Abstract
Abstract Wind farm layout optimization (WFLO) often utilizes engineering wake models where wake decay coefficients serve as a proxy for atmospheric stability. However, the sensitivity of optimal topologies to these stability assumptions and their impact on electrical infrastructure remain largely unquantified. This study employs a Gaussian wake framework and gradient-based optimization to evaluate high-density turbine layouts across three atmospheric stability regimes, statistically validated across 10 random initializations. Results indicate that mean optimized AEP ranges from 293.35 to 348.08 GWh across stability regimes, while optimization yields divergent geometric strategies, most notably a perimeter-loading logic under stable conditions. Cross-evaluation reveals a 3.5% energy penalty in stability-blind designs. Notably, internal cable lengths varied by less than 2% across configurations despite significant layout reconfiguration, suggesting that aerodynamic wake mitigation can be decoupled from infrastructure planning in space-constrained domains. Stable conditions further require approximately 17% more iterations to converge, indicating a more complex objective landscape. These findings provide a controlled benchmark for stability-aware layout frameworks.