#1 cleantech news, reviews, & analysis site in the world. Subscribe today. The future is now.

Clean Power

Published on July 22nd, 2014 | by Guest Contributor


The Continuing Exponential Growth Of Global Solar PV Production & Installation

July 22nd, 2014 by  

By Mike Shurtleff

For years, I’ve been following the exponential growth of Global Solar PV production/installation, starting with this article in 2007: “PV production up 50% in 2007,” in which it was stated that solar PV production had “jumped to 3,800 megawatts worldwide in 2007” and was “world’s fastest-growing energy source.” There is a good plot of PV production growth from 1975 to 2007 included. A number of commenters suggested the high rate of growth (48%) seen in the preceding seven years would not be as easy to maintain at higher volumes of production going forward. I was not as sure. I was also not sure they were wrong. I made a resolve to watch this. Now, seven years later, they have not been completely wrong, but pretty dang close to completely wrong.

The Problem with Human Perception of Exponential Growth

In his book “Tipping Point,” Malcolm Gladwell points out human beings do not properly perceive exponential growth. This is the illustrative example he uses:

“Consider, for example, the following puzzle. I give you a large piece of paper, and I ask you to fold it over once, and then take that folded paper and fold it over again, and then again, and again, until you have refolded the original paper 50 times. How tall do you think the final stack is going to be? In answer to that question, most people will fold the sheet in their mind’s eye, and guess that the pile would be as thick as a phone book or, if they’re really courageous, they’ll say that it would be as tall as a refrigerator. But the real answer is that the height of the stack would approximate the distance to the sun. This is an example of what in mathematics is called a geometric progression.”

His point in the book is people do not see exponential growth for what it is. As a result, the “tipping point” in new technologies usually takes them by surprise. His book does not mention the “tipping point” in solar PV production/installation, but the concept certainly applies.

Another illustrative example of exponential growth is found in this article from November 2011: “100GW of demand, and the coming inflection point in the US solar market.”

This article may be the most important one on solar PV on the internet (imho). The author, Richard Keiser, points out that decreasing solar PV costs lead to an exponential increase in demand. A continuing decrease in solar PV cost will reach grid parity over an ever increasing number of areas (i.e., for an exponentially increasing number of customers). He too provides an easier-to-understand illustration of the power of exponential growth:

“Non-linear systems are often difficult to understand. The famous ‘penny game’ is a good example. In this game, a hypothetical person is given one penny (or one euro cent) on the first day, two pennies on the second day, four pennies on the third day, etc., and then asked to guess the total value of the pennies at the end of one month. Very few people guess correctly – US$21 million – or appreciate that 75% of that value is created on the last two days.”

“Like the penny game, the solar industry is characterized by a number of non-linear dynamics.”

The Scientific Evidence

The data for these graphs is taken from different Internet articles. The sources are listed further below.
(This is really one set of data collected over time and plotted over Christmas last year.)

1 Actual Global Installs - Linear Plot

2 Linear Plot - 41 and 48 Curves extended to 2022

3 Actual Global Installs - Log Plot

4 Log Plot - 41 and 48 Curves extended to 2022

The green line is a calculated exponential growth rate of 48% per year, starting with 3.8 GW in 2007 from the original source article mentioned at the beginning. The blue line is a calculated exponential growth rate of 41% per year, starting with 3.8 GW in 2007. The red line is a plot of the data collected from various Internet sources.

The top two graphs are linear plots. The bottom two graphs are exponential plots of the same data. (Note that the calculated 48% and 41% rate projections appear as straight lines when plotted with an exponential vertical axis …as they should.)

The graphs on the left show the actual data collected from the Internet compared to 48% and 41% rates of growth.

The graphs on the right show a projection of 48% and 41% growth rates out to 2022.

Global Solar PV has been growing exponentially, although it has not followed a well-behaved exponential curve. At this point in time, it has been growing faster 41% per year since that original article in 2007. A growth rate of 41% per year is the same as a doubling of production/installation every two years. Like I said, it did slow down from 48%, but not much.

If a growth rate of over 41% continues until 2022, then the world will be producing/installing over 0.5 terawatts of solar PV panels per year and maybe as much as 1.0 terawatt per year. At this rate, solar PV will become THE major source of power throughout the world. Further, when including any additional growth in production/installation, this will happen in a few years, easily within the next decade. Total global power use is less than 20 terawatts. (This is all of the world’s power use, not just electricity.)

Additional Considerations

The demonstrated exponential growth has occurred in spite of two detrimental events:

1. In roughly 2007 to 2009, there was an under-supply of purified bulk silicon. This was not a shortage of the raw material silicon. There is plenty of silica sand. It was a manufacturing shortage that has since been remedied… and then some. Silicon solar PV panel manufactures learned to produce panels using less silicon. Purified bulk silicon producers scaled up their production and have also dramatically reduced their production costs. Lower-cost purified silicon, and PV panels using less of it, have since resulted in much lower-cost silicon PV panels.

2. More recently, starting in roughly 2011, there was a global economic collapse. (We are still feeling the effects.) This had the secondary effect of causing many countries to reduce their solar incentives. (Germany, Italy, and Spain in particular have scaled back their incentives and reduced their solar growth rate dramatically.)

You can see the respective dips in production/installation during the two pull-back periods mentioned. It would be hard to over-emphasize the fact that solar PV has continued to maintain a high overall growth rate right through these two hard hits. It is now widely recognized that solar PV is entering a new phase of high-rate growth, an over-demand market. There should be no mistaking of solar PV as other than an unstoppable disruptive technology. Opposition by fossil fuel companies and some utilities will not be successful.

(Note: I’m lumping production and installation together. Initially production was pretty much the same as installation. More recently, during the period of solar PV over-production, this was not the case and I started to see numbers of global solar PV installation instead of production. I’m using them somewhat interchangeably to represent global solar PV growth.)


For the last 14 years, almost a decade and a half, global solar PV production/installation has grown faster than 41% per year (compound annual growth rate, aka CAGR). This means global solar PV production/installation has been more than doubling every two years. (I’m using predictions for 2014, but we are on track.) Predictions for 2015 already suggest a continued high rate of growth. Most predictions going forward beyond 2015 will continue to be of linear growth, as has been the case in the past. This is because humans have a psychological blind spot with respect to exponential growth. I submit this may not be the case at all. Global solar PV growth will eventually flatten out, but don’t bet on this happening in the near future.

If this high growth rate continues for another 8 years, till 2022, then we will be already well on our way to providing most of the world’s power using solar PV.

The next time you ponder the growth rate of solar PV, consider the plots above…. Consider the continuing drop in cost…. Consider the resulting exponential increase in demand. Think about the multi-gigawatt production plans of some of the big solar PV manufactures. Will it really take 50 years for solar PV to provide most of our power?… Or will those plots above continue, so that it will only take 10 to 20 years? Think about it.

Tags: , , , , , ,

About the Author

is many, many people. We publish a number of guest posts from experts in a large variety of fields. This is our contributor account for those special people. :D

  • Bob_Wallace

    Mike, if you’re still following this –

    How about taking Fig.1 and expand the horizontal axis to 2022? (Just let the installed data stop at 2014).

    That should give people a feel for where we might be in a few years if we keep growing solar at previous rates.

    And, for fun, you might add GMT’s prediction of 55 GW added globally in 2015.


    I’m thinking this should be an annual tradition.

    (Perhaps this was all said earlier, but I’m not wading through 154 comments….;o)

  • Alastair Leith

    Is it okay to republish one of the graphs in this story? the PV deployment curve. I can remake myself but will save me the effort.

    • Bob_Wallace

      Notified Zach. He can probably answer your question if Mike doesn’t see it.

      • Alastair Leith

        thanks for the fast response Bob!

      • Alastair Leith

        so yes it would be great to know the main source for historical data too, is from factories making panels, installers installing panels, govt estimates added up, what?

        • Bob_Wallace

          Here’s a couple of Wiki pages you might wish to read.



          My feel is that totally solar is an inexact science. It seems to me that the most accurate way would be to report the amount shipped. The number of manufacturers is not all that high.

          Get the shipped numbers and do a little work to come up with a ‘shipped but never installed’ percentage that could be used to true up the number a bit.

          • Alastair Leith

            Thanks Bob.

            When Chinese companies are so heavily featured in the total shipped it’s hard to get reliable numbers isn’t it? Corruption is rife in China even compared to USA, as are issues around state backed loans and dumping rulings in US which might make them inclined to be less than honest. Even in OECD countries there’s commercial sensitivities involved in declaring units shipped.

            I’m not sure how CEC calculates the amount of solar installed each year in Australia but I gather it’s derived from the SSTC and LGC (small and large scale RE certificates) issued by the regulator.

          • Bob_Wallace

            My short answer is that I don’t know how accurate the numbers out of China would be.

            My longer answer would get into whether or not China is any better or worse than any other country when it comes to honesty. How about that VW thing? Dishonest Germans. Who’da thunk it?

          • Alastair Leith

            yeah good point, west has just had more practice at PR and Marketing to hide it from public view in a modern democratic setting. On VW I think it’s been a bit of an open secret for years that emissions testing is rigged for all manufacturers. the VW thing just makes it look a lot more mendacious than when the regulator “works with industry” to compromise the testing regime to the general public. Both have bad outcomes.

    • I’m not actually clear who created these. Mike didn’t.

      • Alastair Leith

        plot thinkens, Dickens

    • Bob_Wallace

      I can’t find the source of the graphs with Google.

      How about using the Wiki numbers?


      And, if you don’t mind, post your version back here.

      • Alastair Leith

        Will do.

        • Bob_Wallace

          I was looking at the Wiki global solar numbers. Decided to do a simple bar graph of year on year percentage growth.

          We’re clearly staying in exponential country….

          • Alastair Leith

            if 42% is a two year doubling (global rate of PV deployment growth over last three decades according to many graphs I see online) I estimate 25% is about a 3.5 year doubling. still significantly stronger than linear growth.

          • Bob_Wallace

            Use the “Rule of 72”. Divide the rate into 72 to get the approximate number of years to double. Or divide the number of years to double into 72 to get the needed rate.

            72 / 25% = 2.88 years.

            Linear growth would be 0% change, year on year.

          • Alastair Leith

            hmm. my back of the envelope sheet said >3 years:

          • Alastair Leith

            hmm. my back of the envelope sheet said longer than 3 years:

          • Bob_Wallace

            Yeah. Rule of 72 is a close approximation of doubling time.

            72, while not totally accurate, gives a close approximation and 72 is a number easily divided.

  • nedford

    I found this post by googling for an answer. I really like it. However I see a major problem, based on the conclusions here in conjunction with an article by Elon Musk written in June (the guy who sent it to me didn’t include a link). He says that if we want to have solar provide 40% of global electricity by 2040 we postulate an annual rate of install of 400 GW’s.

    I assume the current technology permits a solution to electric sector carbon which is about 40% wind and 40% PV. We can talk about energy storage and other renewable technologies, but we don’t have those in hand. 40% is also the level at which either wind or PV starts to run into problems if we don’t have a dispatchable renewable or good energy storage.

    (I also know a lot more about efficiency, and assume the efficiency will pay for all the renewables we need, but that’s a different comment).

    The point I want to make is that the investment community will stop funding new PV factories around 20 years prior to what they see as a saturated market.

    So if we just project replacement of existing fossil and nuclear plants with wind and PV and a few odds and ends to fill in the gaps between existing hydro and biomass, we don’t see that stunning growth your charts project around 2021. Wall Street is a whole lot smarter than most utility engineers, and won’t fund a nuke or a coal plant or a gas plant or a wind turbine factory or a PV factory if they don’t see a market for the product, or if they see their investment competing for a limited turf with existing plants.

    The solution that I see is the abundant electricity future. In the U.S. this means that we need about 50% more electricity in 2040 than EIA presently projects. The additional electricity will fuel electric transportation options, and a whole lot of shifting from natural gas space and water heating to electricity or (solar water heating with electric backup).

    I hope we can keep up the 40% growth. I temper this optimism with the fact that wind, solar and efficiency have all averaged closer to 15% over the last two decades. I’m quite sure that the best way to keep the rates high is to give lots of people a reason to believe that future electric consumption will be much larger than we presently expect.

    • Bob_Wallace

      Without going over the numbers you listed –

      Make sure that whoever generated those numbers was allowing for the much greater efficiency of electricity vs. coal and petroleum.

      Sometimes people make the mistake of calculating that if we’re burning X amount of oil for cars then we need to generate the same amount of energy for EVs when EVs are 80+% efficient and ICEVs are about 20% efficient. And coal plants are only 35% or so efficient.

      Overall we should see a large decrease in energy consumed simply because we should quit wasting a terrible amount of energy as thrown off heat. Check the graphic below. More than half of all energy used is wasted.

      Then storage. We have a perfectly fine solution. We’ve been using pump up hydro storage for 100 years. We’ve got plenty of places to build it around the world and it’s reasonably priced. What we’re doing right now is looking for cheaper (or as cheap) and easier to site storage. Storage is going to be best if we can distribute around the grid and not have to build a lot of transmission lines and deal with land acquisitions. The sort of containerized storage being developed now can basically be trucked to a piece of cheap real estate and plugged in.

      Now, solar and wind won’t continue to grow at 40%, whatever. We’re in the acceleration phase (look at the technology uptake graph a couple of comments below). After a little while longer we’ll probably hit the “steady state”, flat portion of the curve. And, then as the market saturates, the rate will slow.

      (I think I addressed your points. If no, get back to me.)

  • Show me 3D ‘graphs’ of what kinds of money buys these panels exponentially.
    Did everyone buy them exponentially?
    What was their tripping point? 500 hrs of solar info finally got through the media to them?
    Was there a wave of retired debt that allowed this or was it legislation/changes to building codes/ or backlash to dirty sand oil production.

  • Vensonata

    As I read through the 105 comments, I keep thinking the question is: Is 100% solar pv possible today. By that I mean complete supply of all energy demands for a typical family home. This includes cars, heat, air con, hot water, and sundry electrical demands. The answer is yes, and a number of such houses are around. But we need to distinguish between on grid net zero energy, which if intelligently done requires about a 10 kwp panel, and a completely self contained off grid battery storage which would require about 15 kw panel plus heat pump, and some seasonal heat storage in most northerly u.s and Canada. The efficiency needs to be there …ie the house needs to be close to passive house standards, but with careful shopping it is possible today, at reasonable cost. Anybody agree? Anybody shocked? By the way, unless you are up to date on all this, don’t give me views that are even 2 years old, they are obsolete.

    • Bob_Wallace

      I’ve lived in places where 100% solar would not be anywhere close to practical.

      In the Central Valley of California, for example, the Tule fog can settle in for days. Buying enough batteries to carry a house through long periods of no sunshine would simply be too expensive.

      Large grids make for cheaper electricity. Especially if the grid is largely in place.

      • Vensonata

        Yes, there are certainly some places that may be impractical, but I am seeing many articles that show it is possible in places as unlikely as Alaska 64degrees north (see how close Thorsten Chlupp is on three houses) one house is 1800 sq feet and is off grid, all electric plus less than two cords wood year. That is the most extreme example of passivhaus standards in the world. In homepower magazine latest issue a house in Mass. Is complete net zero all electric heat without a heat pump! R 100 roof, r 70 walls, 5 kwh pv array on roof. Basically, he is there. Add air source heat pump and he could easily be off grid zero with modest battery. By the way, the Mass. House cost $100 sq ft. I am deeply impressed.

      • eveee

        I have lived off grid on vacation a few times in Canada. They had not set up any practical means except boating in gas and charging batteries. The inland waterways have heavy daily tides, which could be used if there was a catch basin. A small wind turbine at height could work. But living off grid under those circumstances means running the generator only a few hours a day and using electricity sparingly. Nothing like suburban living. It could still be quite livable. One problem was communications with the rest of the world. Satellite fixed that, but phone calls and internet were expensive. Same thing. Limited use. Its perfectly lovely, and you can live without some of those things, but its not realistic to expect everything to be the same off grid in remote locations.

  • Bob_Wallace

    I’m posting these at Mike’s request. Hopefully this is the correct set of figures.

    (Can’t get them corrected in the text at the moment. Zach is kind of busy being a first time father.)

    • mike

      Thanks Bob!

    • eveee

      Nice job. Its just exponential so far. And doubling at the ridiculous rate of 18 months. Thats and astounding rate akin only to the rates of lithographic reduction in Silicon. The reason I would not be keen to bet against it in favor of saturation soon, is that being in the semiconductor biz, and noting that many have predicted the end of Moores law and saw all kinds of stops… well it did not happen. We might finally be reaching that point. Here is the point. In semiconductors, if you bet against Moores law progress over the years, you lost. I have a gut feeling that is happening is solar. Its incredibly competitive. And it survives like cockroaches. (sorry for the analogy)

  • Let’s check out Gladwells analogy:

    A piece of paper is about 0.004 inches. In miles this is 6.3×10-8 miles. Folding the paper 2^50 times (1.126×10^15 times) and assuming no space between folds, that would equal to 70,931,693 miles thick. The earth is 92,960,000 miles from the sun. So Gladwell is off by about 22,028,307 miles. Sorry, “about” is much less than 22 million. Venus is 67,240,000 miles from the sun so our country’s most famous science explainer/simplifier should say somewhere between venus and earth from the sun. Or Gladwell could use a better analogy. Sorry, the guy gets under my skin. Bottom line, solar power is selling like hot cakes. Or installs are multiplying like flies. An analogy should be simpler to understand than the original subject.

    • drevney

      He meant a 0.005 inchs for this paper.

    • mike

      You’re caught up in the details and missing the larger point. “selling like hot cakes” or “multiplying like flies” doesn’t do much to explain how humans have a blind spot when it comes to perception/conception of exponential growth.

      • Exponential growth and decay is inherent to life like bacteria and humans or any species that doesn’t balance itself well in its environment. The old bell curve use to be ingrained into every student who got a “C” with a 49 percent when the test score range spanned from 0 to 78 percent.

        Fast growth of solar power is just that. Trying to fit deployment growth to exponential or a geometric progression is just curve fitting for affect. A sales technique. Who knows maybe solar power will become so popular that growth is more in line with e^X times error function (x). Now that’s fast. And that would be great for us worried about climate change acceleration.

        Anyway, Gladwell does nerd schtick to sell science-y books. At least he’s a bit more careful in his research and sources, than say Jonah Lehrer.

        • mike

          Not a bell curve, a sigmoid, as already discussed above. In my field this would be similar to the real-world response to a step increase in voltage, eg hitting an uncharged capacitor with 5 Volts DC. The beginning part of of that sigmoid curve can be closely approximated by an exponential curve. That’s what I’m talking about.

          I get that you don’t like Gladwell. Throw out that illustration. Use the example of the penny game. …unless you’re so picky about mathematical exactness that you can’t accept any exponential approximation.

          “Exponential growth and decay is inherent to life like bacteria and
          humans or any species that doesn’t balance itself well in its
          That is exactly correct and we are seeing the same initial exponential curve you see in bacterial growth. The same phenomena that allows some bacteria to overwhelm large multi-cellular animals like humans. That can’t continue forever, but it starts that way and continues for a while. Same thing for Solar PV growth and it is going to overwhelm the power sector. The data speaks if you’d care to listen. A good scientist would.

          • You know I’m being overly pedantic, mathematically speaking? Simply to amuse myself and shoot spitballs at Gladwell. He does do a good job shilling, so there’s that. By the way, a step change is a perturbation in mathematics. It could be an event in a linear, trigonometric, geometric, exponential function or whatever. A perturbation in solar power deployment (commercialization and sales) would be something like a technical breakthrough or a regulatory change impacting growth, i.e a step. It could also be something newsworthy like say the world’s billionaires commingle their billions and plop down trillions into PV solar. That event would perturb or step up solar power installs and generation. On the other hand, something could slow down PV power deployment.

            Here’s a paper on the subject that relates rapid deployment of products and services in the age of the internet:

            “Predicting Electronic Commerce Growth: An integration of Diffusion and Neural Network Models”


            Diffusion as a transport phenomena is modelled mathematically pretty much the same, regardless of what’s being transported, e.g. ideas, chemicals in air, sales of stuff. Transport partial differential equations are bounded, simplified, solved, maybe perturbed and voila.

          • JamesWimberley

            Michael: make us a scenario using your methods. For now, I’ll stick with a simple exponential regression, which obviously gives a much better fit to the data than linear.

          • Who’s us? Is there a comment’s section consortium I have to present this information to? I prefer to be the potshot giver here in the comment’s section. I’ll be more than happy to help at $200/hr + expenses, which, should travel be necessary, includes four star lodging accommodations unlimited meals and entertainment. I discussed more of this issue up thread under Bob Wallace.

            I recommend skimming through the links below to see how econ folks are trying to model growth of technology in the modern world. Take the concepts, but keep the math simple, like you said. Mike’s models and the data don’t seem to conjoin.

            However, when Mike re-does his work (I’m assuming he’ll redo it) he can at least pay homage to how other folks model this kind of thing. Plus buzzwords are sexy.

            Here’s some more from MIT

            Endogenous Growth Theory”

            Here seems to be an old edition PDF:

            And it gets crazy from here and out of my league:

            “Can Second-Generation Endogenous Growth Models Explain The Productivity Trends and Knowledge Production In the Asian Miracle Economies?”


  • DGW

    Why do graph designers hate the color blind or color deficient?

    • mike

      Sorry about that. Looks nice in color. Top curve is green, bottom curve is blue, middle irregular curve is red (actually more like orange here).

  • JamesWimberley

    I would like to see a good argument from a solar sceptic why a continuation of this historic trend (it actually extends back well before 2007) is not going to happen. The talking point “it’s all the unsustainable German/Chinese subsidies” has died down: maybe it was true in the last decade, but it doesn’t explain 2014. We read on this blog of regular small improvements in mainstream silicon and thin-film pv, keeping the 22% learning curve going. We also read of a wide variety of new approaches, some of which will surely take up the baton if vanilla silicon flags. There aren’t any critical shortages of materials on the horizon.

    The greatest immediate risk to solar now is political reaction, with politicians and pundits responding to panicking fossil producers and generators. The remaining subsidies are being wound up faster than they should be, for instance in Australia and Germany. In Spain, the whole sector has been stopped dead. But these moves will only buy a couple of years’ local pause before the solar combine harvester resumes and shreds the rest of the fossil business.

    The probable limit is saturation, Here we can’t use global energy demand as the metric. This is partly because of intermittency, and partly because of uses like aviation, shipping, primary steel, and cement where it isn’t yet possible to go all-electric. Germany will get all its electricity from solar on sunny midsummer days at only twice current penetration (70 GW or so from 35 GW). Expand to sunny midwinter days, and you throw away a lot of midsummer solar. You can’t meet cloudy or night-time midwinter demand at all. So solar saturation depends on the mix, including wind, biomass, geo, ocean and storage. Both solar and wind will be massively overbuilt in relation to mean demand. Since solar will be much the cheapest, the simplest back-of-the-envelope assumption is that solar will be for practical purposes free. So you work out the technical limit to meeting electricity demand from solar, and then fill in the gaps with the more costly other sources.

    • Roger Pham

      The answer to the intermittency and seasonal variation in solar energy is a combination of battery for daily variation and H2 bulk storage for seasonal variation in supply and demand of RE. See my posting above. H2 can also be used in transportation, as well as for fertilizer production, chemical industry, as well as in the synthesis of biomass to liquid fuels.

      • Bob_Wallace

        Not “the” answer, Roger. “An” answer.

        And not likely the winning answer due to cost/inefficiency. But you’re free to keep dreaming. Just try to be a bit more accurate in your posts.

      • eveee

        Why go through all that nonsense instead of charging a bunch of EVs and using them as storage when they are not driven, which is 95% of the time? Its true that there is an abundance of processes that do not really on 24/7 power. Like water pumping and some other industries. Thats a kind of demand management.

        • phoenix

          EVs doesn’t need much power, actually, so it’s doubtful that it would suffice. And I for one wouldn’t agree to let the grid shorten the life span of my batteries.

          • Bob_Wallace

            The idea is not to use EVs to store electricity for the grid, but to use EV charging as dispatchable load.

            Charge EVs when there’s ample power on the grid, drop them out when supply tightens. That will allow easier fitting of supply to demand and eliminate much of the need for storage and dispatchable generation.

          • eveee

            EVs don’t need much power? There are 240 million cars in the US. California has at least a large chunk of those. An average EV has a 24 kwhr battery pack. If it is never discharged more than 70%, the number of cycles in its lifetime is very high. If only 10kwhr of that pack were used as storage, and only 1 million vehicles, you would have 10 GWhr. You could knock a sizable chunk off the peak with that. In 2002, there were 20 million registered autos in the state. That means if only 5% of vehicles in California are electric, and only a fraction of their storage charge is used, the daily peak could be reduced substantially, the drivers could be paid to park their cars and charge the grid, and the effect on the battery life would be minimal. I know what you want to know. How much money could I make. Well, if you got TOU metering in California, you would be paid 30c/kwhr. Tier rates make it higher. So you would get 3 dollars a night. Thats a 1,100 dollars a year.

      • Adam Grant

        Long distance power transmission, e.g. from Africa to Northern Europe will probably be part of the solution.

        • Bob_Wallace

          We’re expanding grids. That’s the solution for minimizing storage and overbuilding of renewables.

          Share the wealth.

          Very soon the West Coast of North America will be one large grid. It’s already connected from Canada to Mexico but those ties are being strengthened and extended into the corners. We should see Wyoming wind flowing to SoCal to help deal with the duck’s head peak in evenings and SoCal solar flowing back to Wyoming when they are locked into deep winter. PNW hydro will do “battery” duty as dispatchable generation. Idaho might even get into the act, selling some of its very abundant hydro and taking back solar.

          • eveee

            Hi Bob. There is some deal about BPA not accepting power from wind in Wyoming and Mac Baucus is stepping in asking help from Obama. BPA has been extremely hostile to wind. Wyoming has good wind and matches load very well. We should use it. Southern California imports a lot of dirty energy from Four Corners. Renewables are tamping that down, thankfully, but we need more growth in wind and solar.

          • Bob_Wallace

            SoCal will suck up that sweet Wyoming wind. CA is dumping coal as rapidly as possible. As coal contracts expire they are being replaced with renewables.

            The Intermountain Intertie that use to keep busy sending Utah coal to SoCal is there for carrying Wyoming wind. And Utah geothermal as soon as they get moving.

            In addition, Utah has talked about closed-loop hydro. Sitting between Wyoming wind and SoCal solar they could set themselves up as energy brokers with “warehouse” facilities.

            Like the teashops on the salt roads in the Himalayas.

    • Vensonata

      In Arizona December gets 70% of the solar production of July. In Berlin about 20% December July . It makes sense to locate large arrays in optimal winter spots. The reflection and cold boost performance.

    • Jan Veselý

      Primary steel is not a problem. There several fully mature techniques to make steel directly from iron ore – iron sponge process, Corex/Midrex process, etc. But it is still cheaper to use coal-based processes. But once you have cheap electricity, …

      • JamesWimberley

        Thanks for that. But steel won’t shift without a tax or regulations.
        Low-carbon cement looks further away technologically.

        • Jan Veselý

          Yep, the best are now old, paid-off steelworks. But even in coal-propeled steelworks, there is a progress. They almost doubled their energy effectivity in last 30 years (old textbook says 1 kg of steel produced = 1 kg of coal burned, present nubers are 1 kg of steel produced = 0.5 kg coal burned).
          But there is still a LOT to be improved. “Small” (1 mil. ton of steel/year) steelwork throws off PJs of high temperature heat just during wet coke quenching (which also wastes water and lowers the quality of coke). If they would switch to dry quenching (Nippon steel patent, given for free use), they would have better quality coke, no need for quenching water and big power plant making free electricity. And there still could be some use for lower temperature heat at the end of the process.
          And there is also sludge cooling, steel cooling, …

    • phoenix

      Well, global wind power installations grew exponentially until 2008 or so. Then they went linear. If we assume 45 GW installations per year and 20 years of life, then wind will level out at 900 GW and supply about 10% of current global electricity demand. Perhaps we’ll see a bit more, up to 20% eventually, if growth picks up a bit again.

      Now, solar is more intermittent, i.e. has less than half the capacity factor of wind, so solar will likely saturate at a lower penetration than wind. The reason is that once you have 10% solar power, new additions are fairly worthless, as they produce power when there is little demand.

      Also, that solar will be cheapest is a fairly bold statement. As it stands, wind is far cheaper and it’s unclear why that would change.

      • mds

        “Now, solar is more intermittent, i.e. has less than half the capacity
        factor of wind, so solar will likely saturate at a lower penetration
        than wind. The reason is that once you have 10% solar power, new
        additions are fairly worthless, as they produce power when there is
        little demand.”
        Nonsense. Solar will achieve higher levels of penetration than wind precisely because it generates power during the day when it is needed most …all across the Southern USA sun belt and across the global sunbelt where the large majority of the Earth’s population lives. Further, low-cost high-cycle-life storage is now coming to the market. This will enable solar to be used at the next highest demand time, the early evening time, and during the rest of the night. As the cost of Solar PV and Storage continue to drop they will become the lowest cost source even at night in many areas. They will have the cost advantage at the end-of-grid.

        “wind is far cheaper and it’s unclear why that would change”
        …because Solar PV is still dropping rapidly in price. …because Solar PV technology has not yet fully matured …because Solar PV and Storage are commodity products, subject to commodity price competition …because Solar PV is easier to install and can be installed in far more places.

        • phoenix

          Yes, solar’s time-of-day characteristics is an advantage, but not enough to offset the worse intermittency in relation to wind. I do agree solar needs very cheap storage to reach high penetrations.

          Wind has a great cost advantage and is also dropping in price. It remains to be seen if solar can ever pass it. Also, nuclear is far cheaper than both in China.

          • mds

            “Yes, solar’s time-of-day characteristics is an advantage, but not enough to offset the worse intermittency in relation to wind.”
            I disagree.
            “I do agree solar needs very cheap storage to reach high penetrations.”
            Cheap storage is already here. It will needs a few years to scale. (examples: Eos, Aquion, Ambri, Telsa LIthium Ion, etc)

            No question wind power will be part of the picture going forward, but source-of-grid Wind will not be able to compete with end-of-grid Solar PV. That IS a big generalization. It will differ in different areas.

          • Guest

            “Also, nuclear is far cheaper than both in China.”

            Maybe some numbers to back up your claim?

          • phoenix

            Well, overnight investment costs for solar, nuclear and wind are all around $2/W in China (and interest rates are held low and operating costs are insignificant). As CF is something like 15% for solar, 35% for wind and 90% for nuclear, the result is quite obvious.

            This is fairly easily googled. Here is a nice birds-eye-view on the nuclear costs:

            China is aiming for even lower nuclear costs, of course, and is exploring a large number of research tracks to make it happen. For instance, this reactor is in an advanced stage of planning and is supposed to get costs down to $0.9/W:

          • Guest

            You gave some numbers for nuclear. All right, $2-2.5/W, according to your link.

            Can you now show wind and solar cost in China?

          • phoenix

            You read my link differently than I did. The nuclear costs I see in there is $1.6-$2.3/W.

            No, I don’t have links to solar/wind costs in China at hand. Didn’t think those figures were controversial, sorry. Please feel free to google yourself or dismiss them.

          • Guest

            China’s industry says solar will cost the same as coal by 2016:


            Wind is being built for $0.6-0.7/W:


            I don’t think that your claim that “nuclear is far cheaper in China” stands scrutiny.

          • phoenix

            Well, your first link on solar vs coal is fuzzy and talks about grid parity. Grid parity is another beast entirely – I’m talking about generation costs. And now, not in 2016.

            The second seems to be turbines only. Then you need tower, installation and grid.

          • Bob_Wallace

            China is installing solar for $1/watt.

            Overnight costs need to be expanded into real costs by including financing costs. Rates may be low, but there’s a huge difference between a few weeks and five years.

            You link to a shill publication.

          • phoenix

            Interesting, do you have an sources to back that up? Still, even $1 per watt puts solar at a disadvantage compared to wind, and I guess that’s why wind dominates Chinese installations.

            I didn’t know nextbigfuture was a “shill publication”. They seem to be reporting on all kinds of cool disruptive tech, including PV, batteries and electric cars.

          • Bob_Wallace

            “Yingli chief strategy officer Yiyu Wang said that project costs for its current pipeline of 130MW in utility-scale solar projects in China are about $1.03-$1.05 a watt.”

            “Wang suggested that Yingli would generate a return in the “higher mid teens” for these projects. “

          • phoenix

            Nice! I’ve googled a bit more and settling for $1.3 for now. See page 39 here (also Yingli):

          • Bob_Wallace

            Here’s another interesting low price report –

            Deutsche Bank said that although the market in Europe had contracted, at least one third of new, small to mid size projects were being developed without subsidies. Multi-megawatt projects were being built south of Rome for €90c/W.
            €90c = $1.20

          • Ulenspiegel

            The average price for 1W onshore wind is around 1.3 USD, for 2 USD you get a Ferrari (=Enercon). 🙂

          • eveee

            Careful. Don’t read too much into overnight costs. Bob has had more about this in the past. There are many factors that affect nuclear costs that are not considered in overnite costs.

            Here is the definition.

            Overnight cost is the cost of a construction project if no interest was incurred during construction,

            But nuclear can sometimes take ten years to develop. At 7% interest rate, the finance cost is equate to the day one cost.
            Nuclear costs are heavily influenced by construction times, delays, and finance costs (interest).

    • mds

      “Since solar will be much the cheapest,”
      ” the simplest back-of-the-envelope assumption is that solar will be for practical purposes free”
      I agree with the rest of what you’ve written.

      HVDC connections across the Midwest have turned Wind into a baseload provider. Consider HVDC for Solar PV from low latitudes to higher latitudes with less sun.

      • JamesWimberley

        I don’t mean that solar will actually be free, just that it will be so much cheaper than all competitors that it’s a useful thought experiment (say for 2050 zero-emissions scenarios) to ignore the very low remaining price.

      • eveee

        Hey mds. Do you have a link to those HVDC connections? I would like to see them.

  • Roger Pham

    This growth can go only so far until the grid will be saturated with daytime solar electricity, unless energy storage can keep up with it.

    The growth of Solar City with plan to open Giga Battery factory for residential electricity storage is a good planning move ahead. Tesla’s battery, Panasonic NCA 18650 can be charged 5,000 times with capacity drop to 80%, which means it will last over 13 years. If the projected price will be $200/kWh, the cost of each kWh stored will be $0.04. If solar PV electricity costs $0.06/kWh, and only 1/2 of it needs storage, then the average cost will only be $0.08/kWh, thus allows adequate profit margin to match grid electricity average rate of $0.13/kWh. This is good for sunny locations with mild and sunny winters.

    For locations with cold and cloudy winters, then seasonal-scale energy storage medium like H2 can be used. H2 can be made from excess solar and wind energy in springs, summers, and falls, stored in underground caverns at very low cost, for use in winters and in rainy, cloudy and calm days. For combined heat and power applications in which efficiency of utilization can approach 100%, this storage medium can rival the round-trip efficiency of Lithium battery, and will beat any other bulk energy storage means, and may even rival NG economically. The days of fossil fuels may be over in the foreseeable future. FCV’s will be able to take advantage of this low-cost synthetic fuel. We should start planning for this right now to catch the low-cost solar energy in the near future.

    • Bob_Wallace

      I’m not betting on H2 as a grid storage medium.

      First, H2 is very lossy. Less than 50% efficient compared to flow batteries at 65% to 75% efficient and PuHS at 85% to 95% efficient.

      Second, H2 is hard to store compared to low pressure flow battery tanks and water reservoirs.

      H2 would have to win based on much lower capital costs and I’m not convinced it will.

      • Roger Pham

        It is true that H2 is lossy as electricity storage medium, with round-trip efficiency of only 66% electrolysis x 60% FC = 40%, vs. Li-ion battery of above 80%. For this reason, daily-use electricity storage should use battery, as I mentioned.
        However, for combined heat and electricity energy storage to last the whole cold winter in northern climate, I challenge you to find a more cost-effective and more efficient medium than H2 when stored in underground caverns and used in FC for CHP.

        • Roger Pham

          For CHP, or Combined Heat and Power, the round-trip efficiency of H2 can be 80%, which can rival Lithium battery round-trip efficiency.

          • eveee

            This requires more careful analysis. Harvesting electrolysis heat for use requires local electrolysis to take advantage of chp for space heating. Then the stored hydrogen can be used later as a heat source. There is a discussion here about it and there are pluses and minuses. Gas cannot just be fed into existing natgas infrastructure due to hydrogen embrittlement. Special equipment and burners must be used.


          • Roger Pham

            Right now, existing NG piping can handle up to 20% of H2 mixed in without problem. If old NG piping are to be replaced, new piping that is H2 compatible can be used, so that in due time, all piping will be compatible with 100% H2 without extra infrastructure cost.

            If electrolysis is done in the summer, this waste heat is not usable and has to be written off as 20% loss of efficiency. Winter use of H2 can take advantage of CHP and therefore the H2 can be used at near 100% efficiency. Mass production of FCEV’s will give rise to affordable FC units for home use for both heat and power. This will insure against power black out due to winter storms.

          • Bob_Wallace

            You might wish to use “could” and “might” in the place of “will”. None of us have a functioning crystal ball.

            Will we use H2 as a long time grid storage technology? That’s pretty questionable. H2 storage is very lossy, well over 50%, which means that the cost of generation and storage would have be be very much lower than that of PuHS and flow batteries.

            Don’t see how one stores H2 as cheaply as water or flow battery chemicals. Most of the infrastructure for PuHS and flow batteries would be paid for by their daily cycling. The cost of storing some long term power in either is simply a matter of enlarging the reservoir/tank.

            BTW, our NG distribution system is very leaky. H2 is even harder to contain than methane.

            I do think you’re in love with hydrogen while knowing very little about it. At least very little about how it stacks up against other storage technology. Did H2 FCEVs play a big role in your favorite movie or something?

        • Bob_Wallace


          • Roger Pham

            Hydro-electricity for RE storage is highly dependent on geography, and does not have much room to grow. Not any more efficient than H2, nor any cheaper. Capacity is limited and does not scale up well.

          • Bob_Wallace

            Wrong, Roger.

            We have. for example, thousands of existing dams which are good candidates for conversion to PuHS. And no shortage of places to put closed loop pump-up.

            PuHS is 85% to 95% efficient. H2 is less than 50% efficient.

            Capacity is not at all limited for PuHS and scales extremely well.

    • eveee

      Hydrogen is unnecessary, costly, and inefficient. Much better to just charge EVs during the day while the sun is shining while at work. Daily commutes are the bulk of the demand, not long trips.

      • Bob_Wallace

        I suspect we’ll see lots of work/school charging outlets in the future. The more we can directly use electricity, the cheaper electricity will be.
        Having EVs as dispatchable load to suck up solar and wind supply peaks will greatly decrease our need for grid storage. Lots of EVs will be able to skip charging for days when supplies are low so that what supply is available will go to more demanding needs.

      • Roger Pham

        True for sunny and mild climatic locations. However, for locations that have cold and cloudy winters, and locations with unreliable solar and wind energy, you will need nuclear energy, or RE with large storage capacity such as hydro-lake-dam or H2.

        • Bob_Wallace

          Where, exactly, are these mythical regions with very limited/unreliable wind and solar in the winter?

          Not Alaska, for example. Lots of wind plus hydro, tidal and geothermal.
          Patagonia? Wind howls.

          • JimBouton

            “Where, exactly, are these mythical regions with very limited/unreliable wind and solar in the winter?”

            Ten years ago, the nuclear folks would have responded to your question with: Germany.

            They are left with the lost continent of Atlantis. It is difficult to put solar panels under water.

          • Bob_Wallace

            Neon tetras.

            We need to genetically breed really, really big neon tetras.

            There’s hope for people in Atlantis….

          • JamesWimberley

            Nah, they should be LED tetras.

          • phoenix

            Germany is still a good answer. Wind + solar penetration is still very low and their costs are high.

          • Bob_Wallace

            People confuse the price of electricity + taxes in Germany with the cost of electricity alone.

            Take a look at how the wholesale price of electricity has been dropping in Germany.

            Wholesale price. Exclusive of the taxes put on retail rates.

            Let me rewrite that last sentence for you –

            Wind + solar penetration is still very low and electricity costs are dropping like a lead zeppelin.

          • Roger Pham

            Oh yeah? A few cold winter days without sun and wind in Alaska and your battery storage will be depleted. A fuel source would be needed, either wood, charcoal, coal, NG, nuclear or H2. Energy demand is so high in the winter that if you can build enough solar and wind for the winter, you will have too much surplus in other 3 seasons.

            Even Chipmunks know that fact and they hoard enough food for the winter.

          • Bob_Wallace

            Geothermal. Hydro. Tidal.

            Alaska is energy rich.

            Roger, do you realize how often you make statements of fact which are, in fact, wrong?

          • eveee

            I am not sure about the wind regime in Alaska. Certainly the coasts have typical diurnal land sea breezes. Inland…. I expect the wind is greatest in winter. That makes an ideal source for heat. Heat can be stored. There will be some need for other sources and/or storage then. Heres a good reference


            Its says

            In 2009, Kodiak Electric Association (KEA) installed the state’s first megawatt-scale turbines and then doubled the size of its wind farm in 2012. The project’s six 1.5 MW turbines now supply more than 18% of the community’s electricity. Combined with the Terror Lake hydroelectric project, KEA can now shut off their diesel generators almost all year. – See more at: http://alaskarenewableenergy.org/why-renewable-energy-is-important/alaskas-resources/wind/#sthash.djUHpENQ.dpuf

            Also says:

            The largest areas of class 7 (superior) wind power in the United States are located in Alaska. Much of coastal Alaska has “good” or “excellent” wind resource – See more at: http://alaskarenewableenergy.org/why-renewable-energy-is-important/alaskas-resources/wind/#sthash.djUHpENQ.dpuf

        • eveee

          I know an area like that. One that does it without nuclear.


          “According to the map, Germany received amounts of sunlight comparable to the region around notoriously cloudy Seattle and arctic Alaska, while most of the states along the Canadian border got 50% more sun than the Southern half of Germany.”

          Since Canada and Alaska are so much sunnier than Germany, they should be able to manage.


        • eveee

          I thought so, too, until I realized Germany was exactly one of those regions. Its as far north as southern Canada and gets about as much sun. Its managed to do with decreasing nuclear, too. So it is possible. Thats really surprising. But Canada and Alaska can do it in entirely different ways. They both have abundant hydro and relatively small populations. Wind looks good, too. Your point that they are less ideal for solar is true. Its just that solar is becoming so inexpensive, the amount of sun is becoming less important.

    • Vensonata

      Lithium. Its hard to pin down the exact cycle numbers, often you will see 5000, but only the first 2000 are above 80%. The last 3000 maybe 60%? Saft battery storage unit in Germany confidently declares 7000 cycles! But not at full charge…darn. At $200 per kw I’d put the price at 6cents kwh for lithium. Still need battery mangement system, maybe 1c more. Bottom line 7c kwh. Definitely closing in to kill king coal.

      • Roger Pham

        Your number is typical of LiFePO4 battery, but Panasonic NCA 18650 has been proven by independent University testing (Penn State) to last to 5,000 cycles to get to 80% capacity. The test was done at 80% DOD and at 3 C discharge rate!

        • Vensonata

          Thanks, I’ll do some research.

          • eveee

            I assume you have the LiFe data, but I will post it here anyway. 5000 cycle at 80% DoD. The numbers are a bit better than you cited.


          • Roger Pham

            Thank you, eveee, for the link. LiFePO4battery has advanced quite a bit since just awhile ago that I can recall. Although this Winston battery at 100Wh/kg cannot match Tesla’s battery at 230Wh/kg, it may be cheaper, while capable of fast charge at 3C or in 20 minutes from 0-100% and burst discharge at 10C, good for PHEV having under 10-kWh capacity, beside for home use storing solar PV.

          • eveee

            You are welcome, Roger. I try to let people know that, because they can make their EVs last a long, long time. It makes no sense to discharge them deeply. You lose too much benefit. Yes, it is getting so cheap, its starting to look good for storage off grid in remote locations. Take a look at the Balqon energy storage units. Complete systems with DC DC from solar to battery and inverter to create 120VAC, plus BMS. A complete turnkey system with 36kwhr storage costs 12k. Thats enough for more than a complete day and 1.5x the storage in a NIssan Leaf.

    • Bob_Wallace

      Roger, I’m going off topic a bit to share with you a report I just discovered.

      It’s a recent (2014) ‘well to wheels’ analysis of GHG emissions for ICEVs, EVs, and FCEVs. It shows the problem of generating H2 from NG, it’s about as bad as running EVs off today’s grid. And it shows that H2 FCEVs run with “clean” H2 would be about the same as EVs charged with renewable electricity.

      That says that H2 FCEVs could be an acceptable way of getting our vehicle carbon problems under control. But the second page has a graph (copied below) that shows part of the expense problem of using H2 as a storage medium as opposed to batteries. H2 is just inefficient.

      Regardless where we produce the H2, at a central station or “at home” and how we transport it (truck or pipe) it would take 2+x more electricity per mile driven.


  • mikgigs

    It is excellent publication with awful graphic plots.

    • mike

      Thanks and ouch 🙂
      Please see Bob’s re-post of correct graphs below.

      • mike

        FIxed in the main text now.

  • Ulenspiegel

    Could it be, that pic 1 and 3 are the exactly the same, and pic 2 and 4 are the same exponential function with different y-axis.

    I do not see any semilogarithmic representation of the data as indicated in the text.

    • mike

      Yes, they are. That is not what I sent in, but got corrupted in reformatting. Plots 1 and 2 should be the same as plots 3 and 4, except the latter two should have a logarithmic vertical axis. That is what I sent in. Also, year was added in to GW value for each point. This makes it harder to read the GW values and really wasn’t necessary since the year is easily determined from the grid shown.
      Sorry about that. I’ll see what I can do about fixing.

    • mike

      1. Linear Graph Showing Exponential Growth

      2. Linear Graph of Exponential Growth out to 2022

      3. Log Graph Showing Exponential Growth

      4. Log Graph of Exponential Growth out to 2022

      • mike

        Dang, didn’t work.

        How did you paste that sigmoid graph in?

        • Bob_Wallace

          Pasting graphs/images.

          1. Get them into jpeg or gif format.

          (I had to do Print Screen captures with your files and then crop away the excess stuff. If you need to know more about that I can fill in.)

          2. Click on the little mountain/moon “Upload Images” icon at the bottom right of the Reply box.

          3. Find the image you want on your hard drive and “Open”.

          Often the image won’t show when your comment appears.

          Refresh the page.

    • mike

      Bob has been kind enough to post the corrected graphs below.

      • mike

        FIxed in the main text now. (Thanks Zach!)

      • Ulenspiegel

        Thank you!

  • Chatteris

    Fossil fuel’s resistance would seem to be futile!

  • Eric Gerber

    Total global power consumption for all types is about 130,000 terawatt hours. The article cites 20 terawatts but 20 tW of solar does not produce the same amount of power in a year as 20 tW of gas fired power plant. 20 tW of solar should produce about 40,000 tW hours. It would be great if the articles would to move to consistently discussing power in terms of energy hours as opposed to just nameplate max output so we could more easily put these developments in to a global context.

    • Vensonata

      Power consumption or electricity consumption?

      • Mint

        Yeah, I think he’s talking about total energy consumption. World electricity consumption is only around 20,000 TWh/yr. US electricity consumption is ~4000 TWh/yr.

        • vensonata

          Yep. Each unit of pv can be multiplied by 1600 to get annual production. Although a single kwh pv on a 2 axis tracker in arizona will bring in about 2100 kwh per year,in New Mexico a little more. Staggering!

          • JimBouton

            My 245 watt panels are clocking in at an average of 400 kwh per year. They are stationary, plus, I suspect my rate will be closer to 420 kwh per panel once I finish up with the July, August, and September months.

            Do you get the 1600 kwh value by assuming a 1000 watt panel?

          • Vensonata

            Yes, one kwh panel produces 1600 kwh year fixed array…so you are bang on at 400kwh at 225 watts. As I say if you are on dual axis tracker it can get as high as 2200 kwh year from one kw panel in say New Mexico

          • Eric Gerber

            Interesting to think about what the steady state solar production will need to be when we reach capacity. Assuming we need 40,000 TWh of solar power globally and a panel life of 20 years, we will need to continuously manufacturer 1 terrawatt of solar every year just to deal with obsolescence. A great challenge to have.

          • Bob_Wallace

            20 years. Almost certainly a bad assumption.

            Our oldest solar array is now 40 years old and at 35 years had lost only 3.85% of original output. About 0.1% per year. Panels in places with higher UV concentration will likely lose more output, perhaps 0.3% per year. But they should still be giving us close to 90% of original after 40 years. And no one knows how long a thin sheet of rock underneath a pane of glass will last…..

          • mds

            I think you meant 3.85% off of the original output, or 96.15% of the original output.

          • Bob_Wallace

            Yes, for the University of Oldenburg array. (And it was a 3.88% loss after 35 years. I forgot a bit.)

            The “close to 90% of original” is for an array mounted somewhere like a high desert that is going to lose something higher. The NREL finds less than 0.4% for panels made after 2000. 0.3% over 40 years would be 88%.


        • Eric Gerber

          Yup, Power consumption. Electricity as you note is a lot less. However, the boundaries between electricity and other forms of power are blurring – electric transportation, electric heating, so it seems like the target should be a significant portion of total power. Note about half of the 130,000 TWh is lost due to inefficient conversion (heat loss) which are not relevant for wind and solar as they are for coal and oil.

    • mike

      Good point, but a coupla additional doublings negates this point …and only adds a few years to the curves I’ve plotted. My point remains. Solar PV will eclipse NG in a decade.

    • eveee

      I think it would be a mistake to assume all power comes from electricity. Passive solar and solar hot water create heat without electricity. Its not very good to heat with electricity except for ground source heat pumps. Super insulation and passive solar might be better. The place where electricity increases would most is in transport. EVs. We need to get that show on the road. Transport is really a much bigger deal for carbon emissions than electric power, except in China. If we make the electric power clean before EVs arrive, the expansion into EVs goes smoothly. IMO, EVs will take off in the next four to six years. The decade after that will be one of almost revolutionary change.

  • Adam Grant

    Wind is also on an exponential install curve, although slower than solar. It’ll have an important role until we can send power from the bright to the dark side of the planet or energy storage becomes cheap.

    • phoenix

      No, wind has entered the linear part of the S-curve, since 2008. Exponential growth is over for wind.

      • Adam Grant

        Keep in mind that wind turbines have continued to get cheaper since 2008, and (as far as I know) the harvestable wind resource is nowhere near tapped out anywhere on Earth.

        Improvements in blade design and modelling allow a farm to get more energy out of the wind, in increasingly low-wind conditions. Combined with local batteries for output smoothing, wind power has been increasing its advantage over fossil energy sources.

        • phoenix

          I agree on all these points except batteries, which are not currently a factor. However, the numbers are what they are.

          Even GWEC, the trade association for global wind, expects wind installation growth to taper from 15% this year to 12% in 2018. A far cry from the 30% we became used to in the years before 2008.

          • Guest

            Actually it shows rising to 60GW annual installations. If 25 year lifetime and 25% CF is assumed, then 60GW/year corresponds to 15% of world electricity.

            60GW is 30000 2MW turbines. World made 800000 airplanes during WW2 in 5 years. That’s 160000 per year. So wind turbines are only about a fifth of that. There’s still a lot of room to grow.

            It could concievably get to 200GW/year, which would give about 50-60% electricity.

          • phoenix

            I’m not sure the airplane analogy is valid, but I do agree that it is fully conceivable from an industrial standpoint to build 200 GW/year. I’m just saying we aren’t moving in that direction at all and that exponentiality for wind is off and won’t come back.

          • Ulenspiegel

            What is the capacity factor of a typical bread and butter turbine in 2008, what in 2014? What is the market share of slow wind turbines in 2014-20?

            IMHO a comparison of installed GW is misleading when at the same time the CF of instlled turbines increases significantly. A picture with produced TWh electricity would be better.

          • phoenix

            You’re right. Just checked the BP numbers and it shows 20% increase in energy delivered in 2013, while capacity is up only 12%. OTOH, there is a lag in TWh during a year compared to end-of-year GW numbers, so the bad 2013 installation rate doesn’t fully show.

          • eveee

            WInd can be elusive. For example, China outpaced US in wind capacity. Which one has more wind energy? US. The Midwest is one of the best places in the world. We have barely tapped it. The two best states are the Dakotas. Very little development.

          • phoenix

            Approximate CF on the installed global base seems to not change very much. If I use yearly TWh and average GW during the year (mean of the year-end value and the previous year-end value), I get the following series for 2005-2013:
            22.2%, 22.8%, 23.2%, 23.2%, 22.5%, 21.9%, 22.8%, 22.8%, 23.7%. (Surprisingly low CF values. Data from latest BP Statistical Review of World Energy)

          • Ulenspiegel

            The result is not a surprise 🙂

            The current turbine populations have in some larger wind energy countries (USA, Germany, Spain, Denmark?) an age structure with a very high share of older smalr turbines with low CF. Therefore the avarage CF you cite is still quite low. However, these older turbines will be replaced in the next ten years.

            Prime example would be Germany: There were 18000 Turbines with 18 GW nameplate capacity in Germany in 2005. They provided 27 TWh electricity, 1500 FLH, less than 20% CF.

            To get a better feeling for potential of onshore wind I check the documented CFs of more modern wind farms with turbines from 2005-2013, here you have 25-30% CF for normal turbines, 30-45% for dedicated slow wind turbines.

            I think it is a save bet, that around 2030 the avarage will be around 30-35% CF.

          • Bob_Wallace

            Looking at CF per year performance for Danish wind farms one can see how CF tends to improve over time.

            From the mid-twenties to mid-forties and tapping on fifty percent.

          • phoenix

            Yes, but I was still surprised at slow CF progress. Since installations have been accelerating somewhat, half the global capacity has been added in the last four years alone. I had expected a larger impact from that.

            But I just did a quick check: The US had a 32% CF in 2013, Germany had 19% and China had 18% CF! Germany is more understandable due to age, but China? Either they are underreporting or they have really lousy installations/operations.

          • Ulenspiegel

            Interesting observation in re China.

            I have to modify my statement: I think it is a save bet, that around 2030 the avarage will be around 30-35% CF in Europe. 🙂

            In N. America with larger turbines it should be >40% around 2030. 🙂

          • eveee

            No. Its all about the site. The same turbine can give different CF in different sites. China has less good wind sites. US wind sites are rockin it. For really awesome CF go to the Orkneys in Scotland. World record, I recall. They are in the 50s and 60s. Annually. They have to be built not to blow over. Its howling.

          • phoenix

            Sure, but China is big and there should be plenty acceptable sites with 30%+ potential. There has to be other explanations.

          • eveee

            Yes. China has plenty of wind potential and its taking advantage of it, too. It had some issues with grid, just like Texas, but it responded just like Texas and those issues are being addressed.

          • Bob_Wallace

            China is currently running HVDC to Inner Mongolia and other areas where their wind resources are the most abundant. I’m not sure why CF is so low currently. Perhaps they aren’t doing a good job of matching turbines/blades to wind conditions.

            I’d like to see someone do a piece on equipment matching to site. It seems like there has been significant progress lately. Not so much ‘one size fits all’ of an approach these days.

            Germany has likely only a temporary low onshore CF. They are busy replacing older, shorter rigs with more modern, taller ones in their somewhat limited good wind areas.

          • Sim

            Probably takes ages to get the window of opportunity to install them in the Orkneys.

      • Bob_Wallace

        I just wouldn’t bet on that. You might be right but many countries, especially in Asia, Africa, and South America are just now starting to build their renewable industries.

      • eveee

        Keep in mind that when we are taiking about exponentials we are talking about compound rates. The S-curve is not a brick wall. I think wind is slowing recently, but still can’t be sure if its permanent or not. I think the probability is that there is some slowing. What happens as it slows is that the growth rate goes from 25% to 20% and on down. That really is not too significant as its gaining even as the growth rate goes down. Consider that if it still goes down to 7 % over 10 years it doubles in less than 10 years. You have to check the math on a sliding interest scale, but one might consider an average of 14% (hypothetical). That would still be doubling every 5 years. So a 15 year period might increase the amount 4x. Purely hypothetical, and assuming it really is slowing. Add EVs to demand, and an increase in demand for replacement generation for coal and its possible to see the growth rate go back up. Since the US is at 4.18% wind now, I feel comfortable in predicting at least one doubling to 8.4% electrical in the next ten years. That is starting to look like real numbers. That is probably conservative.

        • phoenix

          I kind-of agree with you on the numbers, but we need like 4% per year, not 4% in a decade.That’s a problem with much of the renewables’ reporting I see nowadays. It routinely portrays far too small gains as really, really awesome. I feel it promotes complacency when we need urgency.

          • eveee

            I was talking about wind and I guess from your comments you mean going from 4.18% wind to 8.4% wind in ten years. Thats only 7% growth rate annually. Actually, I expect wind to maintain a healthy growth rate somewhere in excess of 15%. Thats a doubling every 5 years.

            I agree we need good growth rates. I think if you look, you will find that we are far, far in excess of 4% per year growth rates. Solar is in the 30 to 40% range. Wind is coming down from the 20% into the teens. Solar is doubling in less than two years. Thats extremely rapid growth. The Chinese certainly seem willing to be that aggressive. I think its easily conceivable that the US could double wind twice in the next ten years. Thats more like 16.8% total wind energy in ten years. The solar rooftop market is wide open doubling every two years. The utility scale solar market is about as good. Since the solar growth rates are so good, we can expect solar to catch up with and probably overtake wind. Month by month, wind and solar installations are replacing FF capacity in the US. Only gas remains. What we don’t need is an increase in demand in China that swallows up all the added renewable capacity and opens the door to more Chinese coal plants.


          • Bob_Wallace

            If we’re installing enough wind now to convert 1% of our annual generation to wind and could double that twice in the next ten years then we’d be doing fine. At a 4% per year conversion over the next ten years we could retire all coal (currently at ~40% and dropping) and a hunk of NG.

            If solar can get to 2% per year in the coming decade then we could replace almost all our fossil fuel generation in a 20 year window with wind and solar. Which would be excellent.

            We don’t need long term installation rate growth. We just need growth until we hit a reasonable rate and then keep on truckin’.

            Accelerate to highway speed. Maintain speed until close to destination.

            (For the nitpickers. Yes, we would have to build storage – yada, yada, yada….)

  • Adam Grant

    Although solar’s fraction of humanity’s total energy use will top out close to 100%, total energy use itself will keep expanding for some time after that as we find new uses for cheap energy, e.g. climatically significant levels of sea water desalination and cleaning the excess CO2 out of the atmosphere.

  • jburt56


  • Hans

    I expect PV growth to be more like a S-curve, exponential at first, but than going asymptomatically to some saturation level (hopefully close to 100%)

    • Bob_Wallace

      That’s typical for technological shifts.

      Exponential during initial ramp-up, settling into something like a steady rate until the change is mostly completed.

      • This math behind those curves is a classic error function or erf, in math symbols. The error function shows up typically when solving diffusion (transport) or dissemination (sales) phenomena. The curve will putter for awhile, go rapidly up, and putter out again. The error function comes about when applying Laplace transformations during equations solution. This is called an analytical modeling versus empirical modeling. Mike is doing empirical modeling, or what happened in the past should happen in the future, give or take some issues.

        Here’s a primer: Error and Complementary Error Functions http://www.mhtlab.uwaterloo.ca/courses/me755/web_chap2.pdf

        I discussed this issue with mike and linked a study paper on electronic commerce deployment further down thread. Analytical solutions for dynamic systems prediction are being used more over the past 30 years for predicting things like climate change, stock fluctuations, business growth, etc. Most of the math stayed in chemical engineering to help oil companies extract and refine oil into usable products. Then analytical modeling moved over to environmental and business later.

        Curve fitting of data is always fun. Modeling phenomena for prediction by partial differential equations is a bit harder. Its an age old argument in modeling of dynamic systems, which solar deployment most certainly is.

        But remember, if you want a straight-ish line on a graph, take the log both the x and y axis. It doesn’t say much via accuracy, but the plot looks straight.

    • mike

      I think you’re missing the point. Clearly exponential growth of Solar PV cannot continue forever. It will change to linear growth and then slow down to a steady production rate, as all new technologies do. That’s a given. The point I’m trying to make is Solar PV is still in the exponential growth phase. It has not even reached a linear growth rate yet, let alone slowing down. The larger point is most people, even those who are pro-solar, think the Solar PV growth rate will slow down in the near future. They think high volumes of production are difficult to achieve, so that is why it will slow down. It’s not. Solar PV growth will only slow down when the global market begin to saturate and this is not close at hand. Solar PV is going to become the single largest source of global power. It’s going to happen fast. It’s going to include market like (North & Central) Africa that aren’t even power markets now. …exactly similar to cell phones.

      Yes, sigmoid curve.
      No, not in the near future.


      • Bob_Wallace

        I agree, Mike. We’re likely in the rapidly accelerating phase of a sigmoid curve. Solar hasn’t even started in many countries and we’re going to see meaningful panel cost drops over the next couple of years.

        There’s no real limit set by manufacturing capacity. We created more silicon processing in less than two years of hitting a supply limit. Glass, aluminum, plastic? We know how to make that stuff in large quantity.

        The earliest adopting countries might be past their rapid acceleration phase (or most of it), but there are a couple hundred countries with most yet to get cooking.

        Solar is unlike any other generation technology. It can be installed large scale, small scale or tiny scale. If utility stalls, rooftop is still free to zoom. And vice versa.

        And we might want to remember that other driver of solar installations – climate change.

      • phoenix

        Aren’t you disregarding some facts that paint a different picture? I mean, global PV installations has been 30, 32, 37 GW in the years 2011, 2012 and 2013 respectively. That’s not linear, but isn’t exponential either.

        I’ve seen forecast for 45 GW in 2014, which would not be exponential as it will have growth dropping further from 38% to 32%.

        • andereandre

          nope, it is more than exponential.
          Exponential would have been if the 6% increase between 2011 and 2012 would have repeated in the other years.
          But the percentages are 6, 15, 21.

          • phoenix

            Sorry, now you lost me. Global growth in PV has been 75%, 43% and 38% in the last few years. What does your figures represent?

          • andereandre

            your figures. 30, 32, 37, 45 GW. So the rise in new installations is more than exponential. You are probable right when talking about installed capacity.

        • mike

          No, I’m not disregarding facts. As I stated, it is not well-behaved exponential growth, BUT the results to date are above a 41% exponential growth curve. Further, this has been the case for a decade and a half, probably more. Aren’t you trying to linearize data that better fits an exponential, even if irregularly so?
          Forecast of 45 GW in 2014 is still well above the 41% exponential curve. We’re seeing the irregular exponential like growth typical of disruptive technologies.
          Again, your point focuses on the details of what a perfectly exponential curve looks like, but misses the larger point. Maybe this is part of the reason humans have a blind spot for the exponential growth of disruptive technologies?
          The real data does not make a mathematically perfect exponential curve, but the exponential growth is there. Real world data rarely fits perfectly. Line fit algorithms are often used in scientific studies.

          • phoenix

            I agree exponential growth in the real world is not that well behaved. But we all agree this will follow an S-shaped curve, right? So at some point exponentiality will cease. How do we know when it has happened? How soon can we say? To me, it seems growth has been subsiding three years straight and thus exponentiality is probably off, I can’t say for sure.

            You say “45 GW in 2014 is still well above the 41% exponential curve” even though it is 32% that particular year. The problem with that is that you can do another curve fitting in a few years and say, “50 GW is still well above the 30% exponential curve”. The curve fitting algorithm can give you plausible exponential answers for a few more years than is warranted.

          • mike

            “To me, it seems growth has been subsiding three years straight and thus exponentiality is probably off, I can’t say for sure.”
            Ok, now I understand your point better and actually agree …partially. It comes down to the “can’t say for sure” part. Neither can I …BUT we could still be in the exponential phase of growth. My guess is we still are. I would suggest the reduced growth for the last few years was due to the double whammy of the global economic crunch and the related pull back of subsidies in some significant markets. In addition to the economic crunch, Solar PV has been transitioning from a largely subsidy driven market to a cost driven market in many areas: Chile, Australia, Hawaii, other sunny islands, the Middle East and North Africa (MENA), Spain, Itally, etc. The economic advantage is there now, particularly at the end-of-grid, AND the cost of Solar PV panels/installation is still dropping.

            Additional evidence for continued exponential growth comes from the high rate of Solar PV growth we are seeing again in 201, and the predictions for a high rate of growth we are seeing for 2015.
            We have gone from a punishing Solar PV over-supply market to an over-demand market in a remarkably short period of time. Classic disruptive growth pattern.

            …”thus exponentiality is probably” ON, “I can’t say for sure.”
            Nobody can. My crystal ball just reads different than yours.

        • Graphite Gus

          Phoenix, You are taking an arbitrary 3 years to make your point. I could take another 3 years to make a different point. The author takes 7 years, but even that can be misinterpreted. Bob Wallace’s graph of technology adoption is probably the best indicator of the very rapid growth – exponential – for solar in the next decade. But as always in these arguments, if we really want to know, we should stick around and find out.

          • phoenix

            Well, when gauging if something is (still) behaving exponentially, the last few years are obviously the most important.

            For instance the global human population growth looks exponential in a graph if you zoom out enough, even though it has been sub-linear since 1990. (Peak population growth in millions was then.)

            The graph of technology adoption is nice, but I don’t think it is really relevant here. Everybody wants commodities like electricity and phones, but solar is a powersource that competes with other powersources. We can hope for its eventual domination, but a lot of hurdles need to be overcome first, and last year, coal and oil had the largest additions to global energy. Also, solar has a long way to go if we look at the graph of power adoption for pioneer countries (nuclear, wind and solar):

          • eveee

            It would be unwise to look at the last three years to find a trend. Even if it was the beginning of a cooling off, you could not be sure, because there is noise in the signal. You would have to wait longer to have confidence. Thats where probabilities play a part in noise. Thats why the chart is a line through a series of points that don’t line up perfectly. Trying to plot a line or curve through the actual numbers over a short period would only mislead you.

          • phoenix

            It’s common practice in extrapolations to give higher weight to more recent developments. How long would you like to wait before acknowledging a broken trend? In this fast-moving-world, three years is a long time. And with high growth rates, getting back on the trend line is hard after three years.

      • eveee

        Semiconductor people are familiar with the same phenomena in electronics. Its called Moores law. Its the reason the computer you are using today, costs no more, but has vastly increased capability.

        Moore’s law is the observation that, over the history of computing hardware, the number of transistors in a denseintegrated circuit doubles approximately every two years.


        Swanson Law is one for solar, but it is a different one.

        “Swanson’s Law is an observation that the price of solar photovoltaic modules tends to drop 20 percent for every doubling of cumulative shipped volume.”


        What they have in common is exponential growth.

        Even in limited terms, its is not hard to see that expanding volume means lower prices, more demand, … higher volume, lower prices, more demand…

        Real factory volume has only started and has not hit full scale. Meanwhile, as prices beat existing competitors, a replacement effect takes place. Solar wont slow down as it gains competitiveness in the marketplace. If anything, it may speed up. Then there is the growth of the power market itself. As solar gets cheaper than any other source, it may create markets that never existed before, because no other source has ever been that cheap. It could behave like a boom, if its EROI improves.


    • Alastair Leith

      we’re a long ways off the S plateau though. when solarPV becomes trivial in cost, i expect a lot of building product manufacturers to add it to cladding, landscaping and roofing products as a point of sale feature, so we could easily get overshoot to 150% RE even by targeting ~90% once the deployment builds a head of steam, so to speak.

Back to Top ↑