Laboratory investigation of frost susceptibility of crushed rock aggregates and field assessment of frost heave and frost depth

Abstract

In Nordic countries that experience seasonal frost, understanding how it affects the service life of linear transport infrastructures such as roads and railways is essential. Frost action both regroup frost heave and thaw weakening processes. This research focuses on frost heave, which occurs when three conditions are met: freezing temperature, water availability and frost-susceptible materials. Frost susceptibility is therefore defined as the ability of an unbound granular material (natural soil or crushed rock aggregate) to form ice lenses due to cryosuction, which is the suction of water to the frost front. Frost heave in a frostsusceptible material lifts the layers above the freezing front. Many kinds of damage are attributable to frost heave, such as cracks in and unevenness of the road surface, making it uncomfortable and even dangerous for road users. Frost heave and subsequent thawing can significantly reduce the service life of roads and are accompanied by high associated maintenance and reparation costs. Knowledge of the frost susceptibility of the materials comprising the different layers of transport infrastructure is therefore crucial to optimal frost design. This research mainly aimed to characterize the frost susceptibility of crushed rock aggregates. Laboratory experiments using a multi-ring frost cell were carried out to estimate the segregation potential of the 0-4 mm fraction of crushed rock aggregates. Twenty-four tests performed on nine different rock types found that crushed rock aggregates with a fines content of <63 μm between 11.6% to 25.5% are highly frost susceptible. The study found poor correlation between <63 μm fines content and segregation potential. The latter was used to characterize the frost susceptibility of the different crushed rock aggregates primarily for its capacity to estimate frost heave magnitude easily. The use of a grain size criterion, as presently used in Norway, seems to be efficient for dividing non-frost-susceptible from frost-susceptible aggregates, but does not allow estimation of crushed rock aggregates’ segregation potential with a good degree of confidence. As frost heave tests require a costly laboratory setup and specialist personnel, the segregation potentials of crushed rock materials were estimated using material index properties such as initial water content, liquid limit, mean particle size of the fine fraction and specific surface area of the fine fraction. This method is used by the Quebec Ministry of Transportation for soils, and the goal here was to assess the suitability of the methodology for use with crushed rock aggregates. It was found that estimation from material indexes gave segregation potential results within a range of ±15% compared to those obtained from frost heave tests. There was a direct correlation between the specific surface area of the fine fraction of a crushed rock aggregate and its segregation potential, from which an equation was developed. The study showed that the crushing phase had an effect on the frost susceptibility of the tested crushed rocks. Aggregates are more frost susceptible after the first crushing stage than after the third or fourth crushing stages. The hypothesis that the fine fraction is enriched by weak minerals at the first crushing could explain this behaviour, but further research is necessary before this theory can be presented with confidence. The design was optimized by modelling three road sections with different frost protection layers using the SSR model and the I3C ME software. The fully-instrumented test site is situated in the Røros municipality where harsh winters are the norm, with the freezing index of an average winter equal to that of regions such as Oslo, Trondheim, Narvik, Tromsø etc. It was found that the SSR model is suitable for both frost penetration and frost heave estimation for thick (2.05 meters) layered road structures. The SSR also permitted the back-calculation of key parameters such as dry density, moisture content and segregation potential, making it useful for assessing existing infrastructure parameters. Using a segregation potential function, i.e. segregation potential according to time, was found to produce optimal frost heave estimations in a transient thermal regime.