Digital twin and data: How Autodesk digitizes plants

The industry's digitalization vision often sounds like new construction, greenfield sites, and perfectly planned factories. The reality often looks different. Many production sites have grown over decades. “A new building is of course ideal for bringing along the digital twin from the very beginning,” says Jan Niestrath, Manager Industrial Machinery at Autodesk. “In practice, however, brownfield projects dominate.”

How digitalization is possible in existing plants is demonstrated by Autodesk at its Technology Center in Birmingham. The “Advanced Manufacturing Facility” was opened in 2018 as the first facility of its kind in Europe. The site serves as a real test environment in order to practically test new manufacturing technologies. The facility was equipped, among other things, with machines from DMG Mori, Hermle, Steifelmeyer and Hamuel as well as robots from ABB and Kuka.

In conversation with Produktion, Niestrath explains that the site is “our own small factory.” There, Autodesk tests its CAM and additive manufacturing software on a real machine park.

From the machine park to the digital twin

The digital twin is also used in Birmingham. Thus, a complete digital twin was created from the existing building from the 1980s. “Since current plans often do not exist, we scanned the building, that is, created a point cloud, and transferred it into a BIM model,” says Niestrath. This model is then supplemented with sensor technology; for example for temperature control, machine states and safety aspects such as fire extinguisher locations. Because anyone who wants to convert or expand an existing factory needs reliable information.

Digital scans can, for example, also help avoid collisions during planning and be transferred into dynamic models. In addition, "material flows as well as the movements of employees in the factory can be simulated in order to identify bottlenecks and optimize inventories," explains the Autodesk expert.

But how can one get started? Especially for companies with a limited budget? Niestrath advocates a pragmatic approach. "One must not create a technological 'beast' that no one understands," he says. Instead, the point is to use existing data, if necessary start with a scan, and connect the sensor technology via tools. In this way, initial use cases could quickly be realized, such as the monitoring of energy consumption in order to counteract immediately in the event of high costs. "It is important to start, even if you cannot do everything at once," says Niestrath.

BMW project shows the benefit of end-to-end data

New buildings are, of course, ideal conditions for taking digital models into account from the very beginning. One example of this is BMW's greenfield project in Irlbach-Straßkirchen for high-voltage batteries. There, the commissioning, acceptance and handover of the plant for the assembly of high-voltage batteries was digitized with the Autodesk Construction Cloud. Among other things, automated handover processes as well as drones for scans and collision checks were used.

Planning, implementation, and operation are not viewed as separate phases, but are connected with one another via data. According to Autodesk, 62 specialist models were merged into one coordination model. Around 25,000 previously manual workflows for data transfer were eliminated, around 450 working hours were saved. In addition, handover data is said to have been prepared with system support for downstream IT systems.

How Schneider Electric is driving AI in industry

Gwenaelle Avice Huet, executive vice president of industrial automation at Schneider Electric, sees AI in industry as the key to efficiency, energy, and sustainability. But whether Europe benefits from it will not be decided in theory, but on the factory floor.

Anja Ringel

AI as assistance and later as physical AI

Artificial intelligence also plays a role in Autodesk's strategy, although at present it is more supportive than autonomous. “We have long been using AI in an assisting function in our products such as Inventor, Fusion, or AutoCAD,” says Niestrath.

AI can now be found in nearly all Autodesk solutions and is used to generate design variants, automate processes, optimize simulations, as well as predict project risks. The Autodesk Assistant additionally serves as a central assistance function to which designers can give commands for workflows or automations in natural language.

In the long term, however, Autodesk is going further. Keyword: “Physical AI.” Autodesk is working with World Labs on this. The goal is, “that the AI ‘understands’ the world without each physical simulation having to be run through individually in a computation-intensive way,” explains Niestrath.

For industrial applications, this could become relevant if AI not only generates texts or facilitates operating functions, but can classify geometries, spatial relationships and physical dependencies.

Technology is not the biggest problem

Despite all technological approaches, implementation remains the critical point. Asked whether digitalization often fails less because of technology than because of change management, Niestrath answers clearly: “Unfortunately yes, often we do not have a technology problem, but an implementation problem.”

The best technology is of no use, “if the employees do not accept it or the commitment of management is lacking.” That is why, according to its own statements, Autodesk relies heavily on implementation partners that offer training and onboarding services related to the products.

Because the shortage of skilled workers is forcing companies toward digitalization. In the past, there were experts “who could hear from the sound of a machine what was missing.” This knowledge was often not documented. If such bearers of experience are missing, companies must find ways to secure knowledge systematically.