Chezy Formula
Chezy Formula
Chezy Formula
WITOLD G. STRUPCZEWSKI
Institute of Geophysics, Polish Academy of Sciences, Ks. Janusza 64,
01-452 Warsaw, Poland
ROMUALD SZYMKIEWICZ
Technical University of Gdansk, Majakowskiego 11/12, 80-952 Gdansk,
Poland
Abstract The Chezy friction formula for steady flow in a uniform
symmetrical channel with constant slope-friction factor is examined
mathematically. Firstly, a wide rectangular channel and a semi-circular
channel are compared in respect of the mean flow velocity using the
Chezy formula with the Manning, Chezy and logarithmic laws for velo-
city. Then the inverse Chezy problem, i.e. the determination of the
channel shape above the reference level for a given depth/discharge
rating curve, is posed and the differential-integral equation for its
solution is derived. The rating curves used for computation are the
results of multiplying the discharge for a trapezoidal shape above the
reference level by an exponential function. To facilitate interpretation of
the numerical results, the relationship between side slope and discharge
is analysed. It is shown by the inverse problem solution that an exponen-
tial reduction of channel flow capacity changes linear channel sides into
convex sides (making the cross section shape wider) while an exponential
increase of capacity causes changes into concave sides (reducing a section
width) which is against common sense.
Analyse des paradoxes rsultant de la formule de Chezy avec
rugosit constante: I. Courbe hauteur-dbit
Rsum La formule de Chzy pour l'coulement permanent dans un
canal symtrique dont le facteur pente-frottement est constant a t
analyse mathmatiquement. On a tout d'abord compar un canal rect-
angulaire et un canal semi-circulaire eu gard la vitesse moyenne de
l'coulement l'aide de la formule de Chzy avec les lois de Manning,
de Chzy et logarithmique pour la vitesse. On a ensuite pos le problme
de Chzy inverse, c'est dire celui de la dtermination de la section
transversale du canal au dessus d'un niveau initial, pour une courbe
hauteur-dbit donne. Une quation diffrentielle-intgrale a t propose
pour rsoudre ce problme. Les courbes utilises dans les calculs ont t
dtermines en multipliant le dbit pour une section trapzodale au
dessus du niveau de dbit nul par une fonction exponentielle. Pour
faciliter l'interprtation des rsultats numriques on a analys la relation
entre la pente et le dbit. On a montr, l'aide de la solution du
problme inverse, que la rduction de l'coulement rend la section plus
large, tandis que son augmentation implique la rduction du canal, ce qui
contredit le sens commun.
INTRODUCTION
v* = fji (4)
or, substituting the value of r0 from equation (3)
V* = yJgRS (4a)
Paradoxes arisingfromthe Chezy formula - depth-discharge curve 663
where p and g are the fluid density and the gravitational acceleration,
respectively. The ratio may then be expressed as a function of the
relative distance from the boundary, i.e.:
y_ (5)
y0
y_ (5a)
y0
with/? = 0 (i.e. v constant which in fact unreal) for Chezy and/? = 1/6
for Manning. It is to be noted that from theoretical concepts of open
channel flow mechanics equation (5) takes the form of the Prandtl-von
Krmn universal velocity distribution law:
1In y_ (5b)
ya
vdA
av*B av*Bh{P+l) (6)
V = yPdy
A
Ay/ V ' (p + DAyf
where B is the width of the channel, while in the semi-circular section:
vA
h(p+2) (7)
V= irav yP+ldy = irav
^y0f (p+2)Ay/
Substituting equation (4a) for v* and expressing the depth h by the
hydraulic radius for the wide rectangular channel yields:
W. G. Strupczewski & R. Szymkiewicz
aRP
V= -4gRS (6a)
V y/gRS (7a)
(P+2)y/'
For the semi-circular form, equation (7a) can be rewritten in the form
of equation (6a) as:
aRP
V = F- -yfgRS (7b)
p
(p^)y0
where F is the shape factor:
In 0.5
y0 R
F = (9)
R_ + 1-X
In
To R
In this case the shape factor tends very slowly to unity for the hydraulic
radius/representative roughness height ratio, goes to infinity, i.e. for
frictionless channels, as shown in Fig. 1.
for other shapes, the distribution of shear stress along the wetted peri-
meter is not uniform and the isovels (lines of equal velocity) will not be
similar to those in the two special cases above.
Paradoxes arising from the Chezy formula - depth-discharge curve 665
MATHEMATICAL FORMULATION
R/y0
Fig. 1 Shape factor as a function of the ratio of the hydraulic radius to the
roughness height for the logarithmic law of velocity distribution.
Above the reference level (h0) the geometrical factor (equation (2)) is:
G=fh{h) for h > h (10)
Given the initial conditions (A0, P0, T0) the task is to find the symmetrical cross
section profile above the initial level, i.e.:
T = 4>(h) for h > h (11)
The restriction of a symmetrical section is applied in order to get the unique
shape for the surface width-depth profile given by equation (11).
By means of the inverse problem solution it will be possible to check
whether the Chezy formula is conformable with common sense with respect to
the channel capacity/size axiom and, if not, to assess whether such noncon-
forming shapes can be frequently met in natural or trained river channels.
Because equation (11) contains only T and h variables, A and P in
equation (2) and their derivatives with respect to depth shall be expressed in
terms of these two variables by means of the following geometrical relation-
ships, the last two of which are valid for a symmetrical section only:
=T (12b)
dh
P = P0 + 2 f [l +(d772d/z)2f5d/z (12c)
= 2[l+(d772d/z)2f5 (12d)
A basic shape (1) will be assumed with its stage-discharge rating curve derived
from the Chezy equation with Manning or Chezy friction. Modifying the stage-
discharge equation over the reference level, the inverse problem solution will
be used to generate secondary shapes (2) which will be analysed from the point
of view of the capacity/size axiom.
For simplicity assume that the basic channel (1) has constant sides of
slope z above an initial level h0 = 0. Then the relationships (12a-d) take the
form:
Ax = A0 + T0h+zh2 (14a)
i l l = T0 + 2zh (14b)
dh
(14c)
Px = P0 + 2hyi+z
dP
I = 2V1 +Z2 < 14d )
dh
so that equation (2) becomes:
(A0+T0h+Zh2r
G1 = r.1 (14)
(P0 + 2hil+z2 )
Therefore the basic section shape (1) of the prismatic channel is sym-
metrical and composed of a reference shape defined at a reference level h0 by
A0, P0 and T0, with an upper part with sides constant slope z above the level
K
A secondary shape (2) related to the basic shape (1) may be found from:
G2(h) = CGx(h) (15)
where:
1 for h < 0 (16)
C= i
C{h) for h > 0
Paradoxes arising from the Chezy formula - depth-discharge curve 667
The solution of the inverse Chezy problem may be obtained numerically from
the equivalence of terms:
1
(17)
d/z d/z d/z
with the initial conditions Ax = A2 = A0, Px = P2 = P0, Tx = T2 = T0 for
h = 0. G2 and Gx are given by equations (2) and (14), respectively, while their
first derivatives with respect to stage h are given by:
GX r(T0+2zh) _2(r-l)Vl+z
= G, (19)
2
An + Tnh+zh f ^
0
Pn+2hyl+z
Substituting equations (18), (19) and (2) into equation (17) yields:
Then substituting equations (12b,d) and (14a,d) into equation (17a) gives T2 in
terms of h and P2:
0.5
d^
1+
^2 2 2d/z
(17b)
l/(r-l) l/(r-l)
4
G, CG,
A2r/{r-l)(P0+2hil+z2 ) (20)
Substituting equation (20) into equation (17b) and introducing the auxiliary
functions:
r(T0+2zh) _ 2(r-l)\/l+z 2
Uh) (21c)
cw*> (po+2hir^F')
one arrives at:
0.5
f 12
dr2 (17c)
.Lji-^.c" 1+
h 3
Cd/z
Reordering and denoting A2 = A, T2 = T gives the differential-integral
equation:
dF = 2C1/(1"r) (22)
3 1 2
d^ ~~ 0 r 0 2 ^ A C d / z
Equation (22) has been solved numerically by the Runge-Kutta fourth order
method. An exponential function may be taken for C(h):
0 < h < /zmax. The step (Ah) of the computation was each time selected to
fulfil two accuracy conditions:
1. there shall be no deviation of the solution from a straight line for shape
(1), i.e. for e = 0; and
2. equation (15) shall be preserved within the whole range of computation,
i.e. for 0 < h < /zmax.
Obviously it is sufficient to check both conditions at the /zmax stage. In general,
a shorter step Ah is applied for small values of the side slope z than for large
values.
For a small depth increment, h, all terms higher than the linear one can be
neglected and equation (23) reduces to:
One can deduce from equation (26) the existence of an upper bound to
e in equation (16a). Confronting equations (17) and (26):
2(r-l) (28)
< 1+z2 -1
Lack of a finite lower bound to e points out the possibility of getting from
equation (22) a shape for G being a monotonically decreasing function of h.
For a pair (z, ,zj) where z-t > Zj and the initial conditions, one can derive
the upper limit of stage interval h* for which:
G{h,z,) < G(h,zp (29)
as the greater of two roots of the equation:
Gih*,Zi) = Gih*,Zj) (30)
1.0
1.
0.8
z
j
=
.5
f
0.2
.0
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.8 1.8 2.0 2.2 2.4 2.6 2.8 3.0
One may argue that due to the discontinuity of the G^h) derivatives at
the reference level, such a shape should not be considered as a simple one, i.e.
Paradoxes arising from the Chezy formula - depth-discharge curve 671
NUMERICAL RESULTS
The previous section led to the conclusion that, according to the Chezy
formula, making side walls steeper results in an increase of discharge over a
finite stage interval, which is against common sense. Furthermore, to avoid the
argument about the discontinuity of the geometrical factor derivatives, one can
replace equation (16a) by:
(T - T 0 )/2
Fig. 3 Secondary shapes related by equations (15) and (16b) to the basic
one shown above the reference level. The basic shape: trapezoidal with the
reference values A0 = 9, T0 = 12 and P0 = 12.324 (z = z0 = 3).
Manning friction, i.e. r = 5/3.
672 W. G. Strupczewski & R. Szymkiewicz
discharge ratio Q2/Q\ equals 1.0228 and .9777, respectively ,while the surface
width ratio T2ITX equals .8959 and 1.1077, respectively. However, most impor-
tantly, note that secondary shapes wider than the basic shape correspond to
lower flow capacity and vice versa. A third observation is that, while for
negative values of the parameter e the solution is unbounded, there is an upper
bound for positive value of e. The upper bound corresponds to the vertical
slope (z = 0), which is in conformity with equation (26). Then the maximum
possible flow increment (26), where subscript "0" denotes the bound, becomes
insufficient to fulfil equation (15). The value of the bound decreases as the side
slope of the basic shape decreases.
It is seen in Fig. 3 that the discharge decreases with a growing channel
width in both the cases of convex and concave shapes. Bringing the capacity/
size axiom to mind, if the rating curve of the basic trapezoidal section shape
got from the Chezy formula with a constant value of Manning roughness were
correct, discharge would be underestimated for all secondary convex shapes
and overestimated for concave shapes.
Quite similar results can be obtained repeating the computations for
Chezy friction as well as using equation (16a) instead of (16b). Therefore, the
blame for the disagreement of the results with common sense should be put
first on the Chezy formula but not on the friction law.
CONCLUSIONS
REFERENCES
Aldridge, B. N. & Garrett, J. M.(1973) Roughness coefficients for streams in Arizona. US Geological
Survey open-file report.
Bakhmeteff, B. A. (1932) Hydraulics of Open Channels. McGraw Hill, New York, USA.
Chow, V. T. (1959) Open Channel Hydraulics. McGraw Hill, New York, USA.
Jarrett, R. D. (1985) Determination of roughness coefficients for streams in Colorado. Water Res.
Investigations Report 85-4004, US Geological Survey, Lakewood, Colorado, USA.
Jarrett, R. D. (1992) Hydraulics of mountain rivers. In: Channel Flow Resistance, 287-298. Water
Resources Publ., Colorado, USA.
Jarrett, R. D. (1994) Circular of ASCE Session on Applied Hydraulics of Flow in Mountain Rivers.
Buffalo, USA.
King, H. W. (1954) Handbook of Hydraulics. McGraw Hill, New York, USA.
Leliavsky, S. (1959) Irrigation and Hydraulic Design. Vol. I, Chapman & Hall, London, UK.
Limerinos, J. T. (1970) Determination of the Manning coefficient from measured bed roughness in natural
channels. Water Supply Paper 1898-B, US Geological Survey.
Strupczewski, W. G. (1996) Warning of application of the Chezy-Manning formula regardless of channel
shape. In: Proc. Int. Conf. Hydrology and Water Resources, (December 1993, New Delhi, India),
371-387. Kluwer Academic Publishers, The Netherlands.
Strupczewski, W. G. & Sua, D. Z. (1983) Rating curves of Liberian rivers. Part III. Hydraulic estimation.
UN ProjectTech. ReportNo. 61, Liberian Hydrological Service, Department of Mineral Exploration
and Research, Ministry of Lands & Mines, Monrovia, Liberia.
Strupczewski, W. G. & Szymkiewicz, R. (1989) On direct applicability of the Chezy formula to natural
channels. In: Proc. Int. Conf. on Channel Flow and Catchment Runoff, (University of Virginia,
USA), 350-361.
Strupczewski, W. G. & Szymkiewicz, R. (1996) Analysis of paradoxes arising form the Chezy formula with
constant roughness: II. Flow area-discharge curve. Hydrol. Sci. J. 41(5).
US Corps of Engineers (1959) Hydraulic Design Criteria. Vicksburgs Experimental Station.
Yen, B. C. (ed) (1992) Channel Flow Resistance: Centennial ofManning's Formula, 453. Water Resources
Publ., Littleton, Colorado, USA.