Accelerator Diagnosis and Control Based on Low Energy Empirical Model

点击翻译按钮获取中文摘要

This proposal aims to develop and validate a new approach to creating online accelerator models, and to further establish it as the basis for accelerator diagnostic and control applications in realistic operation and commissioning environments, especially at the Advanced Rare Isotope Laboratory (ARIEL) project at TRIUMF. The ability to correctly and efficiently model accelerator transport and beam dynamics is critical to the success of any real-time online diagnosis and control task. This is even more so at low energy where key performance parameters are often “frozen in” and too late to change at higher energy. Unfortunately, efficiency and correctness are not always compatible objectives for low energy (low-beta) modeling. In many cases faster analytic algorithms are used online to model effects at low energy, producing incorrect physics. We propose to bridge this gap with an online empirical model to be described later. Failure of analytical models for low-beta beam transport arises from the following: • Velocity changes within cavities • Non-relativistic dynamics, naïve field-rigidity scaling, and arrival-time dependence • RF waveform detail defying sinusoidal approximation but important at low energy • Significant longitudinal-transverse momentum transfer • Higher order aberrations in solenoids and in cavities • Strong impact of chromaticity due to large momentum spread • Elements inside fringe fields of other elements • Increased need to account for space charge at low energy, possibly with subdivided elements • Nontrivial interplay between all phase space dimensions preventing reduction/decoupling of model • Algorithm deficiency to deal with time-varying elements at off-design phase • Inability of S-coordinate based formulation to capture time-of-flight at element boundaries • Any unconventional element defying accurate analytical treatment (now limited to low-beta) While in the highly relativistic regime most analytical accelerator and beam modeling tools can be free of these problems, toward the non-relativistic end, which can encompass range of beta from 0 up to 0.999 or beyond, this dilemma between speed and accuracy is very real, often with detrimental effects. The solution put forth in the current proposal, henceforth referred to as the low energy (low beta) empirical model, attempts to mitigate this dilemma by adopting an efficient empirical approach, independent of the nature and detail of the physical process being modeled as long as a dedicated modeling tool exists that correctly models the process of interest as functions of a limited number of control and beam parameters. The modeling tool can account for all pertinent physical details without need of closed-form formulation or approximation to physical processes. The outcome is rendered into efficient numerical models that can quickly and accurately predict detailed machine and beam behavior. This process is conceptually simple and scalable, and can be applied to any modeling tool for any component. Possible directions in which to extend this approach are many. Basically, complexity in underlying physics does not present additional levels of conceptual difficulty for this approach. Experience gained and infrastructure developed within the current proposed scope of low energy empirical model can serve as the basis for more ambitious projects such as empirical model based on real data from controlled beam based measurements or statistically accumulated machine data. Such development can lead to the capability of deterministically and efficiently controlling the traditionally most volatile sections of an accelerator.