Fuzzy based fault detection and isolation of super heater faults in industrial boiler

Abstract

This paper addresses new concept of fuzzy logic based fault Detection and Isolation (FDI) of Boiler Super heater. Here the convection type super heater which is placed in the path of hot flue gases is considered. The FDI scheme detects and isolates the normal problems occurring in the super heater like Steam tube cracking at welding points, steam tube rupture, blockage inside tubes, scale formation inside the tubes, external ash deposits and high flue gas temperature. The inherent characteristic of non-linear behavior of super heater makes fuzzy logic suitable for FDI of various faults. Fault detection is done by generating residuals, which is the difference between real plant output and fuzzy model output. These fuzzy model outputs are got with the decision making logic in the inference engine which is formed with operator's knowledge, data collected from the power plant and simulation results. Fuzzy knowledge base or Inference engine consists of set of if-then rules, which gives strong insight in to the process and estimates the output for the given input. Residuals will be generated for two output parameters considered very important. If the residual surpasses threshold value which is calculated for on account of process noise, disturbances and uncertainties indicates various types of faults in the super heater. To isolate the faults, fuzzy inference rules are formed by considering magnitude of residual of the two outputs. The faults are isolated based on severity of faults expressed in percentage for various combinations of range of residual. More the severity more will be the fault inside the super heater, which are isolated based on the various ranges of fault severity. The required data for the study is collected from 55 tons per hour capacity water tube boiler available in Madras fertilizer Ltd.(MFL).