Drapes in a room with a fan

Drapes in a room with a fan pointed at them. I took a video of this in Canada. Build a physics engine that can replicate this motion based on _laminar_ CFD calculations, joints for the drapes (3DOF), and collision physics. _Try_ Viscous; might need it but it seems to me that the drapes' motion is a consequence of acoustic disturbances, and the energy driving the wave is transferred when the drapes have collisions with one another or when the acoustic waves interact with the individual drape surfaces. Also, collisions occur, but once the wave is going and relatively steady, are they necessary to maintain the wave motion of the drapes? Which energy transfer mechanism (acoustics or collisions) is most important at each stage of the simulation and how do their orders of magnitude change relative to one another. If a simplification can be made of the model because some terms drop during OOM analysis, the code can run faster.