Proposed algorithm

In order to resolve the issues mentioned in Subsection 6.2.4, an improved fast frequency re- sponse algorithm is proposed.

P

t (s)

Support

Recovery

Transition

Reconnection

Activation

Figure 6.10: A conceptual graph showing the holistic approach used in the proposed controller with its five different sequences: Activation, Support, Transition, Recovery and Reconnection.

The main idea is to deliver a holistic approach to fast frequency response as indicated in Fig- ure 6.10. The five phases or criteria described as:

• Activation criterion: Decides when to activate the proposed controller, a later activation results in a prolonged duration of steep values of d f/dt subjected to large generator trips. The impact of a prolonged duration with steep d f/dt values could cause anti-islanding relays based on d f/dt to disconnect production units further worsening the situation. The activation criterion also decides when to disconnect the MPPT controller of the turbine, to ensure that the requested power will be delivered even though the wind speed might change. This is from now on referred to as the locked operational window.

• Support phase: Delivers the kinetic energy from the turbine to the power system, simul- taneously lowering the aerodynamic efficiency of the wind turbine depending on the size of the kinetic energy used.

• Transition criterion: After the support phase, the transition is based on the frequency derivative to realize when to switch to the recovery phase.

• Recovery phase: Aims to recover the pre-disturbance rotor speed. Depending on the need of the system to withstand additional power imbalances, this can be prolonged or short- ened, alternating the change in recovery power. However, due to the efficiency loss (re- duced Cp(λ,β)) of the turbine during the support phase, the recovery phase energy will be larger than that of the support phase.

• Reconnection criterion: Once re-acceleration has taken place, switch back to MPPT in a controlled manner.

The proposed algorithm utilizes an f -independent support,∆ Pe, until the frequency nadir has occurred. The frequency nadir is identified by filtering the frequency through a first order trans- fer function with time constant Trs= 0.5 s. Followed by calculation of the derivative, filtering it with a time constant Td f = 0.2 s and then comparing the derivative to zero. This ensures that the

strategy delivers its support until the frequency nadir has occurred, referred to as the transition

criterion. The transition is slightly delayed due to the impact of previously mentioned filters (Trs,

Td f) and then transitions to a∆f &d f/dt-dependent algorithm, allowing the re-acceleration to follow the returning frequency swing. Due to interactions of dead bands, delays and surround- ing governor settings, this re-acceleration might not be sufficient. Therefore, a secondary signal is added called REAX calculated through

REAX = (ω0−ωwt)R

d f

dt (6.2)

whereωwt is the rotor speed of the turbine,ω0is the pre-disturbance rotor speed and R is a re- acceleration constant set to -50 determining the impact of the re-acceleration. The REAX signal is not activated until the re-acceleration pattern has begun, i.e. the frequency nadir has occurred. Furthermore, the REAX signal can be used to limit the duration away from MPPT operation.

f [Hz] Prs Dead band 2
2

1 Ties+1 1 Tdfs+1 < 0 Hz/s T/F T/F Pe df dt 1 Trss+1 Pe0 REAX + d
dt d
dt -2HwtK -2Hwt -2H -2HwtwtK * + Pe-min

Figure 6.11: Proposed algorithm with its schematics.

The controller is presented in Figure 6.11.∆Pe is the boost in active power in pu and Pe0is the pre-fault power from the VSWT,∆Pe−min is the minimum limit set for the recovery phase. In the control diagram in Figure 6.11, the upper part of the control diagram is responsible for the transitioning from f -independent to a ∆f &d f/dt-dependent re-acceleration. The lower part, similarly to Figure 6.5, deals with the re-acceleration together with the added REAX signal to ensure a rapid but smooth transition back to MPPT. This transition also deals with any variations

in wind speeds, affecting the time necessary for re-acceleration.

Controller to handle variable wind

As proposed in [126], it is important to evaluate FFR algorithms using a alternating wind speed. The paper manages this with the help of an added controller algorithm. The algorithm to deal with wind variations is implemented in the proposed algorithm. However, [126] was imple- mented on a f -independent algorithm, hence the transition to recovery was rather fast. In order to avoid power bursts and allow the controller to deal with wind fluctuations, an improved con- troller is presented in Figure 6.12.

d
wt/dt < 0 Initiate at t = t0 Aquire P_e0,
_wt0 Disconnect MPPT & set Pe = Pe0+Pe Pe(t)-Pe0 < -0.05 pu Yes No
wt0 =
wt t = t0 + tmax No No Break, re-connect to MPPT Yes Yes Transition criteria Yes Yes No

Figure 6.12: The proposed controller flow used in order to manage variable wind conditions when providing any controllable inertia response from a wind farm.

The main inputs are rotor speed, electric power output from the turbine and the timings for the de-acceleration and re-acceleration of the CIR.

The functionality of the proposed controller is described below. Assume that the CIR is acti- vated. Once the activation criterion is fulfilled, pre-fault power, Pe0, and pre-fault rotor speed,ωwt0, are stored. The power for the support phase is delivered, irrespective of any changes in wind speed, i.e. mechanical power. Once the transition criterion is fulfilled, the algorithm investigates three main criteria and returns to MPPT (using a transition ramp rate limiter for 5 s.) if any of them are true:

1. If rotor speed has returned to pre-fault value, this could occur if an increase in mechanical power has occurred during the support phase sufficient to counteract the change in electric power,∆Peor if after sufficient time of acceleration the criterion is met.

2. If sufficient recovery power, set to -0.05 is reached and the rotor speed derivative still is negative, this means that a significant wind reduction has occurred and the VSWT is better off working at MPPT. A filter on the rotor speed derivative allows for a temporary wind decrease of the turbine while still operating within the locked operational window; as shown in Figure 6.9 these variations would not be very significant due to the small changes in wind speed from second to second. Hence, this filter is designed as a first order transfer function with a time constant of 1 s which reduces the number of accidental breaks from the locked operational window.

3. If too much time has passed, tmax, and the recovery phase is unsuccessful in re-accelerating the rotor to pre-fault rotor speed, ωwt0. This would occur if the decrease in mechanical power balances out the recovery power, hence, keeping the rotor speed derivative positive but not enough to accelerate it back toωwt0.