Development of Computer Control Systems for Molecular Weight Distribution Control in Polymerization Reactors

近年来，合成聚合物在各个领域的应用越来越广泛，同时对其质量的要求也越来越高。聚合物产品的质量与分子量分布密切相关。本文主要研究了自由基聚合反应器中分子量分布的计算机控制问题。我们开发了一种方法，通过该方法我们得出适当的工作温度模式和引发剂的投料速度，以达到理想的平均链长和多分散性。该方法包括两个阶段。在第一阶段，获得瞬时平均链长和瞬时多分散度的分布，从而得到所需的随钻测井曲线。在第二阶段，根据得到的瞬时平均链长和瞬时多分散性曲线计算引发剂的温度分布和投料速度。为了在实际中实现所得到的温度模式，需要一个复杂的计算机控制系统。自由基聚合过程是典型的具有强非线性的化学过程之一。在非稳态操作下进行间歇聚合反应。这种时变过程特性和非线性使得经典PI控制难以跟踪得到的温度模式。针对系统的强非线性和时变特性，提出了两种全新的自适应控制方案（自适应内模控制、自适应推理控制系统）和模型预测控制方案。仿真计算和实验验证了这些控制方案的有效性。＊

The recent increase in the use of synthetic polymers in various fields has been accompanied by strict demand for higher and higher quality. The quality of polymer products is closely related with the molecular weigh distribution (MWD). This research focused on the computer control probelm of the molecular weight distribution in a free radical polymerization reactor.We developed a method by which we derive the proper poerating temperature pattern and feeding speed of initiator in order to achieve a desired average chain length and polydispersity. The method consists of two stages. In the first stage, the profiles of instantaneous average chain length and instantaneous polydispersity are obtained so as to produce the desired MWD. In the second stage, temperature pattern and feeding speed of initiator are calculated from the obtained profiles of instantaneous average chain length and instantaneous polydispersity.In order to realize the obtained temperature pattern in practice, a sophisticated computer control system is required. The free radical polymerization process is one of the typical chemical processes with the strong nonlinearities. Furthermore batch polymerization reaction is carried out in non-steady state operation. Such time-varying process characteristics and nonlinearity make it difficult for classical PI control to track the obtained temperature pattern. To deal with the strong nonlinearity and time-dependent properties, two brand-new adaptive control schemes (Adaptive Internal Model Control, Adaptive Inferential Control System) and model predictive control scheme are newly developed. Simulation calculation and experiments were performed to validate these control schemes. *