Facility energy management remains a critical challenge for modern smart buildings because energy-intensive assets such as heating, ventilation, and air-conditioning systems, lighting networks, plug loads, ventilation systems, and operational equipment must be controlled without compromising occupant comfort, operating cost, carbon performance, or equipment reliability. This paper proposes a novel Smart Facility Sustainability Optimization Algorithm (SFSO) for energy-efficient facilities management using Internet of Things sensor data and predictive analytics. The proposed framework integrates smart meters, occupancy sensors, indoor environmental sensors, equipment-status data, weather variables, electricity tariff signals, renewable-energy availability, and carbon-intensity indicators into a predictive optimization architecture. A Long Short-Term Memory forecasting model was used to estimate short-term facility energy demand and achieved a mean absolute error of 8.91 kWh, root mean square error of 12.76 kWh, mean absolute percentage error of 4.37%, and coefficient of determination of 0.966. The proposed SFSO algorithm was comparatively evaluated against Genetic Algorithm and Particle Swarm Optimization using energy saving, operating-cost reduction, carbon-emission reduction, comfort violation rate, peak-load reduction, maintenance-risk reduction, convergence speed, computational time, and normalized fitness value as performance indicators. Simulation-based benchmark results showed that SFSO achieved 26.78% energy saving, 32.37% operating-cost reduction, 31.87% carbon-emission reduction, 28.76% peak-load reduction, and 31.60% maintenance-risk reduction. It also recorded the lowest comfort violation rate of 2.10%, the best normalized fitness value of 0.501, and convergence within 48 iterations. The results indicate that SFSO provides a more adaptive, sustainability-aware, and computationally efficient optimization approach than conventional GA and PSO for smart facility energy management.