Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Modal identification is one of the key steps in the process for ensuring structural integrity of new products. Although this process is typically carried out manually, both academic research centres and industries would benefit tremendously from an automated identification procedure. This master thesis investigates the opportunity for an automated impact testing process for the V2i company. pLSCF and pLSCE algorithms are first revised to allow for an automated poles detection. An automated experimental process is then set up using a robot-controlled laser Doppler vibrometer and an automated impact hammer. The set-up and revised algorithms are validated on an academic mono-bloc bladed-disk. L'identification modale est une étape cruciale du design de tous nouveaux produits afin de garantir leurs intégrités structurelles. Bien que cette identification se fasse généralement manuellement, les industries et centres de recherches auraient beaucoup à gagner d'un procédé automatisé. Ce travail a pour but de développer un procédé d'analyse modal à marteau d'impact pour l'entreprise V2i. Les algorithmes pLSCE et pLSCF sont tout d'abord revisités pour permettre une détection automatique des pôles du système. Un setup expérimental est ensuite mis en place à l'aide d'un vibromètre laser et d'un marteau automatique. Le setup et les algorithmes revisités sont validés sur un prototype académique d'une roue aubagée monobloc.

Impact testing --- pLSCF --- pLSCE --- Laser Doppler Vibrometer --- Experimental modal analysis --- Automated modal analysis --- Ingénierie, informatique & technologie > Ingénierie aérospatiale

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

In the present master's thesis, a robot-assisted modal analysis method is developed for the company V2i, specialized in vibration identification. The main goal is to position a laser Doppler vibrometer (LDV) on an industrial robot in order to, on the one hand, gain time in the point-to-point movement of the LDV and eventually automating the whole process, and on the other hand, get the ability to perform precise 3D measurements. Indeed, Euler angles are tricky to set manually with a satisfactory degree of accuracy.

A brief literature review is done on the different calibration methods and the existence of this type of application. After installing the robot, the risk assessment due to its presence in the company along with the needed safety installations are also established.

The main part of the work treats the specific subject of tool and part calibration. Indeed, when fixing a LDV on a robot head, the problem is that the beam has no physical end. Thus, the calibration phase needed to define the positions and orientations of the laser beam and the studied part with respect to the robot frame becomes quite different from a physical tool calibration. In this work, several solutions for these two calibrations are proposed and tested. A selection is then made. The retained method consists in combining, on the one hand, a common method for physical tool calibration based on physical contact, and on the other hand, the LDV and part calibration procedure. To this end, a plastic 3D printed tip is attached to the robot end effector with the LDV. When placing the tip in such a way that the laser beam touches its extremity, calibrating the LDV boils down to calibrating a simple tip. Similarly, defining the position of the part with respect to the robot boils down to touching it in several locations with the tip and computing the transformation matrix between the robot frame and the part frame. A validation of this method in terms of repeatability and accuracy as a function of the number of palpated points is performed and gives satisfactory results. To the author's knowledge, no other article treats the calibration problem of a LDV on a robot head with such a both simple and efficient method.

Afterwards, modal analyses using both mono- (1D) and multi-directional (3D) measurements are conducted on an academic compressor wheel. The interest is double: firstly, one is able to show that the calibration precision is sufficient to perform an accurate modal analysis; secondly, one is able to show the advantage of using the robot to perform 3D measurements by comparing with the 1D results. Finally, a discussion on the perspectives of automation and inherited time gain is carried out.

LDV --- Doppler vibrometry --- Hand-eye calibration --- Hand-LDV calibration --- Non-physical tool calibration --- Modal analysis --- Robot-aided --- Robot-assisted --- Ingénierie, informatique & technologie > Ingénierie mécanique

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

The intrinsic structure of composites can lead to defects decreasing their reliability and their
in-mission security. Numerical simulations aiming to improve and support defect detection are envisaged in this work. More precisely, shearography and thermography as non-destructive detection methods are being modeled here. Two defect types are considered: delamination and porosity.

Detection is investigated with external thermal excitation, leading to thermal and mechanical
analysis in simulations. Firstly, an overview of composites, their defects and non-destructive techniques is addressed. Secondly, prerequisites for simulations like governing equations and assumptions made, the heater characterization, and the numerical scheme used for the transient thermal problem resolution are exposed. Then, defect numerical models are constructed and studied. Delamination and porosity are the two types of defects considered. Numerical models for the delamination covered true delamination and artificial delamination like the physic insert and flat bottom hole models. The porosity model is represented by a few flat bottom holes localized in a small region. Finally, an experimental approach compared with numerical results is used as a validation method.

Different delamination models are developed and they show pretty well concordances between
them, except for the Teflon layer (type of physic insert) model for which the mechanical response
was not expected or at least, suggests a further study to determine its validity. The porosity model showed difficulties in this kind of defect detection. Finally, the experimental approach enabled to see that numerical and experimental results were similar but that some efforts on parameter updating remain to be made. Mainly the characterization of the lamp that irradiates a highly non-homogeneous flux.

Numerical Simulations --- Shearography --- Thermography --- Delamination models --- Porosity models --- Experimental validation --- Ingénierie, informatique & technologie > Ingénierie aérospatiale