Fault diagnostic algorithm for precut fractionation column

Abstract

This paper presents an algorithm which can be used to detect and diagnose unexpected process faults in the operation of fatty acid precut fractionation column. The fault diagnostic algorithm is supported by the process history based method and developed by using Borland C++ Builder 6.0. The fatty acid precut fractionation column chosen as a case study is modeled by the commercial simulator, HYSYS.Plant. The discriminator for the detection section is developed by using statistical techniques, where the control limits for each selected monitoring variable were represented in 'High', 'Normal', and 'Low' discrete. Hazard and Operability Study (HAZOP) is used to support the diagnosis task. The algorithm has been successful in detecting the deviations of each variable by testing the data set. The tested data is used to interpret the pattern of the chart, where fault is considered to occur if one variable is out of control limits. The system promptly diagnoses the deviations and gives useful guidance to the user by displaying the causes and consequences of the faults.