Operational Model Predictive Control on the Example of the Stabilization Process for Hydro Treatment of Oil Fractions

Abstract

AbstractA method is proposed for solving operational control problems by predicting the behavior of a model within the framework of the MPC approach, which is an algorithm for constructing using the discrete maximum principle based on statistical mathematical models. During the study, for the parametric identification of the object, the weighted least squares method with an orthogonal decomposition of the information matrix was used, which allows you to build a regression mathematical model of the process, which is updated in real-time. A simulation model is also built, based on which the proposed optimal control strategy is compared with the strategy for which the experimental sample was obtained. As an example of the implementation of the proposed technique, a technological object was considered—the process of stabilization of the hydrogenated product, which takes place in a stripping column and is one of the most important processes of oil refining and, in particular, a hydrotreating unit of a low-temperature isomerization unit. As a result of this process, the stream is not only cleaned of unnecessary components by separating from it dissolved light hydrocarbons, water, and hydrogen sulfide, but also does not poison the installations of subsequent processing stages. This object has many limitations associated with the mode of the technological process, as well as with the parametric uncertainty of the object under study. As the goal of management, the task was set to minimize energy consumption.KeywordsMPCCombined control systemRobust systemSystem robustnessPerturbation