Ground2in: Create a Digital Twin of the Ground for Soil Remediation

PIERRE MARECHAL: Welcome to our class called Ground2in, Create a Digital Twin of the Ground for Soil Remediation. So we're, after a small introduction, we're going to explain how it all started, then how we scale it up, present you the results, and the benefits, classic summary.

My name is Pierre Marechal. I am head of CIM at COLAS.

PAOLO SERRA: Hello, my name is Paolo Serra. I'm an implementation consultant at Autodesk.

PIERRE MARECHAL: So do what is soil remediation, and do you think what's underneath is important? Well, it's all about keeping our planet clean, the ground clean, as we have no plan B, no planet B on Mars, no exoplanet yet. So we have to keep it clean. And these are-- there are big needs in the market to provide answers on sites but also digitally.

My company, we are fully aware of that. And we are working on solutions, trying to solve it. My company is called COLAS. We are a global leader in the construction and maintenance transport infrastructure, opening the way to tomorrow's innovative sustainable mobility. We work on a great diversity of projects from roads and highways, which represent 70% of our activities, to airport runways, ports, but we also work on environmental renewal projects.

In just a few numbers, we are in the top five recycling company in the world. But as you can see, the average project is around 100K euros. So it's quite small project, even if we do 60,000. And our strength relies on its glocal network. So global because we are in 50 countries and all continent, but local because that network is composed of thousands of local offices and industrial assets such as asphalt plants, quarries plants, recycling units.

And we work on innovation for sustainability with soil remediation, which implies traditional earthworks, of course, cut and fill volume truck transports. But we also have very specific and innovative solution with venting techniques. So it's physical, chemical, or biological treatments and connected containers for soil remediation. And, more recently, we started the use of BIM model to collaborate, but also to have a complete digital twin of our assets and processes.

And so, the question, why and how have we been working with BIM on polluted sites? Actually, it all started with one big project, COLAS Dunkirk refinery in France. So it was a big project with big needs of the client to de-pollute a refinery. It was a huge site. We had to deconstruct and clean up, and on which there were real issues, like financial risk, countless data to manage, and a lot of actors to make collaborate. Actually, we even made a class in the EU in 2019.

So first, we were asked to digitalize the site into traditional BIM models, so superstructure for buildings, warehouse, plants, and infrastructure for underground pipes, networks, and foundations. And it started with sorting something like 60,000 handmade plants but, also, the scan of 90 hectares.

At that point, we were working on use case, which were collaboration or a BIM bundle. So it was classical. But then we were asked to digitalize different polluted components and even the water table, so that BIM model would serve as digital twin. And when I say digital twins, it means digitizing of the assets, but also processes so as to make decisions.

So all that things that you cannot see, because hidden underground, are also hardly representative-- representable because it's fluid. But these things and works are highly valuable and costly.

So a few explanation about the objective. It's about cutting the ground into tiny voxel at the top left, field data of each voxel with borehole data, top right, and then play with the display to select processes to treat the ground. We'll explain that.

So good for us at that time, Dynamo was enough powerful and fully integrated into Revit to do it by ourself and with different calculation methods. So it offered different scenarios of pollution spread to make simulations.

And when we would compile all the models, superstructure, infrastructure, and soil, it would look like this. And we could display a certain degree of pollution. So in 3D, it would look like this exactly. Let me skip the slide-- if it works. OK.

But 3D was not the goal. The goal was to discuss about the best options of depollution among different processes. So it was more about deliverables like management plans and surveys than 3D. And the benefits were obvious. You could collaborate, take decision, and it takes much less time to have more options than before.

So we did once on the particular project with quite good success. But most our pollution project are quite small and with short delays. So we knew it was useful innovation for soil remediation, but then we wanted to scale it.

So how to scale it for other pollutants, not one by one, but several ones? How to scale it for smaller projects and also to whole community of non-Dynamo, non-Revit experts. We also wanted that Dynamo script would be more efficient, take less calculation time, and do much more things. So it was time to scale it.

So we decided to make a plugin for Revit with Autodesk and Paolo development skills with six objective, finite elements of the terrain, so to transform the terrain into voxels, take data from borehole from Excel sheets, do interpolation, linear or kriging to put the data into the voxel, create the lithology, and then offer automation for deliverables.

PAOLO SERRA: The plugin enables the digitalization of the analysis process within Revit from start to end. The interfaces studies to streamline the workflow and guide the designer in defining the inputs that are necessary to conduct the analysis.

It starts from the definition of the subregions used in the analysis and the definition of the water table levels. The designer then defines the spacing of the grid used to create the finite element representation of the soil and imports the boreholes data from Excel. Once the inputs are in place, the modeling of the Revit elements can start.

In the settings are defined the share parameters to create and assign for the objects involved. There are more than 100 pollutants with a dedicated parameter for pre- and post-depollution level, plus another handful of attributes to fully localize and describe an element in the study.

The settings are designed to be flexible and ensure consistency of the data contained in the models. Things like the x, y, z global coordinates for the insertion point of the elements, the elevation of the topography surface, the lithology, and the classification of the water level. This example, only two pollutants have value associated, and all the other parameters are hidden to keep the model readable. When these elements are in the model, the actual interpolation workflow can start.

There is a dedicated expander for the linear interpolations. The plugin allows the designer to think in abstract terms rather than focusing on the underlying Revit model. This is to maximize the productivity in this phase and explore multiple scenarios and methods. The plugin takes care of maintaining the relationships between the designer inputs, the business logic, and the Revit elements.

Under the kriging expander, the designer can select the pollutant from a dropdown and visualize how its concentration changes with the distance across the entire site, reading the values from the samples. There are six methods to select from to approximate the start of this distribution as closely as possible.

The designer can fine-tune the parameters interactively with both the visual feedback of the resulting curve and the numerical value for the error. The designer decisions are recorded in an extensive log that is running in the background and provides details of what is happening during the calculations.

The kriging approach involves for each voxel all the samples in the site at the same time. This would benefit from a parallel computation, as each voxel is independent from its neighbors. Currently, the plugin is limited to a single process and it's computing one voxel at a time sequentially. This paired with a client server architecture and a design automation would improve also the usability of the plugin.

The deliverables consist of plan, 2D views, 3D views, schedules, and sheets for each pollutant with a color range filter override depending on the pollutant concentration. It is possible to define a dynamic set of intervals and create view templates to be used across projects for consistency.

This one is the most tedious task that has been streamlined thanks to the plugin. This has improved the overall productivity and simplify the communication of the data through the digital twin. The views are created with a consistent naming convention and are organized by pollutant for the pre- and post-depollution. The view filters are managed by the plugin, and the results are immediately visible to the designer.

The plan views and the necessary levels are managed by the plugin depending on the input distance specified by the designer by the plugin interface. Similarly, the 3D view is created and the smart annotation family is used to create a legend view that can be used on multiple sheets.

So now, let's look at the value delivered with this engagement. Pierre, what would you think is the reason that brought you to write this plugin in Revit?

PIERRE MARECHAL: Yeah, actually, we did it to scale it to our whole environment community at COLAS and to leverage it. And as we had a lack of skills in my company to develop such a plugin, or proper plugin, we used Autodesk consulting to benefit from developers with experience like you, Paolo.

PAOLO SERRA: And what are the benefits for your customers?

PIERRE MARECHAL: So the benefits for our customers are, of course, the user experience, speed calculation, automation for deliverables, flexibility of interpolation method. We have different methods integrated. And, also, it's now a true decision and communication tool. And which can also be used for cost control management.

PAOLO SERRA: Great. So can you recall what were the challenges that you faced during the engagement?

PIERRE MARECHAL: Yeah, of course. The difficulties were about the technical aspects of a new interpolation, like kriging, as we are not pollution expert, neither Autodesk people were. So we've been helped with mathematicians to help us with this kind of thing.

PAOLO SERRA: Yeah, that was fun. And what about the impact on the processes and the workflows?

PIERRE MARECHAL: The plugin interface follows a workflow for traditional depollution projects, but with one section for each step. So treatment of the input, creation of the voxels, selection of interpolation method, deliverable, et cetera. So from schedule to PBI dashboard, we worked with environment expert to capture and integrate their expertise into the most UX-friendly tool we could make to help them with their decision-making.

PAOLO SERRA: Great, thank you.

PIERRE MARECHAL: So since the beginning of the project, we knew how it was important to have a functional workflow between the BIM team, the environment expert, and Autodesk development expert. So the challenge was really to take and capture the knowledge of experts.

So at the end, this data-centric and collaborative approach is a digital way to make business but also to bring value to our clients. The digital twin is not only about objects, but also about the process. And within the objects, it's not only about construction but about soil, which is quite new.

PAOLO SERRA: So maybe we can sum up the outcomes and the deliverables of these engagements. Here in this slide is helping us to see what has been delivered. So we had a plugin with an installer. We had created this capabilities are creating model items involved in the studies, the topography surfaces, the voxels, and the samples.

We had this feature for zone detection. We had these new interpolation strategies, the linear that were already created in the Dynamo script as well as the kriging. We had this logging feature. And, of course, we had the automation of the deliverables, including the views, the filters, the schedules, and the sheets.

Now, let's look at the outcomes and the deliverables to give a sense of the size and the scale of the projects that COLAS is doing. In the data set that was provided, which is typical, we had approximately 21,000 voxels and 800 samples. And it took less than five minutes to model the entirety of the model.

For the interpolation part, on 21,000 voxels for a single pollutant, it took approximately five minutes for the linear interpolation, and it took a little bit more than an hour for the kriging one. And we were also able to include the deliverables with the views and the filters, the schedules with the single or multiple pollutants, as well as the sheets.

So I'm talking about the overall impact. We articulated three principal items, the time-saving, the adoption, and the quality. For the time-saving, it takes less time to treat an area compared to the usual three days of pollution experts, as previously with the Dynamo script. The tedious tasks within Revit are automated. And the linear interpolation performs better than Dynamo.

The plugin replicates the same updating behavior that comes with Dynamo, and it allows the designer to continue using the plugin between sessions and explore multiple scenarios on the same model. This is especially useful for the kriging interpolation that requires several attempts to fine-tune the calculation parameters to choose the best method for estimating the distribution of the pollutants.

When it comes to adoption, the plugin comes with an installer to simplify the deployment. The new workflow still uses the Excel inputs for the boreholes, and it reduces the preparation needed to consume the data in Revit.

The single user interface driving the project allows pollution control experts, who are not Revit experts, to have an easier access to the object properties and the functionalities that would otherwise require a significant amount of training. This is lowering the resistance in adopting the software that is foundational to enable the digital twin approach. Managing more than 100 shared parameters required is done through the plugin, and it allows very easy customization through the settings.

Last but not least, the plugin interface, the tooltips, all the buttons, and the logging, supports both English and French, which is critical for the regions that COLAS is supporting, especially in Canada. For the quality, the model elements are more intelligent. For example, the voxels know where they are with respect to the water table levels. The interpolation steps are captured in the logging, and allows the experts to verify what elements took part in the calculations.

The consistency of the models and the graphical representation of the results make it possible to develop a data-centric approach to the de-pollution. This opens a new avenues that leverage artificial intelligence to improve the estimation and predict the level of pollution in the soil and select the most effective treatments, unleashing the full power of a digital twin.

PIERRE MARECHAL: So back to our planet A, because we still have hope to keep it clean and beautiful for the future. That's what and why, at COLAS, we're working for bringing digital twin technology to the soil industry and sharing it to the-- with you.

Enabling this new way to bring Eden technology and friendly UX to this tool is not only about doing business but also working on sustainable projects. So compared with traditional Revit method, we can do more things, have benefits with larger scale of operations. And I'm sure you can do the same, and you should.

PAOLO SERRA: And from my side, I would like to thank you, Pierre, for this project. I would like to mention also Raquel [? Bascones ?] that helped me during the development of this plugin. And, as Autodesk, we're here to help our customers design and make a better world, and as consulting to unlock area of values and make you successful.

PIERRE MARECHAL: If you want to know more about that plugin, or if you want us to help you on your project, feel free to contact us at this address. Thank you for watching, and see you at AU.