Permafrost degradation driven by climate change poses critical challenges for geotechnical engineering, threatening the stability of northern infrastructure and underscoring the need for predictive tools that support long-term resilience planning. This PhD dissertation presents a novel large-strain thermo–hydro–mechanical (THM) modelling framework to simulate freezing and thawing processes in permafrost terrain. An open-source Python 3 modelling package, frozen-ground-fem, was created to integrate heat transfer, pore fluid migration, and mechanical deformation within a unified finite element framework. The research progresses from a comprehensive literature review and mathematical formulation to numerical implementation, benchmarking, and validation against established results, forming the foundation for applied case studies in contrasting permafrost environments.

A new methodological approach is introduced for THM modelling of permafrost terrain. This includes a procedure for generating physically consistent initial conditions using a coupled THM spin-up technique, which enables subsequent multi-decadal forecast simulations to begin from a realistic, dynamically equilibrated state when evaluating future climate scenarios. Model applications using frozen-ground-fem reveal distinct behaviours between warm and cold permafrost systems:​ warm permafrost exhibits progressive, cumulative settlement driven by seasonal freeze–thaw cycling and incomplete recovery, whereas cold permafrost remains comparatively stable but displays sensitivity to long-term warming. These findings highlight the importance of coupled THM modelling in capturing thaw-induced consolidation processes that thermal-only analyses cannot resolve.

The structured approach to THM modelling of frozen ground delivers a transparent, reproducible, and openly accessible framework that supports geotechnical design and climate-resilient infrastructure planning in cold regions. Beyond the presented case studies, the frozen-ground-fem package provides a versatile platform for future development, including site-specific calibration, multi-dimensional analyses, and integration with laboratory and field datasets, thereby strengthening the link between theoretical research and practical engineering applications.