LSM-SEMINAR

Webinar: Precise simulations of multibunches in high intensity cyclotrons

by Matthias Frey (Paul Scherrer Institut)

Europe/Zurich
https://ethz.zoom.us/j/4705824086 Meeting ID: 470 582 4086

https://ethz.zoom.us/j/4705824086 Meeting ID: 470 582 4086

Description

Abstract:

The one of the main operational goals of every particle accelerator facility is the minimization of particle losses. One reason of these losses is the bunch internal Coulomb's repulsion, that gives rise to so-called halo particles. In circular high intensity accelerators such as cyclotrons an additional source of halo particles are space charge effects due to neighboring bunches. Although efforts have been undertaken in the past to describe these non-linear interactions analytically, this is not possible without strong assumptions. State-of-the-art relativistic beam dynamics computer codes such as the Object-Oriented Parallel Accelerator Library (OPAL) library are therefore indispensable for the research on neighboring bunch interactions and on space charge dominated accelerators in general.

Since realistic and precise large-scale neighboring bunch simulations in high intensity cyclotrons require macro particles in the order of magnitude of O(10^9 ... 10^10), the associated Particle-In-Cell (PIC) mesh consists of O(10^8 ... 10^9) grid points. Hence, the numerical calculation of particle interactions coheres with a considerable computing effort. In this thesis the electrostatic PIC model of OPAL is interfaced with the block-structured AMR library called AMReX. This new PIC model in OPAL is complemented with a hardware portable AMR Poisson solver that is based on second generation Trilinos packages.

A global sensitivity study on numerical input parameters of AMR and of energy binning techniques applied in neighboring bunch helped to identify the potential of a model cost reduction. The analysis is based on Sobol' indices evaluated by means of the polynomial chaos expansion obtained by uncertainty quantification.

The project further includes a multiobjective optimization that matches the turn pattern of the PSI Ring cyclotron with measurements. The optimizations make use of a new trim coil model based on profile measurements.

Organised by

Laboratory for Scientific Computing and Modelling

Dr. Andreas Adelmann