Hybrid Monte Carlo methods for Geant4-based nuclear well logging implementation

Abstract

Monte Carlo method is one of most common numerical methods used in nuclear well logging simulations, but it can be very time-consuming when faced with deep penetration problems.In practice, hybrid Monte Carlo or partially deterministic methods are widely used to improve the conventional Monte Carlo computation efficiency. This work aims to apply one of the state-of-the-art hybrid methods, CADIS and FW-CADIS, to enhance the performance of current nuclear well logging simulation process. This is achieved by implementing CADIS and FW-CADIS method into a framework based on the open-source Monte Carlo code Geant4. Two representative nuclear logging tools designed for field production are selected to verify the performance of the hybrid Monte Carlo Geant4 framework. We evaluate the enhancement by calculating the Figure of Merit (FOM) from the variance of total detector flux. The results show that the simulation efficiency can be improved by using CADIS and FW-CADIS method. The maximum improvement of FOM is 1444 times in the gamma density tool and 159 times in the neutron porosity tool.