Transient simulation of the propulsion machinery system operating in ice – Modeling approach

•Bond graph method is introduced as a generalized method for modeling of the propulsion machinery system exposed to transient ice-related torque load.•Sensitivity of the propeller shaft response to a different number of propeller shaft lumps and modes is reported.•Sensitivity of the produced power and torque in the diesel engine to a different discretization of the crank mechanism is reported.•The proportional-integral-derivative (PID) controller is used to guide the combustion process inside the cylinders.•The most suitable discretization format of the propeller shaft and crank mechanism is recommended for the presented propulsion machinery system, considering the sensitivity of the propeller shaft response and computational time.

The propulsion machinery system is one of the major concerns in the design of ships that operate in ice-covered waters. Numerical models and simulations are important tools for a better understanding of the system subjected to the transient ice-related loads. Design loads and design criteria applied in the typical numerical simulations are described by classification societies. However, the rules do not cover the level and type of discretization of the propulsion machinery system is not covered. Therefore, in this paper we study the sensitivity of the system response to: 1) finite-lump vs. finite-mode propeller shaft sub-models; 2) the number of propeller shaft lumps and modes; 3) simplified vs. complex sub-model of the diesel engine. The propulsion machinery system is modeled using the bond graph method. Dynamic torque and angular velocity responses of all elements of the propulsion machinery system are simulated. Simulation results show that the selection of sub-model discretization and fidelity are important parameters for the accuracy of the dynamic response and natural frequencies. Furthermore, energy flow and computational time are used as criteria to identify the most suitable discretization and fidelity of the sub-models in question.