The power market design column - The largest machine in the world does not have a driver

Texas was hit by a winter storm in February. Power supply to millions of Texans was cut off for hours or even days with catastrophic consequences. TenneT indicated in an article in a Dutch newspaper that such a crisis in the Netherlands is unlikely as Texas is an electric island , while the Netherlands is part of the largest electricity system in the world. If it is tight in the Netherlands, the neighbours automatically jump in. But on January 8th, that system malfunctioned. One incident in Croatia led to a split of the European system in no time. Northwest Europe had a power shortage and the frequency fall rapidly and sharply. So what's the deal? Does this European interconnected system increase reliability or is it a possible source of failure?

(This article is a translated and edited version of an article that was published earlier at Energeia.)

The Texas disaster

A disturbance in power supply can come from a failure of components or human failure. In case of a major disturbance, normally more things went wrong at the same time. Texas is a typical example of a disturbance in which many components failed simultaneously due to extreme weather; low temperatures for which the power plants, and more importantly, the gas infrastructure was not designed for. A total of about 30 GW of gas-fired power plants was unavailable. That's  half the peak demand. That’s is simply far too much. ERCOT, the system operator, had to intervene and shut down large parts of the state to avoid a full blackout. The outage is similar to last year's problems in California - which were caused by high temperatures - but the extent of the Texas outage was many times greater. The interruptions sometimes lasted several days. The consequences for the Texans - who could not heat their homes while the water supply and sewerage system were no longer working - were much greater. Dozens have died and President Biden has declared Texas a disaster zone. There is also a battleground within the sector. Producers who were unable to meet their obligations and suppliers who had to buy on the spot market for their delivery paid a price of USD 9000/MWh. Bankruptcies have already been declared. Seven ERCOT commissioners resigned and those who stayed fired the CEO. More information about the disaster can be found in this publication of the IEA.

History: the sleet disaster of 1987

The Dutch energy supply is of course much more resilient to low temperatures. But there are no guarantees. The climate is changing. Why should a week of temperatures of minus 30 0C be impossible? After all, Texas is located at the same latitude as North Africa and the Netherlands is at the same latitude as Canada! Well, 30 degrees below zero is very unlikely, but a lot of freezing rain combined with strong winds can also have catastrophic consequences, because high-voltage lines and even high voltage pylons can then break. The freezing rain disaster of 2 March 1987 is a good example in this respect. And a cyberattack on electricity supply isn't just happening in thrillers.

A large interconnected system means lower costs

The Texan power grid itself seems to have suffered little from the winterstorm and the blackouts were therefore only the result of the loss of generation capacity. So TenneT's contention that the situation would have been much less serious if Texas would have had strong interconnection with other states seems to be right. But that conclusion can be commented on. If an interconnector is added to an existing system, the reliability of that system will naturally increase. But there are dynamic effects. There would have been much less generation capacity in Texas if it would have had interconnectors. The possibility of import means that less generation capacity would be needed in Texas, in order to achieve the desired level of reliability. These interconnectors therefore do not so much lead to higher reliability, but mainly bring cost advantages. The desired level of reliability is achieved with less installed generation capacity. Still, TenneT's  contention in the specific case of Texas, does apply.  If there would had been less capacity installed, then probably fewer power plants would have failed, while imports would have been possible. A greater variety of resources means that the system becomes less vulnerable to systematic errors.

A large interconnected system means that failures can be imported too

If the main benefit of the large European electricity grid is above all lower costs, then the question is justified whether failures in neigbouring countries can also lead to problems in the Netherlands.

The presence of interconnectors almost always leads to import dependency at some times. This will be the case more often for net importers than for net exporters. A country such as Luxembourg - with 85% imports - almost always depends on imports. But the net exporter will also sometimes depend on imports. After all, the market is looking for a dynamic equilibrium and the possibility of imports will, to some extent, crowd out domestic generation or storage capacity. This effect becomes stronger the more interconnector capacity there is. The Netherlands has gradually expanded its interconnection capacity over the years. And, for example on 21 January, all three HVDC cables to Norway, Denmark and the United Kingdom were out of operation. Coincidence exists, but it does feed the concern about security of supply and import dependence.

On January 8th earlier this year, things went wrong. At least, almost wrong. One busbar coupler in a substation in Croatia became overloaded, and in no time the European system split into two parts. The Northwest European part had a power shortage and the frequency decreased rapidly and sharply. The primary or frequency containment reserve was activated, but that was not enough. In France and Italy, for example, no less than 1.7 GW of load (contracted emergency power) was switched off automatically. With this type of additional emergency measures, the balance was restored and more than an hour later the European synchronous system was restored.

A malfunction due to human faults

Meanwhile, ENTSO-E, the European association of TSOs, has published an interim report of the incident. First, it is noticeable that the malfunction occurred without any obvious reason. No single contingency! No short circuit, no lightning strikes, no line breakage, no power plant outages. There was quite a lot of transport from South-East to central Europe, but that was planned and the measured flows did not deviate much from the planned flows. Still, it resulted in an overload of the busbar coupler in the Ernestinova substation.

A failure of this busbar coupler is not included in the network security calculations. Such a contingency is considered too unlikely. In these calculations, the n-1 criterion is applied, i.e. one component must be able to fail without it being allowed to result in overloads elsewhere in the grid. So the grid was not n-1 securely operated. But much more serious is that the grid was also not operated n-0 secure! There was no contingency but the current through the busbar coupler passed a critical boundary and was automatically switched off. This led to overloads on other lines that then also switched off. A snowball effect followed after which suddenly South-East and North-West Europe were no longer synchronously coupled. The South-East had a surplus of production and North-West-Europe had a deficit of no less than 6.3 GW.

The entire synchronous system keeps 3 GW of frequency containment reserve, which must be fully activated within 30 seconds at a frequency decrease of 200 mHz. Now, a deficit of 6.3 GW had to be met with only part of the European system. So obviously the frequency fell further below 49.8 Hz. ENTSO-E is therefore right to note that the emergency measures worked well, because in the end the decrease was stopped at 49.74 Hz. Automatic load shedding only starts at 49 Hz, so fortunately that limit was not met by far.

The lessons of November 4, 2006

The disturbance of January 8, 2021 has many similarities to that of November 4, 2006. Also then, human failure caused a split of the synchronous system. A line over the river Ems was deliberately switched off to allow a large ship to pass through. The n-1 criterion was not adhered to and communication between the TSOs was insufficient.  However, this disturbance was even more severe than that of January 8. In 2006, the West European part had a deficit of no less than 9.5 GW and the sub-system with the generation deficit, was a lot smaller than on 8 January. The frequency therefore decreased to the critical lower limit of 49 Hz, which led to load shedding.

In this respect, it is worthwhile to look back at the Italian blackout of September 28, 2003. The incident started with a trip of an overhead line in Switzerland. After the tripping of a 2nd line, the then created overloads on the remaining lines in the area became intolerable. The remaining interconnectors towards Italy tripped and the Italian power system was isolated with a deficit of more than 6 GW. A few minutes later the system collapsed, when the frequency reached the threshold of 47.5 Hz.

The report of the 2006 disturbance had five recommendations. It seems that these recommendations have certainly led to improvements, in particular with regard to communication between TSOs. The TSOs now also had better insights in what was happening than in 2006. The so-called ENTSO-E  Awareness System  showed that the system had been split into two parts. On the other hand, in 2006, the TSOs were able to resynchronize the system within 38 minutes , a lot faster than in the recent outage. But the fact that the network security calculations were not in order again, will have to be looked at. The role of the so-called Regional Security Coordinators and the sharing of tasks between these RSCs and the TSOs will certainly be examined. Also regulators will have to ask themselves whether they should be pro-actively monitoring whether the procedures are being carried out properly rather than waiting for another failure.

Why didn't the Netherlands make an extra emergency contribution?

It is clear that the additional emergency measures, taken after the frequency dropped below 49.8 Hz, have been essential to prevent a European blackout. The Italian and French TSOs have contracts in place with industrial users to automatically switch off load in case of emergency. In France this is 1,300 MW that switches off at 49.82 Hz and in Italy it is 400 MW at 49.75 Hz. In Germany and a few other countries, pump storage plants, if in pumping mode, are automatically switched off. That function is mandatory and the plant operator receives no financial compensation. It is unclear why so different approaches apply. The idea of the synchronous system is based on solidarity, with each country making an equivalent contribution to frequency control. The French and Italian TSOs incur costs, which are passed to their consumers. François and Giulia can therefore rightly complain that Jan and Anneke do not have these costs. 

Contributions from the United Kingdom and, in particular, the Nordic system have also been important. These systems are connected through HVDC cables and the frequency in those systems is therefore not equal to the frequency in the continental European system. But some of the HVDC cables apply a function that ensures that additional power is delivered when the frequency on one side of the interconnector drops sharply. The Nordic system delivered 535 MW and the United Kingdom 60 MW. But this feature is not in operation on BritNed and NorNed. Should the Netherlands ever unexpectedly get into island operation, there will be no automatic support via these HVDC cables.

The myth of the interconnector

The aforementioned report on the major disturbance of 4 November 2006 contains a striking passage. The report states that the European interconnected system had been developed in the past for the purpose of security of supply and the possibility of mutual assistance. And that only in recent years the grid would be used for large-scale transport for commercial reasons. That is an example of historical falsification. In reality, economic arguments have always played an important role. In the anniversary book "The 50 Year Success Story – Evolution of a European Interconnected Grid", the TSOs themselves wrote the following: "Fuel economy was the central focus of the joint work undertaken during the first phase of reconstruction, which was still marked by the effects of the war. The main objective [...] was to ensure the optimum operation of electric power plants. For example, a surplus of production in countries where generation was based mainly upon hydroelectric facilities might be used to balance a shortfall in production beyond the frontiers of those countries, thereby allowing savings in coal consumption to be achieved in the neighbouring countries concerned." 
I rest my case.

Texas and lessons for flexibility

Back to Texas. In the aforementioned article in the Dutch newspaper, TenneT also indicates that more flexibility will be needed in the future because Nederland will move from a demand-driven energy system to a mainly supply-driven system. The author of the article points out that this flexibility was lacking in Texas. But that remains to be seen. Especially since Texas is an electric island, with solar and wind, it can be assumed that there is quite a lot of flexibility. As in the Netherlands, the market design is an energy only market,  i.e. without capacity mechanism. This design aims to require market participants to react on (expected) prices when investing or divesting in capacity or to increase the flexibility of existing assets. The market price is therefore set at USD 9000/MWh in times of physical scarcity. Although, there are claims that ERCOT kept this high price for a too long time after the rolling outages were already ended.

Also interesting is that some household customers had opted for a dynamic price. It's probably a relatively small group. For example, 29 000 households had opted for a variable rate  plan with the supplier Griddy. These consumers had to pay $9 plus a surcharge per kWh. These consumers faced a bill of several thousand dollars for the month of February. Delaying a few runs with the washing machine is obviously not a problem, but if it freezes outside and you can only heat your poorly insulated house with electric heaters, then the meter turns fast.

The two crucial conditions for greater flexibility in the system are free formation of prices and exposure to these prices. In times of scarcity, the wholesale price must be able to peak. Not only industrial customers but also households must face variable prices. The European directive states that suppliers are obliged to offer a dynamic price contract. This means that the customer pays the hourly day-ahead market price with a surcharge. This is a fairly popular product in Norway, but not yet in the Netherlands.

So European rules promote a contract that has caused quite a stir in Texas. Dutch suppliers and, in particular, consumers must therefore know what they are getting into and, above all, look for smart solutions. Suppliers can also offer products where, for example, in the winter months all hours between 16h00 and 22h00 are settled at an extra high rate, possibly supplemented by an extra low rate in the summer months between 11h00 and 15h00. Contracts are also conceivable in which only the charging of the e-car or the electric boiler is smartly controlled. Consumption by other essential devices does not have to lead to unexpectedly sky-high bills.

Next to monitoring security of supply, TenneT is also monitoring flexibility. But this makes little sense because flexibility cannot be measured. Assessing the potential for flexible capacity is also an impossible task. Firstly, everything has a price and secondly, it is difficult to estimate how the cost of flexibility will develop through the deployment of new technologies. But what TSOs or regulators can monitor is how the retail market develops. How many suppliers offer dynamic price contracts? What types of dynamic pricing contracts are being offered? How many households have opted for a dynamic price contract? And of course the more basic questions: What about the roll-out of the smart meter? Is the process of settlement (allocation of consumption to suppliers) and reconciliation fully suitable for the correct settlement of dynamic price contracts.

Finally, a few words about monitoring security of supply. The Dutch monitor released annually by TenneT is useful but also has a limited value. A country like Luxembourg only looks at their neighbours. Will the Belgian nuclear power stations be closed down or not? What is the impact of the Kohleausstieg in Germany? And how much does consumption in France increase on a cold winter's day? The Netherlands is, of course, not Luxembourg, but at the same time only forms a small part of the European market and is fully intertwined in this market with interconnectors to five countries. Therefore, it is wiser to pay more attention to the security of supply monitor created annually by ENTSO-E. The mid-term adequacy forecast is a monitor that is made annually, covers the whole of Europe and looks ten years ahead.

The Texas disaster has also taught us that monitoring security of supply should look not only at electricity but also at gas supply, transport infrastructure and their interrelations. If there is no power, the central heating boiler does not work anyway, while the availability of gas-fired power plants is essential . And why does TenneT not dedicate a single word to the reliability of the grid in its security of supply monitor? 

Conclusions

• The Netherlands is not Texas. But here too, things can go badly wrong and so things will go wrong. Perhaps not due to extreme cold, but for example an extreme freezing rain or a cyberattack can also cause a disaster.
• The interconnected European system offers great advantages for the Netherlands, especially economic ones. The desired level of reliability is achieved with lower costs. The synchronously interconnected system is very robust. On 8 January, a sudden power shortage of 6.5 GW was absorbed without reaching the critical lower limit of 49 Hz, where load is automatically switched off. 
• At the same time, it is worrying that the grid could quickly split into two parts without any apparent reason. The system is not designed for a failure of 6.5 GW of power. It seems that grid security calculations were not in order, which means that the grid was not operated secure.
• It is striking that the Dutch system does not make an extra contribution if the frequency falls below 49.8 Hz. In particular, the installation of a mutual assistance function on NorNed and BritNed should be considered.
• The outage on 8 January has shown that the interconnected system can also be vulnerable. The TSOs will have to keep the operational planning and operation of the grid at a high level. No central authority is responsible for this. The largest machine on earth does not have a driver who can oversee the whole from one center. The responsibility lies with many national TSOs, which therefore have to work closely together.
• Monitoring security of supply remains important. But the focus should be more on European monitoring. National authorities can focus more on analysing the proper functioning of the national market, for example regarding the use of dynamic price contracts in the retail market. Finally, it is not only electricity but also gas supply, transport infrastructure and interrelations that need to be looked at.

This is my 30th column on power market design issues. The earlier columns covered the following topics: the importance of ACER, Flexibility and foisonnement, reliability and load shedding, regulation of congestion income, dynamic network tariffs, energy communities, scarcity pricing, the Florence Forum, active system management, network planning & sector coupling, off-shore assets, intraday capacity hoarding and pricing, interconnectors, international comparison of market designs, cross-border capacity calculation, flexibility, cross-border capacity, electric time and unintended exchanges, EU Network Codes, price formation and zero marginal cost generation, simplicity in the Clean Energy Package, smart grids, storage, auto-generation, balancing, VoLL, demand side response, interconnectors and the Economist on market design.

Picture: Courtesy Ercot

Disclaimer: The views as expressed in this column do not necessarily reflect the views of Statkraft

Paul Giesbertz

paul.giesbertz@statkraft.com