Accuracy of phase-decoupled and phase-coupled distribution grid power flow models

The complexity and computational burden of nonlinear power flow (PF) models have motivated the introduetion of various approximations. While these approximations provide tractability and reduced computational burden in PF, for optimal power flow (OPF) and control purpose, the accuracy of these models becomes a major concern in the presence of high renewable energy penetration/increased demand in distribution feeders. This paper compares and analyzes, for the first time to our knowledge, the performance of commonly used PF models in OPF formulations for single-phase and unbalanced threephase distribution systems. Simulations performed on the phasecoupled and phase-decoupled versions of the IEEE 123-node and a reduced model extracted from the IEEE 8500-node feeder with varying levels of negative and positive net loads show that, ignoring mutual coupling impedances, as in the phase-decoupled PF models, is the largest source of error on the voltage profiles. Also, the performance of the linear PF models is acceptable only for a small range around the nominal operating point; hence, they can lead to large errors if used in OPF/control framework that seeks solutions on wider operating range.