Reduced-Order Modeling and Analysis of Droop-Controlled, Inverter-Based Distributed Generation Networks

摘要

As the deployment of inverter-based distributed generation (DG) becomes more widespread throughout the distribution system, dynamic modeling is taking on increased importance for ensuring that the behavior of the system, possibly containing thousands of inverter-based sources, remains both stable and well-regulated. Thus, there is a need for dynamic models of large DG networks that are accurate, computationally-efficient, and analytically insightful.;The major objective of this work is to develop and validate accurate, reduced-order, nonlinear models of grid-forming, droop-controlled, inverter-based DG networks in order to achieve accurate reproduction of the nonlinear network dynamics from first principles. Harmonic and spatial nonlinear order reduction methods are investigated.;To enable harmonic order reduction, a multiple-harmonic dynamic phasor modeling tool has been developed that allows convenient construction of nonlinear phasor-domain models of inverter networks. Using these dynamic phasor models, rigorous time-domain and modal analysis is performed in order to evaluate the influence of higher-order harmonics on grid-forming, droop-controlled inverter dynamics. Results from this analysis are used to determine the circumstances under which certain higher-order harmonics merit representation in a reduced-order dynamic phasor network model. Multiple-harmonic models are used to study the impact of switching harmonics, unbalanced excitation, network asymmetry, dc offsets, and 2nd harmonic on the behavior of the network under a range of inverter control parameter values.;In a separate but complementary approach, coherency-based aggregation is investigated as a means of spatial order reduction for grid-forming, droop-controlled inverter networks. The resulting aggregated reduced-order models are nonlinear, and can provide rapid analysis of large-signal transients and greater physical insight into large-scale network behavior. A rigorous coherency identification technique is adapted to grid-forming, droop-controlled inverter networks. The impact of network characteristics and droop control parameters on inverter coherency is investigated.;In keeping with the systems-oriented perspective of this work, a secondary objective has been to investigate a class of `model-preserving' modifications to the basic droop controller that provide useful improvements to individual inverter behavior, but preserve certain desirable properties of the slow-time-scale, reduced-order network dynamic model. Investigated modifications include coherency enforcement, higher-order power measurement filters, and symmetric droop control for hybrid ac/dc droop-controlled networks.