Fault propagation analysis of oscillations in control loops by combining process causality and topology

Disturbances in large-scale industrial systems can easily propagate through the process units and thereby adversely affect the overall process performance. In particular, oscillations in control loops are very common in industrial processes and lead to poor control performance, an excessive energy consumption and deteriorate the product quality. Identifying the propagation path of oscillations is of great importance due to its ability to disclose the root cause and identify the process units of concern which should be monitored closely. This paper presents a technique for identifying the propagation path of oscillations in control loops using a dedicated search algorithm. The algorithm combines the quantitative results from the data-driven causal analysis and the topology-based model in the form of a connectivity matrix in order to yield an enhanced causal model which illustrates the propagation path. The algorithm has two functionalities: one is finding feasible propagation paths among two elements and second is determining whether each path which had been found is direct or indirect. The analysis is demonstrated on an industrial paperboard machine with multiple oscillations in its drying section due to valve stiction. First, the connectivity matrix is extracted from a P&ID. Secondly, the causality matrix is obtained using the Granger causality method which is then refined using the search algorithm based on the connectivity matrix. Finally, the causal model illustrating the propagation path is given and the results are evaluated.