EPJ Nuclear Sci. Technol. 11, 26 (2025)
https://doi.org/10.1051/epjn/2025014

Regular Article

ASSAS project, Artificial intelligence for Simulation of Severe AccidentS; Simulator development and Isotopic Source Term proposals

¹ Radiation Engineering & Analysis, Westinghouse. Avda. Montes de Oca, 1, San Sebastián de los Reyes, 28703 Madrid, Spain
² Simulation and Modeling, Westinghouse. Avda. Montes de Oca, 1, San Sebastián de los Reyes, 28703, Madrid, Spain

^* e-mail: rafaeljuan.carobenito@westinghouse.com

Received: 13 September 2024
Received in final form: 4 February 2025
Accepted: 24 March 2025
Published online: 9 June 2025

Abstract

Nuclear Power Plant simulators play a central role in the training of nuclear operators. However, most simulators worldwide only cover Design Basis Accident, excluding the management of Severe Accidents. The need for a realistic training on such situations has been strengthened by the deployment of post-Fukushima safety improvements, which include different improvements to better mitigate severe accidents. The European project ASSAS (Artificial intelligence for Simulation of Severe Accidents) aims to address this gap by developing a proof-of-concept for a severe accident simulator. The prototype will feature a basic principles generic Pressurized Water Reactor, whose Graphical User Interface will be developed by Tecnatom-Westinghouse.

In addition to being a demonstrator for more complex industrial simulators, this basic-principles simulator is intended to support education and training activities on severe accident phenomenology. Reaching this objective required to address several competing constraints: representing the complexity of severe accidents without overwhelming the user with information, showing a large variety of phenomenology and operating situations with a limited number of systems and scenarios, running the simulation accurately in real time, etc. The proposed characteristics of the simulator to reach a compromise are presented in the first part of the article.

Besides, the description of the Source Term (ST) released by the accident appeared to need special attention, because of the complexity of its chemical and isotopic composition, which can drastically influence the radiological consequences of the accident. The second part of the article proposes a synthetic description of the source term, considered appropriate for training but also for a fast and still accurate assessment of accident consequences. This last feature opens the way for the use of the simulator to support emergency preparedness and response, possibly with Artificial Intelligence models.

This paper summarizes the work developed by Tecnatom-Westinghouse as leading WP6: Interfacing ASTEC with the simulation environment, of Project ASSAS funded by EU-EURATOM (g. a. no. 101059682) along the first-year project. It includes the development of the Simulator Specifications and Interface, the selected systems included in the scope and the scenarios to be considered as well as the definition of the Graphical User Interface and the main criteria used to fit to the targets of the project. Then, the paper focuses on the discussion on the proposals made to define the potential isotopic composition for the source term estimates, fitting to the main objectives of the project, as they are the education, the training, or the training for the surrogate models defined by the IA methodology, and how these targets impose restrictions to the isotopic characteristics of the ST used but keeping the major simulator features. Even though it is out of the current scope of the project, it has also been considered the potential use of the simulator for emergency preparedness and response training and consequence assessments to be properly founded for the future evolution of the simulator.