At the 4th annual Trimble Construction Summit Middle East, held on 18 September 2018, Denis McNelis, Engineering Manager, BAM Higgs & Hill spoke about the complexities involved in building The Museum of the Future, Dubai.
Give us a bit of background on your role as an engineering manager for the Museum of the Future.
The catalyst was when I first saw the project announcement. Then read the news stories about this amazing structure, I said this would be an amazing project to work on. I was approached by BAM International to come on board due to my expertise in 3D engineering and complex steelwork. That was an opportune moment for me to transition to the project.
What’s your particular expertise that is needed for this project?
I’ve had a career as a structural design consultant and also as the technical director in steel fabrication. I have worked both as a client and as a contractor. So, I have seen projects from many different angles. As Engineering Manager on this project, I’m interfacing between BuroHappold, the main consultant, and Eversendai, the steelwork subcontractor with Maffeis Engineering their analytical expert and the Robert Bird Group who are technical advisors to us. Having an understanding of both the analytical and practical side gives you more of a robust role in managing the different criteria of different companies with different priorities. A consultant has one set of priorities, and the fabricator a different set, or maybe a better term is a different area of knowledge.
As a consultant, a lot of time is spent on the analysis and design details required to achieve the project, but you don’t know necessarily how exactly it will be built. As a fabricator, the skill set is how to design the connections, fabricate and assemble the steelwork into erectable components, the logistics required to get it to the site and last but not least how to erect it to achieve the consultant’s design. However, with a structure like this, you need to understand the requirements of both. This is a project where the stresses and loads vary and change as it is built due to the geometry of the structure.
So, the complexity of the project is being able to understand the different perspectives of different stakeholders?
Definitely, because everybody has a focus as to what is critical from his or her viewpoint. The issue is to get people to understand that you are aware of the complexities that concern them. Each party needs to know that you have a full understanding of the problems, and how to find the best way forward for the team to work together. As long as you’re talking and moving forward as a team, you will always find a way.
What have been the technical difficulties and challenges to date?
For one this would be the scale of the challenges. If you consider it from an analytical point of view, Buro Happold required many models for the analysis due to the complexity of this type of structure. Then there is the construction methodology, for example for the steelwork this would be the erection process by which structure will be assembled and erected on-site. Eversendai has to ensure that all of the connections are designed correctly and fabricated so that they can be erected efficiently on site. So you have all the component sizes and millions of pieces of data that dictate all the forces that need to be balanced at each of the nodes. Eversendai, in association with Maffeis Engineering, analyzed all the connections, producing thousands of pages of calculations, both using automated processes and manually, and then submitted them to Buro Happold for review so as to ensure everybody was satisfied that the design intent was achieved and that the actual building could be achieved efficiently and practically.
After that, Eversendai had to break the building down into erectable pieces. From Buro Happold’s perspective, they designed the building that is a series of nodes with members in-between forming the triangulated diagrid. What Eversendai does is not bringing in one piece at a time, but instead assembling many pieces as a single assembly, and lifting this into place. This breakdown is based on what the available lifting capacity of the cranes. As there are three cranes on-site, each crane has a limit as to how much it can lift. So if the piece is closer to the crane, it can lift a heavier section. The building is symmetrical around the central axis but when you look at the structure you can see that there are different points where it’s all assembled. That’s purely so that Eversendai could fabricate as large a piece as possible in the factory, and deliver it to site intact, meaning fewer welds on-site.
This is one of the most interesting aspects as every steel building is thousands of unique parts. From a fabrication point of view, it doesn’t matter if a column is 10m or 10.1m. These are two different components. So, when you model something like this building, you draw every single piece and produce thousands of drawings that then go to the factory, where one piece is manufactured at a time. So whether it’s a simple rectangular building or a complex curved one, the software breaks it down into individual pieces. In the old days, you would have a person calculate the length of the piece, and prepare a drawing on how each beam and column had to be fabricated. You wouldn’t be able to do that easily with a building like the Museum of the Future. That’s why Eversendai details in 3D using Tekla Structures. The level of detail is so great, especially with the complexity of calculating the setting out of the curves etc. The software does the math required to produce the 2D piece drawings that the fabricator can use and then assemble all the pieces on the ground into the complex 3D shape it needs to be for the erection and for transportation to site. That’s where it’s complex and very difficult. And that’s also where the expertise of the steel fabricator comes into play, because they model it in 3D as per the engineer’s design, and then break it down into the component parts, which are then put on the shop floor in a totally different orientation. However, it’s done in a manner that is deliverable. If I take out any one of these nodes, there could be five tubes that all come together as one piece. Due to the design loads, they have to be joined in a very specific manner. This determines the way they have to cut the pieces so they sit accurately over each individual piece as if they have been moulded on top of each other. The big challenge is calculating all of those cuts so that they fit and can be welded together.
Is this one of the most complex buildings ever in terms of structural steel? Is it pushing the software to the limits?
Yes, it is definitely pushing it. There is additional complexity in this building compared to some other buildings in that you have the initial design with the 3D BIM model, but when you build it, you have to allow for the fact that it’s going to move, so that when it’s finished, it is as close as possible to its theoretical positioning. Because everything is moving as you erect the building, you don’t know exactly how much this will be. As engineers, the math’s has been done for the predicted positions and movements. However, when you physically build something, you’ve got to allow for tolerances or a millimeter here and a millimeter there.
Eversendai is making allowance for the fact that the building is moving as it is built, which is the real world, whereas in the BIM world there is no gravity so nothing moves. Everything is suspended in a digital space. Yet, in reality, everything that moves. Now, if it’s a short span of 4 m or 5 m, the movement is not that significant. When it’s a 100m, as the top of the museum is, it moves a lot more. As engineers, we look at structures and their movement. A key check is a span-to-deflection ratio. With a building like this, where we’ve got trusses spanning 24 m, and we want to get the floor level, there is a pre-camber in the truss. Despite the calculations, it never moves precisely as expected. Hence an engineering judgment call has to be made on how much to allow for.
Another key consideration is the accuracy to which different elements of the building can be physically built to. Everything from the foundations to the concrete, steel, glass or blockwork, all the different materials, have different tolerances, which impacts on the accuracy of the build.
Starting with the piling, a pile can be positioned within a tolerance of 20mm from the intended position in the design. Concrete foundations are cast on top of these to within 10mm. Then the holding down bolts for the steel columns then must be within 5mm. So, every step of the way it becomes more and more rigorous in terms of accuracy. In this building, the steelwork is sitting on a big elliptical concrete ring beam that is 20 m in the air, so it is unlike a more conventional building with a solid foundation for the steelwork. The ring beam is 174 m in circumference made up of 11 individual spans. As the steelwork is built, this beam moves down to carry the additional weight. However, when you think about it from a steel erection perspective, it’s not like a concrete foundation that sits on the ground and doesn’t move much. As the beam has to move this requires provision to be made for it during the detailing and fabrication of the steelwork. Sometimes it’s difficult to contemplate that something that a 2.5m by 2.5m RC Beam is going to move, but it does.
Tell us about other technologies used in the construction process.
The design team developed the initial Revit model for the museum which is now being constantly updated by our team including our subcontractors so that we have a much more detailed model now as more information is available. This was transferred into Trimble Connect to be tested with the Trimble safety helmet incorporating the Microsoft HoloLens.
This is an augmented reality approach, so everything is in 3D. We hope to be able to use it later because the HoloLens is impressive technology pushing the boundary of what is available at the moment.
How do you intend to deploy HoloLens on this project?
HoloLens would be used to assess the installation accuracy of the MEP services. This is relatively easy for the concrete and steelwork because these are big pieces in specific locations. When you come to MEP services such as ducts, conduits, cable trays, firefighting sprinklers, and wastewater, these are all elements present at multiple levels and positions. HoloLens would allow us to see where these are designed and drawn on where to go, based on the 3D model, and where it is actually physically installed on-site.
One of the significant challenges is that much time can be spent in the drawing office on the computer getting everything to fit neatly. However, then on-site a decision is taken to adjust the position of something, which throws out the entire design.
You mentioned the challenge to get everything to fit around the calligraphy on the building.
The calligraphy on the building is a key design feature. It’s several sentences of poetry which forms the glazing incorporated into the façade. I’ve never seen a building with this degree of complexity in the façade of having this free-flowing calligraphy forming the windows. It’s not something etched onto the surface but is integral to the building itself. It’s immensely complex.
How will the façade be manufactured?
Denis: It’s a series of unique individual fiberglass, with carbon fiber, panels forming the structural frame clad with stainless steel sheets and with glass panels for the windows. The technology used includes analysis software used by NASA and Airbus. The nature of these panels involves complex mathematical modeling using finite element software.
Could AI be deployed here?
Like most things in theory, yes. The problem with AI is it needs a lot of information to formulate the rules to solve the problem presented. For example, one of the current challenges with the façade is that the calligraphy, which is a lovely free-flowing 3D form of glass, has to be broken down into flat panes of glass. So it’s all in squares, rectangles, trapezoids, and various other shapes. And as you follow a letter along multiple panels, you have to follow a set of rules to define it. To teach AI how to do this, you would have to have examples so it could derive all the iterations necessary for a solution. It cannot do this automatically.
As the detailers are working through the flattening process, they use software routines for the flattening, and when they find that something doesn’t work, they then rely on their experience and knowledge to tweak the process. It would be highly beneficial to have a program capable of doing this, but the software would require ten to fifteen projects as a reference base in order to come up with a transferable solution.