This paper presents a new methodology for integrating multiple disconnected and irregular boundary constraints into wind farm layout optimization. The method uses polygon representations of boundaries and calculates analytical gradients of distances from wind turbine locations to these boundaries. This allows for a continuous optimization formulation. A key challenge is that wind turbines may be placed within the nearest polygons during initial optimization steps, leading to suboptimal turbine allocation. To address this, two strategies are proposed: boundary relaxation and a heuristic "smart-start" algorithm. The study demonstrates the effectiveness of these strategies in optimizing a wind farm layout to maximize annual energy production (AEP) in a site with five irregular inclusion zones. The results show that combining boundary relaxation with a gradient-based solver achieves an average +10.2% increase in AEP compared to a baseline approach. The smart-start algorithm, when combined with a gradient-based solver, achieves an average +20.5% increase in AEP compared to the baseline and +9.4% compared to the relaxation strategy. The methodology is implemented in the open-source software TOP-FARM, which uses PyWake for AEP calculations and wake modeling. The case study highlights the importance of considering complex boundary constraints in wind farm design and demonstrates the benefits of using relaxation and heuristic initialization techniques to improve optimization outcomes. The results show that these strategies significantly enhance the quality of wind farm layouts by allowing better exploration of the design space and more efficient turbine allocation.This paper presents a new methodology for integrating multiple disconnected and irregular boundary constraints into wind farm layout optimization. The method uses polygon representations of boundaries and calculates analytical gradients of distances from wind turbine locations to these boundaries. This allows for a continuous optimization formulation. A key challenge is that wind turbines may be placed within the nearest polygons during initial optimization steps, leading to suboptimal turbine allocation. To address this, two strategies are proposed: boundary relaxation and a heuristic "smart-start" algorithm. The study demonstrates the effectiveness of these strategies in optimizing a wind farm layout to maximize annual energy production (AEP) in a site with five irregular inclusion zones. The results show that combining boundary relaxation with a gradient-based solver achieves an average +10.2% increase in AEP compared to a baseline approach. The smart-start algorithm, when combined with a gradient-based solver, achieves an average +20.5% increase in AEP compared to the baseline and +9.4% compared to the relaxation strategy. The methodology is implemented in the open-source software TOP-FARM, which uses PyWake for AEP calculations and wake modeling. The case study highlights the importance of considering complex boundary constraints in wind farm design and demonstrates the benefits of using relaxation and heuristic initialization techniques to improve optimization outcomes. The results show that these strategies significantly enhance the quality of wind farm layouts by allowing better exploration of the design space and more efficient turbine allocation.