Droplet size and morphology characterization for diesel sprays under atmospheric operating conditions

The shape of microscopic fuel droplets may differ from the perfect sphere, affecting their external surface area and thus the heat transfer with the surrounding gas. Hence there is a need for the characterization of droplet shapes, and the estimation of external surface area, in order to enable the development of physically accurate mathematical models for the heating and evaporation of diesel fuel sprays. We present ongoing work to automat-ically identify and reconstruct the morphology of fuel droplets, primarily focusing in this study on irregularly-shaped, partially-deformed and oscillating droplets under atmospheric conditions. We used direct imaging tech-niques based on long-working distance microscopy and ultra-high-speed video to conduct a detailed temporal investigation of droplet morphology. We applied purpose-built algorithms to extract droplet size, velocity, vol-ume and external surface area from the microscopic ultra-high-speed video frames. High resolution images of oscillating droplets and a formation of a droplet form ligament, sphericity factors, volume as well as external surface area are presented for 500 bar injection pressure in the near nozzle region (up to 0.7 mm from nozzle exit) under atmospheric conditions. We observed a range of different liquid structures, including perfectly spher-ical, non-spherical droplets and stretched ligaments. We found that large droplets and ligaments exceeding the size of the nozzle hole could be found at the end of injection. In order to estimate droplet volume and external surface area from two-dimensional droplet information, a discrete revolution of the droplet silhouette about its major centroidal axis was used. Special attention was paid to the estimation of actual errors in the prediction of volume and surface characteristics from a droplet silhouette. In addition to the estimation of droplet volume and external surface area, the actual shape reconstruction in 3D coordinates from a droplet silhouette was performed in order to enable future numerical modelling studies of real droplets.