Back in May, Alex Trembath and I explained why the value of wind and solar energy in electricity systems declines as these variable renewable resources reach higher market shares.
That article has since spurred a lot of discussion. David Roberts at Vox penned this great two-part series directly building on our column. The issues we raised popped up in this insightful chat about the future of solar between Varun Sivaram, a fellow at the Council on Foreign Relations (who earned an Oxford PhD in next-gen solar PV materials research) and Vox’s Brad Plumer. And most recently, Ben Paulos argues at Greentech Media that because of their zero-marginal costs, wind and solar “blow up” electricity markets when they reach high levels of penetration.
After a few months of articles and plenty of follow up discussion on this issue on Twitter, I thought it would be worth revisiting the article and clarifying a few key issues.
First, it is important to note that the issues Alex and I presented aren’t some new or revelatory discoveries. I make no claim to originating these ideas and aimed only to synthesize these concepts and ensure they are more widely understood. I cited the various research reports and papers that had brought this issue to my attention, including MIT’s Future of Solar Study, German researcher Lion Hirth’s extensive work on the economics of wind and solar, and Andrew Mills and Ryan Wiser’s great research at Lawrence Berkeley National Lab. Since then I’ve come across many more papers raising these dynamics, including a pair of 2007 papers by the prolific Paul Denholm and Robert Margolis that rather systematically explored the limits to scale for solar energy. Alan Nogee, former director of clean energy programs at the Union of Concerned scientists even dug up a few prescient papers from as early as 1990 that at least discussed these effects in qualitative terms.
It should probably be no surprise then, given the raft of well pedigreed research on this topic, that Ben Paulos effectively agrees with the premise of Alex and my article, that wind and solar have declining value as they scale up in electricity markets. I’ll let him explain:
In a typical market, generators place bids to supply power during a future hour or day. The bids are lined up by price, and selected in order (the “merit order”), from low to high, until demand is met. That last winner sets the price for power for that hour, and all the winners get that price.
This fun merit order calculator from the University of Texas Energy Institute illustrates how it works.
Because wind and solar plants will generate regardless of price, they bid into the market at zero, and take whatever the clearing price is.
If there is enough wind and solar, they start pushing other bidders out of the market due to what is called the “merit order effect.”
The result is a lower market clearing price and lower revenues for wind or solar generators (and everyone else generating at that hour). This is a good explainer of basic electricity market economics, and I encourage you to check out the calculator Paulos links to above.
Unfortunately, Paulos then goes off track a bit:
But with large amounts of wind and solar, the system breaks down. Imagine an hour when the whole system runs on wind and solar, all bidding zero — the clearing price would be zero! Nobody would get paid, including the wind and solar generators.
Paulos assumes that this situation constitutes a “break down” of the market, which is why he thinks too much wind and solar will “blow up” existing electricity markets and force major changes to the way we pay for electricity.
Some important changes to electricity markets may well be necessary for power systems dominated by zero-marginal cost resources. I’m currently working with colleagues at MIT and Argonne National Laboratory to explore how electricity markets might work in a hypothetical future where all generators have zero (or very close to zero) marginal costs (i.e. all zero carbon renewables, nuclear, and energy storage).
But I think Paulos identifies the wrong reasons for concern.
According to Paulos, the “break down” or problem in the market is that the market will not support further investment in wind or solar power.
“[I]f there is enough solar,” Paulos writes, “the market will no longer pay for it, and the system will break down. No more investment in solar.”
Paulos then goes on to suggest “a need for different market design for electricity, where wind and solar are paid by other means.” In short, he advocates paying renewable generators through dedicated feed-in tariffs or long-term power purchase agreements that shelter them from short-run electricity market signals and ensure they get enough money to drive more renewables investment. (Please read the full post here).
Paulos’s logic is effectively: if the market isn’t giving us more wind and solar, something must be wrong, and we should find another way to pay wind and solar outside the market to keep them growing.
This logic is simple, but it’s also backwards, confusing means and ends.
Renewable energy isn’t an end in itself. Rather, wind and solar are a means to various goals, including supplying affordable, reliable electricity without CO2 emissions or air pollution.
In other words, well-designed electricity markets should reward and remunerate electricity generators for the value they deliver by contributing to these ends.
Regardless of how we pay for them, wind and solar simply deliver less value the more they scale up. This is for a variety of reasons.
In very simplistic terms, electricity systems require three things:
- Energy: enough energy (megawatt-hours) generated to supply demand.
- Capacity: enough available power capacity (megawatts) to meet the highest peak in electricity demand.
- Flexibility: enough ability to change the output of generators from hour to hour and minute to minute in responses to forecasted changes in demand, unanticipated errors in demand or renewable energy forecasts, or unexpected failures of power plants or transmission lines.
And if we care about climate change and pollution—and we should!–we need to add a fourth requirement:
4. Emissions: we want low-carbon energy supplies that limit CO2 emissions and pollution.
The problem for wind and solar is that their declining value as they scale up isn’t some artifact of market design. The issues Alex and I discussed reflect real diminishing returns to society of deploying more and more renewables. Let’s look at how wind and solar contribute value on each of these four metrics:
1. Energy value
Wind and solar are fuel-saving resources, which when they generate, displace other, more costly generators. That’s good news, and when wind and solar displace expensive fossil fuels like oil, gas, or coal, they save the system money and deliver value.
But once wind and solar have displaced all fossil fuel-consuming resources at a given moment, adding more wind and solar won’t deliver much value. The additional savings of pushing nuclear, hydropower, or even other wind or solar generators off the market at that time is virtually nil. We just don’t save any more fuel (or other operating costs).
In the most extreme cases, wind and solar production at high penetrations can even exceed total demand (or the carrying capacity of the transmission system) during certain hours. This surplus renewable energy output simply has to be wasted or “curtailed,” delivering no additional value. The following graphic from the MIT Future of Solar study shows how curtailment of solar PV rises as its penetration (here measured as a share of peak demand) increases.
Summary: wind and solar contribute energy value to power systems, but this value diminishes once wind and solar displace costly fuels and when surplus wind or solar output begins to be curtailed.
2. Capacity Value
Wind and solar do contribute somewhat to the system’s need for capacity. But due to their variable nature, their so-called “capacity value” is less than their maximum rated capacity. In short, we can only count on a portion of our wind or solar capacity to be generating when we need them during peak demand periods.
This figure, also from the MIT Future of Solar study, illustrates how adding more solar affects the summer and winter demand in the UK. As the figure illustrates, the midday peak in solar production doesn’t align well with the late-evening peak in winter demand (top panels). While solar helps offset the midday summer peak at the outset (lower panels), eventually the peak in remaining demand shifts into the evening as well, and adding more and more solar has little effect. (Note that while the UK isn’t the sunniest place in the world, the same general dynamic holds elsewhere; solar may have a higher capacity value, but it still declines as solar scales up).
Summary: The capacity value of solar and wind is only a portion of their maximum rated capacity. Solar typically has a higher initial capacity value at low penetration levels, but this value diminishes quickly as solar displaces the midday peak and the remaining demand in the evening is unaffected by additional solar capacity. Wind generally provides lower capacity value from the get go, but faces a slower decline in this value as penetration levels increase.
When it comes to a system’s flexibility needs, wind and solar unfortunately contribute little. Being able to curtail wind and solar when necessary can help deliver some flexibility, but this has limited value.
Indeed, since wind and solar are variable and somewhat unpredictable resources, adding more wind and solar to the grid usually increases the system’s overall need for flexibility. We need to have more flexible generators online and able to respond quickly if renewable energy output differs from forecasts, rising or plunging unexpectedly.
Summary: Solar and wind add to system flexibility costs, rather than deliver value on this metric.
4. CO2 and air pollution
Wind and solar generate electricity without carbon emissions or air pollutants. So long as these renewable sources are displacing emitting fossil fuels, they deliver real value to the system on this metric then.
Unfortunately, the same dynamics behind their declining energy value means the carbon and pollution-reducing value of wind and solar disappears once they’ve already pushed fossil fuel burning power plants offline.
Summary: Solar and wind are emissions free energy sources, but if they are already competing with nuclear, hydropower or other renewables, adding more delivers no additional carbon benefit.
It is not at all clear to me how designing better markets would solve any of these challenges – besides properly valuing emissions-free energy sources of course.
As I noted in our original post, we can always continue pushing wind and solar into markets as far as we’d like—as long as we’re willing to pay the costs.
But at the point where we’ve exhausted the energy value of wind and solar as fuel-savers, where these variable sources contribute little capacity value to the system, and where they are already competing against other emissions-free energy sources, why would we want to deploy additional wind or solar?
Markets are tools, and I believe they should be designed to deliver societal value at an efficient cost. In this context, that means electricity market design should ensure payments to generators reflect the value of these resources to the electricity system in a carbon-constrained world.
But in such a market, variable renewable resources – just like all resources – would still have an economic limit to penetration. Short-circuiting that limit through side payments or subsidies would only add cost while delivering little additional benefit.
If we keep our focus on ends rather than means, I don’t think this should be a controversial point.
A note: I’ve promised David Roberts among others that I’ll devote a forthcoming entry in this series to discussing the various ways we might increase the economically-optimal penetration of wind or solar power, including energy storage, demand response, higher capacity factors, and lower costs for renewables. I’ve been juggling summer travels, research at MIT and Argonne National Laboratory (where I’m a summer researcher), and writing a few papers, so I can’t promise when I’ll deliver that post, but stay tuned. For now, check out this work by Mills & Wiser and this paper by Lion Hirth, and I’ll try not to leave y’all hanging for too much longer…