UK Power Outage
Findings and Lessons Learnt
On Friday 9th August 2019, just before 5pm, large parts of England and Wales were left without electricity following an unprecedented power outage that, among other things, had a severe effect on the country’s transport network. When power was lost just before 5pm, approximately one million people were impacted. Some of the busiest train stations closed during the Friday evening rush hour, while around 300,000 homes and businesses were left without electricity. It was about two hours later when the National Grid Electricity System Operator (NG ESO) confirmed that the issue had been resolved and the system had returned to normal. In this article, I would like to share my thoughts about the significant events that took place on the GB electricity system that day, as well as some of the lessons that can be drawn from them.
How the disturbance occurred
The incident was triggered by a lightning strike which occurred on a transmission circuit (Eaton Socon – Wymondley Main) at 16.52pm. Prior to this, the electricity system was operating as normal, with the demand being similar to that of other summer days and roughly 30% of the total generation coming from wind farms. Other important generation sources were gas, nuclear and interconnectors.
Generation sources on 9th August 2019 (ESO Technical Report – Appendices_0.pdf)
According to NG ESO’s technical report on the events, the protection and delayed auto-reclosure systems on the affected transmission line (whose settings I had produced back in 2015) operated correctly to clear the fault caused by the lightning strike and restore the line in the expected time. However, immediately following the lightning strike there was an unplanned and near-simultaneous tripping of two large generators that connect directly to NG transmission network to export their power, i.e. one gas-fired power station (RWE’s Little Barford) and one offshore wind farm (Orsted’s Hornsea). The total generation lost from these two transmission connected generators (ca. 1.4GW) represented just under 4% of GB’s electricity demand at the time and was above the automatic back-up power response available to NG ESO to manage frequency events (e.g. from battery storage plants). As a result, there was a rapid fall in the grid frequency of the electricity system (normally 50Hz) that dropped to below 48.9Hz, whereas the lower acceptable limit is 49.5Hz. This in turn resulted in automatic intentional disconnection of an isolated part of electricity demand (ca. 5%) in order to protect the rest of the system from a further frequency drop and prevent a larger power outage.
Lessons Learnt
Obviously for a widespread power cut like this, there were a number of factors that did not work well or even not at all. The ESO has published a list of Lessons Learned in their final report about the incident, stating that it is necessary to review the relevant processes, procedures and technical standards that are (or may not be yet) in place for managing events like the one on the 9th of August. Issues like the available level of automatic back-up power response, Loss of Mains protection of embedded generation and communication arrangements between the relevant parties are extensively discussed in the ESO report.
My main comment is about the Grid Code compliance process that every new power generation plant in the UK has to go through before commissioning. This process is run by the ESO and defines, among other things, how the generator should operate during and after a network fault to maintain a stable parallel operation with the grid. It should be noted that it is the responsibility of the generators themselves to demonstrate compliance with the requirmements of the Grid Code by submitting evidence in the form of system studies, simulations and testing results.
To start, lets first look at what Orsted’s technical report regarding the events of 9th of August has pointed out, namely, that “the de-load was caused by an unexpected wind farm control system response, due to an insufficiently damped electrical resonance in the sub-synchronous frequency range, which was triggered by the initial event”.
In line with the Grid Code’s compliance process, Hornsea’s control systems had been modelled during the development stage and physically tested prior to the wind farm’s commissioning to ensure their conformance to all specified requirements, including site-specific criteria stipulated in the bilateral connection agreement.
Avoiding “unexpected responses”
Now, do we want to reach a state where there will be no ”unexpected” control system responses of transmission connected generators? If yes, then we should perhaps define more than just the process that needs to be followed by generators to self-demonstrate Grid Code compliance. Having worked in the power systems consultancy arena for years, I have noticed a lack of consistency in the methodologies employed by different consultants for the various grid compliance studies, which can sometimes lead to oversights, misinterpretations of the Grid Code requirements or suboptimal design decisions. In addition, dynamic modelling requirements and the accuracy of the models used to represent the generator’s performance are often a subject of conflict between the system operators and equipment manufacturers mainly related to intellectual property. And we should not forget the issue of what software packages are being used for the required analysis, with many of them having considerable inherent weaknesses which are not directly obvious to the reader of a study report.
A clear framework
All the above issues could be alleviated by establishing a framework that would clearly define the tasks and responsibilities of both the generator and the ESO, as well as the levels of surveillance of the latter, for verifying the plant’s compliance to the Grid Code requirements. A good example of a similar framework is what NG has established for the preparation, application, verification, dissemination, recording and auditing of protection and control relay settings in the transmission system. More specifically, all settings for transmission protection relays are produced by TP141 trained and authorised engineers, an authorisation granted by NG.
During this process, TP141 engineers must be well familiar with and strictly follow the dozens of policy statements, technical guidance notes, technical specifications, design handbooks, commissioning handbooks and engineering bulletins constantly published and maintained by NG for different types of equipment. By a funny coincidence, the TP141 relay settings framework was initiated following the events of the 2003 London power outage, which was basically caused by wrong relay settings.
Following the example from the transmission protection and control relay settings, I believe that NG ESO should be looking to enforce and administer a more rigorous and systematic framework for undertaking the grid compliance studies of generators that are using the transmission system, a framework that will ensure that the relevant studies are undertaken by suitably trained engineers who follow a consistent methodology and use pre-approved software packages. The ESO should also play a more determinant role in the assurance of these studies.
Of course, this is not an easy process and will require collaboration of many different parties, including equipment manufacturers. But as the focus of renewable energy penetration in the UK is gradually being shifted to large offshore wind farms, having an established Grid Code compliance framework for new generators is going to become more and more significant in a collective effort to protect the grid from equally large single-point failures.
February 2020
George Peltekis
Leader of the Power System Analysis team at Fichtner Consulting Engineers Limited