Modeling and Prediction of Thermal Conductivity in Porous Media: Numerical Simulation, Data-Driven Approaches and Cross-Property Relationships

Thermal properties are among the most important parameters in assessing the behavior of porous media, with a wide range of applications from the geothermal industry to heavy oil exploitation. Among the different thermal properties, thermal conductivity is one of the key parameters that has been widely investigated in the literature under various conditions. A precise evaluation of this variable can ensure the success of any industrial or lab-scale experiment and operation that involves heat transfer. Different methods and correlations have traditionally been developed to model thermal conductivity in porous media based on various rock and fluid characteristics. However, due to the diverse dependencies of this quantity on the structure of the porous medium, no generalized model exists that can accurately describe and predict thermal conductivity across different rock types, compositions, and operating conditions. On the other hand, based on the similarities between elastic and thermal properties, different researchers have attempted to predict one property based on the other. This approach supports the development of thermal conductivity maps using available elastic properties, which can be derived from lab tests, well logs, or rapid computational workflows, especially when thermal conductivity values are already available for some samples. Micro-Computed Tomography (Micro-CT) imaging has become one of the innovative methods for characterizing porous media in recent decades. Progress in computational power and algorithm development has enabled various types of analysis through this imaging technique. In this study, the thermal properties of porous media, specifically thermal conductivity, are investigated in different sections. In the first part, homogenization of thermal conductivity in Micro-CT samples is explored using a hierarchical homogenization technique with a new multi-stage format, along with an analysis of how padding and sample heterogeneity affect homogenization error. In the second part, different data-driven models are developed based on Micro-CT images and the graphical structure of the samples to predict thermal conductivity, incorporating both physically constrained and topology-aware modeling approaches. In the final part, different thermal-elastic cross-property correlations introduced in the literature are assessed under both saturated and dry conditions. The deviation of real Micro-CT porous media data from these models, particularly in saturated cases, is analyzed and evaluated. The results obtained from this research are applicable both in converting elastic data obtained from laboratory measurements or well logs into thermal conductivity, and in supporting the computation of elastic and thermal properties within numerical simulation workflows.

Name thesis:degree_name

Master of Science (MSc)

Discipline thesis:degree_discipline

Engineering – Chemical & Petroleum

Grantor

Schulich School of Engineering

Year dc:date.issued

2025

Author dc:creator

• Madani, Seyed Ali

Advisor dc:contributor.advisor

• Kantzas, Apostolos

Committee members dc:contributor.committeemember

• Hassanzadeh, Hassan
• Chen, Nancy

• Thermal Conductivity
• Porous Media
• Machine Learning
• Graph Neural Networks
• Cross-property Relationships

Statement dc:rights

• Unless otherwise indicated, this material is protected by copyright and has been made available with authorization from the copyright owner. You may use this material in any way that is permitted by the Copyright Act or through licensing that has been assigned to the document. For uses that are not allowable under copyright legislation or licensing, you are required to seek permission.

Language dc:language.iso

en

DOI

https://dx.doi.org/10.11575/PRISM/48993

OAI identifier oai:identifier

oai:ucalgary.scholaris.ca:1880/121403