← All problems

Five Open Problems — Compressible Navier–Stokes near vacuum

Consider the isentropic or isothermal compressible Navier–Stokes equations in several space dimensions, with density ρ\rho, velocity vv, pressure p(ρ)p(\rho), and constant viscosity coefficients μ\mu and ζ\zeta:

∂tρ+div⁡(ρv)=0,∂t(ρv)+div⁡(ρv⊗v)+∇p(ρ)=div⁡T,\partial_t\rho+\operatorname{div}(\rho v)=0,\qquad \partial_t(\rho v)+\operatorname{div}(\rho v\otimes v)+\nabla p(\rho)=\operatorname{div}T,

where

T=μ(∇v+(∇v)T)+(ζ−2μ)(div⁡v)I.T=\mu(\nabla v+(\nabla v)^T)+(\zeta-2\mu)(\operatorname{div}v)I.

In a region where ρ=0\rho=0, the momentum equation leaves the velocity subject to the elliptic equation

μΔv+(ζ−μ)∇div⁡v=0.\mu\Delta v+(\zeta-\mu)\nabla\operatorname{div}v=0.

The question concerns initial data whose limiting density consists of two gas droplets separated by vacuum, and approximating data in which a small density ε>0\varepsilon>0 is added between them. Let Gℓ,GrG_\ell,G_r denote the droplets’ centers of mass and vℓ,vrv_\ell,v_r their velocities in the vacuum-limit evolution. Serre considers the configuration

Gℓ−Gr+τ(vℓ−vr)=0for some τ>0,G_\ell-G_r+\tau(v_\ell-v_r)=0\quad\text{for some }\tau>0,

for which the droplets must merge at a finite time T∗T_*. Does the distance r(ε,t)r(\varepsilon,t) between the droplets in the perturbed evolution remain uniformly positive, or can their merger be delayed in the sense that, for some T′>T∗T'>T_*,

lim inf⁡ε→0+r(ε,T′)>0?\liminf_{\varepsilon\to0^+}r(\varepsilon,T')>0?

This is the multidimensional counterpart question posed for the two-droplet scenario; the distance and the admissible solution class are not specified more narrowly in the source. Serre (2012)

Background

The issue is that constant viscosities do not vanish with the density: even inside vacuum, the velocity must satisfy a nontrivial elliptic equation. Serre describes this residual, or “ghost,” velocity as potentially affecting the boundary of a vacuum region as though the vacuum exerted its own pressure. The concern is not merely formal: such a velocity field can couple gas regions that would evolve independently if separated by vacuum. Serre (2012)

The constant-coefficient assumption is mathematically convenient but, as Serre notes, is not physically realistic near vacuum. A related motivation for considering density-dependent viscosities is that kinetic derivations through Chapman–Enskog expansions give temperature-dependent viscosity, which becomes density-dependent for isentropic flow. The multidimensional question also has a specifically geometric difficulty: gas added between the droplets may escape in directions transverse to the line joining them. Serre (2012) Yang (2000)

References

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.