Propulsion with no moving parts.
Magnetohydrodynamic (MHD) propulsion pushes conductive fluids using only electric and magnetic fields — the Lorentz force does the work. You'll model the physics, simulate the channel and build a working thruster on the bench.
Keep an unstable system standing.
An inverted pendulum on wheels: derive the dynamics, design PID and optimal controllers, integrate sensors and actuators, and watch your robot balance itself. A hands-on introduction to control and mobile robotics.
Teach a network to see stiffness.
Structures become binary meshes; simulations become a dataset. You'll generate the data, train supervised models that predict performance in milliseconds instead of minutes, and validate them against the physics.
A valve that only works one way — with no valve.
Fluidic diodes let fluid pass easily one way and resist it the other, with no moving parts — like Tesla's valve. You'll run CFD, optimize the geometry, 3D-print it and measure it. Applications: respiratory devices, biomedical systems, microfluidics.
Let the optimizer draw the hydrogen highways.
Design PEMFC and SOFC flow fields for efficiency, durability and cost. Multiphysics modeling (electrochemistry, mass and heat transport, stresses), multi-objective and degradation-aware optimization, manufacturing constraints and experimental validation.
Shape the channel, steer the field.
Develop topology optimization for MHD systems to maximize energy conversion and flow control: coupled MHD solvers, new parameterizations, adjoint sensitivities and experiments. Applications: MHD generators, liquid-metal pumps, aerospace flow control, biomedical devices.
Replace hours of sensitivities with one forward pass.
Build ML surrogates that replace expensive sensitivity calculations: CNN, GNN and transformer architectures, physics-informed constraints, hybrid optimization and uncertainty quantification. Applications: lightweight automotive parts, aerospace components, heat exchangers.
Recommended reading
The books and papers we hand to every new member of the lab.
No prior optimization experience required for undergraduates — curiosity and commitment are what matter. Graduate positions are funded through FAPESP, CNPq and CAPES.