May 8, 2016 was a day that garnered much discussion. On that day, Germany’s solar and wind power peaked at 11 am local time, enabling renewables to supply 54.8 GW at a time when demand—according to provisional data from Agora Energiewende, a research institute in Berlin—was running at 57.8 GW. In plain terms, Germany, albeit briefly, got almost all of its power from renewable sources—and roughly 40 months earlier than industry experts expected.
Meanwhile, just as striking but less widely noticed, Portugal has recently run 107 consecutive hours with 100 percent renewable generation. Alongside these landmarks, we cannot neglect to mention the world record for wind penetration held by Denmark, which managed to fulfill 42 percent of its energy consumption with wind in 2015. Nevertheless, while some transmission system operators (TSOs), distribution system operators (DSOs) and power producers are doing an impressive job of inverting the fossil-versus-renewable energy mix, there is still room for improvement. Indeed, it is questionable whether utilities across Europe are ready to reach the EU guidelines for 20 percent of energy to come from renewable sources by 2020.
Wind and sunshine—depending on the region in Europe—are potentially abundant. However, the big challenge is managing their intermittency. Given this, one of the keys to increasing renewables penetration will be improving the accuracy of forecasting and controlling volatility. A more accurate forecast contributes to day-to-day operational effectiveness, through advances such as more effective day-ahead planning (including calculation of required reserves, congestion management and so on), and improved asset management. Greater precision around wind and solar forecasts also underpins business success, as it can help renewables operators make better decisions around energy production and how much they can trade. Going back to the German example, due to market-wide oversupply that day in May 2016, power prices turned negative during several 15-minute periods, dropping as low as minus €130.07 per megawatt-hour (according to data from Agora Energiewende).
In the area of wind power forecasting, there are multiple tools to choose from: Some are proprietary to the utility, some developed by academic organizations, and some are recognized products from technology vendors. These tools are all largely powered by algorithms carefully cultivated on the basis of historical data. And since each one seemingly has its specific strengths, most utilities will have more than one. The question executives are asking themselves is this: Which forecasting tool could help them most significantly improve their estimates, and thus further optimize wind and solar power generation?
As utilities look to answer this question, the good news is that they may not need to rely on a single tool to master the weather. Digital technology techniques such as data analytics and machine learning can be applied to operational data, potentially enabling utilities to develop a more intelligent forecasting approach. For example, by combining the results from multiple tools, data scientists can assess each forecast’s accuracy over short- or long-term horizons, and/or according to different scenarios, such as high-wind conditions. As such, analytics approaches can enable utilities to develop a smarter combination. And over time, by continuing to apply analytics techniques, the outcome would be tweaked to deliver increasing accuracy.
Nevertheless, if it was as easy as applying an analytics application, it seems likely that some utilities would have already tested this option. But the smart combination is not just about combining the forecasts—it’s also about bringing together the appropriate people. More effective wind and solar power forecasts require bringing experts to the table from operations, IT, and data science backgrounds. The results can be powerful, but it takes time for this collaboration to be effective. Through the journey of analyzing, interpreting and comparing the data, the team needs pool together their collective expertise, methods and insights, and bridge differences in language and experience.
By integrating information and operational technologies and data, as well as having practitioners aligned to those domains, utilities can generate a more accurate forecasting approach. Like the code on a safe, achieving the smart combination requires lots of numbers and a sharp focus on unlocking the guarded prize. For those that succeed, the results will be worth the effort.
Guest author, Stéphanie Lakkis, MSc
An engineer in the Grid Operations team, Stéphanie leads Wind Power Forecast Optimization at OMNETRIC Group. Working largely with European transmission system operators and distribution system operators, she collaborates with OMNETRIC Group's data scientist team to identify and develop analytics use cases for improved forecasting.