Calculation of Monte Carlo importance functions for use in nuclear-well logging calculations

Examines four variance reduction approaches: geometry splitting with Russian roulette; weight windows; quasi-deterministic generated weight windows; and integral transport functions applied in MCNP, a general Monte Carlo radiation transport computer code. The latter three are relatively new techniques that offer methodologies for improving the efficiency of the Monte Carlo calculation. Each of the techniques is applied to a neutron porosity tool. The near and far detector response are calculated, and the results are compared. Weight window generation is shown to work very well if iterations are performed on optimization of detector response. Quasi-deterministic generation requires less user interface, reduces statistical noise, and needs no iterative steps; however, with the current algorithms, there are large memory requirements. Integral-transport generation requires setting up and executing the XTRAPT code but provides very good weights, and the method yields global transport solutions useful to tool response understanding.<
>