Optimizing Floating Offshore Wind Arrays: Enhanced Macro- and Micro-Siting for Maximum Yield, Reduced Levelized Cost of Energy, and Improved Generation Stability

Abstract

The current demand on the UK’s transmission network is met primarily by gas-fired generation, a dominant fossil fuel, alongside a growing share of renewable energy sources such as wind, solar, and hydroelectric power, as well as other low-carbon resources like nuclear and biomass. The energy mix also includes interconnector technologies to meet the daily demand for electricity in the country when other resources are not meeting the demand. In September
2024 the UK’s last coal-fired power station, Ratcliffe-on-Soar, was shut down, marking the UK as the first G7 country to completely eliminate coal from its power generation. With rising demand for clean and reliable energy across the UK (50 GW by 2030 - including at least 5 GW from floating offshore wind (FOW)), attention is turning to offshore resources as a key contributor. As the offshore wind (including floating) industry emerges worldwide, including the Celtic Sea a region off the south-west coast of the UK. FOW focus through European
Regional Development-Funded projects (ERDF) have supported the deployment of floating light detection and ranging systems (FLiDAR) within two current Crown Estate refined areas of search (RAS) within the Celtic Sea and the focus of this study. These FLiDAR systems provide high-resolution wind speed data at multiple heights within the atmospheric boundary layer, ranging from 47 m to 272 m above sea level (ASL). ERA5, the fifth generation European Centre for Medium-Range Weather Forecasts reanalysis data, was used and validated by the FLiDAR systems campaigns to quantify and characterize the frequent
occurrence of high wind shear and low-level jet events, both having a considerable impact on turbine operations offshore over a timescale of four decades (validation results r ≈ 0.94 and R^2 ≈ 0.87). Resource characterization involved both theoretical modelling of wind flow dynamics and technical assessment using observational data and simulation tools to evaluate
the quality, variability, and extractable potential of wind energy at the given sites. By quantifying inter-annual, decadal-scale variability and power-law-based wind shear exponents, this research aims to better assess the long-term reliability and stability of wind resources for energy yield and infrastructure planning (Weibull distribution results for four decades, shape k ≈ 2.2 m/s, scale A ≈ 11.1 m/s, mean ≈ 9.9 m/s; a statistical distribution used to describe wind speed variability for resource assessment and energy yield calculations). As the FOW
areas are farther from shore, newer larger turbine power curves were used for a 15 MW unit (242 m rotor diameter, 150 m hub height) and a 22 MW unit (284 m rotor diameter, 170 m hub height) to capitalize on this energetic resource produced by The National Renewable Energy Laboratory (NREL) as reference wind turbine innovation parameters. Each reference wind turbine produces an energy yield of ≈ 89.8 GWh/year, capacity factor ≈ 61.6 (15 MW) and energy yield of ≈ 126.4 GWh/year, capacity factor ≈ 59.3 (22 MW) without wake effect.

Wind resource numerical modelling (PyWake – an open-source Python package for simulating wind farm performance, wake effects, and energy production) and submarine power cable (COMSOL multiphysics – a commercial multiphysics simulation software that models coupled physical processes, for example fluid flow, heat transfer, and electromagnetics) across larger wind farm scales (1.08 GW and 1.23 GW this study) surrounding micro-, meso-, and macro-scale is gaining momentum with the rapid expansion of FOW energy. FOW farm wakes
can extend several tens of kilometres, influencing the wind conditions over large regions and adjacent wind farms. The wind farm engineering wake model used was PropagateUpDownIterative with the combination wake models selected for the final application as, wake deficit BastankhahGaussianDeficit, superposition SquaredSum, rotor averaging GaussianOverlapAvgModel, and turbulence STF2017TurbulenceModel verified by the sensitivity matrix test as the preferred models. The overall efficiency of wind farms is significantly influenced by turbine-to-turbine wake interactions that were found for offshore
wind turbines with an optimal spacing of nine diameters, 9D in a topology representing hexagonal arrays. This significance can be seen here when comparing the annual energy production and capacity factor for a 15 MW turbine (1.08 GW farm size) for 8D (5720 GWh/yr, 60.4%), 9D (5765 GWh/yr, 60.9%) and 10D (5784 GWh/yr, 61.1%) respectively. Similarly for a 22 MW turbine (1.23 GW farm size) for 8D (6347 GWh/yr, 58.8%), 9D (6396 GWh/yr, 59.2%) and 10D (6415 GWh/yr, 59.4%) respectively. Results highlight that to realistically capture annual energy production outcomes, downstream flow variations caused by rotor average models must be accounted for in modelling, and that wake expansion significantly influences induction under rated thrust conditions.

Integrated into this study are submarine power cable specifications namely, rated voltage 132 kilo-volts, cross-linked polyethylene insulation, high voltage alternating current, three-phase three-core design in trefoil formation. The design used COMSOL Multiphysics numerical modelling to pinpoint the optimal conductor size as in this case, namely 300, 800 and 1000 mm^2 cross-sectional area found as optimal. These specifications are required to confirm that a
minimal levelized cost of energy (LCOE) can be achieved even at increased cable lengths when thermal and electrical characteristics are properly balanced. Copper results in superior performance in terms of efficiency, lower resistive losses, and a reduced temperature rise, even when carrying higher currents. Creating a transition from numerical modelling when determining LCOE by incorporating capital cost (CAPEX) broken down into several areas as
development and project management (2.6%), wind turbine (22.7%), balance of plant (29.7%), installation and commissioning (6.4%) as lifetime costs for floating offshore wind. In addition, operational expenditures (OPEX) encompassed as operations and maintenance (31.1%), decommissioning (2.6%), and contingency and insurance (4.7%) that are integrated throughout the operational life of the asset. Outcomes are determined by LCOE sensitive
scenarios in capacity factor (range 50 to 60%) and project lifetime (range 20 to 25 years) that tends to reduce LCOE by spreading costs over more generated energy and time, whereas higher discount rates (range 8 to 12%) increase LCOE by reducing the present value of future revenues (best option £88.79/MWh and worst case £120.84/MWh). Fulfilled by sensitivity scenarios, therefore the base case used to drive this study forward was to opt for LCOE for farm size, 1.08 GW (15 MW turbines) at £106.55/MWh and for farm size, 1.23 GW (22 MW turbines) at £105.83/MWh as the optimized outcome.

Date of Award	2025
Original language	English
Supervisor	Simon Neill (Supervisor)