Load Frequency Control for Electric Water Heater Using DBSCAN Algorithm in real-time

Electric grid frequency stability requires precise balance between supply and demand, a challenge intensified by increasing renewable energy penetration and fluctuating consumption patterns. This thesis develops and validates a novel control strategy for residential water heaters to provide demand-side frequency regulation while maintaining customer comfort. The approach employs Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm to dynamically group water heaters based on their thermal state and operational flexibility, enabling coordinated response to grid frequency deviations. The methodology integrates real-time grid frequency monitoring with thermal modelling of domestic hot water systems. A dynamic control system monitors and manages water heater operations across a simulated population of residential households, responding to frequency signals while enforcing strict temperature constraints. The DBSCAN clustering algorithm groups heaters with similar operational characteristics, allowing selective control that balances grid support with customer comfort requirements. Simulation results demonstrate that the proposed strategy reduces root mean square (RMS) frequency deviation by 62% compared to uncontrolled scenarios, while maintaining water temperatures above the minimum comfort threshold of 55°C in over 99% of cases. The clustering-based approach outperforms simple threshold-based control methods by 40% in frequency regulation performance while generating 15% fewer switching operations per heater. Peak demand is reduced and energy consumption profiles are smoothed without compromising hot water availability. Comparative analysis shows the DBSCAN-based methodology offers superior scalability and computational efficiency compared to existing Load Frequency Control (LFC) approaches. The thesis addresses practical implementation challenges including communication latency, data integration with existing infrastructure, and user comfort. Economic and environmental benefits are quantified, demonstrating potential for significant energy cost savings and carbon emission reductions at scale. Future research directions are identified including advanced control algorithms, integration with other Demand Response (DR) resources, and field validation through pilot deployment.