LHC Machine Protection System: Method for Balancing Machine Safety and Beam Availability

Abstract

The Large Hadron Collider (LHC) at CERN in Geneva, Switzerland, exceeds existing particle accelerators in terms of size and complexity. The most remarkable machine damage potential is held by the amount of stored energy. This thesis introduces a quantitative method for the reliability analysis of the LHC Machine Protection System (MPS) in terms of machine safety and beam availability. It is based on object-oriented modelling of the primary signal path, where the components’ behaviour is described by a simple Markov model with two failure states. The explicit inclusion of machine failure allows for the quantification of five scenarios. They include the safety-relevant scenario of a missed emergency shutdown and the scenario of a preventive shutdown, which is crucial with regard to beam availability. The presented MPS model covers two of the main MPS subsystems, namely the Beam Loss Monitor System and the Beam Interlock System. The model includes almost 5000 individually modelled components. It is implemented twice, first based on Monte Carlo simulation and second based on an analytical model description. The implementation of the analytical description is used for a series of case studies. The case studies yield scenario probabilities that confirm expectations based on operational experience and previous studies. The results are verified by the results of related Monte Carlo simulations, which show good accordance. Component importances are also extracted. A series of sensitivity analysis studies based on input parameter variation are presented. The studies do not reveal any unexpected dependence and as such confirm the stability of the model implementation. In addition, a set of model extensions are introduced which overcome some of the limiting assumptions and simplifications of the initial model.