Freezing and thawing of frost-susceptible soils : (development of a reliable predictive model)

Frost depth is an important factor that affects the design of various transportation infrastructures including pavements, retaining structures, bridge foundations, utility lines, and so forth. Soil freezing can lead to frost heave and heave pressure, which may cause serious stability issues. On the other hand, at the beginning of spring season, the ice starts to thaw from the top down and to a lesser extend from the bottom up. The melted water below the pavement surface is trapped (setting on impermeable frozen materials). It saturates the top part of the upper pavement layer. Consequently, the stiffness of the saturated layer decreases causing substantial decrease in its load bearing capacity and high deformations, which lead to premature and localized failure. To decrease the spring thaw damage, Spring Load Restrictions (SLR) signs are usually placed along the roads. The objectives of this study are to develop accurate and reliable frost and thaw depth and frost heave prediction models, estimate heave pressure and develop a reliable SLR policy. After extensive literature review, various existing frost depth models were identified and tested. These include the finite difference UNSAT-H, the Stefan, the Modified Berggren, and the Chisholm and Phang models. Unfortunately, some of these models require substantial input data that are not available and all models yielded inaccurate results. Therefore, statistical frost depth models were developed using frost depth and air temperature data collected by Michigan Department of Transportation (MDOT); one for clayey soils and one for sandy soil. The two models were then combined using the measured thermal conductivity of clayey and sandy soils. The combined statistical model was then verified using frost depth and air temperature data collected by Minnesota Department of Transportation (MnDOT). Additionally, The Gilpin's mechanistic-empirical model was employed to predict frost heave. The model produced inaccurate and counterintuitive results in some cases. Therefore, the model was modified and the empirical frost depth model developed in this study was incorporated into the model. The resulting model was then simplified to replace some of the required of input data that are not available. The modified model accuracy was assessed using the frost heave data measured at 5 sites in Oakland County, Michigan. Further, the relationship between frost heave and heave pressures were established for four soil types.Moreover, a new statistical model was developed for calculating the cumulative thaw degree-day (CTDD) using pavement surface temperature and air temperate data collected by MDOT. Then, the thaw depth data measured in the state of Michigan were used to assess Nixon and McRoberts thaw depth predictions model. Since the model did not produce accurate and acceptable results, statistical thaw depth models were developed using the calculated CTDD values and thaw depth data collected by MDOT and MnDOT; one for clayey soils and one for sandy soils. The models were then verified using the calculated CTDD values and thaw depth data collected by MnDOT. Finally, based on the results of thaw depth model a new SLR policy was proposed.