Computational techniques for simulation and design of a biological sample irradiation chamber

Abstract

This study covers the computational efforts and theory performed to simulate a biological irradiation facility at Kansas State University’s TRIGA Mark II Nuclear Reactor Facility, as well as model and optimize a path taken by a platform to uniformly irradiate a target sample. The work presented in here covers the steps taken to simulate the reactor as a radiation source, model a flask of water as a target, approximate the dose from a path taken through the beam, and optimize the path to reach uniform irradiation. Previous reactor models were improved with updates to MCNP thermal cross-section libraries and converted to a general radiation source. This source was used to simulate the energy deposited in a voxelized flask of water, following a path in front of a reactor beamport in discrete steps, based on a series of parameters which set the velocity over different sections. The simulation output was used to construct a matrix with rows for each step function velocity and columns for the impact on each voxel, which has a parameter vector that minimizes the deviation in dose. Finally, simulation of a random position machine was performed to investigate how this process may change with the added constraint of flask gravity. The algorithms applied were able to converge to a series of linear motions that brought 99.7% of voxel doses within 10% of the median dose for both random and nonrandom initial parameters. There is a relationship between the parameters defining the path and the expected gravity vector, which can be randomly sampled to obtain paths with the expected gravity. Incorporating dose uniformity with gravity constraints would require both linear and rotational motion. Doing so with current memory use would be outside of the application of the algorithms in this work.