You are on page 1of 8

Available online at www.sciencedirect.

com

ScienceDirect
Procedia Engineering 154 (2016) 1393 – 1400

12th International Conference on Hydroinformatics, HIC 2016

Optimal design and operation of hydraulically coupled hydropower


reservoirs system
N. Afsharian Zadeha, S.J. Mousavia,*, E. Jahania, J.H. Kimb
a
School of Civil and Environmental Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran.
b
School of Civil, Environmental and Architectural Engineering, Korea University, Seoul, South Korea.

Abstract

This paper presents an optimization formulation for reliability-based optimal design and operation of a
hydraulically coupled cascade hydropower reservoirs system, where the upstream power plant’s tailwater is affected
by the downstream reservoir’s water level. The problem is formulated as a mixed integer nonlinear program
(MINLP) with the objective function of maximization of the system’s firm energy production while controlling the
reliability level of hydroenergy production. Particle swarm optimization (PSO) is used as the optimization algorithm
to solve the MINLP. The performance of the proposed methodology is tested through its application in the Karoon2-
Karoon3 cascade reservoirs system in Iran as a real-world case study. Karoon3 Dam has already been constructed,
whereas Karoon2 hydropower system is being designed. The results are used for capacity optimization of the
Karoon2’s reservoir and powerplant as well as optimal operation of the whole cascade system.
©©2016
2016TheTheAuthors. Published
Authors. by Elsevier
Published Ltd. Ltd.
by Elsevier This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of HIC 2016.
Peer-review under responsibility of the organizing committee of HIC 2016

Keywords: Optimization; Cascade hydropwer plants; PSO algorithm

* Corresponding author.
E-mail address: jmosavi@aut.ac.ir

1877-7058 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of HIC 2016
doi:10.1016/j.proeng.2016.07.509
1394 N. Afsharian Zadeh et al. / Procedia Engineering 154 (2016) 1393 – 1400

1. Introduction

Construction of dams to produce hydroelectric energy and waterfall from them can be a good strategy to protect
the environment in terms of renewable energy without using fossil fuels. In this regard, design and operation of
hydropower reservoirs and powerplants systems may be formulated in the form of mathematical programs
(optimization models), with the aim of determining design and operation parameters of the systems.
Optimization models have widely been used in hydropower systems operation (e.g. Gablinger & Loucks [1]; Kim
& Palmer [2]; Barros et al. [3]). Successive linear programming (SLP) (Yeh et al. [4]; Grygier & Stedinger [5]) and
sequential quadratic programming (SQP) (Powell [6]; Diaz & Fontane [7]) are among the optimization algorithm
used in hydroelectric systems analysis. Simonovic and Miloradov developed an optimization model minimizing
energy generation costs. The GEMSPL (Reznicek and Simonociv [8]), as an extension of EMSLP model (Martin
[9]), is a single-reservoir model for optimizing medium-term hydropower operations.
Meta-heuristic optimization techniques are becoming more attractive in solving optimization models. A meta-
heuristic optimization algorithm is a procedure to design a heuristic, stochastic search algorithm that could provide a
sufficiently good solution to an optimization problem. Mousavi et al. [9] employed PSO as a meta-heuristic
algorithm and used a simulation-optimization approach to determining the optimal capacity of a hydropower
reservoir system.
In cascade hydroelectric systems, one of the issues that is of value to be accounted for, is the hydraulic coupling
between the components of the system. Hydraulic coupling is assumed to exist when the forebay elevation of a
downstream plant influences the tailwater elevation of the upstream plant. Therefore, in addition to the plant
discharge, the powerplant’s tailwater elevation depends on the forebay elevation of the downstream plant.
Hawary and Christensen [10] presented an approach for scheduling of coupled hydropower reservoirs. Lyra and
Ferreira [11] used discrete differential dynamic programming in scheduling of a highly coupled system of
hydropower reservoirs. Simonovic et al. [12] developed a short-term operation model for optimal operation of
hydraulically coupled hydropower plants. They used binary variables in an MINLP to choose the tailwater elevation
curve in every time step among different curves, each corresponding to a specific water level of the downstream
reservoir.
This paper presents an optimization formulation for reliability-based optimal design and operation of the
Karoon2-Karoon3 hydraulically coupled cascade hydropower system in Iran. The problem is formulated as a
combinatorial nonlinear mathematical program. The PSO algorithm, a random search evolutionary algorithm, is
used to solve the problem. In the following, the PSO algorithm is presented first, and then it is explained how it will
be used to solve the problem of optimal design and operation of real case study. Subsequently, the results are
presented followed by a summary and conclusion section.

2. PSO algorithm

PSO algorithm is one of the meta-heuristic, population-based optimization methods originally introduced by
Kennedy and Eberhart [13] for solving optimization problems. It was inspired by the simulation of social interaction
observed among birds and fishes. In the PSO algorithm, each particle represents a solution in D-dimensional space,
and has a position vector ( x id ) and a velocity vector ( v id ). Particles move in the search space depending on two
important positions, Pi ( Pi1, Pi2 ,..., PiD ) , that is the best position the current particle has found so far (pbest), and
P g ( P g1 , P g 2 ,..., P gD ) , that is the global best position found in the entire population (gbest). In each iteration,
the velocity and position of each particles are updated according to Equations (1) and (2) respectively:

n  1 F( Z v n  c r n ( p n  x n )  c r n ( p n  x n )
v id (1)
id 1 1 id id 2 2 gd id

n 1
x id n  vn  1
x id (2)
id

where, d 1,2,...,D is one of the search space components; i 1,2,...,N represents the particle’s number, and N is
N. Afsharian Zadeh et al. / Procedia Engineering 154 (2016) 1393 – 1400 1395

the swarm size. F is known as the constraint coefficient, and Z is the inertia weight. c1 & c 2 are acceleration
coefficients, r1 & r 2 are random numbers uniformly distributed in [0 1], and n 1,2,...,N max represents the
iteration number. At the end of each iteration, the inertia weight is updated using the following equation:

Z max  Z min
Z n Z max  un (3)
n max

where Z n is the inertia weight in iteration n , and n max is the maximum number of iterations. Z max and Z min
are the maximum and minimum inertia weights, respectively.
The PSO algorithm starts with generating the initial positions of particles randomly, and then the algorithm
determines the gbest and pbest values for objective function evaluations in each iteration while updating the swarms
using by Equations (1) and (2). The procedure continues until the maximum number of iterations is reached, or the
gbest remains unchanged over a certain number of successive iterations.

3. Case study

Karoon2-Karoon3 hydropower system consists of two powerplants and dams on the Karoon River, the main
draining system of Karoon basin located in the southwest of Iran. Karoon3 Dam has already been constructed with a
high height powerplant, whereas Karoon2 and its run-of-river powerplant is under study, and its design parameters
including reservoir and powerplant's capacity need to be determined. The main characteristics of Karoon2 and
Karoon3 hydropower systems are given in Table 1 according to the available data and preliminary analyses. The
schematic representation of the system is shown in Fig. 1.

Fig. 1. Schematic representation of the Karoon2-Karoon3 system.

Table 1. Basic characteristics of Karoon2 and Karoon3 hydropower systems.

Karoon2
Upper limit of normal level (masl) 672
Lower limit of minimum operating level (masl) 660
Generator efficiency 93%
Head loss (meter) 2.25

Karoon3
Normal level (masl) 845
Minimum operating level (masl) 800
Installed capacity (Mw) 2000
3
Design discharge (m /s) 1321
Design head (meter) 164
Generator efficiency 92.4%
1396 N. Afsharian Zadeh et al. / Procedia Engineering 154 (2016) 1393 – 1400

Head loss (meter) 4.5

4. Model formulation

In hydropower systems, reservoir’s normal and minimum operating levels and powerplant’s production capacity
are the most important design parameters that need to be determined optimally from an economic point of view.
Moreover, the economic optimal design of the system depends on the operation of the system, especially variations
of head on and discharge through the turbines. The problem of optimal design and operation of the Karoon2-
Karoon3 hydropower system considering reliability of energy production and hydraulic coupling can be formulated
as a combinatorial nonlinear mathematical program.
The formulation of the resulting mixed integer nonlinear program (MINLP), including an economic objective
function and a set of constraints, can be stated as follows:

2 T
Max w1 u FE  ( PC  DC )  w2 u ( ¦ ¦ spill i ( t ) ) (4)
i 1t 1

Subject to

DC j( S max ) (5)
2

PC y( Pcap2 ) (6)

S i ( t  1 ) S i ( t )  I i ( t )  Q i ( t )  Spill i ( t ) t 1,...,T i 1,2 (7)

h ( t )  hi ( t  1 )
Ei ( t ) 2.725 u Qi ( t ) u ( i  htwi ( t )  hf i ) u ei t 1,...,T i 1,2 (8)
2

hi ( t ) f ( S i ( t )) (9)

M
h 2 ( t ) d ¦ hup m u Ztail m (t) t 1,...,T (10)
m 1

M
h 2 ( t ) t ¦ hdn m u Ztail m (t) t 1,...,T (11)
m 1

M
¦ Ztail m (t) 1 t 1,...,T (12)
m 1

M
htw 1(t) ¦ g 1m (Q1(t)) u Ztail m (t) t 1,...,T (13)
m 1

htw 2 ( t ) g 2 ( Q 2 ( t )) t 1,...,T (14)

E i ( t ) d Pcapi u nhours t 1,...,T i 1,2 (15)


N. Afsharian Zadeh et al. / Procedia Engineering 154 (2016) 1393 – 1400 1397

2
¦ E i ( t ) t FE u Z ( t ) t 1,...,T (16)
i 1

T Z( t )
¦ tD (17)
t 1 T

S min i d S i ( t ) d S max i t 1,...,T i 1,2 (18)

Q min i d Q i ( t ) d Q max i t 1,...,T i 1,2 (19)

The objective function defined in Equation (4) is to maximize the system’s firm energy yield (FE) while minimizing
the total cost, which includes the construction cost of Karoon2 Dam (DC), as a function of Karoon2 reservoir’s
capacity (Smax2) defined by Equation (5), and its powerplant (PC), as a function of the powerplant’s production
(installed) capacity (Pcap2) defined by Equation (6). The sum of spilled water is also minimized to avoid waste of
energy production as much as possible during high-flow periods.
Equation (7) is the balance equation, where S i ( t ) is the beginning –of-month storage volume of reservoir i ,
I i ( t ) is the volume of inflow to reservoir i , Q i ( t ) is the volume of water passing turbine i and Spilli ( t ) is the
volume of water spilled from reservoir i in period t all in million cubic meter (mcm).
Equation (8) is the basic equation of energy production. In this equation, E i ( t ) is the amount of energy in Mwh
generated by powerplant i , h i ( t ) is the beginning-of-month water level at reservoir i and htwi ( t ) is the tailwater
level of powerplant i . hf i is the head loss, and e i is the powerplant i’s efficiency. Equation (9) defines the relation
between storage volume and water level of reservoirs.
Constraints (10)-(13) are defined for considering the hydraulic coupling between Karoon2 reservoir’s forebay
elevation and the Karoon3 powerplant’s tailwater elevation. In this regard, a set of M rating curves, g1m (Q1 ) ,
have already been determined by hydraulic modeling of the river reach between the two reservoirs, characterizing
the variations of tailwater elevation of the first powerplant (Karoon3) as a function of discharge passing through the
turbine, Q1( t ) , and the downstream reservoir water level. In every time period, selection of the appropriate curve
m among these already-defined curves is done using binary variables Ztail( t ) , in the optimization model
formulation. Constraints (10) and (11) are for the selection of appropriate tailwater curve based on the value of
Ztail m ( t ) associated with curve m with upper limit ( hup m ), and lower limit ( hdn m ) characterizing this curve.
Equation (12) ensures that only one of the tailwater curves must be selected in each time period.
Equation (14) defines the tailwater elevation curve for the downstream powerplant (Karoon2), which is a
function of just discharge passing through Karoon2 powerplant, which is assumed to be a linear function.
Constraint (15) satisfies the upper bound on monthly energy generation in which nhours is the total number of
hours in a month, and Pcapi is the production (installed) capacity of powerplant i .
The required system firm energy yield is met through constraints (16) and (17), where Z( t ) is binary variables
that equals one if the total energy produced in time period t is equal or greater than the system firm energy yield
(FE); otherwise, will be equal to zero. Constraint (17) controls the maximum number of time periods in which the
required firm energy is not produced at the reliability level D .
Finally, constraints (18) and (19) satisfy lower and upper bounds levels on the reservoirs storage volumes and
releases in which S min i and S max i are, respectively, minimum and maximum storage volumes of reservoir i ,
and Q min i and Q max i are minimum and maximum turbine releases, respectively.

5. The solution procedure

In this section, the procedure used for solving the optimization model by the PSO algorithm is presented . The
PSO decision variables are three design variables including normal and minimum operating levels and the
1398 N. Afsharian Zadeh et al. / Procedia Engineering 154 (2016) 1393 – 1400

powerplant’s production capacity of Karoon2 as well as operational variables (releases, turbine discharge etc.) of
both Karoon2 and Karoon3. Therefore, in this two-reservoir system with a 54-years planning horizon and monthly
time steps, the optimization problem will include 1299 decision variables, of which 1296 variables are operational
ones. In this procedure, the design and operational variables are searched by the PSO algorithm, and the reservoirs
system operation is simulated using the sequential streamflow routing (SSR) method to estimate the energy yield
reliability. Considering these large number of decision variables, we paid special attention to the issue of generating
feasible solutions in the first population of the PSO algorithm satisfying water balance equality constraints. The flow
diagram of the procedure is presented in Figure 2.

Fig. 2. Flow diagram of the solution methodology.

6. Results

The solution methodology explained above was used in solving the problem for two reliability levels of (80% and
90%). The results obtained are in Table 2 reporting optimal values of design variables and the mean value of
monthly firm energy produced by the system and each of powerplants. More details on the energy production
distribution are illustrated through energy duration curves presented in Figs. 3 and 4.
N. Afsharian Zadeh et al. / Procedia Engineering 154 (2016) 1393 – 1400 1399

Table 2. Optimal values of design variables and the mean value of system monthly energy yield by each of the power
plants for reliability of 90% & 80%.
Karoon2
Karoon2 Karoon2
production
Reliability System firm Karoon3 firm Karoon2 firm normal minimum
(installed)
level energy (Mwh) energy (Mwh) energy (Mwh) operating operating
capacity
level (masl) level (masl)
(Mw)
90% 332058 187360 132251 666 661 593
80% 390578 216915 152681 667 664.5 694

1800000
1600000
1400000
energy-Mwh

1200000
1000000
800000
600000
400000
200000
0
0 20 40 60 80 100 120
exceedence probability

Fig. 3. System energy duration curve for reliability level of 90%.

2000000
1800000
1600000
1400000
energy-Mwh

1200000
1000000
800000
600000
400000
200000
0
0 20 40 60 80 100 120
exceedence probability

Fig. 4. System energy duration curve for reliability level of 80%.


1400 N. Afsharian Zadeh et al. / Procedia Engineering 154 (2016) 1393 – 1400

7. Summary

The problem of optimal design and operation of the Karoon2-Karoon3 cascade hydropower system considering
reliability of energy production and hydraulic coupling of powerplants were formulated as a combinatorial nonlinear
program. The program was then solved using the PSO metaheuristic algorithm linked to a sequential-streamflow-
routing (SSR)-based multireservoir hydropower reservoir operation simulation model. The results provided the
design parameters of Karoon2 hydropower system, which is being planned as a potential run-of-river powerplant, as
well as operational variables and energy production states of the whole system including those of the already-
constructed Karoon3 Dam as a high-height hydropower system presented by energy duration curves determined.

References

[1] Gablinger, M., Loucks, D. P., Markovs models for flow regulatio, J. Hydraul. Div, 1970, 96(1), pp. 165-181.
[2] Kim, Y. O., Palmer, R. N., Value of seasonal flow forecasts in Bayesian stochastic programming, J. Water Res. Plann. Manage, 1997, 123(6),
pp. 327-335.
[3] Barros, M.T.L., Tsai, F., Yang, S., Lopes, J., Yeh, W., Optimization of Large-Scale Hydropower System Operations, J. Water Res. Plann.
Manage, 2003, 129(3), pp.178-188.
[4] Yeh, W. G., Becker, G. L., Chu, W. S., Real-time hourly reservoir operation, J. Water Res. Plann. Manage, 1979, 105(2), pp. 187-203.
[5] Grygier, J. C., Stedinger, J. R., Algorithms for Optimizing Hydropower System Operation, Water Resour. Res, 1985, 21(1), pp.1-10.
[6] Powell, D., Variable metric methods for constrained optimization. In Mathematical Programming: The State of the Art, pp. 228-311.
[7] Diaz, G.E., Fontane, D. G., Hydropower Optimization via Sequential Quadratic Programming, Water Resources Planning and Management,
ASCE, 1989, 115(6), pp715-733.
[8] Reznick, K., Simonovic, S. P., An improved algorithm for hydropower optimization, Water Resour. Res, 1990, 26(2), 189-198.
[9] Mousavi, S.J., Shourian, M., Capacity Optimization of Hydropower Storage Project Using Particle Swarm Optimization Algorithm Journal of
Hydroinformatics, 2010, 12(3), pp.275-291.
[10] Hawary, E.L., Christensen, C.S., Optimal Economic Operation of Electric Power Systems Academic, New York, 1979.
[11] Lyra, C., Ferreira, L. R., A Multiobjective Approach to the Short-Term Scheduling of a Hydropower System, IEEE Trans, on Power Sys,
1995, 10(4), pp. 1750-1755.
[12] Teegavarapu, R., Simonovic, S., Short-Term Operation Model for Coupled Hydropower Reservoirs, J. Water Resour. Plann. Manage, 2000,
126(2), pp.98–106.
[13] Kennedy, J., Eberhart, R.C., “Particle Swarm Optimization” Proceeding of the IEEE International Joint Conference on Neural Networks,
pp.1942-1948.

You might also like