Energy Management of Community Microgrids using Particle Swarm Optimisation

Abstract

This dissertation focuses on the energy management of a community microgrid to minimise its operational cost under uncertain power generation, power demand and electricity prices. The study proposes some effective solution approaches to minimise the operational cost of the microgrid. Firstly, a real-time energy management scheme that is free from the effect of uncertainty is developed and compared with existing management schemes to efficiently control the battery energy of a microgrid. In the objective function for the scheme, a dynamic penalty function is added to incentivise battery charging during low electricity prices. The proposed method can reduce operational cost by 12 per cent as compared to a well-established existing one over a time horizon of 96 hours. Secondly, a two-hour ahead energy management approach considering the degradation cost of the battery and a penalty function to reflect the true operational cost is proposed. The optimisation problem formulated is solved using a particle swarm optimisation algorithm which is designed. The proposed energy management approach reduces electricity cost by up to 44.50 per cent compared to a baseline method and 37.16 per cent compared to another existing approach. Finally, day-ahead scheduling of the battery energy is proposed while considering its degradation costs due to charging-discharging cycles. The degradation costs with respect to the depth of charge are modelled and added to the objective function to determine the actual operational costs of the microgrid. A framework to solve the optimisation problem formulated is developed in which particle swarm optimisation, the Rainflow algorithm and scenario techniques are integrated. Uncertainties of variables, such as power generation and electricity prices, are also discussed in the study. Simulation results demonstrate that the proposed method for a day-ahead scheduling program can reduce the operational costs by around 40 per cent compared to the baseline method. Results also reveal that uncertainty in power generation and power demand has no influence on the energy schedule of the battery, but variation in electricity prices has an impact on the outcome. Several pragmatic tests verify the effectiveness of the proposed methods.