In a deregulated power systems reactive power support service is procured from private reactive power sources through ancillary service markets that are monitored by an Independent System Operator (ISO). However the implementation of reactive power ancillary service market faces several technical challenges due to localised nature of reactive power and the impact of active power markets on it. Economical challenges like exercising of market power and price volatility also exist in reactive power markets. These challenges have impaired the implementation of a real time reactive power market inspite of its relevance in maintaining bus voltage within the permissible limits. We addresses these issues by proposing value based reactive power market mechanism that considers local nature of reactive power, inherent coupling of active reactive power requirement, and can curtail price volatility and mitigate market power.



Localised nature of reactive power and its inherent coupling with active power is considered in the design of market mechanism by proposing a novel relative electrical distance measure and a relevance factor for each reactive power source in the network. Based on the relevance factor a value function is proposed for reactive power considering voltage support, reserve requirement and spatial relevance of each source. Then a value-based reactive power market (VBRPM) under perfect competition is formulated. For imperfect competition, a three stage reactive power market structure with a single leader multi follower game based pricing mechanism is proposed. The existence of equilibrium is proven in the single leader multi-follower game. Conditions for incentive compatibility and individual rationality are derived which ensures the economical efficiency of the proposed mechanism. The thesis also presents a pioneer work in optimal bidding strategy formulation for participants in reactive power market based on deep reinforcement learning technique. For learning optimal bidding strategies a variant of Neural Fitted Q-iteration (NFQ), i.e. NFQ with Target network and Prioritized Experience Replay is proposed (NFQTP).