Predictive control strategies for safe payload handling in crane-based offshore operations

Crane-based loading operations present an integral part of today’s ocean economy, an industry that is projected to become even more vital due to the emergence of offshore windparks as part of a more sustainable future. At the same time, offshore operations are getting increasingly challenging as drilling platforms and wind parks move from shallow waters to open sea, where more severe sea states are common. So-called knuckle boom cranes (KBC) are deployed e.g. on supply vessels and offer increased flexibility during payload handling thanks to an additional articulated boom. To this day, most loading operations are controlled manually requiring highly trained crane operators and additional personnel overseeing operation. Increasing the level of automatization promises a simplified handling task, reduced costs and improved operational safety. Yet, only solutions for vertical payload stabilization are established in industrial practice. The objective of sway control attracted interest from academia, while an holistic approach to spatial payload stabilization through automated control remains an open gap. In context of this thesis project, predictive control strategies directed towards more efficient and safe offshore operations are researched. After reviewing common modeling approaches, a control-oriented model of vessel, crane and payload is derived, which forms the basis of the investigated model predictive payload controller. Different formulations of the underlying optimal control problem are assessed for control performance and real-time feasibility. In particular, a control scheme is put forward leveraging the differential flatness of the crane-payload system in order to invert the system equations. It further motivates a payload-centric approach to payload stabilization and trajectory tracking. The flatness-based model predictive controller (FMPC) is compared to established linear as well as nonlinear versions of MPC. The considered predictive control topology is complemented by a target selector yielding optimized crane configurations and a receding horizon observer providing estimates of the system state alongside short-time predictions of the vessel motions. The controller designs are studied in simulation for different sea states. The control performance is shown to be directly linked to the available capacity of the crane’s hydraulic actuators. Also, the added benefit of optimizing the crane configuration based on the crane’s manipulability index is demonstrated. Last, first validation trials of a model predictive payload controller in a robot-based test bench are presented suggesting that MPC can be used to induce damping and reduce payload oscillations. The thesis concludes with a discussion of operational safety from an automated control perspective.