Digital Twins will seriously disrupt the industry this decade. That is my firm conviction. The early adopters and first movers have already benefited from the realtime simulation and virtual test capabilities the technology brings to the table. Now, also the first majority is starting to discover that the embrace of digital twinning opportunities quickly leads to a cascade of applications and design options, causing most of them to promptly insert the technology in their critical flow.

By now, every right-minded engineer must have realized that the strict separation between mechanics, electronics and software has gone way too far. When you are developing a complex system and want to beat your competitor to the market, the barriers between the divisions have to come down. Apart from the social and organizational challenges involved, breaking down the silos is also demanding on a technical level. Although very good and continually improving with every new release, current design tools do not speak the same language, and are rarely understood across the borders of their own kingdom. A Digital Twin platform unlocks all the data so the different disciplines can easily work together. With such a communication tool, the design work can be divided more efficiently in individual blocks that can be solved separately, and that can even be reused in future work.

A good Digital Twin platform speaks many tongues, but there is still much to be gained. In the coming years, the industry must tackle the challenge of standardization. For a long time, vendor lock-in has been a quandary. But the practice of forcing clients to use the entire tool suite of one particular supplier is quickly becoming a remnant of the past. The different departments will surely not work with software from such a closed bastion.

In construction, engineers work based on the Building Information Modeling (BIM) standard which contains set rules about how to store your geometries and how to map your metadata to them. In the industry two standards are evolving to maturity: AutomationML and FMU/FMI (Functional Mockup Unit/Functional Mockup Interface). These neutral data exchange formats allow you to connect data objects from different sources. Big players such as Airbus and Daimler recognized the issues years ago and are the major drivers behind these standards. Now, these standards have become attractive to the next level of companies like ASML, Philips and their supplier networks.

 

Version control and authorization

Another important challenge is version control and authorization. With growing development teams working on the same (divided and distributed) model, it is becoming ever more imperative to keep track of who did what and when. Only then can you do a proper rollback when things go sour. Current PLM solutions simply do not suffice. They are not suited for the new reality of digital twinning. Sure, you can create excellent 2D drawings. The tools often contain a GIT implementation, so you can divide the work and register for every block where changes have been made as well as the respective engineer that made them. But they lack three crucial dimensions: time, 3D and a way to couple with other rich data sets.

A modern Digital Twin platform should definitely be able to cope with the pressing issues around version control and authorization, but the industry has not tackled that challenge yet. At universities and research institutes like Fraunhofer and TNO, scientists are trying to figure it out. I suspect the solution will be ready and available within a couple of years.

 

Machine Learning

Even with advanced technologies like digital twinning, we still rely heavily on human brain power. Humans both create and evaluate the data. The logical next step is Machine Learning. Because you have such a rich data structure, you can even take these smart algorithms beyond just machine control. You can use Machine Learning to optimize your design in an evolutionary manner. And, you can do this over three axes: geometry, software and mechatronics, or what I like to call machine phenotyping; where do I put the motors and what are the ideal spots for my sensors?

Waymo did exactly that. This self-driving car developer was born at one of the famous Friday afternoon sessions at Google. Engineers recognized the potential of virtual worlds to tackle the problem of self-driving vehicles. They started out with the world model of Grand Theft Auto and upgraded that to a full-blown testing environment for self-driving cars. In that Digital Twin, Waymo cars have driven millions of miles. The company uses Machine Learning to optimize the system behavior, but also to look for the best location and the ideal number of lidars, cameras and other sensors. The best thing is that Waymo can try the configurations that the Machine Learning algorithms suggest in its virtual world, let them run a thousand times, pick the best result and start a new iteration. Of course, it will have to validate the final design in real life, but the development process has been sped up enormously.

Since innovation speed is the most paramount factor in the current industry, organizations should really consider adopting Digital Twin technology. Think of Volkswagen. Once it was the biggest car manufacturer in the world, but now it is getting overtaken by faster companies like Waymo and Tesla. VW is still stuck in the old way of thinking and cannot transform fast enough. By the way, a lot of European power houses are lagging. Their American and Asian competitors are sometimes five year ahead. And with exponentially accelerating developments, that could turn out to be disastrous for European competitiveness. The bottom line is that you cannot physically test your system ten thousand times. It is time to take a physical step toward the virtual world.

 

About the author: Guido van Gageldonk is co-founder & CTO of Unit040, a visualisation and simulation software company from the Netherlands. He founded the company in 2006 while still being a student at the Technical University in Eindhoven. He is a renowned tech watcher in the Netherlands and is well known for his ground breaking ideas regarding Digital Twin Technology.

Leave a Reply