The economic limitations of wind and solar power

Wind and solar power are booming, rapidly becoming serious players in electricity markets around the world. The question now is not whether variable renewable energy (VRE) is viable, but how far it can go. Just how much of a grid's electricity can come from wind and solar?

In a previous post, I described the challenges that face traditional grids as they attempt to integrate more VRE, along with the emerging solutions to those challenges. At this point, the obstacles are fairly well understood. As the National Renewable Energy Laboratory (NREL) says, there is no technical barrier to a grid running on 100 percent wind and solar.

Rather, what limits exist are economic. At a certain point, in a given grid system, the cost of integrating more VRE exceeds the benefits. NREL refers to this point as a grid's "economic carrying capacity" for wind and solar.

Obviously, economic carrying capacity varies from grid to grid. And it's a moving target — it can be expanded, as we'll see in subsequent posts. This post will simply serve as an introduction to the economic limitations facing VRE on current grids, operated by current rules and markets.

Jesse Jenkins and Alex Trembath recently published an interesting set of posts that address these questions. The first is about the enormous progress VRE has made to date. The second is about how far VRE can go before economic carrying capacity is reached.

Jenkins and Trembath propose a "rule of thumb": "It is increasingly difficult for the market share of variable renewable energy sources at the system-wide level to exceed the capacity factor of the energy source."

Let's unpack that a little.

"Capacity factor" refers to how often a power plant runs and thus how much power it produces relative to its total potential (capacity). Nuclear power plants in the US run around 90 percent of the time, so they have a 90 percent capacity factor. On average, the capacity factor of solar ranges anywhere from 10 to just over 30 percent. For wind, it ranges from 20 to just over 50 percent, averaging around 34 percent in the US.

The rule of thumb doesn't say that driving wind power beyond 34 percent of grid electricity is impossible, only that it becomes "increasingly difficult," i.e., increasingly expensive. A similar story holds for solar.

This is true, the rule says, at the whole-system level. A grid that is connected to other grids, like Denmark's is connected to Norway's, is not a whole system, it's a subsystem. It has a place to export or import energy, to balance out its own fluctuations. The rule of thumb kicks in when there's no longer any place to export to or import from.

So that's the rule of thumb. Obviously it leaves lots of wiggle room. Capacity factors vary from grid to grid; so do available grid flexibility measures; so do public and political willingness to subsidize renewables.

And keep in mind that if wind topped out at around 35 percent and solar topped out at, oh, 25 percent, together they would cover 60 percent of total electricity demand. That would be an absolutely remarkable feat. In truth, say Jenkins and Trembath, there's reason to think the rule of thumb is too generous and VRE will top out at maybe 50 percent of global electricity. Even if they're right, though, that would represent an energy revolution, not some grim disappointment.

Still, it would leave 50 percent of electricity demand to be covered by some combination of other low-carbon sources: hydro, nuclear, biomass, geothermal, and coal or gas with carbon sequestration.

So are Jenkins and Trembath right? The phenomena they discuss are certainly real. The open question is whether and to what extent we can engineer around them. It's complicated.

Before we get to the specific problems Jenkins and Trembath identify, a quick note.

One often hears about places where wind or solar is providing some startlingly high percentage of energy. These figures can be somewhat misleading. For instance, Iowa is said to get 28.5 percent of its electricity from wind. And that's true in terms of markets and accounting: 28.5 percent of the power contracts signed by Iowa utilities are with wind generators.

But Iowa does not have its own grid; it is part of the Midcontinent Independent System Operator (MISO) grid region, which includes all or part of 13 other states. The electrons on the MISO grid cannot be divided into wind electrons and coal electrons. Every load (user of electricity) on the grid is, physically speaking, consuming the same mix of energy. Currently the MISO grid gets 5.7 percent of its energy from wind and, thus, so does Iowa. (And technically, even MISO isn't a whole system, since it is strongly interconnected to neighboring grids and the larger Eastern Interconnection.)

Very few whole grid systems have achieved double-digit penetration of VRE, including Texas's, Spain and Portugal's, and Ireland's. The point is not to diminish the accomplishments of VRE to date, but simply to note that even with its headlong recent growth it remains a small part of global electricity — 3.3 percent in 2013. The challenges of high VRE penetration are still mostly ahead of us.

You can see a problem illustrated in the chart above: As the grid penetration of solar power rises (in a self-contained, Texas-like grid), it reduces prices for itself much more than it reduces average prices. The same basic dynamic holds true for wind power. "In short," say Jenkins and Trembath, "wind and solar eat their own lunch!"

What's behind this is something called the "merit order effect." "Merit order" refers to the order in which utilities deploy electricity sources as demand rises. They generally proceed from the cheapest sources to the most expensive. When demand peaks in the afternoon each day, utilities bring the highest-priced sources online, usually natural gas "peaker plants." That expensive power is a big source of revenue for generation companies.

Here's the thing about wind and solar energy: Once they are installed and paid off, they produce power that is, essentially, free. It has zero marginal cost, which means it always comes in at the top of the merit order. It's always first in line.

That's bad news for generation companies who make their money on peak power. Solar, in particular, tends to be coincident with demand, rising and peaking at midday, right when demand is at its highest. It replaces expensive power with free power, reducing prices. A Fraunhofer Institute study found that thanks to the merit order effect, solar PV reduces average German wholesale electricity prices around 10 percent and peak prices up to 40 percent. That was in 2007; a 2012 study had similar results. Wind is not quite as coincident with demand, so its price-suppression effect is somewhat less concentrated, but it exists.

On its face, this would seem to be a good thing: cheaper power! And it is a good thing from a consumer's perspective. But it poses a problem for VRE owners and investors.

To understand why, imagine a merchant selling Elsa dolls from Frozen. From 2 to 6 pm, a bunch of competitors flood onto his block selling their own Elsa dolls. For four hours, there are thousands of identical dolls available, and the price of dolls plunges to almost nothing. This makes it difficult for any of the merchants to make money — from 2 to 6, at least. But at 6 pm, all the other merchants go home and the single merchant is left, cornering the market from 6 to the following day at 2 pm. All the competition from 2 to 6 suppresses his profits somewhat, but not as much as it suppresses the profits of merchants who only operate in those four hours of glut.

That's what VRE starts doing as penetration rises: It floods the wholesale power market all at once, when the wind is blowing or the sun is shining. The power glut during those hours drives down the "market clearing price" for power, sometimes all the way to zero. The more VRE there is on the grid, the less the next VRE generator can expect to make.

Here's how MIT's Future of Solar study puts it:

[A]s a result of basic supply-and-demand dynamics, solar capacity systematically reduces electricity prices during the very hours when solar generators produce the most electricity. Beyond low levels of penetration, an increasing solar contribution results in lower average revenues per kW of installed solar capacity. For this reason, even if solar generation becomes profitable without subsidies at low levels of penetration, there is a system-dependent threshold of installed PV capacity beyond which adding further solar generators would no longer be profitable.

That "system-dependent threshold" is the current economic carrying capacity of a given grid for solar. To get beyond it, solar would either have to continue getting cheaper and cheaper at its current crazy rate, or be subsidized more and more.

(European analyst J. M. Korhonen also has a great explanation of this effect.)

The second problem Jenkins and Trembath identify is curtailment, when grid operators purposefully throttle back VRE for security or economic reasons.

As the penetration of wind or solar reaches roughly its capacity factor, the power it supplies will regularly swing between zero and 100 percent of demand. If penetration exceeds capacity factor, it will periodically generate more than 100 percent of demand.

Obviously, if it can't be exported or stored, any power in excess of 100 percent of demand will have to be curtailed. But it turns out VRE will have to be curtailed well short of 100 percent.

Security curtailment: For one thing, there's a certain amount of conventional generation that can't be shut down. Grid operators have to be prepared for unexpected fluctuations in demand or VRE output, or an unexpected equipment failure. To maintain system stability, they keep some dispatchable power plants "spinning" at all times, just in case (usually natural gas plants, which are the most nimble). Maintaining those "operating reserves" means that VRE has to be curtailed well short of 100 percent.

This big integration study for the Western US found that VRE would have to be limited to 55 to 60 percent of total demand. Ireland limits it to 50 percent at any given time. That level can be raised, and grid operators are working on raising it, but it's there.

Economic curtailment: This happens because some big, baseload nuclear and coal plants aren't very nimble. It takes them a long time to ramp all the way down, cool off, and ramp back up — and it's expensive. Especially if a spike in VRE is going to be short, under an hour or so, it doesn't make economic sense to shut off these baseload plants. Instead, grid operators will spin them to their technical minimum output and curtail VRE beyond that.

The chart at the top of the section shows security-related curtailment for solar kicking in right around its capacity factor (18 percent) and economic curtailment kicking in well before that.

Where does this leave us?

Note that both these problems are related to variability. Wind and solar generators all come on together — when the wind is blowing or the sun is shining — and go off together, when it isn't. That means both problems can be solved in one of two ways: by spreading out wind and solar generation, making the supply steadier, or by shifting in more demand under the VRE curve, so more people are using electricity when wind and solar generators are making it.

I will be delving into specific solutions for VRE grid integration in future posts — there's a lot of fascinating work going on this area. For now, a few general remarks.

This post is framed around "limitations," but as I said above, if wind and solar both reach global grid penetrations equal to their capacity factors, that would make VRE cumulatively around half of all global electricity. That would be amazing!

And let's recall that in the 1990s and mid-2000s, the conventional wisdom was that VRE would hit a ceiling at about 10 percent of total grid power in the US. Now NREL says that "an economic carrying capacity of 30% [VRE] in much of the United States will require largely understood changes to operational practices as well as transmission capacity expansion." Getting there will involve enormous challenges, but they are challenges we know how to overcome.

Smart research and modeling pushed our credible aspirations from 10 percent to 30 percent. By the time we reach 30 percent, smart research and modeling (along with considerable real-world experience) will have pushed it higher.

These economic challenges don't give fans of wind and solar any reason to slow down or get discouraged. But they are a good reason to start planning ahead, so that as wind and solar continue their rise, they are met by a modern grid that can accommodate them.