Of all the AI-driven initiatives emerging from the enterprise these days, perhaps none is more far-reaching than the digital twin. But what, exactly, is this, and how can it be leveraged for greater productivity and profitability?
What Is a Digital Twin?
In short, a digital twin is a virtual representation of a real object. This could be something as simple as a product or as complicated as a manufacturing process.
Twins can also be used to represent theoretical entities, such as a potential customer or an emerging market. The idea is to mirror these objects or systems as closely as possible in a digital environment, then subject them to all manner of external influences to see how they respond.
If the response is positive, then hopefully, they will be duplicated in the real world.
Keys to Success
In a recent McKinsey Talks Operations podcast, the company’s Kimberly Borden and Anna Herlt point out that digital twins have the potential to streamline operations, reduce waste and duplication, develop customer insights, and perform a host of other valuable functions. But only if they are developed and conditioned properly.
One mistake that many organizations make is to implement simple simulations or CAD modeling and call it a digital twin. But this misses the mark in that a twin typically involves multiple models, all being fed by real-time data. This allows the twin to represent an entire life cycle, from initial inception to development, testing, deployment, and ongoing monitoring and refinement.
In other words, a simple simulation represents a key moment in a process, while a digital twin represents the full value chain of the product or system.
Requirements to Consider
Naturally, one of the key requirements in a digital twin ecosystem is a strong data architecture, which will rely on not only sophisticated technology but the skillsets to direct it toward a robust and accurate representation of both the element being modeled and the environment it will inhabit.
Turning the results of digital twin projects into tangible benefits usually involves targeting them at multiple problems at once, says Nancy White of digital transformation specialist PTC (Power To Create). For instance, when trying to improve customer satisfaction, a digital twin can be used to gauge individual preferences, assess purchasing habits and even streamline predictive maintenance procedures to strengthen brand loyalties and produce positive feedback.
If increased sustainability is the goal, digital twins can help evaluate different materials, analyze carbon emissions, and improve manufacturing, supply chain, and recycling processes – always looking for better – and often cheaper – ways of doing things. The overall idea is to gain increased visibility into the inner workings of the enterprise business model to ensure steady progress across a number of fronts and simplify digital transformation throughout the entire organization.
Not the Metaverse
To some, this may sound like the enterprise’s entrance into the metaverse, but the reality is there are some key differences between the metaverse and a digital twin. For one thing, says Computerworld’s Mike Elgan, the metaverse is all about leaving the real world to exist within a digital, artificial world, whereas the digital twin is about leveraging an artificial world to make the real world better.
A good analogy of the digital twin was the replica of the Apollo 13 spacecraft that was used to test the various methods to get the three astronauts of that ill-fated mission safely back to Earth when the real module suffered a near-catastrophic explosion. By linking the physical module with its twin via a robust communications channel, engineers at NASA were able to accurately predict what was going to happen in space before they took action, greatly improving the chances of success at each stage of the rescue.
In the modern age, this model is based largely on AI-driven processes that can calculate results based on available input data in near-real-time, making it a crucial factor in restoring systems and processes that have failed. Equally important, it can be used to predict failures before they happen and even redesign environments and operations to reduce points of failure or eliminate them altogether.
Are High Costs Worth It?
Creating a digital twins ecosystem is not cheap, however. As InfoWorld’s Isaac Sacolick highlighted recently, you’ll need a range of tools, such as CAD software, some of which will need to be highly specialized, along with a scalable data architecture that will likely reach into the petabytes. This will probably require any number of cloud deployments, all of which should be optimized for AI, AR/VR, and a host of other emerging technologies.
Beyond that, however, organizations will also need a clear vision of what it hopes to accomplish with their digital twins. When a typical twin of a commercial office building can cost upwards of $1.7 million, simply throwing data into the mix and hoping something useful comes out is not a very cogent strategy. Organizations should know ahead of time what sort of game-changing opportunities they hope to pursue and who, exactly, will derive the most benefit – preferably in ways that can be monetized or used to reduce costs.
The good news about digital twins is that, while expensive, they are quickly reaching a cost premium that is within the reach of most large enterprises – meaning they will soon cross into the realm of medium organizations and then to the small business tier.
At the same time, the infrastructure needed to support them also supports a wide range of related digital transformation initiatives, which allows the full cost to be distributed over a range of programs.
With digital twins already seeing broad success in so many mission-critical areas of the business model, the way forward for those just entering the field is clear – and most of the risks are already well-known.