Author(s): Janne Ranta; Jukka Tuhkuri; Arttu Polojarvi; Jani Paavilainen
Linked Author(s):
Keywords: No Keywords
Abstract: In this paper, multivariate linear regression modeling technique and systematic model selection procedure are used in reconstruction of simulated horizontal peak ice load data. The main scope of this paper is to see how different model parameters affect ice loads. The data is obtained from earlier simulations of ice interactions against inclined rigid wall using two-dimensional finite-discrete element method (2D FEM-DEM). Results from over 200 simulations are included into statistical analyses. Simulated horizontal peak ice load histories are divided into three subregions that represent initial, middle and end parts of ice-structure interaction processes respectively. In addition, fourth region is defined to cover the whole simulation. Multivariate regression models based on horizontal peak ice loads are then constructed for all of these four regions separately. As a result, linear regression models are shown explicitly and effects of different parameters on the horizontal peak ice loads are discussed. Regression model suggests that only few parameters are required for having fairly good reconstruction of original simulated peak ice load data. The results also indicate that the rubble pile in front of the structure clearly has an effect on the ice-structure interaction process.
Year: 2014