An Open Source Platform for Digital Twins?

Digital twins are the hype of the day in industrial technology. Software companies are quick to jump on the band wagon, launching products that fulfil at least part of the functionality that is associated with digital twins. But where is the open source platform that allows collaboration on the development of this concept? The current open source initiatives don’t seem to live up to the needs. An initiative on European level would be welcome.

The concept of digital twins is still very much in flux. If you check out the Wikipedia page on digital twins, you’ll find no less than seven definitions. These range from relatively simple collections of IoT data to complete modelling systems. The same diversity can be seen in the recent special issue of ERCIM news: one group uses it to create distributed control systems, another applies the concept to models of complex materials, a third adds to the confusion by introducing the term digital shadow as distinct from digital twins.

The commercial arena shows a similar divergence. General Electric, Siemens and Dassault each offer their own proprietary version of the common digital twin functionality. Microsoft also has a digital twin but for now it seems to focus mostly on just collecting data for further analysis. So does PTC’s offering. The digital twins offered by Akselos perform truly impressive structural simulations in near real time for the offshore industry but they seem less suited as a digital twin for production lines in factories. So, every player in the field puts forward its own interpretation of the concept.

I think this fractured playing field is hampering the progress on digital twin technology. In essence, the development of digital twin technology requires the cooperation of people from many different backgrounds. It includes things like connectivity, to make data about a remote device available to the digital twin; fast computational techniques to simulate complex multiphysics, statistics and machine learning to gain insights from the data that are not readily obtained by physical modelling; cyber security, to make the digital system safe. And these are just a number of topics that need to be addressed. Each of these fields has a thriving research community working on many unsolved issues. Wouldn’t it be nice if all these researchers would have a common framework to work on so that other researchers, software companies and end users can immediately make use of any progress that is made?

This is essentially the call made by Cor van Kruijsdijk in his contribution to the ERCIM news issue titled “Digital Twins as R&D Accelerators – The Case for an Open Source Ecosystem“. Professor Jan Post from Groningen University expressed a similar need for collaboration in his interview with┬áPT Industrial Management (translated from Dutch): “if you want to understand all processes around, for example, a shaving head […] you need to assemble a large variety of models for different processes and providers into a process chain. This takes an immense effort. If you’re lucky you can do this in a European project so that you can share the effort with partners.” Such a collaboration would be greatly facilitated by having a common open platform to work on.

So, at least some people feel that an open innovation environment, probably on an open source platform, would be a good idea. Where, then, is the open source platform for digital twins? When you google for it, a few candidates pop up. Let’s have a look.

One is the appropriately named Ditto project. This seems to be an effort by the German company Bosch at creating an open source foundational layer for digital twins. Like Microsofts offering, it concentrates on the data modelling and connectivity, but it has a place for more functionality in the Feature concept. I didn’t try, but I’m not immediately confident that it can handle really complex simulations easily. Also, Ditto has limited visualization facilities by itself. The strong connection to Bosch is a bit uncomfortable, but the source is available on GitHub so nothing would prevent a fork if Bosch would become less benevolent at some point. However, the competitors of Bosch are probably wary of building their products on Ditto.

The other open source digital twin project that makes it to the first page on Google is iModel.js. Again, it is a commercial initiative, this one connected to the US infrastructure company Bentley. Therefore, it is not surprising that it focuses on infrastructure modelling. As such, it seems to be set up nicely, using BIM as the modelling framework and offering npm packages that should readily do the job. But like Ditto, it doesn’t look like it’s built for the heavy computing that comes with some multiphysics models.

Both are decently set up, but they are limited in scope and, at least for now, tightly connected to a company with commercial interests. That is not necessarily a bad thing. In fact, having a deep pocketed stakeholder behind an open source project can be a blessing. The scope, however, is a bit problematic because it seems that the limits of the current set up will soon be reached by anyone who tries to apply the platforms beyond the context for which they were created.

I think a more ambitious effort is needed. It wouldn’t hurt to build upon the foundations that have already be laid by projects like Ditto or iModel.js. But a careful reconsideration of the architecture is probably needed, taking into account all the various use cases that people associate with the concept of a digital twin. And it should be broadly supported by many stakeholders, preferably comprising also more than one that can provide financial backing.

It would be good if the EU would support such an initiative. It is already supporting the Digitwins initiative which focuses on healthcare and biomedical research. The association of European mathematicians in industry, EU-MATHS-IN has started an initiative around the modelling part of digital twins in their report “Modelling, Simulation & Optimization in a Data-Rich environment”. Let’s hope that the EU will manage to bring such developments together in a common platform for the benefit of industry worldwide.