Optimization of a Semibatch Distillation Process with Model Validation on the Industrial Site

Abstract

An industrial semibatch distillation process with a chemical reaction in the reboiler is optimized to minimize the batch operation time. A detailed dynamic model is employed to describe the process and incorporated in the optimization problem. This model is validated with the measured data from the experiment directly on the industrial site. The optimal reflux ratio and the feed rate policies of the process are developed with an efficient optimization approach. It can be shown that up to 30% of the operation time can be saved with the optimal policies developed in comparison to the conventional operation. The modeling and optimization approach proposed can be readily applied to the optimization of general batch distillation processes.

Optimal operation of distillation processes under uncertain inflows accumulated in a feed tank

“…cient, since all equations that may be very complicated are not included in SQP. It has been successfully applied to sev-Ž eral real processes Li et al, 1998;Wozny and Li, 2000;W et . al., 2000 . Ž .…”

“…Although these approaches have been successfully applied to many chemical plants such as reactive, azeotropic, Ž and batch distillation processes Logsdon and Biegler, 1989; . Li et al, 1998;Wendt et al, 2000 , the optimization results are only applicable when the real operating conditions are included in the optimization.…”

Efficient Simulation of Batch Distillation Processes by Using Orthogonal Collocation

“…Note that the lengths of the intervals are not necessarily the same. Thus, the time points corresponding to the collocation points of an interval will be: (28) and then we have the relation: (29) From (25) and (28) the differential terms in (27) can be described as: (30) The collocation points and the corresponding derivations can be given a priori. The control variables can also be represented in the same way, but for simplicity we use the form of piecewise constant for the control variables in this study.…”

A bilevel NLP sensitivity‐based decomposition for dynamic optimization with moving finite elements

“…In AIChE Journal alone, over 200 studies have appeared since the publication of Pontryagin's classical contribution . Optimal control applications abound in the development of optimal trajectories for separation processes, including distillation and crystallization, determination of optimal material and energy inputs, temperature and catalyst loading profiles in continuous reactors and operating strategies and recipe generation for batch processes . Once these optimal profiles are obtained, they serve as essential targets and reference trajectories to assess process performance, treatment of uncertainty and variability, and timely adaptation to real‐time data.…”