Porosity Inference From Surface Data: An Approach Combining Analytical Models to Generate a Synthetic Porosity Log

This study presents a real-time, cost-effective method for estimating porosity during drilling operations using mainly surface mud logging parameters and gas chromatography data. Traditional porosity evaluation requires core samples or expensive wireline logging tools, often leading to delays and increased costs. The proposed approach integrates a deterministic statistical model ( Beda and Tiwary, 2011 ) with machine learning techniques for automated baseline trend (P0 line) definition and porosity calculation. Real-time parameters such as ROP, WOB, RPM, bit size, gamma ray (GR), and methane concentration (C1) are collected, quality-checked, and processed in a cloud-based environment. The method calculates the Perforability Index (PI) and derives porosity (PHI_ML), with further refinement using GR and C1 data to increase lithological sensitivity. Case study results from a Brazilian offshore well show that the enhanced method reduces porosity estimation error to as low as 2.3% compared to traditional wireline-derived porosity logs. The solution offers automation, scalability, and rapid integration into digital drilling workflows.