Leak detection in an experimental heat exchanger process: a multiple model approach

An application of a multiple model nonlinear filtering technique is discussed for the detection of leaks in a heat-exchanger process. The results of simulation study and an experimental study on a laboratory heat-exchanger process are presented. A simple decision logic scheme (threshold crossing) is used to process the outputs of the leak-detection algorithm to indicate the status of the process, i.e. whether or not there is a leak. The results of the study provide preliminary indication of the feasibility of the proposed diagnostic technique to detect small leaks in an experimental heat-exchanger process. The study also indicates that the technique is robust in the midst of significant flow disturbances on the inlet water flow, using only simple models for processing of the outlet temperature data, but is sensitive to the aggregate heat transfer parameter used in the detection algorithm