Application of voltage sensitivity analysis in a probabilistic context for characterizing low voltage network operation

Abstract

The massively dispersed nature of power distribution networks and their current unobservability will drive distribution utilities to hierarchize the integration of automation in their systems. Given the complete lack of real-time information in Low Voltage (LV) networks, state estimation techniques specifically tailored for these systems will not come for the foreseeable future. Considering delayed Smart Meter (SM) recordings as the only current source of information in LV networks, this paper presents a probabilistic method that uses sensitivity analysis and quarter-hourly SM measurements for characterizing and setting boundary values for LV network operation indices. Such information can be useful in the preprocessing phase of state estimation techniques focusing on the Medium Voltage (MV) or on the LV (in a later phase) level. The proposed method is applied for analyzing a real LV feeder and its outputs are compared to the ones of a deterministic direct sensitivity analysis method, whose accuracy has been previously demonstrated, as well as to the ones of a probabilistic Monte Carlo (MC) simulation.