Monitoring the Pressure Swing Adsorption Process: A Model-Based Approach for Fault Detection

High Resolution Image Download MS PowerPoint Slide This study presents a dynamic model for pressure swing adsorption (PSA) for hydrogen purification in refineries, with a focus on failure detection, particularly valve malfunctions. Developed from first principles, the model simulates pressure profiles across PSA vessels and accounts for valve mass-flow dynamics to accurately capture the process. Validation against actual plant data demonstrated a high accuracy and low computational time. The model is integrated with an extended Kalman filter, which generates and evaluates residuals for fault detection and isolation in a fault scenario. The proposed model serves as a reliable and computationally efficient tool for enhancing the PSA process monitoring and operational safety.