Enhancements to a Simple Pavement Frost Model

Abstract

The ability to predict the timing of thaw is important to the design and the maintenance of roads. Such knowledge allows highway agencies to reduce the damage caused by enacting seasonal load restrictions during thawing. Because measurement by observation or testing requires time and resources, one desirable solution is a predictive model. Simple models, such as the Cornell Pavement Frost Model (CPFM), require only daily temperature weather inputs. Possible improvements to the CPFM were tried with a primary goal of improving thaw timing while keeping the model simple and applicable across New York State and beyond. The proposed modifications included changing the thaw reference temperature, providing better tracking of thermal properties throughout seasonal changes, and driving thaw from underneath with a constant underground temperature. During the investigation, further changes were tested involving ways to treat latent heat during thaw, including decoupling. Changes were tested against frost tube data from Ithaca, New York. The model was evaluated and adjusted at other sites in Maine and New Hampshire. The additional inputs required by the resulting modified model were depth and value of constant temperature, moisture content, and estimated shade. The modified model greatly improved the depth and timing of the model for most sites. The final model used the original formulation with minor modifications for freezing and a latent heat formulation for thawing. With more than one freezing–thawing front, the formulas were decoupled. Further investigation should be done to improve frost depth and timing, and in particular, to find out why rapid thawing from underneath does not occur at some sites.