The Energy Collective Group

This group brings together the best thinkers on energy and climate. Join us for smart, insightful posts and conversations about where the energy industry is and where it is going.

9,904 Members


The Optimal Share of Intermittent Renewables


  • Any modelling study of higher penetration wind and solar scenarios must take into account the cost of intermittency.
  • This article describes one such study which estimates the optimal long-term share of onshore wind and solar PV in north-west Europe at around 20% and 2% respectively.
  • A CO2 price of €20/ton was assumed and technology costs were assumed to fall by 30% for onshore wind and 60% for solar PV.


As someone who makes a living from the modelling of complex non-linear systems (multiphase reactors), I have great appreciation for the work of Lion Hirth at the Postdam-Institute for Climate Impact Research, Vattenfall GmbH. As any good modelling study should, his work on the north-west European power system (covering Germany, Belgium, the Netherlands, Poland and France) reveals and elegantly explains a number of seemingly counter-intuitive effects regarding higher penetration renewable energy futures.

Hirth has authored numerous papers on the subject of valuing intermittent renewables at higher penetrations, but his paper on the optimal share of intermittent renewables is arguably the most insightful. This paper calculates the welfare-optimal deployment mix for north-west Europe, taking into account the negative effect that intermittency has on the value of wind and solar at higher penetrations (discussed in a previous article). This article will give a brief summary of this very interesting paper.

Effect of intermittency on market value

Hirth identifies three easily understandable integration costs of intermittent renewables at higher penetrations: profile costs (effect of timing), balancing costs (effect of forecast errors) and grid-related costs (effect of location). A typical effect of these costs is shown below.

Hirth - integration costs of variable renewables

Profile costs are primarily related to the underutilization of power infrastructure as a result of intermittent electricity surges enjoying priority dispatch. This is typically the largest cost related to the intermittency of wind and solar. Balancing costs are related to the spinning reserve requirement to prevent blackouts from forecasting errors and are typically reasonably small. Grid-related costs arise from the necessity to bring wind/solar power from windy/sunny locations to population centres and varies greatly from one case to another.

For north-west Europe, the effect of these costs on the value of onshore wind relative to the wholesale power price is shown below. Note that the long-term trendline assumes that the entire power system is optimized for increased wind penetration and is therefore only applicable beyond 2030. The mid-term line takes into account the sunk investments in existing capital stock.

Hirth - value of intermittent wind at increasing penetration

Optimal wind and solar shares

The paper details a comprehensive sensitivity analysis to understand how the optimal share of intermittent renewables changes in response to various parameters. The most important of these will be briefly discussed below. Note that all results are long-term results (following the blue line in the figure above).

Storage and interconnection

Energy storage and increased interconnectivity (i.e. the supergrid) are commonly quoted as solutions to the variability of solar and wind. However, the study found that both these options had fairly small effects on the optimal share of wind power (shown below).

Hirth - effect of storage and interconnection

In the case of storage, this was simply because wind variations typically occur on a much longer timescale than the storage capacity of the pumped-hydro facilities considered in the model. Further increases in the degree of interconnection beyond the current level also had minimal effects. It was also found that interconnection is substantially cheaper than pumped hydro storage.

Climate policy

The base-case in the study assumes a CO2 price of €20/ton which is fairly representative of the scientifically calculated social cost of carbon. However, the CO2 price has the potential to vary greatly over coming decades and therefore deserves some detailed investigation.

As shown below, the effect of CO2 price on the optimal share of intermittent renewables is highly counter-intuitive. Under the assumption that onshore wind costs fall by 30% and solar PV costs fall by 60%, the optimal share of intermittent renewables rises to about 25% at a CO2 price of €40/ton and falls thereafter.

Hirth - effect of CO2 price

The trend in the above figure is the result of solar and wind being displaced by the baseload low-carbon technologies of nuclear and CCS. At high CO2 prices, these baseload technologies become cheaper than renewables balanced by unabated thermal powerplants. It should be noted, however, that only baseload lignite with CCS was modelled in the study and the picture could look better for intermittent renewables if gas+CCS or even hard coal+CCS was considered.

The study also looked at the effect of societal resistance to nuclear and CCS. As shown below, the optimal share of wind and solar would increase greatly if no nuclear or CCS was deployed under a scenario with a €100/ton price on CO2.

Hirth - impact of low carbon technology restrictions

Taking this option will have some severe consequences though. Firstly, the electricity cost would increase by a further 15-35% over the already high cost of such a high CO2 price scenario. More importantly though, CO2 emissions would increase by 100-200% over the scenario where all low-carbon technologies are deployed. This is the scenario that opponents of renewable energy technology-forcing are most concerned about.

Fuel price

Three alternative fuel price scenarios were considered: a doubling of coal prices, a doubling of gas prices and a European shale gas boom similar to that of the US. As shown below, a doubling of the coal price gives the expected (although fairly small) increase in optimal wind penetration, while both gas price scenarios result in reduced optimal wind deployment.

Hirth - effect of fuel price

The shale gas scenario is self-explanatory, but the double gas price scenario requires some further explanation. Although a higher gas price would make wind more competitive against gas, it would also cause a shift from gas to coal as is currently happening in Germany. Balancing intermittent wind with coal is more expensive than balancing with gas, thus leading to the counter-intuitive reduction in the optimal wind share with increased gas prices.

Summary of scenarios

The optimal wind share after a 30% cost reduction from today’s levels is shown below for all the scenarios considered in the paper. Most of the scenarios fall in a fairly narrow range.

Hirth - optimal wind share in all scenarios

The four outlier scenarios can be briefly discussed in a little more detail. Firstly, the high optimal wind scenario with a high CO2 price and no deployment of nuclear and CCS will be highly undesirable from a climate viewpoint and should hopefully be avoided. The scenario where thermal plant investment costs increase by 50% and wind power investment costs decrease by 30% also returns a substantially higher optimal share of wind. However, since investment costs are most influenced by raw material costs, this scenario where wind and thermal plant costs diverge so sharply appears unlikely.

On the other side of the spectrum, we should all hope that the €0/ton CO2 price scenario is just a theoretical scenario that will not persist in the long-term. The shale gas scenario, however, is possible within the timeframes of these results (beyond 2030) and should not be discounted.

Discussion of other assumptions

The paper states that the different model simplifications should add up to a moderate downwards bias on the estimated optimal share of intermittent renewables, implying that these results can be read as conservative estimates. To balance this perceived bias, the study considers optimistic price reductions of 30% for wind and 60% for solar. According to forecasts from Fraunhofer, onshore wind costs will stay essentially constant between now and 2030, while solar PV costs will drop by about 30%.

At current costs for onshore wind, the study finds that the optimal long-term share of wind drops from 20% to 2%. It should be noted, however, that the total system costs will only increase by 6% if onshore wind deployment is lifted to 30% at current technology costs (below), thus confirming onshore wind as an attractive technology in north-west Europe under an environment with an accurate price on CO2. However, onshore wind will probably be hampered by other forms of resistance long before it reaches 30% penetration.

Hirth - effect of non-ideal wind deployment

The optimal solar PV deployment will drop from 2% to 0% if costs only drop by 30% as opposed to the 60% assumed in this study, thus confirming solar PV as a fundamentally unattractive technology at latitudes above ~40 degrees where solar resource quality is low and, more importantly, seasonal variations are large and misaligned with seasonal demand.

In the medium-term, when taking into account the sunk investments in existing power infrastructure (especially nuclear), the optimal wind share drops from 20% to 7% (assuming 30% cost reduction), thus emphasizing the degree to which intermittent wind depends on thermal backup. Power infrastructure investments will therefore have to be planned very carefully if Europe wants to move towards a high-penetration wind future.

Schalk Cloete's picture

Thank Schalk for the Post!

Energy Central contributors share their experience and insights for the benefit of other Members (like you). Please show them your appreciation by leaving a comment, 'liking' this post, or following this Member.


Bas Gresnigt's picture
Bas Gresnigt on June 2, 2014

The past few years show already increasing gap between reality and Scenario B:

– Share of gas (expensive) is decreasing further behind that of coal (cheap). As prospects for shale gas are bad (bans), that won’t change fast.

– Gabriel’s attempt’s to increase the share of offshore wind in new installations (now only few percent) seemed doomed. 3-5 states want that those wind turbines will be installed on their ground.

– Solar capacity recently became bigger than wind (off- + on-shore). The difference is now already ~10%. It should go the other way around according to Scenario B.

– Gabriel’s plan to tax owners of solar installations also for the electricity they produce and consume themself, was rejected in parliament. That plan would have delayed the moment that it is beneficial to install solar while the FiT is below the whole sale price. At that moment Gabriel will loose the FiT as an effective tool to control the solar installation speed (in ~2017-2023 period).

If solar generate electricity of 2cent/KWh, it becomes highly economic to install more than needed. That will allow to minimize buying from the grid in winter.

Roger Arnold's picture
Roger Arnold on June 2, 2014

Yes, but the experience curve is a log-log graph of specific cost vs. production volume.  If the market stops growing, the cost stops dropping.  It’s that relationship that I’m concerned about.  Given that the market is still heavily dependent on subsidies, its continued growth is not something that can be taken as a “given”.

You make a good point about soft costs.  That’s indeed the area where the biggest potential gains are to be found — at least in the US.

Bas Gresnigt's picture
Bas Gresnigt on June 2, 2014

Your authors have weird ideas. I cite:
“..we argue that the consumption of self-produced electricity should be treated in the same manner as the consumption of grid-supplied electricity, i.e., the exemption from taxes, levies and other surcharges for the amount of self-produced PV electricity consumed in-house should be abolished…”

So if I clean my house and do not hire a cleaner, and that implies more cleaners are idle and need social benefits, than the value I produce by cleaning my house myself should be taxed. Same for painting, etc. etc.

Clayton Handleman's picture
Clayton Handleman on June 2, 2014

“Given that the market is still heavily dependent on subsidies, its continued growth is not something that can be taken as a “given”.”

Yes, for years I have thought that the exponential growth would outrun the subsidies needed to drive it.  Seems there is always someone stepping in to drive it to the next level.  Right now China is pushing pretty hard on the consumption side.  I have long been frustrated that the US hasn’t done more of the demand heavy lifting.  It would be wonderful if we ended up running the anchor leg and gave it the final push. 

If the EPA program significantly expands cap and trade, perhaps that will provide the needed boost.  Will be interesting to see how the numbers play out.

Bas Gresnigt's picture
Bas Gresnigt on June 2, 2014

I use the 35years ongoing trend of 8%/a decrease and assume that that long term trend continues for the next ~20years. That assumption is based on:

– Predicted increasing volumes of the PV-solar market. History shows that that implies decreasing production costs (more automation in the factory);
E.g. a 40 inch TV is endless more complex than a 21.5% yielding PV-panel. Still both cost about the same. Reason: TV’s are produced in bigger volumes and producers have decades of experience in how to produce TV efficiently. They had more time to develop very specific complex automated producton machines.

– Theoretical physics that showed that PV-cells could have a yield of up to 70%.

– The increasing yields in the labs (max. now ~45%) and in the field; on av. an increase of 0.5%/a. So the present field max. of 21,5% will become ~27% within a decade (standard PV-panels, no concentration).

– The increasing stream of publications that show cheaper materials, etc.

The last two items are thanks to improving knowledge of molecule and atom structures, as well as improved machines to monitor and handle those.

Schalk Cloete's picture
Schalk Cloete on June 2, 2014

Thanks Bas,

I think we understand each others’ viewpoints. Future costs of fully installed solar power is something that we will probably not agree on, but that is OK. You ascribe to the Moore’s law like Ramez Naam while I ascribe to the law of receding horizons as described in this article. As long as we understand where each other is coming from, I think we can have fruitful conversations below future articles. 

Incidentally, the reason I regularly challenge the aggressive technology-forcing policies of the Energiewende is not because I think it will ruin the German economy. After all, developed world citizens spend a tiny fraction of their disposible income on electricity and, if increased costs from wind/solar are distributed equitably, this can be successfully absorbed, especially in Germany with its unique willingnes to create much greater value than it consumes. 

My problem with technology-forcing of wind/solar is that it delays/dampens the true technology-neutral approach that we so desperately need. Germany might even achieve 80% renewable electricity by 2050, but the enormous megacities of the developing world and the enormous supply chains which must serve the material needs of these billions of people living shoulder-to-shoulder will probably still run on large centralized energy solutions. If we put off the development of truly scalable centralized clean energy solutions for too long and climate science proves to be correct, we may well end up with a number of terrible humanitarian disasters towards the middle of this century. 

Math Geurts's picture
Math Geurts on June 3, 2014

Every goverment needs courage and time to explain their people that some very proud policies at the end appear to be not that smart at all. You will be surprised to see how many “weird” ideas will be introduced in EEG 3.0. It will be helpfull that all German solar panel factories will be closed by that time.  

„Nach der EEG-Novelle ist vor der EEG-Novelle. Denn bereits jetzt ist klar, dass das EEG in zwei Jahren nochmals reformiert werden wird, um die von 2017 an vorgesehenen Auktionen zu regeln. Deswegen sollten wir mit diesem EEG – dem EEG 2.0 – schon Erfahrungen sammeln, die wir beim EEG 3.0 dringend benötigen werden“, sagt Dr. Patrick Graichen, Direktor von Agora Energiewende.”

 „Das EEG 3.0 muss ein Gesetz zur Synchronisierung von Stromangebot und Stromnachfrage werden. Das heißt einerseits: Erneuerbare-Energien-Anlagen sollten so ausgelegt werden, dass sich ihre Stromproduktion viel stärker als bisher an der Stromnachfrage orientiert. Andererseits muss sich die Stromnachfrage aber auch viel stärker am Stromangebot orientieren. Ohne ein solches flexibles Zusammenlaufen kommt es zu Verwerfungen am Strommarkt und die Energiewende wird unnötig teuer“

(Agora dd. juni 2014)

In English: it appeared to be very difficult to change the German FiT law but the real overhaul is foreseen for 2017.

Bas Gresnigt's picture
Bas Gresnigt on June 3, 2014

Regarding solar, the US does already oké, by financing far more PV-cell research than the EU.
Important as there is really a lot to win with PV-panels *).
Commercial panels now have yields of max. 22%. That can be up to >30% within a decade if cheap methods are developed to produce double junction cells (which is a good solvable physics/technical problem).
Also the use of other, much cheaper materials than Silicon or GaAs (some are developed already) may bring great progress toward LCOE of PV-solar <$20/MWh.

That research will bring the world far more than subsidies that create a bigger US market.
Especial as those big markets are already there and are becoming self propelling as the costs of PV-solar reaches grid parity in many countries.

*) The chips industry shows that limits can be shifted.
In ~2000 I visited a micro lab and the experts thought that the law of Moore would stop at chip details of 60nano meter. They considered that to be the max possible with present production technology (=shine light on details that should not be etched away). I agreed as the wave length of UV-light is substantial more than 60nm.

But now, using the same low cost production methods, the industry will produce chips with details 10 times smaller!  Partly thanks to a small company in USA that developed an Extreme UV light source.

Bas Gresnigt's picture
Bas Gresnigt on June 3, 2014

You give lot of data and also assumptions. Unfortunately not all is correct. Some:

“If Germany went to 50% percent RE, an EEG surcharge at about 11 c/kWh…”
As you can also see at Agora, the EEG surcharge for 2015 is expected to be lower. And it won’t rise much in the years thereafter.

The EEG surcharge will never reach 8c/kWh, probably not even 7c/kWh. In the 2020-2023 period it will start to decrease permanently despite the rising share of renewable (in 2020 35%, 2025 45%, 2030 50%).

Reasons: The expensive FiT’s ( >50c/kWh) of the first years reach the end of the 20/15 years guarantee period and continue to generate electricity at ~ whole sale prices. The new FiT’s are low (<10c/kWh).

You can download an Excel sheet at the page, and play yourself with different scenario’s and the consequences for the EEG surcharge.

“its energy-intensive heavy industries will be finding a haven in Poland, France, Czech Republic, etc.”
Germany’s heavy industry gets tax-excemptions. So much that our aluminum smelter could not compete and broke down. They pay a lower price for electricity, that is four times more reliable than the electricity of France. So they won’t move out at all.

So competition complained at Brussels and the EU started an investigation.
But Merkel is smart and powerful as well as her aids, so a compromise was reached that made the regulations even more complicated and doesn’t make a real difference.

…EEG energy bought by German utilities is sold at 1/4 of the price on the open market…
It is 1/2 – 1/3 of the price. And the energy after the guarantee period is bought/sold with some profit (not much as competition between the ~100utilities is heavy).

Also, in case of wind energy, each TWh/yr added is produced at a higher cost, as the sites with the best winds are used up, and going offshore produces energy at 2x onshore.
Simply wrong as also shown by the FiT’s for wind that went and go down gradually.
The FiT is the calculated and agreed cost price + a profit of 6%-7%.

Remember also that 4 German states oppose more off-shore wind as they want all wind turbines at their land. One state president (Sleeswijk-Holstein) stated that his state alone could accomodate all wind power that Germany needs. So the plan to increase the share of off-shore to 20% of wind produced energy will not succeed. I estimate it will stick below 5% (now a few percent).



` `

Get Published - Build a Following

The Energy Central Power Industry Network is based on one core idea - power industry professionals helping each other and advancing the industry by sharing and learning from each other.

If you have an experience or insight to share or have learned something from a conference or seminar, your peers and colleagues on Energy Central want to hear about it. It's also easy to share a link to an article you've liked or an industry resource that you think would be helpful.

                 Learn more about posting on Energy Central »