Skip to Main Content
The LHC hardware commissioning has shown a considerable difference of performance between the three dipole manufacturers (Firm1, Firm2 and Firm3). More than 90% of the quenches occurred in the dipoles made by Firm3, less than 10% in Firm2, and no one in Firm1. In this paper we propose a Monte-Carlo method based on the quench performance data of individual magnets that accounts for this behavior. The model relies on the data of the virgin training and on correlations with quench after a thermal cycle as measured on a sample of magnets. This model is used to derive estimates of the training behavior in the machine. A comparison with data gathered during the hardware commissioning shows that the model works well for low fields, and that, starting from fields corresponding to an energy of about 6.3 TeV, a slower training in the magnets of one manufacturer is observed.