Research Projects

The rapid growth of electric vehicle (EV) adoption in India necessitates the expansion of charging infrastructure, with government guidelines requiring urban areas to have charging stations within a 1 km square grid by 2030. Additionally, fast-charging stations are expected for heavy-duty EVs at transport depots. This development demands significant power grid planning and potential upgrades, especially where feeder load exceeds 70% capacity. Current approaches to EV charging infrastructure (EVCI) planning primarily treat it as an optimal placement or hosting capacity problem, often incorporating renewable energy data and traffic flow analysis. However, these methods face challenges in modeling uncertainty and computational complexity, as renewable generation and traffic flow are inherently unpredictable. Standard approaches may simplify these issues with specific distributions, yet such simplifications can overlook real-world variability and accessibility objectives highlighted in the NITI Aayog handbook. As a result, many existing tools do not fully address accessibility, utilization, and grid requirements for effective EVCI planning. This proposal addresses the challenge of evaluating the impact of EVCI on the power grid, focusing on non-grid factors such as EVCI usage, accessibility, and space requirements. It also aims to provide distribution system operators (DSOs) an efficient and scalable computational framework and tools to handle uncertainties related to renewable energy, grid topology changes, and traffic flow fluctuations. The project will adopt an Infra-to-Grid approach using a machine learning-based, computationally efficient scheme to assess grid security for specific EVCI policies and provide actionable insights to grid operators. Key objectives include developing a novel tool for evaluating EVCI suitability against grid constraints, enabling security assessments over 100 times faster than current tools, and avoiding assumptions on uncertainty distributions. Additionally, the tool will facilitate online training of power flow proxies in milliseconds, offering operational and planning options for grid security. The project will be executed in three work-packages 1) power flow proxy and prior-data fitted network development, 2) probabilistic security assessment module development and 3) probabilistic security enhancement module development. This project will support the Government of India's efforts to accelerate EV adoption, help DSOs understand the dynamics between EV charging loads, consumer loads, and distributed energy resources, and contribute to advancements in AI for Engineering and Science, along with enhancing GPU-based computing facilities at IIT Roorkee.