Use of Artificial Intelligence Approach for the Modelling of Electricity- Water-Climate Nexus

Abstract

The global water and energy demand is expected to grow significantly. Therefore, water and energy have become important resources and must be wisely managed for a sustainable future. However, the interaction and dependence between electricity generation and water resources will further exacerbate existing vulnerabilities in the water-energy nexus under the negative impacts of climate change. The aim of this thesis is to investigate the relationship between water and electricity generation under different climate change scenarios and evaluate the basin-based and country-based water consumption and water intensity of electricity production to contribute to the policy perspectives with a new tool and method to sustain the future. Various model studies on the water-energy nexus have been conducted. Still, as a reliable forecasting tool, the adaptive neuro-fuzzy inference system (ANFIS) model evaluation has not been tested for forecasting water consumption during electricity generation. Thus, the ANFIS modelling approach is used in this study to generate reliable water consumption estimates based on electricity generation. In the first part of the study, the data of hydroelectric power plants with dams and fossil-fueled power plants, like electricity generation, fuel type, cooling system types, annual average water consumption, and climate data, including historical and future projections, are gathered from various sources. The ANFIS models of both hydroelectric and fossil-fueled power plants with 85% data splitting gave the best-fitted models. The lowest MAPE of hydroelectric power plants’ ANFIS model resulted in about 9.59 % testing data, while fossil-fueled power plants’ MAPE is about 4.62%. The R2 of the models is nearly and above 0.97, and it can be said that model fitting is acceptable. Water consumption and water intensity of hydroelectric power plants were 2,150 million m3 and 41,841 m3/GWh, while it was 191 million m3 and 892 m3/GWh for fossil-fueled power plants, respectively, in 2021. And Turkey's total water consumption and intensity values are evaluated as about 2,341 million m3 and 8,814.5 m3/GWh, respectively, in 2021. In the next part of the study, climate change scenario applications are analyzed to see the effects on the electricity sector related to water resources. The HadGEM-4.5 climate scenario gave the highest hydroelectric power plant water consumption and water intensity recorded at 4,393 million m3 and 85,487 m3/GWh in 2053, while the GFDL-4.5 had the lowest value at 2,609 million m3 and 50,768 m3/GWh, respectively. Additionally, for the HadGEM 8.5 climate scenario, the largest water consumption by fossil-fueled power plants was recorded in Turkey at 246 million m3 and 1,146.2 m3/GWh. And the highest total water consumption and intensity of power plants in Turkey were seen at 4,635 million m3 and 17,450.7 m3/GWh for the HadGEM 4.5 climate scenario in 2053, respectively. The final part of the study analysed the basin-based water consumption per basin area (BBWC) and ratio (BBR) results and visuals of the mapping tool depending on climate change scenarios. The highest BBWC values were seen at Fırat-Dicle, Ceyhan and Kızılırmak Basins in 2053 at HadGEM 4.5 scenario, while Dogu Karadeniz, Aras and Dogu Akdeniz Basins have the lowest BBWC. The Kızılırmak Basin gave the highest BBR % in the 2053 year for all climate scenarios and followed by the Fırat-Dicle, Gediz, Ceyhan, Kuzey Ege and Sakarya Basins, respectively.