0:01 Hello everyone, my name is Preety Anand, and I’m working as an architect and computational designer at BIG, at the Zurich office. So we have opened up an office just for the airport project, which I will also be talking about today. So my presentation, I will first talk a little bit about the project itself, so we understand the background and the scale. Then I will talk about the computational design processes which we have within our project, and then I will talk a little bit about the digitalization in the timber construction industry, and how this is actually impacting the way we work and the way we develop our geometry.

So a bit about the project. So this is a rendering of the Zurich airport project. It was a competition won by BIG. The original airport was inaugurated in the 1950s and has ever been, has, has ever since been the largest transportation node of the country, and it’s one of the most important airports actually within Europe. In order to maintain its competitiveness and to keep up with the projected passenger growth, there is a substantial need for overhauling the existing structure and also to increase the building capacity. The overall design approach is, of course, to combine the ex, existing with the new, which is a challenge in itself, because the existing buildings, and with the underground structures, they do offer a very restrictive and constraining framework.

1:36 So this is one of the interior views of the new dock. So the stru, as you can see, the building timber structure is space defining, which also demands a close cooperation between architect and the engineer. It’s actually essentially a simple building forum, which has been derived from functional, operational, and technical requirements. And if you look at the cross-section with the wi columns, it has been purely developed from constraints and functional requirements, such as column free traffic routes and height limitations, as well as distances of parked aircrafts to the facade. Geometrically, the V column is defining the dock in its expression, and the, the, the cross-section of the dock, of the we column, is changing successively from one end to the other, which is also partially why we can consider it to be a complex geometry.

2:35 And then, as you see in the image, we have a skylight which divides the dock into two parts, and that skylight is has a increase in width, and it in its pitch, from, from the, the northern, the, the east, for sorry, from the west to the east, and then it becomes a triangular skylight in the root, which you see on the next slide. So the route, as it is called, it’s the triangular node between the dock and the existing buildings, and both parts of the project are very, have a very distinct character due to the exposed timber structure. Timber in itself is a very interesting material, because it has unique load, loadbearing characteristics. And it’s also unique project because it’s a large infrastructure project, and it’s the first time where a timber structure has been implemented in such a big project.

3:43 It’s also a contextual material, if you consider the Swiss context, and it’s a renewable resource, and it is intended to be reused at the end of its life cycle. So it’s in itself a benchmark project, and we also will talk about the prefabrication and the preassembly, so we are considering really a a novel approach to construction, to materiality, and to prefabrication. So just this, a little bit about the project in itself.

4:12 So now I will go into the computational D design processes which we have in the project. So just by looking at the geometry, and to understand why we are using computational design in, within the project, I can point out a few characteristics. So I mentioned the changing cross-section of the building, which is, which is, one on the one hand, coming from the free form roof geometry, but also because the the ground is sloping, which leads to everchanging cross-section of the building. So which means that no cross-section repeats twice, so every element you is actually unique in that cross-section. And so together with the slope and the roof geometry, it produces an overall complex geometry.

5:10 And then another important aspect is the double curved skylight surface, due to which each element within the skylight surface is unique, but also each joint within these elements gets unique. Alt together we have around 1,700 unique timber elements, and it would almost be impossible to model this accurately by hand, because we need to be very accurate, which, and the reason is the fabrication, which I will also talk about a little later, and the the precision which we need in the project is also impossible to be modeled manually. So that’s why computational design is a key part and key aspect of this project.

5:51 So we are, in this slide you see a few processes which are actually simultaneously. So the overall geometry of the envelope is complex in itself, so we we actually used scripts to develop and control the massing, so in order to get an area, to get a control over the area, but also to understand local constraints such as views and SES and other sight based conditions, so we can be very accurate and very precise in developing the massing. And then, because the structure is actually based on the overall massing, we used that volume to develop the timber structure, to develop the joints, to develop the setting out. And on the other hand, we were also working with scripts for developing the skylight surface, for dividing it, but also for optimizing the surface for using it for structural simulations and to understand the curvature.

6:59 And at the moment we are actually using scripts for consolidating the design, but also to develop the timber joints, to use it for preliminary dimensioning. There are other constraints such as the glare analysis going on, there’s drainage optimization, and there’s lots of different scripts running at the same time in parallel. And that also shows that we have a holistic approach to computational design, which starts from our competition phase and will actually end up in the fabrication and assembly model. And that’s kind of the key message behind this slide, is to have a holistic approach to look at different phases, and to constantly change the requirements of the script, but also the content of the script. And as it is based on a parametric design model, it can actually be continuously further developed. There can be new requirements coming, and there can be some requirements moving away or changing, so that a script as a base is a key factor for us to work, keep working on this project.

7:59 As I mentioned previously, there is a geometric interdependency between different building elements. So we started with the grid and the setting out and the building code and the site constraints. Out of these factors we developed the massing and the volume, which was done with it, done and controlled and consolidated by using a script. Then this volume is actually defining the envelope. It is defining not just the the overall envelope, but together with the grid it’s actually defining the setting out of the glass facade or PV facade, so different facade types we have. And on the other hand, the grid and the volume are actually a major aspect for the timber structure, because we are defining the spacing and the the setting out out. And together with these different elements, we are actually also impacting the interior fit out. So, so to speak about the slab outline or the fit out, we have cur, we have inclined glazing which is following the beam inclination. So all these aspects of massing and timber structure are in in return implementing the geometry of the interior. And then, of course, also the areas, and the area is like a hard factor which we need to have a good control over. So if we change the massing, we change the area, we change a certain aspect of the project, so this needs to go back as a feedback into our overall design process. So you see that because we have a complex overall geometry, all the other related parts are also complex, and they need also computational design tools and workflows in order to design them and get a control over them.

10:00 This is U, I would like to speak about the timber structure script a little bit. So this is like a major part of the project, because it’s not just being defined by the overall massing and the grid, but in return it’s also defining the interior and the fit out. But we have a certain, for modeling this timber structure, we have quite a few hard criterias and numbers and data we put in, in order to control the geometry, such as building levels or dimensions of the fit out, dimensions of the structural beams. These are some numeric data inputs, but also for the unique ID of each object we actually put in some text, text elements which will be concatenated to a unique name. And then, within the script, we have certain script clusters which actually take in all these numeric and databased inputs, but also geometric inputs, such as the building grid, which I have spoken about, but also some volumes of internal core structures or interfacing structures, but also of existing buildings, and we also have a few constraints which we are using as planes, which trim the building. So this entire process is automated within the script to give just the final geometry.

11:29 And as an output, we get the timber structure elements and the element axis, but also a un, a unique naming of each element, which is used for identification of that unique element. And on the other hand, we can get some numeric data output, such as setting out points, like element insertion points, but also instance values in terms of length of element, or length of of structural section sizes, and so on, which we would use for further studies or for putting it out to other consultants.

12:08 I would like to just show you an overview of the individual steps which are part of the script, and just go through each step one by one. So we start with the project grid and the building levels. Then we have all the points intersect themselves, which gives us a 2D point grid, which is based on the intersecting grid lines. Then we project these points into 3D space, so we get the setting out points for each element. Then we connect these points in order to get the element axis and the extrusion vectors for each element. Then we can draw the planes at each AIS, and with the A’s as a plane vector, plane normal, so each plane is actually oriented towards its, towards its normal, which has been drawn in the previous phase. And then we use these planes to draw the structural section sizes, and these numbers we had obtained from the structural engineer. They are a result of the structural simulation. And then we, as in the next stop step, we project these sections to their end plane along the normal vector. And then finally we can extrude the structural sections along their normal vector, or we can loft these members if there are a pair of these sections.

13:46 As I mentioned previously, there is, there’s quite a lot of volumetric constraints in this project, such as interfacing elements and interfacing buildings. So we use these elements for inclusion tests, because for example we have an irre, irregular slab outline on each level, and the beam distribution is very much dependent on the slab outline. So we use these slabs for an inclusion test, so you only have beams where you have slabs. So there is lots of processes going on within the script which deal with inclusion and interfacing and trimming. And then finally we use the distinctive planes in order to trim the beams and columns, and to get the final geometry. And then at the end we do have our final timber structure geometry, where each element has a unique ID, which has been also developed out of our input data.

14:48 As an overview, in context with other consultants and with other engineers, it’s the architect who develops the timber structure geometry. We give it as an output, as a stick model, to the structural engineer for simulations. They give us a feedback, which we then in turn use for redesigning or redeveloping the structure. On the other hand, we also have an output of instance values, such as workpoint coordinates and insertion planes and the timber member axes and element attributes. Again, the structural engineer uses that data output for the cost calculation, and on the other hand we use it, we architects, we use it in order to develop the BIM model. And that again, as a result of coordination, we get more feedback which again goes into our timber structure model. And the data which we put out of Rhino is also used by the contractor for developing the fabrication and assembly model. This small animation just shows the uniqueness and the D diversity of all the building elements and how they come together and form the overall structure, and it’s very important to understand that each element is unique in this project, and we would not be able to develop this so precisely without having computational design workflows.

16:20 The importance of computational design and digitalization within our project is also a key factor further down the road, when we we are going to be in the construction phase and coordination phase, because these aspects are actually quite important for timber, for the timber construction industry. So digitalization is an important aspect of the timber construction industry, and we need to look at the historic context. So historically, timber structures have always been benefiting from offside pre fabrication and preassembly, and therefore there is a logical consequence that these have adopted digital fabrication, because they can be a lot more precise and a lot more controlled. That development, in terms of fabrication, has also enabled the construction industry to create really precise beam joints, which, due to which, we can actually create those complex beam connections and complex nodes in an economical way, way. We are now able to create a large range of connections where the beam to beam connection can be complex, or we can actually have curved beams with a simple node. So there’s a wide range of development, and that’s all only because there has been a an adoption of the comp, digital fabrication and computational design.

17:50 That development, so the development in the 3D modeling, and then this thean sead fabrication, has actually led to the feasibility of free form grid shells and geometrically complex timber structures. We have been able to create structures which not only stand for themselves, but they actually are able to have an integrated envelope which has different tolerances and different movements. And also there is now, because of the increased precision, we can also integrate other elements such as MEP ducts and so on. So there is a increased complexity, and that’s only possible because of the modeling techniques and the CNC fabrication.

18:33 I will now talk about the design coordination and fabrication process of the timber structure by showcasing a reference project, because we are still in the schematic design phase. So this is a, to start the discussion on this, upcoming phases, I would like to show a typical process diagram which is extremely focused on the digital design, fabrication, and assembly of the tim, complex timber structure. There is an evolution of the 3D model, but also of the script behind that 3D model, throughout the different faces, and each phase has a clearly defined pro, purpose, and a clearly defined script content. And if you consider this evolution of the 3D model and of the script, it can only be possible with a parametric and rule bas, based modeling process in the background, because you can change requirements and you can add requirements as needed.

19:31 As a reference project, in order to explain the process, I have taken the Swatch roof of the headquarters in Be, it’s a project by Shigar Ruban, and it is quite close to our grid shell of the skylight. So we start with the project grid, as we are also doing that, and we have a reference surface, and that reference surface is based on curves which have a minimum number of control points, and that surface is actually the base of all subsequent modeling. That surface can also be used for testing spatial adequacy, but also for testing the curvature, so the structural engineer has used those surfaces for developing the preliminary section sizes and also to test whether the curvature is feasible. Some, something about the importance of that surface is also that the surface is not randomly generated, but it is actually based on local criteria of site conditions and building codes. And as a next step, we actually project the beam grid onto that surface, and there needs to be a clear relationship between the beam axis and the surface, because we will be using this until the end of the project, so we need to be very clear about whether the surface is center to the beams or bottom or top.

21:05 The next phase after the concept and coord, reference model phase, is the coordination phase, which will be very challenging, because we will be a, we will be attempting to integrate the envelope, the grid shell, and the MEP. It has the challenge of competing goals between different engineers and different parties. It can also lead to novel design technology solutions, but it is going to be a c, there’s certain going to be challenges ahead of us. And we, of course, start again with our parametric model which has been developed in the previous phase, and we will attempt to integrate the envelope, the timber structure, and the ME, and that also needs to be done in a rule-based parametric way. That also requires a certain data exchange between the different parties, and that data can be expressed numerically or compiled in text files or Excel, but for example, with the structural engineers we will still be exchanging stick models for them to test the section sizes. There are certain challenges in terms of interoperability and file format data models, like some people or some companies are working with nervs, some are working with meshes, and every firm has their own tolerance and accuracy because they have different purposes behind their 3D model.

22:41 The next phase is the detailing phase, where we will be detailing all the supports and connectors of each element, and that 3D model needs to be very accurate and precise, because it actually refle, reflects the physical object. It’s not just a representation, it will be used by the CNC machine for immediate production, and so the level of detail which you have in the digital needs to be equivalent in the physical, because they are actually standing for each other, and that’s why a high degree of quality control is also important within that ph. And ideally all these cutouts and support holdes and MEP connections, they need to be also parametrically developed based on our previous scripts.

22:41 The next phase would be fabrication and assembly, where we would also be very much aware of the accuracy we need in the model, because the accuracy you have in the digital immediately affects the physical, and we will also need to do a certain degree of of rationalization, because as you, as you can imagine, in a free form grid shell there are, there’s a large range of beam vectors, but for fabrication and for assembly purposes there’s a need for rationalizing this a little bit, depending on the design intent, and also depending on the structural behavior of the system.

And finally, we need to think about the assembly, but the assembly concept actually needs to be considered very early in the project, because it will have an impact on the joints between each cluster, but also within each cluster. So the assembly strategy is something we need to think about very, very early on this project. And the assembly sectors, setting them up and defining those assembly sectors, is very much dependent on the systemic structural behavior, so you cannot just divide the clusters as you please, but you have to consider local conditions, transportation constraints, and structural behavior of the grid shell itself. And it’s very important to also embed that assembly logic within the naming of each object, because the naming of the object can actually be a clue on how to assemble those objects, and also the geometric uniqueness of every every element can help in assembling it, because the assembly is still going to be manual, and we need to have some clues in order to have a logic which is human readable, which is, which can be implemented by people who are on site implementing the the assembly.

25:41 Finally, back to the actual project. And so, understanding this entire phase and understanding the timber structure industry with its digitalization, we can now imagine the challenge es, will we will be facing in the upcoming phases, and how important it is to implement computational design from right from the start, because it will be a key factor for this project to be successfully implemented. Thank you.