Uncertainties are essential in our understanding of the basic nuclear physics used in modelling and simulation of nuclear systems. Some quantities are exceptionally well-studied, with uncertainties less than 1%, while others may be only theoretically calculated with uncertainties of 1000% or more. Quantifying these uncertainties and developing methods of propagating them into simulations has been a subject of active research over the past decade.
Due to our knowlege of nuclear physics and the maturing of semi-empirical nuclear reaction modelling, we can significantly reduce uncertainties by introducing correlations between different reaction channels and incident particle energies, allowing us to effectively constrain the space of possible physics. These correlated uncertainties, or covariances, are now expected in all major nuclear data libraries.
Correlated uncertainties have been introduced in the early 2000s and the NEA has been at the forefront of this work, launching subgroups within the Working Party on International Nuclear Data Evaluation Co-operation (WPEC) on the evaluation of covariances, development of processing algorithms to handle these data and make them accessible to application codes and studies of the impacts of these correlations on real uncertainty propagation cases.
The NEA's nuclear data evaluation co-operation activities involve the following evaluation projects: ENDF (United States), JENDL (Japan), ROSFOND/BROND (Russia), JEFF (other Data Bank member countries) and CENDL (China) in close co-operation with the Nuclear Data Section of the International Atomic Energy Agency (IAEA).