Multivariable Model Predictive Control for optimal operation of a fluid catalytic cracking debutanizer distillation column.

摘要

The purpose of this work is to discuss the application of Dynamic Matrix Control (DMC), a modality of Model Predictive Control (MPC), in a debutanizer distillation column. The column was analyzed and the required step-testing was performed in order to come up with a reliable process model on which the predictive quality would be based. Current market economics, feed nature, process gains, and utility costs were incorporated within the LP cost factors in order to drive the unit towards maximum profitability with the minimum associated cost. These LP cost factors were directly related to the associated performance index criteria. DMC was successfully implemented with an end result of a 5.5% average increase in the C5 composition of the debutanizer's overhead butane-butylene (BB) stream. The standard deviation and variability of process variables, including the control objective, were also reduced significantly. Pressure, temperature, and reflux flows were selected as the manipulated variables which would target steady-state conditions within process and instrumentation constraints in order to optimally operate the column. This advanced multivariable control effectively proved to be a valuable addition to the unit's productivity by allowing it to operate closer to its limits while maintaining a tighter control on all other associated process variables. This translates into maximum profitability and optimal operation of the debutanizer's performance.