Comparing CAD Models with 3D Scanned Manufactured Parts on the Cloud

Checking the quality and accuracy of a manufactured part against the as-designed status of the corresponding nominal CAD model is a common process in industry. For this purpose, the machined part is scanned with a 3D acquisition device, e.g. a laser scanner, which results in several point clouds typically consisting of millions of measured, discrete 3D points. To determine where deviations between the CAD model and the point cloud exist, the two data sets need to be aligned and matched in a process called registration and distances between the closest points of both models need to be calculated. The end user Stellba is a Germany-based SME operating in the field of hydropower plant maintenance, repair and overhaul. For the green energy sector they are engineering and manufacturing one-of-a-kind products with the goal to optimize energy efficiency. For the manually quite tedious quality checks Stellba so far uses a multitude of different software packages, implying high license costs and high training efforts to be able to handle all the different user interfaces. This created the demand to reduce the amount of time needed (currently 8 hours) by at least a factor of 5 by developing a dedicated solution exploiting HPC resources for fast, accurate and optimized matching of point clouds versus nominal CAD models improving both accuracy and usability. With the help of SINTEF (a research institute in Norway) and Jotne AS (a Norwegian SME in product lifecycle management), the processing time of quality checks at Stellba is being reduced to less than 20 minutes saving more than 7 person-hours of work. The ICT supported manual process containing error prone steps is being replaced by a validated software application that bridges domain barriers and enhances the achieved quality of the inspection. The operator at Stellba can now focus on the quality of the measurement and the produced parts while leaving most of the data processing to the new software application. In this particular case, shortening the elapsed time for point comparison quality checks increases Stellba's capacity for taking on new projects. Since the addressed topic of accuracy checking is of high relevance for many manufacturing branches, hundreds or even thousands of other usages per year of the developed Cloud services are regarded as likely. The corresponding additional revenue can put SINTEF into the position to hire 1-2 new researchers for porting even more functionality to the Cloud. Tackling the challenges of this experiment as a European endeavour has brought together partners from Norway (SINTEF, Jotne), Germany (Stellba) and Slovenia (Arctur as the HPC/Cloud provider) to develop an effective and efficient software solution together with the CloudFlow Competence Center that no single organization would have been able to offer on its own.