The nature of double-fed induction generators for frequency and voltage support from wind turbines
This research improves the understanding and utility of power electronic controlled grid-connected wind turbines by studying their physical capabilities. The primary topic is of inertia and capability for frequency response. Frequency response characteristics of these turbines are not well understood and there is opportunity to improve the art of generator control to make better use of the resource. More specifically, it is hypothesized that physical inertia of the massive power electronic-coupled rotor hub assembly can provide a naturally stable fast frequency response. This is contrary to popular belief that wind turbines can only provide emulated inertia with methods like frequency-power droop control. This work first derives the estimated potential for load transient response considering the rotor mass and without control influence. A new low-inertia approach, in which the grid frequency is assumed to be non-constant, proves that sufficient inertia and response capability exists. It shows that even though a wind turbine rotor is indirectly coupled to the grid via power electronics, there is still an inherent frequency response provided by nature of the system dynamics, even if not intended. Wind turbine controllers in use today work to emulate the effects of synchronous inertia on frequency response, but they do not use the capability in a natural and fulfilling way. In fact, today's state-of-the art controllers are already understood to be insufficient for a 100% wind-powered system, by way of practice and observation. What is lacking is an ability to provide adequate fast frequency response for regulation during transient. To that point, it is shown here that double-fed induction generator wind turbines without well-designed frequency control can actually have a temporary load-rejecting response. Here, the connection of physical inertia and control design to frequency response is quantified with transfer functions derived for the wind turbine electromechanical and electromagnetic systems.
Using the non-constant frequency assumption, this work then proposes a generator PE controller to correct and prescribe the frequency response; voltage response is designed in a similar way. It modifies the dynamic properties and affords the ability to provide adequate frequency response from physical inertia. The controller adds frequency- and voltage-responsive transient-only components to the already-existing generator qd current commands of steady-state torque and reactive power controllers, respectively. Control parameters are designed using the DFIG dynamic model to place poles and zeros of the respective transfer functions. In this way, response is predefined and stability is assured with a prescribed control action. The effect is faster frequency regulation (approximately 1.5x faster) with higher nadir and arrested frequency when compared to the current art of droop control. Additionally, the duration of load-support response is limited only by the stored energy capacity of the rotating mass. With the proposed controller, wind turbine inertia is proven here to be capable of providing all of the necessary transient frequency response, even during severe load change and temporary generator overload for a duration of more than 500 ms and small load change for beyond 15 seconds. This type of performance has not previously been achieved. The added transient response from wind turbines can relieve burden currently placed on other generators and auxiliary frequency-support devices. Tunable control design affords utility of wind turbine inertia for a variety of needs and conditions. Power system operation and planning can be impacted by use of the new inertia resource, potentially reducing the amount of other inertial capacity that is currently required to maintain reliability.
A secondary topic is of reactive power and capability for voltage support. DFIGs have historically been used with grid-connected stator windings. A new method using grid-connected rotor windings shows potential for improved efficiency, and thus suggests an ability for different reactive power capability too. In this dissertation, the capability for reactive power support considering generator and converter nameplate ratings and electrical parameters is modeled. It finds that some generators benefit from the new configuration by having increased reactive power generation capability due to reduced var loss in the generator windings. Nameplate current ratings can also be used more completely and this also increase var capability. Reduced var loss in the DFIG can enable other generators to reduce their reactive output. Equations are derived to estimate the reactive power capabilities in the new configuration and experiments prove the potential. Added reactive power from wind turbines can mean a reduced need for auxiliary voltage support components like static var compensators. Together, added frequency and voltage support capability can improve the value of wind turbines. The enhancements to wind turbine utility services can help support a more complete transition to renewable-fueled electric power systems.