Contributions to a methodology for periodical verification of the parameters of the control systems at Cernavoda Nuclear plant Unit 1

A model identification methodology for periodical verification of the regulating system parameters at Cernavoda NPP Unit 1 was developed. As support to this methodology, the computer program MODELIDENT was implemented in the Java programming language. This program is used for off-line evaluation of the real regulating systems characteristic parameters using an identification algorithm which takes as input data the system response collected for different input excitation signals, a structurally similar model of the analyzed regulating system, and some starting guess value of the unknown parameters. The real values of the parameters are determined during MODELIDENT program execution by applying an iterative algorithm and afterwards are retained as nominal reference values. The success of the identification algorithm is strongly dependent on how appropriately the structure of model's transfer function is chosen. By repeating periodically the identification method, using newly collected data from the process, the current value of the parameters are determined. Any deviations of the new values relative to the nominal reference values are interpreted as de-calibration of the control equipment and in this case corrective maintenance actions have to be taken. With the implementation of the presented methodology at Cernavoda NPP Unit 1 we can make the statement that the preventive maintenance activity is gaining a predictive feature, which can lead to the elimination of major degradation possibilities in the performances of the RS equipment and consequently to increase the NPP availability. On the basis of the experience gained in the practical application of the presented methodology we expect that the identification method will also have beneficial effects in the optimal control of the process systems and also in the activity of Full Scope Simulator software maintenance (the reference values of the identified parameters being used for fine tuning of the simulation models. (authors)