Numerical Predictions of Global and Local Ice Loads on Ships

Abstract

Ice loads represent the dominant load for ice-going ships, and it is important to estimate both global and local ice loads on ship hulls. The global ice load governs the ship’s overall performance in ice, and it is an integrated effect of local ice loads over the hull area. Information on the distributions of local ice loads around the hull can be used for more effective design of ice-going ships both in terms of overall operation and from the structural point of view.

The present thesis focuses on a numerical model for simulating ice–hull interaction and ship maneuvers in level ice. This model is partly based on the empirical data, by which the observed phenomena of continuous icebreaking can be reproduced. In the simulation of a full-scale icebreaking run, the interdependence between the ice load and the ship’s motion is considered, and the three degree-of-freedom (DOF) rigid body equations of surge, sway and yaw are solved by numerical integration. The thickness and strength properties of the ice encountered by the ship are assumed to be constant or predefined based on the statistical data. Accordingly the global and local ice loads on ship hulls can be obtained in a deterministic or probabilistic way.

The convergence tests are carried out to make sure that this numerical method can give a convergent solution of both global and local ice loads. The computation time is also examined with the purpose of determining a balance between the computation time and the convergence. The influences of the assumptions and simplifications made in this numerical model are analyzed by changing different parameters and comparing with an empirical method and the measured data.

The simulation results are discussed through three case studies, in which the global ice load effects on ship’s performance, the probabilistic and spatial variations of local ice loads around the hull and the short-term distribution of maximum ice loads on a frame are respectively analyzed and compared with field measurements conducted in the Baltic Sea.

The ship’s performance in level ice is usually described by the speed that the ship can attain in the ice of a certain thickness. A case study with an icebreaker, Tor Viking II, is carried out by using the simulation program. In this study, the ice encountered by the ship is assumed to be uniform. The ship’s motion is obtained by solving the equations of motion in which the thrust and the global ice load both are identified. The speed that the ship can attain is then simulated in the ice of different thicknesses. The simulation results agree well with the full-scale measurements. The turning circle diameter which is a measure of the ship’s maneuverability in ice is also investigated by using the simulation program. Herein the ship is assumed to turn freely with a given rudder angle, and the simulated turning circles are comparable to the full-scale ice trials. It is also found that the ship’s performance can be considerably affected by the geometry of simulated icebreaking patterns. If the shoulder crushing takes place (i.e. the ice is continuously crushed by hull shoulder without bending failure) both the forward speed and the turning rate of the ship will be significantly slowed down. This phenomenon is also observed in full-scale trials, but it is difficult to learn about its effect on ship’s performance as the actual ice conditions in-service usually are uncontrollable. It is expected that numerical simulations can supplement full-scale tests in providing more details about the continuous icebreaking processes and the global ice load effects on ship’s performance.

The local ice loading process has a clear stochastic nature due to variations in the ice conditions and in the icebreaking processes of ships. A case study with an icebreaking tanker, MT Uikku, is carried out by using the simulation program. In this study, the thickness and strength properties of the ice encountered by the ship are assumed to be constant or randomly generated using the Monte Carlo method. It is found that the variation of simulated ice loading process on a frame is noteworthy even if the ice properties are fixed. If the statistical variation of the ice conditions is considered, then the distributions of simulated load peaks are found to be comparable to the measured statistical distributions. In this case study, the spatial distribution of local ice loads around the hull is also investigated. The simulation results agree with previous experimental studies, that the turning operation may develop a high load level on the aft shoulder area.

Ice conditions and ship operations in ice vary in the short term from voyage to voyage and in the long term from winter to winter. Long-term ice load measurements conducted in the Baltic Sea consist mainly of 12-hour load maxima which are gathered during the normal operation of the ship over several years. A case study with a chemical tanker, MS Kemira, is carried out by using the simulation program. In this study, the statistical data on the strength properties of Baltic Sea ice are applied and the thickness of the ice encountered by the ship are classified referring to the full-scale measurements onboard MS Kemira. The 12-hour maximum ice loads on a frame are evaluated by fitting a Gumbel I asymptotic extreme value distribution to the simulated 10-min load maxima in a certain ice condition. It can be expected that if a reasonable variance of the ice thickness is defined, the simulation results can be used for a preliminary estimation of the maximum ice loads within a 12 hours’ voyage in level ice. By applying the different ice thicknesses in simulations, the probable correlation between the simulated load maxima and the ice thickness is analyzed. A potential way to evaluate the long-term ice load statistics based on short-term simulations is then introduced.

Up to now the main source of knowledge about ice load statistics has been field measurement. While field measurements will continue to be important, numerical methods can provide useful information, since they can be easily used to study the effect of different parameters. As far as we know the present numerical model is the first one to deal with multiple subjects including the ice–hull interaction, the overall performance of icebreaking ships and the statistics of local ice loads around the hull. It is hoped that further studies on this numerical model can supplement the field and laboratory measurements in establishing a design basis for the ice-going ships, especially for ships navigating in first-year ice conditions