Meeting demand is a requirement in every electricity system in developed countries. But it has a very high cost because the non-storability of electricity requires the installation of large production capacities, some of which are rarely used. Why not instead go through supply cuts based on consumers' willingness to pay?
To understand the value of reducing consumption, let us look at the incident that occurred in early 2019 on the European synchronous grid. On January 10, at 9:02 p.m., the European continental power system, which includes 26 countries, recorded for 9 seconds the highest absolute frequency deviation since 2006. The frequency, which must be maintained in a very narrow band around 50 Hz, has dropped to 49.8 Hz. This fall is due to the superposition of two elements, one routine and the other exceptional.
Let's start with the routine. Every day, there is a gradual increase in electricity demand in the morning, followed by a decrease, and at the end of the day, a second increase followed by a decrease to reach a minimum during the night. During the day, these two peaks of demand move from east to west on the continental plate, with some compensation or aggravation depending on the regions concerned from one country to another. On the supply side, as markets are hourly-scheduled, at each hour change, power plants are activated and others are slowed or shut down under the control of national power system managers. In addition, there are uncontrollable variations in renewable energies. As a result, demand, which varies continuously, and supply, which varies more in blocks, are never exactly equal, especially at the time changes corresponding to peak periods. As long as the gap between supply and demand is not too large, the frequency remains within the expected limits. However, sometimes significant differences can reduce frequency quality (e. g. +173 mHz on 24 January 2019 at 6am), which can disrupt the operation of electrical equipment that consumes and produces electricity. The number, duration, intensity and speed of occurrence of these deviations have increased over the past decade or so, a phenomenon that system managers are striving to correct through compliance with guidelines and better collaboration. The fact is that a decrease in the frequency on January 10 at 9pm was not in itself a surprising event.
But an exceptional event accentuated this routine deviation ("deterministic" in the language of the network). It consisted of the following elements: (i) at the beginning of the day on 9 January, wind power production in Germany was high enough (up to 34 GW) for part of it (723 MW) to be exported to Austria via the four interconnections of TenneT DE and APG. This value was recorded by the continental system management software. But (ii) at 13:25, a telecommunications line transmitting information on cross-border flows stopped working (an incident that will only be detected on 11 January) and system management software that did not receive new information continued to record an export of 723 MW from Germany to Austria. In fact, (iii) at the end of the 9th and during the 10th, German wind power production fell sharply (4 GW), thus exports to Austria too, even going so far as to reverse with an Austrian export of 330 MW on the four connecting lines with TenneT DE. For more than 24 hours, the managers of the European system therefore observed a decrease in the frequency without knowing its origin and sought to remedy it on the basis of erroneous information. The correction was made on January 11.
The 50 Hz has been restored thanks to the activation of the primary reserve (batteries, pumping stations, coal-fired power plants) in all European grid countries and, in France, by a temporary interruption of 1700 MW of the supply to certain large consumers whose contract provides for this possibility. The activation of the service interruption clause is considered a last resort. It is decided by the operator of the electricity system (in France, RTE), unlike demand decreases, which are chosen by consumers or private operators to whom they have delegated the control of their consumption.
This interruptible service was created by the NOME Act (December 7, 2010) and is now incorporated into the Energy Code (art. L 321-19). Eligible sites are those with an extraction capacity exceeding 25 MW and capable of interrupting their extraction in less than 5 or 30 seconds. Candidates respond to a call for tenders issued by RTE. Those selected sign a contract by which they accept interruptions decided by RTE "when the normal functioning of the public transmission network is seriously and immediately threatened or requires calls on mobilisable reserves". As compensation for these public service constraints, approved final consumers with an instant interruption profile are paid up to a maximum of €120 per kilowatt per year.
Advantages of service priority
Service interruption is a brutal form of what economists call "priority service". This is a product differentiation method that endogenizes product quality in a menu offered to customers. Quality is represented by the reliability of the delivery service. Depending on the urgency of the needs to be met, there are consumers who are willing to pay a high price for the product on a specific date and time. Others, less hurried, agree to wait but only if they pay a lower price. The probability of being served is therefore an essential element in determining the applicable tariff policy.
Priority service can be seen as a form of demand rationing when demand exceeds supply. However, it is not the only way to ration demand. For example, the grid operator may use random supply cuts. This approach, apparently more "fair" than priority service, is less economically efficient: unlike priority service, it can lead to a situation where a consumer who is willing to pay a high price to avoid a supply disruption suffers one, while a consumer who has little value for being served immediately is. Priority service is not only preferable to random rationing from a collective point of view, but also from an individual point of view. More specifically, it is possible to set up a priority pricing menu such that no consumer is harmed by a switch from random rationing to priority service.
Priority service can also be seen as a particular form of market organisation, which should be compared to the spot market. For residential consumers to be able to participate in the latter, they should be able to observe instant prices and adjust their demand on an ongoing basis or have access to technologies allowing them to program their responses to all possible market fluctuations. Priority service achieves the same result at a lower cost by exploiting the fact that the optimal rationing rule corresponds to the order of priority dictated by consumers' willingness to pay.
Finally, the priority service provides information on the distribution of consumer payment provisions that can guide investment decisions in production capacity. It makes it possible to infer the value placed by consumers on an increase in the reliability of the service (i.e. the probability that there will be no interruption of supply) that would result from an increase in production capacity. The spot market is essentially an algorithm for determining an allocation of resources at a given time and therefore does not allow information on demand for reliability to be inferred.
Although it was the loss of a telecommunications line that caused the incident on January 10, ENTSO-E's analysis of this incident highlighted the gradual decline in the quality of the electrical frequency on the continental grid due to daily changes in the electrical mix that were not adapted to continuous changes in demand. Technical solutions exist: more coordination between network operators, more primary reserves, more flexibility in generation technologies. However, the economic solution of expanding the offer of priority services should be examined. While it is true that the development of information and communication technologies already gives consumers, not just the largest ones, greater flexibility in their electricity extraction from the grid, when it comes to responding within a few seconds to a crisis situation and thus creating a positive externality on the network, the best solution is to rely on the system manager with whom you have signed a contract to consent to service interruptions for a fee.
 Central European Time.
 All operators use the same tolerance margins which, if exceeded, trigger appropriate variations in production or consumption. For example, if the frequency exceeds 49.95 Hz-50.05 Hz for more than 20 minutes, the corrective measures are more drastic than if the duration is only 15 minutes. If the frequency falls below 47.5 Hz or exceeds 51.5 Hz, all electrical devices are disconnected.
 The following technical elements are taken from the report of the European Network of Transmission Companies (ENTSO-E) "Continental Europe significant frequency deviations - January 2019", published in April 2019. According to the report, a third deviation caused by the Serbia-Macedonia-Montenegro system did not play a role in the decline of 10 January because it is permanent.
 H.P. Chao and R. Wilson (1987), “Priority Service: Pricing, Investment, and Market Organization”, The American Economic Review, Vol. 77, n° 5, December, 899-916.