Multi-output radial basis function interpolation of saturated steam table for full scope simulator design

Abstract

Full scope simulators in nuclear or fossil power plants rely heavily on steam tables since most of their equipment operate in the saturation pressure/temperature region. The process of simulating the behavior of such plants involves trade-off between simulation cycle time and the available memory space. It is difficult to store entire steam table in the memory as they may adversely affect the computation and space complexity. This paper presents a novel approach of interpolating the steam table by radial basis function network (RBFN) that overcomes the shortcomings of storing the entire table. The paper uses multi-output RBFN architecture. The paper further shows applicability of forward bulk orthogonal least squares (OLS) algorithm for optimum center selection of RBFN. Various radial basis functions are tested for the regression purpose and their corresponding mean square errors are compared with existing ANSI standard.