Advanced Hydrology (HWRE 6031)

3. Distributed Hydraulic Routing

By: Mulu Sewinet

Debre Markos University

Distribute routing model (ST. Venant equation)
• ST. Venant equation describes both the forward or
d/s wave propagation as well as the back ward or
u/s wave propagation characteristics
• Drawbacks of the ST. Venant equations are:

• The eqns. Are non linear hyperbolic partial

differential eqns. and hence the necessity for
numerical integration.
• Solutions also requires extensive data availability
 Distribute routing model (ST. Venant equation)
Momentum terms consists of terms for the physical process
that govern flow momentum
𝜕𝑉
1) local acceleration: change in momentum due
𝜕𝑡
to change in velocity over time
𝜕𝑉
2) V Convectional acceleration : change in
𝜕𝑥
momentum due to change in velocity along the
channel
𝜕𝑦
3) g pressure force terms: due to change in water depth
𝜕𝑥
along the channel
4) g𝑆𝑜 gravity force term due to bed slope
5) g𝑆𝑓 friction force term: due to friction slope
Distribute routing model (ST. Venant equation)

• Terms 1 and 2 represent the effect of inertia forces

on the flow

• Terms 3 represent the effect of non uniform flow

and together with terms 1 and 2 may be used to
introduce the effect of backwater and u/s
propagation of disturbance for subcritical flow

• Lumped Routing methods may not perform well in

simulating the flow conditions when backwater
effects are significant and river slope is mild
 Classification of distributed routing models
Distributed routing models are classified as
• Kinematic wave model, It neglects local acceleration
and the convective acceleration and pressure terms
• Diffusion wave model: It neglects the local
acceleration and the convective acceleration but
incorporates the pressure term.
• Dynamic wave model: considers all the acceleration
and the pressure terms in the momentum equn.
 Classification of dynamic wave models
• The dynamic wave model may also broken up into
sub components based on flow type as shown
below
Kinematic Vs Dynamic Wave
 Visualization of Kinematic and Dynamic wave model
A wave is a variation in a flow, such as a change in flow
rate or water surface elevation
 Kinematic Wave Celerity
• Wave celerity is the velocity with which the variation
in a flow travels along the channel.
• It may be quite different from the water velocity.
• For kinematic wave the wave motion is described
principally by the equation of continuity b/c the
acceleration and pressure terms in the momentum
equation are negligible,
• So kinematics refers to the study of motion exclusive
of the influence of mass and force;
• This is generally valid when for steep slopes
• This also will have minimum u/s propagation of any
disturbances/changes in flow conditions.
• Hence in this case kinematic wave model is more
valid
 Kinematic Wave Celerity

• Continuity equation
𝜕𝑄 𝜕𝐴
+ =𝑞 (∗)
𝜕𝑥 𝜕𝑡

Momentum equation

The momentum equation can also described in the form

of A= 𝛼𝑄𝛽 (∗∗)
 Kinematic Wave Celerity
• From manning equation with so=sf and R=A/P; A can be
described as

• Differentiating the eqn.(A= 𝛼𝑄𝛽 ) with respect to t

𝜕𝐴
• Substituting in eqn (*) it gives
𝜕𝑡

𝜕𝑄 𝛽−1
𝜕𝑄
+ 𝛼𝛽𝑄 =𝑞 ∗∗∗
𝜕𝑥 𝜕𝑡
 Kinematic Wave Celerity
• Kinematic waves result from changes in Q. i.e 𝑄 =
𝑓 𝑥, 𝑡 ; An increment in flow, dQ, can be written as

• Dividing through by dx and rearranging gives:

𝜕𝑄 𝑑𝑡 𝜕𝑄 𝑑𝑄
+ = (∗∗∗∗)
𝜕𝑥 𝑑𝑥 𝜕𝑡 𝑑𝑥
These two eqn. (***) and (****) are identical if
𝑑𝑄
=𝑞 i.e
𝑑𝑥
𝜕𝑄 𝑑𝑡 𝜕𝑄 𝜕𝑄 𝛽−1
𝜕𝑄
+ = + 𝛼𝛽𝑄
𝜕𝑥 𝑑𝑥 𝜕𝑡 𝜕𝑥 𝜕𝑡
 Kinematic Wave Celerity
𝑑𝑥 1
and =
𝑑𝑡 𝛼𝛽𝑄 𝛽−1
Differentiating eqn. 𝐴 = 𝛼𝑄𝛽 and rearranging gives

𝑑𝑥 𝑑𝑄
Then =
𝑑𝑡 𝑑𝐴
From this kinematic wave celerity (𝑐𝑘 )
𝑑𝑄 𝑑𝑥
𝑐𝑘 = =
𝑑𝐴 𝑑𝑡
 Kinematic Wave Celerity

𝑑𝑥
• This implies an observer moving at a velocity of =
𝑑𝑡
𝑐𝑘 and the flow rate increasing at a rate of
𝑑𝑄
=𝑞
𝑑𝑥
The kinematic wave celerity can also be expressed in
terms of depth y as
1 𝑑𝑄
𝑐𝑘 =
𝐵 𝑑𝑦
Application of kinematic wave

• Flood waves which approximate kinematic waves

travel with wave celerity (𝑐 = 𝛽𝑉) and subjected to
very little or no attenuation.

• In practice flood waves are kinematic if they are of

long duration or travel on a channel of steep slope
 Dynamic Wave Celerity

 As shown in Fig., in moving water there are two dynamic

waves,
 (1) Proceeding upstream with velocity V- cd and
 (2) Proceeding downstream with velocity V + cd.
 For the upstream wave to move up the channel requires V< cd, or,
equivalently, that the flow be subcritical, since 𝑉 = g𝑦 is the
critical velocity of a rectangular, open-channel flow.
 Dynamic Wave Celerity
• The dynamic wave celerity can be found by
developing the characteristic equations for the Saint-
Venant equations (Henderson, 1966):

• and

• cd is dynamic wave celerity and for rectangular

channel is given by

• Where y =depth of flow

Solutions to St. Venant equations

•
 Solutions to St. Venant equations
• Analytical – Solved by integrating partial
differential equations
• Applicable to only a few special simple cases of
kinematic waves

Numerical – Finite difference approximation

 Calculations are performed on a grid placed over
the (x, t) plane
 Flow and water surface elevation are obtained for
incremental time and distances along the channel
 Solutions to St. Venant equations
• Finite difference scheme
FINITE-DIFFERENCE APPROXIMATIONS
FINITE-DIFFERENCE APPROXIMATIONS
• The partial differential equations, in general, must
be solved using numerical methods.

• In numerical methods for solving partial differential

equations, the calculations are performed on a grid
placed over the x-t plane.

• The x-t grid is a network of points defined by taking

distance increments of length ∆𝑥 and time
increments of duration ∆𝑡.
FINITE-DIFFERENCE APPROXIMATIONS
FINITE-DIFFERENCE APPROXIMATIONS

• Numerical schemes transform the governing partial

differential equations into a set of algebraic finite-
difference equations, which may be linear or
nonlinear.

• The finite-difference equations represent the

spatial and temporal derivatives in terms of the
unknown variables on both the current time line,
𝑗 + 1, and the preceding time line, j , where all the
values are known from previous computation
 Finite Differences
• Finite-difference approximations can be derived for
a function u(x) by using Taylor series expansion

Fig. Finite difference approximations for the function u(x).

 Finite Differences contd…
A Taylor series expansion of u(x) at 𝒙 + ∆𝒙 produces
1 2 ′′ 1 3 ′′′
𝑢 𝑥 + ∆𝑥 = 𝑢 𝑥 + ∆𝑥𝑢′ 𝑥 + ∆𝑥 𝑢 𝑥 + ∆𝑥 𝑢 𝑥 + ⋯ (1)
2 6
𝜕𝑢 𝜕2 𝑢
Where 𝑢′ 𝑥 = , 𝑢′′ 𝑥 = , … . . 𝑎𝑛𝑑 𝑠𝑜 𝑜𝑛.
𝜕𝑥 𝜕𝑥 2
The Taylor series expansion at 𝒙 − ∆𝒙 is
1 2 ′′ 1 3 ′′′
𝑢 𝑥 − ∆𝑥 = 𝑢 𝑥 − ∆𝑥𝑢′ 𝑥 + ∆𝑥 𝑢 𝑥 − ∆𝑥 𝑢 𝑥 + ⋯ (2)
2 6
1) By using central difference approximation
 Uses the difference of eqn.1 to eqn.2 i.e subtracting
eqn.2 from eqn. 1 and it gives

where 0 ∆𝑥 3 represents a residual containing the third and

higher order terms.
 Finite Differences contd…
Solving for u'(x) assuming ∆𝑥 3 ≈ 0 results in

 which has an error of approximation of order ∆𝑥 2 . This

approximation error, due to dropping the higher order
terms, is also referred to as a truncation error
2) by using forward difference approximation
 subtracting u(x) from eqn.1.

 Assuming second and higher order terms are negligible,

solving for u'(x) gives
Finite Differences contd…
3) By using Backward-difference approximation:
• Uses the difference defined by subtracting eqn.2
from u(x),

 so that solving for u'{x) gives

 Numerical schemes for a solution
• A finite-difference method may employ either an
explicit scheme or an implicit scheme for solution.
Explicit scheme
• The unknown values are solved sequentially along a
time line from one distance point to the next
• Simpler method
• unstable
• Results are given at the grid points
• It is less efficient than the implicit method and hence
not suitable for routing flood flows over a long time
period.
 Numerical schemes contd….
• Example: Determine the unknown quantity at a point
𝑗+1
𝑖, 𝑗 + 1 at time 𝑡 + ∆𝑡 i.e. 𝑢𝑖 by using explicit
scheme
• Assume the hydraulic quantities u at time t
𝑡𝑖𝑚𝑒 𝑙𝑖𝑛𝑒 𝑗 are known
• Solution
• For the time derivative


 For spatial derivative

Note: A forward-difference scheme is used for the time derivative and a central
difference scheme is used for the spatial derivative.
Numerical schemes contd….
• The numerical stability of the computation depends
on the relative grid size on the x-t plane.
• A necessary but insufficient condition for stability
of an explicit scheme is the Courant condition
(Courant and Friedrichs, 1948).
• For the kinematic wave equations, the Courant
condition is

• where 𝑐𝑘 is the kinematic wave celerity.

• For the dynamic wave equations, 𝑐𝑘 is replaced by
𝑉 + 𝑐𝑑
 Numerical schemes contd….
Implicit scheme
• The unknown values on a given time line are all
determined simultaneously
• mathematically more complicated
• By using computers this is not a serious problem once
the method is programmed.
• The method is stable for large computation steps with
little loss of accuracy
• It is much faster than the explicit method
• It can also handle channel geometry varying
significantly from one channel cross section to the next
 Numerical schemes contd….
• Example: The space and time derivatives can be written
for the unknown quantity at a point 𝑖 + 1, 𝑗 + 1 by
using implicit method as
• For space derivative

For time derivative