Is it Time for a Digital Twin?

Graph shows the current state of adoption of digital-transformation technology in North American plants. Source: Univ. of Tennessee Reliability & Maintainability Center

The concept of digital twins has been around for many decades, but it’s the connectivity of everything (supporting the data flow) that’s possible today that enables it to be more successful.

Most give credit to NASA for modeling/replicating their spaceships in the 1960s. Although others applied the concept, John Vickers coined the term “digital twin” in 2010. 

A digital twin is a physical object and a digital twin model that passes data between them (in two directions) with a synergistic outcome. Once you get to this capability, you can use it to run scenarios to get a better/more accurate look into future possibilities.

According to Gartner, “A digital twin is a digital representation of a real-world entity or system. The implementation of a digital twin is an encapsulated software object or model that mirrors a unique physical object, process, organization, person, or other abstraction. Data from multiple digital twins can be aggregated for a composite view across a number of real-world entities, such as a power plant or a city, and their related processes.” (Definition of Digital Twin – IT Glossary | Gartner)

According to IBM, “A digital twin is a virtual model designed to accurately reflect a physical object. The object being studied, for example, a wind turbine, is outfitted with various sensors related to vital areas of functionality. These sensors produce data about different aspects of the physical object’s performance, such as energy output, temperature, and weather conditions. This data is then relayed to a processing system and applied to the digital copy.” (What is a digital twin? | IBM)

A virtual model that is enriched with good data from sensors and IoT/connectivity can run scenarios/simulations/machine learning, with the results going back into the asset/machine, production line, entire factory, aircraft, building, city, and so on. What makes a digital twin powerful is that it’s a continual back-and-forth of improvement between real-time plant data going to the digital twin and system learnings being applied at the physical asset. Most of us have seen various simulations and possibly have performed them. This is also a virtual model. Digital twins can run many simulations to analyze a process. Similar to an FMEA (Failure Mode and Effects Analysis), you decide the level (granularity of the effort) or level of investigation. For example, the digital twin can be small (for a component) or large (for an entire production facility). 

Digital twins are most valuable when projects are complex, large, and require (and can justify) lots of sensors. This could be for automobiles (design and production), aircraft, complex machines, bridges, locomotive engines, manufacturing/production with numerous machines, and power utilities.

In the big picture, use of digital twins is still in its infancy. My data shows that in North America (on average), when looking at digital adoption level in TPM, remote monitoring is at 48%, wireless sensors, cloud computing, big data, and IoT between 30% and 38%, and machine learning, collaborative robotics, 3D printing, and augmented reality between 8% and 18%. Machine learning was at 10%. 100% equals fully adopted.

A few years back, Gartner found that only 13% of survey respondents already used some form of digital twin. At the same time, 62% were working on it or had plans to do so (Digital twins becoming mainstream: Gartner | CXO Insight Middle East (cxoinsightme.com). I think that more organizations are experimenting with some machine learning on assets, but not referring to it as a digital twin, since it’s early in their learning.

There are numerous benefits that can be attained by applying digital twins. Just a few include:

• Helps reduce unplanned downtime and related costs. Also, this puts people fixing things in less-frequent emergency situations, which will reduce safety risk.
• Tracks failures in real-time.
• Provides more time to intervene on pending issues. It’s a way to better identify when each asset/component is most likely to fail, extending your time on the P-F curve.
• Simulates what maintenance actions provide the best outcome.
• Becomes your focal point for ongoing improvement, with all the current knowledge of the asset performance now in one place.
• Offers the option to link digital twins when appropriate.

There are some challenges that need to be overcome:

• confirming data quality (includes data relevance)
• handling massive amounts of data
• acquiring real-time data and keeping it secure
• integrating and combining asset models and plant functions is complex.

Digital twins provide an opportunity to create good failure models, improve assets, and simulate scenarios for ongoing learning, insights, and question answering. Like all tools and technologies, the digital twin should add value to the business. Once we get better answers to the “what if” questions, we may find out that we haven’t known enough to even ask the correct questions. Digital twins can also be applied to predictive maintenance. If you develop a model of a pump, machine, or production line, you can use it to support predictive maintenance. 

As you move from reactive to a more proactive process (preventive, predictive, condition-based), a digital twin provides you with a model that knows asset performance and changes (real time) with plant-floor asset changes for better predictability. You also have a model that can run any number of simulations to improve predictability based on factors such as hours of operation, age, temperature, and flow rate. Since a digital twin is constantly learning and improving, it provides great opportunity for new insights. This approach delivers all the benefits of a best-practice reliability and maintainability process. In addition, it can improve machinery and equipment performance at the systems-thinking level. EP

Based in Knoxville, Klaus M. Blache is director of the Reliability & Maintainability Center at the Univ. of Tennessee, and a research professor in the College of Engineering. Contact him at kblache@utk.edu.