Hydraulics Prof. B.S.

Murty

40.1 Method of Characteristics

Governing equations for gradually varied unsteady flow in open channel are given by

∂y ⎛ A ⎞ ∂V ∂y
+⎜ ⎟ +V =0 ( 40.1)
∂t ⎝ T ⎠ ∂t ∂x
∂V ∂V ∂y
+V +g = g ( S0 − S f ) ( 40.2 )
∂t ∂x ∂x

Equations (40.1) and (40.2) represent the continuity and dynamic equation, in non-

conservation form, respectively. Here, A = Cross-sectional area, T = top width, y = flow

depth, V = Velocity, g = acceleration due to gravity, S0= bed slope, Sf = friction slope, x

= distance and t = time. These are a set of coupled non-linear partial differential

equations of hyperbolic type. There are no analytical solutions for these equations , and

they are usually solved by numerical methods. Here in we explain the "method of

characteristics" for solving these governing equations. Although Method of

characteristics is seldom used these days for solving the unsteady open-channel flow

equations, this concept helps one to understand (i) the propagation of waves and (ii) the

formulation of boundary conditions. Formulation of boundary conditions is very crucial to

correctly solving the governing equations.

40.1.1 Characteristic Equations

Multiply Eq. (40.1) by an arbitrary parameter λ , add it to Eq. (40.2), and rearrange the

terms to obtain

⎡ ∂V ⎛ λ A ⎞ ∂V ⎤ ⎡ ∂y ⎛ g ⎞ ∂y ⎤
⎢ ∂t + ⎜ V + T ⎟ ∂x ⎥ + λ ⎢ ∂t + ⎜ λ + V ⎟ ∂x ⎥ = g ( S0 − S f ) ( 40.3)
⎣ ⎝ ⎠ ⎦ ⎣ ⎝ ⎠ ⎦

The total derivatives for V and y can be written as

dV ∂V ∂V dx
= + ( 40.4 )
dt ∂t ∂x dt

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

and

dy ∂y ∂y dx
= + ( 40.5)
dt ∂t ∂x dt

Now, Choose λ such that

λA g dx
V+ = +V = ( 40.6 )
T λ dt

Then, Eq. (40.3) becomes

= g ( S0 − S f )
dV dy
+λ ( 40.7 )
dt dt

and
gT
λ=± ( 40.8)
A

Therefore,

g gA
=± ( 40.9 )
λ T

g dx
It can be seen that represents the celerity and represents the absolute wave
λ dt

velocity. Thus

= g ( S0 − S f )
dV g dy
+ ( 40.10 )
dt C dt

is valid along the curve.

dx
=V +C ( 40.11)
dt

While

= g ( S0 − S f )
dV g dy
− ( 40.12 )
dt C dt

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

dx
=V −C ( 40.13)
dt

is valid along the curve. Equation (40.10) is called the Positive Characteristic Equation

and Eq. (40.12) is called the Negative Characteristic Equation. Plots of Eqs. (40.11) and

(40.13) in the x-t plane are shown in the Figure below. In this figure, line LP.

C+ C-
t
L R

xL xR

Positive and Negative Characteristic Curves

represents the Eq. (40.11) and is known as the Positive Characteristic Line, C + .

Equation (40.10) is valid along this line. It may be noted that discretized form of original

partial differential equations can be written between any pair of points, arbitrarily taken

in the x-t plane. On the other hand, Eq. (40.10), which is simpler than original P.D.E s

can be written between any two points in the x-t plane, L and P, only when they satisfy

the condition given by Eq. (40.10). Points L and P indicate that a wave travelling in the

downstream direction takes ∆t time to travel a distance ∆xL . In a similar manner, Eq.

(40.13) plots as line RP, C − in the x-t plane. This line is known as the negative

characteristic line Eq. (40.12) is valid along the line RP. Points R and P indicate that a

wave travelling in the upstream direction takes ∆t time to travel a distance of ∆xR .

Equation (40.10) is integrated along the characteristic line LP as shown below

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

⎛g⎞
P P P

∫L dV + ∫L ⎜⎝ C ⎟⎠ dy =g ∫L ( S0 − S f ) dt ( 40.14 )

Note that the celerity, C depends upon the flow cross-sectional area, A and top width, T,

both of which are functions of time and distance. Therefore, C is not a constant along

the line LP in the x-t plane. In a similar manner, Sf is also not a constant along LP.

However, for the sake of simplification, assuming that C and Sf are constant along the

line LP, and are equal to values of C and Sf at point L, Eq. (40.14) can be shown to give

⎛g⎞
VP − VL + ⎜ ⎟ ⎡⎣ y p − yL ⎤⎦ = g ⎡ S0 − ( S f ) ⎤ ⎡⎣t p − t L ⎤⎦ ( 40.15 )
⎝ C ⎠L ⎣ L⎦

Similarly, Eq. (40.12) can be integrated along the line RP. The result is

⎛g⎞
VP − VR − ⎜ ⎟ ⎡⎣ y p − yR ⎤⎦ = g ⎡ S0 − ( S f ) ⎤ ⎡⎣t p − t R ⎤⎦ ( 40.16 )
⎝ C ⎠R ⎣ R⎦

In Eqs. (40.15) and (40.16), subscripts P, R and L refer to the values at the points L, P

and R, respectively. If the values V, y etc. are known at points L and R, values of V and

y at point P can be obtained by simultaneously solving Eqs. (40.15) and (40.16). By

taking points L and R at the same time level, i.e., t L = t R and

t p − t L = t p − t R = ∆t ,
⎛g⎞ ⎛g⎞
⎜ ⎟ = CL and ⎜ ⎟ = CR
⎝ C ⎠L ⎝ C ⎠R
(
VP + CL y p = VL + CL yL + g S0 − S f L ∆t ) ( 40.17 )
VP − CR y p = VR − CR yR + g (S 0 − Sf R ) ∆t ( 40.18)

Right hand sides of Eqs. (40.17) and (40.18) can be evaluated using the values at

points R and L. Therefore, Vp and yp can be obtained by solving Eqs. (40.17) and

(40.18).

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

40.1.2 Characteristic lines

The physical significance of characteristic lines is explained here. Consider the point P

located at x = xp in the x-t plane as shown in the figure below

NCL R PCL

tQ Q
x
P
x = xp x = xq

A disturbance created at P travels in both upstream and downstream directions.

Disturbance moving in the upstream direction is indicated by the line NCL. Wave

moving in the downstream direction moves along the line PCL in the figure. Any point R,

located in the shaded region of the figure is affected by the disturbance at P. A point Q

located outside the shaded region is not affected by the disturbance at P. What this

means is, arrival time for the wave from P(x = xp) to Q (x = xq) is more than tQ, or

conditions at x = xq would not have changed at the time tQ due to a wave from P.

The shaded region is referred to as the "zone of influence". Now, consider the point p in

figure shown below.

PCL NCL
R

x
L R

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

The positive characteristic line (PCL) and the negative characteristic line (NCL) passing

through the point P are shown in this figure. These lines are drawn backwards in time.

Here, the shaded region is referred to as the "zone of dependence". What this means is,

any disturbance orginating at point R, located within the shaded region, would affect the

conditions at P.

Subcritical Flow

For a subcritical flow, the PCL lies to the right of point P (in the x-t plane), while the NCL

lies to the left of point P as shown in the figure below. This means the disturbance

travels on both upstream and downstream directions.

NCL PCL

x
P

Critical Flow
For a critical flow, the PCL lies to the right of point P while the NCL is a line

perpendicular to the x-axis as shown in the figure below

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

NCL
PCL

x
P
Critical Flow

Supercritical Flow

For a supercritical flow, both the PCL and NCL lie to the right of point P in the x-t plane

as shown in the figure below.

NCL

PCL

x
P
Supercritical Flow

This means that the disturbances can travel only in the downstream direction, if the flow

is supercritical. That is why it is often said that the supercritical flow knows only what

happens on the upstream side. Disturbance created at a downstream location from

point P gets swept further downstream. They cannot travel upstream and reach point P.

The above discussion on physical significance of characteristic lines can be also used

to specify the initial and boundary conditions.

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

40.1.3 Initial and Boundary Conditions

Numerical solution of governing partial differential equations (Eqs. (40.1) and (40.2)

start with calculations at a specified time. At this time, the flow conditions i.e. the flow

depth, y and the flow velocity, V should be known. These flow conditions at the initial

time are termed as the "initial condtions" (refer to figure below).

upstream Downstream
boundary boundary

P
R
NCL NCL Q
t= t PCL
PCL

x
x=0 x=L
xR Initial Condtions

Initial and boundary Condtions - Subcritical Flow

Any physical system has finite boundaries. For a one-dimensional flow, there will be a

boundary at x=0 and another boundary at some distance x=L.

The boundary at x=0 is termed as upstream boundary and the boundary at x=L is

termed as the downstream boundary. In the solution of governing equations, one has to

specify one or more conditions at the boundaries of the system. These are known as

the "boundary conditions".

Consider a point R in the interior of domain as shown in the figure above. This point

indicates that we are interested in finding the flow conditions at a point x = xR at time

t = ∆t . Two characteristic lines, PCL and NCL, passing through this point (for subcritical

flow) carry the information from t=0 to t = ∆t . There are two flows variables, y and V,

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

whose values at x = x R and t = ∆t need to be determined. Existence of two

characteristic lines at this point R in the x-t plane indicates that two characteristic

equations will be available for this purpose. Now, consider a point P as shown in the

figure. Point P indicates that values of y and V, at the upstream end, at time t = ∆t ,

need to be determined. No positive characteristic line, which originates from time t = 0

and from within the domain, is passing through this point P because P falls on the

upstream boundary. Thus only NCL is existent for this point. This indicates that only one

equation (negative characteristic equation) is available for solving for two unknowns at

point P. Therefore, a boundary condition needs to be specified for closing the problem.

This boundary condtion can be a specified flow depth or a specified flow velocity or a

relationship between the depth and the velocity. Now, consider a point Q on the

downstream boundary as shown in the figure. Point Q indicates that we are interested in

determining the values of y and v at x=L, at time t = ∆t . As in the case of point P, only

one characteristic line (positive characteristic line, PCL) is existent here. Therefore, one

needs to specify one boundary condition at the downstream boundary.

Above discussion is valid for the case of subcritical flow. The characteristic lines in the

case of supercritical flow are shown in the figure below. As can be seen from this figure,

no characteristic equations will be available for solving for the two flow variables at point

P on the uspstream boundary.

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

upstream Downstream
boundary boundary
R Q
P
NCL NCL

PCL PCL
t=0 x
x=0 x=L
Boundary Condtions - Supercritical Flow
Therefore, two boundary conditions need to be specified at the upstream boundary, in

case of supercritical flow. On the otherhand, two characteristic equations will be

available for solving for the two unknown flow variables at point Q on the downstream

boundary. Therefore, boundary conditions should not be specified at the downstream

boundary if the flow is supercritical there.

40.1.4 Riemann Invariants

Consider the Eqs. (40.10) and (40.12). Consider a rectangular channel. The wave

celerity in such a case is given by

C = gy (40.19)
or
dc g dy
=
dt 2 y dt
or
dc 2 y dc
= (40.20)
dt g dt

Indian Institute of Technology Madras

Hydraulics Prof. B.S. Murty

Substitution of Eqs (40.19) and (40.20) in Eqs. (40.10) and (40.12) results in the

following.

dv
dt
+2
dc
dt
(
= g S0 − S f ) (40.21)

and
dv
dt
−2
dc
dt
(
= g S0 − S f ) (40.22)

or
d
dt
(
(V + 2 c ) = g S0 − S f ) (40.23)

and
d
dt
(
(V − 2 c ) = g S0 − S f ) (40.24)

In the case of horizontal and frictionless channels,

d
(V + 2c ) = 0 ( 40.25 )
dt
d
and (V − 2c ) = 0 ( 40.26 )
dt
or V + 2 c=J+ =Constant along PCL
and V − 2 c=J - =Constant along NCL

The constants J+ and J-are called "Riemann invariants" since they remain constant

along the characteristic lines. An important point to be noted with regard to the

boundary conditions is that they should be independent of the governing equations and

of the Riemann invariants. For example, one should not specify the value of V-2C as an

upstream boundary condtion. Similarly, value of V+2C cannot be specified as a

downstream boundary condtion.

Indian Institute of Technology Madras