Development of a load scheduling algorithm to aid in demand side management for industrial consumers
Abstract
This document contains a detailed description of our project titled, “The Development of a Load Scheduling Algorithm to Aid in Demand Side Management for Industrial Consumers”, it starts at the introduction where concise information on the general background behind the project’s area of specification and the particular problem that we are addressing, by taking on this project are discussed, along with the reasons why this project is important. This is all done within the introduction of the report. The literature concerning the project is then discussed in the second chapter, here, previous work on the area the project is addressed to, are analysed as we try to identify the gaps in this work and broaden the understanding of our project. The methodology of the report follows in the 3rd chapter where we talk about the various activities, we completed in order to see our project through following the order of the specific objectives i.e., to collect data through an energy audit to feed into the algorithm, to develop the load scheduling algorithm to aid in Demand Side Management (DSM). With the aid of Python programming, we were able to adopt a Binary Particle Swarm Optimization (BPSO) algorithm to make the load scheduling possible and lastly, to evaluate the performance of the load scheduling algorithm by adopting a Genetic Algorithm (GA) within Excel Solver (ES) and running both algorithms against each other. The fourth chapter, contains the results obtained from the project through a series of tests done on both the algorithms, these results are visualised and discussed in details, from this discussion we noted that the BPSO algorithm had 23.7% reduction in cost, 28% reduction in peak demand and registered energy losses of about 21%, as compared to the GA’s 8.13% reduction in cost, 4.13% reduction in peak demand. From the results obtained it is concluded that the BPSO algorithm is superior in cost reduction, peak reduction and shorter convergence time, however it falls short when it comes to energy losses, on that note, recommendations to further study the algorithm to find ways to mitigate the energy losses would improve the algorithm’s performance making it ideal for users to exercise.