The terms digital twin and digital thread often come up in conversations about PLM. Now and again, we tend to use new words to describe old concepts that have been tweaked and updated. In my blog post, I will make the articulation of digital thread and digital twin much simpler, extending from the concept of PLM.
The beginning of digital twin and thread
The digital journey started back in the early mid 90s with CAD systems such as CATIA, ProE, UG, and others. We have been doing 2D or 3D modeling of parts and products for decades, but can we call this a digital twin?
Let’s take the example of a simple product – a copier. We’ve created a 3D model of the copier and called it a “twin” This twin is just a representation of a copier, without any intelligence. To build intelligence, we can start with the requirement specifications and “attach” them to our model, or “twin.” We can further add design specifications, simulation data, and manufacturing plans, etc. With an unique identifier for the twin, we can also retrieve all the added intelligence from it. This is the simplest representation of a “thread.” This concept can be leveraged and extended to create a digital thread of many assets in different contexts.
During simulation of the manufacturing plan, if we detect any manufacturability issues, such as collision, etc., we can rework either the design or the manufacturing plan. Professionals in the PLM field are familiar with and can easily relate to this activity. We have created data related to requirements, design specification, manufacturing plans and stored them in the PLM system for a product. In other words, a thread is already created. This thread is built into, or attached, to the twin where all the intelligence has been captured until now and can be easily retrieved, analyzed, worked on, improved, etc. I would like to emphasize that we have not built the physical product yet and the thread we have created is in the virtual world.
Digital twin and digital thread of a smart product
Due to the rise of connectivity in the physical world, it is now possible to get insights into how a product or asset will behave in operation. Various parameters, or properties, of an asset can be captured using different types of sensors and connectors. Going back to our copier example, let’s assume that we have successfully launched our smart product copier, which has a host of sensors and connectors, on the market. We now have a continuous data stream from our smart copier while it is in operation which can be analyzed using artificial intelligence, analytics, and machine learning.
Let’s consider the following use case:
When the print quality starts to deteriorate, the customer will place a request for cartridge replacement. Typically, the user will have to place an order for the cartridge and then replace it once received, and a substantial amount of time is lost in the whole process. A smart connected copier would come up with an intelligent solution based on the data received and infer that the cartridge needs a replacement in a week. The store would then ship the cartridge to the customer in advance to avoid any downtime. The customer replaces the cartridge and this information is also attached to the twin. This means the twin is now live since we are getting the real-time data and the thread can now also be referred to as a digital thread.
PLM consultants would typically ask how a copier can be identified. The answer is serialization, which will facilitate a unique digital twin. Every configuration in the series will be captured and recorded with its own digital twin representation.
Using digital twins in manufacturing and production planning
Data captured from the manufacturing floor can provide insights into the actual manufacturing under way with respect to the manufacturing plan that was created and simulated in the virtual world. It provides insights into the real-world manufacturing scenario that may necessitate a change in the manufacturing plan. What would happen on a factory floor where we have MES systems and other assets, if there is an issue during manufacturing? The local corrections made during the MES operations will be fed back and stored in PLM system as a record. It is possible to revisit and revise the manufacturing plan for future use. Extending this concept to a typical manufacturing floor, manufacturing assets (machines, tools, robots, etc.) or a cluster of assets can interact with each other digitally and decide instantly if there is a need to switch from one manufacturing station to the other in order to avoid any delays in the cycle time. The inputs come from different sources – the manufacturing floor, the intelligent material itself – and all these are analyzed so that appropriate decisions can be made faster. This is referred to as the twin of a manufacturing floor or a factory with a related thread. This manufacturing twin, though centered around making the factory smart, can also provide insights to our product twin.
Different perspectives of digital twins
In order to get the full benefit of the digital twin and digital thread, organizations are beginning to understand that they should view the twin from different perspectives – a product perspective, which focuses on the product properties in operation; a manufacturing and operations perspective, which focuses on all the insights during the manufacturing; and finally a customer experience perspective, which focuses on how the product is used by the end customer and provides deep insights into future designs and innovations. In each of these perspectives, an enormous amount of data is generated.
With adequate assessment of data from the thread, it is important to ensure that right data is going into the twin. To make the right decisions, we need meaningful data. Emerging technologies – IoT, cybersecurity, artificial intelligence, etc. can be effectively leveraged to work on data and draw meaningful inferences. Data is generated from multiple sources due to multiple IT systems being deployed in an enterprise. Some important data may go undetected, hence it is important to connect the dots with appropriate connections or proper sewing of the thread.
Looking at this from the customer perspective, no customer would want their product to stop functioning, particularly in complex products such as airplane engines. Data from the real world provides so many insights that a failure can be predicted well in advance to take preventive measures and eliminate or minimize downtime.
PLM as the backbone
It is relatively rare that concepts appear out of nowhere without any precedent. In the case of digital twin and digital thread, they have their origins in CAD and PLM, respectively. In order to ensure that the extensive amount of data coming from various sources is leveraged well, PLM will continue to play the pivotal role here in acting as the backbone in storing all the product insights and driving future innovations. Capabilities of PLM systems and platforms are expanding in order to accommodate these changes so that at any given point we have a single source of truth in the form of a digital twin enabled by a digital thread. In the age of smart and connected product development, PLM is further evolving and emerging as an innovation platform to enable digital thread.
Building a robust data model to accommodate the new sets of data and then relating them to different characteristics of the product are the most important steps. An architecture leveraging the latest technologies and cloud is equally important step in ensuring that the true benefits of digital PLM are effectively achieved.
It is important to know that digital twins are not just restricted to smart products but to the manufacturing process, entire manufacturing ecosystem, etc. Although these things sound futuristic, they are happening today.