"That gives us a picture of the global flow scenario," explained Lars Christensen, the Vestas VP responsible for turbine placement and monitoring. "Those models are then cobbled to smaller models for the regional level called mesoscale models. The mesoscale models are used to establish our own huge wind library so we can pinpoint a specific location at a specific time of day and tell what the weather was like."
How detailed is the library? In its early versions, the library could give details for a grid with 27-by-27 kilometer sides (about 282 square miles). But with improvements in computational flow models, Vestas could massively increase the resolution to 10-by-10 meters-just over 1,075 feet, which is about the footprint of an average American home.
With that kind of modeling detail, the Vestas engineers needed to improve the model even further, and the best way to do that was to add data. And that's what Christensen's team did: they planned to increase the wind library tenfold with more weather data over a longer period of time.
So now the IT challenge was significant, because Vestas not only had to accept and store all of the incoming data, it had to be able to analyze that data and all of the historical facts in a timely manner. Initially, that was hard to do. Vestas would have to wait up to three weeks whenever they ran a potential site report.
"In our development strategy, we see growing our library in the range of 18 to 24 petabytes of data," Christensen said. "And while it's fairly easy to build that library, we needed to make sure that we could gain knowledge from that data."
Big data for big wind
Ultimately, Vestas decided to turn to IBM for help. Big Blue's InfoSphere BigInsights software, IBM's Hadoop-based big data solution, turned out to be exactly what Vestas needed.
The new big data framework enabled a serious increase of data storage right off the bat. Vestas was able to increase the data resolution of those 27-square kilometer grids down to three square kilometers, without modeling. According to Christensen, this tightening of resolution drops about a month off the up-front pre-site development timetable.
Those three-week forecast analysis reports are a thing of the past, too.
"Before, it could take us three weeks to get a response to some of our questions simply because we had to process a lot of data," Christensen said. "We expect that we can get answers for the same questions now in 15 minutes."
With the time to build forecasts reduced by 97 percent, Vestas gained a significant edge over competitors, because they could get the jump on forecasting an area when talking to potential customers. Plus, since the reports were based on more accurate data models, the level of returns on turbines increased, even as the level of initial investment decreased.
Like many big data deployments, this wasn't simply a case of Vestas saying "we need a big data solution, regardless of cost." The conclusion that big data would be able to help came only after Vesta's sales and research teams started collaborating together to work on the age-old problem of increasing revenue. It quickly became apparent that by fine-tuning their data, Vestas would have a very good chance to improve. Looking at the potential gains, versus the real costs of deploying IBM's Hadoop solution, Vestas decided the expenditure would be worth it. Their decision has paid off.
And, with the increasing production of wind-generated energy, companies like Vestas can use all the savings they can get.
This article, "Turbine company knows which way the wind blows at your house," was originally published at ITworld. For the latest IT news, analysis and how-tos, follow ITworld on Twitter and Facebook.