Modeling, Simulation and Optimization for CERN’s Quench Protection System
Nowadays, low-temperature superconducting magnets are used in high energy particle accelerators such as the Large Hadron Collider (LHC) at CERN to apply electromagnetic forces on the beam of particles traveling through the structure. This force is used to e.g. bend the path of the beam and keep it focused, and thus keep them on a circular track.
Network of magnets, where each magnet may require a representation by a finite element model in the case of quench propagation.
If temperature, magnetic field or current are above its critical surface then the superconducting material can quench, that is, lose its superconductivity and become resistive. If this happens, the Ohmic losses quickly heat up the magnet and thermo-mechanical stress is created. In particular, the energy density stored in the magnet is such that the peak temperature can lead to irreversible damage of the coils, compromising the integrity of the overall circuit. The reparation of such an accident is very time consuming and costly (see the busbar Quench incident at CERN in 2008) and thus the study of quench propagation and protection is of high importance.
Quench propagation is a highly multi-scale and multi-rate problem, which involves the coupling of several physical phenomenon solved with different equations, e.g. electromagnetic, thermal and mechanical field equations as well as electric circuits (see the image). The STEAM (Simulation of Transient Effects in Accelerator Magnets) project is a cooperation between research institutions, such as CERN and PSI (Paul Scherrer Institut), and the Technische Universität Darmstadt to develop a software platform for quench simulations that uses state of the art models and algorithms for simulation. Therefore, various models and formulations have been mathematically analyzed and implemented in software. New algorithms have been designed such that convergence can be assured. Models and algorithms are rigorously tested and complex simulation scenarios are validated by measurements.
A design decision was to use existing (possibly proprietary) simulation tools that focus on one or several aspects but are not able to capture all the required physical behavior. In either case, a classical monolithic approach, which solves all the coupled subsystems together would have lead to prohibitive computation times by not exploiting the multi-rate behavior. The STEAM platform combines various simulation tools by means of a hierarchical co-simulation and the waveform relaxation technique. This allows to solve the subsystems separately with different time-steps and iteratively exchange information between them so as to converge to the coupled solution. For more details see the overview paper (doi:10.1109/TASC.2017.2787665) or its preprint (arXiv:1801.08957).
We proudly conclude that modeling, simulation and optimization is used to guarantee safety at CERN.