Date of Award

1993

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Physics and Astronomy

First Advisor

Mark L. Williams

Abstract

A contribution response Monte Carlo method is developed and successfully applied to a sample deep penetration shielding problem. The random walk is simulated in most of its parts like in conventional M.C. by introducing the concept of a fictitious response particle. The scoring is peculiar to the proposed method and it need not be made at the detector itself so that computing time can be reduced by an order of magnitude or more depending on the geometry of the problem. The probability density functions are natural. They possess properties not encountered in conventional M.C. methods currently in use. The selection of all random variables from any pdf depends on all defining parameters of the system, namely, the geometry of the problem, relative position of source-detector, volume of detector, nature and magnitude of the detector response function and the material of the shield. The source and the scattering pdfs are continuous functions of the directional cosine and the azimuthal angle random variables. The selection of the parameters of the emergent particle from the scattering pdf is affected by the past history of the particle. The transport pdf is an unusual exponential kernel strongly dependent on the path followed by the particle between collisions. One thousand particles are sufficient to reproduce the same answer obtained using DOT discrete ordinates two dimensional code with a very small fractional standard deviation and with less than five minutes CPU time on a 3090 IBM main frame system. Analog and nonabsorption biasing Monte Carlo were considered.

Pages

118

Share

COinS