Grid parity

by John Q on May 3, 2009

I’ve been following discussions of solar energy on-and-off for quite a while, and it has always seemed as if it would be quite a long time, even assuming an emissions trading scheme or carbon tax, before solar photovoltaics could be a cost-competitive source of electricity without special support such as capital subsidies or feed-in tariffs set above market prices.

But looking at the issue again today, I’m finding lots of claims that this “grid parity” will be achieved in the next few years, and even one company, First Solar, that claims to be already at grid parity with a 12 MW plant in Nevada completed last year. Obviously, Nevada is a particularly favorable location, and there is plenty of room for judgement in cost estimates. Still, looking at a lot of different reports, it seems clear that, with a carbon price of say $50/tonne (about 5 cents/kwh for black coal and 7 cents/kwh for brown coal), solar will be cost-competitive with coal for most places in Australia without any need for fundamental technical improvements.

The big question is whether the industry can ramp up from its current small scale (about 5 GW capacity produced last year) to meet global demands for growth and replace a significant part of the existing 4TW of mainly fossil-based capacity, with even more needed if we are to have big growth in electric vehicles. There’s no obvious constraint in the long run, since silicon is abundant and ubiquitous. But the price of the high-purity silicon crystals used in solar cells skyrocketed as demand rose in recent years (it used to be available cheaply as an offcut from the semiconductor industry, but solar demand has outstripped this source). Prices have fallen with the onset of the financial crisis, but will presumably rise again if demand recovers.

However, in the long run, and with no absolute resource constraints, costs should fall further as all elements of the manufacturing chain scale up. And over 20 years or so, with continued growth of 20-30 per cent a year, the necessary scale would be achieved If so (and assuming contributions from other renewable sources) the transition to a post-carbon economy could be faster and cheaper than most existing estimates suggest.

All this seems a bit too good to be true, so please feel free to point out problems I haven’t noticed.



ingrid robeyns 05.03.09 at 10:45 am

A question: is there in the literature/debate on solar energy any discussion on how this could be a resource for African countries, many of which have plenty of sunshine and space? It may not be possible in all poor African countries (those with civil wars etc.) but at least in some which have stable regimes. Perhaps start in South Africa? Or have these phantasies been dismissed in the literature/debate?


Finnsense 05.03.09 at 11:21 am


mpowell 05.03.09 at 11:31 am

Following on from #2, this is the big problem in solar. You don’t really see any discussion of it today in issues like grid parity because it is a much easier task to get solar cost effective enough to replace say 5% of power demand on the grid. Especially since solar is typically available during peak demand. Now, from what I understand the resources exist to ramp solar power substantially beyond that, but at some point, say 20%, you are talking about replacing the grid’s base power generation. That is where the problems of energy storage come into play. This is an enormous problem because you can’t have fluctuation in the base power the grid is providing. And from what I understand, it’s not really possible to just ramp coal sourced power up and down to accommodate. You have to come up with an energy storage scheme which is difficult from a feasibility perspective much less a cost one. There are a number of solutions being explored, but I don’t really have the expertise to comment on which ones have promise or how far they are from being ready.

As a side point, I think the price of silicon crystal peaked because very there was very little investment in independent production methods. I don’t think the recent drop in price is actually dependent on the economic contraction, but actually reflects an expansion in production capacity.


Moz 05.03.09 at 12:55 pm

There are problems with the rare earth elements needed to dope the silicon. Pure silicon is not a semiconductor, or in fact much use for anything. It’s only once you add small amounts of metal impurities that it becomes useful. Unfortunately they’re called rare earth elements for a reason. But there are “common element solar panels” available, and some of them are human copies of the dominant solar energy accumulators so the technology is well proven, it just needs to be commercialised and have the energy efficiency increased. Ideally we’d also get electricity directly rather than using photoelectric electron exchange to strip carbon off carbon dioxide and build carbohydrates to burn to make steam to turn a generator to make electricity. But yes, organic photoelectric technology is creeping along, the goal being paint-on cells.

Storage is actually less of an issue than it sounds. The main thing is that currently there’s not much demand for wholesale storage which means there’s no money in supplying it. When we get to peak solar input to the grid of 20% or more it’ll be time to start wondering how to store the energy, but at least in Australia the solar peak corresponds quite well with times of high demand so we could go a lot higher than 20% without problems. Europe and North America need that more than we do, because their peaks are later in the day and in winter. Both bad times for solar. But in any case, they can load shift to take advantage of much more solar than they have available.

You’ll know it’s time for major storage when you see “off peak power” being made available at midday instead of midnight, along with power companies running ad campaigns asking you to turn your home aircon on at 1pm instead of 5pm. We might also see a bigger push for extra daylight savings – a two hour shift instead of one hour would make a big difference to the timing of the evening residential power peak.

One other thing is direct solar thermal air conditioners. You put a heat collector on the roof (evacuated tubes, just like a solar hot water system) and that powers a cunningly designed compressor directly without having to generate electricity to power a motor. So the overall efficiency is higher (or can be), and the system is much simpler. But currently only built in Europe and available in bigger sizes (10kW or more). This is important because the major demand peaks in Australia and the US are aircon driven.


Joe Mansfield 05.03.09 at 12:59 pm

The major problem with PV today is the raw silicon cost but that is a currently a consequence of the economics of the primary consumer of Si wafer stock – the IC industry. They sell their end product at prices that range from $100k-$500k per square meter. Each “raw” 300mm diameter wafer costs something in the range of $100-$150, ie $1400-$2100 per square meter but much of that cost results from the wafer industry’s focus on the needs of their IC customers. If there was any significant PV industry the Si producers would be able to deliver raw material at significantly reduced prices.

Ultimately the major component in the cost of Si wafer stock is the energy required. The best numbers I’ve seen indicate that the energy required is about a fifth of the expected lifetime return for a typical PV cell in use in a high sun area (which would be most of Australia) so that should cancel out, eventually. In the meantime though the raw material cost of Si is a huge problem.


salacious 05.03.09 at 1:18 pm

Joe, is that energy figure only for production or is it for the entire lifecycle? Productions+transportation+installation+etc?


Moby Hick 05.03.09 at 2:20 pm

In the future, people will use “Stick it where the sun doesn’t shine” to refer to industrial developments that were foolishly placed away from areas with significant solar generation capacities.


Omega Centauri 05.03.09 at 3:06 pm

There are many issues. Some of the cheaper methods, for instance most of the thin film panels that acheive the lowest cost per unit power today utilize scarce materials that will limit their scalability. Solar grade silicon is produced at industrial scale today -especially in China, scrap from semiconductors is no longer a major source. The cost of solar silicon was very high a year ago, but has collapsed due to both major increments to production capacity, and a slowing of the rate of demand increase due to the GFC. Manufacturers are also getting better at getting more output per kilogram of silicon.

I’m not so sure about grid parity against baseline rates. But, the cost of an incremental increase in pak generating capcity can be quite high. Peaking is usually accomplished by natural gas plants, which have (relative to other plant types) relatively low capital requirements, and can respond faster (in say an hours warning). But, because the capital cost of such plants are not amortized across a large capacity factor (fraction of the peak design power that is used on average), the cost per KWhr is high. So it may be true that system wise adding an increment of PV power to the grid in areas like California with high summer afternoon peak demand on sunny days may already be cost effective.

We also have solar thermal, which uses mirrors to generate heat which is used to drive turbines. Solar thermal plants use no photovoltaics. It is also possible to design them to store the heat for a moderate length of time. This in principle at least means solar thermal plants could generate power into the evening. Of course the longer you want to keep operating at night the more costly storage capacity you need. Since air conditioning demand lags the solar day -it takes time for things to heat up, this provides a better match to demand than PV.

I really doubt that large scale power storage will ever be cheap enough to fully even out the temporal variations in supply/demand. Part of this imbalance must be made up via demand management, i.e. only running some power hungry processes during times of high power availability. Various sorts of smart grid, smart metering have been proposed to facilitate this. Users will see the time varying cost of using power, and schedule their usage accordingly.


Tim Worstall 05.03.09 at 3:48 pm

One problem that has been identified is that the dopants needed to make the cells work might be in short supply. George Monbiot for example has said that the world’s only Gallium mine is running out and what will we all do then?

Probably exactly what we’ve been doing or the past few decades actually, which is to get our gallium from the by products of the Bayer Process (bauxite to alumina that is, and only a minority of the current plants have the necessary capture circuits already installed) rather than try and get it from a complex ore in a forsaken part of the Congo.

Apologies for that aside but these weird metals are part of my day job.

“However, in the long run, and with no absolute resource constraints, costs should fall further as all elements of the manufacturing chain scale up. And over 20 years or so, with continued growth of 20-30 per cent a year, the necessary scale would be achieved If so (and assuming contributions from other renewable sources) the transition to a post-carbon economy could be faster and cheaper than most existing estimates suggest.”

On this I agree with you. Not that I have any technical expertise on solar cells themselves, just that’s the sort of gossip that’s coming back down into the weird metals industry.

However, what I do find most interesting about this is that this sort of continual reduction in the cost of solar was predicted. That solar would indeed become cheaper than coal and so we’d all pretty much switch to using it. And that therefore worrying overmuch about climate change shouldn’t be done. In essence, that “technology will save us”. The predicted date for this cross over, to where we would preferentially install solar rather than coal was predicted somewhere (from memory this is) around 2030/2040.

The prediction was by Bjorn Lomborg in his book the Skeptical Environmentalist and boy, didn’t he get stick for making it?


Finnsense 05.03.09 at 5:11 pm

I should just point out that the overwhelming tenor of this thread suggests that the future of solar will be on an industrial scale. In actual fact, I believe that it will be domestically generated on rooftops. Roughly half the cost of electricity is in transportation so solar becomes economical quite quickly once you take that into account. Installation costs are high at the moment but Nanosolar etc will bring that down with thin-film. The missing link is efficient storage but that problem is likely to be solved before 2020. The idea of storing TWs of energy in industrial plants is a bit daunting. Not so when it’s distributed in households.

Not sure if it’s ever going to take off here in Finland though, unless we manage to store 6 months worth of electricity somehow. Nuclear is definitely our future for better or worse.


Mark 05.03.09 at 6:07 pm


Omega Centauri 05.03.09 at 7:17 pm

Those who think rooftop solar is going to be the big savior need to recognize that BoS (Balance Of System) costs are about half the price of a current installation. Even if the photovoltaics were free, such systems are never going to be cheap. Only for larger scale systems will the economies of scale allow for the best economics. Renewables are unlikely to ever have much penetration at the level of individual homes, but mostly be delivered over the grid. Storage will always be difficult and expensive. Stored and reused power will be much more expensive than power utilized as it is produced. A few areas have pumped hydro available -or have nearby hydro plants that can be turned on/off as needed, but those will be the exception, not the norm. Giving every house enough batteries to bridge gaps is very expensive and environmentally destructive as well. Storage for minutes to hours may be feasible and economic, storage from season to season will not be. Eventually, unless we go for advanced cycle nuclear in a big way, we will have to face up to significant time varying availbility.


gmoke 05.03.09 at 7:18 pm


“There are a lot of amazing NGOs doing work to address the issue of rural household lighting but I think they are at best a fill-gap to an existing market gap. The mass market solution (LED + small rechargeable battery + 1 W solar panel) that will really make a difference will be Chinese and at a price that will encourage extremely fast adoption rates. This is evident from the introduction of LED flashlights in Mali that completely took over the market in less than six months.”

This kind of product (LED, battery, small solar) will be marketed within a year is my guess.

As for lifecycle costs of PV, see

The financing of solar electricity is the main sticking point to market penetration, IMHO. The mechanism that may break that open is the application of an energy services management or ESCO model to homeowners. Through purchase power agreements (PPA), a company will install and maintain PV on your property and sell you electricity at a competitive (or better) price. This is a model that Sun Edison has used successfully with commercial entities and that other companies are now taking to individuals.

As for pricing carbon, the auctions of carbon permits in the Regional Greenhouse Gas Initiative (RGGI) here in NE has been on the order of about $3 per ton but still may be counted a success with the last quarterly auction bringing in something over $100 million.

I have one room off-grid for reading lights and radio, plus battery charging, and did it for less than $200 nearly four years ago. The day I got my solar LEDs, the power went out for a couple of hours but I still had light. How should I account that?

Here’s a Solar Market Snapshot from a March 2009 conference on renewables:

PS: Solar IS Civil Defense and you can learn more about what I mean by that at


Zamfir 05.03.09 at 7:34 pm

Just to add another data point: here in the Netherlands, there is a (limited) subsidy for alternative electricity that is designed to make each of them cost-competitive with traditional sources, to generate some experience.
From the top of my head, the amount ( so on top of market prices) for land-based wind is 9 cents per kWh, for off-shore a bit more, for biofuel from garbage a bit less, but PV-solar needs something like 40 cts/kWh subsidy, so a complete order of magnitude more than the market price. I can’t personally judge how likely this number is to go down, just that it needs to go down a lot to make solar a serious player.

One thing that I have seen in Danish wind power that might affect solar when it becomes a larger part of the grid, is that the market price for electricity goes down a lot when the wind blows strongly, so you always produce most of your power when demand for your power is low, because all those other wind parks are also generating a lot. Basically the economic side of the storage issue.

What it means is that these sources have to be significantly cheaper on average then coal to be competitive on their own, and this effect becomes stronger the more power of the same kind you have installed. Denmark can partially compensate by exporting, but Australia won’t have that luxury.


lemuel pitkin 05.03.09 at 10:24 pm

This is very interesting, and has important implications for the debate between carbon taxes and carbon permits. In static terms the two schemes are equivalent but they behave differently depending on how responsive emissiosn turn out to be.

If you’re worried that it may be harder to decarbonize than we expect, tradable permits are good, because they guarantee you’ll hit the goal. On the other hand, if you suspect — as this post suggests — that it may turn out to be easier than we expect once the incentives are there, that’s a strong argument for taxes. Because with a carbon tax, improvements in renewables mean you get greater carbon reduction; with permits, they mean you get less revenue.


derrida derider 05.04.09 at 2:01 am

I know there has been considerable talk of massive solar farms in the Sahara with very-high-voltage DC lines under the Mediterranean to supply Europe. There are definite economies of scale in electricity transmission – once you get enough to make it worth sending over this sort of cable (you need a lot) transmission loss is not such an issue. Locating your power source close to users is then not so vital.

That also suggests power lines running along lines of longitude might be a way to match peak demands. That might be a good thing for some Arab states to put their money into as a hedge for when the oil runs out.


Alex 05.04.09 at 9:28 am

Tim: it was also predicted by a hell of a lot of other people, including essentially the entire green movement and anyone who knew anything at all about solar, and especially the people who have been struggling on working on it through the 80s and 90s whilst the politicians you support (and who support you, TCS boy) did their damndest to get in the way.


L 05.04.09 at 9:56 am

Re solar as a resource for Africa: it’s been discussed fairly loudly for Algeria, which however has loads of hydrocarbons anyway. Eg: .


mpowell 05.04.09 at 10:00 am

17: Yeah, that comment is pretty ironic. The timing could make a big difference here. If we had done a better job of encouraging solar development so that the transition away from carbon could be happening now, the difference could be a billion lives saved. If it takes long enough that the transition is more or less forced by the exhaustion of carbon based fuels, we bear the environmental risk of using all of those resources within a relatively narrow time window. It’s not as if the time constants involved in solar technology development are magical properties of nature. They are largely driven by the research investment available.


John Quiggin 05.04.09 at 10:53 am

One big problem with Lomborg’s suggested reliance on technology is that he wants this all to happen without any price incentives, which would be pretty silly if you took his “sceptical environmentalist’ self-description as being genuine. I’ll post more on this soon, I hope, but I long ago came to the conclusion that it’s a big mistake to treat Lomborg as someone arguing in good faith.


Ronald Brak 05.05.09 at 5:56 am

I’m afraid First Solar only used the cost of their solar panels and not the cost of installation when claiming they had achieved grid parity, which doesn’t really count. (I’m afraid I have no idea what their costs of installation were.) But point of use solar PV may be starting to obtain parity with retail electricity prices in some parts of Australia. According to Solarbuzz, the cheapest solar panels cost about $3.54 Australian a watt. If a business has a large flat roof that results in a low installation cost equal to the price of the solar panels, then a 1,000 watt system would cost about $7,100 installed. In a sunny place such as Moree that system would generate about 2,700 kilowatt-hours per year. I believe electricity costs about 25 cents per kilowatt-hour in rural NSW, so it would produce about $675 worth of electricity a year. If the system has a 25 year life span then investing in solar PV would provide a return of about 5.5%. This doesn’t seem like a lot, but government subsidies for solar power, feed in tariffs and the introduction of carbon trading in the near future make it much more attractive. If prices continue to drop as they have, then in a decade point of use solar might be the cheapest source of electricity across wide parts of Australia.


Ronald Brak 05.05.09 at 6:12 am

John, I think there is an error in your figures on how much a price on carbon will increase the cost of electricity. A $50 a ton price on carbon dioxide will increase the price of electricity from coal by about 5 cents a kilowatt-hour, but a $50 a ton price on carbon will only increase the price by about 1.3 cents a kilowatt-hour. (Coal plants burn about 260 grams of carbon and emit about 953 grams of CO2 to produce a kilowatt-hour of electricity.)


John Quiggin 05.05.09 at 11:28 am

Quite right, Ronald, I meant price on CO2, but (at least in Australia) carbon price is often used as a shorthand.

As regards First Solar, I suspect you’re right. And 5.5 per cent real is a perfectly adequate return- pre-GFC expectations of more are no longer relevant.


Zamfir 05.05.09 at 11:55 am

Ronald, those numbers look realistic, but also not too encouraging for most of the world. 25 years of service, without significant maintenance costs, seems like a big assumption, and 25ct/kWh is many times the price industries pay in most areas.

Still, it is good to know that there are (perhaps isolated) cases where solar panels do make business sense at this moment. That helps to get experience in the field.


Zamfir 05.05.09 at 11:58 am

Edit: Those are Australian dollars and cents, I realize. Makes the story a lot better looking, but still a bit too far off for most areas… Just for comparison, a 1500 W setup in the Netherlands produces in practice around 1400 kWh a year.


Ronald Brak 05.07.09 at 6:04 am

I’m sorry everyone, but I think my estimates for how much return a PV system in Moree would get might be a bit off. Firstly, I forgot to account for losses resutling from conversion from DC to AC which might be around 7%. And secondly, the figure that I have for insolation in Moree is so high that I suspect that it is for direct beam insolation, which would apply for a solar panel that moves through the day to track the sun rather than for a typical fixed flat panel. Panels that track the sun are more expensive as they have more parts, but generate about a third or so more electricity than a flat panel.

Comments on this entry are closed.