Deep learning approach to nuclear fuel transmutation in a fuel cycle simulator

Jin Whan Bae, Andrei Rykhlevskii, Gwendolyn Chee, Kathryn D. Huff

Research output: Contribution to journalArticle

Abstract

We trained a neural network model to predict Pressurized Water Reactor (PWR) Used Nuclear Fuel (UNF) composition given initial enrichment and burnup. This quick, flexible, medium-fidelity method to estimate depleted PWR fuel assembly compositions is used to model scenarios in which the PWR fuel burnup and enrichment vary over time. The Used Nuclear Fuel Storage, Transportation & Disposal Analysis Resource and Data System (UNF-ST&DARDS) Unified Database (UDB) provided a ground truth on which the model trained. We validated the model by comparing the U.S. UNF inventory profile predicted by the model with the UDB UNF inventory profile. The neural network yields less than 1% error for UNF inventory decay heat and activity and less than 2% error for major isotopic inventory. The neural network model takes 0.27 s for 100 predictions, compared to 118 s for 100 Oak Ridge Isotope GENeration (ORIGEN) calculations. We also implemented this model into CYCLUS, an agent-based Nuclear Fuel Cycle (NFC) simulator, to perform rapid, medium-fidelity PWR depletion calculations. This model also allows discharge of batches with assemblies of varying burnup. Since the original private data cannot be retrieved from the model, this trained model can provide open-source depletion capabilities to NFC simulators. We show that training an artificial neural network with a dataset from a complex fuel depletion model can provide rapid, medium-fidelity depletion capabilities to large-scale fuel cycle simulations.

Original languageEnglish (US)
Article number107230
JournalAnnals of Nuclear Energy
Volume139
DOIs
StatePublished - May 2020

Fingerprint

Nuclear fuels
Simulators
Pressurized water reactors
Neural networks
Deep learning
Fuel storage
Chemical analysis
Isotopes

Keywords

  • Artificial neural network
  • Machine learning
  • Nuclear fuel cycle
  • Simulation
  • Spent nuclear fuel

ASJC Scopus subject areas

  • Nuclear Energy and Engineering

Cite this

Deep learning approach to nuclear fuel transmutation in a fuel cycle simulator. / Bae, Jin Whan; Rykhlevskii, Andrei; Chee, Gwendolyn; Huff, Kathryn D.

In: Annals of Nuclear Energy, Vol. 139, 107230, 05.2020.

Research output: Contribution to journalArticle

@article{3066bc62b8b74210ace2645427c345b4,
title = "Deep learning approach to nuclear fuel transmutation in a fuel cycle simulator",
abstract = "We trained a neural network model to predict Pressurized Water Reactor (PWR) Used Nuclear Fuel (UNF) composition given initial enrichment and burnup. This quick, flexible, medium-fidelity method to estimate depleted PWR fuel assembly compositions is used to model scenarios in which the PWR fuel burnup and enrichment vary over time. The Used Nuclear Fuel Storage, Transportation & Disposal Analysis Resource and Data System (UNF-ST&DARDS) Unified Database (UDB) provided a ground truth on which the model trained. We validated the model by comparing the U.S. UNF inventory profile predicted by the model with the UDB UNF inventory profile. The neural network yields less than 1{\%} error for UNF inventory decay heat and activity and less than 2{\%} error for major isotopic inventory. The neural network model takes 0.27 s for 100 predictions, compared to 118 s for 100 Oak Ridge Isotope GENeration (ORIGEN) calculations. We also implemented this model into CYCLUS, an agent-based Nuclear Fuel Cycle (NFC) simulator, to perform rapid, medium-fidelity PWR depletion calculations. This model also allows discharge of batches with assemblies of varying burnup. Since the original private data cannot be retrieved from the model, this trained model can provide open-source depletion capabilities to NFC simulators. We show that training an artificial neural network with a dataset from a complex fuel depletion model can provide rapid, medium-fidelity depletion capabilities to large-scale fuel cycle simulations.",
keywords = "Artificial neural network, Machine learning, Nuclear fuel cycle, Simulation, Spent nuclear fuel",
author = "Bae, {Jin Whan} and Andrei Rykhlevskii and Gwendolyn Chee and Huff, {Kathryn D.}",
year = "2020",
month = "5",
doi = "10.1016/j.anucene.2019.107230",
language = "English (US)",
volume = "139",
journal = "Annals of Nuclear Energy",
issn = "0306-4549",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Deep learning approach to nuclear fuel transmutation in a fuel cycle simulator

AU - Bae, Jin Whan

AU - Rykhlevskii, Andrei

AU - Chee, Gwendolyn

AU - Huff, Kathryn D.

PY - 2020/5

Y1 - 2020/5

N2 - We trained a neural network model to predict Pressurized Water Reactor (PWR) Used Nuclear Fuel (UNF) composition given initial enrichment and burnup. This quick, flexible, medium-fidelity method to estimate depleted PWR fuel assembly compositions is used to model scenarios in which the PWR fuel burnup and enrichment vary over time. The Used Nuclear Fuel Storage, Transportation & Disposal Analysis Resource and Data System (UNF-ST&DARDS) Unified Database (UDB) provided a ground truth on which the model trained. We validated the model by comparing the U.S. UNF inventory profile predicted by the model with the UDB UNF inventory profile. The neural network yields less than 1% error for UNF inventory decay heat and activity and less than 2% error for major isotopic inventory. The neural network model takes 0.27 s for 100 predictions, compared to 118 s for 100 Oak Ridge Isotope GENeration (ORIGEN) calculations. We also implemented this model into CYCLUS, an agent-based Nuclear Fuel Cycle (NFC) simulator, to perform rapid, medium-fidelity PWR depletion calculations. This model also allows discharge of batches with assemblies of varying burnup. Since the original private data cannot be retrieved from the model, this trained model can provide open-source depletion capabilities to NFC simulators. We show that training an artificial neural network with a dataset from a complex fuel depletion model can provide rapid, medium-fidelity depletion capabilities to large-scale fuel cycle simulations.

AB - We trained a neural network model to predict Pressurized Water Reactor (PWR) Used Nuclear Fuel (UNF) composition given initial enrichment and burnup. This quick, flexible, medium-fidelity method to estimate depleted PWR fuel assembly compositions is used to model scenarios in which the PWR fuel burnup and enrichment vary over time. The Used Nuclear Fuel Storage, Transportation & Disposal Analysis Resource and Data System (UNF-ST&DARDS) Unified Database (UDB) provided a ground truth on which the model trained. We validated the model by comparing the U.S. UNF inventory profile predicted by the model with the UDB UNF inventory profile. The neural network yields less than 1% error for UNF inventory decay heat and activity and less than 2% error for major isotopic inventory. The neural network model takes 0.27 s for 100 predictions, compared to 118 s for 100 Oak Ridge Isotope GENeration (ORIGEN) calculations. We also implemented this model into CYCLUS, an agent-based Nuclear Fuel Cycle (NFC) simulator, to perform rapid, medium-fidelity PWR depletion calculations. This model also allows discharge of batches with assemblies of varying burnup. Since the original private data cannot be retrieved from the model, this trained model can provide open-source depletion capabilities to NFC simulators. We show that training an artificial neural network with a dataset from a complex fuel depletion model can provide rapid, medium-fidelity depletion capabilities to large-scale fuel cycle simulations.

KW - Artificial neural network

KW - Machine learning

KW - Nuclear fuel cycle

KW - Simulation

KW - Spent nuclear fuel

UR - http://www.scopus.com/inward/record.url?scp=85076237414&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85076237414&partnerID=8YFLogxK

U2 - 10.1016/j.anucene.2019.107230

DO - 10.1016/j.anucene.2019.107230

M3 - Article

AN - SCOPUS:85076237414

VL - 139

JO - Annals of Nuclear Energy

JF - Annals of Nuclear Energy

SN - 0306-4549

M1 - 107230

ER -