Development of an intelligent control combinator curve (I3C) for a controllable pitch propeller propulsion system

Abstract

Ships equipped with controllable pitch propellers (CPP) are widely used in maritime transport for their operational versatility. Unlike fixed pitch propellers (FPP), these systems provide two degrees of freedom for controlling vessel motion by allowing simultaneous regulation of the propulsion machinery speed and the propeller pitch to achieve the thrust needed to overcome hydrodynamic resistance. Control commands for these propulsion systems are typically established using combinator curves, developed during the design phase and subsequently adjusted during initial sea trials. These curves are intended to ensure that the main machinery operates within acceptable ranges and can provide sufficient power to meet propulsion demands. However, because they are built with predefined safety margins, they restrict the propulsion system from achieving optimal performance under changing operational conditions. This thesis introduces the Intelligent Combinator Control Curve (I3C) framework, which leverages the ability of Artificial Neural Networks (ANNs) to model nonlinear systems and multi-objective optimization methods (weighted sum, epsilon constraint, and NSGA-II) to formulate adaptive policies for selecting propulsion control commands based on data collected during real sea trials, and to generate combinator curves for different environmental and operational scenarios, with the objective of reducing fuel consumption and ensuring the required navigation speed. The resulting curves are validated using operational data collected during the experimental phase, demonstrating that the proposed reference framework reduces fuel consumption by up to 18.42% compared to conventional curves, without compromising the performance of the propulsion system.