Optimization-based control of diesel-electric ships in dynamic positioning

Abstract

Recent advances in computer hardware and algorithms make it possible to consider more computationally demanding control methods, allowing more effective exploitation of the equipment under control.

This thesis explores new ways of controlling ships (or other marine vessels) that are designed to keep a pre-determined position and heading automatically exclusively by the means of their thrusters – a task called “Dynamic Positioning”, or DP. Special attention is given to the interplay between the thruster system and the power plant that supplies it.

A DP control architecture typically consists of at least 1) a DP control algorithm that considers the current position and velocity of the vessel against the DP setpoint, and calculates the total forces and the moment that the thruster system should produce, and 2) a thrust allocation (TA) algorithm that calculates the forces to be produced by the individual thrusters to match the command from the DP control algorithm. Chapter 2 describes a TA algorithm that enables centralized control over the power consumption in the thruster system. It achieves that by allowing the TA to make short-term deviations from the command it receives from the DP control algorithm; the resulting deviations in position and velocity of the vessel are carefully monitored and constrained, and are usually small due to the large inertia of a typical marine vessel. This enables the thrusters to counter-act load variations from other consumers on the ship, reducing the total variations on the power plant. The TA algorithm is tested on a simulated marine vessel, which includes a realistic marine power plant.

In Chapter 3, a more efficient version of this algorithm is described. The improvement in efficiency is achieved by positioning the vessel against the slowlyvarying component of the environmental forces in a way that increases the acceptable deviation margins in the likely drift-off direction. In Chapter 4, the capabilities of the thruster system to control its power consumption are examined from a theoretical perspective.

Much of the work above required a mathematical model of the power output from a diesel engine; a model that is well-suited for controller design and verification purposes was designed based on first-principle models in the literature. This model was then used to design an improved diesel engine governor (controller) algorithm, which is described in Chapter 5.

The TA algorithms that are described in the literature usually focus on solving one or a few aspects of the TA problem at a time. In Chapter 6, functionality from several earlier publications is gathered into a single TA algorithm. The singularity avoidance functionality is given additional theoretical treatment.

Implementing a DP control algorithm that is aware of thruster limitations such as saturations and rotation rate constraints involves largely heuristic adaptions. Chapter 7 introduces a DP control architecture that avoids having separate DP and TA algorithms, and is instead based on a single MPC-based controller. This allows better coordination between control of the thrusters and control of the ship.