Study of Gluings by Periodic Method : Optimal Design of Experiments

This paper presents the study of gluing thermal parameters by periodic methods. A classical gluing is featured by a thermal resistance in between two identical layers. A study of phaselag versus excitation gives an estimation of the gluing thermal parameters . A 1D model of simulation was the base for a ODE (Optimal Design of Experiments) global methodology. It is shown that front face configuration is well adapted for an NDC application. The optimal frequencies for the measurements of thermal resistance are calculated and consequences of errors on supposed to be known parameters are studied. INTRODUCTION Recently developed scientific and industrial applications use new materials such as multi layers coatings, thermal barriers and gluings in response to an increasing requirement of performance. It is then important to measure the thermal properties of these multi-layers materials in order to evaluate the performances of an application, to optimize the material build up or to predict its evolution versus time and wear. Periodic methods used by Gervaise and al [1] [2] [3] can be applied to the study of multi-layers. A modulated laser beam heats the sample on its front face and the field of temperature is measured by radiometry on either the front heated face or the rear face. The temperature has a continuous component plus a periodic one whose period is identical to the heating beam one. Its fundamental harmonic is defined by its magnitude and its phaselag. A great interest of the periodic method is that the disturbed field area can be limited through the choice of the frequency. Dimension of this area is the diffusion length δ. δ α π = f /1/ Phaselag records versus excitation frequency gives an estimation of multi layer thermal parameter : it is the identification by material transfer function. A global methodology of identification of properties per material transfer function is proposed. An experimental bench dedicated to the study of multi-layer at millimetre scale using the periodic methods is described. A 1D modeling of experiments is presented and is adapted for a three-dimensional one is given. From 1D model, matrix X’X and sensitivity coefficients are used to Optimal Design of Experiments (ODE). NOMENCLATURE α : diffusivity ms CTR : Contact Thermal Resistance δ : diffusion length m ε : vector of measurement noise f : frequency of excitation Hz Φ : heat flux Wm φ : thermal phaselag rad or ° h : thermal loss coefficient WmK λ : conductivity WmK M : thermal modulus °K NDC : non destructive control Rc : thermal resistance WKm