Enhancing cross-correlation analysis with artificial neural networks for nuclear power plant feedwater flow measurement

One of the primary cost-saving objectives of the power plant industry, including the nuclear industry, has long been the efficient operation of plant systems. Since the maximum operating thermal power of any nuclear plant is bounded by the specific licensing requirements of the competent national nuclear authorities, the amount of uncertainty in its calculation has a direct effect on the maximum energy that can be produced. The feedwater flow rate is one of the major quantities used in the thermal power calculations, and Venturi flow meters are traditionally used to measure this flow rate. Venturi flow meters are subject to known drifting problems, mainly originating from the gradual fouling of the Venturi constriction, which leads to an overestimation of the flow, and, consequently, of the produced thermal power. This overestimation has then to be balanced by a reduction of the operating power in order to stay within authority limits. This paper reports on preliminary results of a computational intelligence approach to the enhancement of the accuracy of cross-correlation flow measurements as a potential solution to the problem of Venturi fouling.