Linear Accelerator Modeling: Development and Application

Abstract

Most of the parameters of a modern linear accelerator can be selected by simulating the desired machine characteristics in a computer code and observing how the parameters affect the beam dynamics. The code PARMILA1 is used at LAMPF for the low-energy portion of linacs. Collections of particles can be traced with a free choice of input distributions in six-dimensional phase space. Random errors are often included in order to study the tolerances which should be imposed during manufacture or in operation. Once a machine is built, the simulation can be put to another use - that of modeling the actual, observed behavior. If a model of good fidelity can be found, it can be used in many ways: for example, to understand what is happening at intermediate points unobservable in practice. A most important use is to derive optimum settings for the accessible variables. Often the effect of imperfections can be largely alleviated or circumvented once it is realized that they are present and the actual measurements are used in the model. This technique has been successfully applied to the question of longitudinal tuning of the side-coupled linac portion of LAMPF, and is here applied to the same problem in the low-energy portion of the accelerator. We have incorporated as many of the actual measured physical parameters as possible into the code, and have found it necessary to include a more comprehensive treatment of the beam dynamics in order to obtain agreement between the model and experimental observations.