The Ranque-Hilsch vortex tube is one of thermodynamics' more delightful curiosities: feed compressed air into a swirl chamber, and it spontaneously splits into a hot stream (up to +200°C) and a cold stream (down to −50°C) at opposite ends. No moving parts. No refrigerant. No magic — just angular momentum redistribution as outer gas layers do compressive work on inner ones. Industrially, they're used for spot-cooling CNC tooling. Could we scale one up to cool, say, Manhattan in July?

Manhattan's peak summer cooling load is roughly 10 GW thermal. We want a cold stream about 20 K below ambient (so ~15°C output on a 35°C day).

The mass flow:

Q = ṁ · cp · ΔT
10×10⁹ W = ṁ · 1005 J/(kg·K) · 20 K
ṁ ≈ 500,000 kg/s

That's half a million kilograms of cold air per second. At 6 bar and 290 K, density is ~7 kg/m³, so we need ~70,000 m³/s of pressurized airflow — equivalent to ~250 large turbofan engines feeding the inlet continuously.

The tower dimensions. Vortex tubes scale poorly because the swirl Reynolds number grows linearly with diameter, but viscous dissipation in the boundary layer doesn't help separation — the effect actually degrades above ~10 cm diameter as turbulent mixing erases the temperature gradient. To preserve the effect, we'd need a bundle of, say, 10⁶ parallel 5-cm tubes, each handling 0.07 m³/s. A 200-meter-tall tower 80 m on a side, packed with vertical tubes like a monstrous heat exchanger, gets us there geometrically.

The fatal number — energy. Vortex tube COP (cooling delivered ÷ compression work input) is brutally low: roughly 0.1–0.2 in the regime we want. To deliver 10 GW of cooling:

P_compressor ≈ 10 GW / 0.15 ≈ 67 GW electricity

Compare that to a conventional vapor-compression chiller (COP ≈ 4) at 2.5 GW, or a district-scale absorption chiller running on waste heat at essentially free marginal cost. The vortex tower would burn through ~27× more electricity than existing technology — roughly the entire summer output of every nuclear plant in the Northeast, just to cool one borough.

Why this is so bad: A vortex tube is fundamentally a throttling device dressed up with rotation. The total enthalpy is conserved (it's adiabatic), so you're not creating cold — you're sorting molecules by kinetic energy and throwing half away as a hot exhaust. Half your compressed air leaves as 200°C waste that you then have to dump to ambient, taking ~80% of your input work with it.

The compressor heat problem compounds. Compressing 70,000 m³/s of air to 6 bar adiabatically heats it to ~230°C. Before it enters the tubes, you must intercool it back to ambient — itself requiring another ~40 GW of heat rejection through cooling towers. The Hudson River would visibly warm.

The one redeeming pitch: in places with stranded curtailed renewables (West Texas wind at midnight), maybe a vortex AC tower is acceptable as an inefficient-but-mechanically-trivial heat sink — no refrigerant leaks, no compressors with moving seals, just a giant whistle. It's the thermodynamic equivalent of heating your house by lighting a $100 bill: it works, but…