Monday, November 9, 2015
Exhibit Hall 1 (Salt Palace Convention Center)
In recent years, the need to maintain safe and reliable industrial operations has taken great proportions. In the continuous process industry, in particular, this issue is even more important because the operation is often carried in full capacity, with high production speed, huge impact downtimes and low production flexibility. This scenario has stimulated the development of efficient tools for fault detection. Statistical process control (SPC) techniques are widely used in the industry of manufactured products, where their basic assumptions are often valid: independent and identically distributed measurements with normal distribution. However, in the industrial environment of continuous processes these assumptions are often violated. Thus, this study focused on the adaption of SPC techniques to continuous processes. A methodology for identifying intervals under control of a time series was proposed. Furthermore, a method for updating the monitoring parameters was defined according to the current state of the process and past operational points. Natural gas compression systems of offshore platforms were chosen as case studies using real and simulated data. The results demonstrate the ability of the techniques to detect abnormalities. It is also shown that the identification of stable intervals is adequate and that the methodology of updating parameters along the monitoring is efficient.