Distillation Control

Distillation Control Configuration Selection

Authors: Scott Hurowitz, Joe Anderson, Marshall Duvall, and James B. Riggs

ABSTRACT

Detailed dynamic simulations of three distillation columns (a propylene/propane splitter, a xylene/toluene vacuum column, and a depropanizer) have been used to study the issues of configuration selection for single and dual composition control, feedforward control using a feed composition analyzer, and decouplers using control perfomance results for PI controllers. ATV identification [1] with online tuning for setpoint changes was used for tuning the diagonal PI composition controllers. In addition, controller reliability tests were conducted by inducing reboiler duty upsets.For dual composition control, the optimum configuration changes from one column to another, but it is found to be among a predictable set of control configurations. When the control of both products is equally important, four or eight potential configurations must be considered, which usually requires a number of dynamic simulations to identify the optimum configuration. On the other hand, guidelines are presented for reducing the number of viable control configurations to two for cases in which one product is more important than the other. Feedforward control using a feed composition analyzer and one-way decouplers are shown to offer significant advantages for specific control configurations, but can degrade control performance for others. The results presented here are consistent with other studies in the field, but provide a more unified description of the field.

 

Publication Information: Accepted to J. Process Control

Corresponding Author: Jim Riggs

 
Including Levels in MPC to Improve Distillation Control Control Authors

Authors: Haitao Huang and James B. Riggs

ABSTRACT

This work examines the effects of including levels into MPC for distillation control via rigorous simulation of two different columns: a depropanizer and a propane/propylene splitter (C3 splitter). Three different MPC implementations for distillation control were compared: regular implementation in which levels are controlled by PI regulatory controllers; direct implementation in which the levels are controlled by MPC directly by manipulating flow rates; and cascade implementation in which MPC moves the setpoints to the regulatory level controllers. The results show that directly including levels in MPC improves composition control performance significantly for fast responding distillation columns. The cascade implementation is recommended for columns that exhibit ill-conditioning for the [L,V] control configuration or for columns that exhibit a large time to obtain steady state.

 

Publication Information: I&ECR, Vol 41, No. 16, Aug 2002

Corresponding Author: Jim Riggs