Optimization of the ramp-up time of a superconducting synchrocyclotron

Abstract

This work addresses the optimization of the ramp-up time of the S2C2, a synchrocyclotron developed by Ion Beam Applications (IBA) for proton therapy. The study relies on a numerical magnetothermal model of the S2C2, extended here so that it can be used in a simulation-based optimization procedure. A first contribution is the coupling of the finite-element model to the external electrical circuit of the magnet, so that the simulations can be driven by the set-point current commanded to the generator. A strong coupling approach is shown to provide better predictions of the coil current and is retained for the rest of the study. The ramp-up is then formulated as a constrained black-box optimization problem and solved using the covariance matrix adaptation evolution strategy. The ramp-up computed with the optimization procedure is tested experimentally. A significant reduction of the operational ramp-up time from 4 hours to 2 hours is achieved. Finally, the numerical predictions are compared with experimental measurements for several ramp-ups. The correspondence is very satisfactory for slow and fast ramp-ups, but discrepancies appear for intermediate ones. Parametric studies and a data-driven analysis suggest that these discrepancies are likely due to an additional phenomenon not represented in the present model, with the cryocoolers appearing as the most plausible source of variability.