Experimental study of concrete abrasion due to ice friction — Part II: Statistical representation of abrasion rates and simple, linear models for estimation

This paper is the second of two papers that present, discusses and processes the results from experiments where ice abrasion properties of concrete were investigated in relation to concrete compressive strength (fc), nominal ice–concrete contact pressure (Pice) and ice temperature (Tice). The abrasion testing was done by sliding a vertically oriented fresh-water ice cylinder in a repetitive back and forth motion on the surface of the concrete specimens. The experiments consisted of two series; one containing laboratory cast specimens and one containing field samples from an offshore concrete lighthouse with proven damage from ice abrasion. fication of appropriate probability distribution functions by use of Anderson–Darling statistics revealed that ice abrasion rates from the experiments were well represented by a 3-parameter Weibull distribution. The representation of the experimental data by this distribution was found statistically significant, based on a significance level of 95% (α = 0.05). perimental results were used as input to regression analyses with abrasion rate as response parameter and fc, Pice and Tice as predictors (explanatory variables). A linear regression equation based on these three predictors gave a fairly good representation of the experimental data (R2 = 66.4%). The most important predictors were fc and Pice while the effect of Tice on abrasion rates was limited. For typical values of the three predictors, the relative contributions to total abrasion rates were typically 29%, 17% and 6% for fc, Pice and Tice, respectively. tion of an offshore concrete lighthouse with material quality approximately conforming to the minimum requirements in ISO, 19903:2006/ISO, 19906:2010 showed severe damage from ice abrasion at the waterline. Maximum abrasion depth was measured to 81 mm after 44 years of seasonal sea ice exposure. application of two of the regression equations derived from the conducted experiments (one from each of the two test series) and typical input values of the three predictors provided fairly good estimates of maximum abrasion depth for the inspected lighthouse, with estimated/measured ratios of 0.77 and 0.86.