Abstract The long-term response of an artificially thickened drilling ice platform is investigated using a new creep bending finite element model. The ice platform is simulated using thick plate theory, and the fluid support is modeled as a linear Winkler foundation. The numerical results produced by the model have shown good agreement with a small scale load bearing capacity test. A mathematical technique will be described to perform a probabilistic analysis. This technique allows one to study the effect of a random change in air temperature, grain size, creep properties, elastic properties, and ice cover thickness, on the creep response of ice cover. An empirical formula to predict the central deflection of the ice cover which is based on the field and numerical data is presented which is then combined with the previous technique to carry out a probabilistic analysis of the long term response of the ice cover. The mathematical details of a formula to predict the time of onset of failure, (time of safe operation) based on the critical strain energy failure criterion, is also presented. The effect of random change of a number of different parameters on the time of safe operation is discussed. Introduction The increasing need for oil and gas has led to many schemes for drilling offshore in the Arctic. Artificially thickened ice has been used to make drilling platforms that are about 4.5 to 5.5 meters thick. A number of construction techniques have been developed to increase the first year ice cover thickness of 1.5 m to 2.0 m to this required thickness. A review of the analysis, design and construction of floating ice islands for offshore drilling has been made by El-Tahan.1 The actual methods used to design these artificial ice platforms may be divided into the following categories:Elastic Analysis: The ice platform is treated as an infinite elastic plate resting on an elastic (fluid) foundation. This method can only accurately predict the load bearing capacity for short durations of loading.Viscoelastic Analysis: A number of viscoelastic models have been proposed to predict the time dependent response of ice. These models usually employ spring and dashpot elements which are connected in series and/or in parallel. This method may be used to predict the long term response of a floating ice cover.2Creep Analysis: This method, which may be used to predict the time dependent response of ice, uses empirical equations which have been determined originally from laboratory creep tests. Recently, a creep bending model has been develop3. Regardless of the design method, deterministic techniques are usually used to predict the load bearing capacity of the ice platform. Probabilistic techniques are necessary to fully investigate the effect of random changes of a parameter on the response of the ice platform. A simple procedure is described to predict the time of safe operation for each field operation together with the associated percentage of risk. Deterministic Finite Element Analysis Hamza and Muggeridge3 have developed a new deterministic creep bending finite element model the numerical results of which have shown good agreement with field data obtained during a load bearing capacity test.4,5