Numerical simulation and interpretation of neutron-induced gamma-ray spectroscopy measurements
Ajayi, Oyinkansola Modupe
MetadataShow full item record
Neutron-induced spectroscopy measurements are commonly used to quantify in-situ elemental and mineral compositions of rocks from the processing of measured gamma-ray energy spectra. However, geometrical effects on measured spectroscopy logs, such as thin beds, dipping beds, and deviated well trajectories, can cause shoulder-bed averaging that compromises the assessment of true layer elemental and mineral compositions. Traditional methods of interpreting neutron-induced gamma-ray spectroscopy measurements typically neglect such shoulder-bed averaging effects in the estimation of elemental and mineral compositions. Monte Carlo methods accurately reproduce borehole and formation geometrical effects on spectroscopy measurements but are extremely time consuming and impractical for use in routine interpretation. Reliable measurement interpretation must therefore begin with the development of a fast and accurate forward simulation method that explicitly incorporates measurement physics, borehole, tool, and formation geometry. This dissertation introduces a new algorithm to rapidly simulate elemental and mineral compositions from neutron induced spectroscopy measurements. The algorithm utilizes neutron-gamma ray spatial sensitivity functions to account for environmental and three-dimensional (3D) effects of formation porosity, fluids, dipping beds, thin beds, and arbitrary well trajectories. Simulations assume a logging-while-drilling (LWD) spectroscopy tool furbished with a 14-MeV pulsed-neutron source in the interpretation of gamma ray spectra obtained from high energy inelastic neutron scattering and thermal neutron capture. Results obtained with the rapid simulation method are benchmarked against rigorous Monte Carlo spectroscopy calculations for synthetic conventional and unconventional thinly-bedded reservoirs penetrated by vertical and high angle/horizontal (HA/HZ) wells. The fast simulation method yields calculations in approximately 1e6 the time required by Monte Carlo simulations, with an average difference below 5% between Monte Carlo and fast simulated logs. An inversion-based interpretation method is next introduced to accurately evaluate mineral concentrations from measured spectroscopy elemental logs based on the analytical relationship between elements and minerals through their chemical formulas. In the presence of geometrical effects, spectroscopy elemental and mineral logs are corrected for shoulder-bed averaging by the inclusion of spatial sensitivity maps, which account for such geometrical effects, in the inversion-based interpretation. Calculations are performed with both inelastic and capture gamma-ray spectroscopy measurements which arise from high-energy inelastic neutron scattering and low-energy thermal neutron capture, respectively. This strategy provides two sets of data that can ascertain chemical elements or minerals detectable in only one measurement mode and also independently validates estimated elemental and mineral compositions. In laminated formations, where layer thicknesses are below the vertical resolution of the tool, it is impossible to quantify layer properties with inversion methods. An additional interpretation method based on a new spectroscopy mixing law is therefore developed to estimate elemental and mineral compositions within individual laminae. The new inversion-based interpretation methods are successfully implemented in diverse synthetic and field cases with varying lithology types and well trajectories including vertical and HA/HZ wells. Results show that the developed methods reduce shoulder-bed averaging effects on measured spectroscopy logs by as much as 0.4 yield fraction, 0.17 weight fraction, and 0.34 mineral volume fraction. Finally, a new spectroscopy-based petrophysical interpretation method is introduced that utilizes estimated mineralogy to overcome the common assumption of homogeneous lithology in measured porosity logs, thereby improving the estimation of porosity and water saturation. Inclusion of shoulder-bed averaging effects on spectroscopy mineral logs also increases the accuracy of spectroscopy-based petrophysical interpretation.