Abstract

Additive Manufacturing (AM) of metal parts is becoming increasingly common as a flexible and scalable option for production. However, AM does have its challenges, notably in terms of the quality of the finished product, including the presence of defects that can increase the risk of failure. One key challenge involves comparing as-designed and as-built AM parts, and how these differences between the virtual and physical affect real-world performance. One set of solutions are being developed through a partnership between Synopsys, nTopology, ANSYS, North Star Imaging (NSI), and EOS. By combining industrial Computed Tomography (CT) with simulation, it is possible to virtually test and compare original CAD designs and scans of the actual printed parts. From this data, manufacturers can identify issues with the component (such as cracking or porosity) that could affect performance and lead to costly delays in production. In this presentation, we will discuss a recent case study for a workflow involving rapid re-design of a traditional heat exchanger, and how the combination of methods helps optimize design. We start with design of the part in nTopology software, analysis and simulation using ANSYS software, printing in EOS AM Machine, CT scanning of the part with NSI, image based inspection and image based meshing in Synopsys and finally simulation of the as-built part in ANSYS. The main focus of this presentation, however, will be on the Quality Analysis (QA) process post CT scanning. From the CT image data, inspection and meshing in Synopsys Simpleware software enables comparison of the performance of the original CAD design and the image-based model through simulation in ANSYS. Using this case study, we will demonstrate that the efficient workflow results in a design that is 80% lighter, 40% smaller in form-factor, and 10k more efficient in heat transfer than conventional technology. In addition, we will discuss how these techniques can be applied to similar design and manufacturing workflows and when applied in production, become fully automated by leveraging Machine Learning based AI technologies.