Coupling fast reactor design and scenario calculations: a new methodology applied to scenario optimization

There is currently a sequential relation between reactor design studies and scenario studies. Generally speaking, it goes as follows: after establishing a set of objectives for the nuclear fleet, reactor concepts are designed to meet constraints and performance criteria coming from reactor physics (safety and efficiency aspects) and scenarios (minimizing plutonium inventory, minimizing natural uranium consumption...). Once the design phase is over, numerical models of the reactors are generated. The purpose of these numerical models is to simulate the irradiation by adapting the flux and the cross sections for given fresh fuel compositions, without accounting for reactor performances. In scenario calculations, these models are considered as fixed entries whereas scenario parameters such as reprocessing strategy, deployment dates or batch composition are adjustable. This sequential logic may not be optimal from both reactor design and scenario calculations points of view. From reactor design perspective, it is known for example that plutonium isotopy has an impact on reactor performances such as reactivity loss per cycle or maximum linear power which can lead to limitations on core parameters like cycle length or reactivity control system dimensioning. Taking into account the data of plutonium evolution coming from scenario calculations would enable a better dimensioning of reactivity control systems and a better monitoring of safety estimators. From a scenario perspective, using set reactor models may prevent the access to optimized results regarding given criteria. By considering an adaptive reactor which design can be modified online as an optimization lever, it is expected to get access to new optimized scenarios, while making sure that reactor safety indicators are acceptable at every moment of the simulation. The main goal of this paper is to present a new methodological approach that consists in using adaptive reactor design in scenario calculations in the case of sodium-cooled fast reactors (SFR). The construction of the adaptive fast reactor design using artificial neural network is presented as well as an explanation of the optimization loop that makes the link between the adaptive fast reactor and the scenario calculation. Then, this methodology is applied to an example: the minimization of plutonium inventory in a mixed fleet PWR-SFR. (authors)