Main Article Content
Designing bulbous bows for ships remains a challenging task. Their impact on different design attributes as well as their change in performance when operating off their intended design condition renders this as a multidimensional problem. This paper explores the application of machine learning techniques to a sample of in-service vessel data to develop a preliminary design tool. The ships' data was analysed together with their bulbous bow data to generate machine learning models using a supervised approach. The K Nearest Neighbours Classifier and Regression models were used as the basis of the tool. Together, these models can be used to predict whether to install a bulbous bow and the recommended dimensionless coefficients for new vessels. Generating this preliminary bulbous bow design tool required the introduction of new dimensionless coefficients that discretise the bulbous bow's longitudinal section. The preliminary design tool gives the designer the ability to determine whether a bulbous bow should be fitted and, if so, to obtain an initial estimate of the bulbous bow required for the vessel being designed, based on key input parameters that relate to the ship and its operation. The new design tool is demonstrated to provide preliminary design details for bulbous bows through the case studies.