Abstract
The accumulation of heavy polynuclear aromatics (HPNAs) in hydrocracking units (HCUs) poses significant challenges to catalyst performance and process efficiency. This study proposes an integrated machine learning framework that combines ridge regression, K-means, and long short-term memory (LSTM) neural networks to predict HPNA formation, enabling proactive process management. For the training phase, weighted average bed temperature (WABT), catalyst deactivation phase—clustered using unsupervised K-means clustering—and hydrocracker feed (HCU feed) parameters obtained from laboratory analyses are utilized to capture the complex nonlinear relationships influencing HPNA formation. In the simulation phase, forecasted WABT values are generated using a ridge regression model, and future HCU feed changes are derived from planned crude oil blend data provided by the planning department. These forecasted WABT values, predicted catalyst deactivation phases, and anticipated HCU feed parameters serve as inputs to the LSTM model for predicting future HPNA levels. This approach allows us to simulate various operational scenarios and assess their impact on HPNA accumulation before they manifest in the actual process. By identifying critical process parameters and their influence on HPNA formation, the model enhances process engineers' understanding of the hydrocracking operation. The ability to predict HPNA levels in advance empowers engineers to implement corrective actions proactively, such as adjusting feed compositions or operating conditions, thereby mitigating HPNA formation and extending catalyst life. The integrated framework demonstrates high predictive accuracy and robustness, underscoring its potential as a valuable tool for optimizing HCU operations through advanced predictive analytics and informed decision-making.