A modified walrus optimization algorithm for multi-objective sizing and excess energy management of hybrid microgrid in coastal communities with excess-driven freshwater desalination

• Proposed a novel hybrid microgrid co-optimizing electricity and freshwater supply. • Developed a novel modified Walrus Optimization (mWO) for robust multi-objective sizing. • Introduced an excess energy management strategy using reverse osmosis desalination (ROD) as a flexible load. • mWO achieved statistically superior convergence over 18 benchmarked algorithms. • Optimal PV/WT/DsG/BAT/ROD minimized energy cost to 0.1714 $/kWh with 71.9% renewables. Energy–water nexus optimization is crucial for off-grid coastal communities, where freshwater production is energy-intensive and renewable resources are intermittent. In stand-alone hybrid renewable energy systems (HRESs), temporal mismatches between variable generation and demand often cause surplus energy curtailment. Integrating reverse-osmosis desalination (ROD) with HRES offers a sustainable solution by utilizing the ROD plant as a flexible load for excess renewable output. This study proposes a stand-alone hybrid microgrid for a coastal community in Yanbu, Saudi Arabia, designed for joint energy–water optimization. The system integrates solar photovoltaic (PV), wind turbines (WTs), battery storage (BAT), and a diesel generator (DsG) to ensure a reliable power and freshwater supply. System sizing is performed using a novel modified Walrus Optimization (mWO) algorithm that mitigates premature convergence and local optima issues of the original WO. The mWO incorporates leader-based mutation–selection (LBM), orthogonal learning (OL), and an adaptive exploration–exploitation controller for enhanced search balance. The objective function minimizes the levelized cost of energy (LCOE), loss of power supply probability (LPSP), and unutilized excess energy under operational constraints. The mWO is rigorously validated on benchmark functions and compared with the original WO and seven recent metaheuristics, showing faster and more stable convergence. Applied to the Yanbu case study and benchmarked against 18 algorithms, it achieves statistically superior performance. The optimal PV/WT/DsG/BAT/ROD configuration achieves an LCOE of 0.1714 $/kWh, a cost of water of 0.7715 $/m 3 , a 71.9% renewable fraction, and 1,500 tons/year of emissions. The results confirm that mWO-driven co-optimization offers a robust, low-cost, and sustainable pathway for autonomous coastal desalination microgrids.