Documente Academic
Documente Profesional
Documente Cultură
C9,PAGES22,035-22,047,
SEPTEMBER15,2000
Sean J. Bennett
NationalSedimentation
Laboratory,AgriculturalResearchService,U.S. Departmentof Agriculture,Oxford,Mississippi
22,035
22,036 VENDITTI AND BENNETT: TURBULENCE AND SUSPENDED SEDIMENT TRANSPORT
OBS B OBS A
uneProfile
ß 0.6 rn •
Figure 1. Schematic
of thedunetopography
andinstrument
positions.
VENDITFI AND BENNETt: TURBULENCE AND SUSPENDED SEDIMENT TRANSPORT 22,037
20
0 00 •0 0 •0 0 0• 00 0 00 0 00 00 0 •0 • 0 0 -
15 0 00 00 0 •0 0 00 00 0 00 0 •0 00 0 00 • 0 0 --
0 00 00 0 00 0 00 00 0 0• 0 00 0• 0 00 0 0 0 -
--
o o o o -
10 o o o o _
> 20
- --- ..........................
• 0 O0
0 O0 O0
O0 0 O0
0 O0 00 O0
O0 O0 0
O0 0 O0 0 0
00O0 O0
O0 0 O0
0 O0 0
0 0
0 0 ---
0
• 0
0 O0
O0 O0
O0 0 O0
0 O0 00 O0 O0
O0 O0 00 O0
O0 0 O0
0 O0 O0
O0 0 O0
0 O0 0
0 0
0 0 --
0
• 0 O0
0 0 0 O•0 00 0O•0 000•
0 0 O0
0 0 •
0 O•0 00•
0 0 0 0 O0
0 0 •
0 O• 0
0 0 0
0 0
0 --
0.1
0 4 8 12 16 -4 0 4 8 12 16 -4 0 4 8 12 16 -4 0 4 8 12 16 -4 0 4 8 12 16
-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3
0.1
-' 0 1 2 3 4 -' 0 1 2 3 4 -1 0 1 2 3 4 0 1 2 3 4 -1 0 1 2 3
NormalizedVelocity
Figure 4. Selectverticalprofilesof streamwise(U) andvertical(W) flow velocityandtheir rms values
(opensymbols)normalizedby u,R andcomparedto profilesmeasured by BennettandBest[ 1995] (solid
symbols). Circles are mean values,and trianglesare rms values. Also shownare the mean and rms val-
ues for the normalizedcross-stream (V) velocitycomponent.Positionsare given relativeto the dune
length•.
1:2 ..............................
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
..• ' '. :;!•:.,,
:'i.':';:i.•:•.?:
!•4::.•i.';.::•;
....';'
-1 o I 2' 3 4 5 6 7 -0.002 0.025 0.07.5 0.125 0_175
BoundaryLayerCorrelation(R,w) Eddy
Vi__•scosity
(m2s'1)
r..,•.•.
•-.--......•
..•..-. V '. •.•.
:.L--J.
:•....¾•'0'•'•"2'-•"
.-•" •:'•.' '.-••
:•i
•?"':"•'•.,.,?;-•-:•:::-'::ii4:•',•C;•.•i•':•-,•::'•i;:?..•:".....'..•i?;.i:•?"
•{5•'•;•;
':,.'?',---.-•.•; 5-i;..:.
"'-•;•"T :.---•.
';:•;.'.::";;:•.:•'•'•,•';•
........ ' '.?&..::
:...'•"'...•:....,
. '--........'%.::::.'½- ;:;•,•
..........
..;.-;'-%." "..""•:"•:;•'
' •/-•-"•
...."-:-•--•-•-•-.-•:'.'....
• •'•'
0.20 0.26 0.32 0.38 0.• 0.50 0.• 062 0.68 0 0.0004 0.0008 0.001 0.002
TurbulentKineticEnergy(Pa) Suspended
Sediment
Concentration
(rag!-)
c
0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90
[':•"
::-'":
•:.
•::•
::':•
:::
•::J"--::::':i:•:•i•!':'•:-:'•i:'!l'•::
.'•":':•:•i[•":•:•.•f':•-•
F-• "'•'f-•
•t•1 50 90 130 170 210 250 290 330 370
0.0 3.5 7.0 10.5 14.0 17.5 21.0 24.5 29.0
peakupstreamanddownstream of the peakin TKE (Figures5c and peaksin the broad spectrumcorrespondto the frequencyof
and5f). Distributionof concentrationalsoappearsto be affected an energetic mean or dominant eddy size passing the sensor
by theturbulentcharacteristics
of the shearlayer. As largevalues [Boppeand Neu, 1995]. Confidenceintervalsfor eachspectrum
of ruw and e extendoutwardinto the flow downstreamof reat- were calculatedusing the Z 2 method where upper and lower
tachment,suspendedsedimentis advectedalong with thesere- confidenceintervals are given by
gionsof highturbulence
(seethe130mgL-] contour
line,Figure
5f). Bw f.P(f) Bwf.P(f)
(9)
Shownin Figure 6 are profilesof suspended
sedimentflux 2'(•.95 ' 2'(•.05
qs(z)definedas
respectively[Jenkinsand Watts,1968]. For eachvelocityspec-
qs(z)= C(z)'U(z), (8)
trumthe upperconfidence intervalis 1.2795f. PIf) andthe
at-a-pointtime averagesand all necessary lowerconfidence
where(z) represents intervalis 0.8031f. P{f) at the95% confi-
correctionsfor unitsare made. The largestsedimentflux andthe dencelimit. Sincethe focusof the presentstudyis to evaluatethe
greatestvertical gradientin transportrate occurat the dunecrest.distributionof energyin the frequencydomain,theseconfidence
Sedimentflux decreases in thedunetrough(separa- intervalsarenot includedin theplotsof spectralenergy.
significantly
tion cell) becauseof low-flowvelocities.The outerflow region Spectrafor the velocitytime seriesall generallypeakfrom 1 to
showsvery little vertical variationin suspended sedimentflux. 4 Hz. The streamwiseenergyspectragenerallypeak at lower
Spatiallyaveragedsuspended sediment transportrateis 0.065kg frequencies (-1-2 Hz; e.g., 6-45) than the vertical and cross-
m-I s-1. streamenergy spectra(-2-4 Hz; e.g., 6-51, Figure 7). Differ-
encesbetweenthe peakfrequencies in the streamwise spectraand
othervelocitycomponents maybe dueto eddyshape:eddiesmay
4. Spectral Analysis be ellipticalas theyadvectanddiffuseaway from the shearlayer
presumablybecausestreamwiseeddy lengthsare stretchedand
4.1. Velocity time series
elongatedby the meanflow. In the near-bedand wakeregions,
Univariate spectralanalysiswas performedon velocity time spectralenergiesare generallylarger than in the rest of the flow
series,the locationsof which are identified in Figure 2. Since field andhavewell-definedrecurrenceperiods(e.g.,3-60, 3-51 6-
these locationshave an approximatelyequal spacingin the 36, 6-45, 9-51, and 9-45). Outsidetheseregions,energiesare
streamwisedirection,velocityspectraare plottedas eight sparse smearedacrossa large frequencyrangeand are generallysmaller;
profilesin Figure7. Each82 s time series,corresponding to 2048 that is eddiesare not passingthe sensors with a well-definedpe-
measurements,was detrended, and spectral estimatesfor each riod but rathera rangeof sizes,trajectories,andmagnitudes (e.g.,
velocitycomponentwere calculatedat 0.05 Hz intervalsfrom 0 to 12-36, 12-69, and 12-75). Near the bed, there is a clear decline in
12.5Hz usinga bandwidthBw of 0.8908Hz in thespectralalgo- the peak spectralenergyfrom the regionof flow reattachment.
rithm. By default, spectralresultsare biasedtowardthe low-fre- The spatial distributionof the velocity spectracan be inter-
quencyrange. Velocity spectraare plottedin variance-preserving pretedas characteristicof a perturbedshearlayer havingthe fol-
formwherespectral energydensityP{f) (cm 2 s'2 Hz-1) is lowing characteristics:(1) a "deadzone"or linear stabilityregime
multiplied by frequencyf [see Soulsby,1977; Panofskyand extendingdownstreamof the dune crestthat growsinto a wave
Dutton, 1984; KaimaI and Finnegan, 1994]. The area underthe structureand (2) a nonlinear instability regime where Kelvin-
spectrum of f. P{f) against logf represents thetotalvariance, Helmholtz waves develop and break along the shearlayer that
1.0
0.6
0.4
0.2
0.0 ß . i .... i i i , i
0.05 0.10 0.15 0.05 0.10 0.15 0.05 0.10 0.15 0.05 0.10 0.15 0.05 0.10 0.15
Suspended
Sediment
Flux(kgm-2s
Figure6. Selected
verticalprofilesof suspended
sediment
fluxasa functionof duneposition.
22,042 VENDITTI AND BENNETT: TURBULENCE AND SUSPENDED SEDIMENT TRANSPORT
400
300ju0.85
........
• [u1.20 u 1.65
12-45
u 0.90
OL
u 1.40 J u 1.05 [u 0.70 I:u 1.55
v 1.90 v 1.50
200ß
100 Jw
2.05 w 1.65 w 2.25
400
300
[u9;!8o
ø'200 8FW [uIv9-27
Iv 2.00
1.55W
1.35 9-36 W "9"-5i
W i 9-45
u 0.80
v 1.85
"V•........
u1:00
' 9-60osu 1.25
vU•:•50
[9-69
I:u
os
1:55
[9-75
FW J
I:u0:90
&2.20 ]
100
400
J'6':
•8'F"W
.........
J'l•::2?
'W'.........
JJ'•"-36"V• ........."6"-4S'"V•
.........
t I6:'5i"W
...........
:1[6-60
w ][6-69
Dz 6-75FW
300
[u2.05 r u1.85-4 I [:u1.65 •
ul.85
X t[u0'95 -•,' lru,.,o u0.95
v 2.20
' w 2.90
o .
0.1 1 10 0.1 1 10 0.1 1 10
upperconfidence interval
is 1.4166f. P(f ) andthelowerconfi- seriesto be sampledat the samefrequency. Thereforethe 25 Hz
denceintervalis0.7439f. P(f ) atthe95%confidence limit. velocity time serieswere low-passfiltered to 10 Hz, interpolated
Energyspectrafor the concentration dataare shownin Figure to 100 Hz, and then resampledat 10 Hz. Comparisonof the time
8. Becausethe magnitudeof variancewas unchangedby the seriesused in the cospectralanalyseswith the spectralanalyses
correctionsapplied to the sediment concentrationrecords, all described above showed no difference in results. Each time se-
spectrawerenormalizedby the varianceof eachtime series,pro- ries was detrended, and cospectralestimateswere calculated at
ducing spectrawith a normalizedenergyrangingfrom 0 to 3. 0.05 Hz intervalsfrom 0 to 5 Hz using a bandwidthof 0.4711 Hz
The normalizationprocessappliednegatesany interpretation ofin the cospectralalgorithm. A total of 102.4 s of the velocityand
the relativeenergycontainedat a specificfrequency(i.e., the areaconcentrationmeasurementswere used correspondingto 1024
under the spectralcurve), but the frequencyinformationremains data pointsin eachtime series.
the same. Spectralenergy peaks around0.5 to 2 Hz, with the Cospectralanalysiswas performedat severallocationswithin
strongestpeaksgenerallyoccurringat 1 Hz (e.g., 9-75, 9-60, and the flow field, but only one exampleis describedhere at a point
3-36; Figure8). Theseenergypeaksappearto be coincidentwith near the bed and in the wake (6-45; see Figure 2 for location).
the spectralpeaksin the streamwisevelocitycomponent(Figure Spectraat this location show a defined peak in the u component
7). Given the stronglypeakedshapesof the concentration spec- at 1.85 Hz and peaks in the w componentat 1.90 and 3.80 Hz
tra, it appearsthat the suspendedsedimentis a very sharpdelim- (Figure 7). The cross spectrumof u and w showstwo strong
iter of themosteffectiveeddyscale.Valuesof fs are0.47 using peaks:at 1-2 (largestnegativecovarianceat 1.70 Hz) and 3-5 Hz
l = d and 0.11 using l- H, in agreementwith nondimensional (Figure 9a). The squaredcoherencyspectrumshowsgood corre-
frequenciesfor the u componentof velocitypresentedabove. lation on the rising limb of the spectrumand again at -1.65 Hz
(Figure 9b). The secondarypeak in the u againstw crossspec-
trum is not recognizedin the squaredcoherencyspectrum,sug-
5. Cospectral Analysis gestingit is relativelyunimportantwhen comparedto the primary
Cospectralanalysis,which included calculationof the cross peak.
spectrumand the squaredcoherencyspectrum,was appliedto the The concentrationspectrumat this measurementlocation dis-
time seriesto examinefurtherthe link betweensedimentsuspen- playsa strongpeak at 1.65 Hz in agreementwith the u againstw
sion and the production,advection,and diffusion of turbulence. crossspectrum(Figure9c). Crossspectraof u againstc, v against
Crossspectrashow whetherfrequencycomponentsin one time c, and w againstc all appearto d•splaysomeassociationbetween
seriesare associatedwith large or small fluctuationsat the same velocity componentsand concentration(Figure 9d). However,
frequencyin the other time series. The squaredcoherencyspec- the squaredcoherencyspectrafail to reveal any peak in the ve-
trum is a correlationcoefficientdefinedat each frequency[Jen- locity spectrathat can be correlatedwith the peak in the concen-
kins and Watts, 1968]. Cospectralanalysisrequirestwo time tration spectrum(Figure 9e). Althoughit would appearqualita-
[ 12-75 OL
2 c0.95
c0.95
[c1••c
0.90
/• j c1••[&l•
-•
[0'h .......
] .......
l .... .......... ............
•o ..
•3 .
O 0 .............................................................
0.1 1 10 0.1 1 10 0.1 1 lO
Flow Re.qions
OL-outer layer
W-wake
FW-far wake
2 :0.0 [3-60
S(•
DZ-dead zone
IBL-internalboundarylayer
R-reattachment
SC-separation cell 0 ......
OS-over sep. cell 0.1 1 10 0.1 1 10 0.1 1 10 0.1 1 10 0.1 1 10
Frequency(Hz)
,,_ -5o
-40 6-45 werederivedto determineintegraltimescaleandlengthscale. An
integraltimescaleis the time an eddy requiresto passa given
point in the flow, and an integrallength scaleis the characteristic
-60 eddy dimension. Correctionsappliedto the OBS time seriesfor
1.0 variable turbidity had no effect on the derived autocorrelation
functions.
•0 0.6 rœ
=lR(t)dt,
o
(11)
•(• 0.4 where R(t) is the autocorrelationfunction,dt is the lag distance,
s near the bed to 0.30 s at 0.12 m above the trough (Figure 10a),
c)') o and thesevalues resemblemore closelythe u componentresults,
E i.e., T•(c) • Tœ(u).
-40
Streamwise
eddylengthscalesL•(u) rangefrom0.065 m near
the bed to 0.135 m at 0.12 m above the trough (Figure 10b) and
Q_ -80 are grosslyrelatedto heightabovebed, i.e., Lœ(u)• z. Cross-
streamL•(v) and vertical L•(w) integrallengthscalesare nearly
"•--120
1.0 invariant with distance from the trough, ranging from 0.020 to
0.030 m, and are grossly related to dune (step) height, i.e.,
L•(v) • L•(w)• H. The integrallengthscalefor concentration
0.8
LE(c) increaseswith flow depth,t¾om-0.085 m nearthe bed to
0.160 m at 0.12 m above the trough(Figure 10b), similar to the u
E 0.6 component,i.e., L•r(c)• L•r(u)• z.
(L)
j::: OA
0
Figure 9. Cospectralresultsfor a measurementlocation in the
0.2 near-bedregion (6-45; seeFigure 2) displaying(a) streamwiseu
velocity against vertical w velocity and (b) squaredcoherency
spectrum.Also shownare (c) the concentrationspectrumand (d)
0.0
u againstsuspendedsedimentconcentrationc (thin solid curve),
o.1 1 lO
cross-streamv velocity againstc (dashedcurve), and w againstc
Frequency (Hz) (thick solidcurve) with the (e) squaredcoherencyspectra.
VENDITTI AND BENNETT: TURBULENCE AND SUSPENDED SEDIMENT TRANSPORT 22,045
o
to0.135m at0.12m abovethebedandis grossly
related
to flow
depth.Cross-stream
andvertical
integral
lengthscales
range
from0.020to 0.030m andaregrossly
related
to duneheight.
Theintegral
lengthscaleforconcentration
ranges
from0.085m
nearthebedto 0.016 m at 0.12 m abovethebed,similarto the
½8 streamwise
velocitycomponent.
Notation
6
BW bandwidth.
C,Ct,C', C suspended sediment concentration, its in-
4 stantaneous value, its fluctuation about the
mean, and its mean value.
d flow depth.
2 frequency.
Darcy-Weisbach friction factor calculated
0.00 0.05 0.10 0.15 0.20 using r0and
Integral Length Scale (m) Nyquist frequency.
nondimensional
eddysheddingfrequency
Froude number.
Figure 10. Vertical profiles of spatially averagedintegral (a)
timescaleand (b) length scale.Refer to Figure 2 for locations. g accelerationdue to gravity.
Error bars are standard error of the mean. H dune height.
k time step used for the autocorrelationfunc-
tions.
I length scale.
7. Conclusions
L•r, T•: Eulerianintegrallengthscaleand timescale.
Laboratorymeasurements of turbulentfluctuationsin velocity P turbulenceproduction.
and suspendedsediment concentrationwere obtained synchro- spectralenergy.
nously over fixed two-dimensional dunes using a sediment- qs suspendedsedimentflux.
starvedflow. The observedturbulentflow field agreeswell with R hydraulicradius.
observationsin previous studies over both mobile and fixed Re Reynolds number.
dunes, High-magnitudeReynoldsstresses(particularly ruw), Ruw boundarylayer correlation.
22,046 VENDI'ITI AND BENNETT: TURBULENCE AND SUSPENDED SEDIMENT TRANSPORT