Boundary Effects

Boundary noise can manifest when subsonic waves near the boundary cause errors that may propagate inward with speed less or equal to the sound speed. Thus to consider if such a disturbance has time enough to travel from the boundary to the BE sphere/ambient boundary, I put together this table that shows the sound speed of the ambient, the final time of the simulation, and the distance that a sonic wave would travel in this time. For all cases, except for the stability case, sonic disturbances from the boundary do not have time enough to propagate from the boundary to the sphere's surface. Note, I have taken the boundary to be the sphere of radius = 15, centered on the origin. The physical boundary conditions on the box are set to extrapolating with perpendicular velocities set to 0 using winds on all outward faces, and reflective on the inward faces. The Poisson boundary conditions are set to multipole on outward faces, and reflective on inward faces. Every spatial unit is in terms of Rbe, the Bonnor Ebert sphere's radius. So 1 computational unit on the domain = 1 Rbe.

Case	Sound Speed Ambient	Tsim	Distance sound wave from boundary can travel	Does flow at boundary become supersonic?
Matched	4.491	0.216	0.97 RBe ~ 1/15 Domain	yes
BP	42.82	0.192	8.2 RBe ~ ½ Domain	no
Stability	44.91	1.1	44 RBe ~ 3*Domain	no
1/3 Ambient	8	0.28	2.25 RBe	yes
1/10 Ambient	14.2	0.4	5.6 RBe	no
1/30 Ambient	4.491	0.216	0.97 RBe	yes

Below are 2D slices of pseudocolor plots showing rho and vrad/cs.

Matched

As you can see in the left plot of rho, the density in the ambient is increasing as the simulation progresses, going from darkest blue to lighter shades. This is coincident with speeds in the ambient increasing, likely due to the gravitational acceleration of what resembles a uniform sphere of rho=rho_ambient. I believe new material is not supplied at the boundary, and that any winds coming in from the boundary or going out is just set to 0 there. Along the fictitious bounding sphere of R=15Rbe, material is supplied from the external domain. Speeds however, seem to have a sharp gradient, possibly due to the elliptic solver.

BP

Ambient 1/3

Ambient 1/10

Ambient 1/30