(124a) PCA Based Blockage Detection in an Industrial Solid State Particle Slurry Stripper

Multivariate statistical methods are a well-established technique for abnormal situation management and early event detection (EED). Nevertheless, there are only few publications in the open literature describing real-time industrial applications of EED systems where results are presented to operators rather than advanced control experts. This paper presents the development and on-line deployment of a Principle Component Analysis (PCA) model based EED system for an industrial-scale solid state particle slurry stripper. The developed solution was designed to detect plugging or blockage of the stripping column trays earlier than it is possible using traditional monitoring techniques, and to avoid process disruption and production losses. The paper describes the project steps beginning with the offline development of the PCA model to the online implementation and utilization by operators and plant personnel. In some cases the system was able to detect upcoming plugging events days earlier than with the help of single-signal based alarming. Keys to success included the combination of data analysis and process knowledge, the development of an easy-to understand user interface, and the involvement of plant personnel beginning with the early project stages.