|Sammanfattning: ||This dissertation focuses on correlation between measurement and simulation results on a Body in White (BIW). The purpose is to set up and evaluate a correlation methodology, where each sub-assembly of the car body is assessed separately to be able to localize the finite element model uncertainties. The sub-assemblies correlated in this dissertation are the Floor Rear and the Body Side Outer, of a BIW architecture. To obtain the dynamic properties of the car body, several experimental and CAE based methods are employed and evaluated.
The sub-assemblies of a car body are usually connected using different joining methods, such as spot welds and adhesives. The first step is to set up two FE-models of the sub-systems, from CAD-geometries and a complete FE-model of the BIW. This is done by using ANSA as a pre-processor. Normal modes analysis (NMA) and modal frequency response analysis (MFRA) can then be utilized in MSC NASTRAN to calculate mode shapes, resonance frequencies and transfer functions. These numerical results are then used in a pre-test analysis to decide accelerometer and impact locations for the experimental modal analysis (EMA). Thus, qualitative measurement data can be obtained, capturing the dynamical behavior of the structure. Measurements are then done using an impact hammer, triaxial accelerometers, and measurement software and hardware from Müller-BBM VibroAkustik Systeme (PAK), as well as from Siemens (LMS Testlab).
The correlation of the sub-assemblies focuses on physical differences between the FE-model and the manufactured parts. Uncertainties identified are the adhesives, welds, structural damping, and the thickness distributions. To quantify differences between measurement data and simulation data, and to study the sensitivity of different properties, transfer functions are analyzed. Global modes and resonance frequencies are either identified from real and imaginary parts of the transfer functions, or from experimental and analytical modal analysis.
The first conclusion drawn was that when correlating sub-assemblies, the FE-model must be adapted to the current step in the manufacturing process. One example is the adhesives that are cured at a later stage in the factory, and thus the ultimate material properties cannot be applied directly using the methodology presented in this dissertation. Furthermore, conclusions on the influence of the welds, structural damping, the thickness distribution and the mass is presented in this dissertation.
By correlating the car body piecewise, the sensitivities and uncertainties were localized, and could thus be analyzed in more detail. Furthermore, once the complete body is assembled the only uncertainties lies in the connections between the sub-assemblies, and not in the different sub-systems themselves. The presented methodology can thus be considered more robust than by correlating the complete body directly.|