A

s physical assets become increasingly software enabled and more complicated in their construction, a new model is required to operate them – a model that will not be fully possible without the creation of digital asset replicas. These replicas are called digital twins.

While there is a range of definitions for what digital twins are, the simple definition is “a digital representation of a physical asset.” This digital replica may include all of the components required to run and maintain that asset, from 3D computer-aided design (CAD) renderings to bills of materials and failure codes.

Digital twins are used by organizations to determine how their physical assets perform under certain conditions or how they monitor asset performance in real time. Powered with data from sensors attached to the physical assets, digital twins enable the creation of robust failure models. They help organizations understand asset criticality, right down to the individual parts each piece of equipment is made of, and then share that information across teams. As a result, digital twins become critical in transforming organizations and driving digitization initiatives end to end.

Many organizations want to create models that demonstrate how the various elements of the Internet of Things (IoT) will be brought into their operating reality and merged with existing ISO, Six Sigma and total quality management (TQM) aligned methods for ensuring reliability. To achieve this, they have to apply rigor to specific areas where they define the what, when, how and why of their IoT deployment. These decision points align with the Internet of Things knowledge domain that is found in the Uptime^® Elements:

1. Source: The ready availability of digital twins will revolutionize the process of onboarding complex equipment. Onboarding is often a siloed process, rife with paperwork and prone to human error. When industrial equipment is purchased, maintenance plans must be built based on multiple conversations with original equipment manufacturers (OEMs). Materials lists must be created and entered into systems manually, and stocking occurs based on assumed part criticality and allocated budgets. With each stage managed by a different team, the process is plagued with inefficiencies and can take months to complete. For example, digitizing something as large and complicated as an entire oil refinery could take as long as five or six months. This assumes it would be based on physical blueprints and paper documents from its structure to its equipment inventory, processes and failure modes. An initiative like this would completely dominate the time of plant engineers.

Digital twins can streamline this process to get the system up and running immediately. Rather than rely on slow, manual processes, the digital twin leverages existing digital replicas of the individual components that reside within the larger system. (These digital assets would have been created by the owners of the components during the development process.)

2. Connect: Many in the industry say that IoT sensors continue to drop in price every year. This is to encourage organizations to enable their enterprises with the IoT now. But, people often underestimate the various requirements that come with connecting their equipment, their enterprise and their industry. Additionally, there are many application programming interfaces (APIs) and other development approaches. And, many questions on issues, such as access, information flow and storage, must be answered before an organization can build a truly secure and purposeful digital nervous system. Every connection point creates a possible vector for a cyberattack. Therefore, each connection needs to be designed, developed and deployed to be secure; every single one. This is a lot of work without a platform that helps drive uniformity.

Digital twins that are used in operations with live data, as well as in collaborative development situations with manufacturers, operators and other third parties, require secure workspaces and communications. Otherwise, the integrity of the digital twins and the data they house cannot be maintained. Connectivity with asset sensors allows organizations to get closer to a full operating twin that mimics the experiences of the physical assets. But, it demands that the system is protected and facilitated as it’s shared and updated by multiple parties.

3. Collect: Gartner estimates that there will be 20.4 billion IoT devices by 2020, creating more than 500 zettabytes per year in data. There is no doubt that companies are struggling with what data to collect, where to store it and how to make it usable.

Digital twins create a logical taxonomy of the data from the IoT, as well as the categorization and use of that data. The ability to shape the data set based on physical, digital and electromechanical attributes allows enterprises to manage their data requirements. Digital twins manage by exception and operating processes, such as anomaly detection, asset modifications and customizations. They filter the massive volumes of operating data that flows through the enterprise, limiting what needs to be captured, cleansed, stored, reused and actioned. By definition, the digital twin encompasses this collection phase, pulling in everything it knows to be true about the physical asset.

4. Analyze: As mentioned in data collection, data analysis is more effective based on volume of data. But it’s less efficient based on the need to manage, merge and cleanse the volumes of data to create data sets that analysts and data scientists can use. By employing digital twins to categorize data and visualize and contextualize it, teams can compartmentalize their activities, create hypotheses and communicate results to non-technical or non-analytical peers. The addition of statistical analysis tools and models that execute a plethora of calculations on a rich model with multiple attributes makes it easier to predict failure, flow and feasibility. Modeling tools that enable the simulation of an issue or enhancement in a future state are already common in the virtual reality space, whereas digital twins allow physics to guide the analysis in a pseudo real-world sense. As such, whole production lines, factories, vehicles and systems can be tested before major investments or reductions in investments are made. This leads to the benefit artificial intelligence (AI) can bring to improving operations for physical assets. The digital twin holds the data that is collected and fed into the AI models to uncover key insights.

5. Do: If done correctly, a digital twin allows owners and operators to perform simulation and what-if analysis on their physical assets. If the physical asset is IoT enabled and accepts commands to actuate on the asset, remote and autonomous operation becomes possible. By combining the IoT with AI for anomaly detection and prediction, you create the asset of the future.

The vision of the digital twin is to enable good decisions that deliver value to your physical and digital operations and support your ongoing industry transformation. While the digital twin will solve many issues, the achievement of its vision isn’t possible unless you digitize your physical assets and their subassemblies. You must also work together to develop standards for how digital twin resources should be structured to achieve consistency across consumption technologies.

Regardless of the challenges, there is no question that even in the very physical, asset intensive industries (e.g., energy and utilities, oil and petroleum, manufacturing and industrial products), digital assets are becoming just as essential as physical ones. Digital twins are a revolutionary solution to help ensure that companies achieve efficiency and efficacy in their operations, and that they continue to bring value to their customers.