For over eight decades, the aerospace industry operated under a simple rule.
Smoothness equals speed.
It seems obvious, doesn’t it? If you want less drag, you polish the surface until it shines. This dogma came from Ichiro Tani in 1940. He argued that any surface roughness—no matter how tiny—would trigger chaotic airflow and spike resistance. We built bullet trains, jets, and cars based on that assumption.
But nature is rarely this neat.
In 1989, Tani himself reconsidered old data. He suggested that roughness might not always be the enemy. Decades later, a team at Tohoku University took that whisper and turned it into a roar.
Micro-roughness, macro results
Aiko Yakino and her colleagues have done the unthinkable. They reduced aerodynamic drag by 43.6%.
How?
By making things rougher.
They used distributed micro-roughness (DMR). To the naked eye, these surfaces look smooth. The irregularities are microscopic. Tiny glass beads, diameters of 38 to 53 micrometers, or sandblasted concave pits. From a hydrodynamic perspective, these coatings are barely a blip. Their height is just 1% of the boundary layer —that thin sheet of air hugging the vehicle.
Yet that slight imperfection delays the moment airflow goes chaotic. It keeps the air laminar, flowing in orderly, low-friction sheets for much longer than a perfectly smooth surface would.
“Roughness may not necessarily only promote turbulent.”
— Ichiro Tani (1989)
The measurement problem
Proving this required new tools. Old wind tunnels were cheating.
Support rods. Wires. The very structures holding the test models disrupted the air, masking the subtle benefits of micro-roughness. You couldn’t measure what you couldn’t isolate.
The team used the 1m-MSBS. It’s the world’s largest magnetic support balance.
The model levitates.
Electromagnetic force suspends it 1.07 meters into the wind stream. No rods. No contact. The airflow remains pure.
They tested across a massive range of Reynolds numbers —from $0.35 \times 10^6$ to $3.6 \times 10^6$. This dimensionless number predicts whether fluid will flow smoothly or break down.
The results were stark.
On the DMR-coated models, the critical threshold for turbulence jumped from roughly $1.9 \times 10^6 to $2.2 \times 10^$. The window for efficient, laminar flight expanded significantly. Up to the highest speed measured, the rough surface beat the smooth one.
Not pressure, but friction
So what actually happens in the air?
Drag usually comes from two places:
- Pressure resistance. The air separates from the object’s rear, creating a low-pressure vacuum that sucks the object back. Think golf balls. Dimples force turbulence to prevent this separation.
- Frictional resistance. The physical viscosity of air rubbing against the skin.
DMR attacks friction, not pressure.
To confirm this, the team ran Large Eddy Simulation (LES) models with up to 45 million cells. They painted models with fluorescent oil to watch the flow visually. The numbers didn’t lie. The drag reduction far exceeded the theoretical maximum gain from eliminating separation alone.
If you erased pressure resistance completely, it would only account for about 20% of the savings.
The rest? Reduced friction.
It is a direct contradiction of the golf ball logic. Golf balls invite chaos to stay attached. DMR enforces order to reduce rubbing. Opposite mechanisms. Different goals.
Why sharks miss the point
Shark skin technology—ribs running lengthwise like grooves—has been the gold standard for passive drag reduction. But it is fussy. The ribs must align with the airflow. Change the angle, lose the benefit.
DMR is lazy in the best way.
It is omnidirectional.
Random.
Because the roughness is distributed without a pattern, the angle of attack matters far less. There are no moving parts. No electricity required. Just texture.
The implications are heavy. Lower drag means less fuel. Less fuel means lower costs and smaller carbon footprints. It applies to aircraft, ships, maybe even your car next time.
The Tohoku team says they will optimize the shape and density next.
I wonder how long it will take for engineers to un-learn 80 years of polishing?
