0 votes
174 views
in The Hindu Summary for Competitive Exams by (5.3k points)

New technique can say where spent nuclear fuel came from

Scientists in China have developed a technique to reliably identify whether some nuclear fuel originated in one of two common kinds of nuclear reactors, a difficult task in nuclear forensics, using experimental data and machine-learning (ML).

Their work was published in Physical Review Applied on March 9.

Nuclear fuel is a highly regulated material because of its destructive potential. Countries maintain detailed inventories to safeguard it.

Nuclear forensics use analytical methods to identify the origins of nuclear materials and whether they were used for military applications.

Spent fuel from boiling water reactors (BWRs) is hard to differentiate from that from pressurised water reactors (PWRs) because both “use water as moderator and have similar thermal neutron spectra, so they are quite similar in neutron reaction mechanism,” Shengli Chen, assistant professor at the Sun Yat-Sen University, Guangdong, and a co-author of the study, told The Hindu by email.

Experimental data

The group used experimental data instead of simulations, whose accuracy is unknown.

The reactor type, the fuel’s exposure time inside the reactor, and the extent of the fuel’s enrichment can uniquely identify spent nuclear fuel.

Using a database that contains the composition of different isotopes in spent fuel from reactors over 50 years, the researchers developed equations that related these quantities to each other.

If one quantity can be determined through other measurements such as gamma-ray emissions from the spent fuel, the values of the other two quantities can be calculated.

They also trained data from the database to develop three ML models to distinguish fuel from BWRs from that from PWRs.

In BWRs, the fuel rods are submerged in water. When the fuel fissions, the water boils and the steam drives a turbine. In PWRs, the fuel rods aren’t exposed to the water; only the heat is exchanged.

The group’s ML models correctly identified 91% of the fuel from BWRs, and 95% the of fuel from PWRs.

“There was no explicit information about the accuracy of discriminating between PWRs and BWRs in older papers,” said Dr. Chen.

“Previous works, based on simulation data,” could differentiate between reactors, whereas “our work, based on experimental data, explicitly concludes which reactor type has been used to irradiate the sample,” added Dr. Chen.

1 Answer

0 votes
by (5.3k points)
The scientists in China have developed a technique that can accurately identify whether some nuclear fuel originated in one of two common types of nuclear reactors using experimental data and machine-learning (ML). This is a challenging task in nuclear forensics, which use analytical methods to identify the origins of nuclear materials and whether they were used for military applications. Spent fuel from boiling water reactors (BWRs) is difficult to differentiate from that from pressurized water reactors (PWRs) because they use water as moderator and have similar thermal neutron spectra. The researchers used a database containing the composition of different isotopes in spent fuel from reactors over 50 years and developed equations that related these quantities to each other. They also trained data from the database to develop three ML models to distinguish fuel from BWRs from that from PWRs. The ML models correctly identified 91% of the fuel from BWRs and 95% of the fuel from PWRs.
...