AIRBUS A350XWB Ground Vibration Testing: Efficient techniques for customer oriented on-site modal identification

Ground vibration testing (GVT) of aircraft structures belongs to the most challenging task under the application of experimental modal analysis methods. Besides the size of an aircraft and therefore the required number of sensors, exciters and measurement units, for high demanding industrial programs it is nowadays especially the customer request to conduct the modal test in a very short time period that puts high pressure on the GVT team. This paper describes how state-of-the-art modal analysis techniques are utilised to perform a time efficient GVT campaign using optimised excitation techniques, intelligent signal processing and on-site modal analysis to achieve “real time” high quality experimental modal data. The paper is presented in the context of the ground vibration test on AIRBUS A350XWB MSN1 which was performed shortly before the first flight. The test has been conducted by a transnational GVT team of people from the German Aerospace Center and the Office National d'Études et de Recherches Aérospatiales (DLR-ONERA) in the AIRBUS facilities in Toulouse, France, within only 9 measurement days. During the test an LMS Scadas III data acquisition system was used with a distributed architecture of the acquisition units connected by 300 m of optical fibre cables to minimise sensor cable lengths. Overall a number of 530 acceleration signals, 27 force signals, and 33 other signals have been recorded. The structure was excited from 23 locations by mean of 13 electro-dynamic shakers to acquire more than 180 excitation runs mainly with optimised swept-sine signals and occasionally with random signals. For further information, the phase resonance method (PRM) was also applied for few specific modes. The DLR-ONERA processes for ground vibration testing are strongly focusing on phase separation methods (PSM) and are perfectly harmonised for fast and reliable output. On-site modal identification has been performed directly after data acquisition. The identified modal data (overall more than 2600 modes) have been efficiently stored in a novel SQL-database system. Specialised DLR-ONERA software tools for modal model correlation and quality assessment yield the final modal model and indicate non-linear effects on dedicated modes