Farsi Computational Mechanics Lab
Differentiable simulations and operator learning for Earth science and engineering
I lead the Farsi Lab, where we build high‑performance numerical models to tackle complex problems in Earth sciences and engineering. Our research advances nonlinear solid mechanics, discrete fracture mechanics, contact processes, fluid flow in porous media, and tightly coupled multiphysics systems. We develop and deploy finite‑ and discrete‑element methods, and adjoint‑based data assimilation and optimisation, also using a newly developed finite element‑based operator‑learning framework. Focus areas are:
-
Fracture, faulting, and contact
Research on these processes in heterogeneous solids spans rock and ice mechanics, fibre‑reinforced concrete tunnel linings, and catalyst pellet strength.
-
Multiphase flow and transport
Simulations of flow and transport in porous media address wellbore stability, geothermal wells, and the permeability of growing faults.
-
Strongly coupled multiphysics
Models of strongly coupled processes include Arctic sea‑ice dynamics.
-
Inverse modelling and data assimilation
Adjoint methods and operator learning support missing‑physics discovery and digital twins of Westerly granite.
If you'd like to learn more about our work in rock mechanics, watch my invited seminar for the American Rock Mechanics Association, Bridging Computational Mechanics and Machine Learning in Geomechanics below:
I'm always interested in meaningful collaborations across academia and industry — if you have an idea or project in mind, let's talk. Exceptional students are also invited to reach out about MSc and PhD projects.
Current students
Collaborators
Previous students