Discrete ice lens theory for frost heave in soils

Abstract

The existing segregation potential (SP) method for frost heave prediction in soils is semiempirical in nature and does not explicitly predict the relationship between heave rate, temperature gradient, and other more fundamental soil properties. The SP method assumes that the heave rate is directly related to the temperature gradient at the frost front but acknowledges that the SP parameter is dependent on pressure, suction at the frost front, cooling rate, soil type, and so forth. This paper extends and modifies an approximate analytical technique of Gilpin and accounts for the effects of distributed phase change within the freezing fringe in both the head- and mass-transfer components of the formulation. The approach requires as input a relationship between frozen hydraulic conductivity and temperature and predicts the discrete location of each ice lens within the freezing soil. The solution can be carried out quickly on a microcomputer to obtain the heave, suction at the frost front, ice lens temperature, and other results of interest with time. Furthermore, the discrete ice lens method predicts the effects of changing overburden pressure on the predicted heave rate. A method of extracting input parameters for the discrete ice lens procedure from a series of frost heave tests is proposed. The discrete ice theory has been tested and calibrated against well-documented frost heave test results in the literature, and very encouraging agreement between prediction and observation has been obtained. Key words: frost heave, discrete ice lens, segregation potential, hydraulic conductivity of frozen soil, freezing soil.