Fast level-set-based boundary reconstruction of highly complex body configurations for studying bio-inspired locomotion using a sharp-interface immersed boundary method

The 3-D finite-difference sharp-interface immersed boundary method (Mittal et al., JCP 2008) has been widely applied to various bio-inspired flow simulations. Immersed boundary reconstruction is the first step before the flow solution. Its efficiency has strong impact on the simulations with moving bodies, in which the reconstruction needs to be done in each time step. In this talk, we present a new fast level-set-based reconstruction method for flow simulation of highly complex bodies. The level set values are calculated in the direct neighbor grids around the solid boundary as a signed distance function, and then propagated to other grids through the entire computational domain in a highly-efficient way. Then the interior/exterior grids are identified by the sign of level set value. After this the flow field can be solved with the sharp-interface immersed boundary method. Algorithm analysis shows the new method has smaller computational complexity than the direct searching method and the ray-tracing method. Applications to swimming dolphin, manta ray and trout with dense meshes show the new method not only significantly saves boundary reconstruction time, but also improves the error-tolerance of reconstruction results.