Araqev, a Purdue University-related company, has commercialized quality control software that could benefit aerospace, automotive, consumer products, medical devices, national defense and other sectors that use additive manufacturing technology.
Araqev’s quality control software enables end users to print products in just a few design iterations, reducing scrap and machining time, reducing frustration and improving the overall quality of final printed products. “We estimate that the quality control issue with additive manufacturing can lead to nearly $2 billion in global losses per year based on a model of production costs of metal additive manufacturing systems that was developed by Baumers, Dickens, Tuck, and Hague in their 2016 paper published in the peer-reviewed journal Technological Forecasting and Social Change,” said Arman Sabbaghi, associate professor in the Purdue Department of Statistics at the College of Science, and CEO and President of Araqev.
To use Araqev’s software, customers upload their nominal design files and scanned point cloud data from their printed products.
“Our software uses these inputs to fit machine learning models that can simulate shape deviations for future printed products,” Sabbaghi said. “In addition, machine learning models allow our software to derive changes to nominal designs, called compensation plans, so that when the changed designs are printed, they have fewer shape deviations compared to when original designs are printed.”
Araqev’s algorithms also enable the transfer of knowledge, coded via machine learning models, across different materials, printers and shapes in additive manufacturing systems.
“This means that our software provides a complete platform for a customer to improve the quality of their entire system,” Sabbaghi said.
“The power and cost-effectiveness of our algorithms was recently demonstrated through two validation experiments for the Markforged Metal X 3D printer involving 17-4 PH stainless steel products,” Sabbaghi said. “Our algorithms reduced shape inaccuracies by 30% to 60%, depending on geometry in at most two iterations, with three training shapes and one or two test shapes for a specific geometry involved in the iterations.”
Araqev establishes direct partnerships with manufacturers and 3D printing companies. This, Sabbahi said, will allow the business to grow rapidly.
“We will establish license agreements after demonstrating to companies the cost savings and benefits we can provide for their processes,” Sabbaghi said. “These partners will integrate our software into their systems and sell it to their customers, providing us with an important customer channel.”