PORTLAND, Ore. — IBM Research (Zurich) and Airlight Energy SA (Biasca, Switzerland) aim to boost the output of high-concentration photovoltaic cells by as much as 10-times by using micro-fluidic water cooling with commercial triple-junction photovoltaic cells. The resulting 25kWatt 50-foot-wide dish-based concentrated solar power stations will be constructed in Biasca and Rüschlikon Switzerland under a three-year $2.4 million grant from the Swiss Commission for Technology and Innovation. Also working with IBM and Airlight Energy on the project will be the the Swiss Federal Institute of Technology (ETH, Zurich) and the Interstate University of Applied Sciences (Buchs).
"We are using the same water-cooling technology IBM developed for high-performance computers to attain a 10-fold decrease in the thermal resistance of the photovoltaic cells," said Bruno Michel, manager of advanced thermal packaging at IBM Research. "As a result, we have demonstrated that commercially available triple-junction photovoltaic cells--which cover almost the entire spectrum of solar energy with 80 percent efficiency--can now handle from 2000-to-5000 times concentrated solar power from the sun, compared to 300-to-500 times when air-cooled."
The higher concentration of solar energy is produced by arraying 36 mirrors across the 50-foot dish that tracks the sun as it moves to continually aim their high-energy beams on an array of over a hundred triple-junction photovoltaic cells mounted on the central gantry, each of which generates from 200-to-250 watts. Without IBM's water cooling, the concentrated energy from the solar beam could produce temperatures high enough to vaporize the chips. Instead, IBM's micro-fluidic substrate conducts the heat away from the photovoltaic cells with a hierarchical arrangement of water channels.
High Concentration PhotoVoltaic Thermal (HCPVT) system under development by IBM and Airlight Energy concentrates sun onto hundreds of microfluidic liquid-cooled triple junction photovoltaic cells to provide 25 kilowatts of electrical power.
SOURCE: Rendering by Airlight Energy Click on image to enlarge.
And just as IBM's water-cooled data centers make use of the heat from the water heated by its chips--to heat adjacent buildings--the solar power stations will be adapted to applications that make use of its heated waste water. However, since most of the envisioned installations will be in warm climates, IBM is experimenting with using the warm water in absorption refrigeration systems that substitute for conventional air conditioners, and for thermal water desalination plants.
The final goal of the project is to produce electricity from solar energy the cost of which is on par with coal-fired generators, about 5-to-10 cents per kilo-Watt-hour (kWh). By using inexpensive concrete and pressurized metallic foils, the cost of commercial dishes should drop to about $250 per square meter--three-times lower that existing photovoltaic concentrators today, according to IBM.
What temperatures will the cells operate at? What temperature is the 'waste heat' water?
I wish you the best of luck with this experiment but I'm not holding my breath in anticipation. Five or six years ago, I was touting the same benefits of a HCPVT solution (Menova Energy, now defunct). We too were going to use triple junction cells at ~1000x concentration and use water to keep the cells cool. We found that unless you have a virtually infinite requirement for low grade heat, it was difficult to derive any meaningful (financial) benefit from the waste heat. If you slowed the flow of the water (to get higher grade heat), then you lost efficiency on the PV side; and, given the differing values of electrical vs thermal outputs (i.e., ~12-16¢/kWh vs ~4-5¢/kWh (from NatGas)), it was not good business case...
It sounds like the subtle distinction is that this technology may boost the efficiency of a particular technology by 10x ... but that technology doesn't happen to be the most efficient available technology. It isn't a 10x improvement over the best available.
Here is a video by IBM about their prototype, the facts about which enabled the companies to be so optimistic about their project, and land the $2.4 million to prove they can do it: http://youtu.be/J_zzE8xMdZc
Ok so you are going to start misrepresenting facts in the title?
I ussally like your articles and they are generally acurate. What happened here?
Are you just reprinting what you are given?
You do not get 10x more power for a given amount of solar energy.
80% efficiency # thrown out without the facts?
How much more power is produced for the same concentration level with and without cooling?
What % improvement is created by keeping the solar cells at a lower temperature?
if solar concentration gets as high as stated, how hard is it to focus at this concentration.
if not perfect how much extra losses for ideal constant concentration acroos the whole surface of the triple junction cell?
How does concentration ratio and the resulting cost balance between concentration means and PV collector costs map out in a graph.
How much better ROI if concentration goes up 10X with cooling means compared to aircooled?
'5 to 10 cents per kWh'.
Current panels are on the order of a dollar/W.
In near desert, this produces 1700Wh/year.
To hit 10 cents/kWh - your panel needs to last a little under 6 years.
To hit 5 - 12 years.
(neglecting cost of money).
'cost of concentrated panels could drop to $250/m^2' - the price of existing panels is about half this.
Correct me if I am wrong...
50 feet x 0.305 m = 15.25m diameter
15.25m * pi / 4 = 182m square area
At 25kW peak power at full sunlight (1000W/m sq) that gives us:
25000W / 182m sq = 137W / m sq
Electrical efficiency is a ratio of power out to power in:
n = 137 [W / m sq] / 1000 [W / m sq] = 0.137 = 13.7%
Effective efficiency is 13.7% minus inverters/transformers efficiency.
Hardly 80% or even 30%......
As these concentrators only work correctly in full sunlight and they do cast shadows - they would have to be spaced in order to work. Say we estimate that we could spread them so for one unit of space we would need 3 space units around it to be free. This will lower the area efficiency by a factor of 4 - for every square meter of a space at full sunlight they would generate:
1000 [W / m sq] * 13.7% / 4 = 34W
Please note that concentrated panels work only with the direct sunlight (as opposed by flat panels that harvest energy in proportion to the current insolation).
To build a 1GW power station (still we assume full sunlight for all of the day and night) this would require an area of:
10^9 [W] / 34 [W / m sq] = 29.4 km sq
This would be a 5.4 x 5.4km square. Can we do it?
And remember the Solyndra boondoggle? Solyndra made efficient solar panels into a narrow cylindrical form, which turned out to be inefficient because the sun wasn't hitting the back of the cylinders. (And also Chinese companies entered the market and severely undercut them in price).
I always wondered why they didn't make the cylinders into pipes, then they could do this same trick of making the solar panels hotter and more efficient while also generating power from the hot water by putting them into solar trough farms:
I'm glad IBM is working on these ideas for using more of the sun's power. I hope we eventually get really low cost solar power.
We plan to use triple-junction photovoltaic cells on a micro-channel cooled module which can directly convert more than 30 percent of collected solar radiation into electrical energy and allow for the efficient recovery of an additional 50 percent waste heat.
Here is the full release http://www-03.ibm.com/press/us/en/pressrelease/40912.wss
Blog Doing Math in FPGAs Tom Burke 23 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...