I’m a mechanical engineering PhD student at WPI. My research simulates what decides whether two colliding droplets bounce apart or merge, which matters for separating water from crude oil with ultrasound. A lot of that work is software: building cases, adding physics to the solvers, running them on workstations, clusters and GPUs, and checking whether the answers are right. I’m taking a break from the PhD right now.
Some specifics:
- Two changes merged into MFC, an open-source solver for compressible multiphase flow and a 2025 Gordon Bell Prize finalist: a fix for a ten-month-old bug that corrupted grid coordinates in post-processing, and an
--archiveoption that files each run’s inputs and outputs with a manifest. More - Heat conduction, temperature-dependent surface tension and temperature-dependent viscosity added to MFC, each with its own validation case against an exact solution or theory. More
- 3D droplet collisions up to 96 million cells on about 24 A100 GPUs on NCSA’s Delta, and a stretched grid that ran the same collision about 2.9 times cheaper. More
- A droplet bounce resolved in Basilisk, with 128 nm cells in the film between the droplets. More
- A 53% speed-up I withdrew after tracing it to the boundaries of my simulation box. With the box fixed, the bubble slows drainage instead. More
- Two years on the Fluent testing team at Ansys, after verification and validation work at DEKA.
Research is listed first. The rest is tooling, an older CFD study, and the computers I run at home. I’ve also written up two tools I built for myself: deleting 108,498 emails without losing one, and garbage-collecting my AI assistant’s memory.
