Offshore wind farm design increasingly faces the challenge of adapting layouts to larger turbines as technology advances. A promising approach is down-selection, where layouts designed for smaller turbines are adapted to higher-rated machines by selecting a subset of positions. This thesis investigates two central questions: which optimisation strategies are suitable for the down-selection problem, and what is the impact of down-selection on the final wind farm layout. The first question was addressed through a structured evaluation of candidate algorithms, including a literature-based ranking, parameter tuning, and comparative testing across multiple cases. The analysis showed that Gradient-Based methods and Greedy Heuristics were the most effective strategies, with complementary strengths: Gradient-Based approaches offered scalability and computational efficiency, while Greedy Heuristics achieved higher energy yields under different occupancy conditions. The second question was addressed by comparing down-selected farms with layouts directly optimised for larger turbines. Across all test cases, the down-selected layouts achieved annual energy productions within 0.15% of the optimised layouts, showing that down-selection can closely replicate optimal performance. The small residual differences were governed primarily by spacing constraints, with compatibility between initial and final requirements leading to near-lossless performance. Down-selection also influenced turbine distribution, with more retained along site perimeters, which may affect secondary design drivers such as cabling or support structures. Overall, the results demonstrate that down-selection is a viable and efficient design strategy, provided that algorithm selection and parameter tuning are carefully matched to the problem context, and that future spacing requirements are anticipated at the design stage. These insights underline the potential of down-selection to reduce computational and economic costs in wind farm development while maintaining energy yield, and point towards its integration within robust design frameworks for future turbine upgrades.