Laboratory Experimental Validation of Sensitivity Functions for a Neutron Porosity Logging Tool in Casedhole Environments

Mature producing fields face the challenge of safely abandoning numerous wells. Casedhole logs can play a key role in the integrity and confinement assessment. Therefore, the industrial problem today is no longer to characterize some petrophysical properties of the rock formation but also some components of the well. Due to the large number of variables, a multiphysics inversion simultaneously considering different nuclear logs is mandatory. During the past 20 years, The University of Texas introduced the concept of flux sensitivity functions to predict logs and allow the inversion of formation properties. This approach has been successfully tested in open hole, but for cased hole, it relies on the ability of the fast-forward modeling algorithms to reliably reproduce the nuclear logs. In this work, we first compare two numerical methods for computing sensitivity functions with different Monte Carlo particle transport codes. The first is based on weight window with Monte Carlo N-Particle (MCNP), and the other is based on particle tracking with GEometry ANd Tracking 4 (GEANT4). Both methods provide equivalent results, but only particle tracking allows enough flexibility to be used in cased hole. We also report an experimental validation of a commercial porosity tool by comparing Monte Carlo simulations with disturbance measurements in cased hole. After checking the capability of MCNP and GEANT4 to reproduce count rates, we compare the fast-forward porosity predictions, calibrated with our new sensitivity functions based on particle interactions, with the measured porosity. Results show a good agreement between predictions and laboratory measurements in open hole, and between predictions and Monte Carlo simulations in cased hole, for a variety of well architectures.