On 10 January 2019, the utility frequency of Europe’s power grid dropped to 49.8 hertz. Many factors contributed to the near-blackout that evening, but the incident is not the only one in recent weeks that has shaken the grid.
But, let’s start at the beginning: Thursday , 10 January 2019 at 21:00 CET. Europe was settling in for a relaxing evening on the couch in warmly-lit living rooms at the end of the workday. At that moment, however, things were far from relaxed for employees at European grid stability observation posts. In less than two minutes, the utility frequency plunged toward the critical threshold of 49.8 hertz across Europe and was threatening to sink even further.
It was evident that primary control reserves alone would not be enough to halt the drop, and reserve power plants could not be started quickly enough. The French transmission system operator, RTE, sprang into action, issuing emergency rolling blackout requests to all of France’s 22 large industrial demand response consumers that reduced power consumption by 1,500 MW. That is roughly the amount of power required to keep the lights on in the greater Lyon region. The utility frequency stabilized and was heading back toward normal levels at 21:10; by 21:25, the frequency was back to 50 hertz. Living rooms in Europe remained warm and bright.
Potential production causes of the frequency disruption
The greater European power system is a complex network that can be significantly impacted by several factors. What may seem like a harmless event can, when combined with other harmless events, destabilize the system and cause problems large and small. If large, stabilizing elements go down, the more difficult it is for the rest of the system to absorb the failure.
While a comprehensive analysis of the events of 10 January 2019 is still pending from ENTSO-E, we can draw our own conclusions about the frequency drop based on information available to the public and from our own systems. In the interest of creating more transparency about what happened, we will list these in chronological order.
According to ENTSO-E, 140 MW of available power dropped off the grid at 20:00 from boiler 1 of Spain’s Litoral coal-fired power plant near Almeria. It is not certain, if this had an impact on the utility frequency collapse that occurred about an hour later.
At the French nuclear power plant Penly near Dieppe, ENTSO-E states that reactor 2 went offline at 20:26, taking 850 MW available capacity out of 1,330 MW installed capacity off the grid.
Energy sector news service montelnews.com quoted TenneT press spokesperson Matthias Fischer, who said there was “a short but significant peak of power consumption” when pumps were activated at pumped-storage hydro plants, including the 1.1 GW Goldisthal facility in Thuringia (Germany)
Additionally, another root cause of the frequency drop can be observed every day at the top of the hour. This is when the ‘hourly handoff ’ takes place, regularly causing a frequency fluctuation of 0.1 hertz in the morning and evening hours. It may not sound like much, but it's half of the 0.2 hertz frequency tolerance of alternating current. The hourly handoff occurs at the ‘shift change,’ when one set of large power plants hands over power producing duties to another group of plants.
Part of this can be chalked up to power trading that is still, in some cases, conducted in one-hour blocks. At the top of the hour, the first set of power plants shuts down power production as quickly as possible to hit the handoff deadline and to avoid ‘over-producing.’
The incoming power plants try and take over power delivery as late as possible – payment for power produced begins at the agreed upon time, not before. The issue occurs during the ramp times. If turbines and generators are still ‘ramping’ up or down, they are not at full capacity, which causes a dip in the frequency. Controllable renewable energies such as bioenergy, aggregated in virtual power plants, can provide capacity much quicker – just sayin’.
Become part of the VPP
Let’s take a look at how we observed the frequency timeline at our company headquarters in Cologne. From there, we had full visibility of the European grid imbalance: Point 1 on the graph represents the failure at the Spanish coal plant Litoral, which marks the beginning of the grid frequency disruption. Point 2 marks the failure of the Penly 2 reactor at the nuclear power plant near Dieppe. The grid still manages to cover these shortfalls.
But, at point number 3 – 21:00 – the pumps at Goldisthal kick in. At this point, the situation in the grid was already unstable, and the frequency drops rapidly – until France’s RTE calls for 1,500 MW of rolling blackouts. TSOs in Germany react as well, calling on emergency reserves including 153 MW from the pumped-storage hydro plant Herdecke near Dortmund. But why were the pumps at Goldisthal switched on during such a precarious situation in the first place?
Did a data error cause the frequency failure?
The cause may lie with a data error from the TSO TenneT, as Vienna’s “Der Standard” newspaper reports. The article contains a hypothesis presented by the Austrian TSO Austrian Power Grid (APG): The local control system misinterpreted the situation on the grid, which caused the grid controller to deliver incorrect data and set into motion a series of mostly automated actions – based on false data – that drove the grid frequency downward.
Due to the ongoing investigation, further insights from TenneT on this case are not available. But our experts have also concluded that a grid controller could have been part of the problem. These fully-automated control mechanisms continually measure grid frequency at pre-defined spots, such as border transfer points. If frequency deviates in either direction, the grid controller automatically sends operating commands to power producing and consuming units that are providing control reserve. If a controller malfunctions or delivers imprecise data – an extremely rare occurrence with up to four built-in redundancies – grid frequency can deviate by a little or a lot.
Nevertheless, the massive frequency collapse was the result of a chain of events with multiple causes:
Failures at large power plants in France and Spain destabilized the integrated European grid.
A grid controller data error from the TSO TenneT starts up the pumps at the Goldisthal pumped hydro plant – even though the necessary power is not available in the grid.
The normal hourly handoff between large power plants causes a predictable frequency fluctuation of 0.1 hertz.
A feed-in shortfall, which can possibly be traced back to the data error.
Why weren’t the power market’s reserves enough?
Fluctuations in the European high-voltage grid are designed by the power market to be regulated by primary control reserve. It kicks in instantly, if the frequency recorded directly at an asset providing control reserve fluctuates up or down within a certain tolerance (dead band). We provide primary control reserve in our Virtual Power Plant and could quickly see that all the available control reserve was used up.
So why wasn’t it enough? After all, in the pan-European pool of primary control reserve providers, we’re a relatively small fish, and there are plenty of large power facilities contributing to the mix. The online portal netzfrequenzmessung.de, which tracks Europe's grid frequency, has come up with a prevailing theoretical analysis that cites two possible reasons:
“The discrepancy in the grid's capacity balance was greater than the primary control reserve capacity at the time (more than 2.6 GW).” [...] “Services for providing primary control reserve were not delivered by many power producers during this time."
However, the platform does not provide a reason why these power plants did not fulfill their obligation to provide primary control reserve. In any case, it is reasonable to conclude that providing the necessary primary control reserve to stabilize the grid may have gone against other interests of the plant operators. To put it another way: Providing near-instantaneous control reserve was simply not attractive enough in terms of price.
Why isn't the grid settling back down?
In recent months, stakeholders in the electricity markets are having a tough time shaking the feeling that the usually-stable European grid has fallen into disarray. Three major and several minor incidents have gotten a lot of attention. In addition to the previously described situation on 10 January 2019, there have also been the following incidents:
Rolling blackouts in accordance with Germany’s levy for interruptible loads (AblaV) in December 2018
On 14 December 2018, an incorrect forecast for photovoltaic feed-in led to a grid imbalance. This was corrected by reducing power consumption at certain industrial facilities in accordance with a government levy (AblaV) regarding interruptible loads that can be taken offline to stabilize the grid. Under AblaV, companies are compensated for reducing their power consumption while the measure is in force. We documented the impact of this event in a previous blog article (German only). To summarize: Tempers flared, particularly in the heavy industrial sector, and the FAZ newspaper penned a fiery article.
Dangerous grid frequency increase on 24 January 2019
At 6:00 on 24 January 2019, grid frequency swung in the other direction: Across Europe, the measured frequency was 50.2 hertz. In some local instances, the frequency exceeded the upper limit for safe grid operation. Observers and the industry press spoke of an extraordinary anomaly but have yet to provide an explanation that holds any water. Once again, the 50.2 hertz outlier took place during the hourly handoff.
Renewable energies were hardly involved in this event, if at all, and our experts believe the blame lies in power trading. A power surplus on the grid was recorded the previous day during the hourly handoff at 6:00 as the early shift began at the large power plants. A decline in negative secondary control reserve was also observed, which flooded the grid with more power and drove up the frequency.
We can’t be sure what lead to these unusual events on the grid in the last few weeks. However, we can come up with some theories that are not entirely hypothetical – in some cases, the issues described are already causing concrete problems.
One thing is clear from a hardware perspective: The European grid is a highly complex system, with very small and very large elements working together to stabilize grid frequency. Even with all the safety measures and redundancies, it is still possible that single elements can cause this complex structure to fail. The consequences of a failure are closely tied to the size of the unit that causes it.
In other words, if a biogas plant producing a single megawatt hour fails, the grid won’t crash. It’s a different story when the 1,300 MW from reactor 2 at Penly suddenly disappear. Decentralized power production contains an inherent degree of theoretical system security – in a swarm of units, units that fail can be compensated. Almost every year during the winter months, the lack of flexibility and the cluster risk of nuclear-heavy power production contributes to load fluctuation in France’s very centralized grid.
The approach to grid reserves has changed
For the recent grid disruptions, no smoking gun has yet emerged. However, one fact cannot be overlooked: The last serious Europe-wide grid disruption prior to December 2018 occurred on 4 November 2006. A botched response to a planned shutdown of 380-kV very-high voltage lines in East Friesland provided, in that case, a clear, identifiable cause.
The current disruptions, on the other hand, are much more difficult to evaluate and cannot be fully explained by single events such as power plant failures or grid controller errors. It is therefore likely that there are additional reasons for the fluctuations that have nothing to do with the physical grid. One theory is the introduction of mixed-pricing methods on the control reserve market. The resulting consequences – quickly-exhausted reserves, waning commitment from balancing groups – are destabilizing the entire system.
One thing is certain: This time it wasn’t the renewables. The green scapegoat often takes the heat for grid problems, but in the events of January 2019, there were no significant forecasting deviations for wind, solar, biogas, or hydro power.