Sajith Velmurugan Subalakshmi.

Soft actuators can release stored elastic energy far faster than their drive can supply it, by snapping between two stable states. Designing one means predicting the threshold at which that snap fires - and for a singly clamped preloaded strip, no closed form prediction existed. My MSc research derived one.

Working from Kirchhoff rod theory, I obtained the preloaded shape by constrained energy minimisation with a torsional boundary layer treated by matched asymptotics. Extending the model to the actuated stage turned on recognising that the imposed clamp rotation acts as a twist boundary condition rather than a bend or a tilt - the identification the whole model rests on, reached only after four earlier formulations were built and rejected. The envelope theorem then yields the actuation moment in closed form.

With zero fitted parameters, the model predicts the preloaded tip position to within 0.59% of a geometrically nonlinear shell FEM. Where it disagrees, it disagrees predictably: reducing the response to a single coordinate over constrains the structure, so the model must over predict the moment - and it does, by the anticipated factor. The energy functional coincides term for term with the published ribbon model of Chi et al. (Science Advances, 2022), extended here with a snap threshold analysis that work does not contain.

A differentiable finite element demonstrator in JAX captured the qualitative snap response, with the reaction force obtained by automatic differentiation of the strain energy.

This project converted a four stroke engine to run entirely on compressed air rather than fuel - no combustion, and therefore no emissions. Compressed air is stored, heated through a nichrome coil so it expands, then fed via a divergent nozzle into the cylinder to drive the piston.

The system is automated with a PLC, a proximity sensor detects top dead centre at the cylinder head and signals the controller, which times a solenoid valve to release air at precisely the right point in the stroke. I built the rig and tested output power and efficiency against the theoretical model.

Over several seasons with university racing team I worked on three competition karts, each produced entirely from scratch rather than assembled from a kit.

For the black and yellow car I did the full SolidWorks assembly - chassis, drivetrain, steering, seat and bodywork, and then worked on the build that came off that model. Taking my own CAD through to a running vehicle showed me quickly where a design is elegant on screen and awkward in the workshop.

Across the three builds the work covered producing and interpreting engineering drawings, manufacturing components with hand and power tools, assembling chassis, drivetrain and steering, and inspecting each finished car against race specification. Competition deadlines meant iterating fast and solving problems as a team.