Copernic Mensah

Copernic Mensah

Biography

Copernic Mensah is a computer science undergraduate at Hampton University, working
toward a Ph.D. During the school year, he researches nuclear fusion at The Center for Fusion
Research and Training, modeling Poincaré plots and experimenting with stellarator coilwinding.
He is currently interning at MIT’s Hypersonics Research Laboratory, working under
Dr. Wesley Harris and Dr. Stewart Isaacs to study shape-enhanced aerodynamic dust removal
from solar panels in West Africa. Before that, he served as a Quantitative Developer Intern at
the Nwagbara Group LLC, maintaining high-frequency trading engines in Rust. At the Center
for Applied Biomechanics and Rehabilitation, he designed a 2D robotic hand exoskeleton game
for stroke patients. He received recognition at NASA’s Solar Energy and Science Gateways
hackathons for his work in astronaut augmented reality and LLM data portals for reproducible
science. Across every difficult project, Copernic’s approach remains the same: understand the
problem analytically, then solve it numerically.

Uniform Dust Deposition and Orientation Detection for Wind Tunnel
Testing of Photovoltaics
Copernic Mensah1, Dr. Stewart Isaacs2, Dr. Wesley Harris2
1Department of Computer Science, Hampton University
2Department of Astronautics and Aeronautics, Massachusetts Institute of Technology

Dust accumulation on photovoltaic (PV) module surfaces limits power generation. While
manual, automatic, and semi-automatic cleaning strategies exist to remove surface dust, no
passive cleaning strategy currently exists. This project investigates passive dust removal through
wind tunnel experimentation at 10 m/s on analog PV panels, testing three angles of attack (0°,
15°, and 30°). We first developed a standardized dust deposition method by comparing four
candidate techniques—manual pile-and-push, manual mesh tapping, automatic mesh sifting, and
manual mesh shaking—evaluated on deposition time, evenness, particle uniformity, and spread.
Pile-and-push was fastest (1:30) but produced uneven coverage, corner gaps, and dust clumping.
Mesh tapping (1:47) achieved even, uniform coverage but required strenuous, repeated effort,
making it impractical across many trials. Automatic mesh sifting (2:30) produced even, uniform,
fine, gap-free coverage without manual strain and was selected as our standardized deposition
procedure. Alongside this work, we developed a ChArUco board-based computer vision model
that uses corner detection and PnP (Perspective-n-Point) pose estimation to detect panel tilt. The
findings of this work aim to improve PV module performance.