Between March 2017 and September 2018, in a warehouse in Amsterdam, robotic arms steadily created the world’s first, single-span 3D printed steel bridge. The achievement was made possible by Dutch 3D printing company MX3D in conjunction with a huge number of industrial and academic collaborators. The completed 40-foot bridge was unveiled to the world at Dutch Design Week 2018, ahead of its future installation over the Oudezijds Achterburgwal canal in Amsterdam.
In addition to its unique construction, the bridge is also a living laboratory for data scientists as it’s instrumented with an innovative sensor network, produced by Autodesk, Force Technology, Lenovo, and HBM. Everyone that walks, runs, or cycles over the bridge will generate data about the behaviour of the structure.
The Turing’s programme in data-centric engineering, funded by the Lloyd’s Register Foundation, in collaboration with researchers at a number of universities including Imperial College, University of Cambridge, and Newcastle University, has been developing a ‘digital twin’ of the bridge to help analyse the sensor network data, as well as conducting extensive tests of the physical printed material and using statistical methodology to understand more about the material itself.
This work, in conjunction with that of the wider project’s many international partners, is providing scientific insight not just into how this revolutionary new material and construction process behaves, but into the potential for infrastructure to act as living objects that encourage greater citizen engagement.
Header image credit: Arup and Joris Laarman Lab.
How did it start?
MX3D have been working on the 3D printed bridge since June 2015, bringing it to life through collaboration with various partners, including Joris Laarman Lab which designed the bridge, Arup who have acted as the project’s lead structural engineers, and Autodesk who have provided their knowledge on digital production tools. Gijs van der Velden, CEO of MX3D, says, “Finding partners that understood the potential of the technology and felt the same drive to solve early adoption obstacles was essential.”
“The digital twin concept is highly complex…it only made sense to do it [with] the Turing”
Gijs van der Velden, CEO, MX3D
The Turing’s data-centric engineering programme, funded by the Lloyd’s Register Foundation, became involved when the concept of a ‘digital twin’ was introduced, which will help monitor and analyse the real-time performance of the bridge. “The digital twin concept is highly complex and avant-garde”, van der Velden explains, “It needs to be adaptive to several data sources that are not usually analysed together. Therefore it only made sense to do it when the Turing proposed to commit itself generously to the project.”
Liam Butler, from the University of Cambridge’s Centre for Smart Infrastructure and Construction, got involved because of the need for expertise in advanced forms of sensing for the bridge’s sensor network. Craig Buchanan, one of a number of researchers from Imperial College working on the project with the Turing, explains that they got involved due to their “strong interest in exploring new 3D printing techniques” and the experience of the university’s Steel Structures Research Group in material testing and modelling. Chris Oates, a Turing Fellow from Newcastle University, was brought on board to discover more about the material properties of the 3D printed steel through the use of statistical methodology.
These three researchers represent a fraction of the wider project’s many contributors, but their insights illustrate the diverse expertise and value being brought to the work.
Liam Butler explains the thinking behind the sensor network: “We want to measure things like strain, temperature, vibration, and acceleration of the structure as people and bicycles are traversing the bridge.” Real-time data from these sensors will interface with the digital twin model of the bridge.
“The digital twin also helped influence the design of the sensor network”, Craig Buchanan explains, “the location of sensors has been based upon the anticipated structural response from our computer model.” The bridge’s complex geometry, chosen to demonstrate the capabilities of MX3D’s printing technology, provided challenges in developing and validating an accurate digital twin. Buchanan and his colleagues remedied this by conducting “careful non-destructive tests – of stiffness, strength, ductility – on the partially built bridge and comparing these real, measured responses to those predicted by the digital twin.” These tests culminated in a successful 10+ tonne load test on the completed bridge in September 2018.
The physical testing has been supplemented by statistical approaches to understand the “intrinsically random” properties of 3D printed steel. “It’s an ‘inverse problem’”, Chris Oates explains, “you’re given data about how much the material displaced under a certain force and you have to work backwards to try and determine the elasticity properties that would generate such a displacement.” This approach is helping MX3D characterise not just 3D printed stainless steel, but also a number of other 3D printed materials.
“The Turing has fostered collaborations [that] have undoubtedly led to a more successful project”
Craig Buchanan, Imperial College
“One of the main benefits of the Turing has been to bring together and foster collaboration between academic fields and researchers that do not traditionally work together,” Buchanan reveals, “and these collaborations have undoubtedly led to a more successful project.”
The work is not just giving insights to designers, engineers and data scientists; the work is enabling civic interests too. Butler elaborates, “Thousands of people are going to cross this bridge each hour so the City of Amsterdam are interested at looking at things like foot traffic, CO2 emissions, noise levels, and more.” Craig Buchanan and his colleagues have also been working, as part of a larger team, to develop a long-term structural health monitoring network to ensure that the bridge remains safe during its lifetime.
Mark Girolami, Programme Director for Data-Centric Engineering concludes: “When we couple 3D printing with digital twin technology, we accelerate the infrastructure design process and ensure we design optimal and efficient 3D printed structures. This project has been a world first in engineering and it’s fantastic to have the Turing and our team of talented scientists helping to make it a reality.”
What does the future hold?
The aim is to have the bridge installed in Amsterdam in 2019, where the focus of the project will move to real-time studying of the behaviour of the structure.
Liam Butler on the plan for the future: “We’re planning what kinds of data we’ll collect, and how often, and what kind of computing platforms we’ll use, in collaboration with Autodesk. Where the Turing is going to be so important and really shine is in the long-term contribution when the data from the sensing system is coming through.”
Chris Oates’s work is aiming to produce statistical models which reflect the variation of the printing process. “We’re at an early stage but the work definitely has potential”, he informs.
“The Turing is going to really shine when the data from the sensing system comes through”
Liam Butler, University of Cambridge
MX3D is keen to continue developing its technology with help from the analysis being provided by the Turing and its collaborators. “We want to make the printing more adaptive and fully autonomous,” CEO Gijs van der Velden explains, “Data research from the current bridge is essential for creating a hotbed for 3D printing solutions.”
Craig Buchanan believes the project has the potential to revolutionise the way we think about infrastructure in cities: “Can we create infrastructure that act as living objects, that do more for our cities? Can these kind of objects help regular citizens and our community to better engage with infrastructure?”. These questions represent the dramatic potential value of this multidisciplinary project and the power of combining physical and digital analysis.