Frost Depth: General Prediction Model

Abstract

Accurate frost depth prediction is an important aspect of engineering design for infrastructure including pavement, building and bridge foundations, and utility lines. Actual frost depth is affected by material type, soil thermal properties, and soil water content as well as climatic conditions such as temperature, wind speed, precipitation, and solar radiation. Numerical or analytical modeling techniques could be used to estimate frost depth; however, the required input data are unavailable or expensive to collect. Therefore, the use of numerical and analytical models entails estimating missing data that affect the reliability of the results. The accuracy of three analytical and semiempirical frost depth prediction models (Stefan, modified Berggren, and Chisholm and Phang empirical) were tested with soil temperature data from the Michigan Road Weather Information System. Because none of the models yielded accurate results, revised empirical models developed for different soil types required only daily high and low air temperatures as inputs. The predicted frost depths were more accurate than in previous models. Finally, by using thermal conductivity values for each soil type, the models were combined into one general model that required thermal conductivity and average air temperature as inputs.