Doctor of Philosophy / Department of Electrical and Computer Engineering / Behrooz Mirafzal / Voltage source inverters (VSIs) are an essential interface for grid integration of renewable energy resources. Grid-tied VSIs are employed in power grids to integrate distributed generation units, e.g. photovoltaic arrays, wind turbines and energy storage units, to the utility and extract the maximum energy from the DG units in an efficient manner. However, the stability of VSIs and by extension the entire DG system can be degraded under abnormal grid conditions. In this dissertation, new control and switching techniques for stability and power quality improvement of grid-tied VSIs under abnormal grid conditions are presented. For grids with a low inertia and a low short-circuit ratio, commonly referred to as weak grids, grid connection may make VSIs susceptible to voltage distortion and instability. In this dissertation, through root locus analysis of a detailed state-space model, the design of several circuit and control parameters of the grid-tied VSI are evaluated for improving stability in weak grids. It is shown that grid-side filter inductances can be increased for stable operation of VSIs in weak grids. Accordingly, a virtual inductance emulating the effect of an increased inductance in the grid-side filter is developed in this dissertation, which enables stable operation of VSIs in weak grids without the tradeoffs, i.e. additional voltage drop, increased cost and larger size, associated with a larger inductor. The virtual inductance scheme is realized through the injection of a feedforward current element in the VSI controller through a gain component. The measured grid currents, which are sensed for regular VSI controller operation, are employed as the feedforward component eliminating the need for any additional sensors for the utilization of this control scheme. Furthermore, a direct model reference adaptive control (MRAC) scheme is employed in this dissertation to tune the virtual inductance gain block according to a stable reference model for varying grid conditions. The use of direct MRAC scheme allows tuning of the virtual inductance block without the need for a plant parameter estimation stage. The virtual inductance scheme enables stable operation of VSIs in weak grids without system parameter redesign, thereby maintaining the steady-state performance of the system. The efficacy of the virtual inductance feedforward scheme is verified through hardware tests carried out on a three-phase grid-tied experimental setup. Along with extracting energy from the DG sources, grid-tied VSIs are capable of providing various ancillary services to the utility under abnormal conditions. However, providing ancillary services could drive the inverter voltages beyond the linear modulation region resulting in grid current distortions, which could violate the requirements for grid integration of DGs. An atypical pulse width modulation (PWM) technique is proposed in this dissertation, which maximizes the dc-bus utilization of VSIs, which in turn enables the VSIs to supply the maximum extracted power from the DG units to the grid when providing ancillary services while operating in the linear modulation region. The switching scheme is realized by injecting common mode components in the PWM references, computed based on instantaneous reference magnitudes. The proposed scheme is suitable when providing both symmetrical and asymmetrical ancillary services. In this dissertation, negative-sequence compensation and harmonic compensation are employed as instances of symmetrical and asymmetrical ancillary services. The proposed scheme can be integrated with any control scheme and carrier-based PWM combinations. The efficacy of the proposed atypical PWM scheme is verified through both simulation and hardware tests.
Identifer | oai:union.ndltd.org:KSU/oai:krex.k-state.edu:2097/39329 |
Date | January 1900 |
Creators | Adib, Aswad |
Source Sets | K-State Research Exchange |
Language | en_US |
Detected Language | English |
Type | Dissertation |
Page generated in 0.002 seconds