When baking a cake, what you put in isn’t the same as what you get out. Biting into soft golden sponge doesn’t taste like egg or flour or sugar; it tastes like cake – or it should if you follow the recipe. Additive manufacturing is a little bit like that. You may know your ingredients at the start – the material, the design, the oven you’re using – but there’s chemistry and physics happening in between, a recipe, that causes the part to become something new. It depends heavily on precision. There’s variability – the potential for porosities, warping, surface irregularities – that makes quality and inspection of 3D printed parts a challenge, and as the technology’s adoption in production applications grows, that challenge is only deepening.
In a recent issue of the International Journal of AI for Materials and Design, founded by Professor Wai Yee Yeong, a paper on ‘Machine learning techniques for quality assurance in additive manufacturing processes’ penned by Surajit Mondal and Shankha Shubhra Goswami, stated that “despite its transformative potential, AM poses unique challenges, particularly in the realm of quality assurance.” As opposed to the established quality control methods of traditional manufacturing processes – visual inspection, dimensional inspection, non-destructive testing (NDT), for example – “the dynamic and additive nature of AM introduces new complexities and uncertainties that traditional quality assurance methods may struggle to address effectively.”
“The biggest difference is that additive manufacturing is actually creating a material,” Kirill Volchek, Chief Technology Officer, Oqton, a developer of AIpowered software for additive AM, told TCT. “Physics is very complicated and you’re getting one type of raw material that you’re putting inside the machine and then you’re getting a part out – process creates your actual material, digitally, and that’s a major difference.”
In CNC machining, a billet of aluminium placed onto a 5-axis machining centre will be the same chunk of material at the end, just in its desired shape. But Volcheck says the adoption of process simulation technologies, which allows close monitoring of printed parts in-situ, is enabling higher success rates for AM. Oqton, for example, recently partnered with EOS to integrate Oqton Build Quality, an AI-powered tool specifically for metal powder bed technology, and provide users with full end-to-end traceability of their parts. The software evaluates build performance to detect, prevent and correct anomalies and defects, allowing issues to be spotted and corrected early.
“Additive is just so different than what we’ve done historically,” Noah Mostow, Business Development Manager at Phase3D, a developer of in-situ monitoring solutions for AM, concurred. “With additive, you are creating both the material and the geometry at the same time. With forming, we’re looking at creating just the material property and controlling the material property because of your form, which is your geometry. Within subtractive, you have your material properties, you’re turning it into the correct shape. Within additive, we are doing both of those things at the same time. Where the challenge comes in is, how do you control all of the parameters that go into creating those two key pieces of material and geometry?”
Phase3D’s flagship products, Fringe Research and Fringe Qualification, measure every layer of an AM build, creating heatmaps to help users of primarily metal powder bed machines ensure quality and productivity. Earlier this year, the company unveiled its work with the U.S. Air Force and NASA to develop Fringe Research to validate two materials on two different laser powder bed fusion machines. Phase3D believes it is the first inspection company for AM to measure anomalies during the build that lead to porosity, which is said to be a major cause of part rejection for both organisations.
“We’re now hitting a period, especially in the aerospace world, where designers want to create parts that they either cannot inspect using traditional methods like CT scanning or it’s unbelievably time-consuming and expensive,” Mostow said.
All of the stuff that AM is supposed to be superior at – time compression, freedom of design, and reducing scrap – risk being cancelled out by the challenges imposed by post-printing inspection. The layer-by-layer structure of AM introduces the threat of defects and anomalies at each stage of the printing process, meanwhile, its ‘limitless’ design benefits can render inspection by traditional means largely ineffective for complex geometries.
“With the capability of having more design freedom, more complex geometrical designs have been developed to reduce material cost, reduce the number of parts and/or to enhance performance of the intended part or product,” says Dr David Menzies, Chief Commercial Officer at Additive Assurance. “Hand-in-hand with this new design space, it creates complexities in the manner by which quality control and inspection can be performed. Additionally, metal additive manufacturing processes, such as laser powder bed fusion, are extremely intricate. The smallest changes in the manufacturing process, such as powder batch-to-batch variation or laser output, can lead to different solidification characteristics and result in variability in quality.”
Additive Assurance is an Australian developer of process monitoring and quality assurance solutions for laser powder bed fusion. Its AMiRIS product is an independent in-situ process monitoring solution which uses high-resolution sensors to monitor the metal LPBF process to the micron level and deliver real-time monitoring and quality control. It’s completely machine agnostic and can operate across a fleet of machines. In August, the company announced a partnership with Additive Industries to integrate AMiRIS into its MetalFAB systems.
“Quality is a paramount concern for all manufacturers regardless of the processes that they use to generate their products,” Menzies continued. “Significant investments are made to ensure that quality irregularities can be identified as early as possible in manufacturing processes. The cost of the lost production time and raw materials versus the profitability of the production run dictate how much investment is needed and at what stage of the production process to mitigate this risk.”
Cost remains a huge barrier for AM adoption, and post-printing steps such as inspection can be a major expense.
“The biggest challenge in additive is the price of a part,” says Volchek. “So trying to reduce it and to make sure that part could hit the proper price target, we need to ensure that there are no additional costs involved in the process. So we need to prevent issues.”
Thinking preventatively can save costs further down the line. According to Volchek, having a full end-to-end view of the process is vital – “it’s already too late when you’re at the end of a process.” It also means being able to prepare and account for any deviations that may occur along the way.
“When you are coming to production, we need to not just print successfully, we need to be in required tolerances by customers. So then, iterations come in place,” Volcheck said. “If you don’t have a software that allows you to predict your deviations and account on that during preparation, then you’re iterating and iterating again, coming to a cost.”
After adopting Oqton’s Manufacturing OS at its central manufacturing facilities in Houston, Texas, energy technology firm Baker Hughes is said to have realised a 98% reduction in active monitoring engineering time and saved 136 engineering hours per printer annually. Crucially, it also reported an 18% reduction in costs associated with scrap due to real-time actionable alerts during part production.
“This saving is actually happening just because of data,” Volchek said, “because of data being collected and data being analyzed and data being labeled and used in trainings.”
On an episode of TCT’s Additive Insight podcast earlier this year, Andreas Bastian, co-founder and Head of Product at Lumafield, a Californian company working in industrial CT and AI inspection, described the cost of checking an AM part as “astronomical.” With high machine install and running costs, building a solid economic business case for AM can already be a battle, and high scrap rates and costs of inspection makes the case even harder, particularly as manufacturer’s wrestle with fitting AM into their usual KPIs and established processes.
“The use of these inspection technologies, in my perspective, is becoming emphasised because it’s needed,” Mostow said. “We can’t have as a technology this high of scrap rate because it’s not economical.”
Menzies adds that long production runs of some laser powder bed fusion parts compared to traditional processes, with some builds running for several days, also creates a greater need for productivity and early detection of failures, particularly in circumstances where the profitability of production is thin.
“Lost time and materials in these circumstances have a large impact on financial performance, and any new methods to mitigate this are of clear interest to manufacturers,” Menzies explained. “With increasing adoption of PBF-LB/M, more low profitability production is being opened up and at the other end of the spectrum higher quality requirement production is also increasing.”
This article originally appeared inside TCT Europe Edition Vol. 32 Issue 6. Subscribe here to receive your FREE print copy of TCT Magazine, delivered to your door six times a year.
