Record Details
A computer control algorithm for a nuclear reactor support system
ScholarsArchive at Oregon State University
| Field | Value |
|---|---|
| Title | A computer control algorithm for a nuclear reactor support system |
| Names |
Kosorok, John Roger
(creator) Stone, Louis N. (advisor) |
| Date Issued | 1968-05-09 (iso8601) |
| Note | Graduation date: 1968 |
| Abstract | The High Temperature Lattice Test Reactor (HTLTR) at Richland, Washington, operates at temperatures up to 1000°C. Its low nuclear power requires a 384 kilowatt electrical heating system, that is divided into four circuits, to attain the high temperatures in the ten-foot cube of moderating graphite. This thesis describes the design and evaluation of an algorithm for digital computer control of the heating system. Design specifications for the algorithm were: (1) limited computer memory was available; (2) the ranges for the manipulated variables were limited and were different; (3) the temperature differences throughout the reactor were to be small; (4) the response was to be non-oscillatory; (5) there should be a small steady-state error; and (6) it was necessary to compensate for nonlinearities, The distributed nature of the heating system, its multitude of inputs and outputs, and its non-linearity over the operating range formed a process that was extremely difficult to describe without directly measuring the transfer function. Since the completion of the control program had to coincide with the completion of the reactor, the design of the algorithm could not wait until direct measurements on the system were made. A preliminary analysis of the reactor heating system showed that the design specifications could be met by using a discrete approximation to a continuous proportional-plus-integral controller as the framework around which to design the algorithm. The extensive logic and calculation power of the digital control computer provided the means to extend the basic form of the algorithm to meet the design specifications. A hybrid computer simulation of the heating system was designed for verifying the algorithm and for refining it to compensate for the interactions between circuits of the heating system. Important features of the simulation were: a program to optimize the control parameters by a "hill-climbing" method; and an operating program that allowed extensive changes in the digital portion of the simulation program while it was running. After the reactor was constructed, transfer functions for the heating system were obtained by step function tests to complete the hybrid simulation. In addition to amplitude scaling, time scaling by a factor of 3000 was necessary to obtain reasonable problem solution times for the control system. Extensive tests with the simulation showed that the first five design specifications were met, and that the system was stable. Extensive testing with the real system, which may take months to complete, is required to verify that the sixth specification will be met. However operating experience has shown that the feature of the algorithm to meet this specification will be necessary. General conclusions, that became obvious during the research on this thesis problem, were: simulation should be considered as a primary tool for the practical solution of complex design problems; and more development effort will be required to make advanced control concepts applicable to practical problems. |
| Genre | Thesis/Dissertation |
| Topic | Nuclear reactors -- Computer programs |
| Identifier | http://hdl.handle.net/1957/46714 |
