Grid fault ride through for wind turbine doubly-fed induction generators

Abstract

Wind farms must contribute to the stability and reliability of the transmission grid, if they are to form a robust component of the electrical network. This includes providing grid support during grid faults, or voltage dips. Transmission system grid codes require wind farms to remain connected during specified voltage dips, and to supply active and reactive power into the network. Doubly-fed induction generator (DFIG) technology is presently dominant in the growing global market for wind power generation, due to the combination of variable-speed operation and a cost-effective partially-rated power converter. However, the DFIG is sensitive to dips in supply voltage. Without specific protection to 'ride through' grid faults a DFIG risks damage to its power converter due to over-current and/or overvoltage. Conventional converter protection via a sustained period of rotor-crowbar closed-circuit leads to poor power output and sustained suppression of the stator voltages. This thesis presents a detailed understanding of wind turbine DFIG grid fault response, including flux linkage behaviour and magnetic drag effects. A flexible 7.5kW test facility is used to validate the description of fault response and evaluate techniques for improving fault ride-through performance. A minimum threshold rotor crowbar method is presented, successfully diverting transient over-currents and restoring good power control within 45ms of both fault initiation and clearance. Crowbar application periods were reduced to 11-16ms. A study of the maximum crowbar resistance suggests that this method can be used with high-power DFIG turbines. Alternatively, a DC-link brake method is shown to protect the power converter and quench the transient rotor currents, allowing control to be resumed; albeit requiring 100ms to restore good control. A VAr-support control scheme reveals a 14% stator voltage increase in fault tests: reducing the step-voltage impact at fault clearance and potentially assisting the fault response of other local equipment.