About the author:
Professor Janusz Bialek is Professor of Power and Energy Systems at Newcastle University, UK.
- Email: janusz.bialek@ncl.ac.uk
- Profile: januszbialek.html
9th of August Power Outage on GB system
UPDATED to include reference to the authors, Energy Policy Research Group working paper with Cambridge University¹
The power outage on 9th August 2019 that affected over 1 million customers in England and Wales and caused a major disruption to other critical infrastructures was a major news item and sparked wide-spread discussions about who is to blame. Power outages are like stress tests exposing strengths and weaknesses of the power system as the whole and its constituent elements and other critical infrastructures connected to it so our main aim is to consider the title question: what does the power outage tell us about the state of GB power system?
A uniformly accepted (N-1) reliability criterion stipules that there should be enough fast power reserves to respond to a loss of one power station, as the probability of two power stations simultaneously failing is very low. On 19 August a lightning strike caused two power stations to trip, so it was (N-2) event. Consequently, frequency dropped below the statutory limits to 48.8 Hz which triggered under-frequency load shedding. Frequency was then returned to 50 Hz in about 5 mins and power supplies were restored within 40 mins. The main adverse effect of the blackout was a severe disruption to rail service around London due to an unexpected failure of trains when frequency dropped below 49 Hz. Hence, everything seemed fine as the power system itself responded exactly how it was designed to. Should we then be happy about the state of the GB power system? The answer is: not really. The blackout has uncovered important fault lines which may significantly affect reliability of the system in a near future.
Changing landscape
Over the last 10 years or so the GB power system has changed quite rapidly and significantly with renewables, often embedded in the distribution level, replacing traditional gas/coal generation and increasing deployment of energy storage, active demand and smart grids technologies. To put in simple terms, it means that a lot of new gear and controls were added to the system in a very short time. Hence it is increasingly difficult for the Electricity System Operator (ESO) to fully monitor, model and control the whole system. As a consequence, the probability of hidden common modes of failures, affecting one than more unit, has increased – as exemplified by the 9 August outage. This would suggest that it might be prudent to strengthen the old (N-1) security standard by providing extra security margin.
There were also other issues highlighted by the outage. Embedded generation reached such a high penetration level that it cannot be treated any longer as negative demand. Its importance for real-time power balancing and in a response to disturbances requires a new approach. Traditional under-frequency load shedding disconnects indiscriminately all customers on the disconnected feeders, including embedded generation and frequency response units which are essential for the system to survive. With rapid advances in telecommunication, it should be possible to assess in real time the actual loading on individual feeders so that load shedding has the maximum possible effect and perhaps also implement load shedding at 11 kV level, rather than 33 kV, hence allowing more selective operation.
Lessons learned
As power systems are more likely to be affected by large disturbances due to the reasons outlined above, the ability of critical infrastructures and services to ride through the disturbances has to be closely monitored and tested. Not only back-up supplies have to be regularly checked but also compliance with the regulations must be enforced to make sure that the infrastructures can survive large frequency deviations.
Finally a question arises why some GB outages that affected hundreds of thousands of people over the last two decades attracted a public attention and media coverage and others did not. Our conclusion is that short-duration outages matter only if they affect critical infrastructures, especially transport, in London and the surrounding areas. What really matters to the public is not the number of people affected by a power outage but how the disturbance affects their life. Hence if a disturbance is of a relatively short-duration and does not disrupt significantly critical infrastructures, it does not attract much attention. Also outages affecting metropolitan areas such as London are more likely to attract the attention of media than those happening elsewhere.