Regression-Based Assessment of Shafting Torsional Vibration Key Responses

Abstract

A novel procedure for the assessment of propulsion shafting torsional vibration key responses is proposed in this paper. The procedure is based on the response surface methodology applied to a set of system responses compiled over a selected design space. Design space of 1250 design points comprising the shafting stiffness, propeller, turning wheel, and tuning wheel mass moments of inertia coordinates has been employed to build the first torsional natural frequency, crankshaft peak vibration torque, and shafting peak vibration torque quadratic polynomial approximations. Statistical evaluation performed on a full test set of 2500 design points showed that the mean relative errors of 0.14% for natural frequency, 2.93% for crankshaft peak vibration torque, and 0.41% for shafting peak vibration torque were achieved. Good agreement between the assessed and actual torsional responses stresses the importance and utility of the proposed metamodels for the propulsion shafting preliminary design purposes. An example application based on the 114,000 dwt tanker propulsion shafting is also provided.