Quantitative Full-Field Data Fusion for Evaluation of Complex Structures

BackgroundValidation of models using full-field experimental techniques traditionally rely on local data comparisons. At present, typically selected data fields are used such as local maxima or selected line plots. Here a new approach is proposed called full-field data fusion (FFDF) that utilises the entire image, ensuring the fidelity of the techniques are fully exploited. FFDF has the potential to provide a direct means of assessing design modifications and material choices.ObjectiveA FFDF methodology is defined that has the ability to combine data from a variety of experimental and numerical sources to enable quantitative comparisons and validations as well as create new parameters to assess material and structural performance. A section of a wind turbine blade (WTB) substructure of complex composite construction is used as a demonstrator for the methodology.MethodsThe experimental data are obtained using the full-field experimental techniques of Digital Image Correlation (DIC) and Thermoelastic Stress Analysis (TSA), which are then fused with each other, and with predictions made using Finite Element Analysis (FEA). In addition, the FFDF method enables a new high-fidelity validation technique for FEA utilising a precise full-field point by point similarity assessment with the experimental data, based on the fused data sets and metrics.ResultsIt is shown that inaccuracies introduced because of estimation of comparable locations in the data sets are eliminated, The FFDF also enables inaccuracies in the experimental data to be mutually assessed at the same scale regardless of differences in camera sensors. For example, the effect of processing parameters in DIC such as subset size and strain window can be assessed through similarity assessment with the TSA.ConclusionsThe FFDF methodology offers a means for comparing different design configurations and material choices for complex composite substructures, as well as quantitative validation of numerical models, which may ultimately reduce dependence on expensive and time-consuming full-scale tests.