Does data have all the answers?

RD&I’s newest Data Science (DS) consultants Simon van den Broeke and Ionut Barbu from Bright Cape discuss their work over the past 6 months and how Damen’s data future is in safe hands.

Foreword – Edwin van Buren, Manager RD&I Valorisation

It is time, Edwin says, “to reflect and share the team’s achievements, challenges and surprise findings. Our Data Science (DS) team can now integrate further into the entire Damen organisation with the potential of becoming the backbone of RD&I (Research, Development & Innovation) developments. We’re now prepared and ready to support complex decision-making by executing complete data science projects, building relevant data models and sharing new knowledge. Ultimately, we can dramatically help improve business efficiency and help Damen - and its clients - towards increased sustainability securing our future as a worldwide pioneer in shipbuilding.”

Project progression

The task was clear. The scope was comprehensive. The support required was two-fold. Build a strategic roadmap and provide operational guidance and support to the DS team. Seven objectives were agreed and then the engagement activities began.

“Our first agenda item was to meet with various team leads and colleagues from multiple departments across the business. We introduced ourselves and gauged the sentiment around data science in the organisation,” says Ionut. From these meetings, the team completed an overview of the research, design, and development needs that could be addressed with data science. Together, Simon and Ionut brought the DS team of Quint van Leeuwen and George Drakoulas along on their data discovery journey. They spent significant time guiding the DS team through stakeholder meetings and structuring project setups, from defining scope, objectives, and risks to time and resource planning. More trust in data science was built within the department and beyond and date science Ambassadors were established.

You won’t find them in the basement

You can easily find the team at the RD&I department in Gorinchem. Being visible and in the same room as colleagues has helped increase the overall understanding of what they can do for the business.

“Within Damen, there is a broad spectrum of understanding when it comes to data science. Some colleagues are further along in their willingness to adapt, others are more conservative, and this is understandable. Our goal has been to build trust with the organisation so they have confidence in the data to support their decision-making, to build the best solutions and better ships,” says Ionut.

Does data hold the answers?

“If you ask the right questions, data can support and certainly give weight to your answers. The numbers either add up or they don’t. But the methods we use are reliable and concrete,” says Simon. At Damen, we value solid use cases, and there’s no better way to demonstrate the team’s value than with a real example.

Optimising ship design

Some say data is the cornerstone of informed decision-making. Back in November, the team met with colleagues who, while designing the hull of a vessel, needed to predict its resistance in water and air.

They focused on one type of resistance called the wave making resistance coefficient (Cw). Wave making resistance is a form of drag that affects boats and ships and shows the energy required to push the water out of the way of the hull. The team built a model that enabled the designers to quickly predict what will be the resistance instead of the traditional method of computational fluid dynamics (CFD).

CFD can take a full day to run a simulation. The DS model was built by inputting a lot of past CFD results for multiple hull designs into the machine learning (ML) system. The ML model is trained, i.e. finding patterns in the data to identify parameters with predictive power, on these past results. The trained model is then used to predict the resistance coefficient (Cw) of new hull designs. This prediction takes just seconds for a whole batch of new hull designs.

The designers could then see the most promising hull types and continue using the CFD method to further defi ne the optimal hull design regarding resistance. “The design team was impressed with the results, speed and accuracy of the model and how it streamlined the process. Accuracy is what matters most and gives assurance to the team and the buyer of the vessel,” states Simon.

With this information, there is also potential to reduce resistance and therefore fuel consumption and emissions, helping Damen to meet its goal of becoming the most sustainable maritime solutions provider.

You get out what you put in

As the saying goes, "garbage in, garbage out." Quality data underpins success. While crafting sophisticated ML models is straightforward and takes up 10% of the team’s time, the real challenge lies in sourcing high-calibre data and transforming it into useful information. Moreover, maintaining open dialogue with stakeholders and refining data sources are essential for optimal performance. Ionut explains, “What’s most important for every project is: do we have the data? Is it accessible and is the quality good?” Data coming from vessels in operation, collected and analysed by Triton, is one such data source that the team can use.

“Then we focus on using the right data, pre-processing and preparing it so it is ready for use. Date science is still fairly new in Damen and there is room for improvement to have even better and more data so we can improve the performance of the models.”

A structured team approach

The time and space given to research and development, the entrepreneurship mentality and the complementary team dynamic is what surprised Simon and Ionut most about Damen.

“Structure is the best word to defi ne what we have done with the DS team over the last 6 months. We have identified specific roles and responsibilities and determined the fundamental processes needed to build a robust DS team. We have also facilitated training and data science knowledge development for Quint and George. The way the team has gelled together really impressed us.

I will continue to support and enable the team for the coming six months and together we will explore if data has all the answers, but first, we need you to ask the questions,” says Simon.

On the top image from left to right: George Drakoulas, Simon van den Broeke, Quint van Leeuwen, Ionut Barbu and Edwin van Buren