EXPERIMENTAL VALIDATION OF THE MAXIMUM ENTROPY TECHNIQUE IN TOMOGRAPHIC RECONSTRUCTION TO DETERMINE LOCAL SPRAY DROP SIZE DISTRIBUTION
Résumé
This work presents the experimental validation of a new deconvolution technique that calculates local drop size distributions from line-of-sight laser diffraction measurements. Previous deconvolution techniques are based on the onion peeling method but errors cumulate at each inward step of the calculation and the calculation always reports unrealistic local negative information. These problems are avoided by using a different calculation method based on a maximum entropy formulation that determines the least biased solution at each local position. The experimental validation is implemented by Phase Doppler Particle Analyzer (PDPA). The sprays studied for the validation are sprays of water produced by a solid-cone pressure-swirl injector. These sprays are assumed axisymmetric. Laser diffraction measurements are carried out with a Malvern-Spraytec at several radial positions ranging from the center to the outskirts of the spray for several working conditions. Local drop size measurements are conducted with the PDPA. These two diagnostics report totally different information. Comparisons between laser diffraction and PDPA data are performed on both radially resolved and line-of-sight bases. Comparisons on the line-of-sight basis are firstly performed to indicate that both diagnostics measure the same ensemble of drops and if the local distributions from both diagnostics do not agree, it is due to deconvolution method. The new deconvolution technique is consequently validated by comparisons on the radially resolved basis. The results show a very good agreement in each working condition.