You are on page 1of 6

Hydrology I

Jozsef Szilagyi, Professor of Hydrology


Department of Hydraulic and Water
Resources Engineering
Budapest University of Technology and
Economics

Lecture #10: Flood routing and the loop rating curve

A
P

Typical range for n: 0.01 - 0.1

(1)

Enough to know the IR function.


What is the IR(=IUH) of the diffusion wave?
h(x,t)=x / (2[D]0.5t1.5) exp[-(ct-x)2/4Dt], where x is distance from the upstream crosssection.
So, if one wants to predict the flood values, Q d(t), at a downstream location x-distance away from the upstream
discharge values (Qu):
which is the well-known convolution integral. Note that the Q d and Qu values in the above equation are values
relative to the last non-flood value, Q min. So Qmin has to be added to Qd for obtaining the total flowrate downstream.

In practice one generally deals with discrete data. With


discrete data (within the so-called pulse data framework)
one needs the unit-pulse response, UPR, function:
UPR = u(t) = g(t) g(t-t), where g is the unit-step
response function. What is g(t) for the diffusion wave?

erfc(x)
1

The second term in g(t) vanishes fast for large x, t, or c values.


By the Manning-Strickler formula Eq. (1) can be written as

x
where K is called conveyance. Since A and R are functions of z,
it can be seen that for the same z, different Q values may
z
belong, depending on the z/x term. This predicts a loop
Falling limb
rating curve.
z/x > 0
When z/x > 0 Q < Q| z/x =0
z/x < 0 Q > Q| z/x =0
all at a given z.

z fixed

z/x < 0
Rising limb of hydrograph

Q
R

Q
F

In hydrological flow routing the loop rating curve is replaced by the function
of Q=Q(z). This happens with routing techniques that rely on a first order
approximation of the St-Venant equations momentum eq., i.e., Sf = S0. This
approach is called the kinematic wave approach. One of the most widely
used such approaches is the linear cascade model where the stream-reach
is divided into a series of uniform linear reservoirs and water is routed
through these linear elements. It can be shown that the discrete version
(i.e., when using discretized data) of the model yields the same result as the
discrete diffusion wave model we just discussed. Flow routing techniques
are justified to be used on gently sloping rivers, and with gently rising flood
waves. They cannot account for so-called backwater effects, i.e., when
QQ(z).

Lab #8: Calculate the downstream flowrate at x=3600 m from the upstream
location if c=2 m/s, D=1;40;80 m2/s, t=60 s. The inflow values are: Qin =
t+10, when t=0,, 300 s; and 610-t, when t=300,,600 s. Explain what
happens as D is changing.
Use the following MATLAB script:

clear
c=2;
dk(1)=1;
dk(2)=40;
dk(3)=80;
x=3600;
dt=60;
tdt=linspace(60,3600,60);
qdt1=linspace(10,310,6);
qdt2=320-qdt1(2:length(qdt1));
qdt=[qdt1,qdt2]
qbe=qdt-10;
%Subtraction of the initial q value!
for i=1:3
d=dk(i);
for i=1:length(tdt)
sr(i)=.5*erfc((x-c*tdt(i))/(2*sqrt(d*tdt(i))));
end
u(1)=sr(1);
for i=2:length(tdt)
u(i)=sr(i)-sr(i-1);
%This is the UPR!
end
qout=10+conv(qbe,u);
%Adding the initial q value to the convolution result!
plot(qdt)
hold on
plot(qout,'--')
end

You might also like