International Water Technology Journal, IWTJ Vol. 6No.

3 September 2016

FLOOD ROUTING IN RIVERS BY MUSKINGUMS METHOD WITH NEW

ADJUSTED COEFFICIENTS
Kaveh Ostad-Ali-Askari1*, Mohammad Shayannejad2
1
PhD Student, Department of Water Engineering, Faculty of Civil Engineering, Najafabad
Branch, Islamic Azad University, Najafabad, Iran, *Corresponding author
E-mail: : [email protected]
2
Associate Professor, Water Engineering Department, Isfahan University of Technology, Isfahan,
Iran, E-mail: [email protected]

ABSTRACT

For determining of Muskingum model coefficients, requires to an output hydrograph. Such

hydrograph is not available in most rivers. In this research, the Muskingums new coefficients are
determined by the method that which is not require to output hydrograph and its accuracy is high. This
coefficient is determined based on kinematic wave model with suitable scheme. The comparison
between results of Muskingum model with new coefficients and dynamic wave model, showed that
the new coefficients, are valid at special conditions. The new coefficients were adjusted by
optimization technique for all conditions. The new adjusted coefficients are function of bed slop,
bottom width and Mannings roughness coefficient for the river. The results of these coefficients were
validated by dynamic wave model.

Keywords: Flood routing, Kinematic wave, Muskingum

Received 16 May 2015.Accepted 29, June2016

1 INTRODUCTION

The flood flow in rivers is an unsteady flow and its characteristic is varied with time. These
variations are made by human or natural factors. The flow variations are described by a hydrograph in
hydrology. The flood routing investigates the variations of depth and discharge flow in rivers or
channels. The methods or models of flood routing are different. The full dynamic model is the most
accurate of them, in which the continuity and momentum equations are solved completely. Other
methods such as kinematic wave, the continuity equation and summarized form of momentum
equation are solved. These methods were compared by Samani and Shayannejad (Samani, 2000).
The kinematic wave method is valid if the local and convection accelerations are negligible and slopes
of surface water and bed are same (Chaudhry, 1993). The generated error in results of kinematic wave
model is due to basic assumptions and finite difference numerical solution (Weinmann, 1979). An
usual and simple method for flood routing in rivers is Muskingums method. This method was based
on continuity equation and its equation is following:

O2 C1 I 2 C2 O1 C3 I1 (1)

where I 1 and I 2 =input discharges at t1 t 2 time steps ; O1 O2 =output discharge at t1 t 2 time

steps. C1 , C 2 , C3 =constant coefficients are which determined by a given input and output hydrograph.
The disadvantages of this method are:
1. It requires to an output hydrograph for calculating of its constant coefficients.
2. It determines an output hydrograph only at one point of river.
3. The applied assumptions in this method cause low its accuracy.
The many of researchers have presented coefficients for Muskingums method to remove above
disadvantages. For example Cunge (Cunge, 1965), Ponce (Ponce, 1986) and Bowen and Koussis
(Bowen, 1989) presented a series of coefficients, but the accuracy was still low.

189
International Water Technology Journal, IWTJ Vol. 6No.3 September 2016

In this paper the Muskingums method with new adjusted coefficients have been derived of
kinematic wave and then they have been adjusted by full dynamic model. This model validates the
result of the new method.

2 MATERIALS AND METHODS

The kinematic wave method is combination of continuity equation and an equation for flow
resistance as Chezy-Manning equation. These equations are:

Q A
0 (2)
x t

A K .Q B (3)
3
Where Q =discharge;. A =area cross-section; x =distance; t =time, B and:
5
3
2 5
nP 3
K (4)
S0

Where P =wetting perimeter; n =roughness Mannings coefficient; S 0 =bed slope.
K in equation 4 is determined by considering a given discharge (base flow) and calculation of wetting
perimeter for this discharge.
The derivative of equation 3 is:

A Q
K .B.Q B 1 (5)
t t

The combination of equations 2 and 5 is:

Q Q
K .B.Q B 1 0 (6)
x t

For solving equation 6 by numerical method, its terms are discrete following form (Chow, 1988):

Q QiJ11 QiJ 1
(7)
x x

Q QiJ11 QiJ1
(8)
t t

QiJ 1 QiJ1
Q (9)
2

Where i =local step number; J =time step number; t =time between two sequential time step;
x =distance between two sequential local step.
By substituting equations 7, 8 and 9 into equation 6 gives following equation:

QiJ11 C1 .QiJ 1 C2 QiJ1 (10)

190
International Water Technology Journal, IWTJ Vol. 6No.3 September 2016

Where:
t
C1 (11)
t K .B.C0 x
K .B.C 0 x
C2 (12)
t K .B.C 0 x

B 1
Q J 1 QiJ1
C 0 i
(13)
2

Equation 10 is Muskingums method with new coefficients and a coefficient is equal to zero. In
spite of old Muskingums method, this new coefficients are not constant during time step calculations,
because the coefficients depend on C 0 and it is not constant. Besides for calculating of the new
coefficients do not require to a given output hydrograph and flood routing can be carried out at any
point of river.

The grid size calculation must choose so that the Courant number is equal or less than one. This
number is:

C K .t
C (14)
x

Where C =Courant number; C K =celerity (velocity of wave transport). It is determined from equation
3:

Q 1
CK (15)
A K .B.Q B 1

The result of this new method is compared with the results of full dynamic model that its accuracy
has been validated by real data. The equation 2 and following equation (momentum) constitute full
dynamic model:

V V y
V g g S f S 0 0 (16)
t x x

Where V =flow velocity; y =depth of flow; g =gravitational acceleration; S f =slope of energy

grade line.Equations 2 and 16 are Saint Venant equation and are solved by numerical method. For
this work, they are discrete by Preissmann scheme (Li, 1975). Then for each reach of river, two
equations with four unknown variables are constituted. These variables are depth and velocity of flow
at two ends of any reach. Thus for M reach (with M+1 nodes), 2M nonlinear equation with 2m+2
unknown variables are constituted. Thus two equations are required, that are gained from boundary
conditions. For example, the downstream boundary condition is rating curve and upstream boundary
condition is an input hydrograph. The nonlinear equation is linearized by Newton-Raphson method.
Finally depth and velocity of flow (and then discharge) are determined at any node and any time.
In this paper, the coefficients of equation 10 were adjusted in order to increasing of the new method
accuracy, by nonlinear optimization technique with using full dynamic model. Firstly, these
coefficients were changed to following form:

C1 K1C1 (17)

191
International Water Technology Journal, IWTJ Vol. 6No.3 September 2016

C2 K 2 C2 (18)

C0 K 3C0 (19)

Instead of C1 , C 2 and C 0 at equation 10, C1 , C 2 and C 0 were applied. Secondly K 1 , K 2 and

K 3 were determined by nonlinear optimization technique according to following objective function:
N
F Qmt Qct
2
(20)
t 1

Where F =objective function; Qm =output discharge calculated by full dynamic wave; Qc =output
discharge calculated by equation 10 with new adjusted coefficients (equation 17, 18 and 19) and
N =number of time step.
Thirdly were determined relationship between K 1 , K 2 , K 3 and characteristic of river

3 RESULTS AND DISCUSION

Figure 1 shows the output hydrograph at a distance one kilometer, calculated by full dynamic
model and Muskingums method with new coefficients ( C1 , C 2 and C 0 ) for a hypothetic input
hydrograph and following input data:
n =0.035; S 0 =0.001; bottom width= 20m; t 1min ; x 100m

Figure 1 shows that there is different between results of two methods. In this research was
concluded that with increasing of bed slop and decreasing of bottom width and slope of input
hydrograph, the results of two methods were similar. On the other hand, in these cases, the kinematic
wave is dominated.

For increasing of accuracy of equation 10 its coefficients were adjusted by optimization technique
and values of K 1 , K 2 and K 3 were determined. These coefficients were not constant and were
depended on characteristic of river. K (Equation 4) was chosen as preventative of characteristic of
river, because it depends on bed slope, wetting perimeter (depends on bottom width) and roughness
Mannings coefficient.

Figure 1: Comparison of results of full dynamic and Muskingum with new coefficients methods

Figure 2 shows the relationship between K 1 , K 2 , K 3 and K . The statistical analysis gives following
equation:

192
International Water Technology Journal, IWTJ Vol. 6No.3 September 2016

K1 0.0391LnK 1.0289 (21)

K 2 0.0577 LnK 0.956 (22)

K 3 0.59961LnK 1.4705 (23)

Figure 2: Variations of
K 1 K 2 K 3 related to K

Figure 3 validates Muskingums method with new adjusted coefficient. The following data were
used in this figure are:

n =0.02; S 0 =0.003; bottom width= 30m; t 1min ; x 100m

The accuracy of the new adjusted coefficients is more than Cunnge, Ponce and Bowen ones. Figure
3 shows results of Cunge coefficients for example. Thus Muskingums method with new adjusted
coefficients is acceptable in rivers.

Figure 3: Comparison of output hydrograph by different methods

REFERENCES

Bowen, J.D & Koussis, A.D. (1989) Storm Drain Design: Diffusive Flood Routing for PCS. J Hydr
Eng Div ASCE, 115(8), pp. 1135-1149

Chaudhry, M. H. (1993) Open channel flow, Prentice Hall, Englewood, Cliffs, N J.

Chow, V. T. & Maidment, D.R. & Mays, L.W. (1988) Applied Hydrology. McGraw Hill
Publishers.

Cunge, J.A. (1965) On The Subject of a Flood Propagation Computation Method (Muskingums
Method). J Hydr Res, 79(2), pp. 205-230.

193
International Water Technology Journal, IWTJ Vol. 6No.3 September 2016

Li, R.M.S & imons, D.B. & Steven, M.A. (1975) Nonlinear Kinematic Wave Approximation For
Water Routing. Water Resource Res. 11(2),pp. 245-252.

Ponce, V.M. (1986) Diffusion Wave Modeling of Catchment Dynamics. J Hydr Eng Div ASCE,
112(8), pp.716-727.

Samani, J.M.V & Shayannejad, M. (2000) Comparison of kinematic wave and matched diffusivity
methods with dynamic wave method in flood routing of rivers, International J of Eng Science, 3(11),
pp.29-43,

Weinmann, P.E & Laurenson, E.M. (1979) Approximate Flood Routing Methods: a Review. J
Hydr Eng Div ASCE , 105(12), pp.1521-1535.

194