A method for systematic interpretation of dynamic measurements in a high temperature gas-cooled reactor
Large linear dynamic models for nuclear reactor systems are widely used for simulation and control system design. It is important to be able to verify these models and the parameters in them. Existing parameter identification techniques are very time-consuming for use with large systems. This study presents the development and application of a new technique for practical parameter identification of systems whose dynamics are described by up to 110 first order, linear differential and algebraic equations. A partitioned matrix technique was developed for this purpose. This technique which takes advantage of the matrix form of the system equations, constitutes a very efficient analysis algorithm when implemented on the digital computer.
The identification procedure is to be used to analyze test results from the Fort St. Vrain high temperature, gas cooled reactor.
In order to validate the important physical parameters in the overall system model of Fort St. Vrain, reliable models for each component in the system are necessary. A linear, three node moving boundary model of a once-through steam generator for the Fort St. Vrain nuclear plant was developed for use with other subsystem models developed .at the General Atomic Company.
To demonstrate the efficiency of the proposed technique which was used in the parameter identification algorithm in this study, three system models were analyzed: (a) Oconee I nuclear system described by twenty-nine differential equations; (b) Fort St. Vrain steam generator, described by thirteen differential and seven algebraic equations, (c) Fort St. Vrain nuclear steam supply system, described by fifty-eight differential and twenty-three algebraic equations. Computer programs were written in FORTRAN IV to make the necessary calculations; the programs were executed on the IBM 360 Model 65 computer.
The accuracy and computing time of the new technique were studied. Also, the use of the Fisher information matrix for answering the question of parameter identifiability was studied. The information matrix which was generated could be used to evaluate the goodness of the parameters estimated. Using the new technique, the simulated analysis of the 8lst order model of the Fort St. Vrain nuclear system required only 12 minutes and 50 seconds on the IBM 360 Model 65 computer.
Thesis76b.C325.pdf
5.67 MB
Adobe PDF
c1e714463400f34e723f5ada8fcaeb64