The Fan Wall

Every wind turbine you have seen follows the same logic: one rotor as large as possible, as high as possible. It works brilliantly, and it needs cranes, open land and years of permits. Our question is the opposite one. Can many small rotors, built into the places where cities already are, make electricity at a price that makes sense? The Fan Wall idea was first proposed within the Local Solar System Foundation in 2020; this page is its open, updated model.

This page is written for people who might fund, build or test the idea. It is deliberately not a sales brochure. We ran the numbers for our first design, found where it fails, redesigned it, and ran the numbers again. The short version: a wall of computer fans on a facade cannot work, a crown of larger ducted rotors on a windy roof edge can make real electricity, and it will not beat solar panels on price unless several things we can test turn out well. Every figure below comes from our open model (code, data).

Version 1, and why it does not work

The first idea was a building facade, 40 m long and 15 floors high, covered in about 110,000 fans of 120 mm, each run backwards so its motor becomes a generator. It is a lovely picture. It fails for three reasons, in order of importance.

1. Air cannot pass through a closed wall. A turbine takes energy from air that flows through it. On a solid facade the wind stops (engineers call it stagnation) and slides up, down and sideways. Even if every fan had a 30 cm gap behind it, open at the sides and top, continuity sets a hard limit: all the air entering 1,800 m² of fans would have to leave through about 39 m² of gap, so it could pass the fans at no more than 2.2% of the wind speed. Power goes with the cube of speed, so the wall would keep well under 0.01% of what the same fans would make standing free. Measurements behind real ventilated cladding agree: 0.2 to 0.6 m/s of cavity airflow in winds of up to 5 m/s (Salonvaara et al., 2007). Switch the 3D model above to version 1 to see it: the wind streaks stop at the wall and slide away.

2. Tiny blades work badly. A 120 mm fan blade has a chord of about 2 cm. The air around it moves at a Reynolds number near 10,000, a regime where wings lose most of their lift-to-drag ratio: smooth airfoils improve “more than an order of magnitude” above a critical Reynolds number of about 70,000 (Lissaman, 1983). Measured centimetre-scale turbines reach power coefficients around 9% (Howey, Bansal and Holmes, 2011); a 10 cm computer-fan rotor used as a turbine delivered 28 mW at 5 m/s, about 4% of the wind’s power across it by our calculation (Federspiel and Chen, 2003). Computer fans are shaped to push air, not to be pushed by it.

3. Thousands of tiny generators waste what little they make. A fan motor delivering a fraction of a watt at a few volts can lose 10 to 40% in a diode rectifier alone. Our central estimate for the whole chain from fan to grid is 54%.

Even ignoring the first reason entirely (a free-standing wall of fans), version 1 would cost about $4.20 per kWh on a windy coast and $18 in a city, twenty to a hundred times the price of rooftop solar.

The physics that sets the limits

The power carried by wind through an area A is P = ½ ρ A v³, with air density ρ = 1.225 kg/m³. The cube is everything: at 6 m/s a square metre carries 132 W, at 12 m/s 1,058 W. Averaged over a year, a city roof with a 4 m/s mean wind offers about 75 W per m²; an exposed coast at 6.5 m/s about 321 W per m².

Betz limit. No open rotor can extract more than 16/27 ≈ 59.3% of the power flowing through its area, because the air must keep moving out of the way.

What a second rotor can add. Two rotors in series (like the stages of a jet engine) can reach at most 16/25 = 64% of the power through one rotor’s area, and infinitely many only 2/3 (Newman, 1986). A jet engine stacks stages because it pushes air through a confined core with a huge pressure difference from burning fuel. Free wind offers no such pressure: every extra stage slows the air further, and slow air spills around the device. Measured gains of contra-rotating wind rotors range from 0 to 60%, but well-controlled recent tests cluster at 2 to 10% (Adema et al., 2026 review).

What stator vanes can add. A rotor leaves the air swirling, and swirl is lost energy. At the low tip-speed ratios of small rotors the loss is large: the ideal Glauert rotor reaches 0.42 at a tip-speed ratio of 1, 0.51 at 2 and 0.57 at 5. Fixed vanes behind the rotor can recover part of it. Nobody has measured this on a wind turbine; on aircraft propellers such vanes gave 2 to 5% (Li et al., 2017).

What a duct can add. A duct with a flat brim (the “wind lens” of Kyushu University) lowers the pressure behind the rotor and draws more air through it. In wind-tunnel tests a compact brimmed diffuser raised the power coefficient from 0.37 for the bare rotor to between 0.7 and 0.88, relative to the rotor area, but to only 0.48 to 0.54 relative to the whole brim (Ohya and Karasudani, 2010). That is the catch: per square metre of frontal area, ducts have never beaten the Betz limit, “in practice power augmentation factors above 3 have never been achieved”, and ducted turbines have had “more than 50 years. So far without any commercial success” (van Bussel, 2007). The most expensive recent attempt, FloDesign/Ogin, removed its turbines within six months and was liquidated in 2017 (Gipe, 2017).

Size matters because of Reynolds number. We calculated the best possible rotor at each size with a blade-element momentum model (Glauert’s optimum with wake rotation, tip and hub losses, and a lift-to-drag ratio that falls with Reynolds number, from measured low-Reynolds airfoils such as Selig’s SG6043):

Rotor diameter	Blades, tip-speed ratio	Reynolds number (75% radius)	Best design Cp
4 cm	8, 1.25	4,500	0.23
12 cm (computer fan)	8, 1.5	12,000	0.29
25 cm	8, 2.0	20,000	0.36
50 cm (improved module)	6 to 7, 2.5	43,000	0.39 to 0.40
1 m	6, 3.0	74,000	0.42
7 m (small turbine)	8, 4.5	265,000	0.49

These are ceilings for purpose-built blades, before real-world losses. As a check, a purpose-designed 39 cm portable turbine measured 0.32, where our model gives about 0.39 for that size (Kishore, Coudron and Priya, 2013): we therefore assume first hardware reaches 65 to 90% of the design value, and far less for centimetre rotors, where laminar separation dominates.

Wind direction. A fixed module only uses the part of the wind blowing into it. Power falls roughly with the cube of the misalignment angle for open rotors and more gently for ducts (Tamaro et al., 2024; Richmond-Navarro et al., 2022). With a moderate prevailing wind, a module facing it keeps about 47% of the energy an always-aligned rotor would get. Version 1 assumed 100%.

The improved design: a crown on the windward roof edge

Take everything above and the design almost writes itself: bigger rotors where the Reynolds number is kinder, in ducts that protect and accelerate them, standing free where air can pass, facing the prevailing wind at the place on a building where the flow is strongest. That place is the roof edge. Air that hits the facade rises and turns over the edge; above an upwind roof edge the wind is 1.06 to 1.25 times the speed at roof height, but only 0.38 times when that edge is downwind (Mertens, 2006).

Each module is 725 mm square: a 0.5 m seven-blade rotor in a compact wind-lens duct, five stator vanes that recover some swirl and carry a 50 to 300 W generator, a stainless guard mesh and one sealed power board with maximum-power-point tracking. Two rows make a 1.45 m tall crown on steel rails above the parapet. One person can lift a module out and swap it.

The blades are not guessed. The model designs them for the Reynolds number they will actually see, about 45,000 at 6 m/s: chord 62 mm near the root tapering to 36 mm at the tip, twist from 38° to 10°, a design tip-speed ratio of 2.5. At 6 m/s the rotor turns at about 700 rpm; its tips move at about 30 m/s in a 10 m/s wind, less than half the 79 m/s of a large turbine’s tips.

The honest numbers

Per square metre of frontal area is the fair way to compare: a facade, a roof crown and a highway barrier all compete for the same thing, a surface facing the wind. Central values, with the full pessimistic-to-optimistic range in the model data. Energy includes wind direction, turbulence (5 to 15% loss) and availability (92 to 97%). Cost of energy: 20 years at a 7% real discount rate, maintenance 2 to 5% of the investment per year.

Design	Cp on frontal area	kWh per m² per year, windy site / city	Capacity factor, windy / city	Installed cost per m²	Cost per kWh, windy / city	Simple payback at $0.19/kWh
A. Version 1, stock 120 mm fans (free-standing)	0.045	23 / 5	5.2% / 1.2%	$765	$4.21 / $18.0	never
A2. Custom 120 mm rotors	0.089	45 / 11	5.2% / 1.2%	$805	$2.23 / $9.45	never
B0. 0.5 m open rotors, crown	0.148	125 / 34	9.3% / 2.5%	$1,192	$1.19 / $4.37	never
B. Improved: 0.5 m ducted, stator, crown	0.164	159 / 43	10.7% / 2.9%	$1,074	$0.84 / $3.10	never (best case: 4.5 years windy)
B2. As B plus a contra-rotating rotor	0.174	168 / 46	10.7% / 2.9%	$1,245	$0.92 / $3.40	never
B3. B as a free-standing barrier	0.164	128 / 32	8.6% / 2.2%	$944	$0.92 / $3.66	never
Rooftop solar, same building		225	12%	$410	$0.19	10.6 years
Solar on the south facade		139	7.6%	$630	$0.47	31 years
10 kW turbine, 7 m rotor on an 18 to 30 m mast	0.35 on rotor	31,600 / 10,300 kWh per turbine	36% / 12%	$8,910 per kW	$0.35 / $1.08	27 years windy

What the table says:

• The duct earns its place, just. Per frontal m², ducted modules (B) beat the same rotors in the open (B0) by about 10 to 30%, mostly because they tolerate wind direction better. They also give a guard, a quieter tip and a place for the stator.
• The second rotor does not pay. It adds about 6% energy for 16% more cost. A jet-engine-style multi-stage wind module is physics-limited, not ambition-limited.
• Only windward modules pay. A crown all around the roof earns only 55% per module compared with the windward side alone, because the other sides face away from the prevailing wind most of the time. On our reference building (40 × 20 m, 45 m tall) the sensible crown is the windward side only: 60 m², about $64,000, about 9,500 kWh a year on a windy site, 2,600 in a city. The same money in rooftop solar buys about 155 m² of panels making about 35,000 kWh a year.
• In cities it cannot compete. Building-mounted micro wind has a poor record: 26 turbines in the Warwick Wind Trials averaged a 0.85% capacity factor (Encraft, 2009), and the three turbines on London’s Strata tower were switched off after noise and vibration complaints.

For scale, the US small wind fleet averaged a 14 to 16% capacity factor and $6,680 to $9,630 per kW installed in 2024 and 2025 (PNNL Distributed Wind reports). Our crown lands at about $6,300 per kW rated, in the same range, which is a sign the cost estimate is not fantasy.

What is proven, plausible and speculative

Proven (textbook or measured): the power in the wind and the Betz limit; the 16/25 ceiling for two rotors; that ducts raise power per rotor area but not beyond about Betz per frontal area; that small blades at Reynolds numbers below about 70,000 lose efficiency; that a closed wall has no through-flow; that roof edges accelerate upwind flow; that building-mounted micro turbines in cities have performed badly.

Plausible (supported, but not measured for this design): a 0.5 m rotor designed for Reynolds 45,000 reaching a power coefficient around 0.3 bare and 0.55 on rotor area in a compact brimmed duct; 75% chain efficiency from rotor to grid with 50 to 300 W generators and module-level MPPT; installed cost near $1,000 per m² for a first project.

Speculative (must be measured): how much ducts packed side by side lose to each other (we assume 10 to 30%); the real speed-up and turbulence at a crown on a given building; durability of thousands of bearings over 20 years of gusts, ice and dust; noise and vibration transmitted into the building; wildlife interaction with meshed ducts (large turbines kill an estimated 140,000 to 328,000 birds a year in the US (Loss, Will and Marra, 2013); glass buildings kill 365 to 988 million (Loss et al., 2014); nobody has data for meshed roof-edge modules).

What would have to be true

For the crown on a windy 6.5 m/s site, the cost of electricity depends on two things we can work on: how much power each square metre of frontal area delivers, and what it costs installed.

Cp on frontal area	$300 per m²	$500 per m²	$800 per m²	$1,100 per m²
0.15	$0.24	$0.40	$0.63	$0.87
0.20	$0.18	$0.30	$0.47	$0.65
0.25	$0.14	$0.24	$0.38	$0.52
0.30	$0.12	$0.20	$0.32	$0.43

Where it can win, and where it cannot

It can win:

• Windy tall buildings and coastal sites, as a complement to solar: wind at a roof edge peaks at night, in winter and in storms, when solar panels produce little.
• Roofs already full or shaded. A crown uses the edge, not the roof area, and does not compete with panels, plant or terraces.
• Off-grid power where cables are expensive: sensors, lighting, telecom equipment, weather stations on bridges, masts and coastal structures.
• Barriers that already exist: highway and railway noise barriers, sea walls and bridge parapets, where the structure is paid for.
• Visibility and education. A crown that turns is a public statement a solar panel is not; schools, museums and landmark buildings value that.

It cannot win:

• As a cheap source of bulk electricity. Large wind and solar farms cost $0.04 to $0.10 per kWh; nothing on a building comes close.
• In sheltered city streets and low-rise suburbs, where the mean wind is under 4 to 5 m/s at roof height.
• Flat on a facade. However the rotors are designed, a closed wall has no through-flow.

Development roadmap and what it would cost

Phase	Work	Decision it enables	Estimated cost	Time
0. Design study	Blade and duct design, CFD of one module and of a 3 × 3 array, electrical topology, cost model with a manufacturer	Is Cp on frontal area above 0.2 plausible in an array?	$30k to $60k	3 to 4 months
1. Wind tunnel	Three rotors, three ducts, stator on and off, then a 3 × 3 array; 3 to 15 m/s, yaw 0 to 45°	Measured Cp, array loss, start-up speed, noise	$60k to $120k	4 to 6 months
2. Prototype module	Generator, power board, sealed module; endurance, ice, rain and acoustic tests (IEC 61400-11 method)	Module efficiency and durability	$50k to $100k	6 months
3. Rooftop pilot	10 to 20 modules on a windward edge, ultrasonic anemometers or lidar, a year of data, wildlife camera	Real kWh per m², availability, noise in the building	$80k to $150k	12 months
4. Product path	Structural and facade engineering, IEC 61400-2 design basis, UK MCS 006 or US ACP 101-1 certification, electrical and fire compliance, insurance	A product a building owner can buy	$100k to $250k	12 to 18 months

Phases 0 to 3 need roughly $220,000 to $430,000 and two years. Each phase has a clear stop point: if the wind tunnel does not show a frontal Cp near 0.2 in an array, we stop there and publish why.

Who we need

• Aerodynamicist with low-Reynolds and ducted-rotor experience, to design and review the blades, duct and stator.
• A university or commercial wind tunnel, with a CFD engineer to plan tests and model array interaction.
• Power-electronics engineer for small permanent-magnet generators, active rectification, MPPT and DC collection from many modules.
• Structural and facade engineer for wind loads on the crown (EN 1991-1-4 / ASCE 7), fixings, ice and fatigue.
• Acoustician for tonal noise and vibration transmitted into the building.
• Urban wind specialist to measure and model the wind at candidate roof edges before anything is built.
• Certification and compliance: small wind standards (IEC 61400-2, -11, -12-1), electrical and fire codes, and an insurer willing to cover a pilot.
• Ecologist to design the bird and bat monitoring.
• Manufacturing partner in injection moulding or rotomoulding to price blades and ducts at volume.

A better idea is welcome

The Fan Wall is one answer to a wider question: how can the places where people already live make, store or save more clean energy than they do today? If the numbers on this page convince you that a different approach would do better, we want to hear it. Rooftop solar that also makes heat, thermal storage, energy recovered from water mains, waste heat or smarter timing of demand could all beat a wind crown on many buildings. The idea radar compares them, and each has its own article written with the same honesty we apply here.

Some members of the Foundation may have contacts who could hear a pitch. To be pitched, an idea needs a working, real-life model that actually produces power, with its output measured on an instrument, not claimed; that is exactly why this page ends in a wind tunnel and a rooftop pilot rather than a drawing. Paper ideas are welcome too: if the physics and the numbers are solid, we can publish them here on the website. But raising capital without a working model is very difficult, and nothing here is a promise of funding, returns or introductions. The standard is the one this page tries to meet: show the numbers, say what would prove you wrong, and propose the cheapest test.

How you can help

• Bring your expertise. If you work in any of the fields above, offer a task. Even an hour reviewing our model is valuable.
• Offer a windy roof. A tall building on a coast, hill or open plain with safe roof access could host the pilot. Tell us where it is and which way the wind usually blows.
• Share wind data. A year of wind measurements from a roof, bridge or coastal mast helps us choose the pilot site.
• Check our maths. The model is open. If you find an error, tell us: we would rather be corrected now than after a pilot.
• Propose a better idea. Send it through the contribution form with the problem it solves, the physics check, a rough cost per kWh and the cheapest test that could prove it wrong.

Sources

1. Salonvaara, M., Karagiozis, A., Pazera, M. and Miller, W. (2007). Air cavities behind claddings: what have we learned? Buildings X Conference.
2. Lissaman, P.B.S. (1983). Low-Reynolds-number airfoils. Annual Review of Fluid Mechanics 15, 223-239.
3. Howey, D.A., Bansal, A. and Holmes, A.S. (2011). Design and performance of a centimetre-scale shrouded wind turbine for energy harvesting. Smart Materials and Structures 20, 085021.
4. Federspiel, C.C. and Chen, J. (2003). Air-powered sensor. Proceedings of IEEE Sensors 2003, 22-25.
5. Newman, B.G. (1986). Multiple actuator-disc theory for wind turbines. Journal of Wind Engineering and Industrial Aerodynamics 24, 215-225.
6. Adema, N. et al. (2026). Wind Energy Science 11, 2037-2051, including a review of measured dual-rotor gains.
7. Li, Q., Öztürk, K., Ragni, D. et al. (2017). Design and experimental validation of swirl recovery vanes for propeller propulsion systems. AIAA 2017-3571.
8. Ohya, Y. and Karasudani, T. (2010). A shrouded wind turbine generating high output power with wind-lens technology. Energies 3(4), 634-649.
9. van Bussel, G.J.W. (2007). The science of making more torque from wind: diffuser experiments and theory revisited. Journal of Physics: Conference Series 75, 012010.
10. Gipe, P. (2017). Kiwi pension fund loses $50 million in Ogin ducted turbine fiasco.
11. Giguère, P. and Selig, M.S. (1998). New airfoils for small horizontal axis wind turbines. Journal of Solar Energy Engineering 120(2), 108-114.
12. Kishore, R.A., Coudron, T. and Priya, S. (2013). Small-scale wind energy portable turbine (SWEPT). Journal of Wind Engineering and Industrial Aerodynamics 116, 21-31.
13. Tamaro, S., Campagnolo, F. and Bottasso, C.L. (2024). On the power and control of a misaligned rotor: beyond the cosine law. Wind Energy Science 9, 1547-1575.
14. Richmond-Navarro, G., Uchida, T. and Calderón-Muñoz, W.R. (2022). Shrouded wind turbine performance in yawed turbulent flow conditions. Wind Engineering 46, 518-528.
15. Mertens, S. (2006). Wind energy in the built environment: concentrator effects of buildings. PhD thesis, TU Delft (after Mertens, 2003, Wind Engineering 27(6)).
16. Encraft (2009). Warwick Wind Trials final report.
17. Pacific Northwest National Laboratory (2025, 2026). Distributed wind market reports and technology data updates.
18. US Department of Energy and NREL (2025). Solar photovoltaic system cost benchmarks, Q1 2025: commercial rooftop $1.95 to $1.98 per W.
19. Lazard (June 2026). Levelized Cost of Energy+, version 19.
20. US Energy Information Administration (2025). Average price of electricity by sector, 2024; Eurostat, Electricity price statistics.
21. Loss, S.R., Will, T. and Marra, P.P. (2013). Estimates of bird collision mortality at wind facilities in the contiguous United States. Biological Conservation 168, 201-209.
22. Loss, S.R., Will, T., Loss, S.S. and Marra, P.P. (2014). Bird-building collisions in the United States. The Condor 116(1), 8-23.
23. Grant, A., Johnstone, C. and Kelly, N. (2008). Urban wind energy conversion: the potential of ducted turbines. Renewable Energy 33(6), 1157-1163. A related idea we will also test: ducts through the parapet driven by the pressure difference between facade and roof.