Abstract

In order to close nuclear fuel cycle and address the problem of sustainability, advanced nuclear reactor systems of the fourth generation are in the focus of the research for many years. With a simple goal of supporting this research, machine learning-based methodology for the assessment of the Doppler reactivity has been developed and applied to the European Sodium Fast Reactor (ESFR) in the frame of the ESFR-Safety Measures Assessment and Research Tools (SMART) Horizon-2020 project. In the scope of this study, a database of the precise Monte Carlo (MC) calculations was prepared and used to train artificial neural network (ANN) as a surrogate model to assess the Doppler reactivity across the range of reactor conditions that could occur throughout the life of the reactor core, in fast, yet accurate manner. The database was generated for all the combinations of several core parameters carefully predefined in order to account for both operational and accidental states of the core. Subsequently, Doppler reactivity change as a function of the above-mentioned parameters was assessed by herein developed methodology, as well as by widely used logarithmic dependence of the Doppler reactivity on the fuel temperature and compared to the results of the precise MC simulations. This study proves that, if certain computational resources are allocated to the database generation and ANN training, newly developed methodology yields similar or even more accurate results compared to the classical methodology and at the same time provides a tool for parameterization and interpolation of Doppler reactivity not only on the fuel temperature but also on the other parameters characterizing core of the sodium-cooled fast reactor (SFR).

References

1.
Rineiski
,
A.
,
Meriot
,
C.
,
Coquelet
,
C.
,
Krepel
,
J.
,
Fridman
,
E.
, and
Mikityuk
,
K.
,
2018
, “
Specification of the New Core Safety Measures
,”
Horizon-2020
ESFR-SMART, Zenodo.10.5281/zenodo.1990703
2.
Krepel
,
J.
, and
Losa
,
E.
,
2017
, “
Comparison of Fast Reactors Performance in the Closed U-Pu and Th-U Cycle
,”
Proceedings of International Conference on Fast Reactors and Related Fuel Cycles: Next Generation Nuclear Systems for Sustainable Development
, Yekaterinburg, Russian Federation, Paper No. IAEA-CN245-052.
3.
Leppänen
,
J.
,
Pusa
,
M.
,
Viitanen
,
T.
,
Valtavirta
,
V.
, and
Kaltiaisenaho
,
T.
,
2015
, “
The Serpent Monte Carlo Code: Status, Development and Applications in 2013
,”
Ann. Nucl. Energy
,
82
, pp.
142
150
.10.1016/j.anucene.2014.08.024
4.
MathWorks, 2019, “MATLAB and Deep Learning Toolbox Release 2019b,” The MathWorks, Inc., Natick, MA.
You do not currently have access to this content.