Development of Optimal Distributed Resource Utilisation Approaches Under Connectivity Constraints for Energy Internet

Abstract

The global transition to sustainable power generation has driven a rapid increase in Distributed Energy Resources (DERs), especially from renewable sources. When integrated with energy storage, these localised generation units form microgrids capable of managing supply and demand. Although storage enables time-shifted use of intermittent renewable power, its capacity limitations and fluctuating loads cause energy curtailment, creating spatially distributed dynamic surpluses and deficits. Peer-to-Peer (P2P) power trading addresses the balancing of this by enabling direct local energy exchange, improving utilisation, reducing costs, and enhancing market resilience in interconnected microgrid communities. While full peer connectivity is infrastructure-intensive, sparse connectivity—where indirect exchanges occur via intermediaries through an Energy Internet (EI)—offers a more scalable and cost-effective solution. Thus, P2P power trading via EI raises the challenge of optimising the utilisation of spatially distributed generation under connectivity constraints. This thesis addresses this challenge by modelling realistic microgrid behaviour using multiple real-world electrical load datasets. Initially, internal power scheduling within a grid connected microgrid equipped with solar generation and storage is formulated as a mixe dinteger nonlinear optimisation problem, later relaxed to a mixed-integer linear formulation to reduce computational complexity. Predictive scheduling is employed to enable time-shifted energy usage. The resulting surplus and deficit data form the basis for simulating P2P power exchange within a connected community. First, this thesis designs the Connectivity and Preference-constrained Hop-regulated P2P Trading (CPHPT) approach to evaluate trading under constrained infrastructure. CPHPT models P2P trading as a linear optimisation problem that schedules energy exchange along shortest paths while respecting capacity and predefined hop limits. The internal microgrid scheduling and inter-microgrid trading are coordinated using a distributed control architecture, enhancing scalability and preserving data privacy. CPHPT provides optimal market clearance by performing P2P subscriber matching under connectivity, hop length and trading preference constraints. The sparse connectivity and P2P power exchange via EI is modeled using Graph theory, which is also leveraged to avoid explicit route computation during the hop-constrained P2P scheduling. The P2P power exchange model assumes the availability of dedicated low voltage DC connectivity infrastructure, including ideal power routers, converters, and smart metering, without accounting for investment costs; consequently, the reduction in infrastructure cost reflects only the reduction in the length of P2P connecting links and does not account for the costs of other required components. Theoretical analysis demonstrates that while full connectivity enables the maximum P2P power exchange, increasing the allowable hop count in sparsely connected communities enables near-optimal performance, albeit with higher routing complexity. The total P2P power exchange is chosen as the performance metric and full connectivity is used as the baseline case for comparison against varying levels of connectivity. The performance and scalability of the CPHPT approach is verified with ablation studies performed on upto 100 house communities simulated using real-world electrical datasets. Next, an Optimal Multi-Path Power Routing (OMPR) algorithm is developed to route the individual P2P exchanges in the community, determined from the results of subscriber matching through CPHPT. OMPR uses graph-theoretic principles to assess the connectivity structure and deterministically identify all feasible routes between peers. The power scheduling among the routes is formulated and solved as both a linear and a nonlinear multi-path power scheduling optimisation problem. The OMPR approach divides each multi-hop P2P exchange into multiple individual hop exchanges and solves each step-by-step. While routing multiple concurrent P2P exchanges, the order in which each P2P exchange is routed affects the optimal solution, as each route increases the power flow routing constraints. To overcome this limitation, a Hop Optimised Multi-Exchange Routing and Scheduling (HOMERS) algorithm has been developed to solve all concurrent multi-hop P2P exchanges to obtain the optimal routing paths. HOMERS formulates the routing and scheduling of all concurrent P2P exchanges into a single-step mixed-integer nonlinear programming optimisation problem. This approach efficiently identifies all feasible routes and schedules each power exchange, ensuring conflict-free power flow from the source to the destination in the predetermined number of hops. HOMERS is compared against Dijkstra’s algorithm, which was modified to eliminate power flow violations for performance evaluation, selected as a baseline. The 68.29% reduction in the mean number of hops, and 65.27% reduction in mean runtime compared to the baseline case in a 400 house community shows the scalability of the HOMERS approach. Finally, recognising the limitations of assuming uniform node distribution and connectivity in the CPHPT and random connectivity for OMPR, and HOMERS models, the thesis proposes a more realistic framework named the Sparse Connectivity Approach for Maximum P2P Power exchanges (SCAMPP) in Energy Internet. SCAMPP accounts for non-homogeneous node spacing and variable power generation capabilities across the community, similar to the real world scenario. It identifies optimal connectivity topologies based on geographic separation and resource distribution, yielding improved connection efficiency. The resulting community-specific connectivity model from SCAMPP is shown to have better connectivity utilisation and has improved efficiency compared to the ideal full connectivity. The case study with 50% PV integration resulted in a 98.91% reduction in total infrastructure length and a 76.15% improvement in infrastructure utilisation compared to the full connectivity used as baseline. Through the development of CPHPT, OMPR, HOMERS, and SCAMPP, this thesis substantially contributes to the research on distributed energy systems and P2P power trading. The integration of predictive scheduling, hop-constrained routing, and spatial-connectivity modelling offers a comprehensive and scalable framework for the future deployment of Energy Internet architectures. The research establishes a practical and theoretically grounded foundation for resilient, intelligent, and energy-efficient microgrid communities