By Sanjeev Srivastav, Senior Vice President, Persistent Systems 
Digital twins in the Industrial Internet of Things, the virtual representation of physical products or systems, provide a paradigm that will greatly enhance our ability to monitor and control complex systems. When manufacturing or operating machines, your finger should remain firmly on the pulse of that machinery, to avoid costly mechanical failures.
Imagine that your systems are able to operate at peak efficiency, with the real-time tracking of operations data letting you anticipate repairs when you aren’t within physical proximity. That’s the promise of the digital twins.
Digital twins are composed of, say, a CAD model, the design parameters, the operating state, and the operating environment. Think of the twin as a body whose nerve endings lie in the sensors deployed on the physical machines! These nerve-endings bring a level of sophistication to the digital twin that’s not economically feasible with traditional modeling and simulation techniques.
The importance of the IIoT digital twin
The digital twin paradigm allows manufacturers to do two things – (i) operate their factories efficiently without fear of failure and (ii) gain timely insights into the performance of the products manufactured in these factories.
First, getting the real-time view of shop floors requires the digital twin to monitor the health of machines, when operating under load, so operators can make decisions on operational efficiency. Creating complex models of the machines and using physics-based simulation is the approach that engineers use during the design stage; however, the need for computing the actual response of systems under real-life conditions is an expensive and difficult process and can be replaced by the use of sensors, for key health metrics. The operating state of manufacturing equipment can thus be juxtaposed with the safe or desired operational envelope, allowing a continuous feedback loop to control the machines.
Secondly, monitoring the field performance of the manufactured products via their digital twin provides an ongoing view of actual usage versus the anticipated usage, as reflected in design parameters of the product. This process generates valuable feedback for improving the design process. Running physics-based simulation models for systems deployed in the field is virtually impossible, because you never know the local environment; again IoT to the rescue.
Digital twins in manufacturing
While the notion of a digital twin is similar across various domains, what differentiates the role of the digital twin in the manufacturing process is the granularity of information and the near real-time updates.
In the cloud, a machine’s digital twin makes its current state and prior operating history available at a level of detail that’s only possible with the Internet of Things. Back in the day, these machines were instrumented to catch alerts based on relatively few parameters, say, high temperatures or high vibration levels, managed at the edge. Now, maintenance engineers, designers, marketers, or even the CEO of a company has a window into the life of the machine, viewing the granular detail of specific machines or the aggregated metrics. The result is quicker and more informed business operation.
Constraints on digital twin technology
The cost-benefit analysis of creating digital twins typically involves decisions on the granularity of measurements. Depending upon your domain of operation, you’ll need to consider the key operating parameters that will provide enough understanding of the machine or product that optimizes the size and complexity of the digital twin. Companies can certainly start small and then refine their requirements, based on their experience operating a digital twin.
Gartner cites digital twins among the Top 10 Strategic Technology Trends for 2017, with “substantial disruptive potential across industries.”
Indeed, industrial operations will surely be transformed by the real-time controls on industrial equipment, and predictive algorithms that can leverage investments in big data technologies and machine learning/cognitive algorithms. Digital twins are here to stay, along with the physics-based models (cousins?).
