You are on page 1of 38

Ill

SEPARATION OF BED AND SIDE COEFFICIENT OF

ROUGHNESS AGAINST QPTIflAL CONDITION OF FLOU


3.1 Introduction
In open channel flow, when water is allowed to flow
across a channel, resistance is encountered by it at the
boundary. The resistance is generally counteracted upon
and balanced by the component of gravity forces acting on
the body of the water, resolved in the direction of motion.
In the beginning, when water enters the channel, the velocity
and hence resistance are gradually increasing until a
balance between resistance at the boundary and gravity
forces, acting on the body of irater is reached and flow
becomes uniform. However, if the forces causing flow are
increased continually, a point is reached when work done
by the forces can not be compensated for by the viscous
forces alone and flow, assumes, a more different and
complicated eddy structure in which a mass exchange resis-
tance is produced sufficient to bring the balance against
the forces causing the flow. Since the process of lateral
change in the mass distribution of flow, across a unit
volume which exerts stress over a unit area along the
permimeter, near the wall,is governed by random eddies and
secondary currents, the time average local resistances
would diffier from instantaneous resistance space average
by a random quantity. This renders the local resistance
phenomena stochastic with respect to space and time. An
analytical expression for the instantaneous boundary-
shear-stress has already been provided in Chapter II.

- 21 -
However, the time average mass-distribution of flow could
be conceived as deterministically separated into regions
corresponding to defined portion of wetted perimeter
where the masses are separated with respect to stresses
that they exert effectively at the boundary. The manner
in which this separation occurs in the flow, beyond the
perimeter is difficult to define, except In terms of a long
itudinal surface orthogonal to isovels, across which net
momentum transfer may be negligible compared to transfer
of longitudinal linear momentum. In the present chapter,
a stochastic model has been developed in which the effec-
tive cross-section average resistance is conceived as
weighted sum of time- and local space average resistances
for bed and 3ides such that the flow distributes momentum
between the sides and bed regions of flow in a way that
the resulting resistance would be minimum amongst various
modes of distribution. The deterministic solutions provide
functional relationship between effective resistance or
roughness and material roughness, besides giving component
resistances for bed and sides without resorting to any
side wall correction procedures or reauiring physical
separation of flow. The component roughnesses obtained
are compared against their time average counter parts,
observed independently by various investigators, and
determined by the author using appropriate estimation
procedures, from basic flow data of experimental channels
collected by the author through various sources.
-22-
3.2 Methods of separation

3.2.1 Review

Methods of separating the side and bed coefficient of


roughness have been evolved by various authors following
concept of division of flow between the walls. Such divi-
sion of flow has not been well defined except at the
boundary where there is either physical difference in the
roughness material or which provides different dimensions
of a geometrical cross-section such as sides and bed of
trapezoidal or rectangular sections with or without uniform
boundary roughness. Most of the formulae developed after
such division of flow belong to the class of general
equation

n = ( V, p,.n!h/( S p.) (3.1)


i=l 1 x i=l x

k
where ^ PJ = p, the total wetted perimeter, 'n' is the
i=l x
equivalent Manning's n for the whole section, 5 is the
index of n, and p., n. denote wetted perimeter and Manning's
n for the 'i' th region of flow, if the whole wetted peri-
meter is divided into k regions. The methods of Los
Angels district (1968), Cole bat ch (19 U ), Einstein (1950)
or Horton, Pavlovosky (1931) Lotter (1933) etc. all belong
to this general class with different values of S.

Separation of 3ide and bed roughness, following

- 23 -
Einstein concept of division of flow, when conceived in
terms of shear-3tress T ^ and T for the bed and sides of
a geometrically defined channel cross-section is obtained
by using the formula

T
P = PbTb + PSTs (3
'2}

or given in a generalised form

TP = V- p.T, (3.3)
i=l 1 x

where
TL = average shear-stress at the bed
T = average shear-stress at the sides
T = average shear-stress for the whole cross-section
Tj = average shear-stress for ' i'-th portion of
the wetted perimeter, p*.

The relation (3.3) ha3 physical significance in that it


characterises a net transfer of longitudinal linear momentum
as sum product of such transfers taking place along the bed
and side regions of flow. In an open channel flow, momentum
must be transferred both downwards towards the floor and
laterally towards the wall. Cruff arrived at the above
relation envisaging that the average bed and wall resistance
do not assume values depending only upon the boundary re-
sistance but there is always a process of lateral transfer

- 24 -
of linear momentum from central region of flow to walls
and from walls to the central regions and as such must
be dependent upon the aspecjj ratio.(B/d) was an important
parameter in the functional relation which he obtained
empirically for T\. Komora (1973) arrived at the relation
(3.2)by using the relation

T b = kxu2 (3.4)

where u is the velocity near the wall, measured at a constant


distance and k-, is a proportionality constant. He obtained
an empirical function for his experimental data, of the
form

(Tb/*yS) = f(B/y)

where y is depth of flow, S the hydraulic slope and 3 the


bed width of the channel.

Taylor (1961) used Darcy-Weishbach equation in terms


of friction factors to define effective resistance parameter
for the composite section of a channel. To evaluate fric-
tion factor for rough bed, he utilised Johnson method of
side wall correction of roughness.

3.3 Objective of the investigations

3.3.1 Evaluation of methods

It has been recognised long back by Shields (1936) and

- 25 -
shown later by Lane (1953/ that even in case of roughness
material ttelri& uniform alori^ the entire boundary, the
local distribution of resistance on the bed and sides of
a channel would be different than the average foo the whole
cross-section. The difference is due to the geometrical
configuration of the wetted perimeter. No satisfactory
method is available to evaluate local resistances except
through measurement of shear-stress distribution. The
procedure of sid^ wall correction due to Johnson has been
examined by Taylor ana he found that it hau no explicit
rational basis and though the method gave reliable results
in some situations, its application to widely different
situations coulu only be undertaken with caution. The
valiuity of the aide wall correction procedure was recently
examined by Rang a iiaju and iahaaur (197U; adopting directly
measured values of bed shear-stress as stanaard for compari-
son. They found wiue variation between shaar-stresa values
obtained by the conventional side wall correction nroceaure
and those obtained by airect measurements.

3.3.2 Objectives
The present investigation was, therefore, undertaken
to (i) provide a physical basis and general niouei for
separate local resistances for uniformly the
sa?re material at the boundary,
(ii) determine the effect of ahaoe of cross-section

- 26 -
on the roughness coefficient or effective shear-
stress with uniform boundary material,
(iii) determine the component resistances against
relevant physical considerations,
(iv) obtain structural relationship between unknown
parameters of deterministic general model and
comoonent3 of roughness.

3.4 General model

ivhen the effective resistance parameter n , occurring


in Manning's equation of flow for channel with boundary
formed of uniformly same material, or T expressed as HR3,
the cross-section time average shear-stress is considered
to account for the effective resistance to flow, a-t a chan-
nel boundary, the parameter is not separable for material
and shaDe except in case of very wide channel where n or
e
T. is approximately free from shape effect. In case of a
very wide channel the effective resistance parameter n
would be cl&se to the material roughness. For channels
of moderate aspect ratio i.e. (3/d) not large enough, the
experimentally determined resistance parameter includes with
it, besides the effect of material,the following resistance
effects and random errors on account of
(i) effect of side walls or shape of the cross-
section,
(ii) internal friction of the fluid,

- 27 -
(iii) time and space average element of turbu-
lence and secondary flows entering into
the structures of turbulent boundary
shear-stress,
(iv) air resistance, and
(v) random errors of observations, that would
depend upon magnitude of the observable
parameters and method of sampling the
basic characteristics.

It would be seen that in a rigid boundary channel, the


effect of side walls would render a spatial distribution
of resistance. The internal friction of fluid, normally
taken as dynamic viscosity would be fairly constant (assum-
ing a normal variation in temperature). The time and
space average elements of turbulent eddies and secondary
flow would only be given by certain probability distribution
not yet fully explored by various research investigators
working in this field. The effect of air resistance it-
self would be stochastic in nature, the process would be
more complicated wherein the effective air resistance at
the interface is to be evaluated for moving boundaries of
flowing water surface and accounted for in moderation of
the effective shear-stress at the boundary and for the
whole cross-section. Since the air resistance is uniformly
present along the water surface in all channel flow, in
practical evaluation its effect could only be taken to be

_ 28 -
o
: i o.'

a.

UJ
- Q
en
/ UJn-
r( _
UJ
— u,
' -V CO U.
n. ^_^
.a* 5"
''<'-' O

<^
is

19 -
uniform with mean random variation and negligible relative
variance (relative to hydrostatic pressure). Random errors
of observations, as is known from theory, is dependent
upon measurement techniques and sampling schemes used in
taking the measurements in space and time. The resultant
random error in any hydraulic parameter obtained by certain
mathematical expressions would be proportional to the true
values of various parameters entering into the expression,
if any linear equivalent is considered.

Thus, the formulation of a general stochastic model


for observed time average resistance parameter T(» ,p) or
nU,p) for a unit area of the wetted perimeter vide Fig.
3.1 would be a formidable task, but an approximate stochas-
tic model may be possible to ba formulated as

T(„ ,p) = T (1 + E + £j
(3.5)
T
= T o <X + e
sh>

where p stands for wetted perimeter, a function of


p(y»z)> £ h e ^ e c t of
shape over wall shear stress T ,
C stands for time average parameter of characteristic
Q = »(x,y,z) for given x = x , which generated the proba-
bility distribution of second order random term arising
on account of turbulent eddies and secondary currents
corresponding to the fluctuations in the shear-stress
process at point in the flow, close to wall, along perimeter.

- 30 -
The fluctuations proauce an apparent shearing stress equal
to the cross transfer of momentum in unit time and unit area.

£ stands for random error of observation including


sampling error on account of measurement technique or
method used for sampling. £ may be assumed to be randomly-
distributed over space and time with zero mean. Both £ , .
sh
and tr may vary in space.

3.5 Bed and wall shear stresses

3.5.1 Stochastic model

In a turbulent channel flow, at a cross-section, if


resultant flow is taken in x direction, y along the verti-
cal and z along the lateral direction, with a conveniently
chosen origin at the wetted perimeter of the cross section
viae Fig. 3.1, the finite time average boundary shear stress
close to the wall (boundary; would vary due to shape
effect of the boundary section besides due to average
turbulent eddies and secondary currents for flow regions.
and T
Averaging x(£,p)/along the bed perimeter one may find

where rfe = T Q cb. Similarly, averaging the t(C,p) along


the sides of the channel, one may find

T
< = s { 1 + e s> (3.7)

- 31 -
where C b and £ / averages ox £.- along the bed and sides
of the channel, T = T C ; c, and c are averages for
(1 + £ , ) along bed and sides of a channel respectively.

Identification of the deterministic parts t b and T


is possible through suitable approach of the partitioning
of the flow or resistances for bed and sides or through
experimental measures.

3,5.2 Stochastic Models: Deterministic parts

Inserting the values of T b and T_ in Einstein's


relation (3.2) the observed average shear stress for the
cross section is obtained as

T
'P = T b p b + T
s ps (3
'a)

= xb(i + eb)pb + T S { I + e s ) p s

or T'p = x ^ + T s p 3 + Xtf#b + Tspses

if tep = Tbpb + TSPS, and TepeT= t b p b e b + T s p 3 e 3

where x denotes average effective resistance parameter


of flow , then x' = Ta(l+£);£ denotes total relative
' 9 X X
e r r o r i n x'. And x b P b + T3pg = f 0 ( c b p b + c g p s ) (3.9)

Let cbpb + csps = c p (3.10)

where c is the average shape effect for the channel cross

- 32 -
section. The relation (3.9), using (3.10) reduces to

T
oCp = T
bPb + T
sP S (3
'11]

It may be seen that

(3a2)
%-v
where c is the shape factor for the cross section. If
T K and T are divided by ' C , the resulting shear stress
would be average material resistance for the bed and sides.
Denoting these values as iv and T respectively, we get

= T
V bo p b + Ws (3 13}
'

The relation (3.13) indicates that the wall shear stress


assumes values different than a constant value along the
bed and sides of a channel due to relative or partial
shape effect even though there is uniformly the same mate-
rial in the bed and sides of the channel. The relation
(3.13) is analogous to the Einstein's equation for resist-
ance except that the total effective resistance tp is
i
replaced by the total material resistance T p and so also
the component resistance parameters.
The equivalent of (3.13) for resistance parameter n ,
the Manning's coefficient of roughness for material alone,
free from shape effect, may be obtained along with the
corresponding general model for Manning's coefficients of

33 -
roughness f n ? , as

n 3 / 2 (Q,V) = n 3 / 2 (1 + tr + e ) 1
(3.14)
3/2 3/2
and n = nQ (1 + e s h )

The stochastic equivalent of the Horton's equation is


obtained as

»'3/2P - < ? ' \ • <?<\ (3.15)

n b / 2 (1 + e b ) p b + aj/2 (1 + e s ) | |

- 4-e/z <i• e T >f


-

2
with n 3/ p = n b / 2 p b + n3/ 2 p a

and n3/2e = n3/2pK + n 3 / 2 P e (3.16)


so ^s

3.6 Evaluation of deterministic part : Effective


Components of resistance

To solve for individual components viz. n,,n , or


T
b ' T s mathematically for a given depth of uniform flow
in a rigid channel, one must have one more relation
connecting the parameters in either of the relations
(3.13) and (3.16). Such a relation may be obtained from
the physical consideration of the flow, a constraint of

- 34 -
the resistance balance that the flow tries to adjust local
resistance offered to it at the boundary, within the
vicinity of the viscous layer close to the wall in such a
way that for a given boundary or material resistance it
would stabilise its depth of the viscous sublayer in a
manner that the changes in the boundary resistances at these
local regions of flow is balanced by the gravity content
of flow. In other words, in the process of balancing the
resistance, the flow distributes momentum between side and
bed regions of flow in such a way that the resultant
resistant would be minimum amongst various modes of distri-
bution. Thus, if T, and T are average material resis-
> bo so
tance for bed and sides respectively, we may expect

"'l*te + £ > > - * . (3.17)


« Minimum.

Since this resultant material resistance is balanced by the


gravity content of flow, T can be equated to VR3 as the
effective shear stress for the channel cross section, 7
is the unit weight of the water, A the hydraulic mean radius
of flow and 5 the hydraulic slope.

When Manning's coefficient of roughness is considered,


the equivalent relation may be obtained as

- 35
V < n ^ •< » ^ - ^ (3.18)

= Minimum

where ne is the minimum effective coefficient of roughness

as appearing in the Manning's equation of flow for channel


with rigid boundary, having the same material roughness n
free from shape effect.

To solve for T b o and T g Q or nfeo and n g Q , given pband


p_ as the wetted perimeter for bed and sides, using rela-
tions (3.13) and (3.17) or (3.16) and (3.13) a3 the case
may be, one may consider the two relations as linear in
p. and p with p. + p = p. To solve for component resis-
tances using the non-linear constraint in corresponding
cases,the linear relations could be normalised to obtain
the minimum effective resistance parameters in the two cases
as

and r?e/2= r?J2 [p/(p* + p *) U^ (3#20)

The line corresponding to relation (3.16) has been drawn


in Fig. 3.2, for an hypothetical data on p, , p and n .
The axes are represented by n3/2
' and
3/2 3/2
of . A straight line AB intercepting the axes of n_'
at {TC' p/ps ) and ny at (n^' p/pb) may be referred to.

- 37 -
-36-
Given only the relation (3*16) and two unknowns, any point
on the line would give a solution for n 0 and n. and thus
infinite many solutions may be obtainable. The one which
corresponds to the constraint is the point where the normal
from the origin meets the straight line AB, The co-ordi-
nates of the perpendicular would render the solution for
the component roughness consistent with the assumption
of minimum resistance. The solutions for n^ and n are

ALZ - n o / 2 [p V ( P o + *!>J
(3.21)
B
.o 2 • n o / 2 tP P./(Ph +
Ps>3

Corresponding set of solutions for relations (3.13) and


(3,17) is obtained as

T T p
bo * 0 fcPb^Pb * s^
(3.22)
*so " TotP P , / < P £ +
P.>]

It may be seen that to make use of the solutions given in


relations (3.21) and (3.22) one must have the values n o r
T known. But normally n Q is known from Manning1s equation
only in ease of very wide channel and so also t viz.
when B/d ratio tends to infinitely great values. But for

- 38 -
ail valuta or H/a ratio, commonly mat witn m practice,
the effective coefficient of roughness given by Manning'a
equation reflects a value for material and shape together
aa would be evident from the relation (3.20}. Values of
ao or To ean be obtained from relationa (3*20) or (3*19)

respectively. It oan be seen by comparing the relations


(3.12) and (3.19) that the value of 'c', average shape
factor for the channel cross section is given aa

[p/<Pb* Ps>/ l/2 3 (3.23)

But by definition

T CT
b " bo
and T. * CT
so

+
Whence *be- * 0 CPV<P2 P^ 3 / 2 ]
* %[P PfeApJ • pj)]

and T
ae" %tP Pa / ( Pb • P. >] (3.24)

where T ^ and T - # represent effective bed and side shear


stress, solutions of T ^ and r g respectively consistent with
the constraint of minimum resistance. Corresponding set
of solutions for nfa and n may be obtained as

- 39 -
(3.25)

Fhus from measurable data on n e , p b and p 0 and similar


data for shear stress, one may obtain theoretical values
of effective resistance parameters using relations (3.24)
and (3.25). It may be noted that these solutions do not
include the effects of turbulent eddies and secondary
currents averaged for bed and side regions of flow. The
solutions given by (3.24) and (3.25) satisfy the following
conditions:

n
(i) P b - P> % - V s " °*

(ii) p s - p} n s - n0J ab - 0.

The ease of shallow section in (i) above corresponds to B/d


ratio tending to oc, and deep and narrow section in (11)
refers to B/d ratio tending to 0.

3.7 Evaluation of shape effect


As may be seen from the solutions obtained in relations
(3.19) and (3.20) the shape effects superimposed on material
roughness for the cross section provide effective resistance
parameters for the whole section. Due to shape effect only

- 4© -
o
'<

00

X
»

cr

5
z
<

z
o
H
U
W
U-
U-.
UJ
UJ
CL
<
X
U)

o
z
o
I
l/>

>

'U S.ONINWVW NO A33dd3 3dVHS 3AU\rt2tf

-41-
•X3

CO

*~ [0- f ^)» + tC p/a W('ti!!!A z + p / 9 ^ [°l/(°l - a l ^

-4i-
material resistance parameters assume different rallies
when placed in the sides or bed of a channel. For easy
determination of shape effect, for a given flow condition,
the curves are provided in Fig. 3.3(a) and 3.3(b) for dif-
ferent channel geometries and aspect ratios. It can be
seen that the effect of shape is maximum for rectangular
channels at (B/d) ratio equal to 2 and for trapesoidal
channels at (B/d) ratio equal to 2(m +1) ' . It would be
seen that as an effect of shape the resistance coefficient
n is at worst 27 per cent and T . about 41 per cent higher
than the material roughness parameters n and T respec-
tively. Since absolute values of nQ and T_ are very small
the percentage increase in their magnitudes on account of
shape effect becomes indistinguishable.

Ven-te-Chow (1959) advocates an increase in n 0 of the


order 0.005 for all shapes in general. If n » 0.04 as in
case of very rough channels, Increase in n dueAshape
effect works out to 12 per cent. In case of smooth chan-
nels, with n being of the order 0.01, the 0*005 Increase
on account of shape works out to approximately 50 per cent.
On an average, percentage Increase, therefore works out to
about 31 per cent as against theoretically derived figure
of 27 per cent at the most,

3.d Experimental data


Experimental data utilised in the present studies were

- 43 -
I XJ ••
CM »f\
©
.
© •
• i*
• •
i n go
o
_*
in
CM

HUA11
M» H CM C* A C#> 4.
4-
fro « _ # • • • • _ * •
* <« H rt H • «
CM ri

*> vo o
r» CM CM CM

49 O » • J « » » I » . .»
O CM K> H t - r4 "•"

? tenI 4 A i t t 4 4 ©
si «* V
(6 V
a. w ri o
o* 10 o* c?
<t N * vo
o

o
o
•H

1 d o
U

S P4
A £
o
o m ©
•p • d
m in
tf\ H J i «
© C/» H rt © rt CM
"•' CM r # • • •
§§
N

• • • O O © O o

"si in tn
iH CM
_• •
CM £ . c- m CM
H O O O Q
_» • • • •
o o o o o o o

£ i S £8d &£
in
in
o o O H © CM

S"
* * * « P.
v\ CM <£
|*»

o
*

©
A
CM

d
CM O C* 4-
««•»

o'
n 1*% CM

©
> d
d
d
d
VO MS O tn
CM
S "^ £>
O CM O
d
• • o • •
o o o o
• o o

gS|g CM
vo m o
© J,1 ^ In
S• S• S• 8 i
or* i a o o o o o o d d

I •H
M H
»

m o « •? M
1 « •
o
g o

©
It •

mm
- 44 -
5

« v\
Si I
H
0
El I?
Mart
m

|
I
o
i
U
o

M
If*

a o
o*

t
. It *«&"!!

til
•s;i
« •
<rt © *

s
i

- 45 -
- 46
*J
<VJ
o
OJ
O
o

13Q0W 1V0I13«03H1

-47-
(s d + %/*\)% = 9q i
Q3S 1V3I13U03H1
( iua/uiB ") S S 3 a l S HV3HS

-48-
from the basic data on flume experiments by Basin (1365),
Jayaraman (1970), Yassin(l954), Cox (1973) and data
reported by Kerala Engineering Research Institute and
Andhra Pradesh Engineering Research Laboratory in the Annual
Review of the Central Board of Irrigation and Power, India
(1974). The data from the above sources were in respect
of Manning's n for uniformly same material in the boundary,
Directly measured data of bed shear stress distribution or
average bed shear stress relative to YR5, were collected and
processed by the author from published results of Cruff,
Ghosh et al (1970), Komora (1973), and Kartha et al (1970).
All these data pertained to same boundary material for
sides and bed of the channel. Experimentally determined
values of T ^ from the above sources were utilised for
comparison with theoretical values obtained using rela-
tion (3.24). Data on Manning's 'n' worked out from basic
hydraulic observations were used in the stochastic model

% 3 / 2 - » i / 2 (Pb/P) • »2 / 2 <P./P> • »J /2 e
,1/2
where n^-" denotes Manning's equivalent «n* with random
term which includes e f f e c t of turbulent eddies and secondary
currents since the flow assumes the resistance value i n -
clusive of these factors besides boundary r e s i s t a n c e .
Least square estimates of nfe and n were obtained using
procedures developed by the author (1970), Estimated nfe
and t h e o r e t i c a l nv_a?e expected to d i f f e r by an amount

- 49 -
representing the cross sectional average resistance on
account of turbulent eddies and secondary currents, not
accounted for in the relation (3.25). Estimates of xfe
were also obtained from this data using the relation

*b * * (nb7/Sl^2)3^2S (3.26)

A total of 41 sets of measurements comprising of 451 series


of experiments were studied and different parameters esti-
mated. Tables 3.1 and 3.2 give the range of hydraulic
data used and range of boundary material and B/d ratios.

3.9 Analysis of data


The data was analysed for the range more commonly
met with in practice viz.

(P|/P.) > 2

or [B/d (m2*!)1/2] > 2

This condition reduces to (B/d) > 2 in case of rectangular


channels. Most of the experimental data pertained to the
range of (B/d) ratio between 2 to 20. Estimates of nfa and
t b were compared against their theoretical eounter parts.
The values of correlation coefficients in the two cases
were obtained as 0.96 and 0.99 respectively. Figs.
3* 4 and 3.5 show the plots of experimental and theoretical
measures in respect of the two parameters. To permit a

- 50 -
uniform comparison, data on n. were also transformed to
T b using relation(3.26 ) The oombined experimental data
on tv was compared against Tfe0, obtained by the model.
The correlation was found to be as high as 0.99. The
proportionality constant was estimated by least square
method. The value obtained was (1->0.1D* It may be seen
that the term -0.11 indicates that observed resistance
parameters are less than the laminar shear stress in terms
of deterministic component of the total resistance encoun-
tered by the flow.

Thus B(t b |p 0 ) = 0.89 Te[p PfcApJ • P J ) 3

The distribution function of the cross sectional average


value of C K *>* obtained in the Chapter V.

3.10 Summary and conclusions


The studies carried out into the nature of variations
in resistance parameters In turbulent flow, throw new
light on the effect of shape on Manning's roughness coeffi-
cient fnf or on the cross sectional average shear stress
IfRS. Identification of Manning's »n' with the effective
minimum resistance to flow i.e. n e leads to functional
relation between effective coefficient of roughness and
material roughness in terms of aspect ratio B/d vide relation
3.20. Material roughness is defined as part of the resist-
ance which is due to material alone, exclusive of shape

51 -
» f ¥ * * M f t f f l ^ Boughayta, for Uniform

< * « ) , Bed Computed Corrtotion Abgolatt coe-


to depth Manning * a factor for f f i c i e n t of
ratio n or n ahape c roughaOaat » 0

«.—

4,9 0.0171 1.17 0. 01462


9.8 0.0161 1.10 0.01463
12.4 0. 0160 1.09 0.01468

average 0.0165 0.01464


Percent standard
deviation a m 2,0
Rectangular channel
Botton elope . . 0.001
Bod width Metre* . . 0.714
Material - cruehed l i a e steno.

- 52 -
tffi¥..7»3Jbl

4.9 0.0125 1.170 0.0107


8.0 0, 0118 1.138 0.0104
10.2 0. 0120 1.100 0.0109
15.9 0. 0115 1.075 0.0107

Average 0.0123 0.0107


Percent standard
deviation a = 2.0
lottos * 0.001
Wid* . . 0.736 • .
Material - Safety tread (a s e l f adherent Baterial to
i n c r e a s e roughness)

M P 3»? (t)

4.83 0.0114 1.194 0,00955


6,04 0.0102 1.168 0.00873
7,07 0.0108 1.150 0.00939
9-59 0. 0101 1.118 0,00903
11.14 0.0106 1.102 0.00962
13.81 0.0107 1.085 0.00986

Average 0.0106 0.00956


Pereent standard
deviation a * 4.4
l o t t o n slope . . 0.001
led width .. 0.736
Material .. P l a a t i o ooated plywood.

- 53 -
effect or any other remittance factors. The value of
material roughness nQ corresponds Manning's fn' for very
large B/d ratio. Observed valuea of variation in Manning's
*n* with changes in depth or (B/d) ratio is explained in
ease of geometrically defined cross-section with uniformly
the sane material at the boundary. It thus becomes possible
to separate out material roughness from the values of
effective roughness by correcting for the shape. Such a
correction and separation has been shown in Tables 3.3(a)
to3.3(b)for experimental data of Cox in respect of three
typical material roughnesses. It is seen that effective
roughness is always greater than material roughness.
Alternatively if shear stress parameter is considered

o •• e •"

where TdS is the maximum shear-stress at the bed for a given


depth of flow d. If x% is expressed in terms of material
roughness parameter tQ and shape factor 'e't one would get

T« < * A e < JfdS

Dividing each side of the above inequality by •C one gets

tt c / c) ~ T o - 11**/*) ..... (3.27)

Denoting (l/c) by a, we get

T 0 « $ t0 ^ a(*dS) (3.2d)

- 54
where a is always less than unity and would tend to unity
when (B/d) ratio is very large i.e. when depth is decreas-
ing for fixed bed width. An important oonelusion is deri-
vable from the inequality (3.28J that the actual friction
force t for the material developed on the bottom covers
only part of the expression YdS. This fact derived here
has been demonstrated experimentally by Engels and Schober
(1975 Bogardl) and provides theoretically measure for the
coefficient a proposed by them. Cox's study of the nature
of variation in Manning's »n» for uniform roughness indi-
cated that as width to depth ratio decreased, the value of
»nf increased. This finding is fully vindicated by consi-
dering the Fig. 3.3(a).His qualitative findings are not only
supported by the present investigations but also quanti-
tative assessment in terms of equation (3.20) is now
possible. The observed values of bed shear-stress and
coefficient of roughness, for rectangular channels or trape-
zoidal channels, show close agreement with those obtained
using the concept of minimum over all frictional resistance
offerred by the entire wetted perimeter vide Figs, 3.4,
and 3.5

It is now possible to meaningfully interpret different


resistance parameters occurring in a flow equation and
related phenomena to describe various physical situations:
(i) Parameters representing absolute roughness due
to material alone viz. n and T Q . These are now possible

- 55 -
to be derived from measurable parameters n_ and f. using
relation (3.20) and (3.19).
(ii) The total effective resistance parameters n^' p
or T C P may be partioned into their components for bed or
sides of a flow section without resorting to any side wall
correction procedure or other methods of separation of
roughness parameters.
From the previous studies, the following broad conclu-
sions are arrived at:
(a) Frictional resistance developed at the boundary
for material in terms of roughness parameters n Q or T 0
Is always smaller than the effective resistance encountered
by the flow. The latter approaches the former only when
the channel is very wide.
(b) It is possible to separate out material roughness
from effv ctive roughness inclusive of shape effect as
obtained from Manning's equation, in case of the cross
section having uniformly the same material along the
boundary.
(cj The effective and material roughness coefficients
on the bed, on the sides and for the entire cross section,
achieve different values due to shape effect. The shape
correction for working out the effective bed or side rough-
ness are derived as

- 56 -
1 O
and c s - p Pa/(pb • pj)

(d) In design of lined canal and sediment transport


studies, the effective bed-shear stress could be obtained
to a close approximation by the relation

T
b * °'*9Tbe
* 0.*9 [r9ip Pb /(pJ + p 2)}]

- 57

You might also like