Monday  |  Carnegie III  |  02:30 PM–02:50 PM

#13764, 3D Thermal Imaging for Thermo-Elastic Stress Analysis (TSA) and Numerical Model Validation

Finite Element Analysis (FEA) offers a convenient means to analyse structures and components, however if confidence is to be put in such models it is imperative that they are appropriately experimentally validated. Traditional experimental campaigns make use of sensors, e.g. strain gauges to provide data for the validation of FEA. These sensors typically provide only point measurements, and judicious placements of strain gauges is crucial if useful validation data is to be obtained. Moreover, such localised measurements do not necessarily validate numerical models since they provide no information pertaining to the rest of the component. In contrast, TSA is a full field measurement technique based on thermal imaging providing thousands of temperature measurements across a field of view which can be used to infer the distribution of stress in a component. TSA provides spatially and temporally rich data with significant potential for model validation whereby thousands of points can be used to validate the FE model. However, thermal cameras capture data on a 2D sensor array, whereas FE models are constructed to capture the 3D nature of components and structures. Recent research has focussed on planar specimens or surfaces, where 2D comparisons can be readily made even with 3D models. Yet real structures and components are rarely 2D, and often exhibit significant curvatures, e.g. leading edges of aeroplane wings. Capturing validation data for such surfaces is currently achieved using a single camera which is challenging since perspective effects can skew and distort the view of the surface. The current work utilises aspects of machine vision to capture 3D thermal data using two cameras which are calibrated in order to perceive depth. The aim of the proposed work is to develop a novel method of projecting TSA data to form a 3D point cloud which can then be compared in a like-for-like basis with complex 3D models. The paper shows work undertaken to design a suitable experimental setup for stereo vision in IR, including the design of a calibration plate. Results of validation work carried out on a cylindrical specimens are then presented showing a point cloud capturing 160 degree arc of the cylinder. Preliminary work describing a 360 degree view of a 3D object is then described followed by a demonstration of the technique C-shapped spar specimen.

Geir Olafsson   University of Bristol
Toby Jack   University of Bristol
Janice Dulieu-Barton   University of Bristol