| During the manufacturing of the wing skin, the inspection steps are essential to ensure their conformity and thus allow the wings to ensure the required aerodynamic performances. Nowadays, considering the panel's low stiffness which prevents traditional inspection methods, this inspection is done manually with a template gauge and a jig. Iteratively, as long as form compliance is not reached, the panel goes through an additional dimensional refinement before being inspected in a second time. Because the jig is accurate, it is very expensive and furthermore, the inspection of panels is time-consuming by monopolizing the jig, which cannot be used in the meantime.;Using this consideration as a starting point, this project seeks to provide a response to the practicability of a methodology based on the automation of that king of operation. This by integrating into the process non-contact measuring machines capable of acquiring numerically the geometrical shape of the panel. Moreover, the opportunity of realizing this operation without the use of a jig is also being considered, which would leave it free for other tasks. The methodology suggested use numerical simulations to check form compliance. Finally, this would provide a tool to assist the operator by allowing a semi automated inspection without jig.;The methodology suggested can be describe in three steps, however it is necessary to propose an additional step to validate the results achieved with this methodology. Then, the first step consist of manually acquiring reference values which will served to be compared with the values obtained during the application of the methodology. The second step deals with the numerical acquisition, with a laser scanner, of the object to be inspected settled down on some supporting plate. The third step is the numerical reconstruction of this object with a computer-aided design software. Finally the last step consists of a numerical inspection of the object to predict the form compliance. Considering the large dimensions of the wing skins and of the jigs used in industry, the methodology suggested takes accounts of the available means in laboratory. Then, the objects used have lower dimensions than those used in the industry. That is the reason why a simplifying assumption that the shot peening operation has a negligible effect on the evolution of the thickness of the wing skin is made. Furthermore, the non-contact measurement device is also tested to know its accuracy under real conditions. Those two preliminary studies show that the thickness variation of a plate after being shot peened, with extreme parameters in terms of effects, remains negligible for the study of practicability realized in this thesis. The study on the performance of the REVscan 3D also brings to light that this variation would probably be drown in the uncertainty acquired by the device during the numerical acquisition.;In this project, only the steps two and three are dealt with in depth. This study involves essentially to test the measuring device and the software about their capacity of numerically acquiring an object and then to bring it to another state of stresses with the help of a simulation. Indeed, the validation of the free state step is problematic because it is precisely a state that cannot be obtained in an experimental way. As an analogy, it is suggested to pass from a particular state of stress to another because, in a simplified way, the free state step is equivalent to a change of a state of stress.;The study of the result allows to put forward a particular phenomenon linked to thin plates : it is a sudden change of the form when the plate is in a particular state of stress. The software is then no more able to predict that kind of comportment. Several tests are carried out to confirm the existence of that phenomenon and show that the stress modulus, the point of application of the stresses and the position of the support points are the more influent parameters. However, even by ensuring to avoid this phenomenon during the tests, the degree of accuracy reached by the software is far from being sufficient. Indeed, the uncertainty of the results is still too high and the next studies will have to focus on improving the results.;Currently, the tests realized in this thesis are not enough to validate the steps 2 and 3 of the methodology suggested. Nevertheless, the phenomenon highlighted which can suddenly modify the comportment of thin plates and the information gathered in these tests establishes a base for further research. (Abstract shortened by UMI.). |
