We know that while there can be grand plans for centralising data, often just getting started is the biggest obstacle. We’re going to be discussing this very topic following an insightful webinar Sennen hosted earlier this year on O&M data: “Driving the benefits from Offshore Wind O&M data – The dream, the reality and how to get started.”
This article will explore a specific segment of the webinar, particularly the “reality” side of O&M data. We will share and elaborate on the key challenges that our panel of Offshore Wind industry experts are facing when it comes to this particular topic.
These industry experts include Paul Grimshaw from Sennen, Christopher Gray from i4SEE TECH, Maggie McMillan from Miros Group and Justin Grimwade from RWE, who made-up the outstanding webinar panel.
Play the clip below to watch the full discussion of what Paul, Chris, Maggie and Justin’s biggest challenges in the world of Offshore Wind O&M Data.
Justin’s most prominent challenge when it comes to Offshore Wind O&M data is handling the multitude of data that gets generated from a number of different systems. He raises a number of questions, including ‘How old are the turbines? How do we physically get offshore to support them? Is it an SOV or CTV-operated site? How long will the sail be? Etc..’ All these questions are legitimate factors that need to be fed into the system to make sense of all the data.
My reality is that we’ve got lots of sites, lots of different turbines and lots of different HV systems and different people and so on. Therefore we have a lot of different systems producing lots of different data and the ability to compare apples with apples and so forth across all of them is quite difficult. So standardised in it in some way it’s of some particular benefit.Justin Grimwade from RWE
Moving onto Chris, his focus is to cross the gap between data analytics and operations, ultimately leveraging that information to be used in a manner that allows for tasks to be performed.
He goes on to share a quick anecdote, about how 87% of data science projects fail – the underlying lesson is to ensure you use information in an effective way that brings value. He shares that his biggest challenge is the productive use of information at a larger scale, highlighting the “big jump” between using data that could theoretically solve problems and working on solving real global issues across multiple operating sites.
The data is higher. The system is maintainable. It’s kind of automated data pipelines running on a daily basis where people are getting the information where they want it and how they want it. I think there’s a huge gap between those two scenarios.
It’s about performance analytics and condition monitoring but I think that whether it’s that or whether it’s monitoring wave heights to calculate vessel timings or managing human resources or health and safety all of those things I think to do them effectively at a big scale in an operational environment is very tough.”Christopher Gray from i4SEE TECH
As a result of this obstacle, Chris offers a potential solution, to break down complex problems into smaller incremental problems that can help make actions more manageable.
The layer-by-layer approach is quite problematic because it can take a very long time before you start to generate value that way. So I’m a really big fan of going after specific topics and keeping the focus. I’ve been doing this for the last few years and that’s been one of the most effective ways to reduce complexity.Christopher Gray from i4SEE TECH
According to Maggie, a lot of the Offshore Wind issues actually start during the application and planning phase. She highlights how owner developers are very much focused on the capex side of things rather than the opex side of things, pointing out how interfaces and solutions are somewhat an afterthought. This raises multiple questions, as almost 30-40 per cent of the energy costs of a project are actually during the operational phase.
I think there’s a gap and an education piece, where an earlier engagement with companies like ours can help inform that operational strategy from the beginning. You get that great value-added data from day one. When you look at smart data to inform O&M, from a safety perspective you now have this real-time data at your fingertips.
That can really help and inform the developers earlier on in the process to help them get those economic gains and increase safety on their projects.”Maggie McMillan from Miros Group
To conclude, Paul speaks of a “new” and rising logistical challenge, whereby having and accessing increasingly complex machines has become much more complicated to coordinate. The ideal scenario would be to automate these processes and allow for machine learning to then take place. He elaborates that if we want to learn, specifically through machine learning, then in order for the machine to learn from anything, it requires data. This becomes challenging when the data is not in place or doesn’t exist, because of outdated processes such as using Excel to plan shifts. Those processes aren’t really driven by a system.
The reality is that you need to start on that level and get all the processes automated… You have to get that into the process… The system has to really guide it to enable that data to just be harvested reliably day-in day-out no matter who is operating. I think that’s one of the key challenges we’ve seen; it’s just getting that really basic level. Get the bread and butter systems running, so you have a really great structured dataset that you can then start to take the next steps such as machine learning and AI to get smarter with that data.”Paul Grimshaw from Sennen