2012A&A...540A..77D

Cosmic rays (CRs) are transported out of the galaxy by diffusion and advection due to streaming along magnetic field lines and resonant scattering off self-excited MHD waves. Thus momentum is transferred to the plasma via the frozen-in waves as a mediator assisting the thermal pressure in driving a galactic wind. The bulk of the Galactic CRs (GCRs) are accelerated by shock waves generated in supernova remnants (SNRs), a significant fraction of which occur in OB associations on a timescale of several 10⁷ years. We examine the effect of changing boundary conditions at the base of the galactic wind due to sequential SN explosions on the outflow. Thus pressure waves will steepen into shock waves leading to in situ post-acceleration of GCRs. We performed hydrodynamical simulations of galactic winds in flux tube geometry appropriate for disk galaxies, describing the CR diffusive-advective transport in a hydrodynamical fashion (by taking appropriate moments of the Fokker-Planck equation) along with the energy exchange with self-generated MHD waves. Our time-dependent CR hydrodynamic simulations confirm the existence of time asymptotic outflow solutions (for constant boundary conditions), which are in excellent the agreement with the steady state galactic wind solutions described by Breitschwerdt et al. (1991A&A...245...79B). It is also found that high-energy particles escaping from the Galaxy and having a power-law distribution in energy (∝ E^–2.7) similar to the Milky Way with an upper energy cut-off at ∼10¹⁵ eV are subjected to efficient and rapid post-SNR acceleration in the lower galactic halo up to energies of 10¹⁷-10¹⁸ eV by multiple shock waves propagating through the halo. The particles can gain energy within less than 3kpc from the galactic plane corresponding to flow times less than 5x10⁶years. Since particles are advected downstream of the shocks, i.e. towards the galactic disk, they should be easily observable, and their flux should be fairly isotropic. The mechanism described here offers a natural and elegant solution to explain the power-law distribution of CRs between the ``knee'' and the ``ankle''.