This is a cool talk: Bill Gross talks about his efforts to tap into solar power. It’s a little bit over-optimistic — how much of the desert Southwest would we have to pave over to collect enough energy for the country? — but the really fun part is where he talks about using unguided evolutionary processes to design solar collectors and heat engines. People who claim that chance and selection can’t produce anything new have never tinkered with genetic algorithms.
Pharyngula
Evolution, development, and random biological ejaculations from a godless liberal
Good! It’s about time someone gave solar technology some serious attention.
I have solar-heated hot water, and I like it well enough for what it is, but it’s obvious after even a cursory study of the solar unit that the technology has lots of room for improvement.
Video is broken.
Good question. We may soon find out. Chris Clarke has been thinking about this one for a while, and at the risk of moderation-delat, here is a sampling of some of his posts on the subject of Big Solar vs. desert ecosystems:
http://faultline.org/index.php/site/item/is_a_fish_more_important_than_a_tortoise/
http://faultline.org/index.php/site/item/compare_and_contrast/
http://faultline.org/index.php/site/item/could_green_kill_the_desert/
http://faultline.org/index.php/site/item/solar_rush/
I am curious, what exactly is the way to use evolutionary processes to design solar collectors and heat engines?
“[H]ow much of the desert Southwest would we have to pave over to collect enough energy for the country?”
A square about 230 miles wide on each side to cover all energy (not just electricity) demand in the US. Pretty dinky. The problem isn’t energy production via photovoltaics….The problem is upfront system cost, transmission, and convincing governments that fossil fuel-driven energy companies aren’t sacred cows.
As an exercise for class, I had students work out a simplified approximation how large a surface would be needed for a photovoltaic solar collector sufficient to provide all of the US’s energy needs.
In 2004 the US was generating at about 1667 GW.
In the lower 48, sunlight arrives at intensities ranging from about 19-56 Wm^-2.
Assuming a 50 Wm^-2 (good sunny day in the southwest), and an unrealistic 100% efficiency for the cells (in practice it is about 15%), that’s 50 MW per km^-2. We would need approximately 33,340 km^-2 surface area to collect the 1667 GW needed. That’s a tad more than a tenth of the surface area of Arizona.
Of course, 222,266 km^-2 is needed at current efficiencies of solar cells, or about 75% of the surface area of AZ.
And, of course, for these to work at highest efficiency, all plant life underneath will die (since they don’t get their photons).
That depends whether you’re expecting solar power to be the sole source of energy, or just a significant part of it. I’d suggest that the former is not a good idea, and the latter reasonable.
The answer, by the way, is in the region of several hundred square miles. Right now, it’s hardly a meaningful question, since we don’t have the capacity to manufacture that volume of any kind of solar collectors, but we should be thinking about whether it’s a good tradeoff or not for the time that that choice becomes available. Personally, given the impact of coal burning (which is where most of the US electricity supply comes from) I think it would be a good trade…
I think a better question is how much of the midwest would we have to pave over, and is there any way to increase this number?
PZ makes a very good point with regard to “paving over the Southwest”. However we must remember that there are, literally millions of square feet of potential solar collecting space already developed, they are called roofs, and every human structure has one. The next step is a photovoltaic system that can be produced in a “green” manner at a reasonable price. >1.00/kilowatt seems to be the grail. Difficult but not necessarily impossible. The power pimps are all about centralization, there are other options.
The link didn’t work for me either. You can download/view the released TED videos from miro:
https://www.miroguide.com/feeds/2014
50 watts per square meter is not at all a good figure for a nice sunny day in the Southwest. At midday, you’re looking at more like a kilowatt per square meter.
Personally, I think concentrating solar power (CSP) using either a trough array or, better yet, a power tower configuration is the way to go, not photovoltaics. First off, CSP is already >30% efficient in total conversion, plus with power tower configurations it’s possible to store absorbed solar energy as molten salt in insulated tanks and generate power off this heat on demand, allowing for dispatchable power 24 hours a day. Th cost per installed watt capacity is lower than PV, too.
Holtz: all plant life underneath will die (since they don’t get their photons).
So, why couldn’t the power company put solar cells on the roof of every building in Phoenix and Tucson and Los Vegas and Salt Lake City and Denver? No plants die, the buildings will be cooler, and the electrical wiring infrastructure is already there. Just need to change all the electric meters to bidirectional to feed power back to the power company when the sun shines.
This would not supply 100% of all our power needs, but would replace a significant portion of coal power for electricity.
Yes I know there are lots of technical problems, but the argument that desert ecosystems must die for solar power plants is simply wrong. Just put the solar panels where we have already destroyed the desert ecosystem.
WOW! PZ we are cosmicaly connected by transplanic-aquarian–psi-tendril-quasi-elementals. I was watching this talk yesterday 20:00 GMT-03:00 at home in Rio! There’s also a nice Dawkins TED, and a cool one by Juan Enriquez, a bit over the board perhaps but still cool:
http://www.ted.com/talks/juan_enriquez_shares_mindboggling_new_science.html
I agree with Marcus that we need a more decentralized solution, even if the solar potential is comparatively lower. The same goes for wind- the Dakotas and Texas have more than enough wind potential to power American homes, but this would mean laying out millions of miles of transmission lines that we dont have.
Before converting wilderness to an industrial use, they ought to exhaust other options. Why not lease space on the tops of houses, businesses, parking garages, etc. to the energy companies before converting desert wilderness?
Thomas @ #6,
There probably aren’t a lot of photosynthetic organisms under most of our currently existing buildings and paved over parking lots.
Perhaps the entire concept of centralized power production and distribution systems is what needs to be reexamined.
Link
I seem to recall some calculation that covering an area of the Sahara the size of Switzerland with solar panels would supply all of Europe’s energy needs. That is miniscule compared to the size of the Sahara, much of which is pretty much devoid of any life. of course, the upfront cost would be enormous, but then again future energy zillionaires might well come from Libya or Algeria….
Umilik @ 17,
I wonder if astrobiologists finding as much life on some distant planet as there is currently in the Sahara would be disappointed?
Charlie the banned Bozo is back! Bye-bye Charlie.
You’re overlooking the information that is transmitted to the function by the environment, namely, whether the function is successful — whether it causes an increase in its ability to replicate itself.
Rebecca C., I think you’ve got your numbers wrong. When I ran the calculations, I found that the area you gave (which is about the size of Conneticut) would be required just to replace the energy made with coal in the US, that is, about half of our energy production.
Couple that with a switch from fossil fueled vehicles to electric vehicles running off the grid, and you’re looking at twice that area.
If you’re going to put it all in the southwest, then you’ll need more area to counteract inefficiencies in the electrical grid.
“The problem is … convincing governments that fossil fuel-driven energy companies aren’t sacred cows.”
So happy to hear someone bring up what I have long suspected is the real problem – convincing the government to fund conversion to green-power at the expense of some of America’s largest companies (and therefore, the source of elected politician’s campaign financing.) That seems like an impossible task. Fixing our healthcare system faces the same hurdle.
Well, your first statement is wrong so the rest of your BS is lying gibberish.
Genetic algorithms and biochemical equivalents such as error prone PCR + selection do in fact, produce new information. Since the simple facts of reality contradict your statement, it is wrong.
Try to explain something easier. How about how War is Peace, Freedom is Slavery, and Bush was a good president.
I suppose that Canada would become dependent on US sunshine, since we don’t get nearly enough of it up here. Solar is a fine solution for a country with big, warm, sunny areas, but less useful north of the 49th parallel.
What we need are some kind of funky snow-and-slush powered generators…
@www10.ch.org
Such an algorithm would have to devise a way to encode the surface structure of a solar cell into a “genome”, in the easiest way encoded by a bit-string, i.e. zeros and ones, say 110110.
The algorithm then needs a so-called fitness function. Given a genome, it calculates how efficient the resulting solar cell would be.
You also need to implement the way procreation and gene transfer work, that is: How do the genomes of two individuals form a new genome? One, very simplistic, possibility is to take the left half from one genome and concatenate it with the right half of the other genome, so 110110 and 001001 produce 110001.
You also have to think about mutation. This usually needs a bit of tweaking. Too much mutation means you miss the optimal solution, too little mutation means you might get stuck in local optima where small changes make things worse, but a big change might be much better.
Then you start with a random population of a bunch of individuals. You calculate their fitness and let them procreate according to how fit they are.
The big difference between genetic algorithms and real evolution is that in genetic algorithms we have a single specific function we want to optimize: There is a goal. Therefore, we need to put in a bit of extra information into the algorithm to achieve this goal.
Evolution does not have one single goal, there is not “the single best species”. As long as you manage to live long enough to make babies, it’s fine with evolution.
Noooo! Leave my beautiful Southwest home alone! We’re already expanding too far for our own good!
“You’re overlooking the information that is transmitted to the function by the environment, namely, whether the function is successful — whether it causes an increase in its ability to replicate itself.”
No information is “transmitted to the function” (selection) by the environment.
Selection is nothing more than a sorting function. It acts like a sieve on pre-existing variation.
It follows the same rules that all functions follow. It cannot create anything new.
Unfortunately, no chance-function combination is ever going to produce the information required to produce and maintain living systems.
Uh huh. See how stupid Charlie & the Creationists are.
But this program of using unguided evolutionary processes to design solar collectors and heat engines could play into the Creationists propaganda hands, because there’ll only be change within a species – there won’t be any solar panels evolving into wind turbines. I think they could be stupid enough to use that as evidence that one species cannot evolve into another, completely missing the point, of course.
The bigger question here is how much water will be needed to keep all those panels clean and working at tip-top efficiency? Put another way, given the current stand-off between the states/provinces surrounding the great lakes, and the desert southwest, how long will it be before a deal is struck… southwest gets water from the lakes, in return for cheap electricity to the northeast?
Same thing will happen with Yucca mountain which just got delayed by Obama – do you really think Nevada wanted that thing in their back yard? Hell no! They want the political power that comes with holding the rest of the country to ransom over nuclear waste disposal… “you wanna dump uranium in our back yard, then give us some of your lovely clean water”.
The power (excuse the pun) that the desert southwest will obtain by implementing massive solar, will permit them to pull all sorts of tricks with the rest of the country in order to get the one thing they don’t have – water. Expect to see the great lakes sucked dry by 2050.
Marshall Nelson is banned troll Charlie Wagner. Don’t feed the troll, as PZ will remove his posts as soon as he gets a chance. Bye-Bye Charlie.
I was with you all the way up to this sentence, which is simply wrong.
Algorithms can indeed not generate information, they only destroy it. “Chance” on the other hand, for which I read “stochastic process” most certainly can and does. Go read an introductory text on information theory.
I just watched the video — it is from 2003. Although interesting, if you go find Energy Innovations on the web, their current product is PV — not at all the thing he was pushing in 2003.
The thought of putting one huge (or a few slightly less huge) power farms is terrible from a terrorism/vulnerability point of view. How much of a bomb would it take to shatter a few square miles of glass?
Marshall @18 — algorithms can’t produce information? OK, imagine a needle in a haystack, literally. Pick any algorithm you want: recursive subdivision, random probes, whatever. Eventually the chosen algorithm will find the needle. You now know something you didn’t before, namely where the needle is. That is also called “information.”
And by the way, Charlie, it was Admiral Nelson.
Well, I’m not surprised.
And I agree with those who suggest putting solar panel roofs on all houses/apartments/buildings. If I ever buy a house, I’ll be doing that.
What is it with energy and people and the PR and the thinking that what we hear most about are the big MEGA type projects ideas?
One of the weaknesses of our system of distribution today is its inefficiency. The lose in transmission is I seem to remember 30% or more and the longer the lines the more lose. Roofs are by design deserts and they are close to the end users but they are not owned by the Big Utilities! We seem to be tending toward supporting and growing the business of energy more than generating and using energy. Unlike biological systems it is possible to abandon man made systems when they become no longer practical or workable as conditions change and redesign new systems from the ground up.
Actually, the collectors would act as rain concentrators. By separating dry and wet stretches of ground and shading soil the biosystem would very likely benefit.
Desert ecology is a bit strange that way, but I love it.
Please, nice PZ – can you say a few words sometime on the subject of Craig Venter’s efforts to design new fuel-producing organisms?
His TED talk was a major dazzler, even discounting his history as a self-publicist, and his vested financial interest in rosy scenarios. The real, existing stuff he showed looks like a clear path to the only kind of nanotechnology that is already field tested over billions of years.
It’s not going to be just the Southwest, certainly, since running all of that power across the country wouldn’t be too efficient.
Maybe what matters is that in most places there are some quite marginal lands–and of course the roofs. Electricity from coal takes up quite a lot of land as well, once you figure in mining, but you have to mine coal where it is, so a lot of fine environments are wrecked. Solar can go into strip-mined areas, perhaps some right-of-ways, and into environmentally degraded territories.
We’ll have to ask how much area we want covered with solar cells (roofs help, but hardly could supply America’s electricity by themselves), but it is unlikely that initial production needs to degrade much environmentally-sensitive ground.
And even better, we get to do what life could not, leap over many selectional valleys. That’s what idiot-boy Charlie never addresses, while he drones on with the same useless rot as always.
He must have been one of the worst teachers ever. You rarely meet with such clueless monotony as Charlie spews. You get as much stupidity with many, but rarely are they the droning monotonous long-winded bores that Charlie’s mindlessness has caused him to be.
Glen D
http://tinyurl.com/6mb592
Thanks, Desert Rat. My point being that there is a lot of juice out there and whatever works. I’m not surprised that someone knows more about it than I do. Although the PV paint sounds cool.
I believe the new buzz in energy research is “personalized energy.” Dan Nocera of MIT gives pretty good talks on the subject, even if he oversales it somewhat. But a major movement is underway to perform artificial photosynthesis (which will have to be more efficient and more stable than plant photosynthesis).
Marcus & others,
You immediately picked up on what was the next question on that “The Roof: America’s Untapped Resource”)
Furthermore, as many have noted, solar power can be used for things other than direct electrical generation; for example, heating water.
Another generally unexplored aspect of this situation is the weather and climate implications of changes in albedo on this scale. (And it isn’t like blacktops and roofs aren’t albedo changers themselves.)
Posted by: Virgil
“The bigger question here is how much water will be needed to keep all those panels clean and working at tip-top efficiency? Put another way, given the current stand-off between the states/provinces surrounding the great lakes, and the desert southwest, how long will it be before a deal is struck… southwest gets water from the lakes, in return for cheap electricity to the northeast?”
I’m willingto bet it would be less then is required to process fuel from the tar sands in Alberta, and even if it is more, at least it would be reclaimable..not just end up as toxic lakes like we have now
The problem with PV cells is that it is deadly inefficient, enormously costly, environmentally hazardous to produce, and the energy produced is difficult to store. These all need to be addressed, but the first and last problem are the most prominent.
Why would you assume that cleaning has to involve a lot of water? The predominant problem in a sandy area will be sand fines. Tip the panels over and vibrate them – a bit like camera sensor cleaning devices. Blow compressed air over them. Use static to de-cling them. Employ ex-bankers to wipe them. Breed geckos with hairy underbellies to brush them clean. Use some imagination.
Agreed that independent generating systems cannot not answer America’s needs completely, however, if most (or even many)households could put the energy they required for normal operations (plus maybe even charge a plug-in electric car for commuting)then we could go along way to reducing extractive technologies.
“”Chance” on the other hand, for which I read “stochastic process” most certainly can and does. Go read an introductory text on information theory.”
You read wrong.
Stochastic processes are still random. Only the output is indeterminate. (like the stock market)
Perhaps I erred in using the word “information”, which lacks intuitive clarity. I’m not talking about Shannon information.
I should probably clarify… I don’t think rooftop PV is a bad idea at all. Like others have pointed out, that’s already ecologically destroyed area, so you might as well get some use out of it. Right now, though, the cost of PV versus CSP is much higher per KWh produced, so it would be a very expensive way to go. Plus, you either have to grid condition the power generated, and use a grid-storage model (which will still need some storage nodes to operate) which will cost you about 10% of your total power, or you have to store it locally in batteries and have DC appliances in your house. I’ve helped friends build off-grid setups like that, and I for one don’t really want to have a ton of lead-acid batteries in a closet which need swapping out and recycling on a regular basis. It makes sense for remote installations, but if every household everywhere had a ton of lead in it,and it was being shipped around and recycled, there would be some leaks, some contamination, and we’d see a rise in lead exposure. Of course, there are other methods, like compressed air storage, or vanadium flow cell battery storage, or even flywheel storage, but most of these don’t make sense at a household scale, and barely make sense at an industrial scale. On an indsutrial scale, thermal storage is by far the cheapest and easiest, but it doesn’t work well with PV, since you’d be going from sun to electricity (~16% efficient, depending on cells) to heat (100% efficient), and then heat back to electricity (~35% efficient). With CSP, you go sun to heat (~90% efficient) and heat to electricity (~35% efficient), and with a well designed storage system, you can retain 90% or more of the heat you originally stockpiled and dispatch it when people use the most electricity, which is around the time that the sun goes down. So, CSP handily beats PV for utility scale applications. However, PV is pretty good for locally offsetting the demand of things that run during the heat of the day, like air conditioning systems and office equipment, as well as for remote applications. So, if a solar generated kilowatt hour can be economically harvested and used, it should be, whether the method is PV or CSP. CSP is a good fit to displace large power plants, and PV works to help offset spike in local usage, like air conditioning, creating a hybrid grid that uses both centralized large scale power plants and delocalized small scale generation to achieve maximum efficiency.
Also, the loss due to transmission averages about 7%, not 30%, so it’s not too bad. The bigger issue with long transmission lines is the cost of construction and the environmental damage of installing them. There is some good work being done with experimental superconducting lines, which would bring the distribution losses to zero, but for now the amount of extra cost in constructing them makes it untenable.
Yes, Charlie’s always talking about magic and his gaping vacuum of knowledge, not about anything scientific.
Heads up.
Glen D
http://tinyurl.com/6mb592
Boggle. This is not even wrong. Before lecturing on information theory, it might be a real, real good idea to at least take a class or two on finite mathematics.
You’d still be clueless, but you might actually understand the insult when someone tells you how clueless you really are.
“People who claim that chance and selection can’t produce anything new have never tinkered with genetic algorithms.”
The problem is that algorithms (functions) are incapable of producing
information. And that’s really what the problem is. Where does the
information come from? In most programs, the
programmer sets the initial conditions and the parameters for selection.
The unfolding of the algorithm is guided by intelligent input.
Darwinian evolution depends on chance (mutations) and functions
(algorithms or natural laws). Chance cannot generate information and
neither can algorithms. Functions transmit already present information.
You get out of an algorithm exactly what you put into it, nothing more.
If “improvement” is seen in the outcome, as you suggest, it’s because a
higher intelligence (the programmer) set it up so that it would.
The notion that functions can sift chance and produce information is
wide-spread among evolutionary biologists and forms the underpinnings of
most modern evolutionary theory. The Darwinian mechanism of mutation and
natural selection is a chance-function combination, in which the
variability of the organism provides the chance component, and selection
pressure from the environment provides the function component.
Unfortunately, no chance-function combination is ever going to produce
the information required to produce and maintain living systems.
Desert Rat:
Thomas was averaging over a 24-hour day, which is the correct thing to do.
Oh, and as far as water usage to keep solar collection systems clean… there’s been some very good work with nanotitanate photocatalytic coatings for glass which can be applied very cheaply and greatly reduce the need for cleaning. Right now, they’re a little fragile, but I have seen some experimental work done by a colleague of mine that uses molecular self-assembly of a sol-gel material to create a nanoporous glass coating which can be applied in a spray or dip-coating method, which then allows the anatase photocatalytic material to be deposited into the nanopores. the result is a film which is far more abrasion resistant, since the bulk of the top surface is glass and destructive sand or grit generally can’t get into the nanopores, as well as enhancement of the photocatalytic effect due to better carrier separation as a result of the small isolated patches of anatase.
So, yeah, solar presents many challenges, but we’ve got solutions ready to go. The biggest challenge is really in convincing people that we need to look at the total long range cost of the sources of power we’re using, and that when assessed this way, the initial up front costs of rolling solar generation out on a massive scale are worth it.
Did he remember to make it costly for the algorithm to become Skynet?
marilove:
My youngest brother did exactly that on his house in the Bay area. The local power company buys his excess power, albeit at a pathetic rate. The payback period is about equal to the expected life of the system, so it doesn’t make much sense financially. Knowing that beforehand, he went ahead anyway, figuring that Doing the Right Thing is worth something.
>Thomas was averaging over a 24-hour day, which is the correct thing to do.
Even if you average over a 24 hour period, and around a full year to factor in seasonal variation, the figure is about 250 watts per square meter, not 50.
Why stop there? We have an entire country full of roofs….
I’m sure Marshall’s posts will be removed as fast as he keeps putting them up, but I feel duty bound to correct his ill-informed ideas anyway.
Basically, you have a “fitness” function which takes a “DNA sequence” as its input and gives an output (vector or scalar, depending on how complex you want it to be) which determines, based on the input, how well-adapted the “organism” with that “DNA sequence” is to its environment. You have a “mating” function which takes as its input two “DNA sequences” and gives as its output a third “DNA sequence”, which is based on the two input sequences being combined somehow with a little random variation just to make it more interesting. Your “DNA sequence” is a vector which represents the various properties under consideration.
Then you create a whole bunch of random “DNA sequences”, run each one through your “fitness” function and line them up in order. You discard the least-fit ones, and “breed” from the remaining ones by running randomly-selected pairs through your “mating” function. (Another way to do it is to pick three “organisms” at random, eliminate the least fit and replace it with the offspring of the stronger two. This way, the population size remains constant, but there is no clear boundary between generations. Convention in this case dictates that a new generation is declared after the number of matings = the population size.)
The point is, doing it this way comes up with a result much sooner than the “brute force” method of running every possible “DNA sequence” through the “fitness” function (which, depending upon the dimensionality of the “DNA” vector, may be computationally unfeasible anyway).
Just spitballing here, but doesn’t anyone know of any research into integrating solar energy collection/power generation into paving materials? Even if it were very low efficiency… there’s a fuck of a lot of parking lots in this country, and almost by definition, they’re all mostly unshaded (unlike lots of residential roofs in my part of the country). And just try walking barefoot on one some summer afternoon!
Jus’ wondrin’…
I also wonder whether or not it’s reasonable to use current electrical consumption figures as a value for how much power we need to produce. Just imagine, for example, if LED light bulbs were able to be profitably sold at the same price as incandescents. The amount of power needed for lighting would drop by what, 90%? That’s a lot of juice.
Thomas Holtz:
You’re off in your calculation of the necessary area. I’m not sure which of your numbers are incorrect, but other people who do the calculation (myself included) get something more like a 10,000 square miles, or a square 100 miles on a side, using presently demonstrated conversion efficiencies of about 10 to 15% for PV.
Ah — wait, someone else caught an error for you.
Also: I put a PV system on my roof in the Bay Area. With the state-funded rebate of about 30% of the installation cost, my payback time is roughly 10-12 years. I plan on living in my house longer than that. And the system should be good for longer than that, too. And the power buy-back rate isn’t pathetic at all. If you’ve got a payback time substantially longer than that, you’re doing something pretty wrong. It is important to get the size of your installation correct, and the financial shenanigans would constipate Einstein, but it can be a net financial gain for most people, especially if you own an air conditioner. You listening, Sacramentans?
There was a series at the beginning of this year called ” Ecopolis”on the Science Channel. It featured among many things, a gentleman in Australia using 112 mirror panels to concentrate the sun rays onto a single cell, thereby greatly increasing the efficiency of solar power.
I have one friend who likes the idea of wind power, another for tidal, and I would like to see all of it harnessed.
But solar power has those most immediate results and will most likely be pursued more than any other energy alternative.
Desert Rat:
Possibly, with a two-axis tracking concentrator. Without tracking, the number is much smaller. Most existing PV systems don’t track.
It’ll take a combination of energy solutions. Wind is highly undeveloped in this country. Our energy grid must be redesigned as well. There are choke points in the grid which make some of these ideas difficult to implement.
Enjoy.
Which is supported by the always-reliable Proof By Confident Assertion.
Let us know when you actually learn what the questions mean, Champ.
The first thing I thought of when I read ‘evolving solar collectors’ was the sunflowers on Larry Nivens Ringworld.
I’m confused.
Americans, in general, seem to be unreasonably supportive of most ‘alternatives’ (alternative medicine, alternative schooling, alternative work practices, …)
Why are they generally so against alternative energy? Is it simply NIMBY?
on another point – more people would invest immediately in local PV/Solar (or heat plants, whatever) if their were better incentives to do so. The upfront cost is huge (for individuals) but tiny in comparison to overall energy infrastructure costs.
One challenge (as mentioned above) is the cost of converting/conditioning DC to grid compliant AC. For an individual, that is a huge component of the install cost. For a small community, that cost can be shared by all!
Neighborhood power plants: fed by DC from PV’s, heat pumps, whatever from every roof & vacant lot. Feeding back AC on the existing infrastructure.
But that would be socialism!
I propose the following analogy to counter Charlie’s #50 delusions:
1. Suppose a crazy man posts 100 rambling incoherent statements on a blog. These statements are our “noise”.
2. The audience points out that 99 of those statements are nonsense, but that one is interesting. The audience are our “functions”.
3. The 1 interesting statement has information in it, but where did the information come from? the crazy man? not really. The audience? Bingo! By selecting the signal from the noise, the audience added information.
Both in Firefox and IE
@Marshall Nelson #50
You do not know what you are talking about. It may behoove you to read up on genetic algorithms (there’s a very accessible description and demonstration of them in Dawkins’s The Blind Watchmaker) and their capabilities before speaking about them. Had you, at any point in your life, seen the output of a genetic algorithm, and recognized it as such, you would quite easily understand that a process consisting of random input plus a method of selection (like, say, GA’s, or biological evolution) generates all sorts of useful information.
Algorithms are perfectly capable of generating information. They do this by of generating randomized input data and using fitness functions (algorithms are NOT functions, by the way) to determine whether any of it is useful. Genetic algorithms do just this, only they improve the process by iteratively generating SEMI-random data. That is, at each step it takes the previous step’s best output (as determined by the fitness functions) and outputs a number of slightly mutated (i.e. randomly modified) copies, any of which may be better (more fit) than the input.
What the programmer controls is the amount of variability at each iterations, the total number of iterations, and the particular sub-algorithm which dictates HOW variation occurs. This last part is important, since it depends on the application – for example, the variation may be a random re-ordering of the input, changing data, adding data, deleting data, etc., or any combination of those things.
Hanes, here are my calculations:
The US consumes about 100 quads of energy (not just electricity or coal, and that includes losses via transmission) per year, which is 2.93X10^12 kWh/yr. Most of Arizona gets about 6 kWh/m2/day of insolation (that’s an annual average that takes nighttime, wintertime, and atmospheric attenuation into account). Assuming a low estimate of 12% efficiency for basic PVs, an array will yield 262.8 kWh/m2/yr. Thus, we’d need about 1.12X10^11 m2 (or a square about 207 miles on each side) of PV to supply the whole US’s energy needs. My calculations actually show an area LOWER than what I quoted above, which I just pulled off the internet in a hurry. (Thanks for keeping me honest with my calculations!)
The good news about using PV to power cars is that internal combustion engines have 12% efficiency, whereas electric cars’ engines have 65% efficiency. So, per unit of energy input, you can get more than 4 times as many miles in an EV than with an ICE.
You’re right that right now the infrastructure just isn’t there. Of course it’s silly, both from a national security and a transmission losses standpoint, to power the US with one 43,062-square-mile array in Arizona. And it’s not going to be easy to switch all our fossil fuel-powered machinery (cars, natural gas stoves, oil-fueled heaters) to electricity-powered machinery overnight. Obviously the idea is to have multiple solar farms across the country, and these new electricity-based technologies would be phased in over several years.
P.S. Southern Minnesota gets about 4.5 kWh/m2/day, while Seattle gets about 3.5, Denver gets about 5.5, and Cleveland gets 4.0. Clearly, more PV area will be needed in the midwest and northeast to harness as much energy from Earth’s Yellow Sun.
I seem to remember the inventor of the super-soaker designing a potentially high-efficiency cell, at the cost that it uses hydrogen internally (which has a limited lifespan, as the hydrogen will “tunnel” out, where we might consider harvesting).
Double P.S. The insolation data I used are for static (non-tracking) PVs. Fancy-pants PVs that follow the sun will get even better output per area.
Generating power is only one facet of the problem; transporting and storing it is another. If this development pans out, those problems may be considerably reduced. With a fast-charging, efficient battery, power from the American Southwest could be stored efficiently and shipped safely. The fun thing here is that it doesn’t matter how you generate the power, whether it’s solar (U.S. Southwest, roofs anywhere), wind (U.S. Midwest and Canada), hydro (rivers), tidal (coastline), heat exchange from the highway (U.S. interstate), or a stationary bicycle (any given health club). Any little bit could be captured, saved, and used.
Hmm. New generations of cells that are in development, but still too expensive, are like 20-25% efficient. Second, a **lot** of the power usage and waste heat in homes right now is caused by conversion, i.e. stepping down 120v to 3-12v, while panels produce 12v normally. You could cut probably 20-30% of your power usage if you could just plug your fracking computers, TVs, stereos, etc., into something that used the same voltage range as its own components (and this includes all the new light bulbs, which need high frequencies, but still use the same low voltage ranges). Fact is, the “grid” is only really necessary for things like stoves, microwaves, very high end power tools, and the like. Most of the stuff you own doesn’t need 120v power, its actually built to waste energy stepping it down from 120v to 3-12V, before doing anything with it.
And, that is part of the problem. Its a chicken or egg problem. Right now we have people making 120v chickens, because the egg is actually some huge 120v power distribution system some place, and *except for* a few devices like laptops, cell phones, or the like, which have a big “power block” they plug into, which does the stepping down for you, most of them have no way to “directly” plug into something native to their own requirements. And they will continue to now have that, so long as we rely on “dumb” power grids, that can’t even manage to track their own failing components (until they fail), balance loads, tell you when peek usage is, or reliably “accept” power from systems that are not part of the main grid already. We need to smarten the grid, take houses off the “you must run 120v for everything” idiotic thinking, then hopefully someone will start building TVs and other things that can “us” those, “Oh, cool, its just my range” personal home power systems, instead of losing 1/5th of all the power to heating the room from the step down systems transformers in all of them. Oh, and.. your equipment would probably get hot a lot less often that way too. ;) lol
…”how much of the desert Southwest would we have to pave over to collect enough energy for the country?”
As a wild-assed guess, I’m thinking completely covering Oklahoma with solar panels should be adequate.
When can we start construction?
I’ve always liked the idea of solar collectors in orbit, concentrating sunlight to Stirling Engine generators. Beam the energy down to a antenna farm in the form of (non-ionizing) microwave energy.
Not many others seem to like that idea.
Doesn’t solar thermal have a much higher efficiency, and the added benefit of being able to store heat for generation after the sun goes down?
“3. The 1 interesting statement has information in it, but where did the information come from? the crazy man? not really. The audience? Bingo! By selecting the signal from the noise, the audience added information.”
Ah…a “signal to noise” guy!
Do you know any radio operators? There are not many left in the world,
but I am one. We have an expression that we use when there’s a lot of
noise on a channel, perhaps static or interference. We say the signal is
“down in the mud”. It’s very difficult to pull a readable signal out of
a high signal/noise ratio. It is the same in nature. These variations
that occur must be rather large and important to be noticed. Small
variations, such as those proposed by Darwin are below the noise level
and get lost “down in the mud”. A real-life illustration is that an
antelope being chased by a lion may have a small variation that allows
him to run just a little faster and he would be better fit to survive
such a challenge and pass these traits on to his offspring, and they
would accumulate over time. The real truth is that this kind of scenario
is a fairy story, since the lion fixates on one prey to the exculsion of
all others and chases him till he drops. Speed is irrelevant in this
context. In addition, this small variation will get lost in the noise of
day to day existence and whether or not the antelope gets caught is much
more contingent on whether or not he stubles and falls, not on his
absolute speed.
Don’t tempt an old Texas boy: My kneejerk answer was “not soon enough!” ;^)
Desert Rat:
Apologies. I misread your number as 1250W/m^2, which is close to the peak solar insolation. If that’s what you had written, my comment would make sense. It isn’t, and it doesn’t.
“The point is, doing it this way comes up with a result much sooner than the “brute force” method of running every possible “DNA sequence” through the “fitness” function (which, depending upon the dimensionality of the “DNA” vector, may be computationally unfeasible anyway).”
Ah!…the “cumlative selection gambit!
Dawkins is wrong about cumulative selection. In order for cumulative selection to
work, the outcome must be known in advance or there must be selective value at
every step of the process. There are many programs the purport to demonstrate
cumulative selection. The license plate on my car has 5 letters and 1 number. Use
any of those programs to determine my tag. If you can, I’ll believe in cumulative
selection.
Epikit @ 54,
My brother lives in Germany and rents roof top space for his PV panels and makes a small amount of money by selling the excess power to the local electric company.
Here in Florida, (I know we deserve all the ridicule for our resident religious wackos) however some of the residents of The Sunshine State do occasionally do things right.
Marshall Nelson, the self-confessed crap radio operator, burbled:
Not if you know what the signal should look like. Noise is by definition broadband and if you are only bothered about a narrow frequency band (or even a single spot frequency being keyed on-off) then most of the noise energy lies outside the band of interest. With the right filtering, you can often resolve a clear enough signal.
Except they don’t, because there is some very tight filtering and non-linear amplification going on. The strength of the ultimate effect of a mutation is essentially orthogonal to the measured difference in the DNA.
Wow, speed isn’t the only criterion being selected for — who’da thunk it, eh?
All you’ve demonstrated there is that fitness is a vector (which we already knew anyway). In this case, our antelope has to be both fast enough to outrun at least one of its brothers and sisters (it doesn’t have to outrun the lion; the lion will stop chasing as soon as it’s caught something else), and sure-footed enough not to fall over in the process.
Solar trees — or fake trees that sometimes look like palms (or occasionally, like carports) — are common at shopping malls in California and in Europe. They don’t collect much power, but are cost competitive if you are subject to time of day pricing to account for peak loads. They provide some shade in parking lots. In cloudy areas, they even collect enough light at night to power low watt street lights.
That of course is an empirical claim, and cannot be asserted a priori. Do you have any evidence that this is true for all natural variation? If not, your assertion fails.
Increase in inclusive fitness is the outcome, and it is indeed “known” in advance, in the same way that reduced potential energy is the “outcome” that is “known” in advance for a falling rock.
Not so — if the change is neutral with regard to fitness, it can nonetheless increase in frequency in a population. Such neutral changes can later be useful when the selective pressures are altered.
@78 Marshall Nelson
1. You do know that the majority of the time the lions don’t catch what they chase.
2. Being slightly faster than the rest of the herd is not by definition a genetic advantage unless it is genetic and, well, make the prey get away.
3. Since there are many genetic traits that combine to figure the survivability under the selective pressures a group live under it could be as simple as camoflage > speed, therefore being faster isn’t a primary survival characteristic. Maybe a genetic disposition to being surrounded by others of the herd > camoflage > speed. Therefore the slow bastards that habitually hang out in the center of the herd are fitter than the fast idiots on the edge of the herd.
How you want to define fitness doesn’t matter. Fitness is defined by surviving to pass on your genes.
Yes, I’m well aware that stochastic processes are “still random”. My point stands. Random/stochastic processes do in fact create information.
If you’re not talking about information as understood by, well, everybody who studies and uses information theory (yes, it has applications; it’s not a purely academic discipline) you’re just making stuff up…
jesus h christ. Charlie has had this shit explained to him over and over again FOR YEARS.
don’t be under the misapprehension that anything you say to him here has not been said to him a thousand times before, nor that it will have any effect on him now.
He was banned for a reason.
well, many reasons, actually, including his rather annoying tendency to morph.
Just get lost, Charlie.
Fitness is defined by surviving to pass on your genes.
only in part.
Fitness is really defined by how many offspring you produce relative to the rest of your local population.
surviving to do that is only part of it.
…albeit a necessary part.
:)
The fitness functions derive a selective value at every step. The fitness function is vital to the success of genetic algorithms. It seems as though you are willfully ignoring the actual details of GAs in favor of your own twisted misconception.
What? Cumulative selection is not ESP. It does not attempt to conjure up predetermined information. The whole point of genetic algorithms is to solve a problem that the programmer cannot solve.
Genetic algorithms work. They successfully solve all sorts of problems. We are not discussing mere conjectures and unproven hypotheses. With a little bit of effort, you yourself could create one and see it work, rather than waving your hands around claiming it doesn’t.
Marshall says,
What?!? Are you just making this up as you go?
1 rambling, incoherent statement down, 99 to go. I’m sure we can get information out of this noise if we just look hard enough.
I must admit that having a well-educated,open-minded group of posters explaining things to the stubborn/ignorant/clever IDiots has helped increase my understanding of several aspects of biological constructs in general and natural selection in particular. Thanks.
The license plate on my car has 5 letters and 1 number. Use any of those programs to determine my tag. If you can, I’ll believe in cumulative selection.
This is quite possibly the dumbest thing ever uttered by an ostensibly sentient being.
Marshall,
To find your license plate would require a “surface” of 7 dimensions, 5 letters + 1 number + 1 degree of freedom of where the number is.
There are certain algorithms that will guarantee the answer is found, such as simply testing every combination in a yes/no manner, be it in an ordered fashion or randomly.
But evolution gives us more than a yes/no, it gives the relative fitness. That allows for a much more powerful relationship, distance to the target, to come into play.
With that information, you start with a random “population” of plates. The plates are then judged by how “close” they are to the target value. Then, based upon their closeness to the target value, they are “bred” together, with a small amount of randomness thrown in. The process is then repeated with the descendants.
If it helps, you can think of the information as not being “created” but “found”. In your example, the license plate # is always there, the algorithm just “found” it. Above, the algorithm is responsible for “judging” each plate, in nature, the success of a collection of genes is based on how well it propagates.
This doesn’t necessarily guarantee the exact plate will be found, but that’s one of the beauties of evolution, even inexact solutions can be very good. In nature, perfect isn’t necessary, just slightly better will do fine.
1 rambling, incoherent statement down, 99 to go
to replace: “99 bottles of beer on the wall…”
…at least around these parts.
:)
If it helps, think of the algorithm as rain falling on a 3D surface. It’s randomly falling, but it still manages to find the lowest spot.
I think the main barriers to the idea of putting solar collectors on rooftops currently are efficiency and maintenance.
If we can develop a way to make the PVs really robust, to the point of install and forget, the possibilities become pretty exciting. Rooftops, outside walls of skyscrapers, roads and sidewalks. A city is basically a concrete desert (populated by a monoculture of biped primates and associated parasites) after all.
I’m not sure if your typical city will have the requisite rooftop area to be self-sufficient, though, given current efficiencies.
We need to think beyond just paving over a flat surface for our collectors, though, and start considering deploying them in 3D space, at varying angles and levels. That’s how plants do it, after all. In fact, mimicking the canopy structure of a typical tropical rainforest might not be a bad way to start.
What the guy who’s currently calling himself “Marshall Nelson,” and others of his persuasion, never seems to get (or maybe is deliberately ignoring), is that even random mutation is not completely up to chance. In each generation, the available potential mutations are constrained by the surviving genomes of the prior generation. Certain possibilities will not be available because all potential precursors were eliminated from the prior generation by selection. And each surviving mutation opens a pathway to new possibilities that were not available to the prior generation.
Each generation gets to start higher up on the fitness landscape than the previous, and selection prevents any significant downward movement.
In a genetic algorithm, the fitness function doesn’t usually change from generation to generation. In nature, the fitness function is the environment, and it does change from generation to generation, but as long as the rate of change is slower than the generation time, similar fitness requirements will prevail over several generations, which is all that you need to see directional change.
In those rare circumstances where the rate of environmental change exceeds the generation time, of course, you won’t be able to produce directional change with evolutionary processes. If the definition of fitness changes so quickly that what is fit for generation one ceases to be fit for generation two, then survival will be down to chance alone. We generally see mass extinctions when environmental conditions are that unstable.
amphiox:
Bingo.
No, he can’t. It is already quite thoroughly tested. No creationist can write a genetic algorithm. It works like a kind of magic. They simply can’t even if they try.
“Just spitballing here, but doesn’t anyone know of any research into integrating solar energy collection/power generation into paving materials?”
Posted by: Bill Dauphin
It would be possible to extract heat from a surface such as a parking lot & convert some of it to other forms such as electricity. You might use heat pipes to feed the hot end of a stirling engine. For the cool end, you would need a mass storage of “coolth”, such as water stored in an insulated tank in the daytime while exposed to a clear sky for radiative cooling & evaporative cooling at night.
This would require both a much larger initial cost & higher maintenance costs than simply placing PV roofing over the parking lot. The PV panels would produce more power. And the ground temperature would be reduced by both shading & from the lightweight materials used to build the structure (as they have less thermal mass).
Also, much of the area of most parking lots are shaded by parked vehicles during the sunniest & hottest part of the day, seriously decreasing possible output. PV panels would work at the same efficiency whether the lot is full or empty of cars. And, since PV efficiency is higher with lower temperatures, you could use water to cool the panels & harvest the heat for various purposes, including stirling engine powered air-conditioning for buildings, a present & proven technology for waste heat use.
Win! – Win!
Whatever happened to Black Silicon which was supposed to be a big breakthrough in this area?
Hmm. Actually, with a parking lot, or the like, its really about, “how much efficiency”, Lets say you just lay a lot of pipe under the lot, all tied to a few towers, which have some turbines. The blades in them, if designed like those roof things, can only practically turn one direction, so any heated air in the pipe has to go one way, and one way only. The other end you put some place shaded, which could be nothing more than concrete box, with enough insulation to prevent “it” from heating its own air supply. You get heat all going in one direction, air flow, and, “bam!”, power. Heck, several of the designs being worked on in Arizona for solar power systems use basically “no” fluids at all, but just some variation on the concept I just described.
@103 kagehi
The air in your example is a fluid. Just sayin’.
Carnot set the standards for heat engines a while back, and so far, no one has proved him wrong.
I’ve heard of systems that use pipes in parking lots and sidewalks as pre-heaters for on-demand domestic hot water systems. In the Summer, the water is usually hot enough that the on-demand heater doesn’t even get used. I think the main problem with using it to drive a heat engine of some sort is that the temperature difference is too low. A well designed heat engine can get up to ~60% of Carnot efficiency. The Carnot efficiency is 1-Tc/Th, where Tc is the temperature of your cold reservoir in Kelvin, and Th is your hot side, the closer those two temperatures are together, the crummier the efficiency becomes. So, not only is there less total power to be harvested, but the efficiency with which you can do it becomes poor. As a result, you really want a big temperature difference to use a heat engine. But, heat is useful for warming homes and heating water, so even low level sources of heat, such as parking lots and sidewalks, can be put to use with a well designed system. Rooftop panels for hot water are, of course, more efficient per area, but if you’re going to have a lot of sidewalks and parking lots, you might as well do something with the heat.
One of the main problems, IMHO, with rooftop solar is that it puts serious limits on how dense your architecture can be and still be energy independent. I designed a system a few years back using a new advanced small scale Stirling engine (3 kW peak output) a client of mine was interested in. He was considering investing in the company, and wanted to see what kind of energy independent architecture it could allow for. So, I designed a hypothetical apartment building, with an average of 1500 sq ft apartments, using dual axis tracked dish-engine systems on the roof, with the coolant loop water from the cold side of the engine (“cold” being relative, in normal operation the cold side of the engine can reach 180 F) being used to drive thermal adsorption air conditioners, provide domestic hot water, and provide most of the heating for the building. The rest of the heating, for cold nights in the Winter, came from electric resistance heating. With use of both the reject heat and the electricity produced by the engines, the total efficiency of converting sunlight incident on the roof to useful power of some sort was over 70%, which is staggeringly good. However, the building couldn’t be over three floors tall at a maximum, and it had to be staggered, with a two story wing and a three story wing, to allow the collectors to be staged such that they didn’t shadow each other, if it was to be fully energy independent. And, of course, in a large scale implementation, buildings had to be sited to avoid shadowing each other as well.
Personally I’m against sprawl. It makes transportation energy costs go way up, and it makes it harder to develop mass transit. People and materials have to be moved a greater distance in a given day, commute times go up, and so on. So, sure, you can make low density architecture work with rooftop solar, but you can’t make tall buildings work that way. So, if you’re to achieve high urban density, which is more efficient, then you need to offset it with purpose built solar collector areas, or windfarms, or whatnot, so there’s a certain total area you’ll have to commit to human use if you’re going all solar. Of course, if you build remote purpose built solar power plants, you can put these in areas where they operate the best, without having to have people live in those sorts of extreme desert environments. And, if you build dense urban environments, harvest what energy you can locally to supplement what comes in from large scale solar power plants, you save in both commute energy and commute time, making this the smallest total impact.
Most of American suburbia is low density enough that all power needs could be met by rooftop solar, but you end up with high operations and maintenance costs due to the decentralized installation, and this sort of sprawl based urban planning is ultimately unsupportable because of its dependence on energy intensive personal transportation.
So, it’s the usual case of there’s no such thing as a free lunch. Low density works with local solar, and saves the use of extra land specifically for power generation, but is itself fraught with inherent inefficiencies. High density achieves excellent transportation efficiencies, but cannot power itself without additional land dedicated to power generation.
Why pave over anything? To begin with, its a much better idea to have energy production be decentralized and spread throughout the U.S. than to have it in a single place. Secondly, given the high wind speeds in the Great Plains, and the generally decent wind speeds nationally, any future energy plan that doesn’t include a wind-generation component is deliberately ignoring an important resource. Then there’s the potential benefit of fuel cell home units, which Exxon and the like have had in development for years now. It may take awhile for those to become economical, but having something the size of an a/c unit beside your house providing most of the energy you use is a pretty niffty idea.
Whatever course we take, the approach will be best if it is diverse. Paving over hundreds of square miles of desert so that we can have a single, isolated production facility is just a bad idea all round.
As a resident of the southwestern deserts… there really is a LOT of land out there that has no practical use, except to the indigenous creatures that live there. And in some places, not even that.
I’m sure decentralized generation is awesome, but don’t write off using some of that desert land. It’s not going to be used otherwise.
“using unguided evolutionary processes to design solar collectors and heat engines”
Wow, that sounds like Front-Loaded Evolution. Using evolutionary processes to design inevitable optimum structures. Cool.