Research Initiation Award: "On-Line Optimization of Batch Reactors: Experimental Implementation to Polymerization Reactors"

摘要 - Palanki - 9409577 间歇式反应器生产多种特种化学品。由于间歇式反应器本质上是瞬态的，因此主要目标是在间歇循环结束时优化性能指数（例如产率）。 间歇式反应器操作的一个主要问题是初始条件和环境条件的变化可能导致批次间的显着变化，从而导致资源和时间的显着损失。 为此，有必要制定可以在线实施的优化策略。 在本研究中，将开发一种用于在线优化具有多个输入的批处理过程的方法框架。 重点是在工业上重要的间歇聚合反应器中最佳状态反馈定律的实验实施。 选择苯乙烯与丙烯酸甲酯在甲苯中使用异丁腈（AIBN）共聚作为模型实验系统来演示在线优化策略的应用。 在研究的第一阶段，将开发实验系统的数学模型。 在第二阶段，将开发最佳状态反馈定律，用于具有多个输入的间歇式反应器的终点优化。 最优状态反馈在优化性能指标方面的功效将通过计算机仿真在数学模型上进行检验。 第三阶段将试行在线优化策略。 将进行多次批量和半批量实验，以证明当初始条件和过程扰动存在批次间变化时，最佳状态反馈在线计算最佳操作策略的能力。

Abstract - Palanki - 9409577 A large variety of specialty chemicals are produced in batch reactors. Since batch reactors are transient in nature, the major objective is to optimize a performance index (such as yield) at the end of the batch cycle. A major problem in batch reactor operation is that changes in initial conditions and environmental conditions can cause significant batch-to-batch variations leading to significant losses in resources and time. For this reason, it is necessary to develop optimization strategies that can be implemented on-line. In this study a methodological framework for on-line optimization of batch processes with multiple inputs will be developed. The emphasis is on the experimental implementation of optimal state feedback laws in industrially important batch polymerization reactors. The copolymerization of styrene with methyl acrylate in toluene using aso-bis-iso-butyronitrile (AIBN) has been selected as a model experimental system to demonstrate the application of the on-line optimization strategy. In Phase I of the research, a mathematical model of the experimental system will be developed. In Phase II, optimal state feedback laws will be developed for end-point optimization of batch reactors with multiple inputs. The efficacy of the optimal state feedback in optimizing the performance index will be tested on the mathematical model by computer simulations. In Phase III, the on-line optimization strategy will be implemented experimentally. Several batch and semi-batch experiments will be conducted to demonstrate the ability of the optimal state feedback to calculate, on-line, the optimal operating policy when there are batch to batch variations in initial conditions and process disturbances.