SOME CHARACTERISTICS OF PRESSURE
FLUCTUATIONS ON LOW-OGEE CREST
SPILLWAYS RELEVANT TO FLOW-INDUCED
STRUCTURAL VIBRATIONS
bY
irederick A. Locher
Sponsored by U.S. Corps of Engineers Contract No. DACW39-68-C-004
IIHR Report No. 130
Iowa Institute of Hydraulic Research
The University of Iowa
Iowa City, Iowa
February 1971
TABLE OF CONTLIITS
I. INTRODUCTION 1
II. EXPERIMENTAL APPARATUS AND METHODS OF DATA ANALYSIS 4
Flume 4
Spillway Shape 4
Pressure Measurements 4
Data Analysis 6
III. PRESSURE FLUCTUATIONS I. SPILLWAY CREST 9
General Considerations 9
RMS Value, Pressure Fluctuations, Point 1 13
Boundary Layer Measurements 15
Spectral Density Function,
Spillway
Crest 16Comparison with Prototype Data 17
IV. PRESSURE FLUCTUATIONS II. SPILLWAY TOE 18
RMS Value, Pressure Fluctuations, Point 2 18
Spectral Density Function, Spillway Toe 19
Flow Visualization-Dye Tests 20
Discussion 20
Spillway Toe Pressure Fluctuations, Hydraulic Jump 21
V. DISCHARGE COEFFICIENT 23
Dimensional Considerations . 23
Discharge Coefficient at the Design Head, Hd . . 25 Discharge Coefficient for He/Hd > 1.0 26
VI. SUMMARY AND DISCUSSION 27
VII. CONCLUSIONS 29
Figure 1 Definition sketch
Photographs of the experimental apparatus with discharges cor-responding to He/Ha
1.0, 1.3; 1.5,
and 1.8 . ,Variation
of
the relative- rms value of the pressure fluctuationsat point 1 With He/Hd. No boundary layer trip . . . 55
Variation of the relative rms value of the pressure fluctuations
at point 1 With He/Hci. Boundary layer tripped . . .
.36
Spectrum of the pressure fluctuationsNo boundary layer trip Spectrum of the pressure
Boundary layer tripped . .
Figure 7 Spectrum of the pressure fluctuations
No boundary layer trip.
Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Flgure Figure Figure Figure Figure Figure 13(c) .LIST, OF FIGURES
Figure 8 Spectrum of the pressure
Boundary layer tripped .
fluctuations
..
..
. ... .
fluctuations at point 1, H /H, = 1.0..... .
. . ? . . . at point 1, He/Hd = 1.0. at point 1, He/Hd = 1.3. at point 1, He/Hd = 1.3. 38 39Figure
9
Spectrum Of the preSSure fluctuations at point 1, He/Ha =1.63.
No boundary layer trip 41
Figure 10 Spectrum of the pressure fluctuations at point 1, Hi /HA =
Boundary layer tripped.. .
,
.4
....
42Figure 11 Spectrum of the pressure fluctuations at point 1, H /HA
= 1.8.
No boundary layer trip . . .. .... .
.?"
...
. . 43 Figure12
Spectrum of the pressure fluctuations at point 1, H /Hd =1.8.
e
Boundary layer tripped . 44
13(a) Oscillograms of the pressure fluctuations at point 1, no boundary layer trip
13(b) Oscillograms of the layer tripped
Oscillograms of the pressure fluctuations at point 2, no
boundary layer trip 47
Oscillograms of
the
pressure fluctuations at pOint 2, boundarylayer tripped . 48
OsOillograms of the pressure fluctuations at point
2,
hydraUliojump,
boundary layer tripped .14
Variation Of the relative rMs Value of the pressure fluctuationsat point. 2 with He/Hd.
No boundary layer trip ...
. . 50iii
45
pressure fluctuations at point 1, boundary 146
Figure
16
Spectrum'of the pressure fluctuations at point2,
He/Hd No boundary layer tripFigure 17 Figure
18
Figure Figure20
Figure21
Figure 22Figure 23 Spectrum of the pressure
Boundary layer tripped .
Figure 24
Figure
Figure
Figure 30
Spectrum of the pressure No boundary layer trip
LIST OF FIGURES (continued)
Figure 15 Variation of the relative rms value of the pressure
fluctua-tionS.at point 2 With H /H
d.
. Boundary layer tripped. ,51
e
Spectrum of the pressure fluctuations
Boundary layer tripped 53
Spectrum of the pressure fluctuations No boundary layer trip
19 Spectrum of the pressure fluctuations Boundary layer tripped
\
Spectrum of the pressure fluctuations at pOint
No boundary layer trip .
56-Spectrum of the pressure fluctuations Boundary layer tripped
iv
at
28 Spectrum of the
Hydraulic jump, boundary layer tripped
at point 2, H /Hd = 1.0. e at point 2, He/Hd = 1.3. point 2, He/Hd = 1.3. . . .
2,
He/Hd =1.5.
at point2,
He/Ha= 1.5.
57 fluctuations at point2,
1%-/Hd =1.8.
fluctuations at point 2, He/Hd
=1.8.
OOOOOO
Photographs of the spillway face-with dye injected at the piezo-meter located at point 1, for discharges corresponding to
He/Hd = 1.0, 1.3,
1.5,
and1.8
Spectrum of the pressure fluctuations at point 2, Hydraulic jump, boundary layer tripped lie/
Spectrum of the pressure fluctuations at point
2,
He/H = 1.3Hydraulic jump, boundary layer tripped
pressure fluctuations at point 2, He/Hd =
1.5.
27 Spectrum of the
Hydraulic jump, boundary layer tripped ... . . .
pressure fluctuations at point
2,
He/Hd1.8.
54
55
58
6o
Figure 29 Photographs taken during the experimental rums with '8. hydraulic jump on the dOWnstream apron. Data plotted on Figs.
25-28
. 65 Flow With He/4d = 1.8 and separation, on the spillway face asvisualized by a' dye streak . . . OO O
..
. - 66=1.0.
52 .Figure 25 Figure 2663
64LIST OF kGURES
(continued.)
Figure 31 Discharge coefficient as a function of He/lid . . . 67 Figure 32 Comparison of the discharge coefficient of the present
investi-gation with data published by Corps of Engineers
(1965).
Reproduced with permission of the Corps of Engineers . . . 68
Figure 33 Mean pressure at point 1 as a function of He/lid
..
69 Figure34
Mean pressure at point 2 as a function of He/Hd 70LIST OF SYMBOLS
Discharge coefficient, C = Q/LHe$12
d Discharge coefficient, Cd =
3/.2
- e
C(T)
Auto-covariance function.,G(T) =
X(t)X(t+4)Froude number
d Design bead for, the
spillway
.shapeH Head on the
epiiXway,
He= h
+ haHm 'Maximum head, required to pass the
efdlivey
design flood. Length of thespillway
crest in feet.
SpiiIvall -height
P(f) Spectral Density function, P(f) = C( r) cos(2P0dT.
Discharge in cubic feet per second Reynolds number
11(T) Auto4-correlation function. normalized auto-covariance
function
Uo Velocity of the approach flow
111 Weber number
X(t) Ergodic random process (function of time)
Depth of water on the downstream apron, transducer face
diameter
Gravitational acceleration
Frequency in Hz.
Height of water over the spillway
ha Approach flow velocity head ha = UO2/2g
hd Distance between the line of total head and the free surface at the section where d is measured
Mean pressure
Instantaneous deviation of the pressure from the mean
pressure
1! = p'2 Root-Mean-Square value of the presSure fluctuations
X,Y Coordinates, origin
at
the spillway crest0(f) Normalized Spectral Density function
0(f)
= 4
f
R(T)
cos(2fT)dT0
Geometrical parameters describing the spillway shape Geometric ratios describing the spillway shape
Boundary layer thickness
Boundary-layer diSpladetent thiCkness
Fluid Viscosity
. Fluid density Surface tension
Time delay
SOMEZHARACTERISTICS OF PRESSURE FLUCTUATIONS ON LOW-OGEE CREST SPILLWAYS RELEVANT TO FLOW-INDUCED
STRUCTURAL 'VIBRATIONS
I. inigODUCTION
The
primary
objective of the experimental iprogram reportedherein was to determine whether pressure fluctuations induced on the face of a low-ogee spillWay'under various flow conditions are a possible mechanism
for
the excitation ofspillway
vibration. Some of the moreimportant terms and concepts used to describe
spillways
will be definedat the outset An the interest Of clarity.
With reference to figure 1, the spillway height p
is
the difference in elevation between the Spillway crest and-the approach-channel. floor. The. head on the spillway He is the vertical distance
from
the
spillway crest to the line of total head; thatis, He is the sum ofthe
head of Water over spiflway, h , and the approachvelocity head, ha = U162/2g , where Uc; Is- the approach flow velocity.
The design- head for- the spillway shareHd is the parameter used to
proportion the shape of -the Spillway such that for flaws with He -= Hd
the shape of the spillway will correspond to the shape of the lower time surface as determined from sharp-crested weir experiments (USBR,
1948).
The design head for the spillway shape should not be confusedwith the maximUm head
Hin required to pass the spillway design flood.
It is convenient to group spillways into two classes,
high-overflow spillways and low-high-overflow spillways. High-overflow
spill-ways- are those spillways for which the approach flaw velocities are negligable. Low-overflow or loV-ogee spillways are characterized by an appreciable velocity of approach which affects the desired shape of the spillway cross.section and the discharge coefficient (USBR,
1948;
Although prototype measurements of pressure fluctuations on the.high-overflaw type spillway at Joseph Dmn. (Corps of 'Engineers,
1958)
did not reveal any fluCtUatiOng Which Could reasonably be.con-sidered a possible source of structural Vibrations, there have beenSeveral undocumented oral, reports of vibration of low-crest spillways
for flows with heads exceeding the Spillway height. High-overflow
spillways are usually constructed on good rock foundations. The structure
'and foundation has a-high modulus of elasticity" and the System as a whole is quite, rigid and missive. -PUrthermore, the mass Of
fluid in
motion over theepillway is small
in
comparison with the effective mass of the spillway and itefoundation.. Therefore, vibration of high-overflowspillways has not been of 'particular concern. On .the basis of prototype
experience, satisfactory performnace has been obtained if the crest
height P is equal to or greater than .the head on the crest, provided
the foundationrock is good. (Corpsof Engineers,
1965).
On the other hand, low.-ogee crests are often'used in navigation
and conservation dams, and chute spillways of large structures; in many instances these structures are founded on alluvial soils. The spillway Is then a rigid structure on a relatively elastic foundation. Not only
is the elastic modulus comparatively, law, but ihe mass of water over the
spillway at the maximum head is of the same order of magnitude as the mass of the spillway itself. Thus a significant percentage of the system
comprised of the fluid flawing over the spillway, the spillway, and the spillway foundation is in motion, and should some flow-instability or. hydroelastic control be present, structural vibration of the low-ogee
crest becomes a distinct possibility as the head becomes progressively
greater.
Although there are several possible sources of spillway. vibra-tion (Campbell, 1969) including, for example, earthquake- forces and the .
effects of gates and gate piers-, this investigation was focused specific-ally the, pressure fluctuations generatedon the face of a low-ogee
spillway by the fluid flowing over the spillway at heads equal to or greater than the design head for the
Spillway
Shape,, Hd . The shape3
Chosen Air study was a low-ogee shape. with a 45° upstream face pro
.-portioned according to data on plates 26, 29, and
34
(Corps of Engineers,1965),
refer to figure 1. Two points were selected for measurement ofthe fluctuating pressure.; the pressure fluctuations at point 1 (see
figure 1) are representative of pressure fluctuations at locations on
the spillway crest, and those at point 2 are indicative, of pressure
fluctuations at the Spillway
toe.,
Pressure fluctuations at the spillway toe(point
2) were measured with and without a 'hydraulic jump on thedownstream apron floor.
The quantities Measured Were the
root-Meansquare
(RMS) valUe and spectral density function of the pressure fluctuations. The spectral density functions were Obtained Only:fOr ratiOt Of head on the spillway . to.design head for the spillway shapeof
1.0, 1.33, 1.5, and 1.8..The principal objectives of the Operthental program were:
(1).
To determine whether the pressure fluctuations induced on the face of al.OW-dgee spillway at heads equal toor
greater than the design headfOr the spillWay shape,
without
the action ofa
hydraulic jump on the downstream apron, 'are a
possible Source of excitation Of spillway vibration.
To ascertain the effects of a
hydraulic
jut') onthe characteristics of
thepressure
fluctuations 'at the spillWay toe, and their possible significancewith respect to Spillway
vibratiOn-To establish the discharge characteristics for the
spillway
shape..To review the estabitahed hydraulic deign Criteria
concerned with the
Vibration
of low-ogee spillways and recamMend possible modifications based uponThe experimental techniques, the analysis of the data, and the pertinent results are presented in the following sections.
II. ENPERMENTAL APPARATUS AND NETHODS OF -DATA. ANALYSIS.
Flume. All of the experiments reported-herein were con-ducted in the one-foot wide flume located on the first floor of the Iowa Institute of Hydraulic Research laboratory. The flume has been described in detail by Rouse
(1961).
Satisfactory flow conditions pre-vailed at discharges corresponding to heads less than or equal to thedesign head, Hd For higher discharges, the transition from the head box to the flume produced a standing wave on the free surface. Submerged
screens in the head box and a wave suppressor minimized the undesirable free-surface fluctuations upstream from the model. False floors were installed in the flume to obtain the proper relationship between the
upstream ,approach channel elevation and the downstream apron elevation.
The maximum discharge available in
5.35
cfs, which corresponds toH /Hd = '1.8. e
Abillitay'Shave. The shape of the spillway cross-section
used in this investigation is one of three low-ogee crest shapes suggested
-for generaX.study by the U.S. Arpy Corps of 'Engineers (1965,p. 52). The
ratio of the approach
flow
velocity head to the design head for thespillwayfithape
is
ha/Hd = 0.12, and the ratio of the spillway height tothe design head for the spillway shape is P/Hd
= OA4.
It has been suggested that the results from a,study of this shape are applicable toshapes in the range 0.3 < P/Hd <-0.57.- The 'upstreamquadrant was 'proportioned according to data on plate 26 (Corps of Engineers,
1965),
the doWnstream quadrant according to plate29, and the circular toe curve
according to plate 314 (see figure 1). Photographs of flow over the model
spillway
for, several flow conditions are Shown on figure 2. The Spillwaymodel height P is
0.1 'ft.,
and. the design head. Rd is0.683
ft.study was a Statham Model 131TC, a 2.5 psi differential pressure transducer with a one-half inch diameter diaphragm. Two points were selected for the measurement of the fluctuating pressure. Point 1,
representative of spillway crest pressure fluctuations, was located
at x/Hd = 0.276 and y/Hd = -0.056, with the origin of coordinates
at the spillway crest as shown in figure 1. The location of point 1 was based on observations on a preliminary model. Dye tests in preliminary model showed some disturbances in this region although these disturbances did noireappear in the model used for the pressure measurements. If separation were to occur, it was felt that the
pressure fluctuations at point 1 would be considerably more significant than those measured by a transducer at the spillway crest. Point 2,
representative of pressure fluctUations at the spillway toe, was
located at x/Hd = 1.78 and y/Hd = Measurements of the fluctuating pressure at point 2 were obtained with and without a hydraulic jump on the downstream apron to obtain comparative effects of a hydraulic jump on the pressure fluctuations on the spillway toe. A piezameter opening 1/16th inch in diameter was located at the same
elevation as the center of the pressure transducer so that mean pressure data could be obtained independent of drift problems associated with
the pressure transducer.
'Since the magnitude of the measured pressure fluctuations was
smallihigh
quality sighal amplification was necessary. The output ofthe transducer was first amplified
2500
times by a Dana Model 2850 D.C.,amplifier, and further amplified at the IBM
1800
Data Acquisition and Control System locatedon
the third floor of the laboratory., Temperature Changes caused same drift in the mean value of the voltage from thetransducer. This drift was well-within the transducer specifications, but became. noticablebecause of the high amplification required to
analyze the fluctuating component of the pressure. The mean value
(D.C. voltage) was filtered out to yield a nearIY zero mean Which
facil-itated spectrum analysis. It was not necessary to eliminate the mean for the two minute averaging timetsed to obtain the EMS value of the
by sealing the flume tailgate, filling the flume with water, and then layering the water level in a series of steps by means of a drain
valve. The calibration coefficient remained within one percent of the average value throughout the course of the investigation.
D. Data Analysis. Analysis of the pressure fluctuations was accomplished through use of an IBM 1800 Data Acquisition and Control System which has been described in detail by Glover (1968). No attempt will be made here to be cOMplete in discussing the methods used for
analyzing the data. Suffice it to say that the pressure fluctuations can be described by the theory of ergodic random processes. The objective here is to define and make clear what interpretation may be ascribed to the quantities measured in this study.
-'Pressure fluctuations on the.spillivay-face- are a random phenomenon and cannot be' described adequatelyby. assigning A specific
amplitude orparticulat frequency to the fluctuation, as anyone who has etaminedstrip-chart recordings or,oscillograms
is
well aware, (see. figure 134for
example). Instead, statistical Parameters which provide - a picture of the.average:behavior of the fluctuations are required*A random process contains a distribution of amplitudes, with some occuring more frequently than others. Specifying the mean-square or root-mean-square (E(S) value is somewhat analogous to specifying the amplitude of a deterministic process (a sine wave for example). For an ergodic random process X(t) p where t is time, the mean-square value can be expressed as
/2
3771t= lim
=IT
Itox(t)dtT+=
The ENS value is the square root of the mean-square and is sometimes.
'referred to AS a measure of the Intensity of the fluatUations:
just as it contains a distribution of amplitudes. The spectral density function is a statistical function which displays some of the frequency characteristics of a random process, and indicates, on the average, the frequencies that are most likely to be encountered. The spectral density function can be obtained indirectly by first measuring the auto covariance
function
C( T) = X(t)X(t+T)
where the bar over the quantity denotes temporal average. Notice that
C(0) is simply the mean-square of the processCTET2-) . The random process is correlated with itself displaced in time by the lag time, T .
The auto-correlation function is'the auto-covariance function normalized
by C(0).
R(T)
c(I).
x(t)x(t+T)'
c(o)
It Can also
be
dt0Onstrated"that the spectral density function and the al4O-coVatiance function are related by Fourier transforms (Wiener, 1930).C(T) = P(f) -cos 27rfTdf
Rri72-P(f) = 41 C(T) cos 271...frdT
0
The normalized form of these relationships is used to present the experi-mental data obtained in this study
flt( = cos fce 1)(f)
-77-
= 0(f) cos 2Tuftdf ' C 0 k )car,.
44. 41
R(r) cos (2TrfT)dTOne interpretatioh of the spectral density fUnction
can
be Obtained by8
Hence P(f)df represents the contribution to the mean-square from fluctuations with frequencies between f and f + df . The quantity
(f)df = P(f)/C(0)df therefore represents the intensity of fluctuations of frequencies between f and f + df relative to the total intensity
of the fluctuations i7777= C(0) .
Another property of the spectral density function can be derived from the fact that a sine wave of frequency f = f appears in the spectrum as a Dirac delta function at f = fo; that is, as a
rectangle or "spike" infinitely high, infinitesmally wide and having a
finite area. In practice, periodic functions appear as sharp, high
peaks in the spectrum. Thus, not only does the spectral density function
t(f) present the distribution of the relative intensity of the
fluctua-tions 'with respect to frequency, but it may also indicate the presence
of periodic or almost periodic components as well.
The auto-correlation function was determined by use of the IBM 1800 Data Acquisition and Control System. The computer program has been described by Locher
(1969),
and some references on the theoretical and practical aspects of methods for obtaining the spectral densityfunction are Blackman and.,Tukey (1958), Bendat and Piersol
(1966),
and Jenkins and Watts (1968).In sunmary, the descriptors of the random pressure fluctuation's utilized are the BNB value and the spectral density function, which
are measures of the amplitude and frequency characteristics, respectively.
These two statistical 'parameters form the basis for interpretation of the characteristics of the pressure fluctuations measured in this study.
where
9
III. PRESSURE FLUCTUATIONS I. SPILLWAY CREST
A. General Considerationa. In order better to understand
the prOblam of eiperimental measurements and interpretation of the data associated With the fluctuating presture, consider first the simpler situation posed
by
the mean pressure. The general functionalrelationship for the. mean pressure at agy point on the spillway face it
= f, (p g, Op Uo, P, h ai) where = fluid density = gravitational acceleration p = fluid viscosity a -= surface tension
U = velocity of the approach flow P = spillway height
h height of water over the spillway Crest a. = geometric parameters describing the spillway
1
shape, including k , the equivalent sand grain
.roughness of the spillway surface, and lid
the design head for the $141lway Shape
An application of the IT theorem yields the following relationship:
P gd
, h-, 131) ( 5 )
1F
=/417
a Froude number= Uohp/p ,
a
Reynolds numberNW
Duo/Ph ,
a Weber ftudberIf the approach flow
is
restricted to being subcritids1 (Froude numbers) less than unity), then the Froude number is no:longer.an independent variable, but depends upon the remaining parsdeters On the tight side of(5).: .(See Rouse, 1938, for further discussion of this point.) Further,
tined He := 2(Vo, h), the functional relationship
can_ba
written as1 1;p110 Nis
2---1
2 2 0 -P He et: H H e e d\
and,
after aode-manipulation, in the more familiar formPH
)(6)
(7)
PH k
e 6 -4/ H H H ' d d d(8)
If the model
scale
is sUffiCientlY-.large, the viscous and surface tensioneffects have negligdble effects On the gross flow pattern For a smooth
surface and a particular -spillway geometry, the final form of the
relationship
for
the Mean pressure is.P A '
(1-1
pU
2 (1)`,17d% H.
.2
0.
The validity- Of the., simplifying assumptionshas.been demonstrated
prac-tically
by
model-Prototype Correlations and theoretically by theappli-cation
of potential flow calculations (Cassidy,1964)i
.Turning to consideration of the fluctuating pressure,
p'
similar
reasoning leads to the follotring relationship- fOr':the EMS value . of the pressure fluctuationeHowever, the effects represented by-F,P, andIW .cannot be dismissed
-immediately as in the case of the mean pressure, just discussed..-The key .
effects on the gross flow pattern; pressure fluctuations depend on the fine scale characteristics of the flaw as well.
AMOng the possible sources of pressure fluctuations on the spillway face are 1) waves on the free surface, 2) tutbnience
in
the approach flow, and 3) turbulence originating in the boundary layer. Free surface waves may be either
a
gravitational or surface tensioneffect, and hence they are Influenced
by both theTroude,and-Wehet
numbers. - .In general, the effects of free dUrface waves may be discounted
if there are no waves observed in the model or prototype. It should be noted that unusual free-surface effects can be induced
by
upstreamconditions', as was observed In model tests for the Kaysinger Bluff Dam (Pickering,
1968).
No such effects were observed in the present study. However, the influence ofall
perturbations in. the free surfacecannot be completely ignored
in
the model', as Will be discussed later.In the absence
of
toticablefreesutface effects, and provided that the model scale is sufficientlylarge,
the Froude and Weber numbers need not be considered, as was diteussed previously, fOt the case of the meanpressure,. p. The relationship for the EMS:Value of the ptesdure-fluctuations for a smoOth spillway and a. particular spillway geometry,
(8)
then reduces toA
2
- T7
(9).
The Crux Of the problem may be seen from 4 comparison Of'
(7)
and(9)
above.. Whereas the mean pressure depends primarily on effects of gravity
and inertia, And henCe scales according to the familiar FroUde ctitetion,
the fluctuating 'pressure also depends strongly On effects of viscosity,
bringing in an additional parameter, the Reynolds number. Therefore,
the Model and prototype are not dynatioally similar, if Water is used
in
both cases, as fat as the fluctuating pressure is concerned.The lack of dynamic similarity in the viscous forces is quite evident when the boundary layers in the model and prototype are compared.
12
Since the model Reynolds number is generally much less than the prototype Reynolds number, the proportion of the spillway face over which the boundary layer remains laminar will be much larger in the model than in the prototype. If the boundary layer is assumed to develop from the line determined by the intersection of the
45°
upstream slope and the approach channel floor, then in the present study the distance Reynolds number for point I (see figure 1) isof the order 105. An acceptable value of the transition Reynolds number for about two percent turbulence intensity of the approach
flow is
3.2
x 105 (Schlichting,1960).
Thus, transition from alaminar to turbulent boundary layer probably takes place near point 1
in the model. If a 401 prototype spillway is considered, for example, then the Reynolds number at point 1 in the prototype would be about of the order 107 to 108, and the boundary layer in the prototype would be turbulent over all but the first few feet of the spillway
surface. It is therefore very doubtful that measurements of the pressure fluctuations at point 1 for a smooth spillway model would have been representative of the prototype characteristics, had special
steps not been taken.
A solution to this dilemma is to trip the boundary layer with artificial roughness elements, thereby making the boundary layer
turbulent over most of the surface of the spillway model. A review of data obtained by several investigators (Bull,
1963;
Serafini,1963;
Willmarth,1963;
Blake,1969)
of pressure fluctuations under turbulent boundary layers on a flat plate in zero-pressure-gradient flows revealstwo significant features; 1) The ratio of the EMS value of the pressure
fluctuations to pU.2/2
where Uw
is the free stream velocity, is almost constant with varying distance along the plate; and 2) Thespectra of the pressure fluctuations superpose when normalized with the free stream velocity 'Um and the boundary layer displacement
thickness, (5*. Therefore, the RIC value of the pressure fluctuations
is practically independent of the distance Reynolds number, and if the boundary layer is turbulent on the face of the spillway model, representative measurements of the characteristics of the pressure fluctuations can be obtained that are adequate for most practical
engineering purposes. Utilizing the experimental evidence cited above, (9) can be reduced to the simple relationship
/17
P H1 2 H
Hd
pUo
(10)
B. EMS Value, Presdure Fluctuations, Point 1. The boundary layer was tripped by two tots of artifical roughness elements 1/8 inch square and 1/32 inch high with a center-to-center spacing of 1/2 inch and located near the beginning of the Upstream quadrant (see figure 1). Experimental measurements obtained at point 1 included:
The EMS value\of the pressure fluctuations relative
to 1/2 pUO2. As a function of the 'ratio of head on
the spillway H to design head for the spillway
shape, Rd
-Spectra'of the preasUre flUctuatiOns at ratios of He/Rd
equal
to 1.0,l.33,
1.5, and 1.8.. Boundary layer characteristics With the turbulence :stimulator* in place.
For
purposes of comparison, measurements of the REB values and spectra of the. pressure fluctuations were obtainedvith and without the turbulencestinUlatOts.
Boundary
layer veldcity ptofiles were measured with a stagnation probe, following the procedure described by Cassidy (1964). No.sttempt was made to measure bOundaty layer Characteriatics Withoutthe tuibulence.stimuiatora.
Figures 3 and 4 depict the variation of the relative HMS value of the presaute.fluctuationeWith %/Hd with and without the turbulence
stiMulatorS; Notice that the HMS
Values
of' the pressure fluctuations.for the smooth spillway are higher than corresponding results With the
roughness elements in place. This unsettling disagreement between the
the distance Reynolds number for point 1 is of the order of 105
indicating that transition from a laminar to turbulent boundary layer
should take place close to point 1. Transition does not occur at a.
well defined location, but oscillates back and forth about some mean
position on the spillway face. This random oscillation and the fact that the turbulence characteristics in the transition region are
different from those in a fully turbulent boundary layer are the most likely causes of the difference in the observed values. Furthermore, eddies are shed from the small separation region just upstream from the spillway, and may be convected downstream along the smooth boundary and influence the pressure fluctuations at point 1. The presence of the artificial roughnesses would tend to break up any large scale
eddies near the boundary and superpose a much finer scale of turbulence on the mean flow pattern, which in turn decreases the fluctuations that the 1/2-inch diameter transducer is able to resolve. Furthermore,
as H
e/Hd increases, the Reynolds number also increases and the trend
of the data from the smooth spillway is toward the data obtained with the turbulence stimulators in place, with almost the same results at the highest ratio of He/Hd = 1.8. That _is, the data for the two cases
tend to agree more closely at high values of the Reynolds number. It may therefore be concluded that the _differences in the results with and
without the turbulence stimulators is influenced by the Reynolds
number.
The trend toward higher relative BNB values at the upper and lower limits of the He/Hd values investigated may be explained as
follows. At He/Hd = 1.0 , the absolute magnitude of the pressure
fluctuations is small. The fact that 2/(1/2 p1102) ,is almost a constant
means that g is proportional to the square of the approach velocity U . Therefore, noise from the amplifier system, vibrations, and small
free-surface fluctuations begin to become significant with respect to the magnitude of the fluctuations, resulting in the observed upward trend at the lower values of He/Hd . With He/Hd = 1.8 , the wave suppressor was not as effective in reducing free-surface fluctuations as for lower discharges, and the approach flow appeared to be more turublent.
15
Both of thesefactors would contribute to the observed increase in the
RMS:valUes. It ,seems reasonable to state that for practical purposes,
the value Of p/(I/2
pu02)
at point i with the turbulence stimulators in place is a constant equal to 0..055 in the interval'i.0 g Hea
/H. 5 LT,.C. Boundary Layer Measurements. The velocity profiles normal to the boundary at point I were obtained for He/Rd equal to 1.0, 1.33,
1.5 and 1.8. The displacement thickness 0 and velocity just outside
the boundary layer were required for comparison with published results.
It should be made clear, that this phase of the investigation was not a
study of wall pressure fluctuations under turbulent boundary layers on
curved surfaces. The data were obtained to determine whether the measured characteristics of the Pressure fluctuations were reasonable
in the light of the available information.
Let us consider the effect of transducer size on measurements
of the pressure fluctuations. It is obvious that those pressure
fluctuations which are derived from the smallest eddies in a turbulent flow cannot be resolved ifthe transducer size is larger than some
characteristic scale of the turbulence. Therefore, the larger the transducer diameter relative to some characteristic length, say ,
the less high-frequency fluctuations will be sensed and the lower the
HMS value will be. Spectra showing the drop in high-frequency fluctu-ations measured with successviely larger transducers have been published by Willmarth and Wooldridge (1963) for example. Because the
high-frequency fluctuations are not of significance in this study, the size of the transducer (within limits of course) is not important except
to permit comparison with other data.
One shOuldstrive to make the ratio of transducer face diameter to displacement thickness as small as possible, so that the high,frequenCy content of the speCtrum can be analyzed.. Some of the more recent data were obtained with d/0
of
0.1 (Blake,1969);
in-the present investigation d/0 was of -the order of 20. With this parameter determined, some Comparison with published data is possible.
For a smooth boundary, Serafini (1963) reports a value of
2/(1/2 pUO2) of about 5 x 10-3, whereas others report values from
5 to 8 x 10-3. A rough boundary also influences the magnitude of the pressure fluctuations, as has been reported by Blake
(1969).
Using several roughness patterns, Blake obtained values of 2/(1/2 pUO2) ofthe order of 15 x 10-3 with d/d* = 0.1. The values of 2/(1/2 PU2) of the order of
7.5
x 10-3 obtained in this study are shown in figure 4. The scale at the left refers to data normalized with U0 whereas thescale at the right refers to data normalized with U , the velocity
just outside the boundary layer. These values appear reasonable in
comparison with those quoted above. The trend of the data in figure 4 is the same as the data with
Uo as the reference velocity. The explanation for the higher valueS af 2/(1/2 pD02) at He/Hd = 1.0
and
1.8
has already been discussed. These data indicate that the pressure fluctuations measured in this study are primarily a boundarylayer phenomenon. Approach flaw turbulence and free-surface perturba-tions are present but are of minor importance, and as a first approxima-tion the characteristics of the pressure fluctuaapproxima-tions at point 1 should not differ appreciably from wall-pressure fluctuations under a turbulent
boundary layer.
D. Spectral Density Function, Spillway Crest. Spectral Density functions of the pressure fluctuations obtained at point 1
for values of He/Hd equal to 1.0,
1.33, 1.5,
and1.8
with and without the turbulence stimulators are presented in figures 5 through 12.These data are in qualitative agreement with published spectra (Bull,
1963;
Serafini,1963;
Willmarth,1963;
Blake,1969)
of pressure fluctuations under a turbulent boundary layer in a zero pressuregradient. That is, with increasing frequency, the spectra decrease rapidly at low frequencies, level off with perhaps a very slight rise, and then decrease continuously at higher frequencies. There is no tendency toward periodicity discernable
in
any of the spectra obtainedat point 1. A quantitative comparison of these data with spectra of wall-pressure fluctuations under turbulent boundary layers was not
17
An elimination of corresponding spectra with and without the turbulencestiMulators indicates that there are more fluctuations
at frequencies above 16 Et. (fP/UO > 1 approximately)
for
the Cueswwithout the turbulence. stimulators than for cases with the stimulators.
This increase in the Spectral density is ofa "broad band" character, and IS probably the Consequence of fluatuations caused by the unsteady movement of the transition point from a laminar to turbulent boundary layer, discuised pretioubly. Both the low BNB values and the absence
of any
tendency toward peiiodicity.displaiedby
the spectral density functiOns of the pressure fluctuations at point 1 indiCate that the pressure fluctuations at the spillway crest are nota
likely cause Ofspillway vibrations.
E.. _Comparison With Prototype Data. No pUblished prototype spectra of pressure fluctuations were, found in the course of the
-extensive literature survey, conducted for this study.. However, the
prototype- tests conducted.at Chief Joseph Dam (Corps of Engineers, 1958)
cOntain:some.estimates of a "typical" amplitude of the pressure fluctuations at a point located approximately in the same position
.relative to the spillway crest as point 1 in the present model. Because of the fortunate circumstance that. 2/(1/2 pUO2) iSpractically,a
-constant, and recalling that the BNO-value is -a measure of the amplitude
Of the fluctuations, the amplitude of the fluctuations
in
the model May be scaled accordingto.the Ifroude criterion. Although the Chief Joseph dpillvay4s.a high.-overflow spillway, a reasonable estimate of theamplitude 'can be 'obtained using the ratio of'designheada for the
spill-way shapes as the scale ratio. A "typical" amplitude fOrflow at the design head Was selected from the oscillograms shown In 'figure 13 of
this report and compared with data in Table 13, test No. 21 (Corps of Engineers,
1958):
The amplitudeof the fluctuations recorded at Chief Joseph .Dam was 0.3 ft.'of.Water; When reduced to the present model(approximately 1161 scale ratio) the predicted amplitude is 0.0021 psi. The measured amplitude at the design head for the
Spillway
shape was0.0038 psi. In view of the uncertainties
in
picking a "typical" amplitude from such records, the agreement is pleasantly surprising.Data were also available from a model study conducted by Copp (1962), in which strip chart records similar to those of figure 13
were available. Scaling the amplitudes measured by Copp (1962) to
the present model yielded an amplitude of 0.0047 psi. Thus, the present model data, a previous set of model data, and a prototype test all
produce results of the same order of magnitude. The prototype estimate is lowest, as expected, because small extraneous fluctuations due to the experimental setup are much more significant in the model than in
the prototype. The above result also reinforces the conclusion that the pressure fluctuations at point 1 with the artificial roughnesses in place are the most representative of prototype performance.
IV. PRESSURE FLUCTUATIONS II., SPILLWAY TOE
A. RMS_ Value, Pressure Fluctuations, Point 2. The data Measured at point2 (refer to figure 1) included:
The EMS.:Value of the pressure fluctuations relative to 1/2 pUb7
as
a function of He/Hd for flow canditions With and without turbulence stimulators:Spectra of the pressure fluctuations with and Without turbulence stimulators at ratios of He/Hd equal to
1.33., 1.5, and 1..8.
3
The relative BM
as a function of with and without
from both series
H /Hd > 1.2, the
18
. Spectra and EMS values of the pressure fluctuations
with a hydraulic jump on the downstream apron for ratios of He/Hd equal to 1.0, 1.33, 1.5, and 1.8. value of the pressure fluctuations at the spillway toe
H /H are depicted in figures 14 and 15 for conditions
e d
the turbulence stimulators, respectively. The data of tests superpose very well for H
e/d H < 1.2; for
initial decrease for which both sets superpose, each case has a range
of fairly constant values. This range extends over a larger interval for the case without the stimulators; the data with turbulence
stimulators is slightly higher in the interval 1.2 He/Hd -5- 1.5 in
comparison with data obtained without the stimulators, as is reasonable
to expect. The most significant differences occur for He/Hd > 1.5.
Clearly some effect of the turbulence stimulators is responsible for the increasing intensity of the pressure fluctuations recorded for
He/Hd > 1.5 (figure 15).
In comparison with the experimental results observed at point 1
with turbulence stimulators, the RMS. value of the pressure fluctuations
for both sets of data obtained at point 2 are consistently higher. Part
of this increase is no doubt a result of the adverse pressure gradient
in
the spillway toe curie. Schloemer(1966)
has shown that the RMB value of the pressure fluctuations on a flat plate with an adverse pressure gradient is larger than at corresponding flaws with a zeropressure gradient. Although the pressure fluctuations at point 1 and point 2 are still of the same order of magnitude, spectral analysis shows that the characteristics of the pressure fluctuations at point 2
are quite different from those at point 1.
B. Spectral Density Fuhction, Point 2. As tight have been anticipated from the agreement of the RMS value for He/Hd 1.2, the spectra obtained with and without turbulence stimulators for He/Hd = 1
(figures 16 and 17) are quite similar. The shape of both these spectra differs Slightly from data obtained at point 1 with turbulence stimulators
(figure 5) under the same flow conditions, with a minor rise near
fP/Uo =
0.8.
For flows with H /Hd = 1.3, (figures 18 and 19), thee
minor rise has developed further, and the shape of the spectra, part-icularly those obtained with the stimulators present is now significantly different from corresponding data at point I (figure
8).
Spedtra obtained with He/Hd = 1.5 (figures 20 and 21) depict a small but definite peak near fP/U0
= 0.45,
which definitely does not20
occur.
in
the corresponding data Obtained at point 1. Finally withH /Hd =
1.8,
the peaking occursat a
Still laver value OffP/Uo 0.35
e
-or 0.4. Because of the already large values of the spectral density at low frequencies, this peak is not as apparent as at He/lid .= 1.5.
The spectra of the pressure fluctuations at the spillway toe thus indicate that there is some change in the flow pattern with
increasing values of He/Hd , and the peaking of the spectra indicate a tendency toward periodicity in the fluctuations. In order to obtain a better physical picture of the flaw pattern, flow visualization with
dye was attempted.
C. .Flow Visualization .-- Dye. Tests, 'Dye was injected through
the static piezometer.tap located:at point 1, and its -motion
along
theSpillway face was observed closely. The dye plume was photographed for
flows with
He/Hd
= 1,0, 1.3,. 145, and .1.8 with turbulence stimulators. glued to the spillway crest (figure l).
Figures 24a and 24b depict flaw conditions corresponding to
He/Hd = 1.0 and 1.3, respectively. The dye streak remained close to the spillway face and could be observed well downstream from the model.
For flow with He/Hd = 1.5 (figure 24c) the plume became more diffused in the lateral direction, and with some disturbances visible in the circular toe curve which augmented vertical diffusion of the plume. Finally with He/lid = 1.8 the dye spread out laterally well upstream from point 2 (figure 24d) and backflow along the spillway face was
clearly discernible. This observation indicates that local separation
was taking place. These tests show that the turbulence at point 2
should have different characteristics than at point 1, as was indicated by a comparison of the spectra of the pressure fluctuation at the toe
points discussed previously.
D. Discussion! The presence of an adverse pressure, gradient over approximately the upstream one-third of. the circular toe curve
(refer to Rouse, 1950, P. 539 for an illustration) explains the. above observation. At adverse pressure gradient promotes separation.
Indeed, as the dye test's have shown, separation does take place at H /Hd =
1.8.
The tendency toward periodicity in the pressurefluctu-e
ations
is
a direct consequence of this separation..Some investigations (ratinclaux,
1966;
Locher, 1967) of separated flaws behind idealized gate shapes have shown that the reattachment point of the separation zone is unstable and oscillateswith a dominant frequency. In this case, both the point of separation
and reattachment are not fixed, but oscillate. The peaking observed In the spectra at point 2 is more probably a result of this unstable oscillation. 'Furthermore, the mean position of the separation point
is dependent on the boundary layer characteristics. Thus, changing the boundary layer development along the spillway should change the mean position of the local separation zone and thereby change the
characteristics of the pressure fluctuations at point 2. One might
also speculate that the turbulence stimulators bring the velocity profile into closer agreement with the irrotational velocity profile, and that the mean position of the separation point moves downstream In a manner similar to the shift in separation point on a sphere as
its boundary layer becomes turbulent. The reattachment point would thus move closer to point 2, and continue to do so with increasing
Reynolds number. The turbulence, and hence the pressure fluctuations at point 2, would therefore be more intense with the turbulence
stimulators present than without, as was observed (figures 14 and 15). A series of careful measurements would be required to verify this
hypothesis.
E. Spillway,TOe_PressUre Fluctuations, Hydraulic Jump.
Spectra and RIC Values of. the pressure fluctuations Were also measured
with
a hydraulic
jump on the downstream apron floorfor
.He/Hd = 1.0,1.3, 1.5, and 1.8. 'The. jump Was formed by raising a tailgate at the downstream end of the flume. Some difficulty was encountered in
22
reproducing the data because the pressure fluctuations were quite sensitive to the position of the jump with respect to the spillway
toe. Therefore, the results of these tests should be regarded as more qualitative than quantitative, and indicative of trends rather
than absolute values. These data are, however, of the same order of magnitude as those reported by Vasiliev (1967), but a direct
comparison cannot be made because the Froude number for the two cases
is not the same. The EMS values have been plotted on figure 15 for comparison with the data obtained without a jump. Notice the highly significant difference for He/Hd =
1.8.
Spectra of the pressure fluctuations at point 2 are shown in figures 25 through
28.
Photographs of the spillway model weretaken during the measurement of the spectral density 'function and are shown in figure 29. Furthermore, dye tests were run during, between, and after the measurement for the determination of the spectrum. The dye showed that separation would occur if the lead portion of the jump was above the downstream third of the circular toe curve for He/Hd = 1.3
and 1.5. By moving the jump so that the position is as shown in figure 29, no clearly visible separation occurred. Spectra at H
e/Hd = 1.3 and 1.5 were obtained with no apparent separation.
Sep-aration did not occur with H
ea
/H, = 1 until the jump was positioned well up on the spillway. However, with He/Rd =1.8,
separationoccurred even if the jump were not on the spillwv. The separation for
He/Rd =
1.8
was made visible by dye as shown in figure 30. Notice how far upstream the separation point is located on the spillway face.The spectra measured with a hydraulic jump, on the spillway
toe are generally similar to corresponding spectra without the jump. Despite the rather large increase in RMS value at He/Rd = 1, the
spectra have a similar shape. Apparently the increase in RMS is a
consequence of the wave action of the jump which is randomly distributed
in both amplitude and frequency. Again, a peaking occurs in the
spectra for H /Hd = 1.3 and 1.5. Because the relative height of the
e
23
the jump augments the dominant fluctuations slightly. The tendency toward periodicity seems to disappear almost entirely from the spectra
at H /Hd = 1.8. The change in position of the separation point to a
e
location well upstream from point 2 explains this observation. A
significant increase in free surface fluctuations, as well as a rise in the water surface upstream from the spillway model' was also noted with He/Hd = 1.8. The significant increase in BNB value recorded is
thus a consequence of the change .in flow pattern, increased free-surface
fluctuations; the action of the hydraulic jump, and increased turbulence
in the approach flow. The change in head on the spillway with the jump present indicates a partially submerged condition as well.
V. DISCHARGE COEFFICIENT
A. :Dimensional Considerations. Measurements to obtain the discharge coefficient as a function of the ratio He/Hd were a routine procedure; the data have been plotted in figure 31. In contrast to results fram high-overflow spillways, the discharge coefficient for this model does not increase monotonically, with increasing ratio of H /Hd' but increases to a maximum near H /Hd = 1.35 and then decreases
e e
Slightly. Notice that there is only a 1.3 percent increase in C from H
e/Hd = 1.0 to "H /Ed = 1.35.e This behavior of the discharge coefficients
for low-ogee crest shapes has
been
observedby others as well (Bradley,
1952).
Before proceeding to an explanation of the observed variation
of C with He/Hd
'
let us quickly review some elementary concepts associated with the determination of spillway discharge coefficients. Since the discharge per unit width q depends upon the same variables
as the pressure at a point p , a general form of the relationship for the discharge coefficient may be written as
224
PH
(
H ' a
4/T
d d -dThe parameter .P/Hd represents the effects of the approach velocity, He/Hd. the effects of,headt_other than the design head, k/Hd the
relative roughness, and IR and IW the viscous and surface tension effects, respectively.- Included in the terms $i. are such effects as the slope of the upstream face of the spillway, (45° in this case), the position
of
the downstream apron, and the radius of the spillway toecurve, for example.'
Experiments by Eisner (1933) (see also Cassidy, 1964, p. 11) showed that the discharge coeffiCient decreased as the relative
rough-ness increased. This effect was observable in the present study. Dis-charge coefficients obtained with the turbulence stimulators were less than corresponding values obtained without the turbulence stimulators. All of the data shown in figure 31 were therefore computed from
measure-ments taken without the artificial roughness on the spillway crest. Effects of viscosity and_surface tension on the discharge coefficient
can be ignored if the model scale is large enough. A more complete discussion is given by Cassidy (1964, 1970).
The effects
of
the approach depth, heads differing from thedesign lead, .upstream face slope, and downstream apron floor position
have beeneXtentively investigated by the U.S..Army Corps
of
Engineers(1965) and the U.S.E.R. (1948, 1960). The results
Of
these teststogether with some physical reasoning explains the behavior of the
discharge -coefficient Observed during the present investigation.
For-a given spillway geometry, the functional relationship
reduced to
C =
0 (H(g)
d 1/. -He3/2 e d
C Q/LHea/Z
there results the relationship
C = (He/Hd)
which has been plotted in figure 31.
B. Discharge Coefficient at the Design Head,
Hd. A
com-parison of the value of the discharge coefficient obtained at the design head for the spillway shape Hd with the suggested design
curve shown on plate 32 (Corps of Engineers,
1965)
reproduced herein as figure 32 shows that the value of Cd3.85
is in better agreementwith the
45°
weir curve than with the suggested curve. The suggested design curve for the45°
crest face shown on figure 32 was obtained byincreasing the data for the
45°
weir experiments by three percent. This increase was obtained by a comparison of the 90° weir data with900 spillway model studies (Corps of Engineers,
1965, p. 54).
However,the increase in C for the inclined upstream face is not a constant,
but varies with P/Hd (OBB,
1960).
At a value of P/Hd =0.44,
theratio of C for a vertical upstream slope to C for a
45°
slope isonly about 1.017 according to USBR
(1960, p. 277).
Multiplying the value of C for a vertical upstream slope given by USBR(1960)
bythe ratio 1.017 yields an expected discharge coefficient of 3.89 for flow at the design head. The value C = 3.85 obtained in this study with flaw at the design head is in good agreement with the calculated
value of
3.89
as well as with data for similar shapes presented by Bradley(1952).
Flow over a spillway may be classified as free or submerged. An overflow weir is said to be submerged when the water level on the
downstream side of the weir begins to affect the discharge characteristics
Of the weir. Similarly, flow over a spillway is said to be free if at the design head, Hd , the discharge coefficient does not differ
26
lower nappe surface corresponds to the shape of the spillway crest. Submergence effects reduce the discharge coefficient and may be
caused by water levels in the downstream channel exceeding the normal depth for the channel, or by raising the elevation of the downstream
apron floor. Free flow is a term which refers to flow conditions for which submergence effects cannot be detected; there is no sharp
demarcation between free flow and submerged flow.
A check for effects of the downstream apron elevation shows that the present shape is a "borderline case for flow with He/Hd = 1.0. It has been suggested by the USER (1948) that (hd + d)/He be greater than 1.7 if there is to be no submergence effects due to the apron
floor elevation. The quantities' /id
and d
are defined on figure 1.At the design head Hd , the value of (hd + d)/He for the present
shape is 1.68; the influence of this factor on C at the design head according to the Corps of Engineers (1965, plate 33) is about 0.1
per-cent, or practically negligable.
C. Dischar:ke Coefficient for He/Hd > 1.0. However, as the
head on the spillway He increases, the elevation of the apron floor becomes an important factor. 'For free flow over the spillway at the
design head, the pressure on the spillway crest is equal to or slightly greater than atmospheric pressure as experimental results reported by the USBR (1960) have shown and as indicated by the results at point 1 for the present study (figure 33). If free flow occurs for all heads,
then as H /Hd increases, the pressure at the spillway crest decreases,
e
the minimum value being limited by cavitation.
Figure 33, shows that the mean pressure at point 1 does not
decrease monotonically as He/Hd increases. Therefore, free flow does not exist at all heads for this particular spillway shape, and the value of the discharge coefficient will.be affected accordingly. As He
increases, the pressure in the spillway toe curve increases in proportion to the square of the velocity, since the pressure in the toe curve
27
depends upon the centrepital acceleration v2/r (point 2, figure 34). The effects of this pressure increase reach farther and farther upstream with increasing H, and finally begin to affect the pressure
distri-e
bution on the spillway crest, as may be seen from the data of figure 33. Notice that as He/Hd increases, the mean pressure at point I first
decreases in the interval from He/Hd = 1.0 to about He/Hd = 1.35 as
would be expected, but then increases for He/Hd > 1.3. This change in the pressure distribution is reflected in the behavior of the
dis-charge coefficient, which attains its maximum value between He/Hd = 1.3
and 1.4. Based upon these observations, it is clear that changes in
the pressure distribution on the spillway crest attributable to the
effects of the spillway toe curve and the apron floor elevation are
responsible for the observed variation of the discharge coefficient at
heads{ greater than the design head. Furthermore, the relatively small
decrease in the pressure on the spillway Crest from He/Hd = 1.0 to 1.35 is reflected in the very small increase in discharge coefficient for
these heads.
VI. SUMMARY AND DISCUSSION
The following statements summarize the results, obtained from
measurements of the pressure fluctuations on the spillway Model at
point's 1 and 2.
1, The pressure fluctuations. on the spillway face
,at-pint 1 are primarily a consequence of turbulence generated
in
the boundarylayer. Turbulence in the approach flow and free surface effects are
Data obtained with the turbulence stimulators in place are probably most indicative of prototype characteristics.
Spillway-crest pressure fluctuations at point I did not
exhibit a "dominant" frequency for any of the flow conditions observed during the investigation, although there was a significant increase in the intensity of the fluctuations for H
28
I. The pressure fluctuations at the spillway toe (point 2)
do have a "dominant" frequency at heads greater than the design head
for the spillway shape, Hd . There is no appreciable tendency toward periodicity for flow at the design head Hd .
The pressure fluctuations at point 2 are complicated by the presence
of
an adverse pressure gradient occurring over the upstream one-third of the spillway toe curve. The MB value of the pressurefluctuation is of the same order as those at point 1.
A hydraulic jump on the downstream apron significantly increases the intensity of the pressure fluctuations at point 2. Although there is no significant trend toward periodicity for flows
with He = Hd , some peaking in the spectra does appear for He/Hd = 1.3
and
1.5.
Not one of the measurements
of
the pressure fluctuations on the spillway face showed an unusually high BNB valueor
tendency toward periodicity With flow at the design headfor
the spillway shape.The results of a study of the discharge coefficient data may
be summarized as follows:
The discharge coefficient, C =
3.85,
obtained with flow at the design head for the spillway shape is good agreement with predictions based on USBR data and with discharge coefficients ofsimilar shapes reported in the literature.
The discharge coefficient for the spillway shape investi-gated during this study does not increase continuously with increasing head on the spillway, but reaches a maximum at H
e/Hd =
1.35
and thendecreases. This behavior is attributable to the influence of the
spillway toe-curve and a6ron floor elevation of the pressure distri-bution at the spillway crest as the head on the spillway increases.
These results are significant with respect to present hydraulic design criteria, because the discharge coefficient for high-overflow spillways continues to increase for He/Hd > 1.0. The crest shape is often proportioned for a design head H of
0.75
times the maximum head29
Such Ei crest is called an underdesigned shape. Therefore, the discharge
coefficient for the underdesigned shape for flows at the maximum head will be greater than a crest shape designed with Hd = Hm . The under-designed shape thus passes the design flood with a shorter crest length
than
a
shape designed with Hm = Hd .Tests with low-ogee crest shapes and free flow conditions also
show that the discharge coefficient continues to increase as
He/lid
increases (Corps of Engineers, 1965, plate 6). Thus, there is a trend
toward underdesigning low-ogee crest shapes (Campbell, 1969) although present practice of the U.S. Army Corps of Engineers does not permit low-ogee crests to be underdesigned (Corps of Engineers, 1965, p. 13). As
far as the present model is concerned, the experimental evidence
presented in figure 31 indicates that there is practically no advantage
gained by using an underdesigned shape for this case.
VIII. CONCLUSIONS
On the basis of the results summarized in the preceding section, the principal conclusions of this study are:
Pressure fluctuations on the spillway crest with flow at
the design head for the spillway shape did not display an unusually high RMS value or alarming trend in the spectra and are probably not a source
of excitation of spillway vibration.
The discharge coefficient for this spillway configuration does not increase significantly for ratios of the head on the spillway
to design head for the spillway shape, He/Hd , greater than 1.0.
In View of 2 above, and the trends in the spectra of the pressure fluctuations at point 2 with flows greater than the design head
for the spillway shape, it is recommended that the underdesign of
low-ogee crests proportioned according to data on plates 26, 29, and 34, Corps of Engineers (1965) with 0.3 P/Hd 0.57 and a geometrically similar
30
location of the downstream apron floor not be permitted.
In closing;, it Should
be recalled that the above conclusions . .are applicable only to uncontr011ed ogee .crests, since the effects of gate. piers, gate slots, etc. were not considered.
Glover, J.R. & Giaquinta, A.R
Unsteady Flow Variables. 31
REFERENCES
Bendat, J.S. & Piersol, A.G.
1968
Measurement and Analysis ofRandom Data. New York: John Wiley and Sons.
Blackan, R.B: & Tukey, J.W.
1958
The Measurement of Power Spectra. New York: Dover Publications.Blake, W.K.
1969
Turbulent Boundary Layer Wall-Pressure Fluctuationson Smooth and Rough Boundaries. Dept.
of
Naval.Architecture andMarine Engineering, Technical Report No.
70208.
Bradley, J.N.
1952
Discharge Coefficients for Irregular OverfallSpillways. U.S. Bureau of.Reclamation Engineering Monographs No.
9.
Bull, 14.K.
1963
Properties of the fluctuating Wall-Pressure Field ofa
Turbulent Boundary Layer. ,AGARD Report No.455.
Campbell, F.B.
1969
Report on Spillway Vibration Studies. Vicksburg.Cassidy, J.J.
1964
Spillway Discharge at Other Than Design Head. Ph.D. Dissertation, University of Iowa, Iowa City, Iowa.Cassidy, J.J. '1970 Designing Spillway Crests for High-Head Operation.
ASCE J.
of
the Hydraulics Division.96, 745-753.
Copp, H.D.
1962
Preliminary Report on Hydraulic Sectional ModelStudies -- Wells Hydro-Combine. Institute of Technology, Washington State University, Washington.
Eisner, F. 1933 Ueberfallversuche in verschiedener ModellgrOsSe.
Preussiche VersuChenstalt flit Wasse/4bau und Sthiffbau, Berlin.
1968
Real Time Digital Processing ofJ. of
Rydraulic Research, IAHR.6, 219-231.
Jenkins, G.M. & Watts, D.G.
1968
Spectral Analysis and its Applications. San Francisco: Holden Day.Locher F.A. & Naudascher, E.
1967
Some Characteristics ofMacro-TIlrbulence in-Flow Past a Normal Wall. Proc. XIIth Congress 14HR.
2, 298-307.
Locher, F.A.
1969
Some Aspects of Flow-Induced Vibrations of HydraulicControl Gates. Ph.D. Dissertation, University of Iowa, Iowa City,
Iowa.
Pickering, G.A.
1968
Spillway for Kaysinger Bluff Dam. Hydraulic Model Investigations. U.S. Army Engineer Waterways Experiment32
Rouse, H.
1938
Fluid Mechanics for Hydraulic Engineers. New York: Dover Publications.Rouse, H.
1950
Engineering Hydraulics. New York: John Wiley and Sons.Rouse, H.
1961
Laboratory Instruction in the Mechanics of Fluids. The University of Iowa Studies in Engineering Bulletin 41.Serafini, J. S.
1963
Wall-Pressure Fluctuations and Pressure-VelocityCorrelations in Turbulent Boundary Layers. AGARD Report No.
453.
Schlichting,- H.
1960
Boundary Layer Theory. New York: McGraw-Hill.Schloemer, H.H.
1966
Effects of Pressure Gradients on Turublent Boundary-Layer Wall-Pressure Fluctuations. U.S. Navy UnderwaterSound Lab Report No,
747.
Tatinclaux, J.C.
1966
Pressure Fluctuations in the Vicinity of NormalWalls of Variable thickness. M.S. Thesis, The University of Iowa,
Iowa City, Iowa.
U.S. Army Corps
of
Engineers1965
Hydraulic Designof Spillways.
Engineer Manual2M 11101603..
U.S..Army Corps of Engineers
1958
PrototypeSpillway
Crest Pressures, Chief Joseph Dam. Misc. Paper NO.2=266.
Army Corps of Engineers
1965
Hydraulic Design Criteria. Vicksburg U.S. Bureau of Reclamation1948
Studies of Crests for Overfall Dams.Boulder Canyon Project, Final Reports, Part VI, Bulletin
3.
U.S. Bureau of Reclamation1960
Design of Small Dams. Washington:U.S. Government Printing Office,
Vasiliev, O.F.
1967
.Statistical Characteristics of Pressure Fluctuations in the Region of Hydraulic Jump. Proc. XIIth Congress, IARR.2, 1-8.
Wiener, N.
1930
Generalized Harmonic Analysis. Acta math.55, 117-258,
Wi1Imart,h,
w.w, 4
Wooldridge,C.E.
1963,
Measurement'sof
the CorrelationBetween the Fluctuating Velocities and Fluctuating Wall-Pressure
APPROACH CHANNEL FLOOR ROUGHNERS, ELEMENTS N.
SPILL WAY MODEL
Figure 1.
Definition sketch
X.. 1.893 He
1.324 Ha
hd
Hd Hd Hd H /H = 1.0 He Hd = 1'3 HeHd = 1-5. H /Hd = 1-8 Figure 2.
Photographs of the experimental apparatus with discharges corresponding to H /H
= 1,0, 1.3,
e
d
0.06'
0.04
on2
0.8
0
ees
RMS VALUE OF THE
PRESSURE FLUCTUATIONS
POINT
1
NO BOUNDARYLAYER TR IP
o
Different 'Experimental
a Ruins
He
Hd
Figure 3.Variation of the relative rms value of the
pressure
fluctuations at point 1 with.He/Nci.
No boundary layer trip,.