0:00 My presentation will be about a slightly different topic, as you can see on this slide, it is called analysis of ecological building envelopes using Grasshopper, and it will address the question, how we can use computational design tools, workflows, to promote biodiversity in the urban environment. But before I start, I would like to give you an overview about our company, McNeel, and our product, products. My name is Verena Vogler, I’m part of the McNeel team for nearly 10 years now, back then I started as a part-time Rhino trainer and German tech support at McNeel Europe, while working at the university as a researcher, I used in my research Rhino and Grasshopper to produce, design, and fabricate artificial coral reefs, which was really interesting, to work at the company but also to work in research at the same time.

0:56 And today, the responsible person for R&D activities, and our programming course, program that is for, designed for Rhino users with a strong interest in development, so we offer course, courses for Grasshopper, Python, and C, at different levels, so probably, maybe you have seen it around, and if you want to learn more about it, you can just come up to me and ask me some questions.

1:24 So who are we, the McNeel story in a nutshell. Robert McNeel Associates was founded in 1980, and it is a privately held, employee owned company, with headquarters in Seattle and regional offices worldwide. Compared to other software renters, we are a small company with 130 employees, divided into development and into support, we have more than 500 thousand users worldwide, 10 thousands of them are schools, and furthermore we work with more than 500 resellers, authorized trainer training centers, and third party developers. We strongly support Rhino as an open development platform, so our SDK is, p, is public, to access our geometry kernel, and the 3D file format is also open source and, and freely available, for over 20 years now.

2:30 Our core value is to deliver a product that truly passionates our customers, to naturally promote this product by an authentic type of recommendation, so one user recommends it to another user, and our company also actively commits to empowering developers, and to forer a strong sense of the community, which are our users.

2:52 Yeah, our main products, you might know them, they’re Rhino and Grasshopper. Rhino is a powerful free form surface modeler for Windows and for Mac, and Rhino supports the creation and editing of NURBS curves and NURB surface geometry, mesh geometry, SubD, and point clouds, so as such it is industrial neutral. But as we are in a digital and additive manufacturing symposium, Rhino is used in different industry, and into, today, manufacturing is a powerful application of Rhino, it supports a variety of file formats that are compatible, compatible with 3D printing, and hopefully we will have also soon the 3mf format.

3:54 The example on this slide shows something that you have seen probably in the exhibition area, it’s a footwear designed by Renee Maer, that he realized together with Sparine, the ADMS model by Sparine, and it was created in Rhino, U, using SubD. With the release of Rhino 8 last October, we introduced the shrink WP commands, which creates a watertight mesh around open and closed meshes, nurs geometry, SubD, and also point clouds, so it’s not only IDE for creating meshes for 3D printing but also for reverse engineering, so we have seen how people use it for the conversion from 3D scan data to SubD, to nerves, and then to fabrication. So it’s, how Renee mentioned it, it’s like turning garbage, or rubb it, to gold, that’s how we sell it, to like this command or this tool.

5:01 So Grasshopper, how many of you know and have used Grasshopper? Wow, okay, that’s a lot, I didn’t expect the money actually, okay. So, but I will tell you what it is, a, a grass over is a visual programming environment that’s tightly integrated with Rhino’s 3D modeling tools, it was introduced by David Ratton as explicit history, back in 2007, with the aim to, M, make Rhino parametric modeling tool. So Grasshopper allows you to have a map of the creation process of your design, which is the Grasshopper definition that you might be familiar with, and it also, yeah, and also to make changes in your model’s history that then affect the final outcome without having to remodel.

5:56 So you can see here, you can change, you can go back and change the size of the box, you can change the number of points inside of that box, and then this affects the waro, and then the edges, and then you can also create a SubD fattened kind of structure from that. So this is the way crossover works, yeah, and since back then, crossover allows uses to build custom computational workflows, algorithms, generative forms, parametric geometry, and data models, and David actually has spent a lot of time to make this user experience a joyful one for designers and engineers that do not have a background in programming.

So after brief introduction of our company and our products, I would like to present the work that we’re doing atMcNeel Europe R&D. McNeel R&D participates since 2016 in European research projects, they’re funded by the European Commission, and during the past years we have worked in four Horizon 2020 projects, with a goal to test our new frameworks and open APIs in new fields, such as geomatics, neuroscience, and also ecology.

7:23 So our latest project is called Copes, and in the project a team of ecologists, computational designers, architects, and experts for human nature interactions, they jointly work together on a conceptual idea of ecological building envelopes, which we call Eops. And the Eops project aims to address the negative impacts of rapid urbanization on land capes and ecosystems, which are biodiversity loss, invasive species spread, and reduced ecosystem services, which affect both negatively nature and also human well-being.

8:06 So traditionally, architects have not focused on designing for biodiversity, and yeah, and also ear planners, they often overlooked architecture, as they concentrated on the basis that are free of buildings, which are parks. So in this project, in Eops, biodiversity is a central driving force for architectural design, and we try to find a way to include, besides human needs, also those of other species, so such as microbiota in the soil, plant species, and animals, so we try to include them in the design process, their requirements.

8:50 Well, as we are a software development company, our own, the project is rather technical, we are responsible for integrating an ecological model into a computational design workflow, and we tried to offering it as a free tool for architects and designers, so it would be a free Grasshopper plugin. So first of all, what is an ecological model? An ecological model is a simplified mathematical way, or program, to represent a complex ecosystem, so you could imagine it as a, yeah, some sort of game of life. And the output, there are 2D maps that display by, like you can see one of those here, that display byod diversity values over time at the scale of a region or of a country, and here the UCD Dublin, they developed a biodiversity map of Ireland, so this is mainly this, yeah, this scale these models work in.

9:59 So the questions we asked ourself was, how can we adapt these type of models to work on a much smaller scale, how can we operate them in 3D, and how can we simulate an ecosystem on a, on a building. So on the right hand side here, you can see the validated ecological model, and this is developed by ecologists in the project from the Technical University of Munich, and it computes species dynamics between microbiota in the soil, plants, and animals, in 2D.

10:40 And what we did is, we substituted the 2D grid cells by 3D voxin model, and generated geometry related inputs with new Grasshopper functions, and these functions are related to soil distribution, soil type, soil depth, shading, and there are much more, and we did that as the, at a resolution of one cubic meter, and the ecologists on their side, they adapted their model to this resolution, to one square meter in their case, because they work in 2D.

11:10 So the result is that we managed to bring ecological analysis into Grasshopper, and to link it to a 3D geometry, so we can now simulate dynamics between species and their distribution over time in a 3D modeling environment. Well, and as mentioned before, behind these green areas there is a volumetric data model, a vox model, and that boxy model displays ecological analysis results, such as biomass, then we have biodiversity, which is the richness of the species for cell, and then we can also simulate the occurrence of one specific species, for instance, on this slide you can see the species lavandula.

12:09 And at the moment, our plugin uses a species pool of over 19,000 plants, but not all these 19,000 plants are simulated on one example, because it works with local specific data, so the building is, this building, at this example, ex is in Madrid, so it only uses the plants that you could find around there.

12:30 So we can also simulate the occurrence of herbs, such as sedement, and it grows on several soil types that are distributed over the building and the ground. Then also shrubs, such as Rosemary, it only grows on a soil type that’s placed on terraces and on the rooftop. And in this example, this is an example for a tree, the tree Acer, and it only grows on the ground, as its rooting depth is larger than a meter, on, on the terraces you won’t have soil that’s larger as a meter.

13:06 Well, and you can see here that the plugin works with different geometry inputs and parametric Grasshopper models, so this is one of our earlier test models, and we created these serial plots that illustrate the dynamics, dynamic changes in plant species distribution, and this over time, and we used in that example, as it was the earlier example, much less species, just for testing purposes.

13:39 So let’s speak data, behind each colorcoded voxy cell there is a numerical value, and numerical values is associated with architecture, the environment, and with ecology, so how do we organize the data in a meaningful way. So the answer is that we use the parametric vox model, so in the first step the designer has to define a resolution of this volumetric data model, and all voxel cells are indexed, and each voxel cell knows its neighbor, so it’s a kind of graph data structure. Then the data is added as an attribute to each cell, and the index of each cell is directly associated with multiple attributes.

14:32 And let me briefly talk about the extent of data that we are dealing with, for instance, in our example building, which was just a test case, we have 859 voxel cells and 1,755 attributes per cell, because we compute spa, spatial temporal data. So the challenge of mapping vast amount of data lies in extracting meaningful insights from it, so proper analysis can reveal the correlations between the form of the building envelope and its ecological performance. So simply put, through the data we can actually find out which architectural design supports biodiversity, and which ones negatively impact species richness.

15:28 So yeah, for this reason, as we had a lot of data points, we had more than five million data points, we designed an experiment using a machine learning model. So first, we brought all that data that we generated in our grasser plugin as a training data set into a machine learning model, and we wrote this machine learning model in Python, and on the slide you can see how the data set of the example building is visualized in Python.

15:55 And then in the next step, we try to find out which parameters are relevant for the occurrence and growth of one specific species, in this case lavandula, and we asked the question, what changes need to be made in this example that its form supports the growth of lavandula, so what does need to change in that geometry.

16:15 So for the machine learning analysis, we tried different algorithms, for instance, the decision tree algorithm from the skik kit Sky kit learn package, and cut boost algorithm from Yandex Technologies, and the both of them, they returned similar results. So the machine learning decision tree algorithm identified shading, and height, and soil depth, as key factors affecting the growth of lavandula, and also in this order and in this importance, as you can see on the, on, on, on this graph, our findings show that lavandula thrives in areas with minimal shades, height above 4 m, and soil depth over 15.5 cm, so this was actually interesting information.

17:06 So based on that information, the building form can be optimized, so the optimized building form introduces larger teres facing south, and a structure design, it supports more than 50.5 cm soil on the roof, and all terraces that are above the height of 4 m.

17:29 Well, so a few words about the architecture analysis and simulations and our training database, there are comp, and to put the data in our training database, yeah, we have to deal with computationally heavy processes, so that’s why we used, for all of this, Rhino Compute on a Windows Cloud Server. I don’t know, anyone has heard about Rhino Compute, okay, so we have some people, they know about it, so Rhino comput is basically Rhino inside, Rhino inside, running within a HTTP server environment, and so you can use, it can be used as a backend service to bring process intensive algorithms into the cloud, and the system communicates with local machines through a REST API.

Well, conclusions, also the Grasshopper plugin with the ecological model is a very simple start, or simplified start, it already, but it still already incorporates the dynamics of plant species, and we have this species pull of more than 19,000 plants, and it connects architecture, in some sort of way, with biodivers, biod, biodiversity, and the plugin speaks, yeah, for, for the ones in a way that cannot speak, which are the, the, the nonhuman species, and it considers, in some sort of way, their requirements in the design process.

19:09 Well, the parametric VM model, it effectively represents geometry related attributes at a predefined resolution, and this was the only way to make it compatible with the ecological model. And also, I think GI models, they work in a similar way, so it could be also compatible with them, and volumetric data models, they can be used as training data sets in AI appro, approaches, so we see this actually also in practice, a lot, that people use Grasshopper to generate training data, so it’s a kind of a synthetic data, that’s generated by, yeah, changing some sliders in crossover.

So what’s next, yeah, we, we are working already with landscape architects together, and they really ask us that we need to include existing vegetation, that we start not the simulation from zero, but that you start at, at a point where you already have something growing somewhere. And we want to extend analysis and simulations to the scale of the neighborhood, and consider artificial irrigation systems and extreme climate events, and there is already a model around that is computing the species dynamics between animals and birds, especially birds.

20:32 And yeah, we try to maybe bring that, well, into the plugin, or make it available, and maybe grasshoppers as this year. Well, yeah, upcoming events, Carlos already told you about these two events. Well, so thank you for your attention.