New Methodology for Bias Identification and Estimation – Application to Nuclear Fuel Recycling Process

Abstract

This paper focuses on the data reconciliation technique (DR) in case of numerous biases. DR improves the degree of confidence in available information and generates consistent data. The inventory and analysis of the plant data (position and type of sensors …) enable an evaluation of the process redundancy. Classical Gross Error Detection and Identification (GEDI) techniques delete the biased variables, decreasing the redundancy. This leads to information loss and possibly an inability to apply DR. The methodology proposed here combines DR, based on a reduced model, and rigorous simulations to locate and estimate multiple biases and to make data consistent in case of inter-connected flows. This methodology is applied to the nuclear fuel recycling process within the scope of a state estimation tool built on a process simulation code.