simulation of main flow and secondary flow in a curved open channel
J,oh.G.S.Pennekamp and R. Booij
Report no. 1 - 84
I
Laboratory of Fluid Mechanics Department of Civil EngineeringI
Delft University of TechnologyI
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
Improved simulatien ef main flow and secondary flow in a curved open channel
Joh.G.S. Pennekamp and R. Booij
Report no. 1 - 84
Laboratory of Fluid Mechanics Department of Civil Engineering Delft University of Technology
.01
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
2
SummaryFor the computation of depth averaged flows in tidal
channels and rivers a recently developed fully.implicit finite
difference method of the ADI-type proved vastly superior to the
partly explicit variantsf commonly used. The fully implicit
variant allowed a relatively large time step in combination with
arealistic lateral diffusion coefficient.
The fully implicit method, used, requires a square grid for
the time being. The irregular numerical representation of the
sidewalls of a curved channel or flume gives rise to disturbances
in the computed flow field. In this report a correction scheme
to keep the disturbances small is considered. In this scheme a
correction is realized by a modification of the depth near the
osidewalls. The correction scheme is actually a combination of
twe different corrections. The first correction is aimed at
forcing the velocity at corner points of the computational grid
in the direction of the physical boundary. The second correction
compensates for the local narrowing and widening of the flow by
the irregular numerical representation of the ~idewalls.
The influences of both corrections appear to be small when
applied separately, but the combination of the two corrections
has, however, important consequences. The application of the
combined corrections on the simulation of steady flow in a curved
flume with a rectangular cross-section annihilates nearly the
important disturbances caused by the use of a rectangular grid.
For a curved flume with the geometrical proportions of a river, in
which case the disturbances proved much less important, a less
important amelioration is obtained. For flow in tidal channels
even smaller disturbances and less improvement are ~xpected. The
effect of the corrections is very sensitive to the exact waterdepth
at the sidewalls. In time dependent flows it is therefore
difficult to apply the right corrections. Further improvement
\~iII have to await the possibility te use a curvilinear grid in
I
I
I
This research is aimed at the computation of secondary flow in tidal channels based on the depth averaged velocity field. The improved computation of the depth averaged velocity field
I
brings about an improved computation of the secondary flow. Knowledge of the seccndary flow in a tidal channel is essential for predictions about the morphology of the bottom.I
I
I
I
I
I
I
I
I
I
I
I
.
I
I
I
I
I
"
I
4I
ContentsI
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
·Summary 2 Contents 4 List of figures 5 1. Introóuction 6~
.
...::.
.
Mathematica! description 88 9
2.1 Depth averaged flow
2.2 Secondary flow
·3. The computation of depth avpraged flow
3.1 The flow configurations
3.2 Reproduction of the velocity field
11. 11 12
4. Correction schemes for the computation
of depth averaged flow
4.1 Velocity direction correction
4.2 Flow area correction
14
14 17
5. Improved reproduction of the depth averaged flow 19
6. Reproduction of secondary flow 22
7. Conclusions
References 24
Notation 25
I
I
I
5 Li st of f igUt-es 1. Definition sketch.I
~"_'.
I
4a=. ..J.I
6.I
7.I
8. 9.I
10.I
11.I
12.I
13.I
I
14.I
15.Geometry of the DHL-flume with the plane bed.
Geemetry of the DHL-flume with the uneven bed.
Computational grid with !J.
=
0.40 m.Depfh averaged velocity distributions in several
cross-sections (plane bed configuration).
Depth averaged velocity distributions in several
cross-sections (uneven bed configuration).
Obstruction of the flow at the euter side of the bend at
Widening 6f the flow at the inner side of the bend at 25°.
Depth averaged velocity distributions in several
cross-sections (plane bed configuratien), combined corrections.
Depth averaged velocity distributions in several cr oss-sections' (plane bed configuration) , flow area correction.
Depth averaged velocity distributions in ~everal
cross-sections (plane bed configuration) , velocity direction
correction.
Depth averaged velocity distributions in ~everal
cross-sections (uneven bed configuration), combined cerrections.
Depth averaged velocity distributions in several
cross-sections (uneven bed configuration) , determined by the bottom
friction.
Secondary flow intensity di'stributions in several
cross-sections (plane bed configuration). No correction applied.
Secondary flow intensity distributions in several
cross-sections (plane bed configuration). Combined corrections.
16. Secondary flow intensity distributions in several
cross-I
sections (uneven bed configuration). No correctien applied.17. Secondary flow intensity distributions in several
cross-I
I
I
I
I
"I
I
I
I
I
I
I
I
I
I
I
I
I
,
I
I
I
6 1. IntroductionA thorough knowledge of the secondary flow in tidal
channels with alluvial bottoms is required for the predictions of
their morphology~ because this secondary flow gives rise to
hottom slopes transverse to the main flow. This research, which
is financially supported by.the directorate of the Deltadienst of
Rijkswaterstaat, concerns the determination of the secondary flow
in tidal channels of estuaries like the Eastern Scheldt, based on
a known depth averaged velocity field. The depth averaged
veloeities must be computed with a high accuracy in order to make
possible areasonabie determination of the secondary flow.
For the computation of the depth averaged veloeities.
generally an implicit finite difference method of the ADI-type is
used. In such a method the depth averaged equations of motion and
the depth averaged continuity equatien, together cal led the
'shallow water equations, are solved by means of an Alternating
Direction Implicit computatian using a staggered spatial grid.
Although the velocity and waterlevel parameters are treated
implicitly, in general the cenvective and diffusion terms are,
however, treated explicitly in the difference equations. In this
partly explicit representation a large diffusi~n coefficient is
required in order to suppress a possible instability lest an
unecanomically small time step has ta be used eVreugdenhil and
Wijbenga, 1982}. Such a large diffusion caefficient, compared to
the physical eddy viscasity, severely hampers the representation
of the velocity distributions in the considered steady or
quasi-steady flow (Pennekamp and Booij, 1983 and Booij, 1983).
Recently the Dienst Informatieverwerking of Rijkswaterstaat
developed a fully implicit finite difference method of the
ADI-type. This method is usually referred to as Miniwaqua. In this
fully implicit methad no diffusion coefficient is r~quired'for
stability, so arealistic diffusion ~oefficient can be introduced
(Stelling, 1983). To investigate the reproduction of the depth
averaged flow in circumstances comparable to bends in tidal
channels, using this fully implicit method, computations were
executed for steady flow in a curved flume of the Delft
Hydraulics Laboratory (Booij and Pennekamp, 1983). The
computations concerned two different battom topographies, for
bath of which extensive masurements were available: a rectangular
I
cr-oss-sectian <de Vr-iend and Koc h, 1977) and an uneven bott.omI
I
I
I
I
I
7topogr:aphy as found in river b.ends (de Vriend arrdKoc h , 1978). The reproduction of the depth averaged velocity field was satisfactory in the uneven bed case. The influence of the
.
secondary flow on the main flow cannot be reproduced by Miniwaqua. This influence is not very large because of the gentie curvature of the flume. The influence of the secondary flow is very small in the case of the rectangular cross-section. Here disturbances connected with the numerical representation, which also appear in the uneven bed configuration,
but are not very important there~ are however very strong,
I
because of the large sidewalls in this plane bed configuration.I
I
I
I
I
I
I
I
I
.
I
I
I
I
I
The reproduction of the secondary flow is reasonable. Only in regions in which the depth averaged vel~city field is strongly ·influence~ by the sidewall disturbances, a defective reproduction
occurs.
In this report a correctien procedure te suppress the disturbances ~onnected with the numeri cal representation is investigated.
"I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
8 2. Mathematical description2.1 Depth averaged flow
The computation of the depth averaged flow is based on
shallow water equations of the form (Booij and Pennekamp, 1983).
a
u
a
u
au
aÇ
a
t + u
ax
+ v
a
y + g
ax
/2 2' r guVu +v
_ ~
_ (')...
+ --c-
2- h ph H V (1)a
v
+ u
a
v +
va
v +
as
at
a
x
a
y
gay
J 2 2'1L vVu
+v
_
'rwy+
n
u
è2 h ph +o
(2)a
ça
(hu)
- + ___.;._,-+a
·
t
ox
a(hv)
a
y
=
0 (3)In these equations the following notation is used (see also
definition sketch, fig. 1):
x ,y horizontal coordinated, z is the vertical coordinate;
t time;
u,v depth-averaged velocity-component in x-,y-direction;
waterlevel above reference level;
acceleration due to gravity;
9
h watet-depth;
p ma ss density;
components of surface shear stress;
L ,'r
wx wy
n
Cor-iolis parameter-: 2w sin <P ,wher-e<P is the geographicC
latitude and
w
is the angular velocity of the rotationof the earth;
Chézy coefficient;
diffusion coefficien~
I
I
I
I
I
I
9Shallow water equatio~s ~an be obtained by integrating the
Reynolds' equations for turbulent, flow over the depth, assuming a
hydrostatic pressure distribution along each'vertical. Same
.
additional assumptions about the shear stresses are reflected in
the form of the shallow water equations used (eqs. 1,2 and 3).
Jhe bottom shear stresses are assumed to act opposite to the
directien of the mean velocity vector and to vary with the mean
velocity squared. The effective stresses in vertical planes are
replaced by diffusion terms, with an isotropic diffusion
coefficient,E, which is constant in time and throughout the flow.
I
A goed choice for E is an average value of the lateral eddyviscosity (Booij and Pennekamp, 1983)
I
I
I
I
I
I
I
I
I
~I
I
I
I
(4)The overbar in expression (4) indicates averaging over the
flow field.
The shallow water equations are solved nu~erically with
Miniwaqua.
2.2 Secondarv flow
The flow pattern in river and channel bends is quite
comple}~. A main flow can be defined by the horizontal velocity
component, us(z) in the direction,s, of the depth averaged
velocity. In addition to this main flow a secondary flow,
defined by the horizontal velocity component, un(z), in the
normal direction, n, can be important. The main flow velocity
can be described properly by its depth averaged value, us' but
the depth averaged value of the secondary flow velocity is
zero. The secondary flow can be described by its intensity, i.e.
half the averaged absolute value (de Vriend, 1981),
I
n = 2~
J
Iun(z) I dzdepth
(5)
In tidal channels two contributions to the secondary flow can be disti nguished, curvature of the mai n f1ow and the Corrol is
acceleration. In the curved flume considered in this report the
enly important souree of secondary flow is the curvature of the
I
I
I
10
flows the secondary flow can be assumed fully developed
everywhere. The intensity of the fully developed secondary flow
caused by a curvature of the main flow with a radius of curvature
R is (Booij and Pennekamp, 1983) I
=
n u h c 12--2 R K (6)I
where K is von Karman's constant and c is a function of the ChézyI
I
I
I
I
I
I
coefficient only. The value of c derived in most theoretical e~aminations is slightly lower than the value suggested by
·measurements in flumes (de Vriend~ 1981). For C
=
50 mIls thetheoretical value is about 0.25 but the value obtained from
measurements is about ~ij3 times as large.
The computation of the intensity of the secondary flow in
.this report is based on equation 6. The radius of curvature of
the main flow, R, is calculated using
au
n=
R
uas
s
(7)
where un is the depth averaged velocity component normal to the
direction of the flow at the point in which R is calculated.
The intensity of the secondary flow as given by equation (6) does
only depend on the depth averaged flow. An incor-rect
reproduction of the depth averaged flow is reflected in a
corresponding incorrect reproduétion of the secondary flow.
Disturbances of the depth averaged flow near the sidewalls of the
flume can deteriorate the computation of the secondary flow as
the radius of curvature of the main flow depends on a gradient of
the main flow velocity (see eq. 7). lts correct computation is
I
hence severely impaired by a scatter of the depth a~eraged"velocities in neighbouring grid poiMts, caused by the irregular
I
I
I
I
I
I
I
I
I
I
I
I
I
I
113. The computation of depth averaged flow
3.1 The flow confiqurations
To investigate the accuracy of the computations of the depth
averaged flow by Miniwa~ua a comparison with measurements was
executed CBooij and Pennekamp, 1983). No measurements of flow in
tidal channels, with enough precision and detail to make an
investig~tion of this accuracy feasable, are known. Dnly flows
in Laboratory flumes are investigated thoroughly enough.
The calculations are executed for a flume in the Delft
Hydraulics Laboratory,called the"DHL-flume in this report. In
this large flume, with a rather gentie bend CB/Rf
=
0.12, with Bthe width and Rf the radius of curvature of the channel axis) of
almost 900, extensive measurements were e~ecuted for two
different bed configurations. In the first series of experiments
the bed of the flume was plane and the cross-section rectangular
(see fig. 2) (de Vriend and Kocri, 1977). In the other series of
experiments the flume was provided with a fixed uneven bottom of
I
more or less the same shape as in a natural river bend (see fig.I
I
I
I
3)(de Vriend and Koch, 1978). The flume with the uneven bed is
then also a fair model of a bend in a tidal channel (Pennekamp
and Booij, 1983). The cross-sections in which·the measurèments
were executed are indicated in fig. 2 and fig. 3. The
measurements were limited to steady flow. Measurements of time
dependent flow in a curved flume in the Laboratory of Fluid
Mechanics of the Delft University of Technology are being
elaborated.
The flow is mainly controlled by the bottem friction. The
distributions of the depth averaged velocity in the
cross-I
~ections reflect therefore mainly the depth distributions.Deviations from the measurements are somewhat easier to analyse
.
1
I
I
in the plane bed configuratien than they are in the uneven bed
configuration., Besides, the effects of the irregular numerical
representation of the sidewalls by the square grid, used, are
much larger in the plane bed configuration. The plane bed
configuration is properly speaking too strong a test for the
re~roduction of flow in a tidal channel. It gives however an
indication of the result~ to be expected from the numeri cal
I
reproduction of the flow, measured at the Delft University ofTechnology. The cross-section of the flume in which these
"
I
masurements were executed is also rectangular.-
-
I
I
I
I
I
I
I
I
I
12The distance between neighbouring grid points in the
considered computation was ~= 0.40 m (see fig. 4) ~ and the time
·step used was ~t= 1.5 s. These values were chosen because of
available memory space, accuracy, stability an~ efficiency.
The computations were e~écuted with a free-slip boundary
condition
au
s
an
wal1
=
0
(8).The direction of n is perpendicular tcithe wall. The wall shear
.stre~ses, when using the physically more attractive no-slip
boundary condition~ are much too large because of the relatively
large grid spacing (Booij and Pennekamp, 1983).
3.2 Reproduction of the velocity field
The reproduction by Miniwaqua of the depth averaged velocity
fields of the DHL-flume for both bottom configurations was
I
satisfactory (Booij and Pennekamp, 1983). The comparison of thecornputations and the measurements are given i~ fig. 5 and fig. 6.
Fig. 5 shows the results for the rectangular cross-section and
fig. 6 for the uneven bed configuration. In both figures the
reproduction of the shifting of the maximum velocity to the inner
side of the bend at the beginning of the bend can be appreciated.
This effect shows the flow in this region to behave like a
potential flow. The overall distribution of the depth averaged
velocities are satisfactory. The importance of the
bottom friction is evident and is reproduced in the computations.
This is an important improvement compared to the finite
difference schemes of the ADI-type hitherto, in which hori~ontal
diffusion of momentum appeared to be tOD important.
The influence of secondary flow on the main flow is not
reproduced by Miniwaqua. This influence is very small in the
plane bed configuration, but somewhat stronger in the uneven bed
configuration. Consequently, the gradual shifting of the main
flow to the outer side of the bend is slightly too small in the
computation, as only the shift caused by the bottom topography is
accounted for. This effect is to be expected in the computation
I
of flow in tidal channels tOD, -but it will be somewhat smallerI
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
~I
I
I
I
I
"13th~re, because the flow is relatively shallower.
The most important failure in the reproduction of the depth
averaged flow is connected with the irregu.lar numerical representation of the stdewalls. Two different kinds of
disturbances can be distinguished (Booij and Pennekamp, 1983).
One disturbance is a scatter in the values 6f the depth averaged
velocity in neighbouring gid points, caused by the irregular
boundary. This effect is especially obvious at the outer side in
the first half of the bend and at the inner side in the secend
half, where the irregularities i~ the"representation of the
sidewalls appear to act as obstructions to the flow (see fig. 7).
A more important disturbance connected with the numeri cal
representatien of the sidewalls is found at the inner side in the
first part of the bend and at the outer side in the second part of the bend. There the flow does not follow the local widenings
of the flow~ presented by the irregularities in the representation
of the sidewalls (see fig. 8).
In Chapter 4 correction schemes ta suppress
.
these""I
I
I
I
I
"
I
1
1
1
1
1
1
-
I
1
I
1
1
I
144. Corrpction schemes fer the computatio~ of deptM avpraoed flow
4.1 Velnritv directien correction
Correction schemes' to improve the computation of the depth
averaged velocity near the sidewall boundaries in the curved part
of the f!ume are rel~ted to the exact numeri cal representation of
the flow there. In order to discuss pessible correctien schemes
a short description of the staggered grid~.used, can be helpfull.
In Miniwaqua the surface level~ the b6ttom level and the two
c omporierrts o+ the vel oc ity at-e def ined at; "dif ferent pI aces. The
locations in the horizontal plan~ of the various places, makinç
up a staggered grid, are shown below.
place where the surface level is
+
+
+
defined.place where the bottOffilevel,b,
o
o
is defined.+
+
place where the x-component ofthe velocity is defined.
Cirection
X -
d
i
reet ion
place where the v-component of
the velocity is defined.
Four different neighbouring elements make up a computational
molecule in.which these different elements bear the same indices
although their locations in the horizontal plane are not the
same. 0 0 0
+
+
+
0I
0 0v
" "
b.
"
IJ
IJ
+
+ç
j
j
molecule
i
,
j
"
I
0I
0 0j
t
+
+
+
-?
1I
I
I
I
I
I
I
I
I
I
Whenever an impermeabie ~oundary occurs in the space- ""
staggered grid~ the existence of.this boundary is simulated in
the ~umerical model by set~ing the velocity" of the nearest
velocity element perman~ntly to zero, where the ohysical bo~ncary
crosses a line between two neighbouring surface level points!
V=-O
~I
.:
U=-O
t
+
+
)+
qu
0 0iqv
0+
+
+
u=-o
I
As é consequence in flow regions where the sidewall is not in acoordinate dir~ction some surface level points are surrounded by
I
I
I
I
only two non-zero velocity elements. In a stationary flow the
continuity 'of the flow around such surface level points requires
that the discharges through these two velocity elements are
"equal, q~
=
quo When the local bottomlevel is horizontal, thismeans that the veloeities in the two velocity elements have to be
·equal, u
=
v. The velo~ity at the considered surface levelpoints is the mean velocity of the f6ur surrounding velecity
elements. Consequently, the velocity at the surface level points
will always be oriented at an angle of 45 degrees with respect
to the grid orientation.
This means an important difference between the flow in a
I
fl~m2! river or ti~al channel and its numerical reproductien byI
I
Mi~iwaqua or another ADI simulating system. In the prototype a
velocity direction near the sidewall parallel to this wall is
expected and measured. In the numeri cal simulation the velocity
direction in the considered points does not depend on the
·
·
1
I
I
I
. 16 spec::'fiedfOI sloping bottoms other
angles
angle 1..1th the direct.ion. For a horîzontal local·
beottom level
This ill-fitting of the veloc~ty direction. near the
·beundary is passet on t~ a more extensive region by the computation.
The impcrtance of this boundary directio~ effect depends on
the relative flow depth at the boundaries. Fer a flume with a
~ectangular cross-section a large disturbance is found. In the
DHL-flume with an uneven bed the effect is smaller because of the
s·maller-depth ·~,tthe side~·J2.s112.!l'j the corlsequE:IT. 1\,. '1esser
impor~ance of the flow direction· at the walIs.
·
1
·
even less important effect can te expected.I
I
I
I
I
I
I
I
I
I
I
I
I
Tc correct this iIl-fitting velocity direction~ local
bottem slopes car te assumed. In this way the velocity directien
can be made to correspond with the sidewall directien. These
1oC.?,lbottom sl opes at-eirrtroduc ed by a change of the bottom
eleVë.tions arourid tr:e cc,ns.idered sw-fa.ce level point. The
equality qu
=
qv still applies but the flow direction is changed.This can be understood by the
in an element can be cemputed
following reasoning. The vel ccity
by the division of the local discharge
by the area of the ~ Iow s.ection •..Th e area of the
flow section in the numeri cal simulation is the width of the grid
spacing multiplied by the mean differences between surface level and
battom level in two neighbouring grid points.
The hattom levels of the
computational molecules of which
the surface level point is
just cutside the sidewall
0
I
v
=
0
Ob
A
u
=
O
+
,.
+
qu
Ob
jQV
Ob
B
c
.b ourid ar y , ar e changed togive the good velocity direction!
without violating continuity in the
+
considered corner surface level
point. Tc impose an angle a with
respect to the grid direction in
correspondence with the angle of
th~ sidewall! the bed levels have to s2tisfy
v (9)
I
tal1 a u
I
I
I
I
I
wh~~~ bA~bBlbC are the bottom levels in PDlrts
A,E
and Cre~pectiYely. Substitutin~ cy
=
qu' the values of 0A and bB hBveta s.e:tisfy
tan Cl
=
r - l(b +b )
., 2 A C
(10)
The bott.om le'/el b,,-,~. in point C ma..v not te altered because this
bottam level directly influences other velocitie~.
the '·/a1LI.eof one of the remaining bottom levels bA
t-!
I
determines the other.Very unrealistic battom levels have ta be imposed in regions
I
I
I
I
I
I
I
'
1
I
I
I
I
I
I
with small an~les between sidewall and gGid direction. Other
regions where very unrealistic values car.develop are regions
where the angle is about 45°. In the~.e reç ions the choice of one
bottom level determines a whole chain of bottom
4.2
F
l
n
w
area correctinnThe physical boundaries are translated into the numerical
configuratien by imposing the value of the v~locity in the
nearest velocity elements at zerb permanently. The exa.ct
position of the physical boundary within the grid spacing
is lost.
+
+
+
Espec~211y.wh2n the physical boundary has 2 small angle with
crientation this coarse repre~entation of the.
r=..-, r,
'"
1
'
1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
18boundary impairs the flow simulation. The relation tletween the
local discharge qv and the veloçity v in the numerical
representat ion depends on Iy on the d~pth h. In the prot o.type the
'exact placE of the physical boundary with r~spect to the
coordinates (and so to the grid) is important,.hawever, for the
area of the flow section.
To correct for this deviation in the area of the flow
section in the numerical simulation, the battom level of the
computa.tional molecule of which the surfa~e level point is just
outside the flow is to be changed in order to obtain the correct
area of the flow section of the molecule at the sidewall.
Bath the corrections are dependent on the waterlevel. This
I
I
I
I
I
I
I
I
I
'19=~.
Improved rporoduction of thp depth averaged flowApplication of the c~rre~tion schemes modifies the bottom
configuration near the sid~walls. This modified bottom
configuration is used in the computation of the depth averaged
flow by Miniwaqua. The correction on the bottom configuration
depends on the water depth at the sidewalls, hence the correction
should be determined at every time step, especially in time
dependent flow. This procedure would consume much computation
time. In this investigation the correction~ were obtained from
the water depths computed in the 'comp~tation without a correction
for the numerical representation of the sidewalls.
Computations based on the combination of both correction
schemes as weIl as computations based on each correction scheme
separately were executed. The resulting depth averaged
velocities are compared to those resulting from the uncorrected
computation and to the depth averaged values of the measured
veloeities. The improvement of the reproduction of the depth
I
averaged flow by the various correction schemes can be appreciatedI
I
I
I
in this way. Most comparisons are carried out for the
rectangular cross-section, where the effect of the use of the
correction schemes is largest, because of the serious disturbances
in the uncorrected flow in connection with the high vertical
sidewalls.
The depth averaged velocities computed for an originally
plane bottom configuration, modified by the combination of the
two correction schemes are plotted in fig. 9. (The veloeities at
all the grid points within a strip with a width of 1.5 m around each
cross-section are used to compose the plot, in order to provide a
I
better velocity distribution by the inclusion of more grid points.The use of this relatively broad strip also gives an estimate of
I
I
I
I
I
I
the scatter in the velocity distribution.)
The improvement obtained by the application of this
correction is obvious. From the large disturbances at the inner
side in the first part of the bend and at the outer side in the
second part (see fig. 5) only traces remain. This means that
af~er this correction the flow is capable to follow the flume
wall at the local widenings in the irregular numerical
representation 9f the sidewalls. The cross-section at 13.8
degrees shows a small remainder of the originally very important
I
I
I
I
'
I
'
I
I
I
I
I
I
I
I
I
I
I
I
20The correctien of the bettom levels of the grid points near the
wal I in these cross-sections is e:{tremely 1arge becau,se of the
small angle of the flow 'with respect to the grid direction.
Hence exactly at these cross-sections a deviation of the measured
value will show up easily.
It is remarkable that both corrections, flow area correction
and velocity direction correction, separately barely improve the
depth averaged velocity field (see figs. 10 and 11), whereas the
combined correction is sa effective. The ~orrect reproduction
appears to be quite sensitive to the exact correction applied.
The scatter in the values of the depth averaged velocity~
,mainly caused by the ~ocal obstructions of the flow in the
irregular numerical representation of the sidewalls, is also
diminished. The combined correction gives the best results for
this kind of disturbance tOD.
The corrections have less effect on the reproduction of the depth
averaged flow in the uneven bed configuration (see fig. 12). The
reproduction by the uncorrected computation was already quite
good, because of the low sidewalls in this configuration. The
same low sidewalls make an important improvemént difficult.
Comparison of the results of the uncorrected computation (see
fig. 6) and of the corrected computation show an improvement with
respect to both kinds of disturbances that is small but
relatively not unimportant. Only at 13.8 degrees and at
82.5 degrees considerable disturbances remain because of the
small angles with respect to the grid directions.
The influence of the secondary flow on the main flow is
appreciable in this configuration. This influence is not
accounted for in Miniwaqua, so an important deviation of the
computed depth averaged velocities in the last cros~-seciohs is
to be expected. The velocity field obtained by the computation
appears to be mainly determined by the bottom friction as can be
concluded from fig. 13. In fig. 13 the depth averaged velocities
calculated from the surface slope and alocal Chézy coefficient
are plotted. The local Chézy coefficient is assumed to be only
dependent on the water depth. These velocities that are determined
by the bottom friction and the computed velocities based on the
combined corrections are more or less s~milar except for the
I
first part of the bend, where ~he flow behaves like a potentialI
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
.
I
I
I
I
21flow,.and the region ar ound 84.5 degrees, y.Jhet-ethe flow is pushed
slightly to the imner side of the bend by the remaining
disturbance at the outer sidewall.
It may be concluded that the improvement of the reproduction
pf the depth averaged flow is impressive for the plane bed
configuration. The uncorrected reproduction of the uneven bed
flmaJ vJë\S already better- and the impt-ovement is smaller. A still
less deviant reproduction can be expected far- tidal channels, but
the effect of the corrections will probably be much smaller toD.
In all cases the combined cor-rections for velocity dir-ection and
"
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
226. Reproduction of spcondary flow
The reproduction of the secondary flow in this research is
based on the computed d~pth averaged flow. Deviations in the
.repr-oduct t on of the depth averaged flow ~'Jill,hence, have direct
consequences for the reproduction of ~he secon~ary flow. The
reproduction of the secondary flow in the plane bed configuration
was severely hampered by the disturbances of the depth averaged
flow computed without the bottom level corrections at the
sidewalls (see fig. 14). Especially the SQatter by local
obstructi6ns has a devastating effect. The use of the depth
averaged velocity field computed with the corrections applied'
increases greatly the part of the flume where reasonable values
of the secondary flow appeal'"(see fig. 15).
The reproductien of the secendary flew in the uneven bed
cbnfiguration was already satisfactory when the uncorrected
cemputation was used (see fig. 16). Ä factor- of about (1/3
between the computed and the measur-ed values, connected ~ith th~
uncer-tainty of c in equatien 6, was discussed in chapter- 2. The
reproduction using the corrected depth averaged velocity field is
slightly better (see fig. 17) because of the more correct depth
averaged field, and especially the smaller scatter. The scatter
at 82.5 degrees is caused by the large variation of the radius of
curvature of the main flow over a short distance at the end of
the flume, because of the bad reproduction of the flow there,
I
I
I
I
7. Conclusions
The reproduction of the depth averaged flow in a large
curved flume of the Delft ~ydraulic Laboratory with a fully
implicit finite difference method of the ADI-type, Miniwaqua, is
satisfactory on the whole but certain defects remain (Booij and
Pennekamp, 1983). These defects can, in sofar as they are
I
connected with the numerical representatien of the sidewalls inthe bend, considerably be reduced by means of a correction
I
I
I
I
scheme. The correction scheme combines two kinds of corrections
which have only a slight effect separately. Both corrections
modify the bottom level at the sidewalls to correct the velocity
direction in corner points of the computational grid and to
correct the area of the flow section at the sidewalls.
The reproduction of the depth averaged flow using this
correction scheme is importantly improved. For a rectangular
cross-secti6n, in which configuratien the uncorrected computation
showed quite large disturbances caused by the representation of
I
the sidewalls, only traces of the original disturbances remainwhen the correction scheme is used. The effect of the correction
I
I
scheme is less in the case of a flume with the geometrical
proportions of a river, for which flow the disturbances obtained
by the uncorrected computation are small, however, becausé of the
smaller sidewalls. In tidal channels smaller disturbances and
less improvement is to be expected. Further improvement requires
a more natural choice of the grid configuration of the sidewalls
and the bed. The use of a curvilinear grid is not vet possible
I
in Miniwaqua, but it is foreseen in the near future.The reproduction of the secondary flow, which is based on
I
I
.
I
I
I
I
the computed depth avera~ed flow, is consequently also much
improved when the correction scheme for the computation of the
depth averaged flow is used. The part of the flume, where
reasenabie values of the secondary flow are computed, increases
""
I
I
I
I
I
I
I
I
I
I
I
7.I
8.I
I
9.I
I
I
I
I
24 F:eferences 1. Booij, R., 1983, di~cussion to: Vreugdenhil~ C.S. andWijbenga, J.H.A.! Computation of +low patterns in
2.
rivers, J. Hydr. Div. ASCE, to be pu~lished.
800ij, R. and Kalkwijk, J.P.Th., 1982, Secondary flow in
estuaries due to the curvature of the main flow and to
the rotation of the earth and its development, Delft
Univ. of Techn. ,Dept. of Civil Engrg., Lab. of Fluid
Mech., report 9-82.
800ij, R. and Pennekamp, Joh.G.S., 1983, Simulation of
main flow and secondary fl00 in a curved open channel~
6.
Delft Univ. of Techn., Dept. of Civil Engrg.! Lab. of
Fluid Mech., report 10-83.
Pennekamp, Joh.G.S. and Booij, R., 1983, Simulation of flow
in rivers and tidal channels with an implicit finite
difference method of the ADI-type, Delft Univ.·of
Techn., Dept. of Civil Ençrg., Lab. of Fluid Mech.,
report no. 3-83.
Stelling, G.S., 1983, Thesis, Delft Univ. of Techn •.
Vreugdenhil, C.B. and Wijbenga, J.H.A., L982, Computation
of flm'llpatterns in rivers, ASCE-proc., 108, no.
HY 11.
Vriend, H.J. de, 1981, Steady flow in shallow channel bends,
Thesis, Delft Univ. of Techn.; Comm. on Hydraulics,
Delft Univ. of Techn. i report no. 81-3.
Vriend, H.J. de and Koch, F.G., 1977, Flow of water in a
curved open channel with a fixed plane bed, TOW, Report
on experimental and theoretical investigations, R657-V,
M 1415 part I.
Vriend, H.J. de and Koch, F.G., 1978, Flow of water in a
curved open channel with a fixed uneven bed, TOW,
Report on experimental and theoretical investigations,
I
I
NotationI
I
cI
.
C 9 hI
i ,jI
I
I
sI
t
I
I
u Un un(z) Us us(z) vI
I
y zI
ClI
I
e:I
KpI
T ,T wx wy 4> wI
bottom level bottom level ~n grid point P width of the flumecoefficient in the secondary flow intensity
Chézy coefficient
acceleration due to gravity
depth of flm'l
subscripts indicating a com~utational molecule
secondary flow intensity
local flow coordinate perpendicular to the direction
~f the depth averaged flow
discharges through velocity elements
radius of curvature of the main flow
ràdius of curvature of the channel axis
local flow coordinate in the direction of the depth
averaged flow
time
depth averaged velocity in x-direction
depth averaged velocity in n-direct~on
secondary flow velocity at level z
depth averaged velocity
main flow velocity at level z
depth averaged velocity in y-direction
horizontal coordinate
horizontal coordinate
vertical coordinate
local angie bet ween flow direction and grid direction
distance bet ween grid points
numeri cal time increment
diffusion coefficient
water level with respect to a horizontal reference level
Von Karman's constant
mass density
components of the surface shear stress
1ati t_ude
angular rotation of the earth
1
I
-1
1
I
z
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
h
!
y
I
I
.
-.>: \ \
I
I
\ \ \ \ \ 55.00I
I
I
I
I
I
I
I
I
EI
uc R:; SO.Om 0I
0>.s
I
QJ 00 -0 cu .tlf
- 39
I
-41--r-...:
~ ~f-F':
I-I
-40 -42I
11
t
j , -43 -23.0m-U.5m 0 0 13.8027.50 4:13055.00 68.80 82.50---3>~longitudinal distance along channel axis
-11.5m
I
~3mr ....
12x0.45m 123 4 5 6 7 8 9 10 11 12 -23.0 m CROSS - SECTION1
1
I
·
1
1
I
I
I
I
I
1
1
I
I
I
I
I
1
1
1
I
CONTOUR KEY 1 - 0.2000 2 - 0.2500 3 - 0.3000 4 -0.3500 5 -0.4000 6 -0.4500 -7 - 0.5000 8 -0.52001
I
V
-l
1/
:11\
.>
i-.1/\
1\'\
r--
r--
I-! 1A5m trom inner wall channel axis 1.55m trom outer wall -23.0 m -11.sm 0° 13.8° 27.5° 41.3° 55.0° 68.8° 82.5°---'»longitudinal distance along channel axis
.-. - 25
.,.
0 )( E - 30 ~ c 0 - 35....
0 > ClI - 40 ~ "0 -11.5m ClI - 45 .01
- 50 I.(') -55 0 1 2 3 4 5 6 ~distonce trom outer wall (m)
-23.0 m
-I
1
I
I
1
10
20
30
40
50
60
70
80
90 100 110 120
111I1I111J 111 IjI I "11 11IjII"111 I 1II111I1I II1I I I II'I I I1I I I I1' I I I111111' I I , 111 11II , 11111I ,1' I,,11 I"I'i1111"1 1"1 111111I-=
i180
170
=
i1
70
1
1
I
·
.
130
=
I
1
1
1
90
=
90
1
I
I
1
~
11111111111111 I111 , 111111111 , 111tIllIl' 1111 1,1, 1,111 11" 1,, ,, 1, 11, 11111111"" , 111 , 1" 11,",,11" 1, 11,, 1,, 11 , 11, ,!11' 1, ,, , 1I,,,1, , , 11I1
10
20
30
40
50
60
70
80
90 100 110 120
=
--
80
=
-
-70
=
-=
--
60
--=
-=
50
--
--
40
=
-30
-
=
-20
--
--
---=
-10
-
-1
-80
=
1
1
1
60
=
50
=
1
1
I
INITIRL
GEOMETRY
(VERT.)
NMRX=
193
(HOR.)
MMRX=
120
Fig. 4. Computational grid with ~
=
0.40
m.I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
++++ 02.50EG++ ++ ...++ .,.++ +.
,
...+ ++ ++++ +++ o o +....'" o o o o o o o o o..
+ o 68.8 OEG + + ... ... ... ..t+ .. ·...+ ... ++ + o o ++ + +.,.+ ... .t-+... 1-' f-+ + ... o .. + + 0 o o o o o o o o + + + ++ 55.0 OEG .,.+ -I-+ + + -:- ...+ o +.~'" '''++++.} o '" .,; Ol .....
.
,. o +..
... + o....
++ o + ...ot- o El o"
+ .. 0 + + +..
..
..
o Cl è en + ., +... + + +., +..
..
ijI.3 OEG ++ +... +++ "+ + o o o o o o o El o..
o..
+ Cl Cl..
....
..
+ 27.5 DEy ... + .. + + ... + + +...
......
+..
+ + Cl '" '" .... o o o o o e o..
o o o o o ... Cl o Ol> Ol> + + + 13.6 OEG ++++ ++t+ +++.,. +.t- ..+ ++++..
Cl <> Xc -V> o..
o + o o o o..
o O' o o o o + 0.0 OEG + f> o + + + + .. + o +0..
o + ., o + + o o o o Cl Zo D,O. erf; e +0 +..
+ + o + + + ... + + o o o o o e o <> o o'"
r o o '" " o..
+0 -I- -I-o..
+ ... + + +..
+ o o o o o o o o CJ ,,; <> '" N e +HEASUAEO VALUE tOE VRIlNO ANO ~OCH)
COHPUTEO VRLUE tMINIWROURJ
o
o
°or.-o.,.o--....,.or.-::,j-::O---lr.O-::O:---:1I-:,~:-:O:----:2T.-:O-:O---::2r.::5::0---=,)r.O:-:O:---3::r.-:~-:·O:---~T.-:(I-:O---~r.-5-"O---5r.-O-O<-,O---sr.---G'. (10 t-IIDTH lMI Fig. 5 Dcpth averaged velocity distributions in several cross-sections (plane bed configuration). Cl Il' è ++ 0.0 FOR -23.0 M o If) <, :>: 0-""
oW
..J a: u If) o..
o ~ 0.0 FOA °02.50EG + o 0.0 rOR 68.8 OEG o 0.0 FOR 55.0 OEG o 0.0 FOR ijl.3 OEG o + 0.0 FOA 27.5 OEG +..
+ 0.0 FOA 13.80EG o..
0.0 FOR 0.0 Of.G o 0.0 FOA -11.0 HI
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
+ Cl Cl C) en +... o ...0 o .....
+ o + +..
+ ....
o +..
....
+ o o + o..
() +....,. o +..
.
,
.....·H· o o o +..
o o o .....
..
+..
..
+ o '" c.i + + .. Ó 0.0 rOR -23.0 M C) ;:> ". Cl)..
o o .,......... o o + o I./J <, r 0-.,
c.iw ..J er U lf) + + o ~ 0.0 rOR <>62.5 OEG o <> '" e-o o Cl '" '".,
I./J-
xc cr~ '" Win :r ::;) ..J LLo CJ w"; r... Cl Zo DO ..J' cr '"...
..
e o ... o..
...... + o + o + 0.0 FOA 66.6 OEG Cl o <> '" o'"
en ·t++.. 68.6 OEG..
+ ·f + 0.0 FOR 55.0 DEG o 0.0 FOR ijl.3 DEG + o..
..
o o o ·f o + + + o +..
..
..
o 0.0 FOA 27.5 DEG..
+..
0.. 0.0 FOR IJ.6 DEG o + 0.0 rOR 0.0 OEG o ... 0'.0 FOA -11.0 H o + ·f o o +~ '+ o '"èor.-ûo---o-r.'-jO---'I.-û~0----~lT,-5-0----~2r.~ÛO~----~2r,~~·0~--~3~.-(-10----~JT,~5~0----~~r,~ü~0---~~r"~jO~----r.,,.r(-'0---5'.-5-·0---,6,00
r-IIDTf1 (MI
I
·
1
·f o ++ + o +.. 0 +..
0+ cf 0 + + .. 0++ ..0 + +o..
o ++..
..
+ o + o +..
o..
o + o +'f o + o + +0 + o o +~ ~ MEASURED VALUE IOEVRIlND AND KOCHI
+ COMPUTED VRLUE IHINIWAQUAJ
Fig. 6 Depth averaged velocity distributions in several cross-sections
(uneven bed configuration).
++ ijl.J OEG o ..++ + 27.50EG
..
+ o..
.
,
o..
+...
o + o..
.."..t 13.60EG ++++++++++++ o 0 ..0....... + ., +..
..
..·to·~ 0+ 0+ 0.0 DEG é + o + o..
I) + -11.0H 0+ +0 o .. o. + + o + -23.0 M 0" +0 o .. o .. + o ...1
I
'
1
1
I
I
I
I
.
I
I
I
I
I
I
I
.
1
I
I
1
I
I
/ / / I
-:I / /
I
I
1 /1/
//111/
/ I I I I /
/ / / I 1 I /
!
I I 1 / / /
·I///I//!
I---~
. . 0 .I
I
.
:
.
..
f-j--;-/- ;--/--
!--I--~--~--I
////I/If
:
1//1111.
1
·
·
//1///1
I
.
·
//111/.
:
/ I
I / ..
1
/
I
I
~
I
·
L
_
I
I
I
I
I
I
I
I
I
·
I
I
I
I
I
I
I
I
I
I
I
I
I
02." O[G o ..;0+ o ++o++ ..H·++ o .~+.~+o ++'.. +++ ++-H ++++ o o + o o o o..
..
•
..
+ + o '" o o 0.0 fOR -23.0 H 60.60EG + IJ ++++++o'''0 o o o (/) <, :>:" o~.
,
oW
.J cr U sn + oN>-0::
u o .J o._>W o .+ • 55.0 OEG C> o cO '" o o +..;.+ + ·t ++ o .;.o +Y·t+ 0"-t ...+ ..;. + o + + o ~ O.0 FOA 062.5 DEG o + + .p... ... ..+0++.H-.;o+ ... +0... ...·iOt '1 ... + +o+ + ... 0+... "'·t -t.... o o + 660.0.6 FOOAEG C> o o '" ~1.3 DEG C> o '" CD + IJ H.o 'H .... H-o.
..
o +o..
.. '1 o .. 0+ o..
0.0 FOA 55.0 DEG C> o en ......
..
+ o ++...'''++++..
....
'"
... o 0 0 .. '4+ '0•
+ + o o + 0.0 FOA ~ I.3 OEG o + + + 0.0 FOR 27.5 OEG 27.50EG o o '" .... .. + o o+ .. + .. o o .0 + + + o o..
0.0FOR 13.6 DEG + + o..
o +0 + + .. o + 0.0 FOR 0.0 OEG..
+ IJ .. o+ o o o o o tv~ MEA5UAED VALUE 10EVRllNO ANO ~OCH)
+ •COHPUTEO VALUE IHINIWAOUAI o· 0.0 FOA -11.0 H Cl Cl
°Or:o-O---OT:-~-0---1~:-O~O----~I'r.5~O----~2r:O~O~--~2T:~5~O----~3~:~O~O-·----~j'~.5~O~--~4rl.~OO~----~T:~5~O---5~:~O~O~--~5r:~50~--~6j,OO NIOTI-I (MI
I
I
13.8 DEG ++ ++++++++++++++++ o + + + ... .,. + ...0" ..-t-t ri'" ... o + IJ o o o + o..
0.0 aEG + cl> o + + o + + o o C) Zo DO .J' cr '",.. IJ + + + + 0+ -11.o!o0 H IJ o .. + o o o o'"
.;..,
-23.0t1 o' 0..
o o o ". "..
+ + +..
..
o o o <> Cl .,;Fig. 9 Depth averaged veiocity distributions in several cross-sections
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
02,50EG ++++ ++++ ++- .. o}-++ +++.+ <> '" à El 0.0 rOR -11.0 M ++++ ++++ ++++ ++++ o ." à +'"
<, :.: 0 -'"oW ...J 0: U'"
o "'>-ö~ u o ...J OW' -> à IJ o o o..
o "! 0,0 FOR '"62,5 OEG o o 0 o o o..
o•
0,0 FOR 68,60EG 0,0 FOR 55,0 OEG I) + 0,0 FOR 111.3 OEG I) +• •
0,0 rnn 21,5 DEG I) + 13.6 DEG0.0 FOR o I)•
0.0 fOF! 0.0 OEG o 60,80EG + ++i-+ o + +++-: ... o 0.
..
I) <J 0.0 FOA <J -23.0 MoOr.-00---0-'-. '-jO---'L-O-O---l"T,-~-:' O---ZT.-:0-=-0---::2r.'::-jO::---)::>,-:O:-:O:---J=",-:~-:O---:~T,-=-O-::O::'S::-O----:~r,--::S(:-r.'0::---," r,'-:;0,---.,,6.(10
I'lIDTH (MI
,
I
I
.. ++ + o ++-}o+ • + of. .. 0•
f) o o o + o•
.. +..
..
55,0OEG o o à Ol Cl + MEASUREO VALUE IDE VRllNO ANO ~OCHICOM"UT[Q VALUE II11NIHAQUAI
++ +
..
.
,
"
+..
..
+. +..
+ 1I1.30EG + ++ + H+ +++ o o o '" ". en + IJ o o + e e I) 21.5 OEG.•
+ ,... + +..
..
.
... o + ...+ + ... o + .....
..
... + o e o o o ... o El El I) I)..
o'"
ui co ~. 13.8 OEG ++++ ... + +:++·t o + o + + ++++++++++++++++ o e..
.,
o +..
o o I) o o e 0.0 OEG + • o..
+ +..
..
o +0 + El + + + I) + + o El o o I) -11.0 H 0+ l!l +0 .' o +•
..
+ + + + + I) + El El e I) El c.> c.> c.> '" +0 + • +•
+•
+ • + + o o e I) o El o o o CJ .; c.>'"
c.> '" .;Fig. 10 Depth averaged velocity distributions in several cross-sections
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
<> '" '" en ++++ 0++2.5+... DEG++++ v·..·H. + + + + + + C) 11' Ó .,..
,
0.0 rOR -23.0 M <> o è en <> '" '" en <> '" In...
<> o '"...
<> o (t) '" '-" Zo co ..J' a~'"
'"
<> '" -,.++-" + Il o o .,. .o, c.i o o o () o o..
Cf) <, ::E 0-.,
oW ..J Cl: U Cf) +..
., o o..
o + .. 0 + .,. +"+" .:-60.0 DEG +++- ... ,. ++ + ot..,. ...+ + o + o~ 0.0 FOR °02.50EG + ...,. ·t ++ .:.... +..
..
o o () o + o +. 0.0FOR 68.6DEG 0.0rOR 55.0DEG + + ., 0.0FOR ~1.3 DEG e + 0.0 FOA 27.50EG + + + 13.8 OEG0.0 rOA o o o"
+ 0.0 rOR 0.0 OEG o o"
0:0 rOA -11.0 M '"'"
èor.-o-o----~Or.-~-O----~lr,t-,O~----I~.~~-'O~--~2T.~O~O----~2r.~5~O----~Jr.O~O~----J~.~~~'O~----~~.~U~O---~T.~5~O---5r.-OO---5r.~-'(-,---6.UO I'nOTfi (MI
I
I
+ + o + + 55.0DEG +..+ +++-t .... +..-t- ++ +++ ++ +++ +-t+ 000 0 0 0 o ++ + .... ~++..
...
++ o ++o+ + + + o ++ + + " +' + e» + + + ~1.3 DEG +1-2- +.... + ++,
+
,
"-tot ".
'"
"
o o o o..
.. 0+ + + o o o o o .,
.~. + .y + + .. + 27.5 DEC ....
.,
..
+ + +..
+ +"
o o o o o o o o o..
o + 0 t + 13.00EG ....+ + ...t ++ .... ·t .,....jo·t.. ... + + + o o 0 o o + o o o o o"
+..
..
o.
,
+ 0.0 OE(, + e o o.
,
+ + o o + ..+ o +0 + o o o + + + o + ,,+ -11.0t!> M o"
..
+ " + + + o + +"
"
o o"
+ + .."
.,Il + + + + o..
o Il o o o I) o ~MERSURED VRLUE (DEVRllND RND ~OCHI + COMPUTEO VRLUE (MINIHROURJFig. 11 Dcpth averaged velocity distributions in several cross-sections
I
I
I
I
I
I
I
I
·
I
I
I
I
I
I
I
I
I
o ., o en <> <> "..,
<> <> <0 CD IJ) ... xo cr~ ". UJ'" :>: ::> ...J u..0 <.> UJC;; r ... ,_ Cl Zo 0<> ...J' cr~ <> <> C'"
o <> ". ... o +"
.. O'"
+ ++ + 02.5 DEC o o o..
o....
o o o +e, + o '" è o 0.0 FOR -23.0 M + + .. + IJ.. ..
o..
+.,
o..
a> .. ..+ 0.......
.
0+ + o..
++ o +....'"
<, :>: 0-'" oi.JJ ..J er U'"
o "' >-à~ U o ..J oUJ .~> è o "! 0.0 FOR 082.5 DEC 0.0 FOR 66.8 DEG + 0.0 FOR 55.0 DEG +<0 -#0 0.0 FOR 4 I. 3 DEG +..
o + +..
+ + .. IJ..
IJ..
o o..
60.80EG +o + ...0 +<+ tO + o + ol'..
o 0.0 FOR 27.5 OEG +..
..
0 .. 0.0 FOR 13.8 OEG o..
0.0 FOR 0.0 DEG o + 0.0 FOR -11.0 M + o +0 +....
0+.+ + +0 + + o + +I
o <>°Or.-o-o---or.,-jo---'1-.o-o---I'.-~-O---~T.-O-0----~2r.~~~O----~3r.O~O~----J~.~~~·O~----~'.~O~O----~~T.~5~O---=5r.~O~O---=Sr.S~O~----S~.UO
NIDTfi lMI
I
+ o o o +.. e ++++.,. + + 0 ..+ 55.0 OEG +.. o ++....
+.
.
o + + 41.3 OEG o ++> ++ ++. o o 27.5 DEG"
..
~...
o ...o + e + + o + + ++0 + .. +ct 13.6 OEG ++++++++++4 . .-o 0 '~+'''+ .. e " + .,0++ + 0+ 0.0 DEG 0+ l!l o + + o + o ... -11.0 M•
" + '0 + e + + o + -23.0 M•
+0 0+ o + " +..
o + (!) +HERSURED VALUE IOE VR IlNO ANO ,(OCHI
~ COMPUTED VRLUE IMINIHAOUAI
o + e + o + o + o +
"
+ e + o +Fig. 12 Depth averaged velocity distributions inseveral cross-sections
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
,
I
I
I
,
I
o '" cO en o o o en o o o o N r-. o '" CO IC ~ Zo 00 _J' cr~ o o d '" o o ...; 0000 02.5 DEG Ili5Ó+ o o 00 ot.,
..
o '" o o 0.0 FOA -23.0 M o IJ) <, :r 0-'"oW
_J er u IJ) o 'J >-o~ U o _J oW -> o o '! 0.0 FOA 062.5 DEG ..+ o o 0.0 FOA 66.6DEG o..
0.0 FOA 55.0 OEG 0.0 FOA ~1.3 OEG o o..
o .. "0..
0.0 FOA 27.5 OEG o +..
+ 0 o o..
o ..+ o .. o + 0.0 FOA 13.6 OEG o..
0.0 FOA 0.0 OEG o..
0.0 FOA -11.0 M + + .. + o o o + o oI
o <>°or.-Oo---oT.-~-O---,I.-O-O---,lr.~-·O---2r.-OO----~2T.-5~0---3~.~O~O----~Jr.~~·O~--~~r.~O~O---~T.~5~O----~5r.O~O~--~5r.~50~--~6.00
tolIOTH (Hl
I
_,., H> 41.3 DEG ~~...
.
o + + 0' 00 + o o 0 o..
o o o o o +..
..
o+ 27.5 OEG..
..
o + + o + o +..
..
o..
o .. o o..
Cl> t o + 0000 0000 ++++ 00 13.60EG ++~G>t!lÓ· ..+++ .... 0°0, -t+ ...+ 0°0 o o t o..
o 0000 o ....+ + + o..
+..+ + o o o 0 0000 ++ +.+ + +0 .. 0 + 0000 + + 00 + 00 0.0 OEG ~ ~ Ijl•
o•
-11.0 M tJ '" + o + o -23.0 M•
..
•
o•
•
+ ~ L value determined by the bottom friction
+ computed value (Miniwaqua)
o
Fig. 13 Depth averagcd velocity distributions in scveral cross-scctions
I
I
I
..
..
..
..
I
..
..
....
..
..
..
.. ..
..
..
..
..
..
..
..
..
..
..
..
I
..
...
s
oe: co•
..
..
++....
I
..
..
..
..
..
..
..
...
•
+ ++ + +.+. + ++ +..
....
..
..
..
+ + + +I
..
•
..
o o..
+++'..
I
..
..
55.00..
... ++ +++ ... + ....
..
+ +..
..
..
..
+ +..
+..
+..
+ +I
<>..
+ + <>....
..
..
+I
<>o oe:'"
..
+....
..
+..
....
..
..
+ + ..+..
..
"
..
"
..
....
..
+ + + -, ..+..
..
..
..
..
..
..
++ VI....
xo a:~ wf;'; ::I: ::l ...J "-0o w"; :1:,.. l-C) Zo 00 ...J' a:::: w I-....0 Vlo .;; ++...
..
..
..
..
..
..
~..
..
+ ... + .....
+ +..
..
+ + + +..
..
13.So ...++++++++++ +++ +..
..
+ ++ + + + + + +..
..
..
..
..
..
..
..
+ .. +I
..
..
.. + ..I
s
ó...
I
o o ,,; N..
+..
..
+..
..
+..
..
+ + + +a
..;-I
-11.5 m + .. + + + + + + + + + +..
..
-23.0 mI
o..
..
..
..
..
..
..
..
C>óOi~.O-O---OT:-SO----~I:rO-O---IT:-SO---2~:-O-0---2r:s-O---3T:-o-o----~3:rS-O---qT:-OO---q~~-S-O---Sr~O-O---ST~-S-O----·~6:00
HIDTH (Hl
+