A gentleman by the name of David Fisher has been getting some attention (examples: WSJ, New Yorker, Inhabitat) by describing his design for a new building in Dubai. It would be best for the world if bad ideas like these were ignored and forgotten, but without some knowledge of engineering, it’s not obvious that his ideas are bad.

Fisher’s tower is like a shish-kebab on a vertical skewer, where the skewer is an elevator shaft and the food are the apartments. Each apartment can rotate around the elevator shaft. This alone is perhaps impractical, or ugly, or dumb, but not impossible. If you could find a wealthy fool who wanted to build this, you could probably pull it off.

Where Fisher crosses the line into territory that I defend is with his claims about renewable energy. He says that there will be a wind turbine between each apartment, and solar panels on the roof of each apartment. According to his website, the building will “generate electricity for itself as well as other nearby buildings, making it the first skyscraper designed to be self powered.” As Walter says in The Big Lebowski, “OVER THE LINE!”

Before we even look at the available energy closely, we can be certain that it won’t work. One of the central problems of renewable energy is its low power density. According to the ever trusty Vaclav Smil, wind and solar typically yield 1-10 W/m^2; skyscrapers require in excess of 1000 W/m^2, (Energy in Nature and Society, pp. 311, 317). But perhaps Mr. Smil is wrong. Let’s take a closer look.

Judging by the drawings of Fisher’s tower (since removed), it would be about 300 x 50 m. Ignoring the narrowing of the tower as it rises, roughly 20% of the area is devoted to wind turbines. That’s around 15000 x 0.2 = 3000 m^2. (Fisher has described two versions of the tower, one at ~300 m with 60 floors, another at 420 m with 80 floors. Here, I analyze the shorter of the two.)

Fisher claims that the average wind speed in Dubai is 16 km/h, or 4.4 m/s.

Assuming a Rayleigh distribution for the wind speed, the average power available as kinetic energy in the wind is (6/pi) * 0.5 * (density of air) * area * (average velocity)^3.

The density of air is 1.2 kg/m^3.

That’s (6/3.14) * 0.5 * 1.2 * 3000 * (4.4^3) = 290 kW. If the wind turbines were 30% efficient, which would be pretty good for a vertical axis turbine stuck in a building, the yield would be 100 kW.

This ignores the narrowing of the building, the lack of wind near the ground, and obstruction from other buildings.

The building has around 50 m * 50 m * 60 floors = 150000 m^2 of floor space, so the areal power density is about 0.67 W/m^2. Say a room is 5 m in a side, so it has area of 25 m^2. That gives you 17 W per room.

But let’s not leave out the solar power! Fisher claims that 20% of each roof will be exposed to sunlight. On average, then, if photovoltaics yield around 1 W/m^2, we should expect a power density based on floor area of 0.2 W/m^2, which is another 5 W per room, 22 W total. That might be enough to light a single compact fluorescent light bulb in each room.

Oh, and the average temperature in Dubai is 27 C. I guess they can run the air conditioning when all the lights are off.

I should end by saying that I share Mr. Fisher’s enthusiasm for renewable energy. My concern is that his tower of epic fail gives the work that I spend all day on a bad name. We should be building wind turbines and installing solar panels as fast as we can, but we should do it in ways that optimize their performance. Put the solar panels where they will never be shaded by the floor above them, and put the wind turbines on ridgelines where the wind is strongest. Integrating turbines and panels into buildings with the expectation that they will produce energy to spare is moronic.

(And all you energy reporters should be ashamed of yourselves for repeating Fisher’s void claims without any skepticism. That means you, Paul Goldberger and Evelyn Lee!)

We finally released our first product at GreenMountain Engineering. I’ve been working for various different consulting firms (MindTribe, Ideo, and now GreenMountain) over the past few years, and this is the first time I’ve been involved in shipping a product that wasn’t owned by someone else. I feel proud, and I didn’t even do the hard part– most of the development and testing took place in our San Francisco office.

The product is the called the Trac-stat SL1. It’s a ridiculously accurate sensor for measuring how closely your solar tracker is aimed at the sun. I think the spec is 0.02 degrees for the more accurate of the two sensors it contains.

Max and two of his west coast disciples have been testing it at our secret rooftop testing facility in San Francisco for the last few months. (Pretend you don’t recognize the lights of the Giants’ baseball stadium in the background.) The graph below shows the output of the sensor. This was on a tracker of relatively low precision that we have used for a couple of different concentrating solar projects.

Of the many markets that exist for the SL1, the largest is the group of companies that are trying to build concentrating photovoltaic systems. The key point of leverage behind concentrating solar is that if you can gather the same amount of sun with a drastically reduced amount of solar cell, you can win on cost. An unfortunate side effect is that as your target gets smaller, you need to aim ever more precisely at the sun. This means that to be able to evaluate the performance of your concentrating system, you have to know how well you are pointed at the sun. This alone is a serious research project; there is substantial empirical evidence that building a sensor to track the sun takes lots of work, and that doesn’t even get you any actuation. You can buy a pretty good tracker, but you won’t know how good unless you have some sort of diagnostic instrument that can measure your error very precisely.

When I worked in the Stanford Robotics Lab after grad school, a shrewd man told me on one occasion, “Don’t make everything a research project.” (I think I was proposing writing my own TCP/IP stack or something similarly idiotic that would have taken long enough to preclude completion.) My hope is that the concentrating solar companies will not spend the engineering time it would take them to each build an SL1 equivalent.

And finally, did I mention that it also has a sweet command line interface?

(Perhaps I should note that the opinions listed above do not represent those of my employer. Especially not the unreasoning zeal for command line interfaces.)

Historically, I have spent a fair amount of time deriding Nanosolar for claiming that they would build a 430 MW/year solar cell factory by the end of 2007.

On the one hand, they have only 10 days to prove me wrong. I don’t think they will have ramped production up to 430 MW/year. On the other hand, they are shipping a thin film module created through a printing process. That alone is an impressive feat. They win the prize, soaring above the Aonian Mount, while pursuing things unattempted yet in prose or rhyme (well, maybe attempted in rhyme).

Nanosolar is claiming module costs of 1 $/W; the US solar industry is now selling around 5 $/W.

It looks like they have beaten:

• Ascent Solar (thin film, CIGS, probaby through vacuum deposition)
• Konarka (thin film, printed, organic cells)
• Heliovolt (thin film, printed, CIGS cells)
• SoloPower (thin film, CIGS, but electroplated, not printed)
• Miasole (thin film, CIGS, but not printed)
• Solyndra (thin-film CIGS, unknown process, but they’re hiring people with semiconductor process experience)

But have they hit grid parity? Maybe.

First off, $/W is a dumb metric. The W refers to the peak power, so if you buy a panel that peaks at 200 W for $800, you’re paying $4/W. It’s stupid, but it’s easy to measure, and it’s easy for journalists to quote.

A better metric is $/kWh, where kWh is a unit of energy. That’s harder to talk about, because you have to talk about average production over the course of a year, which changes with location, weather, age of the panel, and so on.

The sun that hits the ground peaks around 1000 W/m^2; you might get 20% of that on average, once you factor in night, clouds, and the like.

There are about 10,000 hours in a year, so I’d expect 200 W * 10,000 hours * 10% efficiency = 200 kWh/year for a square meter. In Massachusetts, where I live, that’s worth about $40 ($0.20 per kWh). (Did I get those calculations right?)

Conventional panels of around 10% efficiency are about $1500/m^2 (say, 2 Evergreen 200 W panels for $750 each?). Given panel life of about 20 years and the added expense of installation, an inverter, and maintenance, I think solar is still off from grid parity by a factor of 2-4 in Massachusetts (not counting Nanosolar).

If Nanosolar is telling the truth, they may have just hit grid parity.