Q&A Summary from Dr. Michael Grieves’ Webcast on the Digital Twin

Blog: Apriso Blog

Dr. Michael Grieves published a white paper in 2014 entitled: “Digital Twin: Manufacturing Excellence through Virtual Factory Replication.” This paper has generated considerable interest on how a digital mockup can be created based on what was actually produced, which can then be digitally compared to what was actually designed. In a perfect world, these two digital designs are “twins.” But, we don’t live in a perfect world. Discrepancies can be readily identified to perform root cause analysis to understand why, and then fix.

Dassault Systèmes had the opportunity to host a webcast featuring Dr. Grieves, which took a closer look at some of the concepts introduced in the white paper.

For those that might not have time to watch the entire webcast, we have summarized the questions and answers that immediately followed, and included below.

Do you see a future where the genealogy of products is digitally captured whereas “as-designed,” “as-built” and “as-maintained” data is all available for simulations, digital mockups and designs?

I am always reluctant to predict the future, and how quickly things change because it seems like technology still moves even faster than I had suggested. But, yes, I do think that the idea to design, manufacture and support products virtually is possible. Then, only when we get it all right do we actually physically move around some atoms. We want to trade off bits for atoms wherever we can. I think that at some point in time you will be able to inspect the virtual product, and you’ll be able to obtain all the characteristics that you would be able to obtain from having possession of the physical product. But I’m not going to predict the time frame.

Do commercial tools now exist that support the concept of a Digital Twin?

It all depends on what part of the Digital Twin concept you are referring to. Development models are now supported in CAD, which is a Digital Twin of what will be built. So, the realization of a physical product from a virtual product certainly exists. We really haven’t done as much as we need to do in terms of taking a manufactured product and creating a virtual or Digital Twin of that, although we do collect some information. I ought to caution that we may not need all the information about the physical product in the virtual product. It is really going to depend on what each use case is. For example, I’ll use a trivial one: if we don’t care what color the product is, we shouldn’t have it reflected in the virtual twin model. There is a cost to creating digital data. You have to figure out what has sufficient value to then make it available in the Digital Twin. With regards to the Digital Twin existing now, as a replication of an as-built product in a factory, I haven’t seen a representation, but I think people are working on making that happen. We have parts and pieces, and very nice factory simulations based on information collected from the factory floor. However, it is going to take a unified data repository to fully make a factory replication a reality.

How can a Digital Twin help with continuous improvement on a global scale?

What happens a lot of times with continual improvements today, Kaizen as an example, is you really have to move actual physical things around the factory floor in order to understand the improvement. If you could simulate those improvements virtually, and then send them off (after validating these improvements), you could ship to all factories and track them, making sure they were implemented in the proper fashion. Too often you see the same problems solved over and over again as each – both across locations and different generations of new products being rolled out across production facilities. The reason is that designers really have very little (if any) connection to the factory floor, so they don’t understand what problems have occurred or why a manufacturing process should be changed. Then, a new version of the product comes along with the same design bill of process from the manufacturing engineers, and the same problem gets solved over and over again. Sharing a Digital Twin from the factory floor with manufacturing engineering might help avoid this repeated and wasteful loop.

What role can connective technology (cloud or IOT) play with the Digital Twin?

One of the elements of the Internet of Things (IoT) is basically that physical products have intelligence in order to do two things: sense and communicate. They need to sense what is happening, and then they need to be able to communicate that. In manufacturing we have the Industrial IoT (IIoT), which includes machines talking to each other. Let’s say I have machines that are changing their program or parameters in order to compensate for other things that are happening in the factory. You really need to understand what they’re changing. So having intelligence in machines is a good thing, but unless you have a Digital Twin that can effectively communicate with humans, to know exactly what is happening, I would contend that those kinds of systems can go out of control quickly and cascade failure.

What benefits does the Digital Twin bring to simulation?

If you are going to do product simulation, you have to get the physics right. For example, let’s assume you are building a building and need a beam that runs a little further. When you add length onto that beam, the simulation must reflect the additional weight at the same time, so when you wind up with the real building, it can hold the weight that it’s supposed to hold. That is a challenge we face in simulation: being able to simulate/replicate what occurs in the real world in the virtual world. That is where the simulation people are earning the real money. The Digital Twin is basically built into the concept of modeling and simulation.

If you liked this article, here are others you might also find interesting: