Computational fluid dynamics simulations incorporating supersonic turbulent gas flow models and a droplet breakup model are performed to study supersonic gas atomization for producing micron-sized metal powder particles. Generally such atomization occurs in two stages: a primary breakup and a secondary breakup. Since the final droplet size is primarily determined by the secondary breakup, parent droplets of certain sizes (1 to 5 mm) typically resulting from the primary breakup are released at the corner of the nozzle and undergo the secondary breakup. A comparison of flow patterns with and without the introduction of a liquid melt clearly indicates that the mass loading effect is quite significant as a result of the gas-droplet interactions. The flow pattern change reasonably explains why the final droplets have a bimodal mass size distribution. The transient size changes of the droplets are well described by the behavior of the Weber number. The present results based on the 1 mm parent droplets best fit previous experimental results. Moreover, the effects of inlet gas pressure and temperature are investigated in an attempt to further reduce droplet size.

No abstract

No abstract

No abstract

This study presents a system using an image processing technique that evaluates the pavement condition from an image. Pavement condition evaluation is an integral part of roads and highway maintenance works, which mostly depends on human inspection. Although recently some researches have been conducted on road condition detection with image processing, these researches used huge databases and deep CNNs that require expansive computer and longer training time, which limits the use of deep CNN in practical problems where huge database collection is not possible always. To solve this problem, in this study, transfer learning in deep CNN is applied and with only 195 images in each category, pre-trained VGG-16 and Inception-ResNet v2 models are used for pavement condition evaluation. VGG-16 achieved more than 90% prediction accuracy, while Inception-ResNet v2 achieved more than 85% prediction accuracy. Moreover, to validate the performance, both models have been tested with random images collected from Google. Evaluating pavement conditions this way would reduce the need for human inspection. Finally, the outcome of the study shows that the transfer learning approach could be useful in research areas, especially in civil engineering, where image data is insufficient.