A data-driven approach for maximizing solar PV capacity at the distribution feeder level under existing operational paradigms

Recent rapid growth in solar photovoltaic (PV) marks a shift away from conventional generation, providing a strategy for stemming carbon emissions emanating from the electricity sector. However, solar PV often appears within distribution systems where existing infrastructure was designed under a central-station paradigm. Fortunately, smart grid technologies can aid integration of distributed energy resources through better characterization of how these resources affect the grid. Using a New York utility service territory as a test bed, we present a data-driven Monte Carlo framework that estimates the maximum installed solar PV capacity at the distribution feeder level subject to existing network constraints. Working with representative days that closely match a feeder's load profile, we probabilistically select PV systems according to current New York trends and stochastically model hourly electricity generation. We found 262,318 kW of solar PV could be added across the entire utility service territory meeting 14.14% of electricity demand.