What is a digital twin?

What is a digital twin?

It really is what the name suggests: a virtual replica of something like a product, system or piece of equipment that mirrors its specifications and properties. But the defining feature of a digital twin is data flowing from the original source (often via sensors) that informs the virtual model in real time.

The idea might sound futuristic or like eerie science-fiction territory (particularly the clone example), but the digital twin concept has been around for longer than you might think. All the way back to the 1960s, in fact, when Nasa used a similar approach to examine spacecraft engines and understand the causes of failures.

Today, digital twins walk among us. Almost 75% of companies in advanced industries had already adopted digital-twin technologies with at least medium levels of complexity, reported McKinsey in 2023. And it’s not hard to see why. They present an abundance of opportunities for businesses to improve efficiency, product quality, and customer experience.

How is a digital twin different from a simulation?

You might be thinking digital twins sound like your average simulation. But don’t be deceived by their shared purposes and benefits – there are crucial differences. A simulation isn’t a substitute for a digital twin. You can create a simulated test or 3D model of a product, component, system, or process, without it being connected to the real-world equivalent. In fact, the subject of a simulation doesn’t have to exist at all.

A simulation is static whereas a digital twin is a live entity. In other words, while data can be fed into a simulation, it doesn’t run off a flow of real-time data. It’s for this reason that a simulation can’t evaluate the performance of a real object or process as accurately as digital twin technology. Simulations are more hypothetical scenarios using fixed data points (or even just a subjective impression), so their insight is theoretical.

A digital twin model, by comparison, stays up to date thanks to a constant data feedback loop, which means it can be used for continuous monitoring. While simulations can be cheaper to create, they are less effective and reliable as analytics tools than digital twin solutions. If a business wants to confidently predict an outcome, a digital twin is a much smarter investment.

What are the different types of digital twins?

Digital twins were initially used in product development, testing, and quality control of assets and prototypes. They then evolved into replicating discrete components. More recently, with the advent of business process management (BPM) and the evolution of process mining, digital twins’ role has branched out to digitizing, analyzing, and understanding business processes themselves.

There are four main types, let’s use the example of a car to illustrate how they each work, increasing in scale.

Component twins

These represent the individual parts of an asset or product (they can be known as parts twins) – an onboard sensor within a car, for instance. This type of digital twin lets companies drill down into asset or product performance, optimizing at an intricate level. This precision allows businesses to get to the root cause of a product deficiency, validating the effect of each component on overall performance. But component twins can also generate incremental performance improvements by perfecting each part, and targeting specific objectives in focused areas, rather than unnecessarily redesigning or overhauling entire products or assets.

Product or asset twins

Components come together to form assets or products. An asset twin could be a car engine, while a product twin could be the car itself. Product digital twins give a more complete and holistic picture of the asset in use, so companies can verify all parts and components work optimally together. Businesses can really start putting a product through its paces with a digital twin, testing performance in real-world scenarios before the product is signed off and sent to a customer.

System twins

Different assets work together as systems. In an automotive context, an example of a system twin could be a model of a maintenance plant or factory floor involving all the different pieces of manufacturing equipment. System twins are therefore useful for optimizing operations, looking at the flow of materials and products, and ensuring there aren’t clashes between any assets. This is significantly more efficient than continually visiting the factory floor and inspecting each asset within a complex system.

Process digital twins

The new kid on the block is a process digital twin. These allow businesses to interrogate a virtual rendering of a workflow or operation, as they would with any of the other digital twin types. For our car example, these digital twins could be the full end-to-end Order-to-Cash process (from the point a customer orders a vehicle to the point it’s delivered and paid for) or a micro-process within the supply chain or Inventory Management (such as from the moment a spare part is ordered from a supplier to when it’s used on the production line). All the same principles apply as with a physical object: a process twin is a digital representation of an operational flow that’s connected to live data from its real-world equivalent. Businesses can therefore explore and scrutinize the process in question as they would a product or asset.

• Find out the uses, benefits, and financial value of digital twins in multi-process environments with this study of 260 enterprise leaders, sponsored by Celonis and IBM

What value do digital twins bring companies?

Digital twin technology enables companies to validate the design of products and ensure they’re fit for purpose. The ability to assess products right down to the material level gives businesses a stronger understanding of robustness and an asset’s lifecycle, preparing for component obsolescence.

Since uses of digital twins are so diverse, they offer industry-wide benefits for any company:

• Safety: Protect customers and workforces by thoroughly testing products and systems, then monitoring them throughout their lifecycle.
• Savings: Minimize waste and costs caused by faulty products or suboptimal processes. Companies can assess modifications using the digital twin rather than building an expensive new prototype.
• Efficiency: Streamline cycle times by slashing rework and delays. McKinsey reports digital twins have cut total product development times by 20-50% for some users.
• Customer service: Improve satisfaction by preventing the delivery of faulty or poor-quality products. According to McKinsey, companies report products starting out using digital twins have 25% fewer quality issues when they enter production, and their enhanced quality can result in 3-5% higher sales.
• Clarity: Give concrete definition to otherwise intangible business processes and any blindspots.
• Proactivity: Predict, plan and prepare for maintenance, market change, or trends in customer preferences. McKinsey finds companies can increase revenues 5-10% by being better able to offer a predictive maintenance service to customers.

How is process intelligence transforming digital twins?

Digital twins have process intelligence to thank for their continued expansion. Process intelligence is accelerating process twins by fueling them with the data and business context they need. It provides the accurate, objective foundation of how a process is running, which businesses can then examine and optimize. And Celonis brings all this together in one powerful platform.

Understand and optimize your processes with the Process Intelligence Graph

The Process Intelligence Graph is the heart of Celonis Platform and it is system agnostic and without bias. The Graph integrates data from any and all sources to create a unified view of your end-to-end processes. Then, it enriches this data with the unique context of how your business works. Your business rules. Your KPIs. Your workflows. The PI Graph provides a process-centric view of your business operations that goes far beyond anything yesterday’s analytics tools or existing systems could provide.

• Process data: everything that comes from source systems, describing what’s going on in your business, including objects (POs and invoices, for example), events (the creation of an order or receipt of payment), and interactions (how POs are related to sales orders).
• Business context: everything that’s going on around the process, such as KPIs and industry benchmarks, as well as any process models.

This combination gives the Process Intelligence Graph the power to measure and guide action on end-to-end processes across business silos and source systems. Celonis also layers in a decade of our own standardized process knowledge and artificial intelligence.

Scaling this digital process twin couldn’t be easier. Businesses can rapidly extend from a subprocess to end-to-end processes by decoupling, linking together, or extending processes in the graph. Tailored end-to-end views allow companies to scope down their analysis, so teams can identify and address root causes in a specific use case, without having to navigate the full complexity. It’s therefore simple to manage and maintain governance as you scale the graph.

The graph helps unify data across functions like Order Management, production, and customer service. You can then see how something like a sales order flows through the business – but also how it interacts with service requests, lending new perspectives to problems by drilling into upstream and downstream dependencies. This then allows you to make decisions and changes thanks to that visibility.

What does all this mean for business value? One automotive manufacturer used Celonis to sync up parallel processes, reducing failed material calls causing production interruptions. A packaging manufacturer, meanwhile, saved millions across multiple processes, including the reduction of excess spare parts inventory through cross-process visibility.

Or take the chemicals company that’s significantly increased transparency to identify and reduce excess inventory. Celonis derives the planning strategy of materials based on other materials related through the bill of materials – building a model that links together procurement of the raw materials to the production of semi-finished and finished goods, and lastly distribution and delivery. All the interconnected material processes involved in creating a deliverable product are joined together. The end-to-end supply chain view this generates drives automated calculations of safety and target stock levels to minimize excess stock.

With the Process Intelligence Graph, businesses can connect all subprocesses for a truly holistic and comprehensive process digital twin.

Let’s meet your digital twin

Celonis users can activate Process Intelligence Graph features for existing processes to get started. From there, the solution is ready and waiting to expand by connecting additional processes to leverage new use cases.

Talk to a Celonis expert to find out more about the platform and how to tailor it to your most important use cases.