Probing the “hows” and “whys” of the critical elements of digital transformation has been the subject of a series of articles from CIMdata (link here). Today’s article focuses on the “how” and “why” of digital twins.
Let’s start with a quick review of what a digital twin is.
A digital twin is a virtual representation (aka a digital surrogate) of a physical asset or a collection of physical assets (the physical twin) that exploits data flow to and from the associated physical asset(s). Fundamentally, a digital twin is enabled and supported by robust end-to-end product data connectivity and model-based systems engineering (MBSE) processes.
The continually updated information in a digital twin makes it a key enabler of new business models, innovation and competitiveness. The digital twin also holds the geometry, materials, components and behavior of an asset or a group of assets within its many iterations. This includes the as-designed, as-produced and as-maintained BOM views. Later in the article, we will delve into this and show where to find this information.
Digital twins take one of three basic forms:
- Product-based digital twins are for physical things and assets such as aircraft and their engines, construction machinery, mining equipment, drilling rigs, locomotives, medical devices, etc. that are complex, long-lived and capital intensive. Product digital twins are often used to maintain and optimize service life over decades of use.
- Production-based digital twins describe the means of production, such as industrial machinery, power-generation equipment and material flow and movement. Production digital twins focus on operating parameters while tracking scheduled maintenance and changes in use.
- Service-based digital twins are for infrastructure-related assets such as power grids, railways and utility systems that are geographically dispersed. Service digital twins usually monitor for disruptions and when or where upgrades are needed. They are always seeking to represent, completely and accurately, the as-is state of the infrastructure.
There is never a single digital twin. Every product, process or service has multiple digital twins addressing specific needs at various key points in each business unit or enterprise. Digital twins will be found in design and development, engineering, finance, production and operations, test and inspection, data analytics, modeling and simulation, maintenance, recycling and remanufacturing, and more. Many digital twins even contain other twins.
Digital twins must be connected to and fed by digital threads to provide value consistently and accurately. These “threads” are really webs that gather continually refreshed data from ever-changing sources. Without digital threads, as I noted in my last article, there is no end-to-end connectivity and digital transformation will fall short of its potential.
Basically, a digital twin without a digital thread is an orphan. At best, users of orphaned twins can maintain and update them manually—by scrounging for the data they need, verifying it on the fly (if at all) and typing it in. Given the opportunity for error, decisions based on an orphaned twin are never reliable. Furthermore, orphaned twins may never represent the true context in which the product was designed, produced and/or operated.
Developers must also bear in mind that each digital twin must have a physical twin or asset with which the digital twin must have associativity, connectivity, traceability, sensors and some form of communication. Additionally, unless changes are infrequent, these communications should be automated and in near-real-time.
The Why of Digital Twins: Benefits from Implementation
To better understand the why of a digital twin, a look back at the forces behind digital transformation is illuminating. As the term suggests, digital transformation converts all formatted data and information to 1s and 0s. Documents become blocks of data and a product’s multiple annotated 2D CAD views are combined into information-rich 3D models.
Getting rid of formats in this way presents an opportunity to anyone seeking information, simplified access and a corresponding challenge: exactly how is the data accessed? This is where the digital twin, digital threads and PLM come to the rescue—accessing, contexting, using and managing the indispensable data and information that had been tucked away in dozens of different formats.
The digital twin combines all these new inputs to illuminate key decisions and their impacts. If we know what went into any given decision—or can quickly find out—prior successes will continue to enlighten us.
Without this information at our fingertips, we will be unable to keep pace with fast-moving marketplaces, tough competitive environments and tight budgets. If that happens, the enterprise’s marketplace clout will shrink. It will lag in new-product development, on-going product enhancements, meeting demands from customers and avoiding threats from competitors.
Visual representations of physical things have been around since the dawn of the Industrial Age. These representations can now be kept relevant—connected end-to-end, bidirectionally and inseparable from their physical, real-world counterparts.
A digital twin’s relevance refers to how effectively it can be accessed, how readily it is updated and how easily it can be managed over time. Clearly, we have moved far beyond exploded diagrams.
Despite all the recent press lavished on digital transformation, CIMdata’s 2020-2021 Cloud SaaS Survey showed that digital twins have yet to take over the world. A recent CIMdata webinar pointed out that completely connected digital twins, and end-to-end digital threads, are still not common, although most enterprises are working on them.
Nevertheless, CIMdata’s Cloud SaaS Survey turned up solid rationales for digital twins. These included better collaboration across engineering functions, more first-time-right products and a better understanding of the product. Survey respondents also spoke of more consistency in applying product knowledge during development, increased business agility and better ways to address complexity.
Calculating ROIs for digital twins is always good for gathering support for implementation. But these numbers should be crunched by each business unit in light of its own goals and needs. Aggregated benefit tallies from user surveys are subject to interpretations that may vary too widely to be meaningful.
The How of Digital Twins: Connections for the Digital Twin
There is an essential need to create links to major sources of data and information that the users of digital twins need. This is not intended to tell developers how to build digital twins. Don’t be misled by the term “link;” it can refer to a myriad of sophisticated digital connections.
Enterprises should make sure their digital twins link to specific output files and data records in all the tools used in creating new products and getting them to customers. This is a long list that includes CAD/CAE, EDA/MDA, PDM/PLM, ECRs, M&S, analytics, QA/QC, MES/MOM, CAM, T&I, TDPs, MRO, and others. These abbreviations should be familiar to anyone who develops and delivers industrial products and services.
The digital twin must also connect to an organization’s ERP or MRP systems to generate manufacturing Bills of Materials (mBOMs). Don’t overlook the all-inclusive Bill of Information (BOI), which is an all-inclusive digital twin construct. Links to key decision points in the enterprise’s supply chain are also recommended.
Databases in sales and marketing, distribution, purchasing and warranties should not be bypassed either. Changes in any of these can directly impact the management of existing products (and any other assets) and the development of new ones.
Feedback from relevant online bulletin boards should be accommodated; every industry has online discussion groups and communities that should not be ignored.
More broadly, digital twins should link to enterprise systems for managing and tracking engineering changes, configuration management and data governance. These are key parts of enterprise-level simulation-driven environments, which can enhance innovation as well as more savvy day-to-day decisions.
Developers should also take care that their digital twins intercept information about evolving regulatory requirements and industry standards as they emerge. A link to the Digital Twin Consortium could also be helpful.
As Digital Twins mature, enterprises should consider the use of:
- Artificial Intelligence (AI) to track developments in all those linked repositories and come up with new insights.
- Machine Learning (ML) to keep the links and other internal parts from becoming obsolete.
- Augmented Reality/Virtual Reality (AR/VR) to simplify comprehension and speed uptraining.
Start small and keep it simple while building capabilities to allow the digital twin to be defined, maintained and to evolve over time. Reach an agreement with digital twin users on the benefits they expect and develop realistic support and upgrade plans. Likewise, respect the concerns of the owners of the data that is to go into the twin—keep in mind the risk of unauthorized access, for example.
As soon as they are implemented and connected, digital twins will help enable faster, more sound decision-making, reveal unexpected opportunities and improve failure prediction to prevent unanticipated downtime.
This means that digital twins will drive innovation and performance. They will help the enterprise improve the customer experience by deepening the understanding of customer needs to develop profitable enhancements to existing products, improve services, tighten operations and enable new business models. Ultimately, digital twins will lead to new collaboration opportunities between the experts on physical-world products and the data scientists who help us understand what our data is telling us.