Buzzwords always go around. I remember "fuzzy logic" being this big thing back in the 1990's, for example.
Digital Twin, though, does have its place. For example, let us say you are going to make a new piece of machinery, including the software to program and control it....
- The software developers will need something to test against. As a developer, you need to be able to answer, "did this change fix it?" without spending a week trying to reserve precious machine time on a prototype machine that might not be working. With the twin, it takes 20 seconds.
- Once the hardware is built, and some goofy thing happens, the twin can help answer the vastly useful question of "is this in hardware or software? If the twin shows the issue, you know it's a software issue, and if the twin does not show the issue, you can pretty much know it's the machine - electrical or mechanical.
- Next, when developing a machine, you need to know ahead of time if things are going to be wonky. A digital twin helps simulate the complex processes to create a definition of truth for how the machine should behave, and helps find issues before committing to building the actual machine. You can iterate quickly, and communicate such iterations easily to all parties of concern.
- People doing training and documentation now have something to learn on, experiment on, and get graphics from. They can also use this for in-classroom training later on, to get folks used to the machine before they risk breaking the real thing.
- Offline programmers can now test their tool paths on the digital twin before sending it to the shop floor. This helps prevent a lot of mistakes, and vastly increases their own knowledge on even how to make those paths, so the operator on the floor does not have to do a lot of rework, and even better, the machine can operate automatically with minimal operator intervention. This also enables much more complex tool paths to be more easily programmed, and even the ability to tweak the tool path from within the simulation itself, as in "this needs to move a bit to avoid that".
- On the machine itself, a good twin can simulate complex tool paths prior to hitting the 'begin' button, potentially saving thousands of dollars on otherwise scrapped parts and broken tooling. We had machines, for example, where the thing about to be cut might be worth a half million dollars, and there was no room to mess that up.
- Back in Tech Support, recordings from a real machine can be played back on the twin, so the support engineer can fully understand what the customer is talking about in the request, and a well designed twin system can have all sorts of playback capabilities synchronized to event logs and the like. Cameras on the machine can supplement this with real-world footage, and one can simply click on the camera in the twin to see that footage.
- There is also the opportunity for digital twins to extend beyond all of the above for the future. For example, the 3D graphics rendering the twin can be used for AR and VR for much of the above, as well as more advanced things such as allowing the machine operator to have a heads up display that shows future tool motions in a preview, projected right on top of the machine in AR. Want to know if that path will collide with that fixture? Just look at the machine and see if the virtual path intersects with the real world object. Want to walk around a virtual machine in VR to program the path, or demonstrate it over the Internet to a potential client? Simple, just join the multiplayer "game" rendered in the same engine. Want to see what the spindle speed is, and how much the motor is under load? Just glance over at it, and have it show up in AR.
- Now, let's say you want to have an overview of the status of all machines on your shop floor. Each one could have a digital twin, and each one could be connected via protocols such as MTConnect, and you can view all that status on a single screen, or a glance down to the shop floor in AR, or the machines can talk to a master scheduler and to each other to coordinate stuff like, "I'm starting to lose accuracy, let's move the low precision work over to me, and schedule the high precision stuff to a different machine", or "I am broken, re-nest those laser parts to fit into the work that waterjet is doing", or stuff like that.
So, I think there is real future for digital twins. Of course, it will be abused by marketing for all sorts of baloney, but those who do it right, and for the right applications, it will be game changing.
