Doctoral Dissertations

Evren Eryurek

Date of Award

8-1994

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Major

Nuclear Engineering

Major Professor

Belle R. Upadhyaya

Committee Members

Rafael B. Perez, J.A.M. Boulet

Abstract

A new integrated fault-tolerant control and diagnostics system was developed and applied to the feedwater flow regulation system of a pressurized water reactor (PWR). Control, signal and command validation, monitoring and diagnostic tasks were integrated into one large-scale system.

The control module of the Fault-Tolerant Control and Diagnostics System (FCDS) includes two nonlinear control algorithms, in addition to conventional controllers. A software-based parallelism was implemented in the design of the control module. The parallel control includes a model-based adaptive controller, two fuzzy logic con- trollers, and proportional-integral (PI) controller. Each algorithm was designed to handle the same control task using different strategies and different sets of plant measurements.

Using family of control solutions requires a systematic approach to select the most suitable control action for the present demand. A new concept, called Command Validation, was developed as part of the Fault-Tolerant Control and Diagnostics System. Command validation is a prediction scheme in which the objective is to provide an estimate of the expected control action using relevant plant variables. The prediction is then used to cross-examine the output of the control algorithm before the final decision is made by the Decision-Making module. This approach enables the control system to detect anomalies in the sensors, actuators and in the control actions.

A Signal Validation module was also developed for validating the plant measurements before they were utilized by the control algorithms. A software-based parallelism was also included in the signal validation and command validation modules by incorporating two different modeling techniques: process empirical models and artificial neural network models. The estimations of selected plant signals were made via a set of process measurements and the developed models.

The FCDS was tested using an application to the feedwater flow regulation sys- tem of a typical four-loop Westinghouse type PWR. In order to test the system, a nonlinear feedwater flow system was modeled with 96 state equations. The model includes four steam generators, two main feed pumps, piping and their control systems. There are two control systems in the feedwater flow regulation of a PWR: the steam generator water level controller and the main feed pump speed controller. Both of these controllers were included in the model.

Reconstructive Inverse Dynamics (RID) and the Fuzzy Logic controllers were devel- oped for water level control and pump speed control systems. The RID controller is a model-based controller and is found to be exceptionally accurate if the control problem is of a trajectory-following type. The fuzzy logic controllers give excellent results during unexpected changes in the behavior of the plant as well as during trajectory-following problems. An extensive testing of the fuzzy logic controller was performed to demonstrate the robustness of the fuzzy logic control approach, since the application of this technique is considerably new in the nuclear industry. The development of fuzzy logic controllers for such highly nonlinear systems was one of the objectives of this dissertation.

The utilization of more than one algorithm not only improves the availability of the control system during sensor failures or inadequacies in the control algorithms, but also provides alternate approaches for different plant operating conditions. The Fault-Tolerant Control and Diagnostics System improves the availability of the control system and provides fault-tolerance through multiple control strategies and the incorporation of signal and command validation features.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS