Term 252

Digital Twin and AI-Assisted Optimization of Refinery Hydrogen Network for Cost and Emission Reduction

Project Type: Self-Initiated

Project Description

This project presents the final design and evaluation of an integrated hydrogen management and crude feed assessment framework for refinery operations. The objective is to improve hydrogen utilization efficiency and reduce supply cost while maintaining safe and standards compliant operation under varying crude feed conditions. The design combines steady-state process simulation in Aspen HYSYS with an AI-assisted hydrogen routing system developed using Python 3.11 and the FastAPI framework to evaluate operational feasibility in real time. Representative crude feed cases were defined to capture variations in flow rate, temperature, viscosity, and crude quality. Hydrogen demand multipliers were applied to assess system responsiveness under different operating conditions. The hydrogen network configuration was analyzed to ensure pressures and temperatures remain within API and ASME allowable limits, and hydrogen purity constraints were embedded into the AI recommendation engine. Automated validation routines were implemented to flag off-spec routing decisions and potential overpressure scenarios. Flammability and overpressure risks associated with hydrogen handling were evaluated, and appropriate mitigation measures were incorporated into the design. Simulation and economic analyses were conducted to assess technical feasibility and cost implications across the defined scenarios. Results indicate that the proposed system improves hydrogen distribution efficiency while maintaining operational feasibility and safety compliance under evaluated cases. The integrated framework provides a structured and scalable approach for refinery hydrogen optimization and supports informed decision-making under variable crude operating conditions.