Utilities weather

Using situational intelligence to create business value

Integrating weather data into utility operations can save millions in costs, the Weather Company demonstrates.
Published: Fri 11 Jan 2019

Utilities traditionally have not needed to take much notice of the weather in their day-to-day operations. But that is now changing. In an era of growing use of solar and wind energy, both highly weather dependent, weather forecasts are becoming increasingly important for optimising the utilisation of such renewable generation and with that a growing number of other use cases.

And new tecniques such as artificial intelligence and advanced analytics are bringing new levels of speed and accuracy to the forecasting process.

“It’s about using data to make better business decisions,” says James Burn, Meteorologist at the Weather Company, one of a growing number of companies that are providing weather service capabilities for the sector.

In an Engerati webinar, Burn and Rob D’Arienzo, Senior Meteorologist Global Strategy Leader at IBM, give some insights on how the Weather Company gathers and models weather data to provide forecasting at a resolution of 500m on 15-minute intervals and the use cases to which this can be applied.

“Weather forecasting isn’t an exact science, but we believe our forecasts to be very accurate and relatively a couple percentage points ahead of others,” says Burns. Indeed, the company claims to be moving towards the most accurate local weather forecasting ever seen.

Weather and utilities

In addition to renewables integration, other areas where weather impacts utilities include the increasing prevalence of extreme events such as hurricanes and storms, and grid modernisation and the management challenges it is bringing with the growth of technologies such as storage and electric vehicles.

As an indication of the importance of weather data for modern utility operations, a survey of 1,000 c-suite executives globally by IBM revealed that all of them believed that weather impacted at least one cost and one revenue metric in their organisation.

“Of the utility executives, almost all believed that they could reduce their annual operating costs by up to 2% or more simply by harnessing better weather intelligence,” says D’Arienzo. However, the key challenges were felt to be inaccuracies and unavailability of such data and how to integrate it into their operations and translate it into insights and decision making.

“The weather data is obviously just one part of the solution and we combine it with machine learning and advanced analytics all in the cloud to create a powerful combination,” he continues. “One needs to have the solution to take all the data and convert it into business value.”

Weather use cases

Explaining that the Weather Company’s forecasting draws on data from thousands of weather stations and other sources such as national meteorological organisations around the globe, Burn outlines four sets of offerings. These are weather data packages including weather APIs and an operations dashboard, outage management including prediction and restoration and vegetation management, renewables and load forecasting and trading, and weather alerting for Internet of Things and field work safety applications.

“The data packages are the building blocks of the solutions and include historical, current, intermediate and the far future up to months and even decades ahead,” says Burn. He adds that information types, in addition to the forecasts, include site-based data, reports, imagery and probabilistic data.

To these may be added other data sets, such as in the vegetation management use case, which also may draw on geospatial satellite imagery as well as local data such as tree types and locations and growth patterns. In addition, new so far untapped data sources are planned for incorporation, including sensor readings from aircraft.

“These are then presented on the operations dashboard which has been optimised for high demand operational use.”

Utility case studies

As an example of the implementation of the solution, D’Arienzo cites Canadian utility New Brunswick Power, which found 70-80% accuracy in the outage prediction model based on forecast up to 72 hours ahead and potential savings of millions of dollars through optimisation of utility field worker mobilisations with that for a single event of the order of $500,000 or more.

In the energy forecasting use case, the same utility achieved a 26% improvement in load forecasting accuracy equating to an estimated saving of $1m per year, with other savings of $1-6m per year in improved hydro forecasting and $300,000-400,000 in wind power forecasting.

Noting that trials of the operations dashboard can be set up within a matter of hours while the use cases have varying degrees of complexity and personalisations to implement, Burn comments: “We see these use cases as very exciting for utilities.”