High strain-rate behavior of polycrystalline and granular ice: An experimental and numerical study

Abstract

We study the stress–strain response of two different types of ice, viz. polycrystalline ice and granular ice, between −1° – 0 °C over a strain-rate range of 100s−1 to 300s−1 employing the split Hopkinson pressure bar (SHPB). Polycrystalline ice samples, prepared by freezing water in plastic moulds, exhibit a compressive strength ranging from 7 to 10 MPa within the considered strain-rate range. The strain at peak stress remains below 0.2%, indicating brittle behavior. The stress-strain curve of polycrystalline ice displays a prolonged tail, suggesting that the damaged ice specimen retains some strength. High-speed imaging during tests reveals the damage mechanism in ice is fragmentation and axial splitting. A user subroutine based on the Johnson–Holmquist II (JH-2) model is implemented in the commercial finite element (FE) software ABAQUS to predict ice's response at high strain-rates, which captures the softening present in the experimental stress–strain curve. Intact strength parameters and strain-rate sensitivity constants in the JH-2 model are determined from our experimental data and literature results, ensuring alignment with experimental peak stress. Fractured strength and damage evolution parameters are determined by matching post-peak responses from simulations to experiments. Temporal damage evolution from FE simulations aligns well with high-speed images from experiments, providing additional validation. Extending the study to granular ice, samples are prepared by crushing polycrystalline ice and refreezing it. The compressive strength of granular ice at a nominal strain-rate of 200±50s−1 is found to be 4±0.7 MPa. The granular ice, which is a mixture of polycrystalline ice and voids, is homogenized using rule-of-mixture to obtain the elastic properties. The FE simulation results utilizing the JH-2 parameters that we determine matches well with the experimental data, demonstrating that the JH-2 model is well suited to predict the high strain-rate behavior of both types of ice.