There may be no concept in digital business more compelling than the digital twin.
I remember an early instantiation, in the late 1990s, where an aerospace and defense firm had a titanium compressor case that had been severely gouged during the machining process. The cost of scrapping this nearly finished part would have been huge, but the part was clearly out of specification.
The engineers needed to make a judgment call on keeping or scrapping this part. They decided to create what we now call a digital twin. They placed this compressor case on a coordinate measuring machine, using it to digitize the case, complete with the gouge. They manually processed this set of data points in the computer-aided design (CAD) system (remember, this was the 1990s). They were then able to take the exact digital model of the damaged part and run a battery of stress and thermal analysis tests, simulating actual flight use cases. This battery of tests determined that the damaged part was not significantly weakened, and could be used “as-is.”
Gartner Hype Cycles define the digital twin as follows:
A digital twin is a dynamic software model of a physical thing or system that relies on sensor data to understand its state, respond to changes, improve operations and add value. Digital twins include a combination of metadata (e.g., classification, composition and structure), condition or state (e.g., location and temperature), event data (e.g., time series), and analytics (e.g., algorithms and rules).
The engineering software world was early to adopt the digital twin, both from a product and plant simulation perspective. With the advent of the Internet of Things, big data, and predictive analytics, the content and power of the digital twin has become immense.
The vendor community is all in on this. From enterprise resource planning (ERP) to product lifecycle management (PLM) to operational technology vendors, they are feverishly working on platforms to support the integration of their developed (and acquired) software assets to support the digital twin.
Here’s the rub:
- Each vendor has its own interpretation of what the digital twin is. This view comes from their perspective — defined, modified and accessed using their software tools, plugged into their platform.
- Enterprise vendors are doubling-down on these technologies, expanding outside of their core domains and adding additional functionality. For example, Dassault Systèmes acquired the Apriso Manufacturing Execution System (MES) in 2013 to extend its PLM capability and has since acquired Quintiq (supply chain planning) and Ortems (scheduling). The intention is to expand its CAD/PLM tools into a supply chain “platform,” and they are not alone.
- “Open architectures” are not immune to this. There are strong commercial reasons for vendors to segregate their strongest platform capability for their internal applications, even while supporting openness (we connect to everyone, but our end-to-end capability is best).
- These platforms will overlap at the edges, as ERP vendors, Supply Chain Management (SCM) vendors and MES vendors all vie for dominance in this space.
- Each vendor has strengths and weaknesses in functional applications. One may have the strongest CAD environment, another the best PLM, and still another the best MES system.
- Industry standards, which typically lag vendor innovation, and are often “lowest common denominator” functionality, will become more difficult to define and maintain as these platforms become more complex.
While some enterprise customers will adopt an end-to-end platform, this will be the exception, not the rule. Manufacturers rarely have the luxury of single-vendor solutions in business, engineering or production. Even if customers choose a platform, they may be settling for weaker application functionality in some areas, compared to the current practice of picking and choosing based on individual application functionality.
In the 1980s-’90s, when the “Big Three” automobile companies adopted CAD/computer-aided manufacturing (CAM) systems with then-new surface modeling capability, each original equipment manufacturer (OEM) chose a different vendor. Tier 1 automotive suppliers were burdened with having to own at least one seat of each of these three environments, in order to share design information with the OEMs in the native CAD formats.
The proliferation of digital twin platforms has the real possibility of recreating this age-old problem, only to a much larger scale, given the more extensive definition of the digital twin.
As predicted in a recent note (“Five Approaches for Integrating IoT Digital Twins”):
- Through 2023, 75% of digital twins of OEM IoT-connected products will involve at least five different kinds of integration endpoints.
- Through 2023, 95% of digital twin integration will be based on custom integration development.
The benefits of the digital twin, tightly defined in a particular scope, are undeniable. The concept of a single digital twin that will magically and seamlessly operate across domains and system boundaries is a great vision, but the benefit of this all-encompassing digital twin may not be worth the cost of ownership.
Rick Franzosa, Research Director, Supply Chain Technology, Gartner