Optimization of the operation of hydro stations in market environment using Genetic Algorithms

Abstract

This paper describes an approach to the short term operation planning of hydro stations in market environment. The developed approach is based on the solution of an optimization problem to maximize the profit of a generation agent along a planning period discretized in hourly steps using a Genetic Algorithm. This problem includes the possibility of pumping since this is an important resource in the scope of electricity markets. The scheduling problem was developed starting with an initial simplified version in which the head loss is neglected and the head is assumed constant. Then, it was implemented a second model in which the nonlinear relation between the head, the hydro power and the water discharge is retained and finally an approach in which the hydro schedule obtained in a given step is used to update the hourly electricity prices used to compute the profit of the generation agent. The short term hydro scheduling problem is illustrated using two Case Studies -the first one was designed to run a set of initial tests to the developed algorithm and the second one refers to a set of hydro stations that mirrors a cascade of 8 stations in Portugal.

Operation planning of hydro stations using genetic algorithms considering their impact on the electricity market prices

“…In order to consider this effect, expression (1) is substituted by (2) in which β represents the head loss coefficient. For each particular value of ik h one obtains a non-linear expression relating ik q and Tik P , that is, expression (2) corresponds to a family of curves as illustrated in Figure 1.…”

“…Other models include binary variables as [9] to represent the state of each station leading to mixed integer linear or non-linear formulations. More recently, metaheuristic techniques started to be applied to this problem including Neural Networks [10], Simulated Annealing [11], Tabu Search [12], Genetic Algorithms [2,13] and particle swarm approaches [14]. Finally, several publications [1,15] use an iterative process in which the head is updated using the value of ik q got in the previous iteration and the β coefficient is given by (3).…”

“…In recent publications we have developed short term operation planning models for hydro stations namely installed in cascade taking into account electricity markets prices that is admitting that hydro stations behave as price takers [1,2] using an under relaxation approach and Genetic Algorithms. However, in countries as in Portugal the share of hydro generation is typically large (in average hydro years it can reach 25% of the total electricity generation) so that taking hydro stations as just price takers becomes far from reality.…”

“…The results provided by this model correspond to the operation strategy that maximizes the profit of the generation agent along the 24 hours of a trading day or along a week. In order to consider the impact on electricity prices, this model is then included in an iterative procedure that starts with a set of initial electricity prices for each hour of the planning horizon together with the demand curve and the estimated selling curve for the remaining generation agents, solves the mentioned profit maximization problem using Genetic Algorithms [2], updates the hourly market selling and buying aggregated curves and makes the intersection of these curves to update the hourly prices. These prices are then input again in the profit maximization problem to progressively refine the operation decisions of the hydro stations.…”

Comparison of Differential Evolution and Particle Swarm Optimization for Optimizing the Power Generation Potential in Hydroelectric Power Plants using Dot Net

“…The selection of the turbine efficiency curves that would be ideal for each operation system leads to a mixed, non-linear programming problem. Modern heuristic optimization techniques based on operational research and artificial intelligence concepts, such as Genetic Algorithm [3], [4], Particle Swarm Optimization [5], Artificial Bee Colony [6], Bacterial Foraging Algorithm [7][8], Ant Lion [9], [10], NSGA-II [11], [12], Big BangBig Crunch [13], [14], Grey Wolf Optimization [15], [16]and Differential Evolution [17] provide outstanding solutions for the non-linear nature of the real world problems. Each method has its own merits and de-merits.…”