Modelling of a Combined Cycle Power Plant Performance Using Artificial Neural Network Model

Abstract

The obvious and disastrous consequences of climate change in recent years have moved world leaders and countries to adopt a shift from conventional electricity production means (fossil fuel power plants) to cleaner or renewable energy systems. Although this shift became an imperative necessity, most countries appeared not to be ready for new electricity production technologies. This was primarily due to the lack of will to transform economies in developed countries and limited financial resources in developing countries. Therefore, fossil fuel power plants are still the primary energy source or are used as a base or bottoming power plants. To improve the performance of these plants and reduce pollution and operating costs, these power plants were hybridized with either a renewable energy system or a gas turbine. During this transition period, where countries are moving from a system dominated by fossil fuel power plants to greener energy generation, hybridization of the fossil fuel power plants remains an option even beyond as long as most renewable energy systems are intermittent. In this paper, the performance of a Combined cycle power plant using fossil fuel and a gas turbine is modelled and analyzed using Artificial Neural Network (ANN) for the training and testing of the model. Published 620 data sets which were collected on a combined cycle gas turbine (CCGT) between 2006 and 2011 were used in this study, with ambient Temperature, exhaust vacuum, ambient pressure, and relative humidity being used as input parameters, while Net electrical energy output was used as an output parameter. The power plant was rated at 480 MW. The results showed that ANN performed well with an acceptable overall coefficient of correlation R-value of 0.9727 when trained with the Levenberg-Marquardt algorithm.