Please note: This master’s thesis presentation will take place in DC 2314.
Clara Kim, Master’s candidate
David R. Cheriton School of Computer Science
Supervisor: Professor Christopher Batty
Liquid simulations typically involve solving the fluid equations over the simulated liquid domain via some volumetric discretization of the domain. Consequently, the majority of schemes which aim to simulate the interaction between liquids and freely moving solids are built assuming a volumetrically discretized liquid model. However, storing and manipulating a surface mesh rather than a volumetric discretization, on top of allowing us to avoid storing interior volume data, has the potential to reduce the number of unknowns in the systems necessary to resolve fluid flow. Motivated by this potential, we present a method for simulating the two-way coupled interactions between solid rigid bodies and an inviscid liquid, where the liquid domain is represented and simulated entirely by its surface. Our work builds off of the surface-only liquids method first proposed by Da et al. [2016]. We are concerned with the 2D version of the solid-liquid coupling problem. As such, the liquid surface is represented as a series of point vertices connected by line segment edges, with the velocity data stored at the vertices.
The surface-only liquid simulation method integrates outside forces, such as forces caused by scripted solids in contact with the liquid, by performing a boundary element method (BEM) solve for fluid pressures using surface tension and solid velocities to set boundary conditions. We perform liquid-solid coupling in a single unified solve by modifying this force integration step to account for solids with dynamic velocities by modeling the momentum exchange that occurs at the liquid-solid interface. We show several examples demonstrating our method's ability to handle liquid-rigid body dynamics, as well as validate our method against analytical solutions derived using the fluid mechanics concept of added mass. We also demonstrate our method's ability to support multiple solid rigid bodies interacting with each other through the liquid domain without need for direct contact between the solids. We hope that our work encourages further investigations into the surface-only liquids framework in the future, allowing for the simulation of an even wider range of interesting liquid phenomena using only a surface discretization of the simulated domain.