The neutronic assessment of a fusion power plant design is usually a challenging and time-consuming task involving experts from several disciplines in order to assemble the geometry, source, as well as carry out computationally heavy Monte Carlo transport simulations. In order to overcome this challenge, we present in this work a deterministic method, which combines all these aspects in a single framework that can directly calculate a key neutronics performance indicator, the neutron wall load (NWL), at minimal computational cost for arbitrary stellarator configurations. As our method is based on simple vector and matrix manipulations, a speed on the order of a few CPU-seconds is achieved, which makes it suitable for optimisation frameworks. We demonstrate with a simple optimisation algorithm that it is possible to use our method to generate a first wall with reduced 'heterogeneity' in the NWL distribution.
Machine learning techniques for sequential learning engineering design optimisation