In response to climate change and sustainability challenges, various incentive programs have promoted solar photovoltaic (PV) economic feasibility and interconnection into the low voltage electrical distribution system. However, the existing power system infrastructure can only accommodate a limited amount of PV generation before reverse power flow (PV generation flowing back into the distribution network) becomes an issue. This limits the ability to achieve Zero Net Energy (ZNE) behind individual meters and in whole communities since ZNE communities typically require large PV deployments. In parallel, district-level energy systems, such as Advanced Energy Communities (AEC) microgrids— electrically contiguous areas that leverage the clustering of load and generation by integrating multiple utility customer-owned Distributed Energy Resources (DER) — that include battery storage, offer a great prospect for integrating high levels of solar PV into the built environment.

Designing the least-cost and technically feasible system to serve a district load has been a challenge to utilities and city planners. Battery energy storage and smart-inverter technologies emerge in this context to enable higher penetration of solar PV by locally regulating voltage and controlling active and reactive power flows. However, there is no straight forward way nor a practical rule or consensus on how to size such assets optimally.

This work proposes a Mixed Integer Linear Program (MILP) optimization to determine the least-cost DER portfolio consisting of inverter-connected solar PV and battery storage. It also allocates it in a multi-node electrical grid and dispatches it considering the existing electric infrastructure limits (transformer capacities and nodal voltage magnitudes). Novel linearization techniques such as polygon relaxations are used to limit otherwise non-linear apparent power flows at transformers. A novel Alternating Current (AC) decoupled linearized power flow is also integrated into the MILP optimization. Moreover, for the first time, smart-inverter droop-control functions are included in the DER optimal allocation problem. Results show that such comprehensive MILP is achievable and tractable for a 115-node AEC.