Trouble shooting and performance audit for a multivariable control system: an industrial case study

This case study is concerned with the application of statistical data analysis and model identification techniques in the de-bottlenecking of an industrial problem. The distillation column, which has a model predictive controller (MPG), frequently gets flooded when the advanced controller is on and hence the operator has to turn off the advanced controller which results in substantial decrease in the production rate. This case study focuses on several aspects of the trouble shooting: inferential model validation, cross-correlation analysis between the observation variable and inferential variable, and the performance analysis of the MPC controller.