Bias identification and estimation based on data reconciliation and first-principle model application to nuclear fuel recycling process.

A. Duterme (1) , M. Montuir (1) , B. Dinh (1) , J. Bisson (2) , N. Vigier (2) , P. Floquet (3) , X. Joulia (3)

1 CEA-DES (ex-DEN) - CEA-Direction des Energies (ex-Direction de l'Energie Nucléaire)
2 ORANO
3 Université Fédérale Toulouse Midi-Pyrénées

At La Hague plant, the PUREX process recycles spent nuclear fuel. Uranium and plutonium are recovered and purified by solvent extraction. The PAREX code simulates the partitioning of the species, the transfer and chemical kinetics. Industrially, high consideration is given to a specific set of hard sensors on main fluxes needed for operation, control and safety issues (multiple sensors, regular checking). A project has been launched in order to automatically estimate the process state thanks to data reconciliation and the knowledge-based simulator PAREX. The first step is to make use of secondary sensors. However, these additional data can have biases that cannot be detected with prior data. This paper offers a methodology to identify and estimate significant and numerous biases within connected fluxes.

Keywords: PUREX; PAREX

Domains: Physics [physics]; Nuclear Experiment [nucl-ex]; Physics [physics]; Nuclear Theory [nucl-th]

A. Duterme, M. Montuir, B. Dinh, J. Bisson, N. Vigier, et al.. Bias identification and estimation based on data reconciliation and first-principle model application to nuclear fuel recycling process.. ESCAPE 2019 - European Symposium on Computer Aided Process Engineering, Jun 2019, Eindhoven, Netherlands. ⟨cea-02338719⟩