ORCID

Abstract

Floating offshore renewable energy requires dynamic power cables to transmit generated electricity from the floating device to the seabed, or to an adjacent device, so that it can be exported to shore. Dynamic power cables are connected at the top end to a floating device, and traverse the water column, subjected to loading at the hang-off, and direct hydrodynamic loading from waves and currents. Their reliability has been a concern within industry and research, particularly due to historically high failure rates of static subsea cables. It is essential that design and analysis of dynamic power cables is accurate, to mitigate damage and avoid costly downtime and repairs, and this requires accurate models. This PhD is focused on global modelling and configuration design of dynamic power cables.One widely-adopted method of modelling dynamic power cables is to use mid-fidelity global numerical models, such as OrcaFlex, which are capable of modelling the full length of the cable in context with the entire system. Another method of modelling floating structures for offshore renewable energy is through small-scale tank testing experiments. The first part of this thesis presents comparison between such experiments and the numerical model OrcaFlex.Firstly, a model-comparison study is presented for a floating offshore wind turbine platform, demonstrating overall good comparison, although with underprediction observed in surge, heave and pitch motion particularly, which aligns with previous research. Then, a novel small-scale modelling methodology was developed, where a dynamic power cable was modelled and tested at 1:70 scale, attached to a floating offshore wind turbine in an ocean basin. These experiments were the most comprehensive of their kind, with motion response of the dynamic power cable captured and analysed using an underwater optical tracking system. The cable was also modelled numerically, and a comprehensive experimental-numerical comparison was conducted. The numerical model is seen to predict motion well for the tests considered, with some small differences observed and discussed. Subsequently, the numerical model was used to develop an optimisation for dynamic power cable configurations. Designing the configuration is a multi-objective problem due to conflicting design requirements. Therefore, a genetic algorithm method is adopted, capable of optimising for multiple objectives. The methodology advances current research with the implementation of machine learning. Random forest surrogate models were trained on numerical model simulations from OrcaFlex to predict key design metrics within the optimisation, increasing efficiency and allowing for full-length simulations to be used along with a fatigue analysis. The methodology was demonstrated for a case study of a lazy wave configuration and a suspended configuration, where it was seen to perform well for the former, but not for the suspended design due to its complexity. Optimal designs are selected for the lazy wave configuration using a weighting technique, and the designs are discussed in terms of the three objectives, where trade-offs are observed.

Awarding Institution(s)

University of Plymouth

Supervisor

Martyn Hann, Scott Brown, Shanshan Cheng, Robert Rawlinson-Smith

Document Type

Thesis

Publication Date

2026

Embargo Period

2026-07-23

Deposit Date

July 2026

Creative Commons License

Creative Commons Attribution-NonCommercial 4.0 International License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License

Download

Share

COinS