http://arxiv.org/abs/1510.01214
We determine how MRI-turbulent stresses depend on gas pressure via a suite of unstratified shearing box simulations. Earlier numerical work reported only a very weak dependence at best, results that call into question the canonical alpha-disk model and the thermal stability results that follow from it. Our simulations, in contrast, exhibit a stronger relationship, and show that previous work was box-size limited: turbulent `eddies’ were artificially restricted by the numerical domain rather than by the scale height. Zero-net-flux runs without physical diffusion coefficients yield a stress proportional to $P^{0.5}$, where P is pressure. The stresses are also proportional to the grid length and hence remain numerically unconverged. The same runs with physical diffusivities, however, give a result closer to an alpha-disk: the stress is proportional to $P^{0.9}$. Net-flux simulations without explicit diffusion exhibit stresses proportional to $P^{0.5}$, but stronger imposed fields weaken this correlation. In summary, compressibility is important for the saturation of the MRI, but the exact stress-pressure relationship is difficult to ascertain in local simulations because of numerical convergence issues and the influence of any imposed flux. As a consequence, the interpretation of thermal stability behaviour in local simulations is a problematic enterprise.
J. Ross, H. Latter and J. Guilet
Tue, 6 Oct 15
52/78
Comments: Accepted for publication in MNRAS
You must be logged in to post a comment.