The growing momentum around systems-based thinking in supply chain planning has given rise to a new concept called the “digital twin”. There has been a lot of discussion around this topic lately and I wanted to offer a few insights, including around the importance of the data model in high-quality decision making using digital twins.
You can think of a digital twin as the ultimate ‘what-if’ scenario planner. It allows you to accurately test the resilience of a complex, multi-echelon, global supply chain. At the recent Gartner Supply Chain Executive Conference in Phoenix, analyst Marc Halpern clarified three key concepts about supply chain digital twins in his presentation “Busting the Myth of Digital Twins to Deliver Value.” He says:
- Digital twins are more than just CAD. These are virtual counterparts to the physical world that model a product’s uniqueness and its lifecycle. They link to and monitor real-world data in a simulated environment.
- Digital twins are seldom exact copies. They may reflect quite disparate aspects of the supply chain and products, ranging from supplier to service characteristics, and at many different levels of detail. But the extent and the thoroughness of the simulation depend on the importance of the information to the value chain.
- Digital twins do not likely anticipate all information. Unexpected needs can lead to ongoing changes and additions.
Digital twins drive benefits such as improved business planning and improved sourcing, procurement and supplier management. For example, this could be measuring the impact of random events on the supply chain, such as an earthquake that shuts down a distribution center. At ToolsGroup, our experience so far has been mostly in using the digital twin concept to precisely predict the impact of sales promotions on demand and inventory trade-offs on customer service levels.
How do you go about building a supply chain twin? The financial and operational parts of your business must be well orchestrated through a sales and operations planning (or similar) initiative. In their blog entitled “Prepare for the Impact of Digital Twins,” Gartner says, “Digital twins are not developed in a vacuum. Both the business concept and model must be tested against an economic architecture – revenue, profits, return on investment (ROI), cost optimization – and a way to measure progress as the products/services are rolling out.”
From a technology perspective, the underlying software must be able to ultimately orchestrate all data that can potentially impact a supply chain. This means it must be able to take many data types into account - from traditional supply chain inputs, to non-traditional corporate sources (like CRM data) and even unstructured, online data that factors into demand sensing. For these reasons legacy ERP or spreadsheet-based approaches aren’t suited for the purpose.
Instead, the key technology of the supply chain twin is a planning software platform that can model the entire supply chain starting from the demand signal through supply. Today’s supply chains are complex, orchestrated and adaptive systems whose processes flow in many directions, often simultaneously. As Gartner explains, the devil in getting digital twins right lies in the detail. They say, “The quality of answers or decisions generated by the software depends heavily on the quality of the data model, or how well the model represents reality at the point in time at which the answers are generated.”
The experience of our customer SKF, the world’s largest bearings manufacturer, bears this out. SKF built a twin, or digital model, of its entire distribution network with master data for 800,000+ SKUs across 40 installations of 5 different systems. It allowed them to transform from a regional to a global integrated planning model. It enabled planners to go from operating locally to being able to make global decisions based on full data visibility and full control of their reference data. Twenty-two warehouses have achieved full “autopilot” for two of SKF's factories.
Gartner explains Digital Twins
Gartner describes a digital twin is a virtual representation of a real object. Digital twins are designed to optimize the operation of assets or business decisions about them. Digital twins include the model, data, a one-to-one association to the object and the ability to monitor it.
“What’s in a Digital Twin?”
A twin may include data:
- External data received from sources outside of the twin
- Observational events received from the physical thing, such as sensor, log or meter data from an asset, or virtual sensor data that is calculated from other primary inputs
- Data received from other sources such as information about the cargo being carried on a truck, the name of the owner of the device, device serial number and historical maintenance records
- Derived data — data computed by logic that is within the digital twin
- Pointers to linked data - data about the thing’s environment (e.g., ambient temperature, local weather conditions) or objects indirectly related to the thing (e.g., owner’s name and address and other details beyond owner’s identity). These are not attributes of the thing itself, so this data does not belong in the twin. But the logic in the twin, or the logic in an application that uses the twin, may need to access this data.
A twin may include logic:
- Logic that is implemented as part of the digital twin, operating on input data within the digital twin or on data that is stored outside the digital twin. For example, a twin could calculate and store the time left before a truck will run out of fuel by applying a formula to the truck’s observed fuel level, fuel tank size and average speed.
- Logic that is physically implemented outside of the digital twin but is accessed by logic that is part of the twin. For example, the twin logic can invoke an API to an external decision service such as a geospatial mapping service that calculates a truck’s expected arrival time. This may use data from the digital twin on the truck’s current location and planned itinerary (identity of the next destination), and external information that is not in the twin (maps and real-time traffic information held in another system).