Monte Carlo Acceleration Techniques Applied in Well Logging Radiation Detection

摘要

A reduced order model (ROM) has been developed to accelerate Monte Carlo simulation for natural gamma attenuation. The model is based on fitting a semi-empirical model to GEANT4 computed gamma spectra at various casing and cement thicknesses, and tool positions within the borehole. The ROM is exercised and is demonstrated that is compares well with direct GEANT4 computations at a reference set of parameters. Since executing ROM only takes seconds of time once it's built, such a model has great potential when quick turnarounds are needed for simulating changing environments and could be easily applied to future sensor design in the scope of radiation detection and measurement. Future work will involve further optimization of ROM. e.g. using machine learning to improve the performance of ROM through user-controlled iterations.