Steady State and Transient State Identification in an Industrial Process

Abstract

A computationally simple method is demonstrated for automated identification of steady state and transient state in noisy process signals of an industrial-scale, single or multi-variable process. This steady state and transient state identification method uses the R-statistic method, which is a ratio of estimated variances, and independent of the process variance. It has been implemented for automated identification of steady state of a single variable water flow rate to an absorption column in the Unit Operations Lab and the multi-variable commercial scale distillation process in FRI. When there is an upset in the process the steady state identifier indicates so. Most often the visual identification of steady state agrees with the statistic-based method of identification of steady state. At the process where the noise is pronounced and confounds identification of steady state, the steady state identifier helps operators to interpret the data.