Joan Onyango is a geotechnical researcher specializing in physics-informed machine learning for earthquake-induced surface fault rupture prediction. Her work integrates high-fidelity three-dimensional discrete element method (DEM) simulations with surrogate modeling frameworks to translate particle-scale rupture mechanics into engineering-scale hazard predictions. By combining computational geomechanics with data-driven modeling, she develops efficient, physically grounded tools capable of predicting deformation patterns, rupture widths, and near-surface soil responses under complex faulting conditions.
Her research bridges traditional soil and rock mechanics with modern AI methodologies, enabling scalable prediction of nonlinear geotechnical phenomena that are otherwise computationally intensive to simulate directly. Through large DEM-derived datasets and interpretable machine learning approaches, she aims to improve infrastructure risk assessment and contribute practical tools for seismic hazard mitigation.
In parallel, Joan’s background in mining engineering centers on slope stability in complex geological environments. Her work examines the behavior of subsurface conditions, particularly karst voids and weak rock formations that pose significant risks to open-pit mine safety and long-term stability. Building on her expertise in physics-based modeling, she is extending her research toward data-informed approaches for slope risk assessment in mining contexts.
Beyond computational modeling, Joan is also interested in nature-based solutions for sustainable mining. She investigates the use of vetiver grass as a deep-rooted bioengineering strategy for slope stabilization and mine land rehabilitation. By integrating quantitative geotechnical analysis with low-cost, climate-resilient ecological approaches, her work advances safer and more environmentally responsible resource extraction practices.