@RISK used by National Grid UK in Electricity Network Restoration Planning


National Grid


Energy & Utilities


@RISK & TopRank


Modellling electricity network restoration

Although total and even major partial system failures are rare, utilities still need to be prepared to respond to them by developing detailed and robust service restoration plans. And because taking real-time measurements of these plans is simply not feasible, some utilities have turned to computational modeling to simulate and measure restoration performance without interrupting service. National Grid UK, the electricity transmission system owner for England and Wales and the system operator for all of Great Britain, uses @RISK’s probabilistic modeling techniques to measure network restoration performance in a variety of simulated failure scenarios. In other words, if the system, or part of it, somehow breaks down, the people at the utility want to know what, and how long, it will take to get it working again, and what the challenges will be along the way.

National Grid plc, which also currently owns four of England’s eight gas distribution networks, actually started using @RISK in 2008 to address an industry concern on distribution network resilience requirements, specifically the usefulness of equipping key electricity substations with diesel and battery back-up capabilities. The answer, a resounding “yes,” led to an agreement between network owners and regulators resulting in a major investment and industry-wide deployment.

Simulating Restoration Performance

For the past several years, however, systems engineer Simon Waters, National Grid’s in-house modeling expert, focuses principally on restoration planning. His work revolves around using @RISK to simulate and measure network restoration performance for a range of failure scenarios using the key technical, operational and human factors associated with adverse sets of circumstances.

“Fortunately, electrical energy (MWh) and time (hours, minutes) are units that lend themselves to numerical representation, so it is all the other less obvious factors that are developed using @RISK functions,” Waters says.

Great Britain maintains a comprehensive package of plans to deal with a wide range of failure scenarios, ranging from extreme weather events to deliberate malicious attacks, including cyber attacks, and National Grid and other utilities must be able to respond to each quickly and effectively. Since most utilities cannot shut their systems down to perform restoration exercises, simulations using data from operational experience and industry-based rehearsals were for many years the sole source of information used to inform planning and decision-making. But the results were limited and lacking in detail. Today, at National Grid, @RISK is used for its sampling ability and to run simulations to map the overall restoration performance against a wide range of uncertainties and conditions.

“The ability to model a wide variety of uncertainties and to run Monte Carlo type simulations has given significant insight into what is effectively uncharted waters,” Waters says. “The results provide a credible set of performance measures against which power congestion points can be identified, investment decisions considered or plans adapted to optimize deployment and efficiency.” The work has been undertaken in collaboration with a range of industry parties and government departments to inform industry and formulate policy.


National Grid UK, the electricity transmission system owner for England and Wales and the system operator for all of Great Britain, uses @RISK’s probabilistic modeling techniques to measure network restoration performance in a variety of simulated failure scenarios.

Priming the Model

National Grid’s model captures all the pertinent engineering, operational and human factors associated with agreed industry restoration processes and the prevailing state of the market. Each model restoration simulation uses approximately 12,000 @RISK samples and a host of other data, and Waters typically runs 5,000 samples to map the wider solution space. The model picks at random an hour in the year in which a total network shutdown is assumed to have occurred. This is used as a key to select necessary data to prime the model with a wide variety of background information including national demand projections, the operational state of each generator set connected to the transmission system, the condition of the transmission network, the state of telecommunications equipment, the state of control room resources, the state of hydroelectric and pumped storage stations, the nature of wind generation and various other operational, failure rate statistics and external variables.

The model produces a range of output graphs and statistics that illustrate the wider performance envelope of the system’s restoration capability. These can be seen on both a national and regional level indicating how quickly electricity supplies may be reestablished. The model also provides a wealth of information on fuel requirements and how to optimize flexible generation resources. Changes in the generation fleet or regional plans can then be explored to optimize resources and/or measure their relative value ahead of time. Other output files produced include data on nationally unsupplied energy, impact to the nation’s GDP, how to best share resources across the system, and implications for particular generation sites.

Adapting to Future Needs

The project, Waters says, is very much an ongoing one, with recent work involving the measuring of variations in restoration performance from 2008 to 2016 in light of changes to a generation fleet that is moving away from coal and gas towards renewable energy sources like wind and solar power, as well as the undertaking of zonal comparisons and plan optimization to make the best use of available resources (i.e. how one area of the country compares to the other and how to make the best use of current resources). Moreover, with growing internal and external interest, Waters fully expects to be a busy man for the foreseeable future.

“We want to be able to prime the model with future market projections in areas like generation fleet changes, demand level trends, and to forecast likely future conditions as well.” He also hopes to be able to extend the analysis to capture the likelihood of an event, the cost/benefit of service options and possibly “least worst regret” analysis. Then there are the needs of government and regulators expressing interest in using the models to set and monitor a new national standard for restoration capability.

The National Grid restoration performance project provides insight into an area that no one wants to learn about first hand. And it all adds up to a lot of work indeed for something that everyone hopes will never have to be put into real-world practice. But it’s an absolutely vital exercise in this increasingly unpredictable day and age. Forewarned, as they say, is forearmed.

Join decision-makers around the world who