Data-Based Modeling Methods for Fault Detection and Isolation in Heat Exchangers

Author

Aaron Patrick Sawyer, University of Tennessee, Knoxville

Date of Award

5-2004

Degree Type

Thesis

Degree Name

Master of Science

Major

Nuclear Engineering

Major Professor

Arthur E. Ruggles

Committee Members

Belle R. Upadhyaya, Lawrence W. Townsend

Abstract

A multivariate analysis method is developed for processing measurements, and for detecting and isolating faults and monitoring performance degradation in heat exchanger control loops. A heat exchanger in a typical temperature-to-flow cascade loop s considered. A mechanistic thermal-fluid model for the components in the system is developed and compared to an installed laboratory heat exchanger control loop. A supplemental model for condenser hear transfer is included. The mechanistic model generate data to develop a data driven model using the Group Method of Data Handling (GMDH) approach. The GMDH model matches the mechanistic model well.

A Fault Detection and Isolation (FDI) rule-base is formulated from results of simulations performed using these models. The rule base allows the identification of faults in a heat exchanger control loop given suitable process measurements. The mechanistic model matches the physical system performance well and is used to create a Fault Detection and Isolation (FDI) algorithm for the system.

Recommended Citation

Sawyer, Aaron Patrick, "Data-Based Modeling Methods for Fault Detection and Isolation in Heat Exchangers. " Master's Thesis, University of Tennessee, 2004.
https://trace.tennessee.edu/utk_gradthes/4799