The development of Laser-Induced Thermal Grating Spectroscopy (LITGS) for diagnostics of combusting and non-combusting flows is described. The first use of LITGS to provide in situ calibration of 2-Dimensional temperature distributions generated using Two-Colour Planar Laser-Induced Fluorescence (TC-PLIF) is reported. Time-resolved measurements of temperature distributions in a firing GDI optical engine obtained by TC-PLIF were made during the compression stroke and calibrated to the absolute temperature scale by simultaneous LITGS measurements. The accuracy and precision of the temperatures derived from LITGS data are evaluated using alternative methods of data analysis - Fast Fourier Transform and Fitting to theoretical models of the experimental data. The relative merits of the two methods are examined for analysis of weak LITGS signals obtained under engine conditions of low pressure and high temperature. The combined TC-PLIF and LITGS system was demonstrated by performing repeated single-shot measurements for 1 in every 10 four-stroke cycles showing excellent correlation of the temperatures derived from both techniques. Direct measurement of the effect of 'charge cooling', of order 5 K, for operation with direct injection is reported. Inhomogeneous temperature distributions were observed during the compression stroke for fired operation with Port Fuel Injection (PFI) and also with Gasoline Direct Injection (GDI). The effects of varying the relative concentrations of toluene and iso-octane in the two-component fuel were investigated. Extension of the LITGS technique to multi-point measurements along a 1-D line is described. By recording signals from 4 points on separate detectors using a fibre-coupled photodiode array the limitations of Streak Cameras used previously for 1-D LITGS measurements were overcome. Demonstration of principle experiments are reported in which simultaneous 4-point measurements were made with 1 mm spatial resolution and a precision of 0.7 % in temperature gradients in gas flows and in boundary layers at surfaces.

Cette thèse présente une étude expérimentale du problème de mélange dans les fluides complexes, étude menée en partenariat avec l’entreprise Sanofi Pasteur. Le mélange est un acte des plus fréquents dans la vie courante et aussi dans l’activité industrielle. On trouve dans la littérature de nombreuses études s’intéressant aux cuves de mélange pour en améliorer les performances à partir d’observations faites à grande échelle. Par contre, à notre connaissance, il y a peu de recherche sur l’hydrodynamique du mélange dans les fluides complexes. Dans notre travail, on étudie des fluides non-Newtoniens formés de solutions diluées de polymères caractérisés par leurs propriétés rhéofluidifiante et viscoélastique. Il s’agit de solutions aqueuses de Polyacrylamide (PAA) ou de la gomme de Xanthan (XG). Afin d’identifier la différence de comportement avec les fluides Newtoniens, une étude expérimentale avec de l’eau est effectuée dans les mêmes conditions que celles pour les fluides non-Newtoniens. Cette étude a été menée, en premier, sur un modèle réduit d’une cuve de mélange de Sanofi Pasteur. Les résultats obtenus, non représentés dans ce mémoire de thèse, nous ont amenés à mettre en place une étude fondamentale de l’écoulement dans un mélangeur de géométrie plus simple. Il s’agit alors de pouvoir contrôler les conditions initiales et de s’affranchir des effets secondaires de l’agitation pour ne s’intéresser qu’au mélange. Pour cela, la géométrie retenue est celle d’un mélangeur en T avec deux entrées perpendiculaires. L’exploration en 2D des champs de vitesse et de concentration de scalaire dans cette jonction en T est assurée simultanément aux moyens des techniques optiques (PIV et PLIF). Les observations montrent un effet non négligeable sur l’hydrodynamique et le mélange lié à la présence de polymères dans l’écoulement. De plus, les résultats obtenus permettent de calculer la tension de Reynolds uv et les flux de masse vc et uc. Ils seront utilisés par la suite pour vérifier leur conformité avec le modèle k epsilon couramment utilisé dans l’industrie. / This thesis presents an experimental study of the mixing in complex fluids which is conducted in partnership with Sanofi Pasteur. The mixture is one of the most common act in everyday life and also in industrial activities. We found in the literature many studies focusing on the mixing tanks with objective to improve performance based on observation of large scale. By cons, in our knowledge, there is few or no research on the hydrodynamics of a mixture in complex fluides. In our work, we study non-Newtonian fluids formed of diluted solution of polymer which characterized by their viscoelastic and shear thinning properties. We used in this study aqueous solutions of polyacrylamide (PAA) or xanthan gum (XG). To identify the difference in behavior with Newtonian fluid, an experimental study with water is carried out under the same conditions as those non-Newtonian fluids. At first, this study was on a reduced mixing tank of Sanofi Pasteur. The results, which not shown in this thesis, led us to develop a fundamental study of flow in a mixer with a simple geometry. The objective is to be able to control the initial conditions and to avoid the side effects of agitation to focus on the mixture. For this, we chose a mixer in a T shape with two perpendicular inputs. Exploring 2D velocity and scalar concentration fields in this T-junction is provided simultaneously of optical techniques (PIV and PLIF). Observations show a significant effect on the hydodynamic and mixture related to the presence of polymers in the flow. In addition, results are used to calculate the Reynolds stress uv and the scalar flux vc and uc. They will be used to check their compliance with the k epsilon model that commonly used in industry.