v where l, u, and v are probe length, air friction velocity and kinematic viscosity of the air, respectively

Y. Papadimitrakis et al.: Turbulent boundary over a mobile and deformable boundary


35

99
I I I I I

I
O

I
U~o

I
= 141
cm

I
s "1

I
=

IIII 47,500 60,300 67,600 77,900 = 115,700 = 135,700 o oA

IIII o o o &[~ %, V

ReSo

30 [] %, O

= 179
= = = = 200 231 346 402

oA AD
[] o A D v '7 v v v O O0 O 0 O O O v V O v

v
0 O O O O0 0

oo o o ~
~

R e = U 6 ~o/t~ so o u+ 26

0 0
0

oo

~G~
v 20 %,

oo 15 101 Fig. 1.

Fig. 1. Mean velocity I


y

I IIIII
lO 2

I 111111
lO 3

I I I II
lO 4

profiles

70 60 50 40 30 20 I0

f
0

)O)I

14

14
[2

Re~o=135,700"

12
I0

P~%
@ "-,,2X I 0

U~ c U'U' o Q o

= 2.58
,

oo

- - -.---f X

Io~

I0 8

vv

e
8 6
Q O Q

Uso

4 2
Q

o Q o
I 0.2 I 0.4 I 0.6 I~ 0.8

Q 0

I 0.2

I 0.4

I 0.6 Y/80

el~ 0.8

I 0.2

I 0.4

I O/ 0.6 v 8

O0

Fig. 2. Typical m e a n turbulent Reynolds stress distribution


1.0

Y/8o

Y/8o
order larger than those o f - u' v' and v' v'. An almost constant stress layer near the wave surface is observed in each profile of - u ' v'. The decrease of shear stress at the lowest portion of the constant stress layer can be attributed to the damping of turbulence by viscosity. The non-normalized mean turbulent Reynolds stresses tend to increase with wind speed. Figure 3 shows a typical distribution of the mean wave-associated Reynolds stress ui uj ( - 10%) as a function of y*/6o. As this figure shows, the sign of the stress changes throughout the boundary layer, in agreement with the observations of Hsu et al. (1981), and Hsu and Hsu (1983). For c~ U~0 = 0.45 and 0.39, ~7v-remains surprisingly enough positive. This picture suggests that the energy drains either from the mean field to the waves or vice versa, depending on c~ U~o and the specific location within the boundary layer. Kendall (1970) and, with some exceptions, Takeuchi etal. (1977) found the wave-associated stress to be positive and negative below and above the critical height (where U = c). This contradicts the results of Lai and Shemdin (1971), who found large negative contributions to the u, v cospectra (a downward flux) at the wave frequency when c/U~o< 1. Our data support Takeuchi etal. (1977) measurements and, with some exceptions, those of Kendall (1970).

the velocity magnitudes are larger than those predicted by the log law. This deviation may be attributed to the generation of the surface drift current, whose magnitude is proportional to the wind speed (Wu 1975). The wake characteristic of the mean velocity profiles was also observed by Hsu et al. (1981), and Hsu and Hsu (1983). The variation of the intercepts of these profiles with increasing wind speed or Reynolds number, as also indicated by the decreasing values of constant C, can be attributed to the changing surface roughness condition (see also section 3.6). For low wind speeds the flow becomes super smooth due to the relative surface stress decrease caused by the drift current, resulting in values of C greater than their fixed-surface counterparts. For high wind speeds (c/U~o < 0.45), ripples become prominent surface roughness on the water waves causing a shear stress increase which is greater than the corresponding stress reduction produced by the drift current, resulting in smaller values of C.
3.2 Mean turbulent and mean-wave associated Reynolds stresses

Typical mean turbulent Reynolds stress profiles are shown in Fig. 2. The magnitudes of u' u' are approximately one

100
35 30 25 20 15 10 5 0
I i ! I

Experiments in Fluids 4 (1986)


12 I0 8 6 4 2 0 -2 0 _o o
- C) i I i I I 4 I I I I

R%=7 7,90, U 0 ~o= 1.48 c dU x[O 4

12 I0 UV 4 US- x l O ~0 8 6
4 q

So

vv ,-T~- x i 0 4 USo

g
% %
1%06 o P e l n 0.2 04 0.6 0.8 1.0 Y'~8o

% @

oo o 0 0o 0
I I I

0
I

0 1.0

2 0 0 0.2

o o

0.2

0.4

0.6

0.8

0.4

0.6

? io
0.8
1.0

Fig. 3. Typical mean wave-associated Reynolds stress distribution

Y/8o

Y 18o

2OO I00
<~'7) o ( Cm 2/sec2_)I 0 0

eo ooo
o oo cpe e

Uso/c = 2.58 o
oo~

- 2 0 0 ~-0 1 0.2
I I I

0.4

0.6

0.8

1.0

t (see) Fig. 4. Typical Ev"phase-averaged distribution

stress distribution for the various wind speeds considered in this study. Typical phase-averaged results for the u' v' stress at the lowest point of measurements, shown in Fig. 5, demonstrate that this stress is relatively high on the windward side of the water wave, with the exception of lowest wind speed run, where c/U6o> 1. This behavior is consistent with the measurements of Okuda et al. (1977), Hsu et al. (1981), Okuda (1982) and the prediction of Gent and Tailor (1976), and can be attributed to the response of turbulence to a varying mean flow. 3.4 Wave-induced Reynolds stresses 2

20( I0( ( u' v ' ) (cm2/sec z) 0 -I00 -20C

c/U8,= 0.39

Re6o= 135,700

.........

GO OJ~Q OOo O ~ O G (9 OO O o ~ O O OO OO O O @ O o- - o - . . . . . . . . . . . . . . . . Q__e__ Q GO Q

~o20 Q ~ O ~ O O O

0.0

0.2

I
0.4 t
(sec)

0.6

0.8

1.0

Fig. 5. Typical

u' v' phase-averaged

distribution

Figure 6 shows a typical amplitude (+ 8%) and phase (_ 10 ) distribution of these stresses. The amplitudes Ii/I are large near the interface and decrease as y*/5o increases. This decrease, however, is not monotonic, but shows an oscillatory behavior. A dip is occasionally observed in the i~i/I profiles with a corresponding phase jump, but no consistent pattern can be discerned. The oscillatory behavior of the I/zi;I amplitudes was also found by Gent and Taylor (1976) in calculations of the interface flow in a curvilinear coordinate system under finiteamplitude wave conditions. A consistent increase of the peak value of ]~i/] with increasing wind speed was also found. The decrease in ]By] as 9 / 6 o ~ 0 is again due to the viscosity damping. 3.5 Classification of Reynolds stresses Figure 7 shows the fractional contributions u vi/u v to the total mean Reynolds stress u v at the lowest point of measurements, as a function of wind speed; u vi represents the contribution of the i th quadrant. It is seen that the second quadrant (ejections) is the largest contributor to the ~ - s t r e s s , with the fourth quadrant (sweeps) the second largest. The contributions from outward and wallward interactions (first and third quadrants) are also 2 Define as: ?)/= <u~ u~) - u~ u~

3.3 Phase-averaged wave-associated and wave-induced Reynolds stresses The main component (fundamental mode) of the phaseaveraged wave-associated stress (~i uj) appears at twice the fundamental water wave frequency, as it can readily be seen from the relationship.
1 (~e~j)=~lc~,ll~jlcos(o~t-o~,)cos(~ot-o~,).

(3.4)

Therefore, the harmonics of the (ffi if/) stress appear at frequencies 4, 6, 8 Hz, etc., for a 1 Hz mechanicallygenerated water wave. This behavior is apparently shown in Fig. 4, representing a typical phase averaged ff~

Y. Papadimitrakis et al.: Turbulent boundary over a mobile and deformable boundary


25 20 A
I rll_~l I 5 i i i i

101

R else= 771900

0,360 270

0g o

= 1A8

EI) Q Q D 13 Q

Uso2x

10410
o

07.
13 I 13 I 13 I
-)t

180
9O

5 0 0 0.2 0.4

Q I

13

0,360 270 1.0 0


I I I I

0.6

0.8

0.2

Y 18o

0.4 0.6 0.8 y')$/8

1.0

I0
8

Re 8 = 7 7 , 9 0 0 _ ^ 6
4

90

~o13

Q 13

Uso= 1.48 c

0,360

-I - r121 O xl

0~12
o
I

270 180 90

13

2 0 0

13

13 o I I

0,360 1,0 0

0.2

0.4

0.6

0.8

0.2

0.4
y*

0.6

0.8

Y}8o
5 4
I 'r221 3 I I I I i i

/S o
i i

Re~ = 7 7 , 9 0 0 U o uS = 1.48 c

07360 270 180 130 Q 13


o o

- - x l O

U8oZ

2
I

O?2a
~2) (~eel 3
I I o
I

90

0
I

0~360 270 1.0 0


I I I -NI

0 0

Fig. 6. Typical amplitude and phase distributions of Vii , F12, and F22
1.0

0.2

0.4

0.6

0.6

0.2

0.4

0.6

0.8

YI8 o

Y 18o

2.0 "-r~/ 1.6 1.2 0.8 0.4


--

"B'

~a B, e 'e

UVi UV

--//-

-0,4 -0.8 -1.2 -1.6 -2.0 L.L// 0.2 ,B


B' B"

014

016 c U8o

018

1.10

1.2

Fig. 7. Fractional contributions uvi/~ at the lowest point of measurements: y*/&0= 0.014; 0.015; 0.015; 0.015; 0.018; 0.019; 0.019 in order of increasing wind speed. ~ 1st quadrant; "~ 2nd quadrant; j~ 3rd quadrant; R 4th quadrant

significant, b u t s m a l l e r . For c/U6o> 0.68, ejections contribute on the average 90% to the total mean stress. This result agrees with the smooth-wall turbulent boundary layer behavior observed by Corino and Brodkey (1969), Kim et al. (1971), Lu and Willmarth (1973), and others. Indeed, this fractional contribution is somewhat greater than the value 77% or 80% reported by Kim et al. (1971), and Lu and Willmarth (1973), respectively. In the same range of wind speeds, sweeps contribute about 77% to the total mean stress, leaving - 67% to the other two negative contributors. As can be seen from Fig. 7 both first and third quadrants have identical contributions (34% and 33%, respectively). Lu and Willmarth (1973) have reported values of 55% and - 32% for the contributions of sweeps, wallward, and outward interactions. The increased sweep contributions were also observed by Nakagawa and Nezu (1977) in their rough-bed, open-channel flow measurements. They attributed this increase to the bed roughness and concluded that, "... as roughness increases and the

102 distance from the wall decreases; sweeps appear to be more important than ejections." For c/U~o < 0.68, a significant change in the behavior of fractional contributions u v i / ~ is observed. The contributions of all quadrants progressively increase, but the ratios u-vS'2/~ and ~ / ~ remain almost unaltered. Values as high as 192%, 174%, - 146% and - 120% were found at c/U~o = 0.39. While the turbulent correlation coefficient R ' = - u' v'/ [RMS ( u ' ) ' RMS (v')] was found to remain constant with wind speed, the correlation coefficient R = - u T / [RMS (u) RMS (v)] decreases progressively from about 0.3 at c/U~o~_0.68 to 0.1 at c/U~o~_0.39. This decrease is caused by the modification of the air flow field due to the wave-induced perturbations, as can be seen by expressing R in terms of the wave-coherent and turbulence stresses viz.

Experiments in Fluids 4 (1986) was also observed by Lu and Willmarth (1973), Nakagawa and Nezu (1977), and others. Close to the water surface and up to the middle of the equilibrium region (y*/6o < 0.3), the detailed distribution of the various events depends on the ratio c/U~o. Except for the highest wind speed, the intensities of all the events increase with increasing y*/6o, reach a maximum, and then either decrease until they reach a minimum or remain fairly constant. The width and location of the two extrema depend clearly on c/U~o. At the highest wind speed, the intensities decrease with increasing y*/3o, in agreement with the results of Takeuchi etal. (1977). Lu and Willmarth (1973) and Brodkey et al. (1974) reported an increase of the various intensities as y*/6o ~ O. Nakagawa and Nezu (1977) observed a similar.behavior for their smooth-bed, open channel flows, but they found a decrease in the intensity of ejections and sweeps in their rough-bed case. However, their rough bed measurements do not extend deeply enough in the wall region (down to about y*/6o ~- 0.085), and thus it is unknown whether these trends continue there. In the equilibrium region (0.1 < y*/6o < 0.6), the intensity of each event is nearly constant, irrespective of y*/6o. The rates of intensity contributed by ejections and sweeps are in excess of 100%, respectively, and the excess stress balances the sum of the contributions of inward and outward interactions. In the free-stream (y*/6o > 0.6), the intensity of each event rapidly increases with y*/3o. The same behavior, observed in the equilibrium and freestream regions, has also been reported by all the abovementioned investigators. The relation between the intensities of each event shown in the previous figures confirms the observation made by Grass (1971) that both ejections and sweeps exist irrespective of the roughnesscondition. The differences found in the water proximity are thought to be the physical consequence of the changing surface-roughness. The surface roughness condition characterized by the roughness Reynolds number Re, = u, yo/v was found to vary from aerodynamically smooth to rough in this investigation; the roughness height, y0 = I/v/'2, where ~' represents the instantaneous ripple height, was calculated from the wave-height records. Other mechanisms, depending on the dynamical behavior of the wavecoherent velocity perturbations and Reynolds number effects, are also responsible for the observed behavior. The fact that both ejections and sweeps, which are the predominant events in the bursting phenomenon, are affected by the roughness condition is very important and was discussed by Nakagawa and Nezu (1977). It is seen that the average value of the ratio ~ / u v 4 , at the lowest point of measurements, is approximately 1.15, much smaller than 1.85 reported by Lu and Willmarth (1973) close to the wall. A sharp rise near the interface is o b s e r v e d for low wind speeds, while for most of the boundary layer this ratio is nearly constant, with values between 1.05-1.20. Lu and Willmarth (1973) reported the same behavior, but they found a value of 1.35. As the

12Y+ u'v'

(3.5)

As the wind speed increases, the stress tr decreases toward zero due to the rapid decrease of Y perturbations, provided the stress u'v' doubles only at the highest wind speed. For c/U6o>0.45, the wave-associated stress remains positive close to the water surface, although small, and therefore luT] < ] u' v' [. In contrast, tg2-~ ~, and '~2 v increase with wind speed (but not monotonically), and thus the coefficient R decreases. Since the fractional contributions u vi/uv increase with decreasing R (Lu and Willmarth 1973), it becomes evident that the wavecoherent perturbations affect the distribution of these contributions for different wind conditions through the variation of the wave-associated stress, which in turn reflects the critical and viscous (Stokes) layer dynamics, as Hsu and Hsu (1983) have shown. A detailed description of both the turbulent and wave-coherent velocity fields in the air can be found in Papadimitrakis (1982). Variations in surface roughness, due to the wave growth, probably affect the skewness and diffusion factors, as also discussed by Nakagawa and Nezu (1977) for open-channel flows, and therefore alter the probability density distribution of the Reynolds stress u v, and ultimately the fractional contributions ~ / ~ , .

3. 6 Distributions of the contributions of the various events throughout the boundary layer
Figures 8 show typical distributions of the fractional contributions of the various events to the mean ( ~ ) Reynolds stress as a function of y*/6o. The distributions of positive and negative fractional contributions to u'v' throughout the boundary layer are also included in these figures (Papadimitrakis 1982). As can be seen, the intensities of the various events satisfy the relation: ejection > sweep > outward interaction > inward interaction within the observed range of y*/6o. The difference between the two interactions, however, is negligibly small. This behavior

Y. Papadimitrakis et al.: Turbulent boundary over a mobile and deformable boundary


5
I I I I

103

"t
3

c/

= 0.45

Reso 1 1 6 , 7 0 0

UBo "n

"D
,o" ~ ,o" 4~

[3"

2 I UVi ( U'V')j

0 -I -2 -3
-4
-

,o"

..o

UV

uV

-5

.0

02

0.4

0.6

0.8

1.0

Y/go
5 4.
I I I I

CA =0.78 UBo

Re8~ 6 7 , 6 0 0

3 2 EL
I

n.

n.

_ - - ,UVi . ( U'V'>i . 0 ~ I .....JU~Ja". . . . . .... UV u V


-I -2 -3
-4 -5

p'

J:

0.2

0.4

0.6

0.8

1.0

Fig. 8a and b. Distributions of ~Y~,/~ and (u~?)j/ u' v' with H = O. ~ 1st quadrant; 'El 2rid quadrant; 3rd quadrant; ~, 4th quadrant; ~ negative part; e positive part

Y/Bo
agreement with smooth, flat-wall observations. It is seen from these characteristics, as well as from the contributions of the various events to u v, that the bursting process near the free stream may consist of smoother and more isotropic events.
3. 7 Classification o f Reynolds stress according to the "hole'" method

wind speed increases, this ratio increases in the equilibrium and most of the free-stream regions, with a tendency to decrease in the centerline channel region. This decreasing behavior towards the water surface for high wind speeds agrees well with the results of Wallace et al. (1972), Brodkey et al. (1974) and Nakagawa and Nezu (1977) and may be attributed to the effect of surface roughness. Reynolds number effects may also be responsible. Away from the interface, where surface effects decrease, the ratio uv2/gF4 increases toward a value of approximately 1.4, in

Typical contributions of the various events of the cycle to the mean Reynolds stress u v as a function of the "hole"

104
2.0 1.6
i 2
X X X X 0 X X 0 X 0 1 I I

Experiments in Fluids 4 (1986)

C/uso=

0.87

Re8o--60,500

0.8 0.4.
i

"IS]

x
~ o

UVi
UV

0.0 -0.4 -0.8 - I .2


-I .6

~"

~'

-2.0 0

Fig, 9. Measurements of the contributions to fly from different events. ~" 1st quadrant; "~ 2nd quadrant; U 3rd quadrant; ~ 4th quadrant; o hole; x fraction of time spent in the "hole"

Z HOLE SIZE

3 , H

,4

size H, at the lowest point of measurements are shown in Fig. 9. The fraction of time spent in the "hole" region is also included in these figures. As can be seen from the contribution curve related to the "hole", u v has an intermittency factor of the order of 0.53. Lu and Willmarth (1973) and Nakagawa and Nezu (1977) also observed the same behavior. Ejections are always the largest contributor to ~ , with sweeps the second largest, irrespective of wind speed and "hole" size, while the contributions from u v l and u v3 are negative and relatively small. However, for c/U~<0.68 the UVl and uv3 contributions increase significantly, and they cannot be considered small. There, though the outward interaction shows slightly larger values than the inward interaction, they both become negligibly small at H ~ 3 - 4 , and consequently their contribution to the Reynolds stress disappears. In this range of H, the sweep contributions are not small but are approximately 45% of the ejection contributions. This behavior is clearly different from that observed in smoothwall boundary layers and smooth-bed open-channel flows. In the same range of "hole" size, there is a 15-25% contribution to ~ from the second quadrant, i.e., uv2/gv --0.15-0.25; the "hole" region is the other large contributor. Thus the importance of the ejection-like events in the turbulent boundary layer over progressive water waves is obvious. The pronounced change in the contributions of ejections and sweeps observed by Nakagawa and Nezu (1977) for both the smooth- and rough-bed flows, as a function of the "hole" size, and our results clearly indicate the effect of surface roughness condition on the bursting sequence.

3.8 Mean periods and scales of bursts and sweeps


The nondimensional mean time intervals T&s --- T~,s U~o/S1 between ejections and/or sweeps as a function of H and c/U~o, at the lowest point of measurements, is shown in Fig. 10; 51 denotes the displacement thickness calculated from the mean velocity data. It can be seen that these time intervals have a remarkably log-linear behavior, but they do not scale with the outer flow variables U~0 and 51, as suggested by R a t et al. (1971) and by Lu and Willmarth (1973). They also increase with both the wind speed and the "hole" size H, except probably at c/U~o= 0.68. Blackwelder and Haritonidis (1983) also found that no unique value of their threshold could be obtained with the VITA method by searching for a region where the ejection frequency (= iPb was relatively independent of -1) the threshold. These observations, however, are consistent with the findings of Chen and Blackwelder (1978), who showed that, whenever a simple threshold level is applied to a function having a continuous probability distribution, the frequency of occurrence will vary monotonically with the threshold parameter. As seen from Fig. 10, the nondimensional mean ejection and/or sweep tie intervals increase with Reynolds number for H = constant, in agreement with the conclusions of Bogard and Tiederman (1981). For H = 3 - 4 , it was found that ejections still contribute up to 15-25% to the mean Reynolds stress throughout the boundary layer, while ~ / u ~ _ _ 0 . 4 5 . Since m a x l u v ] = H l / ~ ] / ~ - , it follows that max l u v I = (H/R) I~b- I. Close to the water surface, R -~ 0.3, for low and moderate wind speeds, and

Y. Papadimitrakis et al.: Turbulent boundary over a mobile and deformable boundary


0 1 2 3 4 103

105

c/U,5 : 1.11 o : 0.87 A = 0.78 X = 0.68 [] = 0.55 v = 0.45 : 0.39 (b) ~ ~


N A

!
102

and/or outer flow variables, however, should not be surprising, because Jackson (1976) has pointed out that none of the data from geophysical boundary flows support this time scaling, at least with the outer flow parameters.

~
|

o
TS U6 -- lO ~

4. Discussion

!
v
10 ,7

~
o

o
II
100

4.1 Influence of the mobile boundary on the product(on of Reynold stresses and the bursting-related phenomenon
For H = 4, the dimensional ejection time intervals 2PB(sec) show a somewhat different behavior, as seen in Table 2, inasmuch as they slightly decrease with wind speed in the range of U&= 141-179 cm/sec, instead of increasing. It is remarkable, however, that these ejection time intervals closely correspond to the time between the largest consecutive negative or positive peaks of the phase-averaged Reynolds stresses, as can be seen by combining Figs. 4 and 5. The following simple analysis also demonstrates the validity of the above argument. To a first approximation: (uL,) ~_ ( ~ ) + ( u ' ~ , ' ) = 1~ I I~1 cos(Zcot-Oa-O,~) (4.1)

| v

~
o

o
o I01
~o m

II v
-1ffl

100 o

10"1

I
1

I
2

I
3 Hole Size, H

I
4

I
5

+ l,~121c o s ( ~ o t - % ) +1 al lel cos(0~- 0,~) + u'L,'.

Fig. 10. Mean time interval between ejections and/or sweeps at the lowest point of measurement

Table 2. Ejection and sweep time intervals

Uao(cm/sec)

141

179

200

231

285

346

402

Reo 4,180 4,620 5,100 6,460 6,050 7,500 9,150 T~ 51 66 80 109 339 611 1,065 Ts 125 201 286 354 928 1,053 1,814 iPe(H=4)(sec) 0.443 0.341 0.326 0.325 0.511 0.673 0.685

therefore, m a x ] u v I ~- 1 0 1 ~ I. For the higher wind speeds, max luv[ will be even greater due to the decreasing correlation coefficient R. These ejections are certainly violent and must come from large spikes in the u v signal. The nondimensional mean time intervals between ejections and/or sweeps T&s= TB,s" u 2 /v, scaled with the inner variables and corresponding to H = 4 were not found to remain constant, but increase with m o m e n tum thickness Reynolds number Reo as shown in Table 2, in agreement again with Bogard and Tiederman (t981). The same holds true for any other value of the threshold and/or location y*/6o throughout the boundary layer. Since the viscous lengths of the film were < 20, we have no reason to believe that our results are contaminated by wire-length effects, as discussed by Blackwelder and Haritonidis (1983) and Alfredsson and Johansson (1983). The lack of ejection and/or sweep time scaling with the inner

Therefore, the main component of the wave-associated stress ffzr appears at 2 Hz, with a time interval between successive negative or positive peaks of about 0.5 sec. The wave-coherent turbulent stress then modulates this time interval according to the phase lags of Oa, Oe and 0e1~.Thus, the time period between successive ejections in interface flows is roughly determined by both the wind conditions and the moving boundary characteristics. This conclusion sensibly agrees with Kendall's (1970) observation that "... an intrinsic time constant is an important feature of the flow", provided the turbulence structure lags behind the wave. Since only a single-frequency external wave oscillation was examined in this investigation, no definite conclusion: will be drawn on the relation between the mean ejection period and the time interval between the two consecutive largest peaks of the phase-averaged shear stress distribution. Further research is being conducted on this aspect.

5 Summary and conclusions

The results and discussions presented in the previous sections suggest the following conclusions: (1) The constant C characterizing the mean velocity profiles decreases with increasing wind speed as a result of the variation of the water surface roughness condition between smooth and fully rough regimes; accordingly, the friction velocity u, becomes smaller and/or greater than its counterpart value over a fixed surface, depending on the local wind speed.

106 (2) The log-linear form of m e a n velocity profiles, their wake free-stream characteristic, and the similarity between the m e a n turbulent Reynolds stress profiles and those observed over a flat surface clearly indicate that the transformed coordinate system given by ( 2 . 1 a - c ) describes best the air flow a b o v e water waves in the presence o f a swell. (3) The data presented show that waves influence the air flow above them. Both the wave-induced and turbulent velocity-field characteristics d e p e n d on c/U6o. The phase-averaged wave-associated stresses with their m a i n c o m p o n e n t at twice the fundamental wave frequency are significant. (4) The presence o f water waves effects the mechanism of Reynolds stress production through the variation o f surface roughness condition and the d y n a m i c s o f the wave-coherent velocity field in the air. The bursting events are generally enhanced by the presence o f water waves. T h r o u g h o u t the b o u n d a r y layer, ejections are the largest contributors to u-~, with sweeps the second largest. F o r c/U6o> 0.68, ejections and sweeps contribute a b o u t 90% and 77% to the m e a n Reynolds stress, while the outward and inward interactions contribute - 3 4 % and - 3 3 % , respectively. F o r c/U6o<0.68, a p r o n o u n c e d rise in the fractional contributions of the bursting events is observed. (5) The surface roughness augments the contribution o f sweeps and makes it a p p r o x i m a t e l y equal to the ejection contribution. It also affects the fractional contributions o f bursting events close to the water surface. (6) The m e a n t i m e interval between ejections or sweeps does not scale with either the inner o r outer flow variables. (7) It is speculated that the m e a n time interval between ejections corresponds to the time between the largest consecutive peaks o f the u v phase-averaged Reynolds stress distribution.

Experiments in Fluids 4 (1986) ceedings, Seventh Biennial Symposium on Turbulence, Sept. 21 - 23, 198 t, Univ. of Missouri, Rolla Brodkey, R. S.; Wallace, J. M.; Eckelmann, H. 1974: Some properties of truncated turbulence signals in bounded shear flows. J. Fluid Mech. 63, 209-224 Chen, C. H. P.; Blackwetder, R. F. 1978: Large-scale motion in a turbulent boundary layer: a study using temperature contamination. J. Fluid Mech. 89, 1-31 Corino, E. R.; Brodkey, R. S. 1969: A visual investigation of the wall region in turbulent flow. J. Fluid Mech. 37, 1-30 Gent, P. R.; Taylor, P. A. 1976: A numerical model of the air flow above water waves. J. Fluid Mech. 77, 105-128 Grass, A. J. 1971: Structural features of turbulent flow over smooth and rough boundaries. J. Fluid Mech. 50, 233-255 Hsu, E. Y. 1965: A wind, water-wave research facility. Civil Eng. Dept., Tech. Rep. No. 57, Stanford University Hsu, C. T.; Hsu, E. Y.; Street, R. L. 1981: On the structure of turbulent flow over a progressive water wave: theory and experiments in a transformed wave-following coordinate system. Part 1. J. Fluid Mech. 105, 87-117 Hsu, C. T.; Hsu, E. Y. 1983: On the structure of turbulent flow over a progressive water wave: theory and experiment in a transformed wave-following coordinate system. Part 2. J. Fluid Mech. 131,123-153 Jackson, R. G. 1976: Sedimentological and fluid-dynamic implications of the turbulent bursting i phenomenon in geophysical flows. J. Fluid Mech. 77, 531-560 Kendall, J. M. 1970: The turbulent boundary layer over a wall with progressive surface waves. J. Fluid Mech. 41,259-281 Kim, H. T.; Kline, S. J.; Reynolds, W. C. 1971: The production of turbulence near a smooth wall in a turbulent boundary layer. J. Fluid Mech. 50, 133-160 Lai, R. J.; Shemdin, O. H. 1971: Laboratory investigation of air turbulence above simple water waves. J. Geophys. Res. 76, 7334-7350 Lu, S. S.; Willmarth, W. W. 1973: Measurements of the structure of the Reynolds stress in a turbulent boundary layer. J. Fluid Mech. 60, 481 - 511 Nakagawa, H.; Nezu, I. 1977: Prediction of the contributions to the Reynolds stress from bursting events in open channel flows. J. Fluid Mech. 80, 99-128 Okuda, K.; Kawai, S.; Toba, Y. 1977: Measurements of skin friction flow along the surface of wind waves. J. Ocean. Soc. Japan 33, 190-198 Okuda, K. 1982: Internal flow structure of short wind waves. Part II. The streamline pattern. J. Ocean. Soc. Japan 38, 313-322 Papadimitrakis, Y. A. 1982: Velocity and pressure measurements in the turbulent boundary layer above mechanically-generated water waves. Ph.D. Dissertation, Civil Eng. Dept., Stanford University, 445 pp. Rao, K. N.; Narasimha, R.; Badri Narayanan, M. A. 1971: The bursting phenomenon in a turbulent boundary layer. J. Fluid Mech. 48, 339-352 Takeuchi, K.; Leavitt, E.; Chao, S. P. t977: Effects of water waves on the structure of turbulent shear flow. J. Fluid Mech. 80, 535- 559 Wallace, J. H.; Eckelmann, H.; Brodkey, R. S. 1972: The wall region in turbulent shear flow. J. Fluid Mech. 54, 39-48 Wu, J. 1985: Wind-induced drift currents. J. Fluid Mech. 68, 4 9 - 70

AcknowLedgement
This work was supported by the National Science Foundation through Grant NSF-CEE-7817618.

References
Alfredsson, P. H.; Johannsson, A. V. 1983: Effects of imperfect spatial resolution on measurements of wall-bounded turbulent shear flows. J. Fluid Mech. 137, 409-421 Blackwelder, R. F.; Haritonidis, J. H. 1983: Scaling of the bursting frequency in turbulent boundary layers. J. Fluid Mech. 132, 87-103 Bogard, D. G.; Tiederman, W. G. 1981: Investigation of flow visualization techniques for detecting turbulent bursts. Pro-

Received July 4, 1985

Das könnte Ihnen auch gefallen