Availability of Particle Accelerators: requirements, prediction methods and optimization

Abstract

From the design phase to operation, machine availability represents one of the key indicators for the performance of a particle accelerator. Availability requirements are typically set at the beginning of a project and should be kept (or demonstrated) during the operation phase. In the early design stages of an accelerator, an effective allocation method is needed to translate the overall accelerator availability goal into availability requirements for each subsystem. This is of particular value for cases in which the detailed design is not known, or where new technologies are developed and no failure data is available. In this thesis a novel method is proposed to allocate availability requirements based on accelerator subsystems complexity. During the design of complex availability-critical particle accelerators, the implementation of a detailed availability model that uses component reliability data for estimating the overall system availability, is particularly useful to demonstrate their feasibility and to identify improvements with high performance benefit. To ensure the completeness and consistency of the studies, a step-wise methodology for the definition of availability models is presented. In operating particle accelerators, availability models are also used to optimize machine performance. In both cases, the reliability of the results strongly depends on the precise knowledge of the input data. Hence, availability-tracking tools are of crucial importance to ensure reliable data capture. This thesis presents the performance evaluation of Linac4 during a Reliability Run using the Accelerator Fault Tracking system developed at CERN. The ultimate goal of accelerator availability studies is to determine the system designs and operation modes that would lead to the best performance of the accelerator at lowest cost. To this end, a sensitivity analysis method is proposed to identify the component upgrades that would lead to the best improvement of system availability for a certain investment. Moreover, the presented sensitivity analysis also helps to identify potential common cause failures (which are not considered in the availability models), and other critical components that may compromise significantly the optimal performance of the accelerator. The proposed methodologies are illustrated with examples of accelerators in the design phase and under operation both for linear accelerators: CLIC and Linac4, and circular accelerators: FCC and LHC.

Date of Award	2018
Original language	English
Awarding Institution	University of Stuttgart, Germany