Вы находитесь на странице: 1из 35

UNSTEADY OPEN-CHANNEL FLOW B:

RIVER ROUTING
RIVER MECHANICS (OPEN-CHANNEL HYDRAULICS) (CE5312 AY15/16)

Dr. Yuan Jing,


Department of Civil and Environmental Engineering
Office: E2-05-20
Phone: 65162160
Email: ceeyuan@nus.edu.sg
Flood in Singapore
Jun.16.2010 Orchard road
4 inches of rain fell during just two
hours

Extreme rainfall produced this flood.


How do we deal with flood?

Retention and drainage (reservoirs and canals)


River routing

• How does a reservoir/pond


reduce the peak of a flood?
Solve for O(t), S(t)

• How does flood wave travel


through a channel?
Solve for Q(x,t), h(x,t)
Principle for river routing
Q A
Continuity equation:  0
x t

U U h
Momentum equation: U g  g ( S0  S f )
t x x

Hydrologic routing: just use the continuity equation, but use some
empirical as an alternative to momentum equation

Hydraulic routing: simultaneously solve two equations, but some


simplifications may be applied to momentum equation.
Reservoir routing: level-pool method
dS
I O 
dt

Given I, find S(t), O(t)

I1  I 2 O1  O2 S2  S1
 
2 2 t

I1  I 2 S O
N 2  N1   O1 N 
2 t 2

Need O=f(N)!
N-O relationship

O  f 2 (h), e.g . weir


S  f1 (h)

S O O

S ( h) O ( h) O  f (N )
S  f1 (h) O  f 2 ( h) N ( h)  
t 2

h h N

O-N can be obtained from known S-h and O-h relationships


General procedure for level-pool method
Given conditions:
 Inflow hydrograph S O
 An S-h relationship from topography information N 
t 2
 An O-h relationship (previous flood record or simple formula)

O
Initial flow condition (O1,N1, I1 at t=0)

O  f (N )
Steps:
 Choose Δt and discretize the time domain
 Develop N-O relationship from S-h and O-h relationships
 Compute Ni+1
 Use the established N-O relationship to obtain Oi+1 N

 Use obtain O and N to get S: N=S/Δt+O/2 I i  I i 1


N i  N i 1   Oi 1
2

Oi 1  f ( N i 1 )
Example: level-pool method
Route the following inflow hydrograph through a reservoir created by a
weir.

Storage above crest line:

H S  6.5  105 H [m3 ]

O  15 H 3 [m3 / s ]

122m
I

Choose Δt=2 hour for simplicity


N-O relationship
S  6.5  105 H [m3 ]
S O
N  O  15H 3 [m3 / s ]
t 2
t  2 hour

250

Polynomial fitting
200
O  2.3  106 N 3  3.2  103 N 2  0.1336 N  0.462

150
O [m3/s]

100

50

* In this example, both S and O are simple functions 0


of H, so S-H and O-H are easy to establish. For real 0 50 100 150 200 250 300 350
M=S/  t+O/2 [m /s]
3
cases, you need measurements to get S-H and O-H
Prediction of O

250

Inflow
Outflow
200

I I 150

N i  N i 1  i i 1  Oi 1
Q [m3/s]

2 100

Oi  f ( N i )
50

Flood peak is reduced due to reservoir


storage. 0
0 2 4 6 8 10 12 14 16 18 20
t [hours]
Muskingum Method Advancing Flood Wave I > Q

Wedge storage in reach I


Q
S Prism  KQ
I Q
S Wedge  KX ( I  Q)
Q Q
• K = travel time of peak through the
reach
• X = weight on inflow versus outflow (0
≤ X ≤ 0.5) Receding Flood Wave Q > I
• X over 0.5 increases flood peak
(unrealistic) I Q
• X = 0  level-pool, storage depends
on outflow, no wedge
• X = 0.0 - 0.3  Natural stream
QI
S  KQ  KX ( I  Q) I I
S  K[ XI  (1  X )Q]
Obtaining routing parameter use past flood record

S  K[ XI  (1  X )Q] K
S
XI  (1  X )O

• Use past flood record (S, I and O).


• Try different value of X. With X=X2, the straight-line fit works best.
• X2 is the value for X, and the corresponding slope of the fit is K.

X  X1 X  X2 X  X3
S S S

X 1 I  (1  X 1 )O X 2 I  (1  X 2 )O X 3 I  (1  X 3 )O
Example: Muskingum method’s routing parameters

t
Si 1  Si  ( I i  I i 1  Oi  Oi 1 )
2

X=0.2 X=0.3 X=0.4


20 20 20

15 15 15
S [m3/s*day]

S [m3/s*day]

S [m3/s*day]
10 10 10

5 5 5

K=0.8 [day]
0 0 0
0 20 40 0 20 40 0 20 40
3 3 3
XI+(1-X)O [m /s] XI+(1-X)O [m /s] XI+(1-X)O [m /s]
Some algebra
dS I1  I 2 O1  O2 S2  S1
I O   
dt 2 2 t
S  KQ  KX ( I  Q)

I1  I 2 O1  O2 K [ XI 2  (1  X )O2 ]  K [ XI1  (1  X )O1 ]


 
2 2 t

Basic equation
for Muskingum O2  C0 I 2  C1I1  C2O1
routing
t  2 KX
C0 
t  2 K (1  X ) • 2KX < Δt to ensure C0>1
t  2 KX
C1  • C0+C1+C2 = 1
t  2 K (1  X )
2 K (1  X )  t • X=0.5 and K= Δt: Oi+1=Ii
C2 
t  2 K (1  X )
General procedure for Muskingum routing
Given:
• Inflow hydrograph: I(t)
• routing parameter K and X
• initial condition (I, Q, S)
t  2 KX
C0 
t  2 K (1  X )
Steps: t  2 KX
C1 
t  2 K (1  X )
• chose a time interval Δt 2 K (1  X )  t
C2 
t  2 K (1  X )
• Calculate the routing constants
• Routing for O O2  C0 I 2  C1I1  C2O1
• Use obtain O and given I to get storage S
t
Si 1  Si  ( I i  I i 1  Oi  Oi 1 )
2
Example: Muskingum routing
For a channel reach the routing parameters are K=0.8 [day] and X=0.3. Determine the
outflow hydrograph with the following inflow hydrograph.

Choose Δt=1 day for simplicity

t  2 KX 1  2  0.8  0.3
C0    0.245
t  2 K (1  X ) 1  2  0.8  (1  0.3)
t  2 KX 1  2  0.8  0.3
C1    0.698
t  2 K (1  X ) 1  2  0.8  (1  0.3)
2 K (1  X )  t 2  0.8(1  0.3)  1
C2    0.057
t  2 K (1  X ) 1  2  0.8  (1  0.3)
Example: Muskingum routing (cont.’)
Oi 1  C0 I i 1  C1 I i  C2Oi
35
inflow
outflow
30

25

Q [m3/s] 20

15

10

5
0 2 4 6 8 10 12
t [day]

Flood peak is attenuated.


Momentum equation
U U h
U g  gS0  gS f
t x x
pressure gravity friction

inertial

Typical values:

S0  4.9  103
h Very possible that we can
 9.5  105
x neglect inertial and pressure
U U terms
 2.4  105 ~ 4.7  105
g x
1 U
 9.5  105
g t
thx ThLU
U ˆx0ˆUˆ
000th

Another way to look at this


Normalize momentum equation

U 0  Uˆ ˆ Uˆ  h0 hˆ Sf


 ˆ  U    1  0
gS0T0  t xˆ  U 0T0 S0 xˆ S0

We can neglect the pressure and initial terms if:

U 0T0 S0 gS0T0
1   1 2   1
h0 U0
S0  0.001
h0  10 m 1  17.2  1
U 0  1 m/s  2  432  1
T=1 day=86400 s
Simplification of momentum equation

• Kinematic wave: gravity and friction


• Diffusion wave: gravity, friction and pressure
• Dynamic wave: gravity, friction, pressure and inertial
Kinematic-wave routing: Governing
equation
 A  C Rh S0 Chezy

S0  S f Q   1 2/3
 A  Rh S0 Manning
 n

In general:
Q  D ( h) S 0 Q

Ant any point and any time, the flow is “quasi-steady”,


so the stage-discharge relationship of steady uniform
flow can be applied.

Q A
S0
The unsteadiness comes
 0
from the continuity equation x t
Kinematic-wave equation
Q Q Q
 ck 0 ck  0 Klietz-Seddon law
t x A

DQ dx
 0, along  ck
Dt dt

Kinematic wave moves


downstream without changing
shape, if ck is constant (pure
translation)
Celerity of kinematic wave
For a wide channel:

   3 / 2 Chezy
ck   U 
  5 / 3 Manning

Celerity increases with velocity:

steepen Kinematic shock

As kinematic waves moves downstream, the peak moves faster than the front and
tail, so the wave becomes steepen and eventually becomes a kinematic shock
Kinematic wave vs. Muskingum
Q Q X ( I 2  I1 )  (1  X )(O2  O1 ) (O  I )  (O2  I 2 )
 ck 0  ck 1 1 0
t x t 2x

I1  I 2 O1  O2  x / ck [ XI 2  (1  X )O2 ]   x / ck [ XI1  (1  X )O1 ]


 
2 2 t

Muskingum: I1  I 2 O1  O2 K [ XI 2  (1  X )O2 ]  K [ XI1  (1  X )O1 ]


 
2 2 t

• Muskingum is just a finite


x difference representation of
K kinematic-wave equation
ck
• K is the time for kinematic wave
to pass the channel reach Δx
Kinematic wave vs. Lagrangian wave
Kinematic wave ck   U

Only moves downstream


x

Lagrangian wave ck  U  gh

Can move in both upstream


x and downstream direction

• Kinematic wave is a HUGE wave that gravity and friction are important
• Lagrangian wave is a SMALL wave that inertial and pressure are important
• They are idealized “waves” based on different simplification of momentum
equation.
• A real dynamic wave should have their features.
Kinematic-wave routing: applicability
Recall:
U 0  Uˆ ˆ Uˆ  h0 hˆ Sf
 ˆ  U    1  0
gS0T0  t xˆ  U 0T0 S0 xˆ S0

Ponce (1989) gave the following criteria for applying kinematic wave:

T0U 0 S0
 85
h0
Diffusion-wave routing: governing equation

h
S f  S0 
x
Q2
Sf  Q Q  2Q
D  cf  2
t x x
  f P(h)  1
3 
  Darcy  Weisbach  Q dD
  8 g A(h)  cf  Celerity of diffusion
 1
D dA wave
  1 P ( h) 
3 
D ( h)    2 Chezy’s Equation 
  C A ( h )  D2 Diffusion parameter

 4/3 1


 n 2 P ( h )
 Manning’s Equation  2bs Q
  A(h)10/3 

Behavior of diffusion wave
Q Q  2Q
 cf  2
t x x

x
As a diffusion wave moves downstream, its peak is attenuated, and the
wave becomes longer (spread out)
Diffusion-wave routing: celerity

cf Q dD Q A dD A dD  h  A dD 1 A dD
 /( )     b 
D dA D dh  A  D dh
s
U D dA A bs D dh

ck A D

U bs D h c f  ck

Diffusion-wave celerity is not necessarily equal to the kinematic-wave celerity,


because the flow velocity U in this two routing method are based on different
momentum equations. However, the difference is negligible in most cases, so cf=ck.
Diffusion-wave routing: the diffusivity

D2 Q
 
2bs Q 2bs S0

The diffusivity μ apparently increases with Q, but decreases with S0 and bs,
meaning that diffusion will be more severe if the slope is close to horizontal, or
the river is very narrow, or the discharge is very large.
We have see the diffusion feature before!
35
inflow
outflow
30

25

Q [m3/s]
20

15

10

5
0 2 4 6 8 10 12
t [day]

We see the diffusion feature with Muskingum method!

But Muskingum method is proved to be a finite difference representation of


kinematic wave, which should not have any diffusion.

Why?
Numerical diffusion
Q Q
 ck 0
t x
Weighted finite differencing

Q X (Qi  Qi )  (1  X )  Qi 1  Qi 1  0.5(Qik1  Qik )  0.5  Qik11  Qik 1 


k 1 k k 1 k
Q
 ck   ck  error
t x t x
 1   Q
2
Truncation error: error  ck x   X   2
(Taylor expansion)
 2   x

X (Qik 1  Qik )  (1  X )  Qik11  Qik1  0.5(Qik1  Qik )  0.5  Qik11  Qik 1 


 ck 0
t x
The discretized equation is equivalent to the following equation

Q Q  1   Q
2
 ck  ck x   X   2  0
t x  2   x
Numerical diffusion (cont.’)
Q Q  1   Q
2
 ck  ck x   X   2  0
t x  2   x

Q Q  1   Q
2
Compare:
Q Q  2Q
 ck  ck x   X   2  ck  2
t x  2   x t x x

1 
  ck   X  x
2 

By choosing X properly, the numerical diffusion is equivalent


to the actual diffusion.

X cannot be over ½, otherwise μ<0, meaning the flood will


increase in amplitude
Muskingum-Cunge routing
A Muskingum method is equivalent to kinematic wave method if:

x
K
ck

A Muskingum method can represent the diffusion wave if:

1  1 D2 
  ck   X  x X  1  
2  2  xck bs Q 

Muskingum method’s routing parameter K and X can


now be analytically determined (Muskingum-Cunge
method)

Вам также может понравиться