Combining Nuclear Magnetic Resonance (NMR) Logging to Predict Free Fluid Porosity through Seismic Attributes Utilizing Probabilistic Neural Network: An Empirical Investigation within an Iraqi Carbonate Reservoir

Authors

  • Jawad K. Radhy Department of Geology, College of Science, University of Basrah, Basrah, Iraq
  • Muwafaq F. Al-Shahwan Department of Geology, College of Science, University of Basrah, Basrah, Iraq

DOI:

https://doi.org/10.46717/igj.2025.58.2E.2

Keywords:

NMR log; Free fluid porosity; Seismic attributes; Probabilistic neural network; Seismic Inversion

Abstract

Effective porosity, often referred to as Free Fluid Porosity, can be calculated through the particular processing and interpretation of NMR datasets. Given the considerable costs associated with NMR logging, only a select number of wells are prepared with an NMR log. This limitation can be alleviated through modern techniques, such as the generation of attributes derived from seismic data, which can be physically linked to the reservoir properties. The inversion process employs the Geoview module, which converts three-dimensional seismic data related to the Mishrif Formation into an acoustic impedance model. Initially, the well logs were tied to seismic data by generating a synthetic seismogram at an individual well, incorporating the sonic and density logs. Subsequently, it combines various seismic attributes to evaluate specific reservoir parameters of interest, facilitating the prediction of rock property volumes through the integration of well logs and seismic data attributes. The multi-attribute transformation that utilizes a probabilistic neural network constitutes a complex mathematical interpolation framework that influences neural network architecture for its execution. The predictive outcomes for Free Fluid Porosity derived from the probabilistic neural network exhibit a strong correlation with the actual log data, as evidenced by a correlation coefficient of 0.96. The qualitative and quantitative findings suggest that the application of the utilized probabilistic neural network in combination with model-based inversion represents the most effective approach for prediction of free fluid porosity throughout the entire reservoir.

Downloads

Published

2025-11-30