There's been a catch-22 on new solar technology--projects can't go forward because they only have a PowerPoint.
In the scheme of things, this plant is just five megawatts (about enough to power 1,500 homes). How do you scale?
Gross: A 46-megawatt plant takes one quarter mile, so to build a gigawatt in California would take 20 of those--or 5 square miles. We purchased that land for $30 million in cash, all in small quarter-mile plots, all adjacent to transmission, so we've completely eliminated the owning and permitting issue.
(Project development companies like) NRG will buy the project from us (for a planned installation in New Mexico) and own and operate the plant for 20 years.
So it's large scale but done in a distributed way?
Gross: Exactly. The ultimate distributed solar is to put (photovoltaic) solar panels on every rooftop. The ultimate centralized solar is get 2,000 acres of BLM land and then you have a transmission problem because you need to build a gigawatt of transmission. We're in between--we're large-scale utility but we're still distributed--distributed in small enough pieces.
Two years ago we didn't know how prescient it would be, but we looked at the entire transmission grid of California and where there was 46 megawatts of available capacity, we would go and buy a patch of land right next to that. We inverted the problem. Others said, "Let's build where the sun is best because I need to buy 2,000 acres at once to get economies of scale. And then I'll try to lobby for 10 years to get 100 miles of million-dollars-per-mile transmission built."
Now environmental groups of all things are protesting people using BLM land, and they have a point. Solar is great but you don't want to destroy the pristine desert...Our land is already being used for something (such as farming).
Precision tracking for each mirror allows eSolar to write the company's name at its Lancaster, Calif., plant.
(Credit: eSolar)On the technology side, how much more productive or efficient is this solar tower than existing solar trough technology?
Gross: It is a little more productive than solar troughs. Solar troughs run at between 27 percent and 30 percent efficiency, and we are at 34 percent efficiency. But the real thing is we're half the cost. It's not the efficiency that we're much better at, it's the price--that's really the breakthrough. The reason we're so much less cost is that we use hundreds of thousands of small flat mirrors, instead of long, long rows of huge curved mirrors. The troughs use a mirror that is 5 meters wide by 100 meters long. They pay the same price as we do for the mirror--it's the same high-quality Belgian or German suppliers--but that's only 10 percent of the cost.
The main cost of the solar thermal plant besides the mirror is the steel and the actuator (for controlling mirrors)--that's 90 percent of the cost...The steel (is needed) to hold the mirror in shape without distorting, to stay in a perfect parabola. Because we use a one-square-meter mirror, we use half the steel. Imagine if you take a piece of flat glass and put a tripod behind it, it'll stay flat. But you need far more steel to bend glass against its will.
So why haven't other solar companies broken up their mirrored troughs into smaller bits?
Gross: The problem is historically it's been a software control problem to track hundreds of thousands of small things. The benefit of one big row is you only need 20 motors to turn troughs--all pointing at the sun--and software control is trivial. We have 24,000 individual mirrors, all pointing in slightly different directions to point at one spot. We're basically making a dynamic parabola in software where they are making a static parabola in steel.
In the last decade, there's been a 1,000-fold increase in computational power, so now we can put a $2 microprocessor in every mirror and it costs almost nothing--almost one and half percent of the (material) cost. So every mirror that is tracking the sun during the day has its own computer. And the computational power of a microprocessor today is mind-boggling. It's a 16-bit microprocessor with eight I/O ports. It's like an IBM AT (PC) in every mirror--that was a $5,000 computer in 1985. This completely wouldn't be possible without Moore's Law.
It's interesting that you've come from the computer world into solar. Will there be other stories like eSolar to come?
Gross: I definitely think so. eSolar has been grown right in the same building as other Idealab companies with all the benefits of IT they had--it uses all the servers built for Internet companies with all the experience and hardening capabilities. And it even uses many of software developers from prior years that we've hired back from places like Yahoo.
If anything, we're more a software company than a solar company. Of course we're a solar company, but software is 50 people out of the whole company. There are 135 people--100 are in engineering, 35 are out running power plants, so half the (engineering people) are in the software group, which is an amazing percentage for a solar company.
A view of the mirrors on eSolar's Lancaster, California plant.
No comments:
Post a Comment