(446b) Industrial Experiences of On-Line Fault Detection and Process Monitoring

Process chemometrics (chemometrics and multivariate statistical modeling methods applied to chemical process data and manufacturing quality data) is a technology that has gained increasing significance in Dow Chemical’s plant troubleshooting and proactive quality improvement efforts.  Process upsets and product quality issues often result in loss of asset capability and/or customer concerns.  When these unfavorable conditions happen, it is important to effectively and promptly identify the root cause(s) to recover to normal operating conditions.   Solving these problems has lead to multimillion dollar savings for the company.

This presentation will show several examples of successful application of On-line Fault Detection and Process Monitoring using multivariate techniques such as Principal Component Analysis (PCA) and Partial Least Squares (PLS). The selected examples come from batch and continuous processes and show the versatility of the technology.   Practical issues such as variable selection, outlier detection, on-line model implementation, and operation training will be discussed as well.