The Next Evolution of Additive Manufacturing

As more companies adopt additive manufacturing for end use parts and other applications, the need to print the right part the first time (and do so at scale, reliability and repeatedly) has exposed gaps in the development cycle that design, simulation and 3D printing software vendors must address with new types of tools.

At the Additive Manufacturing Strategies conference in New York City in early March, a number of key software vendors in the additive space gathered to discuss the future of digitalization, and the new requirements this would place on the software needed for 3D printing and other digital manufacturing applications.

Based on presentations at the show, it is clear that the future of additive manufacturing will rely on increased simulation of the printed parts and the printing process itself, along with more compute power and intelligence on the shop floor and throughout the lifecycle of the product. Major software vendors in the space have embraced the use of NVIDIA RTX™ GPU acceleration as a key enabler of these changes, and more powerful engineering workstations will be required to effectively support these processes.

According to a presentation by Smartech Analysis, excluding dental applications there were roughly 3.5 million metal parts printed in 2020, and nearly 1.5 million of those were end-use parts. Demand for additively manufactured parts and products will continue to expand over the next decade.

During the third day of the show, a number of panelists emphasized that additive manufacturing software (as well as the design software used to create the parts) would need to evolve to meet that demand. 

Igal Kaptsan, software general manager at GE Additive, in the software keynote, discussed the criticality of data and data management as manufacturers shift toward Industry 4.0 concepts. Currently, the data used across additive manufacturing applications exists in silos – on workstations, on 3D printers, in the post-processing department, etc. The impulse has been to centralize the data, but doing so without a common data model creates new problems.

“Today you have a distributed mess,” Kaptsan said. “When you put the data in one place, you have a centralized mess. You have to define a layer that knows what data needs to come in what format.”

He noted that additive manufacturing creates new types of data that cannot be managed using traditional PLM, MES and ERP tools. (At GE, a single print can generate more than 1TB of data.) Better management will require a mindset shift from the printer OEMs as well as manufacturers, who need to have visibility into the data associated with the entire design and production cycle. 

NVIDIA Omniverse Enterprise has emerged as a platform for open collaboration across these different types of design tools and a technology infrastructure for digital twin applications.  You can learn more about Omniverse in our previous coverage here.

Simulation of additive machine performance will also play a critical role. Manufacturers will need to apply different types of physical models to predict and monitor 3D printer performance in order to ensure quality parts. 

Tim Bell, head of CoC for additive manufacturing at Siemens Digital Industries USA, said that his company is tying its additive manufacturing activities directly into the concept of a digital twin – a virtual model of a machine built on simulation data as well as (eventually) data coming directly from the manufacturing floor as the item is built. 

“You allow software tools to iterate many times before you print anything. You can iterate fast by doing all of this testing in a simulation situation so we are not typing up machines. That way, you can ensure the first time print is the right print,” Bell said.

“How do we know the material is performing correctly? We do this through simulating each step of the process from build preparation to distortion compensation,” he added. 

Additionally, Siemens has leveraged simulation to help customers model entire factories to determine what additive equipment they will need, and whether their production approach will be cost effective. 

“We talk about digital twins and simulations, but this is not just pulling digital twin data out of machines,” Bell said. If we can tell the machine every attribute that is important to a successful print, feed the machine all that data, and then spit that data back out as a digital twin at the end of the process, the machine can tell you what it did. If we do this, and the digital twin is within 2% of your tolerances, then you can build some error tables to predictively manage the [ensuing] prints. You can know you got the first part right.”

This requires a new level of computing power. Software vendors like Siemens, Autodesk, PTC and others are leveraging NVIDIA RTX™ GPU acceleration in their simulation, design and visualization tools to enable the development of advanced digital twin applications, and hardware vendors like Dell Technologies are providing advanced workstations, edge computing solutions, and other infrastructure to support digital twin implementations. (You can learn more about these efforts in our previous coverage here.

Bradley Rothenberg, CEO and founder of nTopology, participated in a panel on additive software. “Not until design software is provided to enable the technology to make new types of parts and new designs, can we get the advantages of this whole process,” he said. “It is the responsibility of we who make design software to push the limits. Software right now is the bottleneck. We need to rethink how design software is built to take advantage of new technology to get the benefits across the full product development pipeline.”

nTopology now supports NVIDIA RTX™ GPU acceleration, which provides a performance improvement for visualizing workflows and complex geometries. Other vendors like Desktop Metal, Dyndrite and Hexagon have also embraced RTX™ GPU acceleration for 3D print simulation. 

Dyndrite, for example, offers the Dyndrite Additive Toolkit (with a fully native GPU kernel) for improving 3D printing workflows, which directly imports CAD design files and uses the data to drive additive manufacturing processes. Live Sinter software from Desktop Metal leverages the GPU to simulate the deformation of parts during sintering. The Digimat software from Hexagon and e-Xstream enables businesses to simulate the 3D printing process and calculate the total cost of producing each part

(You can learn more in our previous coverage here.)

These new design tools will be a key enabler of allowing companies to take the vast amounts of data they are collecting in manufacturing, and then feeding it back into the design process to inform new product development.

Companies like Autodesk and others have added functionality that makes it easier to propagate design changes across all downstream processes, create additive-specific features like support structures automatically, and quickly re-run process simulations.

You can access the Additive Manufacturing Strategies sessions on demand by registering here.

More Dell Coverage

Artificial Intelligence for Design and Engineering Workflows

In this white paper, learn how artificial intelligence and machine learning can improve design and simulation.

More NVIDIA Coverage