Dynamic bending of an ice wedge resting on a winkler-type elastic foundation

Abstract

For most ice - sloping structure interactions, the incoming ice floes' failures are determined by the formation of circumferential cracks. This physical process can be simplified as analyzing the bending failure of an ice wedges resting on a fluid foundation. In history, closed-form analytical/empirical solutions have been developed for the static bending problem; and numerical solutions have been attempted for the dynamic scenarios. This paper revisits this classic problem and conducts extensive Finite Element Method (FEM) - based simulations on the dynamic bending of an ice wedge resting on a Winkler-type elastic foundation. The simulations are based on inputs (i.e., ice wedge geometry, loading radius and loading rate) within ranges that are typical for engineering applications. Based on the simulation, a database of 'ice breaking load' and 'ice breaking length' is constructed. Then we applied the Artificial Neural Network (ANN) method to establish the general relationship between the varying inputs (i.e., ice wedge angle, loading radius and rate) with the target outputs (i.e., breaking load and length). Such relationship is expressed in simple closed-form (i.e., Eq. (13)) allowing for easy, efficient and wide engineering applications. In the process of developing the ANN model, based on extensive FEM-simulations, we managed to extend Nevel's (1972) analytical solution. We also quantitatively demonstrated many well-known dynamic effects in this classic problem, e.g., a faster loading rate leads to a larger ice breaking load and a shorter ice breaking length. In addition, we also uncovered the failure pattern transition of an ice wedge, i.e., when an ice wedge's angle is below 100, the circumferential crack will develop first; however, when the ice wedge gets wider than around 100, depending on the loading radius and interaction velocity, the radial crack is more prone to develop first.