3.2 The Successful Integration of Offshore Wind Energy into the British Power Grid may Require a Miracle.

Monday, 29 January 2024: 2:00 PM
347/348 (The Baltimore Convention Center)
Kevin F Forbes, Energy and Environmental Data Science, Malahide, D, Ireland

A high-voltage electric power system is only stable and secure if the system operator can ensure that electricity supply is well matched with demand. In the case of Great Britain, this requires that generators adhere to the schedules they reported to the system operator one hour before real-time. Errors in the generation schedules necessitate balancing actions by the system operator to stabilize the grid. Unfortunately, these balancing actions are becoming less effective because of declining system inertia, which, as the name suggests, tends to stabilize the power grid. Consistent with this view, the system operator in Great Britain is increasingly challenged to maintain system frequency within its operational and statutory limits. For example, on August 9, 2019, a series of unfortunate events resulted in measured system frequency plunging to 48.8 Hz, a value well below the statutory limit the system operator must maintain. The total collapse of the power grid was only avoided by disconnecting about one million electricity customers during the middle of rush hour. To put it mildly, the consequences were chaotic but could be part of the new normal unless the power grid’s operational performance is improved.

With the above background information in mind, it is noted that Britain’s energy future includes large planned increases in offshore wind energy generation but that the operating performance of the offshore wind farms currently in operation is not encouraging. Specifically, the operating performance of these wind farms is vastly inferior to electricity generated using natural gas, pumped storage, or nuclear. Indicative of this, there is almost no visual relationship in Figure 1 between the ex-ante scheduled and actual generation* at the London Array, a 175-turbine 630 MW wind farm located 20 kilometers off the Kent coast in the outer Thames Estuary (Note: actual generation* in this figure equals metered generation plus the quantity of electricity not produced per the instructions of the system operator).

Improved modeling of offshore wind energy conditions can improve matters. Specifically, this paper presents evidence from a machine learning time-series model that places more weight on expected meteorological conditions and also represents the dynamics of wind energy generation). Using this model, more accurate predictions of wind energy generation at the wind farm level are found to be possible.

Unfortunately, the system operator in Great Britain has chosen to measure the errors in wind energy generation using a flawed metric that makes the average error appear small, especially when the capacity factor is low. Indicative of its flaw, applying the system operator’s preferred metric to the data presented in Figure 1 would have the reader believe that the generation scheduling error is about 10 %, while a conventional calculation of the error indicates that the scheduling error is about 27.2%. The view here is that the 10% value is not consistent with the scatter diagram depicted in the figure. Unfortunately, managerial decision makers who are unaware of the flaw in the metric may be inclined to believe that the error is on the moderate side but not terrible.

The system operator has been informed of the deficiency of its preferred metric, but there is no evidence of any change in its thinking. Instead, its current metric, is informally defended on the basis that it is following the guidance of the wind energy community. There is also the point that the system operator, using this visually flawed metric, has publicly reported its high level of satisfaction with its ability to predict wind energy generation. In short, an error metric that makes the average error appear small is likely to foster complacency.

Under these circumstances, it may take a miracle to achieve offshore wind energy’s potential contribution to the energy transition in Great Britain, even though the evidence presented in this paper indicates that the technical challenges to offshore wind energy development are real but solvable. Revised advice from the wind energy community would obviate the need for a “miracle,” but wind energy forecasters appear to be quite firm in their unsubstantiated belief that weighting the errors by capacity is the best approach. The suspicion here is that wind energy forecasters have the incentive to prefer capacity weighting because it makes the reported error appear small regardless of the reality. Forecasters are likely completely unaware of how this approach to error reporting can foster complacency in power grid operations that will ironically most likely prevent offshore wind energy generation in Great Britain from achieving its potential.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner