Data-driven interpretable machine learning for prediction of porosity and permeability of tight sandstone reservoir - research paper

Q: Why are porosity and permeability central to reservoir evaluation?

They determine identification of high-quality reservoirs and favorable “sweet spot” zones and support predicting hydrocarbon development potential.

Q: What problem does the study address regarding data collection?

It targets the difficulty of collecting enough core samples on both vertical and horizontal planes due to time and cost constraints.

Q: What is the main contribution of the proposed method?

It introduces an interpretable Permutation Importance-Set algorithm to select logging curve inputs and reduce redundant data while improving interpretability and prediction accuracy.

Aug 4, 2026

Upload by Asher

Porosity and permeability are key indicators for identifying high-quality reservoirs and favorable “sweet spot” zones, yet obtaining enough core samples for reliable analysis across vertical and horizontal directions is costly and time consuming. Logging data can capture continuous, non-linear relationships that machine learning can exploit, but model inputs must be selected efficiently while preserving interpretability. This study proposes an interpretable Permutation Importance-Set feature-selection approach and develops basin-scale prediction models using multiple algorithms for tight sandstone reservoirs in the Sichuan Basin.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages15

Reading Time38 min

AccessFree to download

Introduction

Petrophysical importance and reservoir evaluation

Challenges of core sampling

Logging data as an alternative

Advances in OrigiGeo-Enal articlenergy Research Vol. 16, No. 1, p. 21-35, 2025 Data-driven interpretable machine learning for prediction of porosity and permeability of tight sandstone reservoir Liu Cao 1 ,2 , Fujie Jiang 1 ,2*, Zhangxing Chen 1 ,3 ,4 , Yang Gao 1 ,2 ,5 , Lina Huo 1 ,2 , Di Chen 1 ,2 1 National Key Laboratory of Petroleum Resources and Engineering, China University of Petroleum (Beijing), Beijing 102249, P. R. China 2 College of Geosciences, China University of Petroleum (Beijing), Beijing 102249, P. R. China 3 Institute of Digital Twins, Eastern Institute of Technology, Ningbo 315200, P. R. China 4 Chemical and Petroleum Engineering, Schulich School of Engineering, University of Calgary, Calgary T2N 1N4, Canada 5 Research Institute of Petroleum Exploration and Development, PetroChina, Beijing 100089, P. R. China | Keywords:
Data-driven modeling interpretable machine learning permutation importance-set reservoir characterization
Cited as:
Cao, L., Jiang, F., Chen, Z., Gao, Y., Huo, L., Chen, D. Data-driven interpretable machine learning for prediction of porosity and permeability of tight sandstone reservoir. Advances in Geo-Energy Research, 2025, 16(1): 21-35 .
[https://doi.org/10.46690/ager.2025.04.04](https://doi.org/10.46690/ager.2025.04.04) | Abstract:
Porosity and permeability are crucial indicators in the identification of high-quality reservoirs and favorable “sweet spot” zones, as well as key parameters when predicting and evaluating the development potential of fossil fuels like oil and gas. However, it is impracticable to collect enough core samples on vertical and horizontal planes for analysis due to the associated time and cost demand. Machine learning algorithms have shown remarkable capabilities in predicting the petrophysical properties by capturing non-linear relationships among logging data. In this study, to quantify the selection of logging curves and reduce the redundant logging data input, a novel and interpretable Permutation Importance-Set algorithm is proposed on the basis of logging data from the Upper Triassic Xujiahe Formation in the Sichuan Basin. The results indicate that, because of compaction, burial depth is the primary feature affecting the physical properties of tight sandstone reservoirs. Acoustic and spontaneous potential logs are critical for porosity, while density and spontaneous potential logs are pivotal for permeability, reflecting the complex diagenesis caused by the widespread sand-mud interbedding. Basin-level prediction models for porosity and permeability were developed using ten machine learning algorithms, then ablation studies confirmed the effectiveness of our feature selection and the reduced model complexity and over-fitting. This study offers a concise, interpretable prediction model with superior accuracy and interpretability for tight sandstone reservoirs. | | --- | --- | 1. Introduction To avoid costly drilling mistakes due to the geological uncertainties and varying resource potential, the primary risk assessment indicator for oil exploration must be taken as the hydrocarbon enrichment potential of prospective reservoirs. The petrophysical properties (porosity and permeability) of the reservoir are among the crucial evaluation indicators of hydrocarbon enrichment potential (Wang et al., 2020) . Within the context of continental sedimentation in China, tight sandstone reservoirs refer to a porosity < 10% and permeability < 1 mD of oil and gas reservoirs (Zou et al., 2012) . Compared to conventional oil and gas reservoirs, tight oil and gas reservoirs have less favorable physical properties, stronger heterogeneity, lower reserve density ratio, and they present challenges in predicting favorable “sweet spot” zones and effective reservoirs (Zhao and Chen, 2014 ; Sun et al., 2019 ; Ampomah et al., 2017) . Therefore, it is essential to derive a reliable and convenient technology that can predict reservoir porosity and pe

FAQs

Why are porosity and permeability central to reservoir evaluation?

They determine identification of high-quality reservoirs and favorable “sweet spot” zones and support predicting hydrocarbon development potential.

What problem does the study address regarding data collection?

It targets the difficulty of collecting enough core samples on both vertical and horizontal planes due to time and cost constraints.

What is the main contribution of the proposed method?

It introduces an interpretable Permutation Importance-Set algorithm to select logging curve inputs and reduce redundant data while improving interpretability and prediction accuracy.

Aug 4, 2026

Upload by Asher