Year 2014
Vesa-Matti Tikkala

Integrated fault fetection system for a board machine

The current process industry faces remarkable challenges due to global competition, tightening environmental regulations, and the increasing complexity and integration of process plants. Especially, the pulp and paper industry has been under pressure in recent years to improve the efficiency of operations and to optimize production. Managing abnormal events, such as disturbances, faults and failures is an essential part of improving the operation of process plants. Traditional plant automation systems are able to handle the typical faults and disturbances and to restore the process into a normal state. However, in order to address more complex faults, the automation systems must be accompanied by fault detection methods which provide the plant operators and maintenance with additional information about the faults. This thesis presents the development of an integrated fault detection system for a board machine. The system was developed according to a created methodology which exploited the decomposition and control strategy of the process as well as fault analysis. The presented fault detection system consisted of four fault detection algorithms that addressed the faults having the most significant effect on the economic performance and operability of the process. The fault detection system comprised of a valve stiction detection system employing a parallel configuration of four different stiction detection algorithms, a robust detection method for non-stationary oscillations, a dynamic causal digraph -based method for detecting consistency sensor faults, a detection method for leakages and blockages in the drying section using non-linear parity equations, and a self-organising map -based process monitoring method for detecting caliper sensor fouling. The individual fault detection algorithms were tested and validated in case studies using simulations and industrial data. In addition, industrial experiments were carried out at the board machine. The obtained results were very promising and showed that the presented methodology provided a systematic approach to the development of a fault detection system. The testing results indicated that the fault detection algorithms provide useful information for improving the operation and maintenance of the board machine.

Thesis electronical version can be downloaded from here

This info last modified 28 Apr 2024 by Jukka Kortela