Global MFG - Sep 23, 2019
Additive Manufacturing: The Quest for AutomationAustin Weber | Assembly
However, additive manufacturing, which enables engineers to consolidate parts and produce components in complex shapes, is quickly becoming an alternative. The technology is evolving from the world of low-volume prototypes to high-volume production-ready parts.
Until now, one of the only things holding 3D printing back from wider use has been a lack of automation. But, recent applications involving automated guided vehicles (AGVs), collaborative robots, high-speed printers and other equipment will enable manufacturers to boost throughput, control quality, reduce cost and speed time to market.
“Labor represents more than 30 percent of the cost of producing production parts,” says John Dulchinos, vice president of digital manufacturing at Jabil Inc., a global manufacturing services company that is at the forefront of applying additive manufacturing technology on its assembly lines. “Typically, this is in post-processing and inspection, [which drives] up costs and leads to inefficiencies and inconsistencies. Adding automation can minimize touch labor, save costs, and improve consistency and quality.”
In addition to direct labor cost, there are hidden costs such as operator time for scheduling print jobs, uploading CAD files and loading machines, as well as low printer utilization rates due to lack of availability of operators to remove parts or the cost of print bed removal from parts in post production.
According to Dulchinos, automation will help additive manufacturing take the next step and enable it to move beyond just being a tool for low-volume applications. Jabil is using robots to automate part handling in secondary processes and computer vision to automate inspection processes.
“Today, 3D printing focuses largely on prototyping,” explains Dulchinos. “End use applications require high levels of consistency, tight tolerances and low costs. Automation can help reduce material handling and increase quality.”
“Automation will be critical to making additive manufacturing work in high-volume applications,” adds Harold Sears, technical leader for additive manufacturing at Ford Motor Co. “It will help address problems with variance and inconsistent part builds. Automation will add stability and enable us to transform 3D printing from a batch process to a continuous process that will work much better in a volume production setting.”
“We’re hearing more interest in automation from companies such as Tier One auto part suppliers and contract manufacturers,” says Larry Lyons, vice president of product at Desktop Metal Inc., a company that has attracted investment from BMW and Ford. “They have told us that additive manufacturing equipment needs to fit into an existing factory ecosystem that is already quite automated. They’re demanding machines that can coexist with conveyors, robots and factory management software.”
Lyons claims that automation is essential to bring 3D printing into the high-volume manufacturing market where it has barely touched the surface so far. “Building out factories of hundreds of printers won’t work economically without automation,” he points out. “To get the right part price, automated material handling is critical.
“We’ve been incorporating a lot of hooks in our hardware and software so that our equipment can be easily migrated into a traditional automated factory setting,” notes Lyons. “We’re taking lessons learned from traditional manufacturing and applying them to the development of next-generation, high throughput additive systems.
“For instance, we’re designing our equipment so that it is flexible and can easily interface with robots and other items typically found in an automated industrial environment,” says Lyons. “We’re putting vacuum conveyance systems in place to move powder more rapidly. Automated depowdering also enables the machine to remove loose powder around parts without the need for operator involvement.
“Many manual processes are time consuming, labor intensive and ergonomically challenging, plus they introduce safety concerns,” adds Lyons. “The process of removing a build box from a printer is a good example of something that’s typically handled manually. An operator rolls up a cart, lifts the build box from the printer and transfers it to the next stage in the production process. In the near future, the cart will be replaced by an AGV, a conveyor or a robot.”
Using additive manufacturing even for rapid prototyping applications typically requires many manual steps. Parts need to be designed taking into consideration process constraints and limitations. In addition, parts need to be inspected after printing. Support material also needs to be removed and post-processing steps need to be carried out.
“Automation becomes crucial when we transition additive manufacturing to production environments,” claims Wojciech Matusik, an associate professor of electrical engineering and computer science at the Massachusetts Institute of Technology (MIT). “In this scenario, cost of manufactured parts cannot depend on the manual labor that is required to design and produce these parts.
“Additive manufacturing automation will enable companies to go from functional specifications to a completely customized product with minimal human labor,” says Matusik, who leads the Computational Fabrication Group at MIT’s Computer Science and Artificial Intelligence Laboratory. “Each stage that is automated will effectively reduce tedious human labor.
“Automation of all stages will ultimately drive down the overall cost of manufactured products,” explains Matusik. “This, in turn, will allow additive manufacturing to be a cost-effective solution for a wide range of products.”
“Automation is an essential aspect for successful additive manufacturing programs primarily because it has a positive impact on part cost,” adds Andrew Edman, industry manager for product design, engineering and manufacturing at Formlabs, a 3D printer manufacturer that supplies companies such as Ford, Gillette, Sony and Tesla. “As part costs fall for additive manufacturing, it gets closer to achieving parity with traditional manufacturing techniques, especially for high-mix, low- or medium-volume production scenarios.
“Beyond pure part cost considerations, automation helps manufacturers develop new models of how and where parts are made,” notes Edman. “On-demand spare parts, just-in-time production, shorter supply chains—all of these things become more likely when the means of production is more automated.”
As with any manufacturing process or technique, Erdman says it’s really about identifying the right applications and running the numbers. There are part geometries and applications today that make sense to produce with additive manufacturing, but often those come from taking the same kind of design for manufacture and assembly approach that you would with any other production method.
Automation will expand the reach of additive manufacturing, enabling it to be used for less optimized designs or simpler parts.
“When it comes to cranking out hundreds of thousands or millions of the same part, additive is further off,” warns Erdman. “Matching the production speed of traditional manufacturing is not the core value proposition of additive manufacturing, it’s flexibility.
“Ask your current plastic part vendor to run five of one part, 11 of another and 300 of something else, all in different materials and have them to you by the end of the week without a $500 setup fee on each one,” says Erdman. “If they’re using traditional manufacturing methods to produce those parts, it’s going to be a problem.
“If they’re using additive manufacturing, that level of automation at the machine level can easily make those numbers work,” claims Erdman. “Additive manufacturing is great at addressing complicated demand and supply issues or helping maintain a lean inventory with low risk.”
Companies such as Fast Radius Inc., Voodoo Manufacturing Inc. and Xometry Inc. are already writing the next chapter of additive manufacturing. The service bureaus operate banks of machines that print a wide variety of production-ready parts in high volumes. Automation plays a key role in their business plans.
“Currently, much of additive manufacturing is boutique, being manipulated on a case-by-case, file-by-file, basis,” says Greg Paulsen, applications director at Xometry, which works with manufacturers such as BMW, Dell and General Electric. “To scale, the results require some level of automation to give a predictable and repeatable result end-to-end.
“[Increased use of automation will] solve two issues: trust in the product and throughput,” claims Paulsen. “Right now, there are a small handful of processes from the 3D printing umbrella that can scale to high volume.
“Most of these are plastic, laser powder bed fusion processes like selective laser sintering,” says Paulsen. “This is because the materials are limited and the post-processing is relatively standardized. Many processes are still very high touch, from build planning to post-processing. This limits how effectively you can scale.”
“One of the misperceptions about additive manufacturing is that you can just buy a printer, plug it in and get good parts out,” adds Lou Rassey, CEO of Fast Radius. “Actually, it’s a much more complex process that requires many variables to be dialed in and controlled.
“Additive manufacturing has now crossed the tipping point,” claims Rassey. “It’s gone from a low-volume prototyping tool to a high-volume process that can mass-produce parts of a caliber that can be bolted onto a plane or attached under the hood of a car.”
Fast Radius mass-produces plastic and metal parts for customers in a wide variety of industries, including automotive, aerospace, consumer goods and medical devices. Customers range from Husqvarna to Steelcase.
“Many new additive manufacturing machines are designed for high-volume production rather than one-off prototypes,” explains Rassey. “By applying automation, we see an opportunity to improve efficiency. For instance, we’re using robots to automate the build process by removing trays and washing parts.”
Another company at the forefront of implementing additive automation is Voodoo Manufacturing, which operates a 3D printing “farm” comprising more than 200 printers. It eventually plans to scale up its operations to 100,000 printers. To achieve that ambitious goal, Voodoo recently embarked on a “Project Skywalker” initiative. The goal is to harness automation to cost-effectively handle large production runs that can compete with traditional plastic-injection molding.
“Automation is more than simply a way of cutting costs,” says Jonathan Schwartz, chief product officer at Voodoo Manufacturing. “It’s the only way we’ll survive to be a large company that employs hundreds, if not thousands, of people. In taking on such a massive and deeply entrenched industry such as injection molding, automation is going to be our primary weapon.”
To improve additive manufacturing efficiency and throughput, companies need to automate all of the downstream processes that occur after a part is printed. That includes unglamorous steps such as removing build boxes from printers, recycling excess powder, loading and unloading cure ovens, and testing and inspecting parts.
The simple task of removing 3D-printed parts typically requires manual labor. “It remains an on-going challenge for operators,” says Steph Sharp, CEO of 3DQue Systems Inc., a Canadian startup. “One of the most common ‘automation’ solutions relies on removing the part and the print bed from the printer as a single unit either using a conveyor-type system or robotics. These solutions do not reduce the cost or time required for part removal; they simply defer it to the post-production stage.
“Manual tasks in post-production include separation of parts from the print bed and redressing the surface,” explains Sharp. “Due to the relatively low equipment and material cost, even an extra minute of manual labor significantly increases the unit cost and inhibits the ability to scale operations.”
“Post-processing is a challenge for every printing process, and [each one] has different requirements,” adds Formlabs’ Edman. “During printing, the part geometry has to be supported in one way or another, but when it comes time to get your part, those same support structures become the enemy in terms of time, automation and material waste. In some cases, you have the option to design a part geometry to be self-supporting, which greatly reduces the time that goes into post-processing.”
Steps with the highest concentration of labor include part unloading and inspection.
“Automating the unloading of parts can increase printer utilization,” says Jabil’s Dulchinos. “Automating post-processing and inspection can reduce labor, improve operating efficiencies and reduce defects introduced by part handling.
“Automating part removal from printers is easiest because the parts are in a known location and the working envelope is well defined,” notes Dulchinos. “That said, today’s 3D printers are designed around batch processing. A build starts, is completed and stops until parts can be removed, new material loaded and the printer starts the next cycle.
“Tomorrow’s printers need to be designed for continuous production, which includes automated material loading and part removal,” Dulchinos points out. “This will enable higher throughput and lower part costs.”
Post-processing is much more difficult to automate because parts often vary significantly in shape and size. In addition, many part locations are not well defined. Post-processing steps include freeing parts, removing supports from the build plate and recycling powder. It also includes processes relating to surface finish, such as painting, machining and sanding.
“There are many variables involved, depending upon the application,” says Ryan Martin, principal analyst for smart manufacturing at ABI Research. “For instance, a low-volume powder bed-based aerospace part may not require a lot of post-processing other than removing the supports. A high-volume automotive interior part may require a better finish, because fit, finish and cosmetic appearance need to be taken into consideration.
“Furnaces can be expensive to operate and it takes time for parts to sinter and harden, so some processes could be automated,” explains Martin. “In addition, automation is needed on the test and inspection side of additive manufacturing.”
It’s important that future automation solutions break the bottlenecks related to post-processing and material movement. Tasks must be broken down appropriately to maximize throughput.
“Machine loading and unloading is a natural priority if you can standardize part handling for various geometries or use flexible gripping and workholding solutions,” says Joe LaRussa director of seat industrial engineering at Brose North America Inc., a Tier One auto supplier.
“Material delivery and management is another area that’s ripe for automation, because you have these kinds of systems for other processes that require powder or pellet delivery like coatings or injection molding operations,” adds LaRussa. “Of course, these are more infrastructure-driven, but they can still alleviate a lot of indirect labor needed to keep machines fed with raw material.
“Additive machines bring a small additional challenge unique to the powder-based processes in that they require some initial cleaning or extraction prior to moving to the next process,” notes LaRussa. “This is different than a lights-out subtractive process where machining centers can make parts that are then removed from their fixture and are ready for the next process immediately.”
According to LaRussa, quality control is one of the most challenging aspects of additive manufacturing automation.
“The classical training for process design and development is to implement processes that are stable, in control and capable (statistically) to produce quality,” says LaRussa. “With traditional manufacturing methods like stamping, forming and machining, this kind of approach is natural.
“With additive manufacturing, however, the process is inherently variable for every single part made,” claims LaRussa. “So, traditional methods for statistical process control are challenging, particularly for safety-critical components.
“In this case, manufacturers are forced into 100 percent verification of quality for every piece, and this will require high automation both on the shop floor and in the quality function, since you have to deal with voluminous data generated from the production system,” LaRussa points out. “Manual analysis will never be able to keep up with the production system.”
Brose’s biggest challenge right now is addressing the product and process validation topics. LaRussa and his colleagues also face unique challenges because a lot of their components are large and are produced at a volume where additive manufacturing is still uncompetitive vs. traditional manufacturing methods.
“A lot depends on the specific process,” says Davide Sher, senior analyst at SmarTech Markets Publishing LLC, a market research firm that specializes in additive manufacturing. “In metal powder bed fusion, the biggest issue is post-processing: primarily support removal, which takes time and limits geometry.
“In metal binder jetting, the main issue is furnace-sintering, so the biggest challenge is automating the design process to account for part reduction during sintering,” adds Sher. “Automating powder removal and sieving, powder recycling and powder supply is a challenging issue in all powder-based processes.
“There are also significant challenges in automating process monitoring to the point where the machine is able to recognize errors and stop the process or even fix them automatically,” warns Sher. “This requires implementation of sensors and advanced artificial
Additive manufacturing has not yet reached a maturity point where a design can be generated and fabricated automatically based on given requirements of an application. Generative design software cannot properly handle complex constraints and requirements or make sure the designs it generates are manufacturable. Digital workflows also cannot predict and fix problems before they happen.
“When combined with slow fabrication speeds, expensive materials and an inexperienced workforce, additive manufacturing becomes an expensive trial-and-error experiment that is [less] suited for volume manufacturing,” says Ersin Uzun, vice president and director of the System Sciences Laboratory at PARC, a Xerox Co. “However, additive manufacturing processes are continuously getting faster, and usually higher material costs can be justified with better designs that use less material to provide better performance.
“The financials for additive manufacturing will start to make sense for volume applications only when we have automated workflows that can help generate and validate the best designs for given application and engineering requirements,” claims Uzun.
“AI is already making its way into generative design and topology optimization, as well as into process planning, modeling and monitoring that combines multi-physics modeling with machine learning, planning and advanced reasoning,” explains Uzun. “Those are really important and hard problems, and when solved they will amplify the value proposition of additive manufacturing with better designs, [resulting in] stable, predictable, repeatable and more automated manufacturing processes.”
Robots to the Rescue
Increased use of robotics promises to reduce variation, improve consistency and boost throughput.
“Additive manufacturing is a very competitive process until you try to scale it,” says Joe Campbell, senior manager of strategic marketing and applications at Universal Robots. “If you scale it without automation, it requires a lot of labor. Eventually, robots make economic sense.
“Because of the batch nature of additive manufacturing, multiple machines are running different parts and companies are in constant setup mode,” explains Campbell. “Low-volume, high-mix production environments are common. Typically, there’s a constant influx of new parts, so rapid redeployment is essential.
“Collaborative robots are a perfect solution, because operators can be in and around machines as needed without having to shut down for changeover,” claims Campbell.
Several companies are currently using UR10 machines to streamline their additive manufacturing operations. For instance, Fast Radius uses a robot attached to a mobile platform that can be manually moved around its facility on the West Side of Chicago. Voodoo Manufacturing uses a robot to harvest (loading and unloading build plates) a cluster of more than 25 printers at its factory in Brooklyn.
In Germany, engineers at Daimler AG recently developed an integrated process cell that features six-axis Fanuc and Kuka robots, in addition to Grenzebach AGVs.
“The AGVs transport the additively manufactured components between individual stations,” says Jasmin Eichler, head of future technologies at Daimler. “They enable high flexibility in the factory layout, leading to an easy-to-scale additive manufacturing production cell.
“Robots are used to automate downstream production processes,” explains Eichler. “All transport, placement and removal activities in the post-processing area are done by robots.
“They take the build platform with the parts from the setup station and place it in a furnace for subsequent heat treatment,” Eichler points out. “The same robot then removes the platform again and takes it to a 3D optical measurement system for quality assurance purposes. Finally, the build platform is conveyed to a saw, which separates the parts from the platform, making the components ready for further use.”
Robots can also be used to do the actual printing. Instead of having a 3D printer, which typically print in 2D on a stacked horizontal plane, the printing head can be mounted to the end of a four- or six-axis robotic arm to provide an extra degree of motion.
“Robots are ideal for additive manufacturing applications involving large parts,” says Nicolas De Keijser, assembly and test solution line manager at ABB Robotics. “They feature a wide build envelope that can’t be achieved with traditional 3D printers.
“The extra degree of motion enables manufacturers to change the path of dispensing,” notes De Keijser. “Instead of a 2D path, they can increase the structural dynamics of the part being printed and make it stronger.
“In addition, robots can be used to tend standard 3D printing machines,” explains DeKeijser. “The programming for a machine-tending operation can be readily adapted from what has been long used in the plastic-injection molding industry for a variety of tending and post-processing applications.”
A new generation of 3D printers are addressing the need for high-volume production and fast throughput. Some machines feature built-in conveyors and material feeders, while others are designed to easily interact with robots.
“This is an interesting time for the additive manufacturing industry,” says ABI Research’s Martin. “We’re going to see many things happening in the next four years in regards to high-volume production systems. There will be a lot more interest in integrating that equipment as part of regular workflow.”
For example, Brose recently invested in a German startup company called AIM3D, which is developing machines such as the ExAM 255 equipped with automated material feeders.
“The company has developed a new method for additive manufacturing,” notes LaRussa. “It makes it possible for the first time to print materials typically found in automotive series production.”
The ExAM 225 machine can print either metal or plastic parts. In addition, it’s the first industrial 3D printer that can process pellets that are conventionally used for injection molding, enabling it to use a wider range of materials than any other competitor. The machine contains an automatic material feeder and can take up to 1 liter of material per extruder. The material hopper can also be refilled during the printing process.
“Until now, plastic 3D printing has failed to meet today’s manufacturing needs due to the high cost of part removal and lack of end-to-end automation,” says 3D Que’s Sharp. To address that issue, the company recently unveiled QSuite, which eliminates the need for manual tasks such as job scheduling, part removal, print bed reset and printer restart. Instead, QSuite uses a proprietary suite of hardware and software technologies to mass-produce high quality plastic parts.
“There’s no need for specialized or dedicated operators,” claims Sharp. “QSuite comes complete with fully automated dynamic scheduling that reprioritizes jobs based on changing deadlines or parts and operates 24/7 in a continuous production loop. It provides fully autonomous part removal and delivery and gives operators complete control through real-time reporting and management data—all managed remotely so users can do other work while the printer delivers parts.”
3DQue also recently developed a module called OPoD that features internal conveyors and collection bins, enabling “truly autonomous fabrication of plastic parts.”
“Increasing interest in automation helps validate that additive manufacturing is getting broader adoption and indicates that things in the industry are speeding up,” says Scott Turner, director of advanced research and development at 3DSystems Inc. “You don’t need to automate a slow process.
“Companies are becoming more automated as they scale up to serial production,” explains Turner. “That’s why we launched a module that automates material handling and completely takes the operator out of that step.
“The DMP Factory 500 is a scalable manufacturing system designed to transform metal [printing] through simplified workflows to build higher quality seamless metal parts with lower total cost of operation,” claims Turner.
The customizable machine features five function-specific modules, including a powder management unit that automatically de-powders parts on build platforms, automatically recycles unused powder materials and prepares for the next build.
“The modular design of the DMP Factory 500 enables continuous function of all metal 3D printing and powder management modules to maximize uptime, throughput and operational value,” says Turner. “The integrated automation minimizes manual processes to reduce total cost of operation.
“Zero point clamping enables optimal positioning of the build plate, facilitating a quick transition from the 3D printer to post-processing steps,” adds Turner. “This integrated feature reduces setup times and provides enhanced flexibility by quickly transitioning the build plate from the additive process and sending it downstream for post-processing, saving significant time and money.
“This [machine] will contribute significantly to increased automation possibilities and reduction of build plates costs,” claims Turner.
Another printer addressing the automation challenge is Desktop Metal’s new Production System, which uses a binder jetting process that ‘glues’ metal powder by printing through an ink-jet head.
Two state-of-the-art print bars containing 32,768 nozzles work in conjunction with powder spreaders to disperse metal powder and print in a single pass across the build area, jetting up to 3 billion drops per second. Parts are automatically removed from the build box and cleared of any loose powder that remains in channels and crevices in preparation for sintering. Heated to temperatures near melting, remaining binder is removed causing the metal particles to fuse together and the parts to densify.
The Production System features bi-directional printing where all steps of the print process—powder deposition, spreading, compacting, ballistic suppression and binder jet printing—are applied with each pass over the build area.
“Whenever there is movement, there is printing—making it the fastest way to print complex metal parts,” claims Lyons. “It delivers the speed, quality and cost-per-part needed to compete with traditional manufacturing methods. It’s the fastest way to print metal parts at scale.
“The vast majority of metal printing systems currently available are laser-based systems that melt powder layer by layer,” notes Lyons. “That’s an inherently slow process—each layer can take 30 seconds or more—that does not lend itself to high-volume applications.
“We wanted to take a different approach to 3D printing,” says Lyons. "Our Production System has a much higher throughput than anything that's currently available. In fact, it's about 100 times faster.
“We have customers already planning to use the machine for high-volume applications that call for 100,000 or more metal parts per year,” explains Lyons. “That type of drastic throughput advances should be achievable within the next two years. It will be a stepping stone on the road to print runs of 1 million, followed by the next threshold of 10 million.”
Fast Radius recently became one of the first companies in the world to use the Production System.
“[It] represents an entirely new manufacturing platform that will enable new part geometries, improved functional performance, and a more favorable cost structure for innovative product developers,” says Rassey. “The machine can print a wide variety of metals, including tool steels, low-alloy steels, titanium and tool steels. It also boasts the speed, quality and cost-per-part needed to compete with traditional manufacturing processes."
To further address the automation issue, Rassey and his colleagues developed a software platform called the Fast Radius Operating System. “It supports customers across the product lifecycle and is a key enabler of the Fast Radius Application Launch Program,” he explains. “The platform helps them identify potential applications, conduct engineering and economic evaluations, accelerate new product development, and ultimately manufacture industrial-grade parts at scale with the latest additive technologies.”