Simulations of convection, phase change and solute fluxes in a porous mushy layer
ORAL
Abstract
Sea ice is a reactive porous material composed of ice crystals and an interstitial saline fluid (or 'brine'); a so-called mushy layer. Dense brine tends to sink through the ice, driving convection. Downwelling at the edge of convective cells leads to the development of narrow, entirely liquid channels, through which saline brine is efficiently rejected into the underlying ocean.
We consider numerical simulations of ice growth and convective brine transport. This natural convection during sea ice formation leads to patterns of varying porosity and fluid flow across multiple scales, with brine channels occupying just a few percent of the sea ice volume and evolving in time. To resolve the narrow brine channels at acceptable computational cost, we use the Chombo software framework to implement an adaptive computational mesh. We model mushy sea ice growth in a limit where the liquid ocean has salinity far less than the eutectic composition.
Using these techniques, we calculate the ice-ocean salt flux during steady-state growth and find a sublinear dependence of the flux on the porous medium Rayleigh number. Transient simulations reveal complex behavior; neighboring brine channels interact with each other, whilst their spatial density tends to decrease over time.
We consider numerical simulations of ice growth and convective brine transport. This natural convection during sea ice formation leads to patterns of varying porosity and fluid flow across multiple scales, with brine channels occupying just a few percent of the sea ice volume and evolving in time. To resolve the narrow brine channels at acceptable computational cost, we use the Chombo software framework to implement an adaptive computational mesh. We model mushy sea ice growth in a limit where the liquid ocean has salinity far less than the eutectic composition.
Using these techniques, we calculate the ice-ocean salt flux during steady-state growth and find a sublinear dependence of the flux on the porous medium Rayleigh number. Transient simulations reveal complex behavior; neighboring brine channels interact with each other, whilst their spatial density tends to decrease over time.
–
Presenters
-
James RG Parkinson
University of Oxford
Authors
-
James RG Parkinson
University of Oxford
-
Dan Martin
Lawrence Berkeley National Laboratory
-
Rich F Katz
University of Oxford
-
Andrew J Wells
University of Oxford