DfAM Requires Balancing Creativity and Constraints

In 2023, we published an article titled “The Hidden Cost of Post-Processing” (May 2023), prompted by the Additive Manufacturing (AM) industry analyst Wohlers Associates’ Specialty Report called “Wohlers Specialty Report on Post Processing: Final Steps of AM Part Production.” The report reveals, “Post-processing is one of the most crucial stages in producing high-quality and useful parts in additive manufacturing. It is commonly understood that the time and cost to perform post-processing is high. It is even higher in some cases than the actual time and cost of printing the part itself.” In this article, we explore a tough balancing act: adhering to DfAM (design for AM) rules without hampering creativity.

Manufacturability Check for AM

Bart Van der Schueren, chief strategy and technology officer, Materialise, believes DfAM is no different from the other CAM (computer-aided machining) software tools to help you check a part’s manufacturability. The process DfAM simulates and verifies just happens to be additive rather than subtractive.

“We tend to boast that with AM, you get complexity for free. In theory, that is true. You can design a complex part and get it printed, but you would also need to do post-processing, machining, and surfacing to finish the part. That’s the cost you pay for complexity,” says Van der Schueren.

Generating the support structures needed to print a part is the easy part, Van der Schueren notes. The challenge is, “to design something you can easily remove afterward. This has an impact on the overcall cost of your printed part. At the end of the day, you want to optimize the part for cost too,” he points out.

Since 3D printing produces a part by building the geometry one layer at a time (as opposed to chiseling it out of a block by machining), the vertical Z direction is also its structural vulnerability. In AM, the most sensible way to print parts with holes is to keep the holes in the vertical position. Such treatments can be part of the AI-driven manufacturability check for AM.

But introducing these constraints in the conceptual design phase is risky, Van der Schueren observes. “You don’t want to put too many limitations on the complex design you can explore,” he adds. “You don’t want AI-based rules to limit the creativity of the designer.”

Materialise offers Materialise Magics, which allows you to import your CAD design and verify its printability. Materialise also has a partnership with nTop, the AM-targeted software with implicit modeling features.

In June 2024, announcing the partnership, the companies revealed, “This collaboration has already demonstrated success in producing a high-performance, 3D-printed cylinder head for technology group Wärtsilä, significantly enhancing cooling performance and reducing weight by 60%, which was printed by Nikon SLM Solutions.

“Magics can show you different orientation options and the impact they would have on the support structures needed,” says Van der Schueren. “It can also show areas that you will have difficulty reaching in post-processing.”

So far, the rules of manufacturability in Magics are based on Materialise’s own print experience and cumulative intelligence. But the company is aware that some DfAM software users may want to add additional intelligence, to address the unique workflow and industry requirements of their markets.

“We plan to offer our customers the ability to reconfigure these rules or add other parameters,” says Van der Schueren. “We think that is important.”

Support for Various Supports

Chris Robinson, senior product manager for AM and Metal Forming, Ansys, says, “In general, you could reduce overhang angles, different thicknesses and transitions, and channel cleaning with DfAM. You can also orient the holes for best printing. If you take care of these in the earliest stage of your design, you could eliminate a lot of the post-processing work.”

As a simulation software maker, Ansys champions the use of process simulation to visualize and reduce printing problems in the manufacturing phase. For AM projects, the recommended workflow is to use Ansys Discovery, a design and analysis tool available for local installation or in the cloud, along with Ansys Mechanical Enterprise and Ansys Additive Suite.

“In Ansys Discovery, the topology optimization tools let you optimize your design for AM,” says Robinson. “Here, you can, for example, indicate the direction in which you want to print your part, and the software can highlight the down-facing surfaces that might become a problem. You can also let the software optimize the geometry for you to reduce or avoid overhang angles.”

With machine learning and AI fusion, Robinson believes AM software will soon be able to make process-specific recommendations. “We can do that for the laser powder bed fusion process with the Granta additive capabilities,” he says.

In 2019, Ansys acquired the materials intelligence firm Granta. Announcing the acquisition, Ansys wrote, “By integrating Granta’s technology into Ansys solutions, engineers will have access to high-quality materials information, selection and intelligence tools. This acquisition will also give Granta users an introduction to Ansys’ multiphysics capabilities. This will help Granta users improve their workflows with simulation.”

Deformation Compensation

In print preparation, the use of DfAM to test out the print orientation itself can reduce a tremendous amount of post-processing work and the associated costs. “If you have multi-segmented holes in your part, first you need to make it printable. The part may have been imported from the CAD program with the wrong orientation or the holes have multiple orientations, but in Oqton’s 3DXpert, you can make the holes printable with a single click,” says Roy Sterenthal, vice president of Software, Oqton.

This is a precursor to what DfAM software will be like when AI and machine learning are integrated, according to Sterenthal. Eventually, “The DfAM software will do what’s necessary to make your part printable. You won’t even have to think about the changes needed,” he says.

Oqton’s 3DXpert is described as an all-in-one AM software, covering preparation, optimization and print check. 3DXpert is optimized for the powder bed fusion printers.

Some post-processing work, such as removing the powder trapped in a chamber, is simply unavoidable. “But the important thing is to design it so you can remove the trapped powder easily,” Sterenthal points out. “To make that possible, you’d have to use DfAM with analysis and simulation tools.”

Support structures are also one of the unavoidable burdens in AM. They keep a part propped up to remain in place during the print process, but they have to be removed afterward as they’re not part of the intended design. Many print-preparation software [solutions] can now automatically generate the support structures needed, but, Sterenthal warns, “The software might have also generated those within internal or inaccessible channels, so you need a way to ensure an easy way to remove them afterward.”

Also, in AM processes that involve intense heat, parts invariably get deformed. DfAM software like 3DXpert allows you to design a part with what’s known as deformation compensation. “What you can do is to use DfAM with simulation, so you can simulate and compute the degree of deformation, then design the part to compensate for the deformation,” says Sterenthal.

Many CAD design software packages still use the BREP (boundary representation) method to describe geometry, but in 3D printing workflows, mesh models and implicit models (exemplified by software like nTop) are also commonly used. “Our strategy is to provide the users with support for all three types,” says Sterenthal. “This allows the users to use whatever method they see fit, and also sidestep the disadvantages of each.”

Calling for Closed-Loop Feedback

Brian Jeong, senior research and design engineer on DfAM at Autodesk Research, points out, “Removing metal supports is much more difficult than polymer. It’s important that your DfAM software allows using pin support—something you can snap off easily at the end.”

Jeong spoke from experience, as he was once involved in a project where removing the ill-conceived support bars inside a hollowed printed part took 30-40 minutes of strenuous manual labor, involving pliers.

Jeong recommends the integrated CAD/CAM/CAE tools in Autodesk Fusion for the workflow. “The major benefit is that it’s a seamless all-in-one workflow from designing in CAD, to DfAM with support generation, part orientation and process parameter optimization before directly sending it to the printer,” he says.

DfAM tools in Autodesk Fusion represent a mix of internal development and acquired technologies. They show footprints of Netfabb, NEi Nastran and Within Labs, all acquired by the company at different times in the last decade.

Many print-prep software and even mainstream CAD packages can identify problematic regions and topologies in parts destined for 3D printing. Whether it’s simple rule-based automation or AI-driven is up for debate. However, Jeong felt introducing such features too early in the design phase might hamper creativity.

“In the CAD design phase, you want the designers to have flexibility to experiment, to design whatever they want to design,” he points out. “It’s only when you import the part into the slicing software [for print preparation] that you would want alerts about these issues.”

There is high hope for AI-powered DfAM, especially in metal AM, Jeong suggests. “With metal printing, there are so many factors that can influence the printing process, but we also need lots of data to develop an AI model that can understand all these,” he notes. However, competing DfAM software vendors are not about to contribute their data to the common pool for AM training, because “many see data as the gold mine,” Jeong says.

Best practice at the moment is to develop a closed-loop feedback system, Jeong advises. “If you’re printing with laser powder bed fusion, for instance, you could collect the data from the printing process and results and use it as feedback to adjust the parameters and make error corrections,” he says.

This was the reason Autodesk struck a partnership with the startup 1000 Kelvin. As a result, 1000 Kelvin’s AI-powered DfAM software AMAIZE became available as a plug-in for Autodesk Fusion.

More Ansys Coverage

More Materialise Coverage

More Oqton Coverage

Best Practices In Moving Towards Large-scale Production Planning In 3D Printing

Planning and scheduling for additive manufacturing at a factory scale opens up an incredible array of possibilities, but also corresponding challenges. At the top of those challenges is the fact additive manufacturing - or colloquially 3D printing - tends to...