One of the most important topics of nuclear research for the moment is investigation of possible options for treating spent fuel from nuclear reactors. Two main options are currently under consideration: (1) final disposal of the fuel or (2) burning of a part of the spent fuel in conventional and/or dedicated systems such as Accelerator Driven Systems (ADSs). The second option could be employed to decrease amount and toxicity of nuclear wastes and to produce energy.

An ADS contains a sub-critical core driven by an external source of neutrons (produced by protons interacting with a target). It may allow use of special fuels, e.g. with high content of minor actinides as Am and Cm. That is why ADSs are investigated at many research institutions, in particular at FZK. ADS transient simulations /1/ are important for safety analyses and interpretation of experiments.

ADSs may demonstrate a quite different (compared to conventional “critical” reactors) response to possible perturbations: that is related to the sub-criticality and presence of the external neutron source. In particular, possible variations of the proton beam may initiate very fast (i.e. in the microsecond range) transients. For simulating these transients, which are important for experimental studies and ADS diagnostics, available codes should be extended and/or validated.

In a transient driven by a very fast variation of the external source, usually localized in the center of the core, the neutron flux may show a high level of anisotropy (in angle and space) in the short time scale (comparable with the life time of neutron).

This consideration suggests the use of a transport code that could provide a fast solution over large homogenous region with accurate description of flux shape inside these regions, such as VARIANT/KIN3D /2, 3/. VARIANT is a static nodal neutron transport code developed in the US. KIN3D is a kinetics/perturbation extension for VARIANT developed at CEA and FZK. VARIANT/KIN3D is included in the ERANOS /4/ code system that is widely used in Europe and in the US for fast reactor studies.

The original version of KIN3D was already employed for analyses of the MUSE /5/ experiment performed at Cadarache. The new version is under development at FZK. A more accurate modeling of anisotropic effects in neutron distributions appearing after an external neutron source pulse has been implemented in KIN3D recently /6, 7/.

A very short characteristic time scale of the within group transport equation and a strong angular anisotropy suggests a fully discretization in time of the flux angular harmonics using an implicit time integration scheme (to avoid instability of the solution).

Currently our effort is focused on mitigating the cost of the implicit scheme (with respect to time). Possible solutions are the use of the SPN approximation (reduced set of angular harmonics) that shows a good compromise between time consumption and accuracy of the angular description. In addition, the introduction of energy group dependent adaptive time step control scheme is under development.