Wikipedia elaborates further on the purpose of twinning: “Digital twins integrate artificial intelligence, machine learning, and software analytics with data to create living digital simulation models that update and change as their physical counterparts change.” They function as “living models,” continuously learning and updating from the data streaming from the embedded sensors in their real-world double. Over time, the digital twin collects a valuable, complex information base that can be very useful. That knowledge can enable real-time monitoring and troubleshooting for predictive maintenance of an entire fleet of deployed products. Long term, the data can also be an innovation platform for an iterative design process—producing an evolving redesign based on the experience of the products in the real world.
GE was instrumental in the introduction of digital twins, and it describes four different types: parts twins, product twins, process twins, and system twins.
Some of the manufactured products that can benefit from digital twins include aircraft engines, automobiles, and locomotives. In one demonstration, a GE spokesperson describes how heat sensors in a turbine alerted the engineers back at the company about excessive friction and consequent wear that would likely shorten the life of the engine. After the sensor data streams were analyzed, the company decided to shift one of the processes in that part of the turbine to another system within the same machine. The demo even included the speaker donning augmented reality glasses to “actually see” within the digital twin where the problem was located.
Both IBM and Siemens describe their approaches to twinning with three practical modes for their applications. IBM describes their uses as design, manufacture, and operation, and Siemens has different terms for the same three applications: ideation, realization, and utilization. Digital twins are widely used for all three of these stages of manufacture and application. They can be an important tool, as well, in designing the very assembly line on which the subsequent twinned products will be assembled.
ORIGINS
Digital twinning isn’t a new concept. CAD (computer-aided design) models have been illustrating products for decades in the same industries that today are the primary adopters of digital twins. These include automotive, shipbuilding, aerospace, and many others. But until just a few years ago, creating digital twins, though possible, was prohibitively expensive and difficult. The massive amounts of data that digital twins process require computing, network, and storage resources that prevented most from even considering it as an upgrade to the CAD models already in use.
But today, the development of Big Data management tools, affordable cloud computing and storage, and machine-learning algorithms of artificial intelligence to assist in the analysis of floods of data have all met at a juncture that can support digital twins in industries as disparate as automotive and security or, for that matter, that can create twins of entire buildings.
NICELY POSITIONED
BigLever Software of Austin, Texas, is an industry leader in software product line engineering (PLE) with its Gears Product Line Engineering Tool and Lifecycle Framework (www.biglever.com). Founded in 1999 by Charles W. Krueger, BigLever provides companies with the tools to engineer a portfolio of similar products or systems with variations in features, all as a single production system instead of separate multiple products. The tools dramatically reduce the time spent by engineers trying to manage the complexity of added sophistication or product variation, such as with automobile makes, models, and the confusing arrays of options.
By the time research company Gartner identified the digital twin as one of its Top 10 Strategic Technology Trends in 2017, Krueger and his associates had already integrated IoT and digital twins into their feature-based PLE Factory. Now, when a manufacturer sets out to create a product line with the PLE Factory, a digital twin for each product is automatically generated in the process. The process provides the manufacturer with a complete and simplified view of each of product’s twin.
A self-professed optimist, Krueger says he expects “the birth of 30 million digital twins a year.”
It’s difficult to guess how quickly digital twins will be added to product lines, but a quick look at the advantages seems to indicate there’s no turning back now.
March 2018
