May 1st, 2012

Will campaigning deliver the transition to carbon neutrality? Or the governments that still can’t agree to a replacement treaty to Kyoto? Or a change of heart by energy-using corporations like WalMart? Or some miracle new technology like fusion? Hum.

We don’ t deserve this, but there’s now a partial answer. We now have definite predictions from the PV solar industry that grid parity will be reached within the next 5 years.

  • Mark M. Little, the global research director for GE :15c a kw/hr in 3-5 years.
  • Rob Gillette, the CEO of First Solar: 10-12c per kw/hr in California in 2014.
  • Thomas Dinwoodie, founder of SunPower Corp.: solar already competitive with gas and coal.
  • More on the same lines from McKinsey.

PV-rex hatching

According to MM Breyer and Gerlach, engineers for German company Q-cells, a grid parity wave will wash over much of the world by 2017. Pdf paper, apparently dated 2010. Breyer & Gerlach offer a systematic and properly sourced inside view, so I’ll mainly use that. Their work deserves a wider audience and scrutiny. A summary of their grid-parity predictions for the next 5 years at the end. They claim that residential PV is already competitive in Brazil, so we should see lots of panels falling off the backs of lorries on to favela rooftops in Rio very soon. [Update: their model is based on German costs as a benchmark for everybody else, an important qualification; see comments.]

Should we trust these insiders, or discount their claims for boosterism?

One, in revolutions insiders really do have a privileged perspective. Intel, TI, and ARM engineers have a better handle on whether Moore’s Law will hold up than even very knowledgeable outsiders. Second, there’s no obvious commercial advantage in predicting a rapid fall in prices, rather the reverse. Third, and most important, these predictions are entirely in line with past experience. Here’s Breyer & Gerlach’s experience curve for PV modules:

The percentage number is the learning rate, the reduction in unit cost for each doubling of cumulative production. In case you are worried about the 2002-2009 deviation from the trend due to shortages of silicon feedstock, just add the missing 2011 data point: prices have fallen more than 20% in the last year, and more or less returned to trend.

The historic growth rate in volume is 45% per annum, corresponding to a doubling in 2 years and an annual reduction in unit price of 10%. Balance-of-system costs have historically tracked PV module ones, though they vary a lot between large- and small-scale installations.

One quibble with Breyer and Gerlach is that the cost-of-capital discount rate they plug into the formula for a levelised cost of electricity (LCOE) is only 6.5%. This may be fair enough for German solar investors in a benign environment. These Australian analysts use 8.5%, which looks more plausible for other advanced countries. If you want to work through the sensitivity of the LCOE of different electricity sources to the discount rate, see Chapter 5 in this IEA study. Raising the discount rate would at first sight only make a year’s difference to the arrival of grid parity, which will in any case be staggered by many other noisy factors – subsidy régime if any, regulatory hurdles, relative insolation.

Interestingly, most insiders can’t bring themselves to believe that this explosive rate will continue:

Annual growth trend for the last 15 years has been 45% (Figure 1). Nevertheless, consensus of scientific researchers and financial analysts is a growth rate of about 30%.[1,73,28,29,31,74-76] However, it has been very common to underestimate both near and long term growth rates of PV.

Their Figure 1 actually shows that the growth rate of 45% holds all the way back to the invention of the solar cell in 1955, so it’s a 57-year trend not a 15-year one. Are there any reasons apart from herd caution to think the trend will not continue? Breyer and Gerlach:

Leading PV experts estimate the achievable longterm cost potential for PV module technology and respective industry below 0.30 USD/Wp [51] and Pietzcker et al expect PV system floor cost, i.e. long-term cost level, of 0.60 USD/Wp [52].

The remaining doubts must lie on the demand side, for example with the recent cuts in subsidies in Spain and Germany. But German installations have not stalled, and the number of markets is increasing, making for a steadily more regular pattern. Any near-term supply glut would drive prices below trend. In fact, as we reach the discontinuity of grid parity, you would a priori expect demand to accelerate.

So the egg has hatched and PV-rex is upon us. Apart from offering a hearty welcome to the toothy little carbon-neutral darling, what can we expect from it? I have no expertise to add on when grid parity arrives. But it’s possible to make some predictions on what will happen after it does.

1. No new or replacement coal power plants will be built in OECD countries, and a rapidly declining number in Asia. The economics are marginal and the downside regulatory risks enormous, over the long planning horizon of a new plant. Coal is unsuited to be the balancing power for a renewable-heavy generating portfolio. Carbon capture won’t be ready in time to offer resistance to PV-rex.

The phasing out of old, and fully amortised, coal generating plants depends on the regulatory environment. But with the writing on the wall, and coal companies and utilities looking at exit strategies, it’s hard to see a successful resistance to internalising health and climate costs. There’s a good chance PV-rex will eat up coal within 10 years.

2. Breyer and Gerlach say that grid parity will arrive in some countries first on residential and commercial rooftops, in others as large utility-scale power plants, depending on the retail price of electricity. We should think of two categories of investors.

Professionals have spreadsheets, LCOE formulae, and access to long-term capital: utilities, corporations like WalMart, universities. Their behaviour should reflect quite precise cost tipping points. Utilities in particular will want to invest, or buy in, up the point where the system costs of covering solar variability (ignored by the LCOE formula) become large. This will depend on the mix of other sources (wind, hydro, despatchable gas, geothermal) installed or available, and won’t be uniform. Let’s make a pure guess and say the limit will be a third of total demand.
The second group are Joe and Jane Average, without benefit of spreadsheets. As the price carries on dropping, at some uncertain point (50c/w a module, $1/w all-in?) residential solar becomes a no-brainer and demand explodes. NIMBY planning restrictions will collapse, as everybody wants a panel.
Joe and Jane don’t care about system balancing costs. It will pay them to install panels up to the peak daytime load in their house, probably air-conditioning. The utilities are stuck with the variability costs of all these extra panels, whether or not they are connected to the grid. They still have a duty to “keep the lights on”, but daytime demand and revenue has cratered. Utilities will have to raise residential tariffs, reinforcing the attraction of home installation.
Interesting conflicts are in store, which homeowners will presumably win through the ballot box.

4. Will PV-rex go on to eat up the oil industry? Sadly, not so fast. The hurdle here is the capital cost of car and truck powertrain batteries; per mile, electric cars already run a lot cheaper than gasoline ones. Cheap solar electricity doesn’t change this equation much. However, it will steadily widen the running cost gap, and enable probable improvements in batteries to drive a massive shift to electric vehicles. Solar electricity will only be one of several killers of Big Oil, like the conspirators against Caesar.

5. Similarly, he won’t quite destroy the gas industry by himself. For now, gas remains indispensable to electricity generation as the main despatchable source to fill gaps in supply created from the variability of wind and solar. For an indication of the magnitudes here, see these simulations (page 40) for two historical weeks in Minnesota – the scenario is for a fully renewable supply, so the gap-filling is marked as storage (eyeballing, less than 10%). The full replacement of gas from electricity generation does indeed require storage or hydro or geothermal. This looks like something that can wait a decade for better technologies.

6. Cheap solar power will change geopolitics, mainly in good ways. The solar resource isn’t concentrated in a few nasty Middle Eastern autocracies and is very widely spread across the globe. One entire pillar of US foreign policy, ensuring access to Middle Eastern oil, will crumble into insignificance. PV-rex will help poor and sunny African countries. It’s hard to see how any replacement monopoly in the PV supply chain is feasible, or political leverage with boycotts. Within countries, PV doesn’t lend itself to kleptocratic rent-seeking either. Any landowner can put up panels.

7. The other benefit is of course the end to the power and influence of the oil companies. They won’t be replaced by any Seven Sisters of PV; this is a manufacturing rather than a resource extraction industry, and there are few barriers to entry.

Kochs, BP, Chavez, Ahmadinejad, King Abdullah, Vladimir Putin: go, Towser! Tasty nibbles!

*     *     *     *     *     *     *

Breyer & Gerlach’s grid parity predictions, 2010-2017

Selected data from Table 3 in their appendix, which gives the assumptions.
R = residential, I =industrial installations. The date given is the first grid parity event; the other follows within a year or two. (IMHO, the former is disruptive and likely to be followed by policy changes which will anyway invalidate the assumptions of the predictive model).
China, India and the USA are divided into large regions.

Burkina Faso IR, Cambodia I, Chad IR, Cyprus IR, Dominican Republic I, Fiji I, Gambia I, Grenada IR, Guyana IR, Haiti I, Jamaica IR, Liberia IR, Madagascar, Mali I, Senegal I, Seychelles I, Uganda R

Afghanistan IR, Brazil R, Cuba R, Denmark R, Malta R, Palestine I, Portugal R, Spain R

Austria R, Belize I, Burundi I, Central African Republic R, Chile R, China West I, Cote d’Ivoire R, El Salvador I, Guatemala I, Germany R, Italy I, Japan R, Philippines IR, Suriname IR, Uruguay R, Togo I

Belgium R, Benin R, Cameroon I, Gambia R, Ghana I, Guinea I, Hungary R, Israel R, Kenya R, Luxembourg R, Mauritius I, Mexico I, Morocco IR, Namibia R, Netherlands R, New Zealand R, Nicaragua R, Pakistan I, Panama IR, Peru R, Puerto Rico R, Rwanda I, Turkey IR

Australia R, Burma I, China East I, Croatia R, Greece R, Honduras I, India West I, Ireland R, Jordan R, Lebanon R, Malawi I, Namibia I, Nicaragua I, Niger R, Slovakia IR, Slovenia R, Sri Lanka I, Sweden R, Syria R

Argentina R, Bangladesh R, Brunei IR, China Central I, Colombia I, Congo DR I, Czech Republic R, Finland R, France R, India East I, Israel I, Mozambique R, Norway R, Poland R, Romania R, South Korea R, Sudan R, Switzerland R, Tanzania I, United Kingdom R, United States East R, Vietnam I

Bulgaria R, Costa Rica I, Ecuador R, Gabon R, India East R, Latvia R, Nigeria I, Thailand I, Tunisia IR, United States NW R

Bolivia R, Botswana R, Estonia R, Honduras R, Indonesia I, Lao PDR I, Lithuania R, Malaysia I, United States SW IR

49 Responses to “PV-rex hatches”

  1. matt wilbert says:

    You don’t get to point 6 (irrelevance of Middle East) until you get past point 4 (this doesn’t have a big effect on the attractiveness of electric cars).

  2. Rob in CT says:

    I think it’s plausible to see alternatives kill coal (at least in developed countries) over the next couple of decades. I see less reason to think they’ll kill oil. Oil is super duper useful, and not just as fuel (plastics).

    Anyway… I’ve got a PV array going up on my roof this summer.

    • Rob in CT says:

      I should note, however, that the $ only works on our project because of the subsidies available. Without those, it wouldn’t work. Partly that’s because of where we live and partly it’s the orientation of the roof (good, but not ideal), but it’s also partly the basic cost of the materials & labor.

      The continuation of the trend, however, would make it more and more of a win, even if the subsidies phase out (as they should at some point).

      • Dan Staley says:

        the $ only works on our project because of the subsidies available.

        Here in the very sunny states, we have an expanding business model that leases roofs and sells the power back to the owner. I expect to get a good deal at the big RE conference coming here in a couple weeks, and hope to close on a good leased system during the conf. This business model is one of the key points of my presentation, and I expect it to find its way up to NE in a couple years.

    • Dennis says:

      Long ago, on a campus far away from my current campus I took an honors seminar with our college’s Dean running the seminar. The seminar topic was resource depletion and allocation. The Dean (a chemist before he turned to the dark side) said to us, “One day within your lives, we are going to realize petroleum is much too valuable to burn.”

      • Dan Staley says:

        One day within your lives, we are going to realize petroleum is much too valuable to burn

        I imagine we might be able to replace some plastics with hemp, but replacing industrial agriculture?? Yes, our kids will see this play out in force.

  3. Philip says:

    There are people who write papers, and people actually in the field. Grid parity, at least for residential home installations, is quite far off.

    I work in the MA solar installation market. It costs $25K to put a typical 5KW system on a roof (I can direct you to a spreadsheet that shows you every install in MA over the last several years with the price of the system to support that if you’d like.)

    Of that $25K, only $6K or so is for the panels and equipment. Maybe $7K. Most of the cost is labor and overhead. Great if that could be reduced, but it doesn’t tend to follow Moore’s law (except to the extent that solar panel efficiencies go up and you get more out of every panel installed, but that has been a decreasing part of the solar landscape over the last decade).

    You get around 5 to 7 MWh of electricity per year out of that 5KW system here in Massachusetts. At retail electricity rates here – 15 cents per KWh – that’s work $750 to $1050 in electricity value per year. And that’s with two significant implicit subsidies: first, net metering is required because the power you need (beginning and end of the day) doesn’t match when your panels generate most of the electricity (middle of the day) and the utility nevertheless has to buy your electricity whenever you generate it; second, the utility has to ‘buy’ it at you location for the retail price – even though that assumes distribution has already been paid for, when in fact it still has to be re-distributed.

    Let’s ignore the two implicit subsidies. Would you, based solely on economic arguments, spend $25K to save $1000 per year? How about if we include those two implicit subsidies and value the electricity you generate at more like 5 cents a KWh (the actual generation you are likely displacing; although the electricity you generate is probably worth less given more peaking power is necessary to make up for your volatility) – then your panel’s generated electricity is worth $250 to $350. How great an investment does that sound like?

    Note that including external factors like the implicit cost of CO2 emissions adds some to the solar benefit – but anyone talking about grid parity in the context of developing Asian economies isn’t being serious if they say CO2 costs are a substantial part of the grid parity calculation in those countries.

    The only reason solar is growing in MA is because of subsidies. You get a 30% tac credit from the federal government, a $2,000 rebate and a $1000 tax rebate from the state government. Net cost of system once these are collected: $14 to $15K. ON TOP OF THAT you get an SREC credit for every MWh your panels generate (due to state mandated payments from the utility companies) which are priced at auction (current price at last auction: $540 per credit) but with a $300 floor price. That’s another $1.5K to $3.8K annual cash stream subsidy. Note this cash stream subsidy is many multiples the size of the value of the electricity generated. With subsidies, a $15K net investment in a system (gross cost $25K) generates an electricity value plus SREC cash stream with a return in the 15%-30% range.

    Absent subsidies, solar goes nowehere in MA – except perhaps (as one salesman told me) as a ‘prius on the roof’ reflecting your lifestyle choice. in economic terms it makes no sense.

    Get rid of explicit subsidies and you need installation prices (for a typical 5 kW system) well under $10K to justify the $1K per year (at best) value of the electricity generated. Get rid of implicit subsidies, and you need installation prices well under $5K to justify the $350 or so value of the electricity generated.

    And remember, if the panel prices go to zero, you’re still nowehere near the needed price levels – you’re still near $20K in labor and overhead costs per system in MA.

    Aks those paper-writing dudes why states like MA have to offer 50 cent per kwh annual cash payments (three times the value of the electricity generated) ON TOP OF 30%+ federal tax rebates to support solar activity, if we’re so close to grid parity. They’ll wave their hands and say a bunch of long sentences, but in the end the facts on the ground prove them wrong.

    • David Carroll says:

      Those installation costs seem surprisingly high — but not outrageously so. That’s the cost of more than 8 residential fuse-box-to-circuit-breaker conversions (at $2400 in the route 128 area) for instance. Some other residential remodeling costs are here Remodeling cost vs value report, 2011-2012.. Your installation costs are higher than a mid-range bathroom remodel at $17,460 and about the same as a roofing replacement at $22,827. Still, even if installation costs were to fall to 10k (a credible floor given how expensive any home remodeling is in MA due purely to labor) and PV cell costs fall to 3K for the system you describe, it would still require 13 years to pay back, which doesn’t seem great (but is actually faster than a window replacement which can take a long time to pay back). But that payback time period isn’t awful and at that cost level there may be a case for some government support merely to hedge against future energy price spikes.

      On the other hand, according to this Scientific American article, “Fortunately, installation costs have also dropped at a similar pace to module costs.” Maybe that article was written with larger, standardized installations on big-box stores, malls and office buildings in mind? Or possibly they are thinking that competition will force PV installation to be done by lower-skilled workers moving forward, e.g. roofer-level workers rather than electricians?

      • Jamie says:

        The interesting thing to me is what happens on normal replacement cycles. Almost nobody will replace a perfectly fine fridge, even if it sucks power. But when the repair/replace cycle comes up, people look at energy savings as well as features.

        When replacing a shingle or tar roof with a PV installation is a normal option to think about, I suspect the cost equations will look a bit different.

    • Breyer and Gerlach work in the business. That’s why I considered the hypothesis of bias, but impractical ignorance doesn’t look like a starter. As I read their Table 3, the grid parity projections are simply based on current electricity prices, so incorporate current policy if any on emissions – usually none – but don’t assume any renewable-friendly changes.

      My Australians (academics but not fools) give current best-practice BOS costs at $1.2/w or 40% of the system, maybe 70% for a small residential one (section 5.2.2). Your 3 or 4 : 1 ratio in Massachusetts looks out of line. Why is it so high? Admittedly the US market is fragmented compared to the reference German one (much bigger and with a uniform and favourable regulatory environment), but prices elsewhere are bound to converge on Germany as volumes go up. (German electricians don’t come cheap either.)

      I suggest you look seriously at your cost drivers if you plan to stay in business. Most PV companies won’t, any more than Silicon Valley startups in 1980. Good luck, but you are in for a rough ride.

      Permitting costs will have to give too. I read somewhere that these can be 50c/w in the US. That’s unsustainable when the rest of the system costs $1/w. Either rules are drastically relaxed, or they will just be ignored.

      • James Wimberley says:

        PS: The German Solar Industry Association gives the latest average system price before tax of rooftop installations up to 100kw – so covering commercial as well as residential uses – at €1.969 per watt or $2.6/w. Still looking for a residential-only figure.

    • NCG says:

      What about those arrangements where the homeowner leases the panels, and the company takes the risk? If enough people did that (and I suppose the public might be backstopping it somewhere, early on), could it bring down the costs more quickly?

      Psychologically that would be a much easier sell, assuming the company was reliable. (I assume you can’t look on Yelp for that yet. But I’d be interested to know how to find out.)

      And would the company only want to lease to people who used relatively little energy to begin with?

      • Rob in CT says:

        My mother is about to do that. She’s putting down a chunk of money up-front, but this results in no monthly payment. The company takes care of the system and insures it. 20 years later, they will either remove the system (at their cost) or re-up the lease. It’s a pretty good deal… based on the proposal it’s better than the deal I’m doing (a straight-up purchase).

        • NCG says:

          If you get a chance, I’d love to know how it goes for both of you. For most people, I think this sort of thing is a big risk, and there are many ways it can go south. Here in LA, at least in the paper, people are reporting lots of issues with the local utility having seriously cr*ppy billing procedures and so forth. I expect though that these issues will be worked out. I hope so anyhow. It seems a no-brainer out here to do it. Someday, I hope to have a solar hat that will, I don’t know, fan me as I walk down the street.

    • Rob in CT says:

      Just to add data to this, since I’m one state away in CT:

      We are putting in a 9.7kW system. We got three bids. The cost (pre-subsidies) was $36,500 – $40k.

      About half is labor. If I recall correctly, the panels alone are about 1/3 of the total.

      Anyway, the CT subsidy + federal tax rebate will end up basically halving the cost to us.

    • Warren Terra says:

      A very helpful comment, but it seems focused on distributed use of solar, not centralized solar electricity. Even when the numbers add up for eventual amortization, the high capital costs of residential solar were always going to be a problem, especially for homeowners expecting to move within a decade; and aren’t there also major problems with adapting our electrical grid to having many small inputs?

      But while your points are well taken, might not the developments discussed in the post herald large-scale solar facilities operated by the utilities?

    • Josh G. says:

      Would installation costs be cheaper if solar panels were integrated into new residential construction, rather than being retrofitted into existing houses after the fact?

    • Dan Staley says:

      I see the repetition of “subsidies” several times @Philip. Pull fossil subsidies decades ago and make fossil pay full cost of generation and you are good. The market is distorted.

      Nevertheless, several commenters pointed out ways around labor, permitting, etc. including SmartGrid, which here in Colo the major traditional provider was insistent on sabotaging.

  4. James Wimberley says:

    A lot of residential installations in Germany are planned for more than one house, at a village or cooperative level. That drives down prices and paperwork costs a lot.

    One thing I didn’t say about experience curves. At first sight, it’s very odd that you should see a roughly constant trend in a technology as it moves through very different technical and economic challenges, as with Moore’s Law. The common factor is surely human resources: a faster-growing industry puts more engineers and managers to work solving its problems. The expectations come to be self-fulfilling.

    The problems come to include regulatory and political ones, and labour training. There’s obviously no longer a permit difficulty in Germany over residential rooftop installations, or a technical one over connections to household cabling, or a shortage of qualified labour.

  5. Foster Boondoggle says:

    The OP writes: “Joe and Jane don’t care about system balancing costs. It will pay them to install panels up to the peak daytime load in their house, probably air-conditioning. The utilities are stuck with the variability costs of all these extra panels, whether or not they are connected to the grid. They still have a duty to “keep the lights on”, but daytime demand and revenue has cratered. Utilities will have to raise residential tariffs, reinforcing the attraction of home installation.”

    I’m not sure the inference on utilities’ costs is right. Here in sunny CA, peak loads happen in the summer between noon and 7 PM, when everyone more than 10 miles from the Pacific turns on their A/C. That’s also when the PV systems are generating maximum power. So PV actually provides load-leveling for the utilities, allowing them to avoid buying electricity from the highest cost producers in the stack. Their average cost per watt goes down, not up.

    • In my scenario, the daytime load doesn’t level, it collapses because everybody has roof panels over their house or shop or factory. In theory, and assuming a smart grid, the utilities can avoid this by lowering the daytime price steeply. Either way, they will be hurting.

      • Foster Boondoggle says:

        We’re a long way from that situation, no? By the time we get there, perhaps storage technology will have improved enough to solve that problem — molten salt, compressed air, artificial lakes or whatever… In any case, I would expect that’s far enough in the future that the present value of that downside is pretty small.

        • James Wimberley says:

          We are not “a long way from that situation”: 5 years at the outside, probably too soon for any other big change in technology to cut in. My point was the institutional asymmetry – the downside is all on the side of the utilities. I’m not weeping for them.

          • NCG says:

            This is not my area, clearly, but at least in theory, aren’t we all going to be driving green(ish) electric cars, too? Might that might bump up demand and maybe cushion the utilities a bit? (I am one of those late-adopters, in case it wasn’t obvious. And I don’t like to give up something that works, in case the new thing fails.) Maybe a home system couldn’t power all those cars. (Or maybe they could.)

            And maybe if our utilities used natural gas, we could still save the planet from frying. Which would be awesome.

          • Brett Bellmore says:

            Aren’t you implicitly relying on them to provide leveling “storage”? You better weep for them, or at least care if you’re going to drive them into bankruptcy before you don’t need them anymore. The only reason solar looks even vaguely near achieving payback is the huge subsidies, and the expenses being dumped on the utilities are a big part of that.

            The problem here if you’re using the utilities for load leveling, instead of storing the energy on site, you’re requiring them to maintain, ‘hot’, a KW of generation capacity for every KW you can produce. Maybe you ought to figure the cost of that into your solar expenses, to be complete?

  6. Kaleberg says:

    Bloomberg had an article last month on how PV electricity has changed the marginal electrical power market in Germany. Germany has the largest installed solar electrical power base, and this has dramatically cut the marginal dispatch price of daytime electricity. Basically, it almost always makes sense to buy more solar at zero euros per KWH. In fact, the premium prices are for night time power. Sure, the installation subsidies have been cut, but no one is taking down their solar panels.

  7. Keith Humphreys says:

    James, you say “there’s no obvious commercial advantage in predicting a rapid fall in prices, rather the reverse” which is true for many industries but not in the case of a heavily subsidized industry that needs to persuade its public sector patrons to keep the subsidies coming because independence is just ’round the corner.

    • That can work both ways. Optimistic forecasts encourage the patrons (as in Germany) to plan to cut the subsidies; and makes prospective customers think about delaying purchases. I expect the Brazilian government is being told by its advisers that it’s not worth the trouble to introduce a PV feed-in tariff for only a couple of years. The payoffs from trying to game expectations with false predictions are very dodgy. The risks from making an honest informed prediction are low, as expectations will anyway converge on reality sooner or later. Occam’s razor suggests that the industry spokesmen are making honest predictions.
      The clincher, however, is that these predictions are in line with a 57-year observed past trend. It would need some definite reason (we are running out of indium or something) to think that the trend will not continue, not generic scepticism. Some things are too good to be false.

      • dave schutz says:

        “…makes prospective customers think about delaying purchases….” Yes, absolutely. We are rebuilding our house after a fire, and our strategy is to run conduit from the attic to the basement circuit breaker box area so that in three years we can put it in without opening up our walls again.

        On the ‘installation’ front – Westinghouse has been working on kits for standardizing installation and cutting costs quite a bit there, which should lower the proportion of costs which go to installers.

        For us, in a hot-summer area, daytime is when our air conditioning consumption is highest, so solar intrinsically makes sense, whether or not we sell a lot back to the power company.

  8. Philip says:

    Here’s a link to the Mass CEC’s list of solar installations with gross cost (CEC administer’s the state’s subsidy programs so anyone installing in the state has to send in this info to them to apply for subsidies):
    (scroll down the sheet and click on the installers, costs etc. link to download the spreadsheet).

    The recent prices are about $5 per watt. Note that there is a big fixed vs variable cost here, so smaller installations cost more per watt and large ones cost less per watt – and all of the difference is in the labor side as panel prices per watt are steady across installation sizes at this level. The typical home is about 5KW – the 9.7KW system referred to in CT above is pretty big (must have a big south-facing roof) which explains part of the reason the cost is lower and the percent labor is lower, although his figures are a bit better than I’d expect in MA. Could have to do with the details of the roof layout, the type of panels (mono vs. poly), how complicated the inverter set-up (micro’s or a single big one?) which would be driven in part by shading pattern, etc. My figures are averages.

    Interesting comments about leasing model above – in MA it has gone from about 10% share a couple of years ago to 80% share now – a huge jump. (You can see it in the spreadsheet above by analyzing the third party owner column.) And yet the leases are not good deals for consumers. Consider the two options. Local installers (all of whom lack the financing ability to offer a lease) will sell you a system for $25K, you apply for rebates etc to get net cost to $14K, and then you collect $2K to $5K in cash per year – a cash flow with an IRR in the 20-ish% range. OR, national installationc ompanies have entered MA focused on leasing, and here are the typical lease terms in MA: you put $500 or $1K down (and pass a 700 credit score requirement). You get NO REDUCTION (or a token reduction) in the initial price you pay the installation company for the panel-generated electricity the panels they install give you – they charge the same as the utility (usually their sales pitches show a $1 decline in your utility bill – I assume for marketing purposes). The benefit they pitch to you is that the electiricity price they charge you will go up a guaranteed 3% a year, while the utility price could go up (they say) 6% a year. We can all argue about where we think electricity prices are going (recently with low NG: not up) but accepting this on its face yields a few hundred dollar NPV value to the consumer. The rest of the benefit – the tax credits, the rebates, the SREC income (and the electricity payments from the consumer) go to the installation/financing company for a tidy 20-ish% return, mostly driven by those subsidies. Great business if you can get it. Consumers are taking leases like hotcakes in spite of the very large value difference between offer types. A side topic but one I find quite interesting. Wonder how the state will look back in five years once the current subsidy program runs out and finds (if current market dynamics continue) that most of the profit has gone to out of state investors with leasing consumers possibly having HIGHER utility bills than there neighbors, or perhaps bills with little or no discount vs. their neighbors.

    The solar/grid parity topic gets my blood boiling because of the amount of poor analysis out there. I consider myself very well versed in the residential installation market in MA, but not in the utility-scale market. When I read analysis that I know from my residential experience gets it wrong or ignores huge problems, it suggests potential problems across all of their analyses. Personally I see more likelihood that utility-scale plants will be the first to reach true (unsubsidized) grid parity, but I don’t believe the analysis I see because the weaknesses in their arguments. Some examples:

    1) Labor is a huge part of the cost of solar. Why do they always show the experience/cost curve for panels, and never for labor? Where is the detailed analysis of where the labor portion of costs has been and is going? Throwing around the cost curve for the panels themselves is becoming increasingly less important and yet it is the single piece of hard evidence put forward to ‘prove’ that solar costs are very close to grid parity. Maybe they can show evidence (which would contradict my peronal experience and research) that labor costs are dropping and give a convincing argument (as they do for panels) as to why. I haven’t seen it (beyond the panel efficiency-generated reductions in labor costs per watt, which are petering out for the silicon panels used in home installations). Here’s a tip: if they don’t get into the details of labor costs, if they go on and on about panel prices but barely talk about labor costs or analyze them only cursorily, they are ignoring the central issue, and should not be taken seriously.

    2) On the residential install side, analyzing grid parity by comparing to retail electricity prices is ridiculous and yet the norm. Why is it ridiculous? because comparing to retail electricity prices includes big indirect subsidies. The value generated by a residential solar installation is much less than the retail electricity rate prevalent in the area times the total number of kWh generated over a year. Because the value of electricity is closely tied to WHEN and WHERE is it generated, and the retail electricity rate you get (a) time-averages the time component and (b) gives you all the benefit of it being delivered to your door inspite of the fact that when your panels generate it and put it into the grid, it now has to be re-distributed. One clue to the size of the disparity between retail electricity prices and the actual value of solar is the PPA terms worked out for large installations. When you put upa residential installation and net metering is in force, the PPA (purchase power agreement) terms between you and the utility are implicity set – they will buy whatever power you generate whenever you generate it at the average retail electricity price. For a large installation that has to negotiate a PPA with, say a munipality running its own grid (14% of the grid in MA is this way; and these guys are not some big hard to negotiate with utility), the terms are NEVER this generous because the grid does not derive that kind of benefit from solar. PPAs of this type usually value solar-generated electricity a factor of 2X or more lower than retail rates even beyond the differential distribution costs. Residential solar avoids facing these issues because it is such a small percentage of the grid (less than 0.5% of MA electricity now; no more than 2% in 5 years after current subsidy program runs its course) that the implicit subsidies are rounding errors.
    So: if you read an analysis of residential solar approaching grid parity that uses retail electricity prices, you are reading an analysis that can not be taken seriously, and assume at least another factor of two or three in solar cost reduction needs to occur to reach parity.

    3) lack of close analysis of the residential vs. utility grid parity situation. right now, to a rough approximation (and ignoring the PPA negotiation advantages for residential installations) home installations and utility-scale installations could be viewed as in a similar position with respect to grid parity using the normal assumptions in these analyses. Utility-scale installations cost about $2-$2.50 per watt, all in, and compete with 5 to 6 cent per kWh baseload power sources (ignoring the solar volatility issue, which is a big one and means solar is probably must reach 3 or 4 cents per kw to make up for its volatility). Residential installations cost twice as much per watt (most of that difference due to labor), but ‘compete’ with retail electricity rates which are typically twice the baseload generation price. Residential has the current advantage because of the PPA negotiation advantages I mention above, but it is artificial due to implicit subsidies. Utility-scale has the advantage that continued panel price drops will have a bigger impact on the overall price per watt. That’s a BIG difference. It is unlikely the two approaches (which have quite different cost structures) will both be substantial portions of the grid in 50 years- one will win out. I never read an analysis of that, I always read generic solar grid parity analysis that suggest that grid parity for the two is a similar story and we’ll have large amounts of both in a few years. Unlikely.

    I’m a believer in solar. I expect in 30 or 50 years we will have a large solar installed base. But I’m also a businessman, and whether we reach grid parity in 3 years or 15 years makes a BIG difference to me, my investors, my customers, etc. The weak analysis out there, and the constant over the last 20 or 30 years stream of ‘solar is ALMOST at parity’ news stories leads to inevitable disappointment and makes it harder to get ongoing support.

    Sorry for the diatribe.

    • Rob in CT says:

      Good post. I think it’s very important to not oversell a bill of goods. The fact is that the subsidies (which add up to ~1/2 of the system cost) are needed right now just to make it kinda-sorta a good deal. Even if you cut the component cost in half, you’d still need some subsidization to make it a good deal for the homeowner.

      Re: “must have a big south-facing roof”

      Bingo. Southwest, actually. South would’ve been even better. 32 305-watt panels (Siliken) will fit up there.

    • NCG says:

      Don’t be sorry. A fact-based diatribe is a thing of beauty, especially around here.

  9. Philip says:

    Couple of quick responses to the links James put at the top of his post – news stories about imminent grid parity:

    Mark Little/GE:
    -What he calls grid parity is parity compared to retail electricity prices. This isn’t really grid parity because of the implicit subsidies. Also interesting that his company makes thin film panels and he is commenting on the residential market – thin film panels are rarely used in residential installs because they generate less energy per unit area and residential roofs are generally space-limited. THin film gets used in large commercial/utility installs where space is not an issue.

    Rob Gillette/First Solar:
    -What he calls grid parity is comparing to peaker plants, not baseload. These plants are expensive to operate (ten cents per kwh or more), but exist int he grid because (a) they can be turned on and off quickly, and (b) they are economical when operated at low utilization rates, ie they have low fixed costs. Solar is in no way a replacement for these plants; in fact, adding solar to the grid with its volatility increases the need for these plants.

    -Nuclear is expensive – it has actually gotten more expensive per unit of electricity generated over the last few decades as regulations and safety measures have increased. Not a valuable comparison as nuclear prices are not at parity themselves now for building a new plant in most municipalities. ‘already competitive with NG and coal’ portion of his comment – patently false unless he is using logic like gillette or little

    Note one point about volatility. right now, whenever you flick a switch, you always get power. utilities have to manage the demand-side volatility that causes. on an average day, between middle of the night and the daily peak demand, there is roughly a 2X difference in demand and utilities deal with this by having baseload and load-following plants who’s cost structures are optimized to be most economical in those two applications. However comparing the peak demand on a specific day to the peak demand the next day, or the peak-to-trough demand hour-by-hour, and you get 30% to 50% fluctuations in demand. To deal with that, utilities have ‘peaker’ plants which turn on and off quickly. they might only get used 10% of the time, so they have to be cheap to build (low fixed cost) because there is less electricity generated to spread the fixed cost over. in general they cost 2X to 4X more to operate than a baseload plant. That peaker plant operating cost is the one some solar people compare there costs to even though solar panel electricity is can not be used to replace them, it can only replace baseload/load following (and usually requires more peaker plant capacity to deal with the solar generation volatility).

    • Rob in CT says:

      To be fair, regarding “implicit subsidies” ideally we’d have no subsidies (implicit or otherwise) for other energy sources. Implicit subsidy for coal: externalizing the environmental/human health cost of their emissions. Implicit subsidy for solar: net metering using retail rates. Meh. I’d say that’s *at least* a wash.

      Doesn’t natural gas work pretty well for peaker plants?

      • Philip says:

        Yeah, I roughly agree on the subsidy argument. The problem from a business perspective is: utilities can argue rather effectively that net metering means their costs (and so the public’s electricity rates) go up due to net metering since the connection is very easy to make and requires extra cash be spent now, which puts the net metering subsidy at risk once it becomes a substantial burden to electricity companies, while historically the more complicated tie between CO2 emissions from power plants to the negative and costly impacts on the environment (emissions now, cash impact later) has been less directly impactful of fossil fuel-driven energy businesses (although it is growing, the price you pay for gas/electricity etc. has little external CO2impact priced in). If ‘grid parity’ means i can compete without what we all agree are subsidies, then I worry quite reasonably based on precedent that ‘net metering’ is a subsidy in most people’s eyes while ignoring CO2 emissions (particularly in the current political climate) is not.

        NG – definitely used for baseload and peaker plants. Different types of plants though. For baseload, you build a very complicated combined cycle gas turbine – expensive to build, particularly the combined cycle part, but very effiient at extracting energy fron NG, perhaps 50% efficient. If you run the plant all the time (night and day), the increased efficiency more than makes up for the high capital costs.

        For peaking plants, though, you don’t spend the capital to get combined cycle efficiency because the plant will only run 10% of the time and the variable cost savings (higher efficiency burning of NG) won’t make up for the extra up-front capital costs. so you build a cheaper plant but it only burns NG at, say, 30% efficiency. And the cost of the lectricity it generates is considerably higher than a combined cycle plant used at 100% efficiency – 2X or more. But it IS the cheapest way to serve peaks in power demand – certainly cheaper than building an expensive combined cycle plant and then using it only 10% of the time; where the fixed costs would kill you. It’s this differnce in plant type that drives a significant portion of electricity price variation, but solar is only competitive with baseload, ie the lower end of this variation. I.e. if you are running a portfolio of baseload and peaker plants serving a geography, adding solar allows you to reduce your fossil fuel baseload but requires more peaker – so the right cost comparison for solar is the electricity cost from the baseload. This isn’t controversial at all as far as I know,. and yet often solar parity proclaimers neatly skip by the inherent error and say ‘solar prices are now competitive with peaking fossil fuel plants’ which is only true in a very narrow sense not relevant to what is conventionally meant by the term grid parity.

      • paul says:

        natural gas works pretty well for peak generation, but the price/kwH can be kinda terrifying.

        Meanwhile, it’s not clear to me whether retail net metering is so much an implicity subsidy as a question of market power relations. Residential distribution costs are (insofar as I understand, which is just the way a physicist would) essentially sunk costs. It wouldn’t cost the power company any less to run their distribution network if residential customers weren’t generating.

  10. My American commenters such as Philip seem to disbelieve the German BOS costs and trends, which seem pretty well documented to me. I cited three independent sources. B & G do not ignore BOS costs, they just – based on experience – project that they will continue to fall in line with module costs.

    The model is based on the effective average PV system price in Germany in 2010 of 2.70 €/Wp and 2.40 €/Wp for residential roof-top systems and industrial large-scale roof and accordingly PV power plants, respectively.

    Their predictions for other countries therefore assume German costs and trends, not current local ones. They don’t know what an installation costs today in Togo, probably a silly price. Is this approach unreasonable? The German market is huge – half the world total. Deviations from German prices represent market inefficiencies from small scale, obsolete regulations, and lack of know-how. (Similarly you would base your predictions for global computer memory prices on factory-gate prices in Taiwan or San Francisco, not Berlin or Sao Paulo.) These divergences will presumably iron out over time. There’s nothing secret about German methods and their labour cost is about the highest in the world.

    However, the qualification means that their model predicts grid parity applying German standards and efficiencies, which is optimistic. It’s a fair bet that the gap is wider for residential than for industrial applications; contrast planning a a 50MW installation in Kenya and a 5kw one. So I would expect their model to be more reliable for the diffusion of industrial than residential PV.

  11. NCG says:

    PS- I love the baby hatchling dinosaur. Very nice.

  12. Frank Schmitt says:

    Regarding point 4: I’m a big fan of battery electric vehicles, but as an illustration of what they’re up against, it currently takes around $5000 worth of battery to store each dollar’s worth of electricity in a car-compatible size and weight. Note also that most of the astounding gains of mobile-device battery life have come from improved data-processing efficiency. Unfortunately moving heavy things at speed, even at 100% efficiency, still takes real power.

    Until someone perfects the vanadium-boride air cell (or equivalent), liquid and/or gaseous fuels will likely still be our primary energy source for what we think of right now as automobiles. However I do see an increasing amount of battery-powered imperfect substitutes coming online: lightweight vehicles, autonomous vehicles, battery-swapping systems, or even just a mass realization that a reliable 30-mile range with overnight “refueling” at home meets 95% of the travel needs of 95% of the population.

    • The batteries are indeed the big hurdle. However, you are a little behind the technology. The Nissan Leaf claims 100 miles per charge, or a travel radius (aviation style) of 50 miles. The opportunities for extending this by recharging en route are limited if you only have access to household outlets, but any gas station (in 240v Europe anyway) should be able to offer a 30 amp/240v supply for get-you-home recharging. It will be difficult to get absolutely stuck. In two years, with a little policy push, I suggest it’s reasonable to hope for dedicated fast rechargers (125 amps, 30 minutes) on all motorway service stations. For commuters, cheap household outlets in car parks will get you home even at only 5 miles per hour recharge.
      A team of British researchers have published a report on car battery prospects: basically, only incremental improvements in power density and therefore range before 2020, thereafter new technologies (lithium-air or lithium-sulphur) could change things and get us to a 500 km range or 150 mile radius. A large amount of money and effort is going into all this.

      • Rob in CT says:

        I recall reading that the real range of the Leaf is a bit lower than that, in a friendly climate. In an unfriendly climate (like, say, New England), it’s even less.

        Since I’m not in the market for a commuter car at the moment, I don’t really care. But at some point, hopefully not until I’ve got ~200k miles on my civic, I will. And a non-very-close-to-zero chance of running out of power would be a big deal. My commute: 42 miles round trip. Sometimes I need to run an errand on the way home. So call the bare minimum range I need 50 miles. To feel safe about it, say 60. I don’t know for sure, but I suspect the Leaf, subjected to January in Connecticut temps, clears that bar but not by much.

        • Rob in CT says:

          And if you read the whole Wiki article, it points this sort of thing out. The range varies quite a bit depending on temp, driving conditions, and useage of heat/AC. The battery pack is guaranteed for 10 years. Then what? It’s freaking expensive.

          You have to weigh that against buying a significantly cheaper, efficient gas-powered car. By the time the Leaf owner recoups the price premium, are they staring down battery pack replacement?

          I’d like to have something like the Leaf, really. Without the catches.

          • Rob in CT says:

            Gah, 8 years, 100k miles. Not 10 years. The break-even point at $4/gallon gas is calculated at 7 years. So yay, you catch up in 7 years and then a year or two or three later the battery pack is shot?

    • PS: in case anybody is still looking a this thread,the point of plug-in hybrids is to get us through the transition to all-electric motoring while we wait for really good batteries. At present, hybrids like the Volt are basically internal-combustion vehicles with an extra battery for short shopping runs. Even the Chinese Qin – made by battery company BYD – has a 1.5 litre engine. Over time, I expect this balance to shift, as the battery expands and the engine shrinks. A dinky 2CV engine (the original was 375cc; you laugh, but they can run for 400,000km without a rebuild) would get you home or to a charging station, rather slowly.

Post a Comment