Ice loads time series should be treated in the other way than separate independent ice loads observations. The stochastic process approach can provide information about such important characteristic as mean length of signal’s outcome beyond some critical level and expected number of such outcomes.
The paper considers global ice loads registered in an ice tank experiment with a cylindrical indenter of 100 mm width. The autocorrelation function is fitted in a manner that the observed load process is differentiable. The study conducted in the paper demonstrates that characteristics, such as the number outcomes beyond some critical level and the time spent off this level, are governed in the same way as parameters of a stationary differentiable normal process.
Normal stationary model of ice loads process allows its simulation, if autocorrelation function is given. In the paper, such simulation is performed.