Flawed arguments: A critique of “Upper Limit to Variable Renewable Energy”

My readers will pardon that I am, for the third time in a row, writing a piece inspired by a post by David Roberts of Vox. Unlike most of the media David Roberts is writing intelligent articles about what I consider to be the most critical issue of our time, the global transition to renewable energy.

I applaud Roberts for this, even if I find his analysis to sometimes miss important details, and if I find his choice of sources disappointing, per the latest article. As I did in the last post, I’m going to go straight to the source: an article by Jesse Jenkins and Alex Trembath, both of whom proclaim their association with the Breakthrough Institute (BTI).

The arguments made by this article rest on faulty assumptions, including setting conditions that no zero-carbon generation can currently meet. While additional technologies will need to be deployed and extra costs incurred, various means of storing energy remove any “upper limit” to deployment of variable renewables.

Wind and solar in wholesale electriciy markets

Jenkins’ and Trembath’s first argument is that wind and solar “eat their own lunch” in competitive electricity markets. There are a couple of phenomenon at work here.

First, as shown in the MIT Future of Solar study (p.177-178), deployment of solar PV results in a “peak shaving effect” in all parts of the United States except the Pacific Northwest. In every other region it reduces peak demand from air conditioning use during summer months and thus reduces overall system costs by avoiding the use of expensive imported power and “peaker” plants. It’s worth noting that under nearly all existing policies solar PV owners are not rewarded for this service. As you deploy more solar PV, eventually this demand is met and the remaining peak to be met shifts to after sundown.

The second is that, as the authors note, putting more of something online at a given time in a competitive market decreases its value. Thus, if we deploy high levels of wind and solar photovoltaics (PV) or concentrating solar power (CSP) without storage, power prices for all generators on the wholesale market become very cheap. In Germany, moderately high penetrations of wind and PV are already driving wholesale electricity prices into the negative.

The same dynamic holds true for BTI’s chosen form of electricity generation: nuclear power. Nuclear is the least flexible form of electricity generation, in terms of its ability to ramp up and down. Additionally, like solar PV, most of the costs of a nuclear power plant are fixed. These have to do with construction, licensing and decommissioning, not fuel.

For both these reasons, nuclear reactors typically run as near to 24/7 as possible. This includes overnight, when demand is low and thus the wholesale market value of this power is low.

If you were to power a grid entirely with nuclear reactors, you would hit a similar phenomenon of uncompetitive power to that described by Jenkins and Trembath in their piece. This will occur at a higher level of overall system power than with wind and solar PV, but it will happen as nuclear hits minimum overall demand. This is during the night and leaves the problem of meeting all of daytime demand, including peak demand.

Markets, subsidies and nuclear power

Looking at wholesale market value, Jenkins and Trembath argue “If renewable energy is ever to become truly subsidy independent and earn its keep in electricity markets, that means there is a natural stopping point at which a marginal increment of wind or solar will become unprofitable”.

However, by describing a free-market electricity system independent of subsidies or intentional market design, such as feed-in tariffs, Jenkins and Trembath ask solar and wind to work with conditions that are not only unfavorable for all zero-carbon energy, but impossible for nuclear power.

The United States has provided generous funding to support nuclear power for nearly 70 years, and this currently represents the majority of the Department of Energy’s (DOE) budget. To spare words I’m going to skip a lengthy explanation of the details, except to note that this includes federal insurance and loan guarantees, both of which are necessary conditions for the deployment of nuclear power as the private market would not touch nuclear projects without them, as well as R&D and disposal of waste. Additionally, utility customers are typically forced to foot the bill for nuclear power plants during construction before they ever produce one electron.

If you take all that away, then we can discuss zero-carbon energy sources “earning their keep” without subsidies and/or market design. Under those conditions no nuclear power plants would exist, and therefore they would not be part of this discussion.

Overall, I don’t find this frame of power sources “earning their keep” as helpful. Such “free-market” approaches inevitably end up favoring incumbents. And given that subsidies and/or market design have played a critical role for the deployment of all our zero-carbon options – including wind, solar and nuclear power – what we would be left if these had not been utilized is fossil fuel-fired generation. It’s worth noting that this benefits from not only subsidies but externalized health costs and climate impacts.

The global climate emergency demands that we move rapidly to world of low-to-zero carbon electricity, which requires political decisions and either a design of the market to favor zero-carbon electricity and/or incentives until battery storage costs fall further.

Economic curtailment

The second point by Jenkins and Trembath that I’d like to discuss is economic curtailment – the actual practice of shutting down wind and solar when they conflict with baseload power on the grid. Wind and solar don’t “play well” with baseload power stations, which turn off slowly and can take days to come back on. These power plants are mostly coal-fired and nuclear, and the share of both is decreasing in Europe and the United States.

We’re closing down coal-fired power plants in the United States, starting with the oldest, because they are no longer economic and because they are the dirtiest plants in terms of health impacts. We’re not building new ones, for obvious reasons having to do with CO2.

We’re also closing down nuclear power plants, starting with the oldest and most problematic, because they reach a point where the cost of upkeep exceeds the revenues they are expected to earn. Many of these plants probably should be closed sooner due to safety concerns, but this choice is typically made by utility profit calculations.

We’re not building many new nuclear power plants, because even with a plentiful array of subsidies, they’re still quite expensive (see also EIA). They simply never come in on budget (or on time), and decision-makers are wise to this. This is why the construction of new nuclear reactors was in decline globally before the Chernobyl Disaster.

Other than wind and solar, what is being built in the United States at this time is natural gas-fired generation. These plants are more flexible and ramp faster to fill the gaps presented by wind and solar, something which is covered in detail by MIT.

Under current market rules this conflict between variable renewables and baseload will force more of these plants offline. As noted by MIT, due to the merit order effect and the lower marginal cost of running nuclear power plants, solar (and wind) will push off the coal plants first. This is a net societal benefit.

Thus I am not too worried about “economic curtailment” in the United States and Europe. I expect baseload nukes and coal to be largely phased out (again, for economic reasons) before we get around to reaching the kinds of wind and solar penetrations that we are discussing, at least in the United States.

In other nations, the penetrations of renewable energy are higher, and here many nuclear power plants are or will be closed down by the political will of the German, Swiss and Japanese public. For the same reasons new ones will not be built in Spain or Italy.

Security curtailment

In addition to “economic curtailment”, Jenkins and Trembath cite concerns around “security curtailment” – in other words the point when you need to shut down high levels of wind and solar because they are producing more than total demand and have become a danger to the grid. There is a sound technical basis for this argument, in that when deployed on their own, the amount of solar PV and wind that a grid can absorb is limited.

I must note here that this concern is mostly at a grid level, not the individual nation, and that very few places in the world have reached this level of wind and solar penetration at the grid level. The authors point to Ireland, incorrectly stating that this nation is the world leader in integration of variable renewable energy on a grid level. Actually, Spain and Portugal are, as the Iberian Peninsula features both a higher portion of wind and solar generation and lower degree of interconnection. In Spain, which makes up the large majority of this market, the spot penetration of wind alone has increased above 60% of total system demand.

In the past, grid operators in many nations have made repeated assumptions about the levels of wind and solar that the system can absorb, and they keep revising them upwards as they are exceeded in the real world.

But the larger issue here is that the concern over security curtailment in Jenkins and Trembath’s article and as cited in the MIT Future of Solar study is based on a model that does not include any form of energy storage or other means of soaking up extra demand.

In particular, Jenkins and Trembath never mention CSP, which can integrate low-cost thermal energy storage (TES). The MIT Future of Solar study notes that deploying CSP with TES can increase the total amount of solar that can be deployed and improve the economics of PV on the grid. In the recently released study on moving the United States to 100% renewable energy, Professor Mark Jacobson and his team give a large role to solar CSP as a means to meet demand, at over 10% of total annual energy/electricity.

Two others are worth mentioning here:

  • Pumped hydro. This technology is often overlooked in discussions of storage, but is the most widely deployed form of energy storage globally, with over 100 GW available. There are significant amounts in Europe (almost 5% of total generating capacity) and Japan (10%).
  • Converting surplus electricity into heat for use in district heating systems. Incidentally, Jacobson also gives a significant role to this in his study.

This is not a hypothetical, as all three of these technologies are currently deployed at varying scales in different nations. To assist with the integration of solar PV and wind, we will have to build more CSP with TES and electricity-to-heat systems.

The rule of thumb

The conclusion of all these arguments is Jenkins and Trembath’s “rule of thumb”, which they claim is supported by “a growing body of research”:

“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”

I found this “rule of thumb” quite curious, so I asked for the opinion of Craig Morris, who has been writing in great depth about renewable energy for over a decade, and who knows more about the technical details than almost anyone I know.

Here is what he had to say:

The potential share of wind/solar without curtailment or storage depends on far more than the capacity factor. Most importantly, it depends on the particular power demand profile. Solar has a capacity factor of around 10 percent in Germany, and I have said since 2010 that 10 percent is the level at which we start needing curtailment/storage.

The potential penetration level of wind in Germany and Denmark without curtailment/storage is higher than the capacity factor of wind because most wind is generated in the winter, when power demand is also higher. Denmark aims to switch to electric heat to drain off excess wind power, which will have the same effect as air conditioning does for PV in the US – it will provide more space for the renewable energy source to grow. By leaving out these factors, Jenkins oversimplifies the matter.

I have additional concerns. The authors presents this rule, but does not state clearly if this is due to the technical or economic factors that they cite. So what exactly are they talking about?

If they are referring to technical constraints, then I must refer to the section on security curtailment and note that this argument is only valid when you consider deploying enormous amounts of solar PV and wind without any energy storage, which no one is actually considering.

Which leaves economic constraints, and it appears that the authors are attempting to to push their flawed model as an economic law, by arbitrarily deciding that no currently available (and deployed) forms of energy storage can be deployed at larger scales or new locations in the future because they have deemed them too expensive.

I must note here that in addition to the forms of energy storage (CSP with thermal energy storage, pumped hydro and electricity-to-heat) that I have listed above, there are other solutions that make this task easier by shifting demand to coincide with periods of high power production, including demand response, and time-of-day pricing of electricity. These are all far less expensive.

And this is before we even get to batteries, which are currently being deployed at significant scales in California, Germany, Italy, and Spain. And as we deploy more batteries, you can expect the cost of batteries to fall, due to both economies of scale and technical improvements.

Additionally, as we transition to electrified transport – an essential feature of any model for 100% renewable energy that goes beyond electricity – the use of time of day pricing and electric vehicle charging can soak up output at times of high production.

I don’t expect any one of these solutions to solve these issues and get us to 100% renewable energy on its own, or even the 80% (by 2030) which I consider the more significant goal. Currently different combinations of these solutions are deployed in diverse regions due to geography, the decisions of policymakers, and actual needs.

You can also expect is that the least expensive of these (demand response, time-of-day pricing, and probably pumped hydro and CSP with storage) will be deployed first as we move to higher penetrations of renewable energy. Thus, it is absurd to allude only to the current cost of battery systems when discussing the cost of storage.

Flawed arguments

As such, these concerns voiced by Trembath and Jenkins don’t make a case for their rule on the upper limit of solar and wind deployment. In the first point, as to wind and solar prices, the phenomenon is real but it affects all low-to-zero marginal cost generation, and all zero-carbon generation has been shielded from the free market. As such this is an interesting phenomenon, but not an effective critique of wind and solar when you consider the only other form of zero-carbon electricity available in all geographies at this scale: nuclear power.

In the second point, as to economic curtailment, this is a current practice, but as we are turning off inflexible forms of baseload generation in the United States and Europe, the effect will be limited. In the third point, security curtailment, this depends upon no energy storage and no soaking up of extra demand – which no serious model for moving to high levels of wind and solar considers.

The same is true for the “Rule of Thumb”, which is an imprecise concept which ignores variation in actual demand profiles. More importantly, actual limits in the penetration of wind and solar PV are removed when we deploy various solutions, including CSP with storage, pumped hydro, electricity-to-heat, or EV charging and time-of-day use. Before we even consider stationary batteries.

So instead of some “upper limit” to the penetration of wind and solar, instead it is clear that there is a point where, in order to reach higher penetrations of wind and solar, you need to utilize energy storage and means of shifting and absorbing additional demand. This is necessary for both economic and technical reasons.

This isn’t new information to anyone working in these fields. To present a “rule” for renewable energy, based upon unstated and absurd assumptions is misleading. It’s also typical for the Breakthrough Institute.



  1. […] Christian Roselund goes into more detail. He agrees with me. There is no reason why the share of solar in some grid or other should be […]

  2. […] Christian Roselund goes into more detail. He agrees with me. There is no reason why the share of solar in some grid or other should be […]

  3. Peter Farley · · Reply

    The point is almost all grids already have storage to optimise thermal power stations. It is called Spinning reserves and hydro.

    All thermal power stations are backed up by one or other of these techniques. If it is ok to have thermal power stations running below capacity to share load and back each other up i.e. have spinning reserves which actually costs money in terms of both capital and operating costs.

    In addition almost all grids have dispatchable backup in the form of hydro, landfill gas, biomass, geothermal etc. In addition many grids have pumped hydro which is installed to smooth the load on Nuclear or FF stations.

    Given that the thermal grids have all this backup why is it not ok to have excess wind and solar and use the same backup techniques that are already used for thermal grids. At least they have no operating costs and are not damaged by cycling on and off.

    1. I think there are two separate but related technical concerns here. As you move towards 100% renewable energy and max out the amount of wind and solar PV that a grid can handle w/o storage or curtailment (which no large grid has done yet), you hit both “peaks” – where solar PV and wind are producing more than demand – and “valleys” where they are at minimum output.

      Spinning reserves and flexible hydro can help to fill in the valleys. The peaks require some some sort of mechanism to store the extra output. Since we tend to have fewer technologies deployed to deal with the peaks, this was the focus of this piece.

      But yes, you are right – many grids have some degree of pumped hydro that can be used to accommodate some of the additional demand. Also there is electricity-to-heat, as is done in Denmark, and EV charging. All of these can deal with the peaks, before we even get to stationary batteries.

      I agree with what I see as the essence of your post – that as we have these solutions for conventional power already, this is not some new uncharted territory.

      1. Peter Farley · ·

        I think we are pretty much talking about the same thing, The peaks and valleys problem exists now, particularly with nuclear and to a lesser extent coal. The difference is that we were coping with peaks and valleys in demand, now we are talking about peaks and valleys in supply. In practice it is variation in flow, the cause requires slightly different configuration of the grid but it is a minor issue in the overall scheme of things.

        However the advantage we have now, is that storage of all sorts is become cheaper and more efficient while capabilities to shift loads is much better and the cost of overbuilding renewables is nowhere near as expensive as it was

      2. There’s a third consideration: The spinning reserve itself is thermal generation, usually fossil-fuelled, which is what we are proposing to retire by replacing it (largely) with intermittent renewables.

        The valleys, the increasingly brief periods where demand exceeds the instantaneous supply of what clean energy sources can provide, are currently filled with dirty energy sources. Since both the depth and breadth of these valleys is declining, that’s perfectly acceptable for the time being, but we mustn’t lose sight of the simple fact that it means we need to keep sufficient dispatchable fossil-fuelled generation in a serviceable state until we are able to fill the valleys with stored clean energy instead.

        Power-to-gas lets the very same infrastructure be used for burning fossil fuels and for releasing stored clean energy. Unfortunately it’s being pursued in earnest only in Germany, where gas-fired power has actually become uncompetitive and almost all valley-filling jobs at present are done with coal, except in the dead of winter. Hopefully there’ll eventually be sufficient penalty for the pollution associated with coal-burning (and/or sufficient subsidy to storage) that gas, power-to-gas and other storage technologies can compete with coal for the valleys, just as wind and solar now compete with it for the peaks.

      3. Yes, I anticipate that the valleys will be filled with natural gas for the time being in the United States. As for Europe, I anticipate lower prices globally for LNG will cause Germany and other nations to shift back to gas.

        For both nations, there are other solutions that can lower fossil fuel usage. Fast-ramping biogas and biomass can help to fill some of these valleys, but we will need market design and/or incentives targeted specifically for those technologies and those uses.

        I expect gas to be the last conventional source of electricity to be eliminated under the Energy Transition, as we will need to fill these valleys until we reach much higher deployment of renewables and mass roll-out of various forms of energy storage.

  4. Nick Tedesco · · Reply

    Viable storage solutions are on the way, so this issue will not be a major concern in the coming years.

    1. If I may offer a friendly correction: Additional viable storage solutions are on the way. We already have CSP w/ TES, pumped hydro and electricity-to-heat.

  5. […] my last post, I detailed a lot of those options to store and/or utilize energy without batteries in order to […]

  6. […] my last post, I detailed a lot of those options to store and/or utilize energy without batteries in order to […]

  7. […] are two key takeaways. First: As Kaua’i is proving, any alleged “limits” to levels of wind and solar integration are a joke when energy storage (including pumped hydro) is considered. Second: if Kaua’i can […]

  8. […] limit solar’s penetration levels to near its capacity factor (solid critique of this argument here). Meanwhile, Shayle Kahn and Varun Sivarum argue that the solar industry can mitigate solar value […]

  9. […] limit solar’s penetration levels to near its capacity factor (solid critique of this argument here). Meanwhile, Shayle Kahn and Varun Sivarum argue that the solar industry can mitigate solar value […]

  10. […] limit solar’s penetration levels to near its capacity factor (solid critique of this argument here). Meanwhile, Shayle Kahn and Varun Sivarum argue that the solar industry can mitigate solar […]

  11. […] the past many of my arguments about the feasibility of high levels of renewable energy featured CSP as an important component of high-renewable energy scenarios. This was in part because many of the […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: