A recent research demonstrates real-time control of weakly conducting fluid flows using Lorentz Forces, powered by Koopman-based machine learning.
Controlling how fluids flow isn’t just something for nature documentaries—it’s at the cutting edge of engineering research. From designing faster planes to creating more efficient industrial mixers, fluid control has huge applications. But how do you steer a flowing liquid, especially one that can’t be touched?
That’s exactly what a team of researchers from the Czech Technical University and LAAS-CNRS tackled in their latest study on real-time control of magnetohydrodynamic (MHD) flows. And the best part? They did it without touching the fluid at all—using only Lorentz Forces, and clever machine learning tricks.
Let’s break it down in plain language—engineer-to-engineer!
In simple terms, MHD deals with how electrically conducting fluids behave in the presence of electric and magnetic fields. Think seawater, liquid metals, or electrolytes.
Why it's cool:
Where it’s useful:
But here's the catch: fluids are already notoriously hard to control because of turbulence and chaos. Add electromagnetic effects and things get way more complicated.
The researchers used a shallow dish filled with a weakly conducting liquid (water + sulfuric acid). Around it, they placed:
Here's the magic sauce:
Electric field + Magnetic field = Lorentz Force, which pushes and pulls the liquid in different directions.
By tuning the electricity and magnet strength in real-time, they could shape the flow into vortices, jets, and other patterns!
Now, you might wonder: how can we predict and control such a chaotic system in real time?
Instead of solving complex fluid equations (which take hours to compute), they used a data-driven approach powered by the Koopman operator.
They trained the model by observing how the fluid moved in response to different electric/magnetic inputs. Using this, they built a Koopman-based Model Predictive Controller (KMPC) that predicts future flow and adjusts controls on the fly!
One of the most impressive feats? The whole thing runs on a standard laptop!
The team conducted multiple experiments to test flow control:
They created different flow shapes:
Result?
In all cases, the fluid adapted to the desired pattern within seconds and remained steady—an incredible feat for chaotic fluids!
They even controlled vorticity (how much the fluid spins):
This study isn’t just a fancy lab experiment—it shows that:
This research shines a spotlight on a future where engineers don’t have to wrestle with complicated fluid equations, but can train models and use data to control complex systems in real time.
The fusion of:
And all of this is run on a laptop—proving how accessible and scalable this technology can be for engineers worldwide.
| Feature | Details |
|---|---|
| Control Type | Electric & Magnetic Fields |
| Method | Koopman Model Predictive Control (KMPC) |
| Setup | 4 electrodes + 4 electromagnets + camera |
| Hardware | Standard Laptop (real-time) |
| Controlled | Flow velocity & vorticity |
| Applications | Mixing, cooling, robotics, aerospace |
This work is an excellent showcase of how engineering, machine learning, and fluid mechanics can merge into practical, high-performance systems.
As fluid engineers, researchers, or industrial designers, this approach offers a new way to think about flow control—faster, smarter, and contactless.
Magnetohydrodynamics (MHD) - How liquids that conduct electricity (like saltwater or plasma) behave when exposed to electric and magnetic fields.
Lorentz Force - The invisible push a fluid feels when electric and magnetic fields interact—like steering water with magnets!
Model Predictive Control (MPC) - A smart controller that predicts the future of a system and adjusts actions in real time for best performance. - More about this concept in the article "Turning Waste into Watts | How Smart Control is Powering Energy-Free Wastewater Plants!".
Koopman Operator - A mathematical trick that turns complex, chaotic systems into simple linear models so computers can predict them fast.
Vorticity - How much a fluid spins or swirls—high vorticity means strong circular motion, like a mini whirlpool!
Particle Image Velocimetry (PIV) - A camera-based technique that tracks particles floating in liquid to measure how the fluid flows.
Adam Uchytil, Milan Korda, Jiří Zemánek. Real-time control of a magnetohydrodynamic flow. https://doi.org/10.48550/arXiv.2507.12479
From: Czech Technical University in Prague; LAAS-CNRS, Université de Toulouse.
