Research Information System
Transportation Geotechnics, vol.60, 2026 (SCI-Expanded, Scopus)
Frost-induced damage to pavement infrastructure remains a critical engineering challenge. Accurately predicting frost heave in subbase and subgrade materials is essential for ensuring long-term durability of pavements. Determining the frost susceptibility of soils via direct laboratory testing remains the most reliable approach, however, the requirement for specialized equipment and extensive testing time makes it impractical for routine projects. Consequently, there is a need for reliable classification systems that can accurately predict frost behavior using simple index tests. While traditional classification systems like USCS and AASHTO are widely used, they primarily rely on grain size and plasticity, which often fail to differentiate the frost-heave potential of soils with similar fines content but different mineralogy. Hence, these systems are limited in their capacity to constitute a reliable framework for the assessment of frost heave behavior. This paper addresses the difficult problem of assessing the frost susceptibility of subbase and subgrade material focusing particularly on new soil classification parameters that have a potential for improving soil index based frost classification methods. This study utilized a custom-designed frost susceptibility testing system to conduct step-load freeze–thaw tests on twelve gravel-sand mixtures with fines contents ranging from 5.2% to 18%. Three distinct fines were tested: non-plastic silt, kaolinite, and a high plasticity clay. The mixtures were analyzed using the Revised Soil Classification System (RSCS) to evaluate its predictive potential compared to traditional methods. Detailed test results for all twelve mixtures (each with four duplicate samples) are shared including data pertaining to sample index parameters, heave rates and amounts, sample boundary temperatures and residual heave. The results demonstrate that frost heave rates are profoundly impacted by fine mineralogy rather than mass percentage alone, with kaolinite producing the highest heave rates. A significant finding is that RSCS parameters—specifically the liquid limit tested with brine, LLBrineC and electrical sensitivity, SE —successfully differentiate the three fine types into separate soil classes (NI, II, and IL). Furthermore, normalized heave rates were successfully modeled using non-linear regression, revealing that the threshold fines content expected to produce high-frost susceptibility (8 mm/day) varied drastically from about 5% for kaolinite to 74% for YK clay. These findings suggest that the RSCS provides a physics-inspired, data-driven framework capable of improving Level II frost classification. Future work should build on these initial findings by contributing to coupled RSCS and frost-heave databases, refining reference heave rates for fine and coarse soil fractions and addressing post-thaw strength loss.