International Journal of Power Electronics and Drive System (IJPEDS)

Vol. 16, No. 2, June 2025, pp. 1356~1367

ISSN: 2088-8694, DOI: 10.11591/ijpeds.v16.i2.pp1356-1367  1356

Adaptive control scheme of variable speed wind turbines for

frequency support

Leonardo Javier Ontiveros, Antonio Ernesto Sarasua, Cindy Madrid-Chirinos

Institute of Electrical Energy (IEE), Faculty of Engineering, National University of San Juan (UNSJ), San Juan, Argentina

Article Info ABSTRACT

Article history: Wind generation has experienced significant growth in power systems, due
to the high availability of the primary resource and the maturity of its
Received Apr 24, 2024 technology, which allows a fast control of active and reactive power.
Revised Mar 5, 2025 However, its main disadvantage is the lack of controllability over the
Accepted May 6, 2025 primary resource. This leads to unwanted frequency oscillations that affect
the power system security as wind penetration increases. This is due to the
inertia of wind turbines is decoupled from the inertia of synchronous
Keywords: machines connected to the power system. Based on the aforementioned
information, this paper analyses the state of the art of control strategies that
Adaptive control scheme allow wind turbines to participate in the frequency control of the power
Frequency control system. The main contribution of this work is the novel control strategy
Transient response proposed, which implements a virtual synchronous machine controlled by an
Virtual synchronous machine adaptive control system to enhance the transient response of the wind
Wind generation turbine. This scheme allows efficient management of the turbine's rotating
reserve without the need to reduce its output power or use expensive energy
storage systems. This solution is suitable for power systems with high wind
penetration (above 20%). The validity of this proposal is demonstrated
through dynamic simulations in a test system.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Leonardo Ontiveros
Institute of Electrical Energy (IEE), Faculty of Engineering, National University of San Juan (UNSJ)
Av. Libertador 1109 West, San Juan 5400, Argentina
Email: [email protected]

1. INTRODUCTION
The depletion of world fossil fuels reserves and the growing environmental pollution have
originated the development of renewable energy sources in recent decades. Currently, renewable energies
have been established around the world as an important source of energy. Due to its rapid growth, the energy
sector of developing countries is shifting to “energy transition” where renewable energy plays a leading role
moving from a conventional scheme based on thermal, and nuclear power plants to an alternative generation
scheme.
Within the renewable energy portfolio, wind generation has experienced a considerable growth in
recent years, currently it is one of the most attractive alternatives [1], [2]. The advantages are the high
availability of the primary resource, the maturity of the technology with increasingly competitive costs, and
the connection to the electrical network through electronic power devices allows rapid control of active and
reactive power [2], [3]. Among the disadvantages of wind generation include high installation costs and lack
of controllability over the primary resource [2]. Depending on a primary energy source that is not easy
controllable and under a significant penetration level, wind generation leads to conflicts in the electrical
power system. For this reason, electricity companies are forced to evaluate the influence of wind generation
on various aspects of the electricity system [4]. Among them, we can highlight the power quality (e.g.,

Journal homepage: http://ijpeds.iaescore.com

Int J Pow Elec & Dri Syst ISSN: 2088-8694  1357

voltage violations, and/or insufficient reserve of reactive power) and the power system security (e.g.,
problems associated with the generation reserve due to the impossibility to control the primary energy
source). Considering these problems caused by the incorporation of wind generation into the power system, it
is necessary to analyze and develop solutions to minimize its impact.
This work evaluates an important aspect of the power system security: the frequency control.
Frequency control is seriously affected in systems with a high level of wind penetration (WP). This is due to
the inertia of wind turbines (WTs) is decoupled from the inertia of synchronous machines connected to the
power system. In the case of WT with a full-scale power converter, the mentioned converter decouples the
rotor speed from the system frequency. In this way, when a disturbance occurs, these WTs do not offer an
inertial response; therefore, the dynamic response of the power system is less damped than the case when
the WP is low. Then, if the WP level continues to increase, the total inertia of the power system reduces,
and it is necessary to increase the generation reserve in the electrical system. In WTs, this generation
reserve is obtained by reducing their output power, this causes the dispatch of conventional power plants in
order to cover the same demand, leading to greater pollution and contradicting the purpose for which they
were designed (clean energy). On the other hand, if the generation reserve is not increased and a disturbance
occurs, the system presents a reduced inertia value that leads to greater frequency deviations from its
nominal value, which can cause the disconnection of the load or even, in extreme cases, the collapse of the
electrical system.
Based on the aforementioned information, first this paper analyses the state of the art of control
strategies that allow WTs to participate in the frequency control of the power system. Then, a novel control
strategy is proposed, which implements a virtual synchronous machine controlled by an adaptive control
scheme to enhance the transient response of the WT. This control proposal takes advantage of the kinetic
energy stored in the inertia of the rotating mass, thereby obtaining the generation reserve without reducing
the output power of the WT. This solution allows increasing WP without the need to increase the generation
reserve of conventional generation plants. This alternative is ideal for power systems with high WP (above
20%), characterized by their stochastic, non-linear and time-varying behavior. This is demonstrated by
dynamic simulations in a test system.

2. METHOD
Current technological developments enhance the control capabilities of WTs, allowing wind farms
to participate in the frequency control of the electrical system and consequently, to improve the power system
security. This leads to the revision of the current grid codes by system operators. Nowadays, there are three
principles to perform frequency control with wind farms [5], [6]: droop control, damping control, and inertial
control. In the droop control, the power of the WT is proportional to the deviation of frequency (Δf), as
occurs in conventional generators. Generally, a droop factor (R) between 0.03 and 0.05 pu is used. For
example, if the value of R is 0.05, and the actual frequency varies from 1.0 pu to 0.95 pu, the power of the
droop control 𝑃𝑑𝑟𝑜𝑜𝑝 varies from 0 pu to 1.0 pu is (1).

(𝑓0 −𝑓) ∆𝑓
𝑃𝑑𝑟𝑜𝑜𝑝 = = [𝑝𝑢] (1)
𝑅 𝑅

Where 𝑓0 is the rated frequency (1.0 pu) and 𝑓 is the actual frequency in pu. Note that 𝑃𝑑𝑟𝑜𝑜𝑝 is positive when
𝑓 is less than 𝑓0 .
The damping control calculates a power component 𝑃𝑑𝑎𝑚𝑝 that helps to attenuate unwanted
oscillations. This term is not always used; generally, it is applied when the system presents an underdamped
response or frequency variations caused by renewable generation. In order to calculate the value of 𝑃𝑑𝑎𝑚𝑝 , a
damping coefficient 𝑘𝐷 is employed as (2).

𝑃𝑑𝑎𝑚𝑝 = 𝑘𝐷 ∆𝑓 = 𝑘𝐷 (𝑓0 − 𝑓)[𝑝𝑢] (2)

In inertial control, the power generated by the WT is proportional to the derivative of the system
frequency. In the strict sense, this component is obtained from the differential equation that describes the
dynamic behavior of the WT, given as (3).
𝑑𝜔𝑟
𝐽𝑊𝑇 = 𝑇𝑚 − 𝑇𝑒 [N.m] (3)
𝑑𝑡

Where 𝐽𝑊𝑇 is the moment of inertia of the WT in kg.m 2, 𝜔𝑟 is the rotor speed in rad/s, 𝑇𝑚 and 𝑇𝑒 are the
mechanical and electromagnetic torque respectively in Newton-meter (N.m). Usually, the parameters
Adaptive control scheme of variable speed wind turbines for frequency support (Leonardo Ontiveros)
1358  ISSN: 2088-8694

employed in dynamic stability studies use the inertial constant (H) instead of J, which is defined as the ratio
between the kinetic energy stored in the rotating mass and the rated power of the WT. Generally, in power
systems, typical values of H vary between 2 s and 9 s (4).
1
𝐽𝑊𝑇 𝜔02
𝐻=2 [s] (4)
𝑆𝑟𝑎𝑡𝑒𝑑

Where 𝜔0 is the rated speed of the WT in rad/s, and 𝑆𝑟𝑎𝑡𝑒𝑑 is the rated power in VA. The quotient between
𝜔𝑟 and 𝜔0 is the rotor speed in pu as in (5).
𝜔𝑟
𝜔𝑟𝑜𝑡 = [𝑝𝑢] (5)
𝜔0

Considering that the product between the torque and the rotor speed is the rotor power, and replacing (4) and
(5) in (3), the differential of the WT is obtained as (6).
𝑑𝜔𝑟𝑜𝑡
2𝐻𝜔𝑟𝑜𝑡 = 𝑃𝑖𝑛𝑒𝑟𝑡𝑖𝑎 [𝑝𝑢] (6)
𝑑𝑡

The (6) describes the inertia power component of a WT, obtained from the rotor speed and the inertia
constant. However, full-scale converter WTs decouple the inertial response of the rotor from the power
system. This is accomplished by the speed regulator of the WT, which sets the reference power to the
electronic converter. Based on this, it is possible to set another value of the inertial power component, by
employing the system frequency measured at the point of common coupling of the WT (f), and a virtual
inertia constant (Hv). Analogously to (6) a new inertial power component is defined in (7).

𝑑𝑓
2𝐻𝑣 𝑓 = 𝑃𝑖𝑛𝑒𝑟𝑡𝑖𝑎 ′ [𝑝𝑢] (7)
𝑑𝑡

Thus, the general expression of the power flow of WTs that participate in the frequency control can be
obtained as (8).
𝑑𝑓 (𝑓0 −𝑓)
𝑃𝑊𝑇 = 𝑃𝑂𝑃 − 2𝐻𝑣 𝑓 + 𝑘𝐷 (𝑓0 − 𝑓) + [𝑝𝑢] (8)
𝑑𝑡 𝑅

Where POP is the optimal power of the WT, obtained from the tracking characteristic curve [7], [8]. From (8),
different works have been developed. The papers presented in [9], [10] propose a droop control system,
which is activated in case the frequency of the system exceeds the dead band of the frequency regulator (FR).
The research carried out in [11], [12] suggests the use of a damping control scheme, with the purpose of
mitigating oscillations in the power system. On the other hand, the recent papers presented in [13]-[16]
propose an inertial control scheme with the purpose of attenuating the frequency variation at the beginning of
the disturbance. In other words, the WT provides the instantaneous reserve to the power system.
Some research suggests combining two or more control methodologies discussed above. As an example,
the combination of control droop, control damping, and the reduction of the output power of the WT. This
scheme corresponds to the primary frequency control, which is treated in the works [17]-[19]. In summary, as (9).
(𝑓0 −𝑓)
𝑃𝑊𝑇,𝑃𝐹𝐶 = 𝑃𝑠𝑐ℎ𝑒𝑑 + 𝑘𝐷 (𝑓0 − 𝑓) + [𝑝𝑢] (9)
𝑅

Where 𝑃𝑠𝑐ℎ𝑒𝑑 is the scheduled output power of the WT, which is always less than POP to set the required
generation reserve. For instance, in [18] the generation reserve is obtained by adjusting the pitch angle of the
WT; consequently, the WT operates at a non-optimal operating point. In a power system with a high level of
wind penetration, the reduction of wind generation causes the dispatch of thermal power plants since it is
necessary to cover the same energy demand. This control scheme is opposed to the postulates enunciated in
the introduction; that is, the reduction of wind power produces greater pollution and the depletion of fossil
fuels. For this reason, this scheme will not be treated in the present work.
The papers published in [20]-[23] employ all the terms described in equation 8; namely, the WT
behaves like a virtual synchronous generator (VSG). This constitutes the current research trend for WTs to
participate in the frequency control of the power system. The concept of VSG is to emulate the behavior of a
synchronous machine, or a particular characteristic of it. This concept is not only limited to wind generation
applications, but also in other renewable generation systems that have stored energy, or in applications that
involve energy storage systems (ESS). Unlike the works cited above, which obtain the generation reserve by

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1356-1367
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1359

reducing the output power of the wind turbine or by using ESS, this work proposes to take advantage of the
energy stored in the inertia of the rotating mass of the WT by means of a change in its control system. This
topic is addressed in the next chapter and constitutes the main contribution of this work.

3. A NOVEL CONTROL SYSTEM FOR THE WIND TURBINE

According to the previous section, in this work, the VSG concept is proposed for the WT. In order
to achieve that goal, the FR has to incorporate the three control loops defined in section 2: droop control,
damping control, and inertia control. Additionally, during normal operation, the WT must operate at the point
of maximum efficiency. For this reason, the output power of the WT is not reduced, so the generation reserve
is obtained from the energy stored in the rotating mass of the rotor.
Under these premises, the performance of the WT for frequency regulation varies according to the
output power, the stored energy, and the parameters of the FR. This means that a set of specific values of the
FR at high wind speeds may not be optimal for low wind speeds and vice versa. This is due to the non-
linearity between the rotor speed, the power generated by the WT, and the energy stored in the rotor. This
problem has not been addressed in the literature, and it is proposed to be resolved, being the main
contribution of this work.
In this work, an adaptive control strategy is proposed as a solution to the problem of nonlinearity of
the WT. The adaptation is the mechanism by which the controller is able to change its parameters to respond
to a change in the operating conditions of the WT (rotor speed). According to Figure 1, the control system
developed has the typical tracking characteristic curve and the adaptive gain scheduling control system
(AGSCS). The AGSCS is composed of the FR and a fuzzy inference system (FIS), which calculates the
parameters of the FR. In the case there are no frequency disturbances, the FR is deactivated, and the value of
PFC is zero. Therefore, the power reference of the WT (PWT,ref ) is equal to the power optimal point (POP) of
the tracking characteristic curve. In this way, the WT presents the typical dynamic behavior. Due the
controller uses the kinetic energy stored in the rotor; it is convenient to estimate the energy stored as a
function of the WT rotor speed. This is discussed in the next section.

Figure 1. Model of the wind turbine with the AGSCS

3.1. Energy storage capacity of the wind turbine

The use of the WT as an ESS to provide power during frequency regulation requires the estimation
of its storage capacity. This is carried out by means of dynamic simulations that allow estimating the energy
stored in the inertia of the rotor. The principle of energy storage is carried out in the following way: when it
is necessary to increase the wind power injected into the network, the process is carried out by decelerating
the rotor, which produces a decrease in the kinetic energy of the rotating mass of the WT. On the other hand,
when it is necessary to decrease the power, the process is carried out by accelerating the rotor, producing an
increase in the kinetic energy stored in the rotating mass of the WT. Namely, during frequency regulation, the
operating principle of the variable speed WT is similar to a flywheel ESS [2]. The energy stored in the
rotating mass of the WT is directly proportional to the moment of inertia of the turbine-generator set JWT (JT
+ JG) and the square of the rotation speed ωr. Therefore, the variations of the rotor speed produce increases
or decreases in the kinetic energy of rotation. This can be calculated as (10).
1
∆𝐸𝑊𝑇 = 𝐽𝑊𝑇 ∆𝜔𝑟2 = 𝑘1 ∆𝜔𝑟2 [J] (10)
2

Where, ∆𝐸𝑊𝑇 is the variation of the kinetic energy of the rotor, and k1 is constant. In this sense, the energy
analysis is carried out by numerical calculation through dynamic simulations. The maximum power point

Adaptive control scheme of variable speed wind turbines for frequency support (Leonardo Ontiveros)
1360  ISSN: 2088-8694

tracking (MPPT) of the WT is modified, applying positive power steps (+∆𝑃𝑚𝑒𝑐 ) and negative power steps
( ̶∆𝑃𝑚𝑒𝑐 ). Due to the non-linearity between wind speed (𝑣𝑤𝑖𝑛𝑑 ) and the mechanical power produced (𝑃𝑚𝑒𝑐 ), the
magnitude of the power step is limited to 10% of the reference power defined by the MPTT (linearization of
nonlinear systems around an operating point). Then, the positive power step and the time (∆𝐸𝑊𝑇 ) required for
the WT to decelerate to its minimum rotor speed ( 𝜔𝑟,𝑀𝑖𝑛 = 0.5) are registered. Analogously, the values of
negative power steps and the time required for the WT to saturate ((𝜔𝑟,𝑀𝑎𝑥 = 1.0) are recorded. Based on the
variations in power and the associated times to reach the minimum and maximum speed of operation of the
WT, the energy that can be stored or extracted from the rotating mass of the WT is determined with (11).

∆𝐸𝑊𝑇 = ∆𝑃𝑚𝑒𝑐 ∆𝑡𝑊𝑇 [J] (11)

The results obtained for the detailed model of a 2 MW type 4 WT that is part of the Simulink® software
library [24] are shown in Table 1. Note that the rotor speed and the state-of-charge (SOC) are used
indistinctively in the second column of Table 1; where SOC = 0.5 corresponds to the minimum load of the
WT (𝑃𝑚𝑒𝑐 = 0.064 pu); and at SOC = 1.0, the WT is at full load (𝑃𝑚𝑒𝑐 = 1.0 pu).
It is important to note that as the rotor speed increases, the time available to supply reserve power
decreases. It is possible to find an explanation for this phenomenon by means of the analytical study of the
equations that describe the static behavior of the WT. In this sense, the mechanical power of a WT (Pmec) can
be found with (12).
1 3
𝑃𝑚𝑒𝑐 = 𝜌 𝐴 𝑣𝑤𝑖𝑛𝑑 𝐶𝑝 (𝜆, 𝛽) [W] (12)
2

Where ρ is the air density (1.225 kg/m2), A is the rotor área (m2), 𝑣𝑤𝑖𝑛𝑑 is the wind speed (m/s), λ is the tip
speed ratio, β is the pitch angle, and Cp(λ, β) is the power coefficient of the WT, which describes the fraction
of the wind power that is converted by the turbine into mechanical power. The value of λ can be found as (13).
𝜔𝑟 𝑅𝑟𝑜𝑡
𝜆= (13)
𝑣𝑤𝑖𝑛𝑑

Where 𝑅𝑟𝑜𝑡 is the rotor radius. Considering that the WT operates near the MPPT, the values of λ and Cp
remain constant [25]. Replacing (13) in (12), obtained (14) and (15).

1 𝜔𝑟 𝑅𝑟𝑜𝑡 3
𝑃𝑚𝑒𝑐 = 𝜌 𝐴 𝐶𝑝 (𝜆, 𝛽) ( ) [W] (14)
2 𝜆

∆𝑃𝑚𝑒𝑐 = 𝑘2 ∆𝜔𝑟3 [W] (15)

Where k2 is constant. The time of loading/unloading of the WT can be found employing (10), (11), and (15).
∆𝐸𝑊𝑇 𝑘1
𝛥𝑡𝑊𝑇 = = [s] (16)
∆𝑃𝑚𝑒𝑐 𝑘2 ∆𝜔𝑟

The (16) denotes that for high variations in rotor speeds produce shorter times for the WT to saturate or
exhaust its stored energy. As an advantage, the output power of the WT is proportional to the cube of the
rotor speed. These considerations must be considered in the design of the controller responsible for
monitoring the state of charge of the WT. This aspect is addressed in the next section.

Table 1. Loading and unloading capacity of the WT

Operating point of the WT Unload Load
𝑣𝑤𝑖𝑛𝑑 [m/s] ωrot = SOC: [𝑝𝑢] 𝑃𝑚𝑒𝑐 [MW] ∆𝑃𝑚𝑒𝑐 [MW] ΔtWT1 [s] EWT1 [MJ] ΔtWT2 [s] EWT2 [MJ]
5 0.5 0.128 ±0.0128 0 0 1290 16.51
6 0.555 0.234 ±0.0234 60 1.40 671 15.70
7 0.626 0.374 ±0.0374 43 1.61 396 14.81
8 0.713 0.56 ±0.056 36 2.02 183 10.25
9 0.809 0.8 ±0.08 33 2.64 78 6.24
10 0.907 1.098 ±0.1098 30 3.29 17 1.87
11 0.984 1.468 ±0.1468 22 3.23 2 0.29

3.2. Frequency regulator

The frequency control scheme is based on the standard model employed in speed regulators of
thermal turbines, which is displayed in Figure 2(a). The FR is composed of two parts: the lower control loop
defined by (1) and (2), and the higher control loop that synthesizes the inertial response (7). In case the

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1356-1367
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1361

frequency of the power system varies from the reference value f0, the FR calculates the value of the three
power components defined in (1), (2), and (6). These power components depend on the parameters Hv, R and
kD; which are calculated by the fuzzy inference system.

3.3. Fuzzy inference system

The aim of the fuzzy inference system, for high rotor speeds, is to maximize the use of the kinetic
energy stored in the WT. For medium rotor speeds, the fuzzy system prioritizes the stable operation of the
WT; this leads to a less utilization of the stored kinetic energy. For low rotor speeds, the energy stored is
minimal and the FIS disables the FR. According to Figure 2(b), the FIS is composed of three sub-systems: a
fuzzification stage, an inference engine, and a defuzzification stage. The input variable to the FIS is the rotor
speed 𝜔𝑟𝑜𝑡 , which varies between 0 and 1 𝑝𝑢. Therefore, the membership function of the Fuzzification Stage
is set in the aforementioned interval. Different membership functions have been evaluated, such as gaussian,
trapezoidal and triangular. According to the results obtained, the trapezoidal function is the best option, thus
it is chosen in the present work.
As reported by Table 1, if the rotor speed is 0.5 pu, the power of the WT is minimal (0.064 pu), and the
energy available for unloading is zero (EWT1). Thus, the membership function sets the threshold at 0.51 pu for low
rotor speed. Then, the fuzzy variables are defined as follows: l = 1, m = h = 0. If the rotor speed varies between
0.55 pu and 0.7 pu, the energy stored in the WT is in the range of 1.4 MJ to 2.0 MJ. Therefore, the fuzzification
stage sets this interval as medium rotor speed, due there is enough energy stored to perform frequency regulation.
The fuzzy variables are m = 1, l = h = 0. On the other hand, if the rotor speed is greater than 0.8 pu, the energy
available for unloading is in the range of 2.6 MJ to 3.3 MJ, thus there is plenty energy to carry out the frequency
regulation. Therefore, the membership function sets the rotor speed as high. The fuzzy variables are h = 1, l = m =
0. For the intermediate intervals: 0.51-0.55 pu and 0.7-0.8 pu, the trapezoidal membership functions take the form
of two linear functions, one with a positive slope and the other with a negative slope. Within the range, the sum of
these lineal functions is equal to 1.0 pu; namely: l + m = 1 and m + h = 1.

(a)

(b)

Figure 2. Control system of the wind turbine: (a) FR and (b) FIS

In the inference engine, the fuzzy variables l, m, and h are multiplied by the parameters of the FR. In
that sense, it has been defined three values of the parameters R, Hv, and kD. For low rotor speeds, it is adopted
Rl = inf, Hvl = 0, and kDl = 0. For medium rotor speed: Rm = 0.05, Hvm = 2, and kDm = 100. For high rotor speed,

Adaptive control scheme of variable speed wind turbines for frequency support (Leonardo Ontiveros)
1362  ISSN: 2088-8694

Rh = 0.03, Hvh = 9 and kDh = 200. The values of R and Hv have been adopted according to the typical
parameters of synchronous machines. On the other hand, the value of kD has been obtained through dynamic
simulations using the trial-and-error method. In the stage of defuzzification, the parameters of the FR are
obtained, considering that for any value of ωr, the sum of the fuzzy variables is equal to 1.0 pu.

4. TEST SYSTEM
The 230 kV test system is shown in Figure 3. This model is based on the IEEE 9-bus modified test
system which consists of 2 synchronous machines with IEEE type-1 exciters. At bus 2, it is connected a 600
MW steam turbine, at Bus 3 it is connected a 200 MW hydraulic power plant, while at bus 1 it is connected a
270 MW wind farm. In total, there are 9 buses, 6 transmission lines, 3 transformers and 3 constant P-Q loads.
The wind farm consists of 135 variable speed WTs (type 4), the maximum power of each WT is 2
MW if the wind speed is 13 m/s. The cut-in wind speed is 5 m/s, and the cut-out wind speed is 25 m/s. The
rotor speed varies between 0.75 rad/s and 1.5 rad/s; if the base rotor speed is 1.5 rad/s, then the rotor speed
varies between 0.5 and 1.0 pu. For rotor speeds less than 0.5 pu, the WT is disconnected. The models of the
Type 4 WT and the wind speed were developed in [25]-[27].

Figure 3. Test system

5. SIMULATION RESULTS
The simulations were carried out considering 3 load conditions, each of them with a different wind
speed: low wind speed (6 m/s), medium wind speed (9 m/s) and high wind speed (11 m/s). Table 2
summarizes the dispatches of the test system; the values reported in loads are approximate and include losses
in lines and transformers. Regarding the power factor, a value of 0.97 was adopted (lagging). In all cases, the
turbine speed is modified by adjusting the FR power PFC, Figures 1 and 2(a).

5.1. Case A: low wind speed

In this case, the wind speed is set constant at 6 m/s, the wind farm produces 32.01 MW and the WP is
4.39 %. It is important to note that wind turbulence has not been considered to facilitate the analysis of the inertial
response of the wind farm. Due to the frequency variations caused by renewable generation are minimal, the
parameter kD is set to zero. At time t = 40 s, Load 1 increases by 60 MW; this produces an imbalance between
generation and demand with the consequent decrease in the frequency of the system as shown in Figure 4(a).
According to Figure 4(a), the wind farm enhances the transient response of the power system if the
AGSCS is activated. According to the initial value of 𝜔𝑟𝑜𝑡 , the parameters of the FR are: Hv = 2, kD = 0 and R
= 0.05 (case 2). In this case, the minimum frequency is 59.62 Hz. If the AGSCS is deactivated (case 1), the
minimum frequency is 59.28 Hz. If the FIS is deactivated, and the FR is activated with Hv = 9, R = 0.03, kD =
0 (case 3), the minimum frequency is 58.33 Hz.
Figure 4(b) shows the rotor speed of the wind farm, it keeps steady at 0.5573 pu if the AGSCS is
deactivated (case 1). When the AGSCS is activated (case 2), the rotor speed decreases to 0.5013 pu then
recovers to its previous state (0.5573 pu). This allows the wind farm to increase its output power to
compensate the load imbalance (Figure 4(c)). If the FIS is deactivated, and the FR is activated with Hv = 9, R

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1356-1367
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1363

= 0.03, kD = 0 (case 3), at t = 45 s the rotor speed drops below 0.5 and the wind farm is disconnected, so the
power output is zero shown in Figure 4(c). This leads to a severe disturbance because the steam turbine and
the hydroelectric power plant cover the power shortage, further accentuating the frequency excursion as
shown in Figure 4(a), case 3).
The output power of the steam turbine and the hydroelectric power plant is illustrated in
Figures 4(d) and 4(e). Comparing case 1 with case 2, it is clearly shown that the AGSCS reduces the dynamic
stresses of conventional generators. If the FIS is deactivated, and the FR is activated (case 3) it is observed
that the Steam Turbine reaches its rated power at t = 60 s, so the hydroelectric power plant must cover the
power shortage with a slow transient response shown in Figure 4(e), case 3. Therefore, for low wind speed,
case 2 is the best option (AGSCS activated).

Table 2. Dispatches of the test system

Wind speed Wind farm Steam turbine Hydro power Load 1 Load 2 Load 3 Total load WP
[m/s] [MW] [MW] [MW] [MW] [MW] [MW] [MW] [%]
6 32.0 516.7 179.5 250 215 263.2 728.2 4.39
9 109.7 516.2 179.5 250 300 255.3 805.3 13.6
11 208.8 515 179.4 250 300 353.2 903.2 23.1

(a)

(b)

(c)

(d)

(e)

Figure 4. Dynamic response of the test system for low WP: (a) frequency, (b) rotor speed of the WTs,
(c) wind farm power output, (d) steam turbine power output, and (e) hydraulic power plant power output. (1⚫)
AGSCS deactivated, (2) AGSCS activated, (3◼) FIS deactivated, and FR activated (Hv = 9, R = 0.03, kD = 0)

5.2. Case B: medium wind speed

In this case, the wind speed is 9 m/s, and the wind farm generates 109.68 MW, thereby the WP is
13.62%. Similar to the previous example, the turbulence of the wind has not been considered in order to
highlight the inertial response of the wind farm; therefore, the parameter kD is set to zero. At time t = 40 s,
Load 1 increases from 250 MW to 310 MW, which produces a decrease in the frequency due to the
imbalance between generation and demand shown in Figure 5(a).
Adaptive control scheme of variable speed wind turbines for frequency support (Leonardo Ontiveros)
1364  ISSN: 2088-8694

Figure 5(a) shows that the AGSCS enhances the transient response of the test system. For case 1
(AGSCS deactivated), the minimum frequency is 59.3 Hz and the rotor speed remains constant at 0.8134 pu
Figure 5(b). For case 2, the AGSCS sets the parameters of the FR at Hv = 9, R = 0.03, kD = 0; this leads to a
minimum frequency of 59.71 Hz and ωr at 0.7539 pu. If the FIS is deactivated, and the FR is activated with Hv
= 2, R = 0.5 kD = 0 (case 3); the minimum frequency is 59.62 Hz and the lowest ωr is 0.7729 pu. This shows
that for higher values of the parameters Hv and R, a greater use of the energy stored in the rotating masses of
WTs is made, as shown in Figures 5(b) and 5(c), and a better transient response of the test system is obtained.
This simulation and the previous one demonstrate that the FIS chooses the best parameters for the FR.

(a)

(b)

(c)

Figure 5. Dynamic response of the test system for medium WP: (a) frequency; (b) rotor speed of the WTs;
and (c) wind farm power output, (1⚫) AGSCS deactivated, (2) AGSCS activated, (3◼) FIS deactivated,
and FR activated (Hv = 2, R = 0.05, kD = 0)

5.3. Case C: high wind speed

In this simulation, the average wind speed is 11 m/s, the wind farm produces 208.8 MW, and the
WP is 23.11%. The turbulent component of the wind speed has been simulated, and the wind speed
equivalent model for the wind farm considers the aggregation phenomena. These models are obtained from
[26], [27], and the results are shown in Figure 6(a) (see Appendix).
Figures 6(a)-6(c) show case 1 (AGSCS deactivated), case 2 (AGSCS activated), and case 3 (FIS
deactivated and FR activated with Hv = 2, R = 0.05, kD = 100). Clearly, the AGSCS enhances the transient
response of the test system, which sets the parameters of the FR at Hv = 9, R = 0.03, and kD = 200. In particular,
the increase of the kD parameter allows the reduction of the frequency excursion (Figure 6(b)), at the expense of
a major utilization of the energy stored in the inertia of WTs' rotational masses, shown in Figure 6(c).
The active power generated in the test system is shown in Figures 6(d)- 6(f). Clearly, it is observed
that the AGSCS smooths the power generated by the wind farm (Figure 6(d)) and reduces the mechanical
stress of conventional generators, as shown in Figures 6(e) and 6(f). According to Figure 6, the best transient
response is obtained when the AGSCS is activated.

6. CONCLUSION
This work presents a novel frequency regulator for wind turbines to improve the transient response
of power systems with a high level of wind penetration. Simulations show that the wind farm behavior
depends on the operation point, which in turn varies depending on the wind speed. Because it is a non-linear
and time-variant system, an AGSCS based on a FIS is required in order to calculate the best parameters of the
frequency regulator during the normal and emergency operation. The results show that for low wind speeds,
the AGSCS computes low values for the frequency regulator parameters. Although the contribution of
reserve power to frequency regulation is low, so is the penetration of wind power. In this case, conventional
power plants are the ones that contribute the most to recovering the frequency to its nominal value. As for
medium and high wind speeds, the AGSCS calculates high values for the frequency regulator parameters.

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1356-1367
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1365

This increases the reserve power provided by the wind turbines, helping the power system to restore its
nominal frequency faster. Additionally, it is shown that for high levels of wind penetration, turbulent
variations in wind speed can cause undesired fluctuations in the system frequency. These variations can be
mitigated by adjusting the damping parameter of the frequency regulator; although further studies are
required regarding the adjustment of this parameter, since its value depends on the dynamic behavior of the
power system under study. Nevertheless, in this case it is demonstrated that the parameters calculated by the
AGSCS are the most suitable for performing frequency regulation.

FUNDING INFORMATION
This work was partially financed by the Council for Science and Technology Research (CONICET),
the Argentinean National Agency for the Promotion of Science and Technology (ANPCyT) and the National
University of San Juan (UNSJ).

AUTHOR CONTRIBUTIONS STATEMENT

This journal uses the Contributor Roles Taxonomy (CRediT) to recognize individual author
contributions, reduce authorship disputes, and facilitate collaboration.

Name of Author C M So Va Fo I R D O E Vi Su P Fu
Leonardo Javier Ontiveros ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Antonio Ernesto Sarasua ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓
Cindy Madrid-Chirinos ✓ ✓ ✓ ✓

C : Conceptualization I : Investigation Vi : Visualization

M : Methodology R : Resources Su : Supervision
So : Software D : Data Curation P : Project administration
Va : Validation O : Writing - Original Draft Fu : Funding acquisition
Fo : Formal analysis E : Writing - Review & Editing

CONFLICT OF INTEREST STATEMENT

Authors state no conflict of interest.

DATA AVAILABILITY
Derived data supporting the findings of this study are available from the corresponding author, [LO],
on request.

REFERENCES
[1] World Wind Energy Association report, “WWEA Annual Report 2023,” World Wind Energy Association report.
https://wwindea.org/ss-uploads/media/2024/3/1711538106-40ab83f2-3e01-4c0a-9d28-e0a21bff72e6.pdf
[2] G. O. Suvire and P. E. Mercado, “Active power control of a flywheel energy storage system for wind energy applications,” IET
Renewable Power Generation, vol. 6, no. 1, pp. 9–16, Jan. 2012, doi: 10.1049/iet-rpg.2010.0155.
[3] S. Guan, X. Yan, G. Cai, and J. Jia, “An active and reactive power coordination control strategy during LVRT period,” in 2024
3rd International Conference on Energy, Power and Electrical Technology (ICEPET), May 2024, pp. 1759–1763. doi:
10.1109/ICEPET61938.2024.10627099.
[4] M. Gustavo and J. Gimenez, “Technical and regulatory exigencies for grid connection of wind generation,” in Wind Farm -
Technical Regulations, Potential Estimation and Siting Assessment, InTech, 2011. doi: 10.5772/16474.
[5] S. Xu, H. Wang, Y. Cao, S. Igarashi, J. Li, and X. Cai, “Inertia response control of self-synchronous voltage source doubly-fed
wind turbines in the whole wind speed range,” in 2024 IEEE 10th International Power Electronics and Motion Control
Conference (IPEMC2024-ECCE Asia), May 2024, pp. 151–156. doi: 10.1109/IPEMC-ECCEAsia60879.2024.10567551.
[6] Z. Zhu, S. Du, L. Zhang, and Q. Qi, “A new coordinated control strategy to improve the frequency stability of microgrid based on
de-loaded capacity of wind turbine,” in 2020 5th Asia Conference on Power and Electrical Engineering (ACPEE), Jun. 2020, pp.
645–651. doi: 10.1109/ACPEE48638.2020.9136560.
[7] W. Yan, X. Wang, W. Gao, and V. Gevorgian, “Electro-mechanical modeling of wind turbine and energy storage systems with
enhanced inertial response,” Journal of Modern Power Systems and Clean Energy, vol. 8, no. 5, pp. 820–830, 2020, doi:
10.35833/MPCE.2020.000272.
[8] M. H. Ravanji and M. Parniani, “Small-signal stability analysis of DFIG-based wind turbines equipped with auxiliary control
systems under variable wind speed,” in 2020 IEEE International Conference on Environment and Electrical Engineering and
2020 IEEE Industrial and Commercial Power Systems Europe (EEEIC/I&CPS Europe), Jun. 2020, pp. 1–7. doi:
10.1109/EEEIC/ICPSEurope49358.2020.9160721.

Adaptive control scheme of variable speed wind turbines for frequency support (Leonardo Ontiveros)
1366  ISSN: 2088-8694

[9] Y. Zhang, J. Duan, J. Tao, Z. Li, and L. Li, “Fuzzy droop control strategy for doubly-fed wind turbines considering frequency
influencing factors,” in 2024 6th Asia Energy and Electrical Engineering Symposium (AEEES), Mar. 2024, pp. 529–533. doi:
10.1109/AEEES61147.2024.10544770.
[10] W. Chen, T. Zheng, H. Nian, D. Yang, W. Yang, and H. Geng, “Multi-objective adaptive inertia and droop control method of wind
turbine generators,” IEEE Transactions on Industry Applications, vol. 59, no. 6, pp. 7789–7799, 2023, doi: 10.1109/TIA.2023.3307368.
[11] Z. Li, Z. Ma, C. Yi, X. Cheng, Y. Zhang, and X. Zhang, “Study on oscillation of doubly fed wind turbine based on additional
damping control strategy,” in 2022 5th Asia Conference on Energy and Electrical Engineering (ACEEE), Jul. 2022, pp. 79–84.
doi: 10.1109/ACEEE56193.2022.9851824.
[12] K. Gunther and C. Sourkounis, “Active damping control for variable-speed wind turbines with VSM as grid-side control,” in 2021
22nd IEEE International Conference on Industrial Technology (ICIT), 2021, pp. 304–309, doi: 10.1109/ICIT46573.2021.9453518.
[13] R. Azizipanah-Abarghooee, M. Malekpour, T. Dragicevic, F. Blaabjerg, and V. Terzija, “A linear inertial response emulation for
variable speed wind turbines,” IEEE Transactions on Power Systems, vol. 35, no. 2, pp. 1198–1208, Mar. 2020, doi:
10.1109/TPWRS.2019.2939411.
[14] C. Nikolakakos, U. Mushtaq, P. Palensky, and M. Cvetkovic, “Improving frequency stability with inertial and primary frequency
response via dfig wind turbines equipped with energy storage system,” in 2020 IEEE PES Innovative Smart Grid Technologies
Europe (ISGT-Europe), Oct. 2020, pp. 1156–1160. doi: 10.1109/ISGT-Europe47291.2020.9248807.
[15] C. Wang, C. Li, K. Zeng, Y. Chen, L. Xiao, and C. Sun, “Research of inertial control strategy of wind turbines under low
frequency situation of hydro-dominated system,” in 2021 IEEE 4th International Electrical and Energy Conference (CIEEC),
May 2021, pp. 1–6, doi: 10.1109/CIEEC50170.2021.9510587.
[16] J. C. Martinez, S. A. Gomez, J. L. Rodriguez Amenedo, and J. Alonso-Martinez, “Analysis of the frequency response of wind
turbines with virtual inertia control,” in 2020 IEEE International Conference on Environment and Electrical Engineering and
2020 IEEE Industrial and Commercial Power Systems Europe (EEEIC/I&CPS Europe), 2020, pp. 1–6, doi:
10.1109/EEEIC/ICPSEurope49358.2020.9160718.
[17] C. Xiu, Z. Lv, Z. Zhou, J. Wang, and Y. Guo, “Deloading control strategy of permanent magnet direct drive wind turbine with
adaptive droop gain,” in 2024 IEEE 7th International Electrical and Energy Conference (CIEEC), May 2024, pp. 5058–5063.
doi: 10.1109/CIEEC60922.2024.10583672.
[18] B. Lu, G. Zhu, C. Zhang, L. Ding, S. Wang, and Y. Guan, “Pitch deloading method of wind farm for optimal frequency control
performance,” in 2023 6th International Conference on Energy, Electrical and Power Engineering (CEEPE), May 2023, pp. 344–
349. doi: 10.1109/CEEPE58418.2023.10166677.
[19] W. Bao, L. Ding, Y. C. Kang, and L. Sun, “Closed-loop synthetic inertia control for wind turbine generators in association with
slightly over-speeded deloading operation,” IEEE Transactions on Power Systems, vol. 38, no. 6, pp. 5022–5032, Nov. 2023, doi:
10.1109/TPWRS.2022.3224431.
[20] J. Xi, H. Geng, and X. Zou, “Decoupling scheme for virtual synchronous generator controlled wind farms participating in inertial
response,” Journal of Modern Power Systems and Clean Energy, vol. 9, no. 2, pp. 347–355, 2021, doi: 10.35833/MPCE.2019.000341.
[21] C. Shao, Z. Li, R. Hao, Z. Qie, G. Xu, and J. Hu, “A wind farm frequency control method based on the frequency regulation
ability of wind turbine generators,” in Proceedings - 2020 5th Asia Conference on Power and Electrical Engineering, ACPEE
2020, 2020, pp. 592–596, doi: 10.1109/ACPEE48638.2020.9136431.
[22] Q. Meng, Y. Ren, and H. Liu, “Frequency stability analysis of grid-forming PMSG based on virtual synchronous control,” IEEE
Access, vol. 12, pp. 84134–84148, 2024, doi: 10.1109/ACCESS.2024.3414424.
[23] A. Mishra, A. S. Mir, and N. Prasad Padhy, “Improving system inertial support from PMSG based wind turbines with adaptive
inertial and damping control,” in 2023 IEEE International Conference on Power Electronics, Smart Grid, and Renewable Energy
(PESGRE), Dec. 2023, pp. 1–5, doi: 10.1109/PESGRE58662.2023.10404366.
[24] “Wind Farm - Synchronous generator and full-scale converter (Type 4) average model,” MATLAB/Simulink Website.
https://www.mathworks.com/help/sps/ug/wind-farm-synchronous-generator-and-full-scale-converter-type-4-detailed-model.html
[25] L. J. Ontiveros and P. E. Mercado, “Modeling of variable-speed wind farms for power systems dynamic studies,” in Proc. 14th
Ibero-American Regional Meeting of CIGRÉ, Ciudad del Este, Paraguay, 2011, pp. 1–8.
[26] L. J. Ontiveros, “Incorporation of redox batteries in wind farms to perform secondary frequency regulation,” San Juan National
Univ. (UNSJ), 2011.
[27] L. J. Ontiveros, P. E. Mercado, and G. O. Suvire, “A new model of the double-feed induction generator wind turbine,” in 2010
IEEE/PES Transmission and Distribution Conference and Exposition: Latin America (T&D-LA), Nov. 2010, pp. 263–269. doi:
10.1109/TDC-LA.2010.5762892.

APPENDIX

(a)

(b)

Figure 6. Dynamic response of the test system for high WP: (a) wind speed models; (b) system frequency

Int J Pow Elec & Dri Syst, Vol. 16, No. 2, June 2025: 1356-1367
Int J Pow Elec & Dri Syst ISSN: 2088-8694  1367

(c)

(d)

(e)

(f)

Figure 6. Dynamic response of the test system for high WP: (c) rotor speed of the WTs; (d) wind farm power
output; (e) steam turbine power output; and (f) hydraulic power plant power output (1⚫) AGSCS deactivated;
(2) AGSCS activated; (3◼) FIS deactivated, and FR activated (Hv = 2, R = 0.05, kD = 100) (continued)

BIOGRAPHIES OF AUTHORS

Leonardo Javier Ontiveros was born in San Juan, Argentina, on October 9, 1977.
He graduated as an electronic engineer from the UNSJ, Argentina, in 2004. He received his
Ph.D. from the same University in 2011, carrying out part in the COPPE institute, in the
Federal University of Rio de Janeiro in Brazil. Dr. Ontiveros is currently a professor of
electrical engineering at the UNSJ and a researcher with CONICET. His research interests
include simulation methods, power systems dynamics and control, power electronics modeling
and design, and the application of wind energy and energy storage in power systems. He can be
contacted at email: [email protected].

Antonio Ernesto Sarasua was born in San Juan, Argentina. He graduated as an

Electrical Engineering from the National University of San Juan (UNSJ), Argentina in 1994,
and received the Ph.D. degree from the UNSJ in 2005. Since 2006, Dr. Sarasua is Professor at
the Instituto de Energía Eléctrica – CONICET-UNSJ. His research interests include power
system stability and bifurcation, simulation methods and the application of energy storage in
power systems. He can be contacted at email: [email protected].

Cindy Madrid-Chirinos received the B.Eng. degree in chemical and electrical

engineering in National University of Honduras in 2011 and 2012 respectively, and currently
student of M.Eng. degree in Electrical Engineering from National University of San Juan,
Argentina. Her research interests include the field of electrical power systems, photovoltaic power
systems, harmonics, remote sensing data analysis via geospatial technologies, and batteries
implemented in power system. She can be contacted at email: [email protected].

Adaptive control scheme of variable speed wind turbines for frequency support (Leonardo Ontiveros)