Enhancing Electric Grid Management through Climate-Driven Energy Demand Forecasting Models

Energy demand is directly impacted by fluctuations in temperature. Climate change affects energy demand by altering the extent and intensity of daily and seasonal heating and cooling needs and may pose a threat to infrastructure and communities during extreme events. Forecasting demand patterns under diverse future climate projection scenarios is influenced by the ongoing adoption of cutting-edge technologies like electric vehicles (EVs) and the deployment of solar photovoltaics (PVs). In this study, we used energy demand data from ISO New England (ISO-NE) for Connecticut (CT) during 2011-2021 and the daily resolution Coupled Model Intercomparison Project Phase 6 (CMIP6) dataset for climate projections for the period of 2011-2050. Employing machine learning and neural network models, we developed predictive models to predict energy consumption as a function of dynamic climate variables. Performance evaluation for the models was conducted using MAE, MAPE, and RMSE regression evaluation metrics. We also assessed energy demand across three scenarios: i) prevailing technology adoption; ii) projected technology adoption; and iii) accelerated technology adoption. Our proposed energy demand forecasting models hold potential for improving grid management by providing critical insights for resilient energy infrastructure planning, informed decision-making, and sustainable resource management in the face of changing climatic conditions.