| Résumé | A size-effect study has been done for the first time on simply supported beams under three-point loading. Previous scale effect studies on ice were done under compressive stress. First-year sea ice with a columnar grain size on the tensile surface of 2.4 mm, laboratory-grown freshwater columnar ice with a columnar grain site of 3.4 mm on the tensile surface, and granular iceberg ice of 5.8 mm grain size were loaded to failure at a calculated strain rate at the extreme fibre of 10/sup -3/ s/sup -1/ with failure times of 0.1 s. Four orders of magnitude in beam volume was tested in the sea ice, three orders of magnitude in the laboratory ice and two orders of magnitude in the iceberg ice. The sea ice strength was found to decrease as V/sup -1/12/, similar to the dependence reported for coal and granite, and for ice under compressive stress. There was little discernible freshwater ice strength dependence on volume, and the larger iceberg beams had a higher strength than the smaller. The three-parameter Weibull distribution was fitted to the failure strength data in each of the nine data sets and the values of the parameters evaluated. The predictions of the weakest-link model implicit in the use of this distribution, quantified with Weibull parameters thus obtained, were compared with the change in mean strength with sample volume, and there was very poor agreement. A calculation of the flaw density at mean strength was made and compared with the grain density. It was concluded that, although the three-parameter Weibull distribution fits the failure strength distribution well in any given data set, simply supported beams of ice loaded in flexure do not fail according to the weakest-link hypothesis. Instead a crack initiation criterion seems to be what controls failure. |
|---|
