Hichem Khalil Cherfouf
A Machine Learning Framework for Caprock Sealing Efficiency: Prediction of Breakthrough and Snap-off Pressure from Petrophysical Properties.
Rel. Vera Rocca. Politecnico di Torino, Corso di laurea magistrale in Georesources And Geoenergy Engineering, 2026
|
Preview |
PDF (Tesi_di_laurea)
- Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (4MB) | Preview |
Abstract
The geological storage of carbon dioxide and hydrogen depends on the capacity of a caprock to retain a buoyant fluid column, a capacity quantified by two capillary sealing pressures: the breakthrough pressure, at which the fluid first forms a connected path across the seal, and the snap-off pressure, the residual bound retained after re-imbibition. Direct laboratory measurement of these pressures is slow and the resulting datasets are small, so the empirical permeability correlation of Thomas et al. (1968) has remained the standard predictive tool for over half a century despite being calibrated on a narrow range of rocks. This thesis develops a machine-learning framework to predict caprock sealing pressures from routinely available petrophysical properties (porosity, permeability, and confining pressure) and benchmarks it against the Thomas correlation under source-aware validation.
The framework is built on a database of 121 plug-scale measurements compiled and quality-controlled from the experimental literature, spanning shale, mudstone, clay, anhydrite, and carbonate seals tested with CO2, N2, CH4, and H2
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
Aziende collaboratrici
URI
![]() |
Modifica (riservato agli operatori) |
