Maglev Hyperloop Model

This project is a 1:20 scale physical model of a magnetic levitation transit system, built to validate passive Halbach-array levitation over a conductive aluminum track. The goal is to understand the force-gap relationship empirically before committing to a larger prototype, and to build intuition for the electromagnetic trade-offs that define real hyperloop engineering.

Most student maglev demonstrations use active electromagnets with feedback controllers — the levitation is real but the control complexity obscures the underlying physics. I wanted to work with passive levitation, where the lift force emerges purely from the interaction between a Halbach array and induced eddy currents in a conductive guideway. The challenge: passive lift only exceeds drag above a critical velocity, which means the system must be moving to stay aloft. Designing a track geometry that lets a slow-moving model reach that threshold was the core constraint.

I started from first principles using the method of images to estimate lift-to-drag ratio as a function of array geometry and track conductivity. After running parametric sweeps in MATLAB, I selected a 4-magnet Halbach array (NdFeB N52) and a 6061-T6 aluminum track of 4 mm thickness. The model uses a closed-loop belt drive powered by a brushless DC motor to achieve the minimum levitation velocity of ~1.2 m/s. Force measurements are taken via a strain-gauge load cell mounted to the chassis.

The chassis is laser-cut acrylic with 3D-printed motor mounts. The Halbach arrays were hand-assembled using a custom jig to maintain pole alignment to within 2°. Wiring runs to an Arduino Mega that logs force, velocity, and gap distance at 50 Hz. The track is a single straight section, 1.2 m long, which is long enough to reach levitation speed but short enough to fit on a workbench. Current work is on the data acquisition pipeline — I'm using Python to parse the serial output and plot lift/drag curves in real time.

The most important thing I learned is that passive levitation systems are inherently unstable in the lateral direction — you get vertical lift "for free" but you have to engineer lateral guidance separately. I underestimated this in v1 and the model derailed at speed. The fix was adding flanged guide rollers, which introduced friction I had to account for in the force budget. More broadly: physical systems always have a constraint you didn't model. Build early, measure often, and treat every unexpected result as data.