I recently sat down — virtually, of course — for a chat with Dr. Michael Grieves, the visionary who coined the term Digital Twin. Currently the Chief Scientist for Advanced Manufacturing at Florida Institute of Technology, Dr. Grieves is a leading researcher and advisor on manufacturing technology. He consults with some of the largest and most progressive manufacturing companies, and researches transformative technologies for manufacturing.
It didn’t take long to realize Grieves is a passionate advocate for remaking manufacturing processes. He likes to start public presentations by showing an aircraft carrier, using it to make a bold prediction: by 2075 it will be possible to print an aircraft carrier — “or whatever piece of naval architecture replaces it by then,” he told me. “Not parts of it to be assembled,” Grieves was quick to point out. “The whole thing.”
Grieves starts with the bold prediction because he is a big believer in two things. The first is the power of imagination to help people believe what can be achieved. The second is the power of exponential computational increase to continue pushing the envelope of what’s possible in additive manufacturing (AM).
Usually when people talk about exponential increase, they reference Moore’s Law. For Grieves, the actual Moore’s Law has become mythical. “Counting transistors is just a proxy for increase,” Grieves points out. The real exponential growth in computational power comes from multiple factors, not only the number of transistors on a semiconductor. Bandwidth increases, the rise of powerful Graphics Processing Units (GPUs), and algorithmic advances in software are all part of the continuing exponential rise in capabilities. “It is an exponential curve, going straight up,” Grieves notes. “It provides phenomenal computing power to do what we could not do before.”
Turing got it wrong
Ready for another myth of computation to be busted? “Turing got it wrong,” claims Grieves. As a reminder, Alan Turing was the English computer scientist who broke German codes during World War II, and is noted for the Turing Test, a method of trying to gauge whether or not a computer is capable of thinking like a human.
Turing got it wrong, Grieves says, because we are discovering it is better to have artificial intelligence model the world instead of the human mind. “We can simulate everything but human intelligence,” and in doing so we increase the utility and value of simulations.
“Put an observer in the room and show an image of a virtual product and a physical product side-by-side,” Grieves suggests. Five years ago anyone could tell the difference between a photo of the physical product and an image of its virtual twin. “Today almost no one can tell the difference.”
As a result, companies are doing fewer physical prototypes — and saving money in the process. As for the number of companies using additive manufacturing processes for more than prototyping, the trend is clear. There is a hierarchy of use cases, Grieves says. From the average manufacturing company printing prototypes to companies like GE Aviation printing their Leap engine, “people are at every stage” of the AM usage continuum.
As important as improved hardware is to the growth of additive manufacturing, the real disruptive breakthrough will be in materials. “The future is going to be custom materials,” Grieves says. Custom materials will be selected for the job based on virtual twin simulation. Manufacturers will go to their supply chain and request a material based on defined characteristics. “Companies will tell a materials manufacturer ‘I need a material that does thus and so’ and they will custom make it.” It is about tailoring the material to the idea, not the other way around, Grieves says.
Key to the effectiveness of using custom materials will be changes in our software tools. Design for Additive Manufacturing (DFAM) “will become more important. Today we worry about geometry, but we should worry about designing for this material on this machine.”
A new versatility is coming based on designer materials, Grieves says. Like other major innovations, it will take time. “Ford had to get the assembly line running,” he notes. “We have to create the processes of how to interact with custom materials established.” It will be a complex interaction of material and printer, Grieves notes. “What is the laser wavelength? What is the size of material particles? This must be in place before we can tinker with custom materials. Process always precedes content.”
The Twin Dilemma
Grieves’ notion of process proceeding content ties right back to the increased use of the virtual twin to not only correctly model reality, but to drive manufacturing processes. This comes to another myth Grieves wants to bust, that a virtual twin is produced after the physical model is developed. “At first we were looking at usage in service, using both a physical version and a virtual version.” The change requires “moving from functional-centric to product-centric engineering,” and populating the product digitally in the model first.
“It is clear the virtual twin precedes the physical version,” Grieves sums up. “At least, it is clear to me. It has always been my intention that the virtual twin would be in the entire product lifecycle.” ##
Editor: Watch Dr. Grieves presentation, The Future of Additive Manufacturing, as part of Dassault Systèmes 3DEXPERIENCE: A Virtual Journey series, Episode 3, dropping on September 23, 2020. Register for access to this, and other online content. Featuring two conference tracks, Episode 3 delves into how Modeling and Simulation have evolved over the years to become a critical and strategic component of the design transformation process, and how Additive Manufacturing is disrupting traditional manufacturing process and shifting how industries think.