Data-based control of an industrial tubular reactor

Abstract This paper describes the empirical modelling and control of a tubular reactor producing alpha-olefins at the Mizushima (Japan) site of Mitsubishi Chemical Corporation (MCC). Initially, the reactor was operated manually resulting in inconsistent run-to-run performance. Using archived data from several past runs and by performing statistical and time-series analysis of the data sets, satisfactory and consistent automatic feedback and feedforward control of the reactors was achieved. An overview of the modelling efforts involved and the performance of the automatic control strategies is provided here.