Bengt J. Olsson
Twitter: @bengtxyz
LinkedIn: beos
With the increasing amount of renewable and weather-dependent power, firming or balancing power becomes more important. The question is: which are the major contributing balancing power sources in the pan-European power system?
One way to measure the contribution from balancing powers in the power mix is to look at how they vary with the variation of residual load. The residual load is the remaining load after wind and solar power have been accounted for. Essentially, it’s the load that must be served by all the other power sources.
In this scenario, we analyze the aggregated power production and load for all of Europe during 2023 up to September 2024—almost two years. Data is sourced from EnergyCharts. EnergyCharts calculates Residual Load as (Total Load – Onshore Wind – Offshore Wind – Solar Power). However, we define Residual Load here as the sum of the complementary power sources. These two methods should yield the same result, but in practice, slight differences may occur due to data imperfections.
We group power sources that contribute to serving the Residual Load to make them easier to overview. Data is sampled on different time scales to see how much balancing sources contribute at each scale. Some sources contribute little to balance on an hourly scale, but more on a seasonal basis, and vice versa.
Method
Some mathematical relationships:
- Residual Load = ∑ (“Complementary power production”) (that is all power production except wind and solar)
- Variance(Residual Load) = ∑ Covariance(Residual load, Complementary power component’s production) (summed over the components)
The variance of the residual load is hence the sum of the covariance between the residual load and each power source. This makes sense because the balancing power sources adjust their output to meet residual load needs. And thus this sum provides a structured way to measure each power source’s contribution.
This method has been used by the Swedish TSO Svenska Kraftnät in their modelling of a future Swedish power system. In the picture below the balancing contributions for each future scenario are scaled to a reference value of a probable scenario for 2025. The most remarkable scenario is the “EF” scenario for 2045 that shows a strong need for balancing. This scenario correspond to a power consumption need that is more than duplicated versus today, mainly because of the need for hydrogen production for the iron ore industries. In the EF scenario the VRE share is very high, the present nuclear power plants are phased out and no new built. The strongest balance contribution comes from import. (The EF and EP scenarios are further analyzed in this post).
Here we extend the analysis to include all of Europe’s power production and consumption. One difference is then that import/export on the national level are implicitly included when looking at the whole Europe like “one country”. That is, Europe as a whole can not count on using import/export outside Europe as a balancing power source.
Also note that arguments like “wind power produces more in the winter and thus follow load on an aggregated scale” is void here, since we are looking at residual load, that is, after wind and solar power have been accounted for.
Now some results.
Absolute Contributions from Balancing Power Sources
This graph shows how the individual power components “build up” the total residual load variance. It can also be noted that the residual load varies more on a shorter time scale than on a longer one, which is to be expected with less averaging at shorter timescales. With an hourly sampling period, the standard deviation of residual load is 46 GW. Averaging or sampling over months reduces the standard deviation to 23 GW. The total variation of residual load is roughly between 120 and 400 GW seen on an hourly scale (and between 200-300 GW on a weekly scale which can be seen in the dispatch graph above).
Note that balancing must be performed at the hourly, or shorter, time scale. The longer time-scales gives, however, an indication of the durability of the balancing power sources.
Relative Importance of Balancing Power Sources
This view is the same as the previous, the only difference is that the variance on all timescales have been scaled to unity, in order to more easily compare the balancing contributions.
More than half the contribution comes from fossil fuels. Pumped hydro contributes significantly to hourly balancing but less to longer time scales, as expected since it’s used for daily balancing. When battery storage will be included in the statistics, we can expect that it will look similar to pumped storage in the graph. Nuclear, on the other hand, shows less (but not zero) balancing contribution on the hourly scale but more on the seasonal scale. Essentially, nuclear “follows residual load” seasonally by being more available in the winter than in the summer.
For longer sampling periods (weeks and months), the “All Others” power source actually anti-correlates with the residual load. The main contributor here is Hydro run-of-river, which varies out of phase with the Residual Load over longer periods.
Key Takeaways
- Fossil Dependency: More than half of Europe’s balancing power comes from fossil fuels on all time scales. Eliminating this dependency requires either lowering the residual load variation or having non-fossil dispatchable power sources handle much larger variations.
- With a higher share of variable renewable power sources, it’s likely that the residual load variance instead will increase (see for example the graph from the Swedish TSO above), despite potential increases in load flexibility. Therefore, dispatchable power sources other than fossil fuels will need to step up, not only to replace fossil balancing power but also to increase the total balancing power.
- National vs. Pan-European Scale: On a national scale, increased import/export could provide additional balancing resources. However, on a pan-European scale, this becomes a zero sum game. Multiple countries may want to import or export simultaneously, limiting this source of flexibility.
- The Growing Gap: Europe faces a dilemma where residual load variation increases due to a higher share of variable renewable energy, but more than 50% of the balancing power, which is fossil, must be phased out. This creates a growing gap that needs to be addressed.
Update 2024-10-27
Upon a question in this LinkedIn thread, I divided the data into 2023 and 2024 YtD (to October), respectively. Although the statistics may be a little sparse, and we’re missing data for the power-hungry Q4 quarter of 2024, we can indeed see some trends. Here is the hourly and weekly sampled data:
Fossil fuel usage is decreasing as expected, with more solar and wind in the power mix. Nuclear, hydro reservoir, and pumped hydro are all increasing their balancing contributions to replace fossil fuels. The increase in nuclear may be a sign that nuclear power is starting to run in a new, more load-following pattern. We have seen this recently in the news from France.
It will be interesting to follow up on this in the coming years. Also, it would be really nice if battery charge/discharge data was included in the ENTSO-E statistics.
Update 2024-10-30
This blog post became an article in the Swedish online forum for the energy sector, Second Opinion. It is a Swedish version of the text:
Second Opinion article
Observera att det smugit in sig ett fel i SO artikeln, där det står: “den absoluta variationen ligger mellan ca 40 och 400 GW”. Det ska vara “mellan ca 120 och 400 GW”.