Thesis for the Degree of Master of Science


Year 2008
Di Zhang

Plant-wide disturbance detection and diagnosis

The plant-wide disturbance can affect the product quality, manufacturing costs, process safety and environment. Early detection and diagnosis of the occurrence of an abnormal event in an operating plant help avoid these problems. The detection and diagnosis problem in single control loop has been well studied during the last two decades, and many methods have been used in process performance monitoring. However, neither the equipment nor control loops are isolated from each other; instead, they may have interactions. The poor performance of one control loop can be caused by a fault that has propagated from the other part of the process. If the origin of the plant-wide disturbance is identified, then the secondary fault will be determined without further actions. Hence plant-wide approach is receiving more and more attention.

The aim of this master thesis was to design a general plant-wide performance monitoring concept. The following modules are part of the system: data collection, plant-wide disturbance detection and root cause diagnosis.

In the literature part, the plant-wide disturbance detection methods are reviewed and then a summary of the root cause diagnosis methods is presented and discussed.

In the experimental part, the general concept of plant-wide process monitoring is proposed first. Then one specific plant-wide process monitoring case is introduced, and used to evaluate the concept. A paper making process implemented in the APROS Paper simulator is used as a testing platform. In the case study, the integral of absolute error (IAE) method and the auto covariance function (ACF) method are used for the plant-wide disturbance detection. The causal diagraph method is subsequently used for the root cause diagnosis. Finally, an analysis of the results achieved is provided and conclusions are drawn.

This info last modified 21 Aug 2018 by Jerri Kämpe-Hellenius