Automated T1T2 NMR Log Interpretation: From a Single Well to Standardized Multiwell Analysis

A novel method for interpreting T1T2 NMR log data enables near real-time, automated analysis, applicable when NMR data is recorded with other petrophysical logs or when insights from reference wells are applied to new wells. Using non-negative matrix factorization (NMF), T1T2 log data is decomposed into 2D factor-maps and 1D factor-weights. The optimal number of factors is determined by analyzing misfit trends. The Water-Level- Separation algorithm breaks complex factor-maps into dominant and secondary mono-modal, Gaussian-like signatures, each representing a distinct fluid source. These are mapped to interpretable fluid types (e.g., clay- bound water, free water, hydrocarbons) based on their summit positions on the T1T2 Fluid Interpretation Mask, with support for fluid type customization. From the fluid volumes, key properties like hydrocarbon saturation, oil producibility index, and permeability are computed. For new wells in unexplored areas, the factor basis is derived from T1T2 data, and the Fluid Interpretation Mask is optimized to align NMR-derived total water and clay-bound water volumes with other logs. For subsequent wells, the established factor basis and factors’ fluid interpretation are reused, ensuring consistent, efficient, and accurate interpretation across multiple wells. This approach reduces computational effort while improving reliability. We present several case studies of the new NMR T1T2 interpretation method applied in unconventional reservoirs. NMR interpretation is fully integrated with interpretation of other petrophysical logs. Independent estimation of water saturation from NMR and Resistivity are consistent, while NMR provides needed sensitivity to moveable hydrocarbon in tight and unconventional reservoirs.