Hybrid Singular Value Decomposition (SVD)-Based Interval Inversion Algorithm for Robust Petrophysical Parameters Prediction

The study explores a hybrid optimization algorithm designed to address ill-posed well-logging inverse problems, combining Singular Value Decomposition (SVD) and Damped Least Squares (DLSQ) techniques. The research leverages interval inversion, utilizing series expansion for identifying petrophysical properties across depth intervals, ensuring precision amidst nonlinearity challenges. Initially employing the SVD approach ensures stability by thoroughly evaluating eigenvalues before shifting to the faster DLSQ method, optimizing efficiency as convergence nears completion. Testing on synthetic and real field data from a gas-bearing reservoir in Egypt’s Western Desert validates the algorithm’s capacity to predict reservoir characteristics such as lithology, porosity, and saturation with improved accuracy. The findings highlight the heterogeneous nature of the reservoir, mainly comprised of sandstone layers interspersed with shale laminations. This approach not only provides computational efficiency but also enhances characterization and decision-making in reservoir development by accommodating uncertainties and capturing variations in reservoir quality.