Network Modeling Influence on Small-Signal Reduced-Order Models of Inverter-Based AC Microgrids Considering Virtual Impedance

The development of accurate and computational-effective microgrid models is crucial to assess the performance of microgrids. Therefore, in this paper, three microgrid reduced-order (R-O) models including a virtual impedance representation are explored from accuracy and computational burden point of view. The main difference between R-O models is the network modeling technique, so three alternatives are considered: i. A complete network dynamics (CND) representation which consists of including the network states; ii. A network approximation (NAP) based on Taylor series and iii. A static representation (without network dynamics-WND). A complete small-signal microgrid model including all the system states is also developed to assess the accuracy of each R-O model. Thus, the accuracy of each R-O model is evaluated under different network and virtual impedance conditions, based on the following analyses: i. The eigenvalues; ii. The parametric stability boundary; and iii. The relative error of the damping factor and natural frequency. As a result, the limitations introduced by each network modeling technique in the overall microgrid model are established, and new insights about the virtual impedance influence on microgrid dynamic behavior are pointed out. Finally, the computational burden of each R-O model is assessed.