The Trouble with Solar Power

Solar power is, without doubt, the renewable kind of energy that is most heralded as the solution to provide energy. The argument sounds tempting and reasonable, but it is certainly not fully thought through. Let’s have a look at the characteristics of solar power to see how much truth there is in those claims and how far solar power can take us.

Sunlight seems to have a lot of energy. One square meter facing the sun receives 1000W of power or 1GW per square kilometer. But unfortunately we are only able to convert a fraction of that into electricity. Where money is no object, this fraction can reach the lofty heights of 30% in multijunction cells. When you can spend hundreds of thousands or a million dollar on a kilowatt of power, there is indeed a lot you can do. But that’s like suggesting people should replace their cars with something like a Learjet.

But money is an object. Nobody can afford to spend hundreds of billions or a trillion dollars on 1 GW of nominal peak power (delivering 100-200MW on average). What we need are mass-fabricated cheap panels and those are working at efficiencies on the order of 10 to 15%, namely thin-film and poly-crystalline cells. Higher efficiencies getting cheaper in mass production can perhaps be hoped for, but are not currently in sight.

I am aware of the regular slew of announcements of new technologies, but practically all of them have caveats attached and are basically to be understood as a plea for more funds. Not that there is anything wrong with that, but you should not assume that they are indeed ready for production any time soon. Very often drawbacks such as a short lifetime or low efficiency are beyond improvement. The problem there, however, is that very little money is put to work in fundamental research in solar power, while the general impression may be different.

A lot of the money spent on solar power goes to the banks financing the installations. Assuming a 2% interest rate for the credit means that almost 20% of the money goes to the bank and investor. (Assuming that he will also want to get a 2% rate of profit.) Of course you would be deluding yourself if you expect those investors to get such conditions or expect such a low rate of interest. Assuming a 5% interest means that almost 40% of the money will go to banks and investors.

Most of the rest goes towards investment into production facilities and cost of production. Even of the R&D budget that is left, only about one tenth or less is poured into the kind of fundamental research required to develop such technologies. Which should serve as an explanation why there has in fact been little in the way of progress.

This is a problem of the incentives provided by constantly and predictably decreasing feed-in tariffs. Waiting another three or five years to develop better technologies means losing out on the high early tariffs that were paid on a per kWh basis. The result was that as little as possible was spend on newly developed technology to produce as much capacity as possible as soon as possible for the least amount of money – no matter how primitive or inefficient the actual product was. Under time pressure, new technology is a luxury, an additional cost and an additional risk. Companies may have been aware of that fact, but there is no use being aware of things going wrong, when you would go bankrupt long before you can prove your concept. In the end it meant that the whole branch of the industry was led into a dead end – as we now see with the bankruptcies of solar power manufacturers. But lets have a look at the current state of the art.

Rooftop Solar

We need to differentiate between two applications of solar power. The first is rooftop solar power. Rooftops are constrained. The owner of a house has limited space and will probably go to some length to optimize the profit he can gain. Usually, tariffs for rooftop solar are higher, but so are installation costs and budgets of house owners, compared to industrial investors. The constraints of space means that the owner will have an incentive to use more efficient panels up to a point, in order to maximize the amount of profit gained – so long as the budget is sufficient. This profit may not just be money, but also energy – depending on the reason for buying the panels. The rate of profit is often of secondary importance.

If you have 20 square meters of roof space, you may want to prefer 15% efficient panels over 10% efficient panels, if you can afford them, even if they are somewhat more expensive on a per watt basis – so long as the increase in marginal cost is below the increase in marginal profit.

The problem for the solar industry is, however, that it is running out of house owners both willing and able to buy solar panels at a time when it was expecting their numbers to be limited only by the producers ability to deliver more panels. To paraphrase Henry Ford: “Empty roofs don’t buy solar panels, people do.” This misunderstanding led the industry to overestimate demand and to invest in too much production capacity. This has now forced prices down to the point where producers go bankrupt.

It is worth pointing out that the decrease of prices is not mainly driven by a decrease of costs through innovation and improvement of processes (although especially the latter is perfectly unavoidable). The main driver of the prices is oversupply. Had demand held up to expectations (some 10-20GW in Germany, instead of maybe 7 GW), prices would have had no reason to fall in the first place. The decrease in cost is driven by Chinese manufacturers – by externalizing cost of environmental protection and wages, not by an improvement of process or technology.

That said, I am in favour of installing as much rooftop solar as possible in terms of the technology. There is little reason not to do it, if it wasn’t for the way feed-in tariffs are financed. Solar panels are a rich man’s game and so is house ownership in most countries. There is nothing wrong with that, but the money they earn is paid by all private consumers of electricity, most of whom are not rich at all and cannot profit from solar power for lack of money and real estate. (At least in Germany, industry is exempt from paying for financing the feed-in tariffs.)

To put it (only slightly hyperbolic) in contemporary terms, it’s the 99% paying for the subsidies of the 1%. If such tariffs were financed through taxes, mainly paid by those who could afford to buy solar panels, it would be a very different situation. If you are rich and you don’t want to give your “hard earned” money away to all the other rich guys, you better invest in solar power. This way things turn into a prisoners dilemma among the rich. While the most profitable outcome for all concerned would be to have no solar panels at all (nobody would pay the solar tax), every single participant can profit from this arrangement or avoid damage by installing solar panels. And experience tells us that this is going to be what people will do.

I will deliver an estimation of the potential energy gained this way further on.

Solar Power Plants

Industrial solar power plants are a very different matter. There is little incentive to use higher efficiency solar panels (or other technology) if this has an impact on the potential rate of profit to be gained. Buying a larger area of agricultural land is much cheaper than using more expensive technology. And since solar power is perceived as environmentally friendly per-se, there is usually no investigation on environmental impact to speak of, if you compare it to much lower impact investments such as power lines or rail roads.

A typical installation will use thin film solar panels. For reference, I will use the installation in Lieberose, Germany. It is a new, tightly packed installation and about as efficient as it gets in terms of static layouts. The cells have an efficiency of 10%. The park has a size of 1.62 mio square meters and delivers 52 mio kWh of electricity per year, which corresponds to an average power of 5.9MW.

As we can see, such a park will deliver about 4W per square meter in Germany. Allowing for losses due to long and medium-term energy storage (in order to make this power rating comparable to conventional power plants in terms of quality – as I discussed at length, again assuming that about one third of the eventual consumption is fed from stored energy), this must be reduced to about 2W per square meter.

Now you might be suspicious of this number. Doesn’t roof top solar provide 10W per square meter of collector area? It does. But collector area is less than half the story. In our example, 700,000 panels adding up to a total collector area of 500,000 square meters were installed on an area of 1,650,000 square meters. That’s because some distance between two panels is unavoidable. the rows of solar panels to avoid casting shadows on them. This is less of a problem closer to the equator, where the sun won’t be in the south at noon, but basically overhead. (This factor is already accounted for in the average annual insolation maps.) On the equator panels also don’t need to be installed at an angle, but can be put down flat. In any case, you need additional space between the rows for accessibility with vehicles during construction and maintenance.

The size of such installations is staggering. In order to build the equivalent of a 1 GW power station, an area of 500 square kilometers or 50,000 hectare must be occupied with solar panels as tightly packed as the ones in the picture. For comparison, the city of Berlin has a size of 891 square kilometers. A power plant of similar size would provide merely 2.5% of the total electricity demand of Germany.

It is easy to see that this is a problem. Even major advances in technology (such as panels with 20 or 30% efficiency at a fraction of the current cost of 10% efficient thin film) would still require the area of dozens of the largest cities in Germany to be completely paved with solar panels in order to provide the majority of electricity through solar power.

So, isn’t rooftop solar enough then? Well no. Rooftops are in fact only make up a fraction of the area of a city, on the order of 10-20% (especially when you have to consider whether or not the roof has a favourable southern inclination). Most is occupied by roads, walkways, parks, lakes and so on, areas in which solar panel cannot be installed. Of course, solar panels can be installed above streets, parking lots, railways etc., but even that has its limits. People do actually want to be able to have a clear view of the sky, even (and especially) in a city.

There is no doubt that rooftop solar can make a major contribution to the electricity needs of any country, but this is on the order of maybe 10%. (The total contribution of solar power to German electricity was just under 2% in 2010, despite the preponderance of solar farms. There is still a lot of potential left.)

Given the necessary size of solar farms, however, I despair of the current enthusiasm for this technology. Putting the grandiose plans of some environmentalists into practice would lead to environmental destruction on an unprecedented scale. And I won’t even discuss their monetary cost. (At the end of 2010 the amount of money that will be paid over a span of 20 years as feed-in tariffs for solar power has reached 100bn Euros in Germany, despite delivering a vanishing quantity of the total amount of electricity needed.)


6 thoughts on “The Trouble with Solar Power

  1. I just saw this entry mentioned via Charlie’s Diary. See the NREL technical report “Supply Curves for Rooftop Solar PV-Generated Electricity for the United States.” The base case estimates that up to 419 terawatt hours could be generated annually from residential rooftops and 400 terawatt hours from commercial rooftops, assuming 13.5% efficient panels. That would put solar a bit ahead of nuclear in terms of total electrical production in the US (22% of 2006 electrical demand, all in). That’s from rooftop space alone, without developing any farmland/wilderness or relying on future panel tech breakthroughs or huge new supplies of tellurium/indium/gallium.

    If you want to feel optimistic for a bit, SunPower already sells panels based on boring old silicon that achieve slightly above 20% efficiency. Other manufacturers are not too far behind. At 20% efficiency the same rooftop systems produce over 1200 terawatt hours of electricity, about 1/3 of US electricity demand in 2006.

    It requires equally daunting change to make good use of such vast amounts of intermittent energy, but the potential to generate energy is there if people can figure out what to do with it.

    • I hope it was clear that I didn’t mean to imply with the blog that solar power is useless and we shouldn’t do any of it. Far from it. We should do as much as reasonably possible – and just how much that is, is what I like to discuss. Unfortunately, there is a lot of misinformation about the actual usefulness that’s behind the numbers being circulated.

      Let’s start with the 1200TWh/a figure. In order to achieve this with solar power, you need an installed peak power capacity of 800-1000GW. That’s a lot more electric power than the entire US in consuming at any one time – but still a lot less electric energy than the US is consuming in any one year. Energy storage is absolutely necessary at this scale two reconcile this contradiction – and the necessary energy storage at this scale optimistically has efficiencies on the order of 33%. In the end, you need to deal with conversion losses on the order of 600TWh/a. A bit more (~700TWh) if you only end up using 1/3 of the generated power directly, a bit less (~500 TWh/a), if you can use 2/3 of generated power directly or from some very efficient shortterm storage (batteries, pumped storage, flywheels).

      The other number used in a misleading way is the efficiency of solar cells vs. the ready-made panels. This is going to sound like I’m a bean counter, but I’m not. I’m a photon counter. 😉 Seriously though, I once did this kind of calculation for photovoltaic farms and ended up thinking of myself as being an unsufferable pedant, sure to have accounted for far too many losses etc, … but to my surprise my result ended up being a bit too high!

      So let’s have a look:

      A 20% efficient cell sits behind two glass surfaces, which reflect about 5% of the incoming light. (Mainly the outer one. I assume the inner one to have anti-reflective coating – those coatings are too thin and fragile to withstand the weather. And you can’t coat the coating without incurring another round of reflection losses …) There are conductors on the cell, conducting the energy away from the cell, covering some 5% of the area. There are gaps between individual cells in a panel covering some 2-5% of empty space. There will be some amount of dust/dirt/snow/pollen etc. on the panels either some or most of the time taking 5% of the light (probably more).

      Solar panels also degrade as they age. Assuming a 10% degradation over 20 years (warrantee is typically 80% left after 20 years) and a 10 year average cell age, we must deduct another 5% for the average installed capacity. (That’s the most optimistic scenario. Probably house owners will wait 30 or 40 years to have them replaced, if in any way possible. It also assumes no bad surprises in degradation, as no currently sold panels have actually reached that age. The more realistic figure is 10-20%.) Finally, there is a frame around the panel to protect the cells, covering at least another 2-5% of the area.

      All told, only considering the panel itself, the effective efficiency is on the order of 0.95*0.95*0.98*0.95*0.95*0.95=75% (+/- 3%) of nominal efficiency. Thus, a panel using 20% efficient cells will have an effective efficiency of about 15%. You would need 26% efficient cells to make a 20% efficient panel. The optimistic case is thus quite a lot more optimistic than it seems to be.

      I also notice that the NREL assumes in its “optimistic” case, that there will be 25% more roof area available. Well, I certainly hope for the USA that this is not going to be the case. If anything, the USA should reduce its roof area, by reducing the number of houses and making urban living more attractive. The suburban livestyle is one of the main culprits for the huge energy demand of Americans compared with the rest of the much more densely populated rest of the world. (Australia being part of America for the purpose of this argument. ;))

      Because of that, the most optimistic scenario would have Americans reduce their roof area, increase panel efficiency (although 20% is too optimistic if you ask me) and basically cut their energy use in half (energy, not just electricity) – which would do roughly ten times more than anything PV could ever do under even the most optimistic of the NREL scenarios.

      Still, getting almost 20% from solar isn’t bad at all. If this is combined with large amounts of wind power, maybe about half the electricity (not half the energy!) could be provided with wind and solar power, without having too large an impact on the environment. Unfortunately, most of the energy is not consumed as electricity but as gasoline, natural gas etc. for traffic and heating (about 2-3 times as much as electricity). Wind and PV is absolutely worth doing!

      But it’s just not a silver bullet and it’s not cheap either – the two points currently being widely misrepresented to say the least.

  2. I think you’re double-counting some efficiency losses. NREL’s base-case 13.5% efficiency figure, and my speculative 20% efficiency figure, is at the panel level, not the cell level. It already accounts for glass reflection and fill factor. My optimistic case, unlike NREL’s optimistic case, is simply based on using more efficient panels, not assuming more roof space or better capacity factors. Look up the SunPower E20; it’s the most efficient commercially available panel targeting stationary applications, at just over 20% efficiency, and it’s built from silicon.

    According to EIA statistics, in California solar systems achieve an average 16.4% capacity factor. That’s based on real-world annualized production data and has pollen, dust, bird droppings, maintenance downtime, and every other real-world insult incorporated. The rule of thumb I use is 6 to 1: 6 peak watts of PV to generate as many joules in a year as 1 watt of nuclear.

    10% degradation over 20 years may also be too high, at least for silicon based panels. I have read a paper or two reporting that field data on PV degradation rates shows slower than estimated declines. Manufacturer claims are on the conservative side.

    PV panel prices have declined substantially from 5 years ago. 5 years ago I thought that nuclear was clearly the low-carbon future, with wind where it makes sense. I didn’t think PV made sense anywhere with a grid connection, but I feel differently today. The French and Finnish EPRs are both tremendously late and over budget. Fukushima has given the world fresh nuclear jitters. Solar efficiency is increasing and prices are going down. I think you’re too quick to credit Chinese exploitation of workers and the environment for decreasing solar prices: even producers outside China are constantly driving costs down. For several years that just improved margins in a seller’s market, but with oversupply buyers are suddenly seeing all those production cost savings (and then some) applied to prices in a short time. I expect the long term price trend to be downward with considerable oscillation overlaid on it.

    I freely admit: PV is still too expensive for me personally, certainly too expensive to land on every roof in the near future. Not only is the cost per peak watt higher than megawatt-scale wind, it has worse capacity factors than wind. Without cheap storage or revolutionary changes in energy use patterns, intermittent renewable penetration in general will be limited.

    On the reasons-to-be-optimistic side: There are many technical advances in the pipeline (in prototypes and pilot stage manufacturing, not merely theory or lab experiments) that credibly promise to continue lowering production costs. PV deployment is not bottlenecked by rare element availability or open land; silicon on rooftops can easily produce much more (intermittent) renewable electricity than people know what to do with. If utilities move to real-time pricing, the financial payback time for PV systems will be slashed in places with high summer air conditioning demand. If PV prices decline to the point where rooftop solar makes sense even without generous feed-in tariffs, PV also has the advantage of cutting out the middle man: electricity price is exactly equal to cost for a fully owned system on your roof. Electrification also increases efficiency in many applications: a joule of electricity in an EV can transport a passenger 3-4 times as far as a joule of liquid fuel in an ICE vehicle.

    • Then I guess we should agree on the volume of the liquid being 50% of the vessel it is contained in – and stop arguing over whether or not the glass is half full or half empty.

      However, as you said yourself in your previous comment, rooftop solar does not provide more energy than people know what to do with when averaged over the year. It’s just 30% of the electric energy (about 15% when considering storage losses). And that’s despite the fact the US has larger houses than most countries in the world (and thus larger roofs) and more people live in their own houses than in most other countries (which is part of the reason for the current economic troubles). Unfortunately, I don’t think that anybody made statistics comparing roof-area per person in different countries. 😉

      I absolutely agree that some sort of real time pricing scheme is unavoidable for any efficient system. (However, there must be a certain fuzziness in the system, otherwise you run into the same problem as in stock markets – with everyone switching their air conditioning on or off within one or two seconds, when prices cross the 10 cent limit would bring a lot of unnecessary trouble for the utilities. You’d likely need something like a 0.1 cent fudge factor to prevent this.)

      Finally, when it comes to electrification – I absolutely agree that it makes things much more efficient. But now you are extending the area we’re looking at. It’s no longer just the electricity but the rest of the energy as well.

      In countries with thermal-based electricity generation, about one third of the primary energy is used for electricity and two thirds for heating/vehicles etc. The conversion factor of (thermal) primary energy to electricity is about 40%. (For wind/solar the conversion is assumed to be 100%.) That’s to say that Germany needs 600TWh of electricity or about 2000 PJ, generated from 5000 PJ of primary energy. The total use of primary energy in Gremany, however, is 15000 PJ …

      Even if you manage to replace 5J of ordinary energy with 1J of electricity, the electricity use would still double. (Probably more.) Now, rooftop solar would drop from providing 15% (after storage losses) to about 8% of the total under almost perfect conditions in an optimistic scenario for the US.

      As to the EPR – each reactor has a rated power of 1650MW, a capacity factor of about 90% and a planned life-time of 40-50 years. Last year Germany installed 7500MW of PV with a capacity factor of 10% and a planned life-time of 20 years, for an annual power generation about half of one reactor. The legally guaranteed price of the power generated by those PV installations over the next 20 years amounts to about 30bn Euro. Despite cost overruns, the Finnish EPR is expected to cost 6bn Euro for twice the power generated. Even assuming equal lifetimes the reactor costs about one tenth. Cost is not the issue – there are others though.

      (I will discuss nuclear power in multiple postings at some other point – roughly when I’m reasonably confident that I won’t throw my laptop against the wall in the process.)

  3. I think reactor cost is actually a major issue. In an economy with rational state planning/support for these sorts of things (e.g. China) they can simply pick their technology from a sort of national technocrat’s viewpoint. In the USA — perhaps not to such a great degree in the EU — private investors don’t want to touch reactors without state loan guarantees, utilities and regulators are wary of utility-destroying or ratepayer-soaking cost overruns and delays, and there is not much in the way of state support for reactor construction. Even if principled or NIMBY protests and/or lawsuits don’t hold up the project, the sheer scale and complexity of a reactor increases the investment risk. High risk means fewer projects actually undertaken, so there is less historical data leading to more uncertainty leading to even larger reactor scale to try to better amortize costs…

    Even if the average levelized cost of nuclear energy over 50 years is a modest fraction of solar’s cost, it may well be easier to convince people to install more solar capacity. It’s rather like the difference between getting millions of individuals to buy cars vs. getting them to collectively support quality passenger rail. Even if the rail system offers lower long-term costs per passenger, it also requires much more investment and coordination up front, plus it’s easy for a dedicated minority to stop it. YMMV depending on the local political environment. I have some hopes that small modular reactors may solve some of these problems, but real optimism awaits real-world deployments of successful SMRs.

    When I said that rooftop solar offers more potential energy than we know what to do with, I meant that storage is expensive so annually generating even a few hundred TWh from solar is on the optimistic side for the foreseeable future. The practical limit to solar, until/unless there’s a storage breakthrough (or huge EV/PHEV deployment to serve as dispatchable demand), is going to be as a minority of total grid-tied generating capacity, and almost all electricity generated needs to be consumed immediately. But I think we should try to expand deployment up to that limit, so long as PV prices continue to decline, and we shouldn’t sacrifice any more wilderness or farmland to deploy solar. There is more than enough space on land already covered with buildings.

    • Ok, so you weren’t so much talking about cost, but financing. And that’s quite true.

      On the other hand, the cost of solar is more of an issue than is currently being admitted. Germans are already facing a bill of some 130bn Euro for the 27GW of PV currently installed (100bn Euro of that is pure subsidy). Those account for 3% the electricity use – but also eat into the amount of wind power that can be provided. (Germany has increased its wind power capacity by more than 25% since 2007, but power generation has been slightly below that level in 2011, despite favourable weather.)

      This kind of spending is going to end rather sooner than later, now that even the “but it’s creating jobs” argument is no longer tenable. In the long run, the high cost is a larger problem than the trouble of financing large investments. Unfortunately, Germany is currently set to replace its nuclear plants with coal power plants …

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s