Speaker
Andrei Olegovich Kovalev
Description
See the full Abstract at http://ocs.ciemat.es/EPS2018ABS/pdf/P4.1009.pdf
Fast Simulation of Local Radiation Fields for Synthetic Diagnostics
A.R. Polevoi1, A.O. Kovalev2, R.N. Rodionov2, E.I. Polunovskiy1, L. Bertalot1, Yu.A. Kashchuk2,
D.V. Portnov2, A. Loarte1, M. Loughlin1, S.D. Pinches1, M. De Bock1, E. Veshchev1
1 ITER Organization Headquarters, Route de Vinon-sur-Verdon, CS 90 046, 13067 St. Paul-lez-Durance Cedex, France
2 Project Center ITER, Moscow, Russia
We describe a technique based on the concept of Green’s functions, which enables fast
simulations of the local radiation fields and diagnostic signals for arbitrarily distributed
volumetric sources. The technique is applicable to simulations of diagnostic signals for a
range of neutron and optical diagnostics in tokamak plasmas. The development and
validation of control algorithms for plasma scenarios in tokamaks require fast techniques for
the simulation of the signals from plasma diagnostics. The signals should be realistic, i.e.
correctly describe the source dynamics, geometry of detectors, and take into account
reflection, scattering and dissipation of radiation in the real design geometry. Typically,
accurate simulation of the signals requires the use of time consuming Monte-Carlo or
ray-tracing codes using detailed 3D models of the reflecting, scattering, and absorbing
elements affecting the diagnostic signals from the sources of radiation distributed in the
plasma volume. Thus, computation of a realistic signal from evolving distributed volumetric
sources of radiation becomes too time-consuming and expensive an exercise. For cases with
additive impact of the sources on the diagnostic signal it is possible to use an approach based
upon Green’s functions, considering the function which transfers the source of radiation to
the registered signal as a generalized operator. The matrix Gij of the signals measured by a
diagnostic from a set of unit ring sources on a chosen grid (Ri,Zj) is calculated by
comprehensive time consuming codes just once. This is then used for interpolation of the
Green’s function, g(R,Z), to simulate a signal from an arbitrarily distributed evolving
volumetric source, S(R,Z;t): F(t) = ∫S(R,Z;t)g(R,Z)dV. This analytic approximation enables
simulations of the signals from sources computed on the grid of different complexity adapted
to a specific source to keep the required accuracy and resolution. As an example, we discuss
the application of proposed technique to simulations of the signals from the neutron
diagnostics in ITER. Applications of this technique have enabled a reduction in the
computational time needed for a signal from dozens of hours of parallel computations by the
MCNP code to less than a second of CPU time. The Green’s function matrix Gij and base grid
(Ri,Zj) approach are recommended for the creation of synthetic diagnostics for ITER in the
frame of IMAS.
