Home > Downloads > Performance Analysis of Model Based Predictive Controllers, an Industrial Case Study

All Downloads

MatrikonOPC Drivers

Portal Login

Enter your email address to receive updates about new developments.

Performance Analysis of Model Based Predictive Controllers, an Industrial Case Study

Performance analysis of dynamic matrix controllers (DMC) is a challenging research problem of significant industrial relevance. The objectives of this product included, (i) establishing and applying multivariate performance measures, (ii) isolating the causes of poor controller performance, and (iii) suggesting remedial actions, for the propylene splitter DMC. Routine operating data was available from three different operating conditions - normal, high and low load conditions. Multivariable minimum variance benchmarking, the design vs. achieved objective function approach, and the historical benchmarking methods were proposed for evaluating DMC. Controller performance was found to be poor in the low and high load regions. Diagnosis of poor performance was approached on a case study basis. The role of constraints in analyzing performance is emphasized through this case study. Activation of constraints was found to be responsible for loss of performance in the low load region. Oscillations were observed in the setpoints and some implementational issues peculiar to the DMC structure were found responsible.

Get the full Technical Paper Now!

Details:
Technical Paper (222 KB , PDF)
Author: Rohit S. Patwardhan, Sirish L. Shah, GenichiEmoto, Hiroyuki Fujii

Description:
Matrikon™ Control Performance Monitor Technical Paper.