Abstract: The method for solid wall virtual particle boundary handling is a main boundary handling
method in fluid simulation, while it cannot ensure that no fluid penetrates the solid boundary, and
requires a huge amount of computation. To prevent fluid particles from penetrating the boundary, in
this paper a damp zone is set up near the boundary, and each fluid particle in this zone will be applied
by the boundary with an elastic force and a damping force that is opposite to the moving velocity of
the fluid particle. Therefore, the particles near the boundary become more stabilized. To reduce the
amount of computation, this paper proposes two adaptive sampling methods for boundary particles
based on different numbers of particles around the boundary: one is to adaptively sample particles of
different masses; the other is to adaptively sample particles of the same mass in different layers.
Compared with the traditional method for solid boundary particle sampling, these two adaptive
sampling methods reduce the number of boundary particles, accelerate the simulation speed and save
the computer memory. In addition, all of the 3D fluid simulations in this paper are based on the GPU
acceleration technique, and can interact in real time.

Key words: solid wall virtual particle, damping zone, elastic force, damping force, adaptive sampling
of boundary particles, GPU acceleration technique