The link between reactor design studies and scenarios calculations is usually sequential. From a list set of objectives, a reactor design is produced and passed to the scenarist in the form of a numeric irradiation model. This approach assumes that the reactor design is fixed from the scenarist perspective. The method presented in this article proposes to use a flexible reactor model, built with artificial neural networks, that gives the possibility to the scenarist to change a reactor design directly during the scenario calculations. Doing so, the reactor design is no longer an imposed parameter but a tool to find new optimal trajectories. Moreover, this flexible model is able to exploit the historical loaded fuel compositions generated by...