The BMW Group has introduced computer tomography (CT) into prototype development, production and analysis – a first in the automotive industry. Thanks to this technology, the full range of vehicles, from MINI to Rolls-Royce, can now be quality-controlled even in the early stages of development.
Scans are performed by four robots, which move around the outside of the prototype to produce several thousand cross-sectional images. These are then used for detailed examinations of innovations, new materials and bonding technologies. Until now, vehicles have had to be dismantled for analysis, but CT allows checks to be carried out with the vehicle completely intact.
The new X-ray system is based in the BMW Group Pilot Plant in the Research & Innovation Centre (FIZ) in Munich, at the intersection between Development and Production. BMW Group engineers and the Fraunhofer Development Centre for X-Ray Technology (EZRT) spent two years developing the new system, which is the only one of its kind in the automotive industry.
This degree of detail is required for a range of reasons, such as to check welds and punch-screw connections, and to verify body condition before and after painting, where extreme temperatures can affect adhesive bonds. Findings from the scan are then used as a basis for making targeted modifications to series production.
The scan itself is performed by four coordinated robots. Once the vehicle is in position in the system, the robots move around it. Working in pairs, they send X-rays through it and across to their counterparts. The data they collect is then put through a specially developed computer programme that calculates a multi-layered, three-dimensional image. This forms the basis for a detailed analysis of the internal workings of the vehicle, offering information on objects as small as 100 micro-metres – approximately the width of a human hair.
Engineers are currently carrying out research to establish how far artificial intelligence might be used to evaluate findings. By processing large amounts of data, the software can learn the many different patterns that occur, link individual items of data, and gradually evaluate findings automatically and accurately.