Time to head outside and enjoy the sunshine. Most of us can hardly think of anything else when the Dutch weather institute (KNMI) predicts a sun-soaked day. But for the congestion officers at Enexis, that’s exactly when things heat up. On days like these, their job is to prevent overloads on the electricity grid.
Congestion Management at Enexis
Recently, Enexis has been relying on large customers who return a lot of power to the grid. When there’s a risk of overload in a specific part of the network, the congestion officers ask these customers to temporarily shut down (part of) their solar panels. A tool—developed by their colleagues from the data science department—tells the congestion officers exactly which customer needs to shut down how many panels, and at what time.
Shutting Down Solar Panels
If too few panels are shut down as a precaution, it can lead to power outages and all the consequences that come with them. But shutting down a few extra panels just to be safe isn’t an option either, as Enexis has contractual agreements with these customers about compensation for lost energy production. And in the end, that adds up to a significant amount.
Dream Team
So, no pressure—but tool and team need to operate like a dream team. That wasn’t always the case, according to Martijn Tilma, manager of the data science department. “The congestion officers felt the tool didn’t fully meet their needs, while my team believed they had developed the best possible solution based on the business requirements.”
Neutral Expert
Tilma realized intervention was needed: the usability and output of the tool had to be critically reviewed—but not by the teams themselves. “Due to the high pressure of being part of a new process with massive stakes, it was hard for the congestion officers to maintain the necessary distance. At the same time, we data scientists tend to cling to the feeling of being right—without actually being acknowledged as right. That’s why I felt a neutral expert would help.” He brought in Pipple for the job.
The Impact of Timing on Error Margins
Pipple’s technical assignment at Enexis involved researching ways to improve congestion forecasting. This led to a complex, labor-intensive analysis to determine how variations in weather affected the error margins of predictions.
Tilma explains: “Predicting how much sun solar panels collect at night is easy: none. It gets tricky once the sun comes up, because every weather forecast already carries a wide margin of error. That unpredictability increases the more precisely you try to pinpoint timing. So even on a predicted cloudy day, it’s still mostly guesswork as to when exactly a cloud will block a particular panel. And the impact of that margin of error differs—it’s bigger in summer than in seasons with little sun.”
Interviews
Enexis also asked Pipple to uncover the gap between the needs of the congestion officers and the solutions the data team could provide. Pipple conducted interviews, which yielded more than just data, according to Tilma. “Wouter and Joep are great, approachable guys with knowledge and tact. They asked some tough questions but still managed to build a bridge between the bits and bytes on my work floor and the reality faced by the congestion officers. They also spoke the language of senior management and the board.”
Making Reality SMART
The advice and data analysis Pipple delivered were very useful. But what Tilma valued most was the process that led to those insights. “It really brought people together. We’re no longer reacting to incidents and details, but steering based on a shared, zoomed-out reality that Pipple made concrete and SMART for us. The business’s wishes are now clearly defined, as is what we deliver. Talking to each other about the right things has become much easier as a result.
