Tag Archives: NREL

Seba’s Solar Revolution Part 4 (A Question of Where and When)

Posted on February 3, 2019 | 7 comments

Over the last decade, the efficiency of solar panels has gone up and cost has come down. Accordingly, if we could move solar-generated electricity seamlessly through time and space, even a relatively poorly endowed country like the UK (in terms of solar irradiation and land availability) could meet its energy needs through allocating around 5% of its land mass to solar panels (as I discussed in my last post).

If the world were run by some kind of benevolent green dictator, he or she could possibly just issue a decree mandating a mass solar power build out which would replace all existing fossil fuel plants. In reality, the only dictatorship we face is that of ‘the market’. For solar to spread, therefore, the market must recognise solar as cheaper than existing fossil fuel alternatives.

Moreover, in order to reach a Tony Seba style 100% solar nirvana, solar must transition through a two-stage process. First, it needs to take out all the fossil-fuel competition with respect to new energy generation facilities to be built from now onward. Second, solar must push out all fossil fuel competition in the form of existing energy generation facilities. The first task is much easier than the second.

Energy generation costs are composed of two principal components: 1) the energy generating facility and 2) the ongoing operating and maintenance expenses. The second part is relatively easy to imagine. How much fuel and maintenance is required to produce X amount of energy, say a kilowatt hour (kWh) or megawatt hour (MWh)? For solar, the obvious answer to this question is “not much”. Once you have your panel set up, it just sits there generating electricity when the sun comes up every day. You may occasionally have to clean it and also prevent your local neighbourhood yob writing graffiti all over it or stealing the wires connecting it to the grid, but that’s about all. In economics speak, we describe this situation as one where the marginal cost of generating an additional kWh or MWh of electricity once a panel is in place is close to zero.

The marginal cost when producing 1 kWh or 1 MWh of electricity from a coal or gas-fired facility is, however, not zero since you need to put coal or gas in at one end to get electricity out the other end. For an automobile, you need to stick gasoline in at one end to get motion out the other end (in this case the via engine and the four wheels). Sorry, I know this bit is blindingly obvious.

The more complex bit of the LCOE calculation relates to the capital cost of the energy-generating plant required. For a utility scale solar farm, you will need to secure a large area of land (buy or lease), purchase the requisite number of solar modules, mount them, connect them up and then covert the electricity generated into a grid-compliant standard through the use of inverters and transformers. A 2017 report by the United States National Renewable Energy Laboratory (NREL) shows the cost breakdown of a variety of solar installations by size and also through time in the US.

Untitled 2

Once we know the total cost of the installation, it can then be apportioned over all the electricity generated through the expected lifetime of the facility. Simplistically, the capital cost per unit of energy produced is combined with the operating and maintenance cost of each unit of energy produced to arrive at a single number: the levelled cost of energy (LCOE). The LCOE also takes into account how the project is financed and the time value of money. The NREL provides a more detailed explanation of the LCOE calculation here and also an LCOE calculator that you can play around with here. For those of you who don’t have a financial background and are not familiar with discounted cash flow (DCF) methodology, you can just think of the LCOE as the price at which a project needs to sell its electricity in order to breakeven and stay in business.

Accordingly, if a utility scale solar project has an LCOE of $40 per MWh (which is the same thing as 4 cents per kWh, the financial press switches between the two), then the owners will be very happy bunnies if they can sell their electricity at $50 per MWh. Likewise, an energy consumer may want to enter into a power purchase agreement (PPA) with an energy generator for a set amount of electricity over a set period of time. If a solar utility is offering to enter into the PPA at 4 cents per kWh while a coal-fired facility can only go down to 5 cents per kWh, you will likely go with the solar – other things being equal.

The wording “other things being equal” is critical. Presuming no battery storage is involved, the solar facility can only supply electricity during the day and nothing at night. A factory operating 24/7 needs electricity 24/7. If its weekly requirement is, say, 100 MWh the fact that the solar farm can deliver at $40 per MWh versus $50 for the coal-fired plant will not be a sufficient condition for it to win the contract since it can’t supply the electricity both day AND night. At times, Tony Seba and other commentators can be rather disingenuous in claiming that renewable energy is cheaper than fossil-fuel generated electricity for just this reason. Having an LCOE for renewables lower than that for fossil fuel plants is a necessary but NOT a sufficient condition for renewables to displace fossil fuel. As I stressed in my last post, a kWh or MWh of energy that is not located in time and space is a pretty meaningless concept.

That said, I am not suggesting that we throw LCOE out the window. For renewables to replace fossil fuels, we first need to get the LCOE of renewables below that of fossil fuels and then we need to open up the gap between the two. If solar is generating electricity at $40 per MWh and coal at $100 per MWh, then $60 per MWh is available to transfer the solar generated electricity through time and space. The money could be spent on some form of storage (time) or some form of connection (space). The bigger the gap, the bigger the incentive for markets to try and arbitrage away the cost difference through putting in place mechanisms to transfer energy through such time and space.

With all those caveats in place, it’s time to look at some LCOE numbers. A well-respected benchmark annual appraisal of competing LCOEs is published every November by the investment bank and asset management firm Lazard. The entire slide deck is well worth flipping through and you can find it here, but I will just extract three charts.

First up, you can see that for new build facilities, the LCOE of both wind and utility scale solar is now below that of gas combined cycle and coal. Accordingly, if we didn’t have any issues with respect to the provision of energy in time and space it would be cheaper to deliver all new energy generation through renewables.

Untitled 2

Tony Seba’s claim, however, is that solar will not only be the energy generation vehicle for the future but it will also replace all the old fossil fuel facilities that have been constructed in the past. That is a much tougher hurdle. Remember the LCOE has two principal components: the ongoing operating and maintenance costs and the cost of the facility spread out over all the energy generated over the useful life of that facility.

When looking forward, the cost of building a brand new gas combined cycle or coal facility will be included in the LCOE number, when looking back it won’t. That is because that money has already been spent: it’s a sunk cost. So if solar is to mothball existing fossil fuel power stations, its LCOE must be cheaper that the LCOE of the gas or coal plant made up of the operating and maintenance (O&M) expense alone. The good news from Lazard’s November 2018 report is that wind and solar have got so cheap that they are starting to fulfil that condition as well. The cheapest solar facility at $36 per MWh is cheaper than a large proportion of coal-fired power stations whose operating and O&M costs are between $27 and $45 per MWh.

Untitled

Even more encouraging is the fact that solar has been consistently coming down its cost curve just as Tony Seba predicted.

Untitled

In addition, in areas of high solar irradiance we have seen 20 year power purchase agreements (PPAs) signed with solar utility scale projects at $20 per MWh or lower. The chart below shows the situation in the US, with new PPA price records being set in states like Arizona and Nevada (source: here). Presumably these PPA prices are higher than the projects underlying LCOE otherwise these projects would be loss-making and the solar utilities wouldn’t sign the agreements.

Untitled 2

Fortunately, the cost declines have been such that even in countries with poorer solar irradiance profiles, like those in northern Europe, solar has become increasingly competitive. The chart below is taken from a report by the German Fraunhofer Institute for Solar Energy Systems (ISE). At the time of this post, one US dollar bought 0.87 euros. Keeping that exchange rate in mind, ISE forecasts that the cheapest utility scale solar installations will see their LCOE drop from 4 euros per MWh to 2 euros around 2032. At that price solar will be far cheaper than coal and combined-cycle gas turbine plants.

Untitled

The upshot of this analysis is that countries that are well endowed in terms of solar irradiance already have solar plants that are cheaper than their fossil fuel competition on an LCOE basis, and countries that are less well endowed will see their solar plants winning out over the fossil fuel competitors on an LCOE basis over the next 10 years or so as solar costs continue to decline.

And now for some push back from the renewable energy skeptics Euan Mearns and Roger Andrews from the blog Energy Matters. In November 2018, Spain announced that it intended to move to 100% renewable generated electricity by 2050. Compared with Tony Seba’s claim of 100% solar by 2030 across the entire energy spectrum, it doesn’t seem so aggressive, but let’s put that to one side. In a post in November, Energy Matters took umbrage over the Spanish government’s claim and proceeded to show why such a target would be impossible to achieve.

At the heart of Energy Matters Roger Andrews’ argument is the claim that if we adopted renewables entirely to drive the electricity grid, it would become impossible to transport sufficient energy through time. Solar and wind’s intermittency would lead to large gaps in energy generation, and these gaps would be impossible to fill economically through the use of storage or by any other means. For his analysis, Andrews picked out the average electrical energy consumption and production patterns for two months in Spain: January and July. I will just concentrate on July here, but recommend you read the entire post to follow his argument from beginning to end. Here is the current contribution of renewables production to electricity consumption in July in Spain:

Untitled

And after scaling up renewables production:

Untitled

At this point, it’s worth reproducing the commentary accompanying this chart:

Obviously Spain plans to fill the hole with wind and solar. This approach has one thing going for it – the peaks and troughs in renewables generation are a good match to the demand peaks and troughs. But when we scale up July 2018 wind and solar generation (by a factor of 4.5) to match July 2018 demand we see that the amplitudes don’t cooperate.

Andrews then goes on  to produce a chart showing the deficits and surpluses:

Untitled

Which could only be equated by putting into place 1 terawatt-hours (1 TWh) of storage in his view:

Untitled

In January, the mismatch is even worse with a requirement for 2 TWh of storage to solve the intermittency problem. Roger Andrews further speculated that potentially 5 TWh to 10 TWh may be required to achieve energy security across the entire year. As an aside, how much would a TWh of storage cost? If we are going to provide such storage via batteries, the RNEL in a recent reported issued in January 2019 estimates the cost of a 4-hour system at $380 per kWh.

Untitled

Unfortunately, a terawatt-hour is a billion times bigger than a kilowatt hour. So to provide 1 TWh of battery storage would cost $380 billion at current prices. So to get the amount of storage Andrews suggests, we need trillions of dollars. Given Spain’s gross domestic product was only about $1.3 trillion in 2019, buying trillions of dollars worth of batteries looks unrealistic. Of course, there are other storage options like pumped storage, which Andrews briefly considers, but these have major environmental impacts.

So has Roger Andrews thrust a dagger into Tony Seba’s dream of golden world of solar? Well, these Energy Matters posts are certainly thought-provoking, but if I had a criticism it would be that they suffer from an awful lot of confirmation bias. Andrews appears intent on skewering renewables at the outset and then builds his argument to achieve that end. Accordingly, he frequently makes assumptions in his calculations that look somewhat dubious. In my next post, I will subject Roger Andrews’ skepticism to bit of my own skepticism to see if we can resurrect Tony Seba’s dream .

 

7 Comments

Posted in Energy revolution, renewable energy, Technology

Tagged , , , , , , , , , , , ,

Seba’s Solar Revolution Part 3 (Where to Put the Panels)

Posted on January 27, 2019 | 5 comments

In my last post, I mentioned that the late Cambridge University professor David Mackay was skeptical over the ability of solar to play a lead role in decarbonising the world’s energy infrastructure. MacKay’s highly influential book “Sustainable Energy Without the Hot Air” is rooted in basic science. Yet, despite the text being peppered with scientific identities, it also includes a number of value judgements that touch on the world of economics. And it is from these value judgements that MacKay’s skepticism arises.

MacKay’s book is principally concerned with what it would take to decarbonise the UK economy. Tony Seba, in contrast, forecasts that solar can power the globe not just the UK. In this post, I will stay with the UK, although I will look at other countries in future posts. Nonetheless, for Tony to be right, each and every country must be able to secure its energy needs through solar including the UK (though the solar energy may be imported from abroad). Accordingly, if Mackay’s argument is right (that is that the UK’s solar resource in inadequate) then Tony’s is wrong (notwithstanding the import argument).

Two of the major pushbacks against solar rest on the land mass requirement for sufficient energy generation and the intermittent nature of solar that puts unbearable stresses on the grid. As a former economist by training, I regard such arguments as second-order ones. They are both really subsumed under cost issues. Land is just a scarce resource like any other, and if the return on the land used for solar is higher than that for any other use, then it should be allocated to solar-power usage (that calculation can take into account the cost of climate change and the public good value of land).

Moreover, the unit of energy we are working with in this post, a kilowatt hour, is quite simplistic in economic terms. Energy is demanded at a particular place and at a particular time (hour of the day, and day of the year). A kilowatt hour generated in mid-summer in Spain in July, it not the same thing as a kilowatt hour consumed in mid-winter in London in January. The levelised cost approach (I will have a lot to say on that in future), which is used to compare different energy-producing assets, doesn’t take time and place into account.

In reality, we can think of the energy market as composed of 8,760 hour-long blocks (24 hours times 365 days) with a GPS attached to each one. In each of these GPS-stamped timed blocks, the market will equalise supply and demand at a certain price.

MacKay’s analysis only implicitly addresses the economics. Nonetheless, before we start moving energy through time and space, we must ensure that we have enough energy to move in the first place. MacKay does tackle that question.

In a section of his book titled “Fantasy time: solar farming”, Mackay conducts a thought experiment within which he covers 5% of the UK land with 10%-efficient solar photovoltaic panels.

untitled

He starts by calculating that “the average raw power of sunshine per square metre of flat ground (in the UK) is roughly 100 W/m2″. However, with a 10% efficiency photovoltaic panel, of the 100 W/m2 only 10 W/m2 is converted into electricity. From my last post we also know that if we leave a 40W light bulb on all day, it will use up nearly 1 kWh of power (0.04kW times 24). So if we generate 10 W per a one metre squared solar panel, we will get a quarter of that in energy, or 0.25 kWh. MacKay in his calculation has allocated 5% of the UK’s land mass to be used for solar power, which gives 200 m2 to each UK citizen. Times 200 by 0.25 kWh and we get 50 kWh per person per day, which compares with total energy demand of 125 kWh per person per day.

Also, as an aside, note that his calculation goes from power (solar irradiance measured in watts or kilowatts) into an energy number (solar insolation measured in watt hours or kilowatt hours).

At this point, let’s take a step back and look at that allocation of 5% of the UK’s land mass to solar panels. The UK land area is 25.25 million hectares and the population 66 million. Divide one by the other and we get around 0.38 hectares per person, (or just under an acre), which is the same thing as 3,800 m2. MacKay gives each person in the UK 4,000 m2 of land each since the population of the UK was about 5 million smaller when he was working out the maths. However, these numbers are near enough.

Switching from metres squared per person to number of persons per square kilometre, which is the standard measure when comparing countries, I have put together a table of population densities for selected countries, mostly ones with large populations, below. Note that a hectare (10,000 m2 ) is 0.01 of a square kilometre, so 0.4 hectares (40,000 m2 or 0.004 square kilometres) per person translates into 250 people per square kilometre.

untitled

 

From the table we can see that only the Netherlands, Japan, the Philippines, India and Bangladesh have population densities higher than the UK. So if the UK can become energy self-sufficient via solar it bodes very well for the rest of the world (putting differing solar irradiance numbers for each country aside for the time being). Moreover, the really profligate energy users, like the USA and Australia (which get through over twice the energy per person than the UK), have the advantage of having a lot of land.

Back to the UK and MacKay’s fantasy time solar farming:

untitled

That 50 kWh per day per person amounts to 40% of the UK’s energy consumption of 125 kWh per person per day. Accordingly, if we hold our 10% panel efficiency steady, then to meet 100% of UK energy requirements we would need to cover 12.5% of the UK land mass with solar panels (about 500 m2 per person).

Critically, MacKay headed his calculation “fantasy time” since he felt the calculation rested upon an unrealistically high cost. Fortunately, this is one area where MacKay was wrong (and Seba right): those fantasy cost reductions have come true (from Bloomberg‘s New Energy Finance (BNEF)‘s New Energy Outlook 2018):

untitled

In short, MacKay was far too pessimistic when it came to the cost curve. BNEF calculates a learning rate of 28.5% for solar PV. The learning rate 28.5% means that every time production capacity for solar PV panels is doubled, the cost of those panels comes down by 28.5%. This is an example of a virtuous circle: lower costs spur greater demand for the panels, which spurs greater production, which spurs future cost cuts and thus greater demand — and so the cycle goes on. (Of course, the panels are not the only components that go into a utility sized solar farm and all the other components will have their own learning curves and, hopefully, declining cost curves. We will come back to that in a later post.)

We are 10 years on from when MacKay wrote Without the Hot Air and already solar is overtaking all existing sources of fossil-fueled energy production in terms of cost competitiveness. Of course, there is a big caveat here: production costs are very different from the cost to deliver energy to a customer at a particular time and at a particular place as I have flagged above. Nonetheless, MacKay was worried about how solar stacked up cost-wise on a production basis out to 2050. That worry was misplaced.

How audacious is this plan? The solar power capacity required to deliver this 50 kWh per day per person in the UK is more than 100 times all the photovoltaics in the whole world. So should I include the PV farm in my sustainable production stack? I’m in two minds. At the start of this book I said I wanted to explore what the laws of physics say about the limits of sustainable energy, assuming money is no object. On those grounds, I should certainly go ahead, industrialize the countryside, and push the PV farm onto the stack. At the same time, I want to help people figure out what we should be doing between now and 2050. And today, electricity from solar farms would be four times as expensive as the market rate. So I feel a bit irresponsible as I include this estimate in the sustainable production stack in figure 6.9 – paving 5% of the UK with solar panels seems beyond the bounds of plausibility in so many ways.

A second observation (or criticism) is that MacKay seems to have also been too pessimistic in term of not just his cost assumption but also efficiency. In the above calculation, MacKay used 10% efficiency panels:

I assumed only 10%-efficient panels, by the way, because I imagine that solar panels would be mass-produced on such a scale only if they were very cheap, and it’s the lower-efficiency panels that will get cheap first.

In reality, those crystalline-silicon PV modules shown in the BNEF report above are far more efficient. From the United States Department of Energy:

Crystalline silicon PV cells are the most common solar cells used in commercially available solar panels…..

……Crystalline silicon PV cells have laboratory energy conversion efficiencies over 25% for single-crystal cells and over 20% for multicrystalline cells. However, industrially produced solar modules currently achieve efficiencies ranging from 18%–22% under standard test conditions.

 

True, these efficiencies are at the panel level not at the solar farm level. A utility scale solar facility will also need room for inverters, control panels, transmissions mechanisms, maintenance huts, security facilities and so on. Yet, we are already at around 20% efficiency levels for commercial products in 2019. Even if we knock off a few percentage points of efficiency to take account of ground cover occupied by stuff needed for a solar installation other than the panels, we are still far above MacKay’s efficiency figure.

A second area where MacKay was far too pessimistic with respect to the technology relates to the Shockley-Queisser limit. This limit sets the maximum theoretical upper efficiency limit of a single layer solar cell to around 33%. However, a new generation of multijunction cells has hopped over the Shockley-Queisser limit. With a two-layer cell your theoretical ceiling is 44% and with three layers 50%. The US National Renewable Energy Laboratory (NREL) shows the major improvements achieved in the past and those predicted for the future. The energy academic Varun Sivaram also devotes a chapter in his book “Taming the Sun” to these frontier PV technologies.

untitled

Currently, the really super-high efficiency panels that are up at 40% are not cost competitive enough to adopt for commercial use. Further, most have drawbacks in terms of manufacturing cells at sufficient size and also with respect to building cells durable enough to be deployed in real-world field conditions. Yet results to date suggest the more efficient panels have kept migrating out of the laboratory and into the marketplace at an ever-falling price.

Given where we are now in terms of panel efficiency and where we will likely be in 10 years time, it is possible that the 200 m2 of land allocated by MacKay to every UK citizen for solar panels could actually meet all the UK energy needs; that is, 125 kWh per person per day if we were deploying 25% efficiency panels (provided that the energy could be transferred though time and space). Further, once solar PV technology can be incorporated into roof tiles and road pavings, not all of the required space need be taken from agriculture land (figure below taken from Without the Hot Air“).

untitled

Then, of course, we could add energy generated from wind into our mix. Each additional kWh coming from wind energy means one less kWh needs to come from solar energy. Tony Seba’s focus was on solar, but I see solar and wind as inseparable twins.

Overall, Mackay was far too pessimistic over the ability of solar to come down its cost curve. In my next post, however, I want to look at an even more potent argument against the future primacy of solar. The blogger Euan Mearns and his co-contributor Roger Andrews are not huge fans of renewables and feel the displacement of solar is a pipe dream of green fantasists. We shall see what they have to say.

 

5 Comments

Posted in Energy revolution, renewable energy, Technology

Tagged , , , , , , , , , ,