Digital Twin: Weaving Physical-Virtual Convergence into Tangible Business Value
By Atos staff
While designing systems for connected environments, engineers have long faced a fundamental question: How to ensure the reliability of components that are either beyond the physical reach or are too costly to be pushed beyond prototypes without a guarantee? NASA encountered this problem when it attempted to study mission-critical systems’ behavior and verify their safety in deep-space environments back in the 1960s. It gave rise to the concept of employing duplicates or twins of the space assets to probe their integrity in actual deployments. Thankfully in the era of ubiquitous connectivity and high sensor densities, operational data can be readily shaped into digital twins of physical processes and instances to gain insights into their viability well in advance.
Building models that support systemic studies
In the 21st century, the idea of virtually tracing the lifecycle of products and processes evolved at the University of Michigan in 2002. Modern embodiments of digital twins are complex mathematical models and software algorithms that interpolate historical and real-time operational data and scientific insights to predict systems’ real-world behavior. They can be integrated with Industry 4.0 constructs, AI engines, and advanced analytics for elevated visibility and outcomes.
The tangible benefits of such simulation-based approaches may include high system reliability and availability, improved production, optimized maintenance cost, faster time to market (TTM), and of course, reduced operational risk. For instance, an automobile’s design insights can be assimilated with the live feeds from sensor arrays embedded across its components and transmitting data on mechanical and thermal stress in real-time when the asset is operational. The digital twin makes preemptive maintenance feasible by predicting when a component is likely to break or wear out and recommends practical alterations to improve its performance and durability. Similarly, Atos reports that in smart pharma manufacturing environments, the synchronization of digital twins with online sensor line-ups and hybrid models has accelerated the pace of innovation while materializing high-orderability and control. Reduced wastage can trim up to 20% of the product cost, and high-quality outputs can witness a 10% boost in business margins.
Undeniably, digital twins are fast emerging as a force multiplier and a strategic enabler for enterprise operational realities, with a market size that is forecasted to pull in revenue in the upper double-digit billion range by as early as 2030. As the technology proliferates, three of the most prominent voices in the industry examine how businesses can build upon the physical-virtual convergence to accentuate their use cases and derive an optimized return on their modernization investments.
A networked approach to system efficiency
With the increased mainstreaming of the Internet of Things (IoT), sensor and edge technologies, the fusion of their inputs into a replication or twin of the physical world seems like logical progress in the digital transformation trajectory. However, considering a digital twin as merely a model or mirror image of physical realities is a misnomer. Because it endows the engineers with the unique abilities to gather, integrate and analyze information about systems to which they have either limited or no access in the real-world, unlocking a frontier of technical possibilities and business value streams. The digital twin concept is also an evolution in the asset visualization philosophy. While earlier, operators were more focused on isolated assets, today, their interests extend to the behavior of networked systems like supply chains and shipping logistics. The digital twin approach improves the feasibility of gaining insights into the state of the ecosystem partners involved to predict the outcomes and terminal business impacts with greater confidence.
However, from an innovation standpoint, digital twin is an idea in progression, with incremental maturity. For instance, in a smart factory environment, all the assets from different OEMs had to be ideally enabled with IoT and real-time data transmission capabilities to render digital twin constructs with high fidelity. While we are not there yet, significant headways are being made in the right direction. Niels Thomsen, Group Vice President at Atos, gives an example where his organization helps one of its digitally forward clients simulate its new production lines in 3D. “It will be populated with real-time data from a live production line, helping those building the new plants to understand actual system behavior in parallel, improving the quality and time to value for the upcoming projects,” he says. “This approach can have profound implications for training personnel who work in high-risk industrial environments like oil and gas facilities, where AR/VR technologies can be employed to virtually experience crisis scenarios, without putting them in the harm’s way,” agrees Evan Woollacott, Senior Analyst, TBR.
Justifying the digital twin use cases
Nevertheless, a successful transition to digital twin-driven operations should be justified by the underlying use cases. It is a tool to bring respite for a clearly defined set of business problems. For instance, improved time to market, amplifying yield, or eliminating resource wastage. Digital twin’s contextualization is as vital as the technology itself to drive optimum ROI from it, reminds Niels Thomsen. He says, “Digital twins may range from simple mathematical models to full-scale systems with substantial physical presence. Organizations must develop the consciousness to relate the capabilities of the digital twins with the enterprise complexities they are intending to solve for realizing their full benefits.” Ezra Gottheil, Principal Analyst, TBR, observes that while the technology has obvious benefits, at times, it can be a challenge for businesses to establish their financial viability. Here, the vendor’s role escalates not only to develop affordable digital twin interfaces and empowering applications but also to help the customers discover their profitable use cases in the respective industry backdrops.
The digital twin requirements of an OEM engaged in manufacturing industrial systems will be different from that of their end-users who may employ digital twin solutions to find out how those systems actually behave within a networked environment, on a smart factory floor. Similarly, it is vital to inspect the steps a potential digital twin user needs to take from an IT modernization perspective to enable the desired outcomes and whether they possess financial rationality. Evan Woollacott says, “such transition strategies need substantial capital investments in either retrofitting the existing infrastructure or purchasing systems with embedded IoT connectivity to harness the data, solving the industrial networking protocol and data orchestration and standardization challenges to attain the desired Common Operational Picture (COP).” Does such expenditure make enough business sense?
Niels Thomsen points out that this is why Atos commenced its co-creation and co-innovation program a few years back. The accelerator merges the client’s specific use cases and business rationality with Atos technical expertise in building modular digital twin architectures that are agile and platform agnostic. They can be integrated and scaled easily at the client’s shop floor on-demand to address the targeted business requirements. The approach helps build relevant solutions that work and possess enough business traction to push them beyond the Pilot-Trap into the industrial tech mainstream. “As technologists, it is the job of the digital vendors to break down and explain digital twin constructs as potential solutions to the real-world business challenges that their customers are facing,” adds Ezra Gottheil.
Integrating AI with digital system models
It is equally noteworthy how AI is consistently being converged with the digital twin technologies, simplifying their use cases for the customers with smart control features. digital twin brings a series of standalone industrial systems within connected frameworks, setting the ground for machine consciousness to uncover less obvious kinds of synergies and interactions between them. Niels Thomsen shares his experience with a paper manufacturing client that realized the benefits of connected systems and AI assists. “With standalone systems, a local alarm goes off within the production facility only when an operational parameter is already breached at that point, halting production. However, within a highly networked environment where operational variables like temperature, pressure, and throughputs are consistently correlated and monitored across the ecosystem, an AI solution can predict a possible disruption at least 15 minutes in advance. It allows the controllers enough headway to initiate a shutdown or assume proper countermeasures,” he says.
Nevertheless, to make such smart control rooms feasible, experts recommend exceptional cohesion between the companies’ digital and operational disciplines. Niels Thomsen recalls one of its industrial partners dealing with the challenge of monitoring the live engine performance of marine vessels from their bridge. While digital teams were concerned with connecting the engine room sensor arrays with the onboard computers, to make it relevant, the connectivity had to be achieved within the ambits of strict marine operational safety protocols.
The strong intent for digital modernization needs to be reflected in the operational modernization perspectives as well to integrate assets, improve human-machine interactive experiences and amplify operational outputs. Here the digital and operational vendors have an overarching role to play by partnering to develop standards, nurture cross-functional insights and amalgamate expertise to help their customers propel futuristic projects beyond their Pilot-Traps. Such synergy is essential as Ezra Gottheil considers digital twin technology an excellent opportunity to abstract up from the individual Key Performance Indicators (KPIs) and concerns of the respective disciplines like IT, OT, support and services, and developing perception at a whole business system level.
Accelerating time to value and quality
Digital twin technologies, backed by the right synergies and cohesion, empowers companies to respond better to the market complexities and crisis. For instance, in the pharma sector, where the innovation and operational limits are being tested daily in the wake of the COVID-19 situation. Niels Thomsen looks back on his interaction with a pharma client, struggling with an inefficient vaccine production line, wasting costly raw materials. “We assessed the critical points in the production flow and identified the stage where the foundation for the vaccine is being developed by mixing the precursor chemicals. Atos teamed up with stakeholders who have a broader view of the enterprise and developed a digital twin of the chemical mixing process where a mathematical model indicates how the components should combine and react optimally. These theoretical inputs are compared with the actual chemical mixtures’ attributes by analyzing them with chromatography in real-time. The deviations are communicated back to the operators with the recommendation of the proper corrective measures,” he says. The approach allowed the pharma client to reduce wastage and optimize their OPEX and energy budget. The additional benefit is that the model can be quickly reverse-engineered to replicate its success across the other pharma production environments.
Rationalizing HR bandwidth and optimized talent management
However, for digital twin-based operations to succeed, automation and machine intelligence need to integrate with human supervision seamlessly, not only for safety but also to magnify the abilities of the experts to contribute better to the enterprise goals. Ezra Gottheil says, “It allows human resources to focus on their core competencies rather than wasting valuable bandwidths on iterative tasks that can be automated, improving the quality of throughput.” Evan Woollacott agrees. According to him, much of the process errors result from the over-involvement of human interventions. “Human-driven processing of inputs that can otherwise be automated may lead to persisting quality control issues, costing millions for companies. Here, valuable resources can be saved through intelligent Q&A monitoring across the entire manufacturing cycle of the products,” he says. Also, apart from addressing inconsistencies, digital twin allows companies to pace up the R&D process. With access to actual trial data repositories, the digital models can test the products virtually. The approach can predict the real-world behaviors and chances of success for the products, minimizing their development cost and time to market.
Addressing the security imperative
While using digital twins to accelerate enterprise goals, the businesses must also factor in the pervasive security implications of leveraging models to control actual processes that can be incredibly vulnerable to sabotage and intrusions. Niels Thomsen predicts a worst-case scenario where a baby food manufacturer’s production systems can be hacked to change the ingredients with apparent negative brand implications. “The particular business is now a potential ransomware target and offers enough opportunities for extortion and IP rights violation. Such lapses often result from unauthenticated resources like pirated copies of applications ending up in the connected operational environments,” he says. For instance, in Florida, malicious actors tampered with the city water supply system remotely to render it unsuitable for human consumption. Also, from a business perspective, ironclad cybersecurity is a prerequisite if the system integrators and their customers intend to push the digital twin projects beyond their pilot stages. It assures the various OEMs and components manufacturers involved to confidently share their proprietary information and device data, making a genuine system of systems with end-to-end visibility at the shop floor feasible.
In the current market context, characterized by scarcity of capital, fast shifting customer expectations, and hawkish regulatory oversights, digital twin technology offers an excellent tool to evolve and sustain the pace of innovation. However, it is a means rather than an end to scoring strategic business wins. The organizations will have to resolutely drive cultural convergence, gain cross-departmental insights, and foster trust and dependability across the value chain to successfully realize the benefits of digital twin technology in navigating market disruptions.