guest ::
| Home > Publications database > A Practical and in Parts Adjoint Based Gradient Computation Methodology for Efficient Optimal Magnetic Divertor Design in Nuclear Fusion Reactors |
| Home > Publications database > A Practical and in Parts Adjoint Based Gradient Computation Methodology for Efficient Optimal Magnetic Divertor Design in Nuclear Fusion Reactors |
| Abstract | FZJ-2015-04531 |
; ; ; ;
2015
Abstract: Divertor particle and power exhaust system design is still a key issue to be resolved to evolve from experimental fusion reactors to commercial power plants. In particular the excessive heat load to the divertor geometry structure needs to be tackled. The divertor design process is assisted by computationally extremely demanding codes simulating the complex physics of the plasma edge. In order to reduce design costs, advanced adjoint based automated design methods from aerodynamics have recently been adapted for use in divertor shape design. A similar methodology is sought to enable these adjoint based methods for magnetic configuration design. However, adjoint sensitivities of the plasma edge grid generator are difficult to obtain, as those grids are aligned with the magnetic field to overcome numerical noise arising from the strongly anisotropic plasma flow. First of all, this results in computational grid boundaries that change with varying magnetic fields. A second difficulty is the necessity to use a curvilinear coordinate system attached to the magnetic field that is implied by the grid generator.A Practical and in Parts Adjoint Based Gradient Computation Methodology for Efficient Optimal Magnetic Divertor Design in Nuclear Fusion ReactorsIn this paper, these difficulties are overcome by using a combined finite differences/continuous adjoint gradient computation. An adjoint plasma edge simulation is used to overcome the high computational cost of the plasma edge simulations during gradient calculations. At the same time, the computationally less demanding yet more difficult grid generator adjoint sensitivities are avoided by making use of a finite difference approach to differentiate the remaining terms within a Lagrangian multiplier approach. Moreover, an extensive analytical derivation of the partial derivatives of the plasma edge equations with respect to the geometrical parameters is avoided by using the finite difference approach through the forward plasma edge solver as well. Specific attention is hereby needed for the implementation of the boundary conditions.
|
The record appears in these collections: |