Genetic Algorithm in Tidal Range Schemes’ Optimisation

Abstract

Tidal energy has a significant advantage over many other forms of renewable energy because of the predictability of tides. Tidal Range Structures (TRSs) are one of the main forms of tidal renewable energy. Designing the operation of TRSs is one of the challenging aspects in early stages due to the large variety of scenarios. Traditionally this has been done using a grid search. However, grid search can be very elaborate and time consuming during the design of TRSs. This paper proposes a novel and more efficient method to optimise the design of the operation of TRSs by maximising their electricity generation using a Genetic Algorithm. This GA model is coupled with a 0-D model which breaks the tides into small units and considers flexible operation. This approach delivered more than a 10% increase in electricity generation when compared to non-flexible operation, i.e. using fixed heads for all tides, just by optimising the operation. The GA model was able to achieve the same amount of electricity compared to the best grid search method with flexible operation more efficiently, i.e. with about a 50% reduction in simulation time. The feasibility of the elite operational scheme is validated through a developed 2-D model.

Design of tidal range energy generation schemes using a Genetic Algorithm model

“…There are a variety of mutation and recombination methods used in GA approaches. It has been shown that implementing a Sequence Mutation Method (SMM) and a Ring Recombination Method (RRM) [18,36] can lead to better optimisation results with higher efficiency, in comparison to other methods [20,37]. The mutation will restart from the first gene in the next generation when the end of the gene sequence of an individual is reached during the mutation.…”

“…This process is repeated as the evolution continues. The 'Ring' in RMM denotes that chromosomes of individuals are considered in a ring form [18,38,39] and a random section of the chromosomes are selected, which is not limited to the length of the chromosome. All individuals are evaluated using a fitness function before the next generation is produced.…”

“…This simplified approach provides a higher level of uncertainty in the early stages of design, which can be refined further as the design progresses using either 2-D or 3-D hydrodynamic modelling tools. Optimisation of the entire scheme with some degree of confidence is particularly crucial in the early stages of operational design, due to the significant influence of these multi-variables on maximising the annual electricity generated and therefore the feasibility of any proposed scheme [18]. Hence, a more advanced optimisation algorithm with high computational efficiency, and which provides more certainty of the operational outputs of a scheme through a cost-benefit analysis, is needed to facilitate the development and operational design of TRSs, particularly in terms of maximising energy generation to meet demand.…”

“…Evolutionary Algorithms (EAs) have been developed to solve a wide range of optimisation problems in power systems [19,20], with Genetic Algorithms (GAs) constituting one of the metaheuristic EAs frequently used to combine variables to analyse optimisation problems, and providing a high level of satisfactory results [21], especially when compared with the brute force algorithms including the grid search method [22]. EAs have been used in the optimisation of tidal energy generated from TRSs, and have already been used to provide design details in terms of the number of turbines and sluice gates to use for such schemes [18,23,24]. Leite Neto et al discussed the maximum energy gains obtained with dispatchable turbines by using GAs for optimisation [24].…”

“…They investigated the potential of using GAs to optimise the operation of such schemes [24]. More recently, GAs have been refined to determine operational optimisation with the findings being possible for the implementation of GAs in the optimal design of TRSs for higher efficiency [17], particularly in comparison with other alternative algorithms [18]. Furthermore, Kontoleontos and Weissenberger [23] highlighted the scope for using EAs in optimising multiple variables, including operational modes, speed of turbines etc., to maximise energy generation.…”