Computational Approach for Estimating Petrophysical Exponents and Residual Oil Saturation From Maximum Likelihood Estimation Method Over “FAS” Field, Offshore Niger Delta

ABSTRACT

This study developed a Python‐based computer application, RESOILSAT, for estimating petrophysical exponents and residual oil saturation ( S _or ) in “FAS” Field, Niger Delta. The motivation arose from the need to address the observed lapses in the Archie method for estimating petrophysical exponents and fluid saturation, especially in shaly sand reservoirs. It provides an alternative approach that complements existing fluid saturation models for shaly sands, such as the Simandoux, dual porosity, Waxman–Smits, and Indonesian techniques. The study utilized eight wells (“FAS”‐01, “FAS”‐02, “FAS”‐03, “FAS”‐04, “FAS”‐05, “FAS”‐06, “FAS”‐07, and FAS″‐08), which consist of well logs (gamma ray, resistivity, neutron, density, and sonic) and core S _or data from wells (“FAS”‐01, “FAS”‐02, and “FAS”‐03) as input data for the developed application. The application was developed around the conventional (Archie) method and maximum likelihood estimation (MLE) method, which involved the gradient descent (G‐D) optimization algorithm to estimate petrophysical exponents and S _or in the delineated reservoirs (Res‐A and Res‐B). Results show that, with the conventional Archie method, tortuosity ( a ) and cementation ( m ) are constant, 0.21 and 2.15, respectively, in “FAS”‐01 to “FAS”‐08, whereas the saturation exponent ( n ) varied from 0.40 (Res‐B, “FAS”‐05) to 8.58 (Res‐B, “FAS”‐07), and S _or varied from 0.18 (Res‐B, “FAS”‐05) to 0.97 (Res‐A, “FAS”‐06). With the Indonesian and Simandoux models, S _or varied from 0.88 (“FAS”‐05) to 0.94 (“FAS”‐06) and 0.91 (“FAS”‐03) to 0.98 (“FAS”‐04, “FAS”‐05, and “FAS”‐08) respectively. With the MLE using the G‐D algorithm, a varied from 0.37 (Res‐A, “FAS”‐04) to 2.80 (Res‐A, “FAS”‐03), m varied from 0.05 (Res‐B, “FAS”‐01) to 4.11 (Res‐A, “FAS”‐04), whereas n varied from 0.66 (Res‐B, “FAS”‐05) to 8.59 (Res‐A, “FAS”‐05), and S _or varied from 0.11 (Res‐B, “FAS”‐05) to 0.95 (Res‐A, “FAS”‐03). The percentage deviation of the computed S _or relative to the core data varied from 1% to 7% in MLE and 3% to 64% for conventional. t ‐Test of the computed S _or relative to the core data ranged from 0.44 to 10.10 for MLE and 3.85 to 15.27 for conventional method. The relatively low percentage deviation of the MLE method relative to the core‐derived S _or lends credence to its higher reliability in computing S _or in the study area.