Current research · 01
Wound-Rotor Synchronous Machine Testing
Graduate research in Columbia University's Motor Drives and Power Electronics Laboratory — putting a wound-rotor synchronous machine on a bench, measuring what it actually does, and holding the model to account against it.
Context
Machines are only as good as the measurements you can take of them.
A wound-rotor synchronous machine has something most machines don't: a field winding you can drive independently of the stator. That extra degree of freedom is what makes the machine interesting — excitation becomes a control input rather than a fixed property of a magnet — and it is also what makes it hard to characterise, because now the machine's behaviour is a surface over two inputs instead of a curve over one.
My work in MPLab sits on the experimental side of that problem. The machine has to be instrumented, driven through defined operating points, and logged cleanly enough that the resulting data can be compared against a MATLAB/Simulink model rather than merely described by it.
This is active research. This page describes the scope and method of the work rather than reporting results, which are not yet published.
Where I sit in it
- Machine testing. Bringing a wound-rotor synchronous machine to defined operating conditions and recording its response.
- Drives & power electronics. The converter side that supplies the machine — what it commands, and what it actually delivers under load.
- Data acquisition. Voltage, current, position and speed captured with enough fidelity and time alignment to be worth analysing.
- Experimental validation. Comparing measured behaviour against MATLAB/Simulink models, and treating disagreement as information rather than noise.
- Controls. Structuring excitation and test sequences so each run isolates the effect being studied.
System
The bench, as a signal path.
Power flows left to right along the top. Measurement flows back along the bottom. The interesting part is the loop that closes between them.
Block diagram of the test arrangement. Illustrative — not a wiring schematic.
Principle
Two inputs, one rotating field.
Drag the sliders. Stator excitation sets the rotating field; rotor field current sets how strongly the rotor locks to it. In a wound-rotor machine you control both — which is exactly why it needs testing across a surface of operating points rather than a single curve.
Illustrative animation of machine principle — not measured laboratory data.
Method
How a run works
- Define the point. Fix speed, load and excitation so the run answers one question.
- Instrument first. Decide what has to be measured, and at what rate, before energising anything.
- Acquire. Log electrical and mechanical quantities together, time-aligned, so they can be cross-checked afterwards.
- Compare. Run the same conditions through the Simulink model and put the two traces on the same axes.
- Interrogate the gap. Where measurement and model disagree, decide whether the model is wrong, the setup is wrong, or the assumption underneath both is.
Why this matters beyond the bench
Every traction drive, generator set and electrified powertrain is designed against a model long before hardware exists. The value of a lab like this is that it is where the model gets its credibility — or loses it. That habit, of trusting measurement over expectation, is the same one that shows up in the rest of the work on this site.