Optimisation of the Koeberg nuclear power plant controls by implementing customised transfer functions
Abstract
Koeberg Nuclear Power Station, like most of the current operating plants, was built with the technology of the 1970s and 1980s. Most of the plants build around that time were using analog technology (IAEA, 1999). The shift in technological development has led to the progression of digital technology which resulted in the analog Control and Instrumentation (C&I) Systems being replaced with digital C&I Systems. While the analog C&I Systems have proven to be safe and operable for years, digital technology is not only safe but also has advantages over the analog system. That is, digital systems are free of drift, process data at a faster rate, have higher data storage capacity, and are easy to troubleshoot, calibrate and maintain their calibration better (IAEA, 1999).
The new, advanced Control and Instrumentation systems have been implemented successfully in Nuclear Power Plants (NPPs) around the world using digital technology. Almost all NPPs that operate in North America, Europe and Asia are partially using digital C&I systems (IAEA, 1999). The deployment of digital C&I systems has allowed these plants to operate more productively and efficiently than the old analog C&I systems. The use of digital C&I systems is estimated to reduce C&I - related operations and maintenance costs by 10% and increase plant power output by 5% (IAEA, 1999).
Koeberg Power Plant (KPP) is replacing its aging analog C&I systems with digital ones. Some of the analog C&I systems have already been replaced with digital C&I systems which includes the Rod Drive Control System and the Generator Control and Governing System. The KPP management has established an Engineering Department responsible for replacing the analog C&I System with digital ones.
Optimization of the selected Koeberg Power Plant Controls by implementing the customised Transfer Functions has been studied. The four KPP Controllers that have been selected for this study are the Primary Temperature Controller, Pressurizer Pressure Controller, Pressurizer Level Controller and Steam Generator Level Controller (Eskom, 2008). The simulation software Matlab® has been used for analysis of the current KPP analog controllers and for the optimization and analysis of digital controllers. This study intention is to explore the possibility of optimizing the old controllers to obtain better performing controllers in digital form. This was achieved by first obtaining the Koeberg Power Plant analog controllers from the KPP manuals and by using the process variables or plant dynamic equations obtained from literature survey. The four controllers are analysed using Matlab® Simulation software to obtain four performance values which are the overshoot, rise time, peak time and settling time. Optimization of the KPP controller is achieved by developing new controllers using PIDTUNER which is a Matlab® optimization function used to develop optimize both analog and digital controllers. The new controllers are obtained in digital form and analysis is done to obtain similar performances which are mentioned above. Comparative study has been done to determine the performance of these two types on controllers. Verification is performed for the two controllers which are the Digital Cascaded Steam Generator Level Control (SGLC) controller and Pressurizer Level Controller.
Optimizations of the KPP Controls by implementing customised transfer functions have been achieved. The developed digital transfer functions perform better when compared with the current analog controllers. The developed optimized digital controller have the settling times of the has less than 3 minutes which is reasonable compared to number of days provided analog controller.
Therefore, KPP could use the opportunity of digital controller upgrade program to implement optimized digital controllers when converting from analog C&I system to digital C&I systems. Furthermore, it would be interesting for KPP to consider additional studies, in order to verify, validate and improve the performance of the controllers. Literature study shows that it is possible to improve the performer of controllers by using a different method. The other optimization techniques, such as, the fuzzy-neural network, Zeiger Nicholes and Tyreus Luyben tuning could be employed.
Collections
- Engineering [1403]