Iran-Water Resources

Research

1388 2
Volume 5, No. 2, Fall 2009 (IR-WRR)

15-17

Extended Abstract

Optimization of Multireservoir Water Resources Systems Operation Using Genetic Algorithm

D
I

S. M. Ghadami 1*, B. Ghahraman 2, M. B. Sharifi3 and H. Rajabi Mashhadi 4

Introduction
Due to the variation of rainfall and inflow regimes in
different years, it is necessary to adopt a suitable
management model for reservoir operation. Water
requirement may differ at different demand sites and
therefore it can be satisfied with one reservoir or a
network of reservoirs. Different methodologies are
adopted for optimal operation of a system of reservoirs.
These may be classified in different groups; Stochastic
Dynamic Programming (Braga et al., 1991, Ghahraman
and Sepaskhah, 2002), Fuzzy Rule Based Programming
(Faye et al., 1991), Linear Programming (Mohan and
Raipure, 1992) Deterministic Dynamic ProgrammingRegression Based (Karamouz et al., 1992), Mixed
Integer Non-Linear Programming (Teegavarapu and
Simonovic, 2000), and Dynamic Programming And
Artificial Neural Network (Chnddramouli and Raman,
2001),. Since nearly all dams in Iran regulate water for
agricultural uses, a wise discipline should be adopted
for best allocation of limited water (especially in
droughts) considering the uncertain future river yield.

o
e

iv

h
c

r
A

Objectives
In this study, a deterministic Genetic Algorithm model
was adopted for optimal operation of a multireservoir
water resource system in the northern part of Khorasan,
northeastern Iran.
Methodology
Figure 1 schematically shows the system of the
reservoirs under study.
1- Former Graduate Student of Water Engineering, Civil Engineering
Department, Ferdowsi University of Mashhad
2- Associate Professor of Irrigation, Ferdowsi University of Mashhad,
bijangh@ferdowsi.um.ac.ir
3- Assistant Professor of Civil Engineering, Ferdowsi University of Mashhad
4- Associate Professor of Electrical Engineering, Ferdowsi University of
Mashhad
*- Corresponding Author

S
f

This system consists of two reservoirs in series and the

accompanied agricultural lands. According to the
cropping patterns, the year is divided into dormant and
non-dormant (active) seasons of different lengths. The
proposed cropping pattern is winter wheat (27 and 18%
for fields 1 and 2, respectively), winter barley (30 and
26% for fields and 2, respectively), and sorghum (43
and 56%). Wheat is sowed in late October, barly in mid
October, and sorghum in the first week of June.

The optimization objective function is as follows:

Max c A c [Bc ,k (Ya c ,k Yp c ,k ) Cc ,k ]

f

(1)

k =1

in which f is the total number of agricultural fields, Ac,k

is the cultivated area in the kth field (ha), Bc,k and Cc,k
are benefit and cost for the cth crop in the kth field,
respectively (Rls/ha), and Yac,k and Ypc,k are actual and
potential yield for cth crop in kth field, respectively
(kg/ha). Relative yield (Ya/Yp) is a non-linear function
of actual evapo-transpiration (ETa) (Rao et al., 1998).
The soil-water balance for every crop in every field and
the reservoirs water mass balance are the main
constraints for equation 1.
Four penalty functions were considered. The first
penalty was due to the decreased soil moisture below
the soil permanent wilting point (PWP). The second
penalty was attributed to the ETa exceeding the
potential evapo-transpiration (ETp). The third penalty
was put to the Eta/Etp ratios over 0.5, guarantying the
mild water stress for crops. Finally, the fourth penalty
was due to the fact that the storage of water in the
reservoir must always remain over the dead storage.

15 94

www.SID.ir

logical combination of state variables (volume of

available water at the beginning of the non-dormant
season and the river flow and the rainfall for each time
period).
Results and Discussions
Twelve different scenarios were considered and the
performance of the model was traced accordingly
(Table 1). The effect of the dry regime is more
pronounced for the crop relative yields in Lotfabad
farm. This is due to the fact that (a) Shorkal reservoir is
smaller that the other reservoir, (b) the yield of the river
charging the second reservoir is highly uncertain, and
(c) acreage of this farm is larger than the Dargaz farm
(3780 vs 1770 ha).

D
I

Figure 1- Schematic representation of the

multireservoir system under study, located in
northeastern Iran.

The non-dormant season were divided into 27 ten-day

irrigation intervals. The decision variable was the
actual evapo-transpiration of crop c at time t. Five
genes were assigned to each decision variable. So for
each of the 30 chromosomes, 810 genes were needed.
A racing method was used for selecting the next
generation of chromosomes. Values of 0.7 and 0.006
were considered for the cross-over and the probability
of mutation, respectively. The model is run for every

Conclusion
The developed model is capable of considering local
conditions such as soil properties, cropping pattern,
irrigation efficiency, and root lengthening algorithm.
The model outputs encompas different aspects of the
mutual soil-water-crop relationships including soil
moisture,
deep
percolation,
and
actual
evapotranspiration. It also grasped the attributes of the
reservoirs, such as reservoir release and the evaporation
and spill losses. Under the dry regime, total benefit is
sensitive to the available water in the reservoir at the
beginning of the season. This is not the case under the
other regimes. These results are highly local and every
change in the input variables, may change the results
drastically.

S
f

o
e

v
i
h

Table 1- Relative yield of crops and relative benefit at differnnt combinations of state variables for
consecutive dams in north of Razavi province, Iran.

Case

Relative
volume of
water in the
reservoir at
the
beginning of
the season

1
2
3
4
5
6
7
8
9
10
11
12

0.25
0.25
0.25
0.5
0.5
0.5
0.75
0.75
0.75
1
1
1

c
r

Field 1

Field 2

Rainfall
class

Riverflow
class

Relative
yield of
wheat

Relative
yield of
barley

Relative
yield of
sorgom

Relative
yield of
wheat

Relative
yield of
barley

Relative
yield of
sorgom

Relative
benefit

dry
normal
wet
dry
normal
wet
dry
normal
wet
dry
normal
wet

min
mean
max
min
mean
max
min
mean
max
min
mean
max

0.94
0.92
0.92
0.92
0.93
0.92
0.93
0.92
0.96
0.92
0.94
0.96

0.86
0.89
0.91
0.88
0.91
0.89
0.91
0.89
0.94
0.94
0.96
0.89

0.92
0.92
0.88
0.94
0.91
0.92
0.9
0.92
0.96
0.94
0.92
0.93

0.67
0.95
0.95
0.87
0.96
0.95
0.92
0.95
0.9
0.78
0.91
0.95

0.59
0.96
0.93
0.8
0.93
0.96
0.96
0.96
0.94
0.93
0.94
0.97

0.6
0.95
0.96
0.78
0.95
0.95
0.78
0.95
0.96
0.79
0.95
0.96

0..6374
0.9263
0.9243
0.8007
0.9249
0.9263
0.8191
0.9263
0.94
0.8197
0.9298
0.9415

1388 2
Volume 5, No. 2, Fall 2009 (IR-WRR)

16 93

www.SID.ir

Ghahraman, B. and Sepaskhah, A.R. (2002), Optimal

allocation of water irrigation a single purpose
reservoir to an irrigation project with predetermined multiple cropping patterns, Irrigation
Science, 21: pp.27 -137.

Keywords: Optimization, Multireservoir Systems Operation,

Genetic Algorithm, Water stress, Optimal water allocation.
References
Braga, B.P.F., Yeh, W.W-G, Becker, L. and Barros,
M.T.L., (1991), Stochastic optimization of
multiple-reservoir-system operation, Journal of
Water Resources Planning and Management,
ASCE, 117(4): pp. 471-481.

Karamouz, M., Houck, M.H. and Delleur, J.W. (1992),

Optimization and simulation of multiple reservoir
system, Journal of Water Resources Planning and
Management, ASCE, 118(1): pp. 71-81.

Chandramouli, V. and Raman, H., (2001),

Multireservoir
modeling
with
dynamic
programming and neural networks, Journal of
Water Resources Planning and Management,
ASCE, 127(2): pp. 89-98.

Mohan, S. and Rapure, D.M. (1992), Multiobjective

analysis of multireservoir system, Journal of
Water Resources Planning and Management,
ASCE, 118(4): pp. 356-370.

Faye, R.M., Sawadogo, S., Lishou, C., Mora-Camino,

F., (2003), Long-term fuzzy management of water
resource systems, Applied Mathematics and
Computation, 137: pp. 459-475.

Teegavarapu, R.S.V. and Simonovic, S.P. (2000),

Short-term operation model for coupled
hydropower reservoirs, Journal of Water
Resources Planning and Management, ASCE,
126(2): pp. 98-106.

D
I

S
f

o
e

v
i
h

c
r

1388 2
Volume 5, No. 2, Fall 2009 (IR-WRR)

17 92

www.SID.ir