Carolyn Cairns—Global Industry Marketing Lead, Manufacturing, HPE
The concept of a digital twin is not new—it means a digital representation of some aspect of the physical world. For manufacturers, that could be a product, the assembly line that manufactures that product, or an entire factory.
According to IDC, 20% of G2000 manufacturers will include the industrial metaverse in their digital twin road map to address simulations, collaboration, and safety by 2026 (IDC 2023 FutureScape: Worldwide Manufacturing 2023 Predictions).
Fifty years ago, what can arguably be called a digital twin was used to determine how to bring the Apollo 13 astronauts back to earth after an oxygen tank explosion damaged their craft beyond repair. The “,” comprising 15 simulators networked together with the most advanced computing of the day, allowed mission control to work in collaboration with the astronauts to simulate various options and test “what-if” scenarios on the spacecraft—before the astronauts actually took action that could waste valuable oxygen and power. If you watched the movie, you know the three astronauts arrived safely home.
Connected digital twins
Advances in IoT, sensor technology, edge computing and increased computational power ushered in a wave of “connected” digital twins that allow simulation models to be updated based on how the physical object is performing in real time. Formula 1 racing enthusiasts are likely aware that racing teams such as Red Bull Racing Honda have become edge-to-cloud, data-driven enterprises. They are virtually designing a new car every two to three weeks in hopes of winning the next race.
Before the driver even sets foot on the track, the teams have run a host of race simulations based on track conditions, the latest design mods to the car, anticipated maintenance, and pit stops, just to name a few. On race day, real-time vehicle and driver performance is relayed back to the model to allow the racing team to make real-time adjustments on the fly.
Today, digital twins owe much to the gaming industry and the strides made in 3D graphical representations of the physical world. It’s now possible to create a lifelike representation of a product from concept to testing and simulate its manufacture through the assembly line—or simulate the construction of an entire factory. The benefits are obvious—faster and more-effective simulations can improve product quality and accelerate R&D cycles. The more simulations are done digitally, the less physical prototyping and resources are required, which reduces the amount of scrap and rework needed if mistakes are made and defects occur.
But simulations alone are primarily physics based. They show the impact of conditions on products such as wind resistance, extreme temperatures, or how plant floor asset and assembly line performance will be influenced by temperature, dust, and vibration.
Generative AI and “what-if” scenario planning
With artificial intelligence (AI), you can start giving your factory brains by incorporating historical data with real-time factory data. It allows for “what-if” scenario planning by using the past to predict the future.
Advances in generative AI and large language models have made it possible to summarize large amounts of text ranging from plant asset manuals to production logs, product performance data, and customer sentiment. Imagine being able to talk to your industrial digital twin. It would be like having a digital manufacturing mentor on the shop floor to help reduce guesswork and provide such recommendations as what designs have the greatest likelihood of success in the market and what simulations should be used to yield the fastest, most accurate testing results—without compromising product quality or safety.
And what if such a digital mentor could advise on changes needed to the assembly line to manufacture this product, or predict when plant floor equipment needs to be maintained, and talk the worker through the steps required?
If you are interested in learning more about large language models and digital twins, I highly recommend Dr. Goh’s talk at the recently held Bosch World convention. He also addresses swarm learning. For manufacturers, having the ability to share insights across multiple factories, or even multiple assembly lines, can help improve productivity and profitably. Data sharing back to the factory digital twin is difficult due to data silos, proprietary systems, and the need for digital sovereignty and compliance. With swarm learning, the insights from the data can be shared without having to move the data, to help create more adaptive, self-healing, and autonomous factories that learn from each other.
It sounds simple, but it’s far from easy. While it’s relatively simple to implement a proof of concept (PoC) on a single line in a single factory, many manufacturers struggle with scaling successful AI initiatives across their manufacturing enterprise. Data comes from many disparate sources, and none of it is the same. Sensor data, video streams, manufacturing, and PLM data all must merge with enterprise data in the digital twin. At Hewlett Packard Enterprise, we’ve developed a data-centric architecture framework based on the HPE GreenLake edge-to-cloud platform to help manufacturers unlock value from all their data, and successfully move proofs of concept into production.
Join HPE and Intel at Hannover Messe 2023
We’ll be in hall 14, booth H48. Please stop by and speak with our experts to learn more about how HPE and our partners are modernizing manufacturing from edge to cloud to create more intelligent factories using industrial digital twins, generative AI, and large language models.
Stop by the Aleph Alpha demo to see how HPE, in collaboration with Intel, is tackling challenges such as quality assurance, maintenance and repair, and unexpected manufacturing malfunctions that can be costly in terms of downtime, lost productivity, and risk.
For more information or to request a meeting, please visit our Hannover Messe 2023 registration page.
Carolyn Cairns is responsible for developing marketing strategy, content, and priority solutions for the manufacturing and industrial sectors at HPE. Her diverse business experience includes business development, product management, and marketing roles in the telecommunications, public sector, and IT technology and services industries. She has supported numerous pursuits and accounts within the food and beverage, oil and gas, and industrial sectors. She holds an Honors BA from the University of Toronto.
HPE_Alliances
HPE Alliance Partners
