A Comparative Analysis of Simulations and Experimental Outcomes: Slow Extraction Driven by RF Transverse Excitation at the CERN Proton Synchrotron

Radio-Frequency Knockout (RFKO) is a method of resonant slow beam extraction which is able to produce continuous particle spills over durations longer than can be achieved with fast single-turn or non-resonant multi-turn extraction. By using transverse excitation to gradually drive the circulating particles into resonance, spills can be produced over many thousands of turns, at a low intensity, and with a flat time profile. These spills can be delivered to facilities—such as fixed-target experiments or hadron therapy gantries—which require a continuous, low-intensity particle flux of durations significantly longer than the particle revolution period. In order to accurately and efficiently simulate this extraction process over a wide range of timescales, new modelling tools and computing platforms must be explored. By utilising optimised computational hardware—such as General Purpose Graphics Processing Units (GPGPUs), and next-generation simulation software (such as Xsuite), computation times for simulations can be reduced by several orders of magnitude while remaining accurate compared to empirical measurements. This thesis presents the novel implementation and simulation of RFKO slow extraction at the East Area (EA) experimental facility, located at CERN's Proton Synchrotron (PS). The results of the experiments outlined seek to confirm both the accuracy of the simulated frequency response and time structure, and the benefits of GPU-accelerated simulation software. These results will also provide an analysis of optimal parameter selections for both the extraction control system—optimising for spill quality—and the simulation system architecture—optimising for performance.