Forecasting renewables: an evolving science

Renewable energy forecasting keeps up with advances in climate science and technology.


Weather-based business forecasting has become a serious business. IBM recently partnered with The Weather Company to crunch 2.2 billion unique forecast worldwide and generate more than 10 million daily forecasts for businesses affected by weather. The Climate Corporation processes 50 terabytes of weather data daily to provide farmers with hyper-local forecasts used to plan planting  now and 25 years from now.

From the Home Depot store manager predicting snow shovel sales to energy executives calculating future renewable investments, advancements in forecasting technology are having a tremendous businesses impact . The nature of forecasts is that outcomes are sometimes far from what actually transpires. Still, no prudent investor or utility will get involved in a renewables project without some pretty serious prognostication.

There’s a rough continuum from higher to lower certainty in making predictions. For the forecasting done to determine the viability of prospective wind or solar farms, we can look backward and do a solid job of projecting forward to predict potential availability of sun and wind at a given site.

Current forecasting systems also do a good job of helping asset managers and maintenance crews schedule service more efficiently. We see less of the run-to-failure or fixed-interval approaches  to decide when to dispatch crews. Instead, asset managers are increasingly assessing actual asset health information to forecast probable mean time between failure. That enables service at the time when it’s really needed but before the turbine blades stop spinning or the PV panel dies.

Short-term forecasts of renewable energy availability continue to improve due to more sophisticated sensors, monitoring and data processing capabilities, and highly granular fundamental climate predictions. Both renewable-energy providers and utilities need to know how much power can be produced and available in the coming hours in order to optimize generation and ensure a sufficient supply of power.

As the time frame moves out from hours to decades, we enter the realm of big-picture, long-term forecasting that is far more challenging. It’s tough to answer questions like “What is the value of this wind farm over the next 25 years?” with a great deal of certainty. Which isn’t to say that this kind of forecasting has no value.

It does a tremendous job of providing scenarios related to technical and commercial factors. What happens to profitability if there’s a breakthrough in battery technology, like the ultrafast-charging battery recently announced by Stanford?[1] What happens if the portfolio of available renewables goes up 20 percent in 10 years? Forecasting helps answer some really interesting questions about what should be done to best profit from renewables investment.

Forecasting will never be perfect, but it can be very accurate. Even when it isn’t, it is helpful for planning around the volatility and uncertainty of policy, weather, prices and a host of other issues. Considering the critical need to predict commercial, technical and operational aspects of renewable energy generation, forecasting is certainly something to that investors, planners, and operators need to devote considerable attention.

[1] http://blogs.scientificamerican.com/plugged-in/2015/04/14/stanford-researchers-unveil-new-ultrafast-charging-aluminum-ion-battery/