Feedwater flow measurements: challenges, current solutions, and 'soft' developments

This report presents an early progress of a feasibility study of a computational intelligence approach to the enhancement of the accuracy of feedwater flow measurements in the framework of an ongoing cooperation between Tecnatom s.a. in Madrid and the OECD Halden Reactor Project (HRP) in Halden. The aim of this research project is to contribute to the development and validation of a flow sensor in a nuclear power plant (NPP). The basic idea is to combine the use of applied computational intelligence approaches (noise analysis, neural networks, fuzzy systems, wavelets etc.) with existing traditional flow measurements, and in particular with cross correlation flow meter concepts. In this report, Section 2 outlines relevant aspects of thermal power calculations on electrical power plants. Section 3 reviews from the available literature possible approaches and solutions for feedwater flow measurement, including ultrasonic flow meters, cross-correlation flow meters, and 'Virtural' flow meters with artificial neural networks. Section 4 reports typical experimental measurements at the Tecnatom's facility. Section 5 presents an integration approach and preliminary experimental tests. Section 6 discusses the role of soft computing techniques in the context of feedwater flow measurements related nuclear fields, and Section 7 highlights the future research direction. (Author)