Knowledge of basic nuclear physics is essential for the modelling and safe operation of nuclear facilities, including all nuclear power reactors, waste and reprocessing facilities, nuclear medicine, accelerator-based systems and more. National and international communities have run programmes to evaluate and store our best current knowledge of nuclear reaction physics in so-called "nuclear data" libraries.
The de facto international standard format, the Evaluated Nuclear Data File 6 (ENDF-6) format, was designed originally for 1960s era punch-card readers. The replacement of this format and affiliated system of codes has been recognised as an important initiative, in order to better capture required physical data, allow robust quality assurance practices, interface with modern computing systems and transfer knowledge and expertise to the next generation.
In order to translate basic nuclear physics information into application-ready data, a series of numerical processing algorithms are required. These processes translate between different formats, including some with limited or non-existent documentation and requiring complex software to generate application-ready data, including all physics and uncertainty information.
The ability to use increasingly high-fidelity nuclear physics, coupled with state-of-the-art uncertainties, is crucial for the advanced simulations that have driven the investment in new nuclear technologies. New uncertainty quantification methodologies allow operators to better understand their systems and margins, enhancing safety and providing more predictive modelling capabilities that increase efficiencies and lower costs. This requires more detailed and accurate data, which in turn requires improvements to the standards for data storage that are out-of-date with modern physics and computing. Accurate uncertainty quantification requires a relatively tremendous amount of correlated input uncertainty data, as shown by the trends over the past 30 years, which must take numerous complex forms that depend on the physics being considered.
The demographics of nuclear expertise are shifting and knowledge transfer is universally recognised as a priority for the field, yet technologies based on punch-cards are still in service today. Training the next generation with these approaches is not only difficult, but misses the opportunity to innovate and improve practices and motivate aspiring experts. The process of replacing these systems is challenging and requires a co-ordinated international strategy and consensus to launch a replacement with a full plan for implementation.
In 2013, the NEA Working Party on International Nuclear Data Evaluation Co-operation (WPEC) launched a project to review the requirements for an international replacement for the ENDF-6 format. The recommendations prompted the creation of a new Expert Group on a Generalised Nuclear Data Structure (GNDS) in 2016 that has used these requirements as the framework for a new format specification. Following rigorous international review, version 1.9 was unanimously approved by the Expert Group for publication in 2019.
Alongside the publication, a simple but effective process has been established for the proposal of additions to this extensible new format, with a network of experts to review, endorse and adopt proposals, with a steady flow of new ideas being proposed and reviewed as the Expert Group annually prepares and approves new versions of the standard.
In parallel to the GNDS specifications, users will require interfaces for the new data in order to integrate them into the ongoing workflows at utilities, technical support organisations, regulators, the research community and others. A WPEC subgroup has been established to design and implement new application programming interfaces (APIs) to jump-start this process and prepare users to be able to immediately start adopting the new standard. This has already resulted in three different APIs being released as open-source projects and more are in development. Advanced nuclear simulation codes such as Geant4 (European Organisation of Nuclear Research), SCALE (Oak Ridge National Laboratory, United States) and Mercury (Lawrence Livermore National Laboratory, United States) have already developed full or partial capabilities to interpret GNDS data, with plans to transition to the new standard as validation cases reach maturity.
GNDS as a format can be implemented with different technologies (e.g. Hierarchical Data Format [HDF], eXtensible Markup Language [XML] or others) and immediately interpreted with standard libraries in any modern computer programming language. As shown in Figure 2, it has been engineered as a replacement to, and extension of, the ENDF-6 format, and maintains a strict one-to-one translation capability with legacy files, while other standard processed outputs may be generated by the open-source FUDGE code system.
