Distributed Model Predictive Control for the Atmospheric and Vacuum Distillation Towers in a Petroleum Refining Process

This paper develops a distributed model predictive control strategy for the atmospheric and vacuum distillation tower, which constitutes a key distillation process involved in refining petroleum. When considering an MPC implementation, computational complexity can be reduced and flexibility improved if the system is first decomposed into multiple smaller dimensional subsystems. Optimally exploiting the functionality and structural characteristics of the modern computer networks available in the industry, a novel distributed predictive control algorithm is developed for the atmospheric and vacuum tower system, which is assumed to be part of a wider system comprised of a number of sub-systems connected in series. For each subsystem, given the availability of mutual communication channels between subsystems and by using an iterative calculation approach, it will be seen that Nash optimality can be achieved. A low-cost solution that is readily implementable online is seen to achieve the control objective. The effectiveness of the approach presented in the paper is validated by the results of nonlinear simulation experiments.