Solute redistribution during freezing of sands saturated with saline solution

Abstract

Columns of saturated saline sands were frozen under hydrostatic conditions with constant surface and base temperatures. Nine freezing tests were conducted using a silica sand with a permeability of about 10$\sp{-11}$ m$\sp2$ and salinities that ranged from 1 to 100 ppt but were generally near 35 ppt. Surface temperatures were generally 3 to 5 $\sp\circ$C colder than the freezing temperature of the solution and base temperatures were generally 0.5$\sp\circ$C warmer. A 4 to 10 day long period preceded the onset of convection and redistribution of the solute. The increased freezing rate due to the solute, the effects of brine expulsion and a small amount of water movement independent of the salt were measured during this period. Movement of water to the column surface was not associated with either vapor transport or salt sieving. The interface between the solid and liquid was a vertically diffuse interface rather than a sharp ice-bonded interface. Convection of the pore fluid occurred throughout the entire column. Pore fluid velocities were estimated to be on the order of 0.1 to 03 $\rm{m\over day}$ and do not exceed 1.4 $\rm{m\over day}$. Convection consisted of pore fluid in one half of the column moving down and pore fluid in the other half moving up and was associated with radial asymmetries in salinity, water content and ice-bonding. The effects of convection could be measured in the salinity profiles, but not in profiles of water content or temperature. A stability analysis showed the unstable density gradient in the partially frozen region was not sufficient to lead to convection. It was tentatively concluded that convection resulted from dense brine in the partially frozen region overlying less dense brine in the thawed region. Methods for estimating the final salinity profiles were not satisfactory since the BPS theory could not be applied to the experimental results and a stability theory for the pore fluid could not be developed which matched the experimental results. Application of these results to field situations is limited because of the restricted horizontal and vertical length scales. However, solute redistribution by convection is probably limited to freezing soils with large solute concentrations and large permeabilities.