Abstract
Distributed energy systems (DES) have increasingly become a viable way to integrate more renewable resources while combating highly volatile energy costs. They include small-scale energy generation and storage resources which are located near consumers. Ensuring that grid-connected DES are designed to meet consumer needs, while operating symbiotically with existing alternating current (AC) power networks, is imperative to their wider implementation. Although DES design has been studied extensively using optimisation models, these have either excluded or oversimplified nonconvex AC power flow constraints that underpin electric power networks. Existing studies that incorporate AC optimal power flow (OPF) within DES design frameworks have overlooked the fact that electricity distribution networks are largely unbalanced due to their radiality, i.e., the voltage and current magnitudes and their respective angles across each phase are unequal. This thesis aims to investigate the implications of including detailed and appropriate power flow constraints on the DES designs proposed by optimisation models.
The thesis presents novel optimisation frameworks to obtain DES designs that can feasibly operate within existing distribution networks, whether balanced or unbalanced. It does so by consolidating OPF and multiphase optimal power flow (MOPF) formulations with DES design. The inclusion of MOPF ensures that resulting DES designs are compatible with unbalanced distribution networks, which is another novelty of this work. The proposed algorithms employ pragmatic decompositions and exploit existing commercial solvers to find feasible and locally optimal solutions for the resulting large-scale mixed-integer nonlinear programming (MINLP) problem. A framework for obtaining discrete DES designs, which includes electrified heating and storage technologies while simultaneously considering MOPF constraints, is also presented.
Comparisons with existing frameworks illuminate the necessity of considering detailed and appropriate power flow constraints when designing DES. Not doing so could lead to ill-suited designs, resulting in greater costs for consumers and reduced renewable energy generation. The proposed optimisation algorithms find feasible solutions where the state-of-the-art solver for nonconvex MINLPs fail to do so. Overall, the contributions of this thesis inform wider implementation of DES to support current decarbonisation efforts.