Hi, I'm Andrea Pavan. I specialize in fluid dynamics and aircraft architecture.
View my workI have been 3D printing as a hobby since high school, designing and making most of the objects myself from scratch. At first it felt like magic, turning my ideas into a tangible object. However, I soon realized that the CAD stage was a major bottleneck; the time spent drawing sketches often left my printer sitting idle for days.
During my engineering studies I began exploring simulation-based design, numerical optimization and topology optimization tools. These approaches are a match made in heaven with 3D printing, taking full advantage of the design freedom it offers. They also cut the time spent on CAD modeling and achieve better performance with fewer design iterations and less material waste.
For my final project I joined the LiftUp Student Team. I implemented a multi-fidelity inverse-design method coupled with a genetic algorithm to design the airfoil and wing of the Kowalsky drone. I am proud of this work; the drone competed in the Air Cargo Challenge and finished with good results.
Since then my focus has shifted towards multidisciplinary simulations. For my master's thesis I developed a meshless GFDM method and applied it to the thermo-mechanical simulation of the 3D printing process.
In my free time I have also written several analysis tools, including a 3D panel method and a compressible CFD solver.
After graduation, I completed my education with an Advanced Master in aircraft architecture and an end-of-studies internship on propeller aeroacoustics.
Developed a meshless method for unsteady heat transfer and linear elasticity, using it to investigate warping and inter-layer bonding strength.
Implemented an Embedded Multi-Fidelity Inverse Design (EMFID) method, integrating XFoil and a genetic algorithm to optimize the shape of the airfoil.
A C library reimplementing QPROP for propeller aerodynamics analysis, enhanced with support for airfoil polars. It includes Python bindings and a web UI.
Currently developing a FVM solver from scratch for high-fidelity aero-thermo-mechanical simulations on 3D unstructured meshes.
Implemented a low-order 3D panel method for quick low-fidelity aerodynamic analysis, suitable for preliminary design of wings and rotors.
A simple flight simulator running in real-time on a web browser, solving the nonlinear equations of motion of a F450 quadcopter drone.
Analyzed experimental data on subsonic jets, airfoil pressure distributions and wakes, turbulent BL (hot-wire anemometry) and vortex shedding (PIV).
Redesigned the bellcrank in the Sparker TL-Ultralight aircraft's flight control system for LPBF, cutting weight by 65% while meeting all requirements.
Optimized wing geometry and structure with a genetic algorithm to minimize fuel use while meeting aerodynamics, weight, structural, range and CS-23 requirements.
Developed a minimax chess engine that uses a MLP neural network as evaluation function, trained on Stockfish scores to play at low search depths.
A simple slicing software for 3D printing, running entirely in a web browser. Developed just for fun, serving as a sandbox for testing out new ideas.
Designed and built a CoreXY 3D printer using 3030 aluminum profiles, a V6 bowden extruder, a standard Marlin firmware and OctoPi.
Hacked together a simple quadcopter drone using a F450 frame, MN2213 motors and a PixHawk board with ArduPilot.