http://arxiv.org/abs/2303.18016
To simulate the heated exterior of a meteorite, we impact a granular bed with melted tin. The morphology of tin remnant and crater is found to be sensitive to the temperature and solidification of tin. By employing deep learning and convolutional neural network, we can quantify and map the complex impact patterns onto network systems based on feature maps and Grad-CAM results. This gives us unprecedented details on how the projectile deforms and interacts with the granules, which information can be used to trace the development of different remnant shapes. Furthermore, full dynamics of granular system is revealed by the use of Particle Image Velocimetry. Kinetic energy, temperature and diameter of the projectile are used to build phase diagrams for the morphology of both crater and tin remnant. In addition to successfully reproducing key features of simple and complex craters, we are able to detect a possible artifact when compiling crater data from field studies. The depth of craters from high-energy impacts in our work is found to be independent of their width. However, when mixing data from different energy, temperature and diameter of projectile, a bogus power-law relationship appears between them. Like other controlled laboratory researches, our conclusions have the potential to benefit the study of paint in industry and asteroid sampling missions on the surface of celestial bodies.
H. Huang, P. Tsai, C. Lu, et. al.
Mon, 3 Apr 23
44/53
Comments: 6 pages, 5 figures