Monte Carlo computations of a pulsed-power bremsstrahlung source

The research for this thesis computationally simulated a pulsed-power bremsstrahlung source with Monte Carlo techniques. The bremsstrahlung source consisted of a pulsed current of electrons impinging on a target with a high atomic number. A differential absorption spectrometer and a thermoluminescent detector (TLD) array measured the transmitted photon energy spectrum and the absorbed dose at points in the bremsstrahlung radiation field. Experimental measurements of the source were documented by T.K. Cotter in "Measurements of a Pulsed-Power Bremsstrahlung Source" (Cotter, 1993). Two computational methods for simulating the source are compared in this thesis. Both methods use the stochastic mathematical tool known as the Monte Carlo technique. First, an existing Monte Carlo electron/photon particle transport code, the Integrated TIGER Series (ITS), modeled the electron current, the bremsstrahlung production, and the resulting radiation field. This full transport method calculated the transmitted photon energy spectrum, the absorbed dose in the thermoluminescent detector (TLD) array, and the fluence in the TLD array. The experimental TLD dose values 2.794 cm from the source differed from the full transport by an average of 51%, and by an average of 73% 12.7 cm from the source. Discrepancies are a combination of statistical variation in the Monte Carlo calculations, and errors in the experimental measurements. The full transport absorbed dose calculations cost 22.4 hours of CPU time on a MicroVax 3500, and delivered results with maximum statistical uncertainty 37%. The second simulation method was the a hybrid Monte Carlo method. The hybrid method was designed to decreased the CPU time required to calculate the absorbed dose in the TLD array. The hybrid method uses the Integrated TIGER Series, and a Monte Carlo integration program written for this thesis. The method was verified by comparing hybrid and full transport calculations of fluence in the TLD array 2.794 cm from the source. The hybrid fluence agreed with full transport fluence within 14% when the TLD array was 2.794 cm from the source. The full transport fluence values contained an average statistical error of 23%. The hybrid method was extended to calculate KERMA in the TLD array 2.794 cm from the source. If charged particle equilibrium exists in a volume, the KERMA is equal to the absorbed dose. The comparison of hybrid KERMA and full transport absorbed dose reveals the aluminum equilibriator in the TLD is too thin to ensure charged particle equilibrium. In conclusion, the hybrid method fulfilled its purpose by calculating fluence values within the range of the full transport statistical error in about 1/5 of the CPU time spent on full transport fluence calculations. For a volume in charged particle equilibrium, the hybrid method should approximate the absorbed dose in a fraction of the time it would take for a full transport calculation. Future research should test this hypothesis by comparing hybrid KERMA and full transport dose for a volume in charged particle equilibrium.