Multi-Objective Optimization of Heat Load and Run Time
for CEBAF Linacs Using Genetic Algorithms
Student: Cody J. Reeves
School: University of Florida
Mentored By: Balša Terzić and Alicia Hofler
The Continuous Electron Beam Accelerator Facility (CEBAF) consists of two linear accelerators (Linacs). Each Linac consists of 200 niobium cavities that use superconducting radio frequency (SRF) to accelerate the electrons. The gradients for the cavities are selected to optimize two competing objectives: heat load (the amount of energy required to cool the cavities) and trip rate (how often the beam turns off within an hour). The resulting system is a multidimensional, multi-objective, nonlinear system of equations that is not readily solved by analytical methods. The study employed a genetic algorithm (GA), which applies the concept of natural selection, to solve this system of equations. This paper enumerates several methods to significantly reduce computation time without degrading solution quality. It also demonstrates ability to employ GA for operational use for any Linac-based facility.