Gulf Cooperation Council Countries’ Electricity Sector Forecasting : Consumption Growth Issue and Renewable Energy Penetration Progress Challenges

Alharbi, Fahad and Csala, Denes and Wang, Ziwei and Campobasso, Sergio (2023) Gulf Cooperation Council Countries’ Electricity Sector Forecasting : Consumption Growth Issue and Renewable Energy Penetration Progress Challenges. PhD thesis, Lancaster University.

Text (2023AlharbiPhD.pdf)
2023AlharbiPhD.pdf.pdf - Published Version
Available under License Creative Commons Attribution-NonCommercial-NoDerivs.
Download (38MB)

Abstract

The Gulf Cooperation Council (GCC) countries depend on substantial fossil fuel consumption to generate electricity which has resulted in significant environmental harm. Fossil fuels also represent the principal source of economic income in the region. Climate change is closely associated with the use of fossil fuels and has thus become the main motivation to search for alternative solutions, including solar and wind energy technologies, to eliminate their reliance on fossil fuels and the associated impacts upon climate. This research provides a comprehensive investigation of the consumption growth issue, together with an exploration of the potential of solar and wind energy resources, a strict follow-up to shed light on the renewable energy projects, as currently implemented in the GCC region, and a critical discussion of their prospects. The projects foreshadow the GCC countries’ ability to comply with future requirements and spearhead the renewable energy transition toward a more sustainable and equitable future. In addition, four forecasting models were developed to analyse the future performance of GCC power sectors, including solar and wind energy resources along with the ambient temperatures, based on 40 years of historical data. These were Monte Carlo Simulation (MCS), Brownian Motion (BM), and a seasonal autoregressive integrated moving average with exogenous factors (SARIMAX) model model-based time series, and bidirectional long short-term memory (BI-LSTM) and gated recurrent unit (GRU) model-based neural networks. The MCS and BM prediction models apply a regression analysis (which describes the behaviour of an instrument) to a large set of random trials so as to construct a credible set of probable future outcomes. The MCS and BM prediction models have proven to be an exceptional investigative solution for long-term prediction for different types of historical data, including: (i) four types of fossil fuel data; (ii) three types of solar irradiance data, (iii) wind speed data; and, (iv) temperature data. In addition, the prediction model is able to cope with large volumes of historical data and different intervals, including yearly, quarterly, and daily. The simplicity of implementation is a strength of MCS and BM techniques. The SARIMAX technique applies a time series approach with seasonal and exogenous influencing factors, an approach that helps to reduce the error values and improve the overall model accuracy, even in the case of close input and output dataset lengths. This iii research proposes a forecasting framework that applies the SARIMAX model to forecast the long-term performance of the electricity sector (including electricity consumption, generation, peak load, and installed capacity). The SARIMAX model was used to forecast the aforementioned factors in the GCC region for a forecasted period of 30 years from 2021 to 2050. The experimental findings indicate that the SARIMAX model has potential performance in terms of categorisation and consideration, as it has significantly improved forecasting accuracy when compared with simpler, autoregressive, integrated, moving average-based techniques.The BI-LSTM model has the advantage of manipulating information in two opposing directions and providing feedback to the same outputs via two different hidden layers. A BI-LSTM’s output layer concurrently receives information from both the backward and forward layers. The BI-LSTM prediction model was designed to predict solar irradiance which includes global horizontal irradiance (GHI), direct normal irradiance (DNI), and diffuse horizontal irradiance (DHI) for the next 169 hours. The findings demonstrate that the BI-LSTM model has an encouraging performance in terms of evaluation, with considerable accuracy for all three types of solar irradiance data from the six GCC countries. The model can handle different sizes of sequential data and generates low error metrics. The GRU prediction model automatically learned the features, used fewer training parameters, and required a shorter time to train as compared to other types of RNNs. The GRU model was designed to forecast 169 hours ahead in terms of forecasted wind speeds and temperature values based on 36 years of hourly interval historical data (1st January 1985 to 26th June 2021) collected from the GCC region. The findings notably indicate that the GRU model offers a promising performance, with significant prediction accuracies in terms of overfitting, reliability, resolution, efficiency, and generalisable processes. The GRU model is characterised by its superior performance and influential evaluation error metrics for wind speed and temperature fluctuations. Finally, the models aim to help address the issue of a lack of future planning and accurate analyses of the energy sector's forecasted performance and intermittency, providing a reliable forecasting technique which is a prerequisite for modern energy systems.