Digital technologies are becoming an important tool for keeping up with growth and supply-chain disruptions in the manufacturing world. The smart factory connects data across the value chain, from design all the way to production, in order to drive intelligence through manufacturing processes and enable decision-making in real-time.
Among these technologies are simulation software and Internet-connected sensors used in conjunction with one another to create what’s called a digital twin: a high-fidelity, virtual representation of a product or process that accurately replicates the form, functions and configuration of its real-life counterpart, in real time. Although digital twins have gained much attention for the role they can play in designing products and monitoring the performance of civil infrastructure, they have powerful and sometimes surprising applications in the manufacturing environment.
Digital twins have a broad scope because they can run countless performance simulations from unique perspectives to study and optimize a product across its lifecycle. Unlike traditional CAD-based simulations that are limited to parameters introduced by designers in hypothetical scenarios, digital twins use real-world data from their physical counterpart via sensors. Due to this constantly updated two-way information loop, as a product or process matures and evolves throughout its lifecycle, so does its digital twin.
By leveraging real-world data, digital twins can also help eliminate process duplication, improving productivity across teams and maximizing throughput. They can be used to track the accuracy of processes, along with the waste they produce, the time they take to complete and the energy they consume. The real-time data also facilitates quick response to customer demands and ever-changing market conditions.
Here are five key applications of digital twins in manufacturing:
1. Optimizing Factory Layout
When it comes to an existing factory’s layout, engineers and planners traditionally spend hours poring over drawings and models to figure out how to better utilize the available space. Miscalculations often arise due to the inherent limitations of drawings/models in accurately rendering the factory to scale and offering different vantage points. Reorganization projects tend to be trial-and-error ventures that prove costly and time-consuming when done wrong.
By contrast, digital twins offer a much wider visualization scope than models or drawings. They take multiple factors into account, helping manufacturers to precisely envision machine layouts and assembly flows so that pinch points and bottlenecks can be identified. Manufacturers can use digital twins to avoid unforeseen hazards such as robot arms swinging too close to human workstations, or automated guided vehicles interrupting walkways when they could take a better route.
Digital twins can help to improve worker ergonomics too. Existing data on employee postures and movements can be utilized by manufacturers to identify opportunities for automating certain parts of the assembly line to prevent injuries. Digital twins can also be used to guarantee safe passage toward exits during any kind of emergency.
Re-configuration is another important consideration. During the early days of the pandemic, the world witnessed automakers pivoting to make ventilators, and distilleries shifting from liquor to hand sanitizer. Digital twins simulate the retooling or repurposing of equipment, and predict the impact of such changes for faster decision-making and increased flexibility.
2. Expediting Operational Processes
In a similar vein, digital twins can optimize operations in the factory, such as pre-commissioning and assembly line processes. Programmable logic controller (PLC) codes can be validated virtually and integrated with manufacturing and enterprise systems—cutting commissioning time down to a fraction.
Digital twins can be used to simulate countless scenarios to arrive at the best outcome, as opposed to traditional approaches that employ guesswork. For example, the most expedient assembly and disassembly processes can be calculated for every model that a design engineer introduces into the digital twin. The digital twin can also be harnessed to create work instructions for manufacturing engineers to build and inspect the end product as efficiently as possible. Once the real-life product is ready, its quality is verified and data is generated for further training the machine learning models within the digital twin.
3. Improving Product Design and Sustainability
Digital twins’ capacity to leverage real-world data can be used to optimize product design as well.
Take the example of an autonomous vehicle and its digital twin. Real-world data from sensors on the physical vehicle can feed simulations on the digital twin, helping predict functionality, schedule maintenance and serve as a true-to-life testbed for future upgrades, such as battery chemistry modifications. Digital twins can even be used to test software updates to ensure that no performance issues arise should the updates be rolled out to the real-world counterpart.
One example is Tesla, which creates a digital twin of every vehicle it sells. Data from sensors on thousands of working vehicles streams into the factory, not only providing insights on how the vehicles are performing but also guiding Tesla’s updates to improve car performance. If a problem is found, Tesla can roll out a software update to remedy the problem—for instance, fixing a rattling door by adjusting the car’s hydraulics. Tesla uses the two-way information loop of digital twins to go beyond predictive maintenance and software updates, and optimize the design, prototyping and testing of newer vehicle models.
Manufacturers can also use digital twins to improve sustainability for future iterations of their products. Rather than determining the environment impact of an updated product after completing its development—which is often too late—digital twins can be used to simulate that product across its entire lifecycle, from design and raw material sourcing to reuse/recycling potential and end-of-life.
The virtual environment of digital twins offers the additional benefit of reducing physical prototypes, leading to less material usage during multiple iterations of product design.
4. Training Industrial Robots Through Simulations
Another innovative way of incorporating digital twins in smart factories is using them to train robots.
Machine learning requires an enormous amount of training data that encompasses thousands of scenarios. For instance, training a robot in bin-picking requires exposing it to a variety of shapes, sizes, colors, textures, lighting conditions and so on. Traditionally, this means robots would require training across every single one of these scenarios.
A digital twin is useful not only for training perception systems using a combination of real and synthetic data, but also for testing whether the algorithms work in simulation before the code is deployed to physical robots. This saves a massive amount of time and allows manufacturers to achieve industrial automation much faster.
5. Training Operators Through Virtual Reality
We’ve talked about training robots, but let’s not forget about the humans in the factory.
Training workers in a physical factory setting can be expensive due to machinery downtime. Every wrong move by a worker can risk their personal safety. The training is also less effective as there is a limit to the number of trials and vantage points that workers can be exposed to, especially if practicing on a machine that’s already costly to run.
Digital twins enable manufacturers to implement high-fidelity virtual reality (VR) training to accelerate proficiency in subject matter and improve knowledge retention. Immersive learning allows workers to retain processes by repeatedly practicing their hands-on approach in a flexible setting, considerably decreasing their likelihood of making errors in the future.
When it comes to safety training, workers can strap on a VR headset to prepare for emergency situations without putting their lives at risk. Unlike with a real-life training exercise that involves time constraints, VR training within a digital twin allows workers to step through a process as many times as they deem necessary to master it—without needing to worry about wear-and-tear on the machine or other costly downtime. As in the case of robots, digital twins allow workers to train themselves for different scenarios using a combination of real-time and synthetic data feeds.
Digital twins can help manufacturers realize some serious benefits—from optimizing factory design to facilitating worker training. In doing so, they may provide companies with the competitive advantage that’s needed in today’s rapidly changing world.