You are on page 1of 10

2186 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 27, NO.

4, NOVEMBER 2012
Solution of Power Flow With Auto-
matic Load-Frequency Control Devices
Including Wind Farms
Luis M. Castro, Claudio R. Fuerte-Esquivel, Senior Member, IEEE, and J. Horacio Tovar-Hernndez
AbstractThis paper proposes the integration of steady-state
models of several types of wind generators into a power ow al-
gorithm with automatic load-frequency control. Since the system
frequency deviation is considered a state variable to be computed
by the power ow solution, this formulation helps identify the op-
erating point of wind generators after the action of the primary fre-
quency control when power imbalances have occurred. The math-
ematical formulation of xed-speed wind generators is presented
based on the steady-state representation of the induction gener-
ator. Furthermore, as variable-speed wind generators keep gaining
prominence in power systems, their potential contribution to fre-
quency support is also analyzed herein. These models are formu-
lated within the power ow approach by using a unied single
frame of reference and the Newton-Raphson algorithm. The pro-
posed approach is then applied to the analysis of a three-machine,
eight-bus system and the IEEE-14 bus test system.
Index TermsNewton-Raphson algorithm, power ow analysis,
primary frequency regulation, wind generator.
I. INTRODUCTION
W
IND power generation is growing worldwide because
of economic and environmental benets. However,
from the technical standpoint, several challenges arise from
the wind power penetration into electric power systems. In this
context, the service credibility is one of the primary goals to
achieve even though wind power generation is stochastic in
nature.
The system reliability can be degraded by power imbalances
between load and generation, which are inherent to the daily
operation of power systems. In this context, the impact of a
power imbalance to the system frequency can be quickly deter-
mined by a power ow algorithm in which automatic load-fre-
quency controllers are included [1]. This approach considers
the frequency deviation as a state variable, which is simultane-
ously computed together with all magnitudes and phase angles
of nodal voltages in a unied frame of reference. Furthermore,
Manuscript received October 18, 2011; revised February 03, 2012 and April
02, 2012; accepted April 04, 2012. Date of publication May 09, 2012; date of
current version October 17, 2012. This work was supported by CONACYT
Mexico under the scholarship 209744 and under the research project 106198.
The work of L. M. Castro was supported by the Consejo Nacional de
Ciencia y Tecnologa (CONACYT) Mxico, and the University of Michoacn
(U.M.S.N.H). Paper no. TPWRS-00979-2011.
L. M. Castro and C. R. Fuerte-Esquivel are with the Electrical Engineering
Faculty, Universidad Michoacana de San Nicols de Hidalgo (UMSNH),
Morelia, Michoacn, 58000, Mxico (e-mail: lcastro@dep.e.umich.mx;
cfuerte@umich.mx).
J. H. Tovar-Hernndez is with the Instituto Tecnolgico de Morelia, Morelia,
Mxico (e-mail:horacio.tovar@yahoo.com).
Digital Object Identier 10.1109/TPWRS.2012.2195231
the voltage and frequency dependence of loads can be taken into
account to represent their variation according to different oper-
ating conditions. This methodology is rather attractive because
it permits the estimation of the steady-state operating point after
the action of the primary frequency regulation.
Bearing the above mentioned in mind, suitable wind gener-
ator models must be developed and included into a power ow
program with automatic load-frequency control (PFALFC) to
assess the role of wind farms in frequency regulation. To achieve
this goal, the preliminary work presented in [2] is expanded by
introducing two additional wind generator models to accom-
plish a complete modeling idea of wind generators considering
the frequency deviation of the electric network.
The mathematical modeling of wind turbines developed
in this paper relies primarily on the power injection concept.
For this purpose, a suitable representation of xed-speed wind
generators (FSWG) is developed, which includes their active
and reactive powers in terms of the induction machine stator
and rotor parameters as well as the system frequency deviation
[3]. Furthermore, variable-speed wind generators (VSWG)
are mathematically modeled based on both doubly-fed induc-
tion generators (DFIG) and permanent magnet synchronous
generators (PMSG). These models are developed with the
understanding that they are able to provide primary frequency
regulation by maintaining some spinning reserve [4][10]. The
PFALFCreported in [1] is then expanded to incorporate the pro-
posed models. In this context, these models are implemented in
such a way that the electrical network state variables, including
frequency deviation and those related to wind generators,
are solved simultaneously during the iterative process. This
solution method is selected with the aim of taking advantage of
the quadratical convergence of the Newton-Raphson method.
The proposed approach is presented in detail as follows:
Section II addresses the basics of the potential contribution of
wind generators to frequency support. Section III reviews the
PFALFC. The mathematical models of several types of wind
generators are presented in Section IV, while the initialization
of their state variables is given in Section V. Study cases are
provided in Section VI; nally, Section VII points out the
conclusions of this paper.
II. CONTRIBUTION OF WIND GENERATORS
TO FREQUENCY SUPPORT
The increasing integration of wind power into power sys-
tems has provided the momentum for exploring alternatives to
endow wind turbines with frequency regulation capability. In
the case of the primary frequency regulation, the key aspect is
0885-8950/$31.00 2012 IEEE
CASTRO et al.: SOLUTION OF POWER FLOW WITH AUTOMATIC LOAD-FREQUENCY CONTROL DEVICES INCLUDING WIND FARMS 2187
Fig. 1. Achieving a spinning reserve: (a) by using the blade pitch mechanism,
(b) by increasing the mechanical rotor angular speed.
to attain a spinning reserve of active power in order to com-
pensate active power imbalances. This goal can be achieved by
forcing the wind generator to operate in a suboptimal power
output level, which is referred to as deloading. Thus, each de-
loaded wind turbine will not extract the maximum power from
the wind during normal conditions in order to maintain a spin-
ning reserve, which will permit a release of power when a power
imbalance perturbation occurs [4][10].
A technique employed to deload a wind turbine calls for the
action of the blade pitch mechanism, as can be seen in Fig. 1(a),
[5], [7], [8]. In this case, the pitch angle is increased from to
or (depending on the required level of deloading), resulting
in different output powers obtained at the same rotor angular
speed . Therefore, the pitch angle controller acts like a
conventional speed governor.
On the other hand, by shifting the operating point to the
right of the wind turbine power curve, as shown in Fig. 1(b), a
wind turbine can attain a spinning reserve of active power to
be delivered in case of a power imbalance [4], [9]. However,
a rotor overspeeding is presented with this deloading
technique which limits its application beyond the rated wind
speed because of mechanical stress [6]. Lastly, both discussed
approaches can be simultaneously applied as proposed in [10].
Note that shifting the operating point towards the left of the
power curve, shown in Fig. 1(b), may also deload the wind
turbine. However, when requiring frequency response, a rotor
speed increase becomes necessary. In such a situation, a portion
of the energy extracted from the wind will accelerate the rotor
thus reducing the amount of power delivered to the network [4].
III. POWER FLOWS WITH AUTOMATIC
LOAD-FREQUENCY CONTROL METHOD
The power ow formulation with automatic load-frequency
control devices must include the system frequency deviation as
a state variable, which is calculated together with all the net-
works nodal voltages [1]. Therefore, when a load or generation
perturbation is simulated, the operating point obtained by the
PFALFC method will be at an off-nominal frequency value.
In this context, all wind generator models developed herein
can be readily included in the power ow formulation as shown
in (1), where and represent the power balance
equation of each wind generator and its associated state vari-
able, respectively. Power mismatch equations are denoted by
and at each network node, and is the system fre-
quency deviation from its nominal value:
(1)
Furthermore, this formulation can manage multiple gener-
ator regulators that respond to frequency changes as well as
voltage and frequency dependent load models, as briey de-
scribed below. Note that all equations are given in the per-unit
(p.u.) system, unless otherwise stated.
A. Synchronous Generator Models
Two synchronous generator models from the four presented
in [1] are addressed in this paper. The rst model represents a
generator regulator that adjusts its active power output and
reactive power output according to (2)(6), which are func-
tions of the frequency deviation value estimated at each power
ow iteration:
(2)
(3)
(4)
(5)
(6)
where and are the specied active and reactive
powers of the generator, respectively, is the nominal active
power, is the speed droop of the corresponding generator,
and are the coefcients of the reactive power genera-
tion, and the subscripts and represent maximum and
minimum values. The second model used in this work corre-
sponds to the conventional PV generator, where its generated
active power and its terminal voltage magnitude are set to
specied values during the simulation process. Constant voltage
operation is only possible if the generator reactive power de-
sign limits are not violated, i.e., . If the
generator cannot provide the necessary reactive power support
to constrain the voltage magnitude at the specied value then
the reactive power is xed at the violated limit, and the voltage
magnitude is freed. In this case, the generated active power
and reactive power are specied, while nodal voltage mag-
nitude and phase angle are computed.
For the purpose of this research, only two generator models
are used, since this paper focuses on wind generator models
within the context of the PFALFC. Please, refer to [1] for details
of the rest of existing synchronous generator models.
B. Load Model
Most of the loads in power systems are voltage and fre-
quency dependent, and their demanded power varies according
to changes in these variables. The static representation of their
active power and reactive power is described by the
following [1]:
2188 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 27, NO. 4, NOVEMBER 2012
Fig. 2. Steady-state equivalent model of the induction machine taking into ac-
count the system frequency deviation.
Fig. 3. Response of directly grid-connected wind generators to frequency vari-
ations in the network.
(7)
(8)
where and are the specied active and reactive
powers drawn by the load, respectively, is the nominal op-
erating voltage of the node where the load is connected,
and are the coefcients of the load-frequency characteristic,
whereas , , , , and are the coefcients of the
load-voltage characteristic.
IV. MATHEMATICAL MODELING OF WIND GENERATORS
AND THEIR INCLUSION IN THE PFALFC
The steady-state representation of the induction generator
shown in Fig. 2 is a function of its slip given by (9). This
implies that active and reactive powers generated by a directly
grid-connected wind generator will vary according to changes
in the system frequency and in the mechanical angular speed of
the rotor , as shown in Fig. 3
(9)
Based on the above mentioned, the stall regulated and pitch
regulated xed-speed wind generator models can be suitably
derived as a function of the frequency deviation by using the
power injection concept. The mechanical angular speed of the
generator rotor is employed as the wind generator state variable
within the iterative process. Hence, the power converted from
the mechanical to electrical form , the generated powers
and , the rotor current and the stator current are
expressed as follows:
(10)
(11)
(12)
(13)
(14)
where the coefcients are dened as
,
, ,
, ,
, , ,
, ,
, ,
, ,
.
A. Stall Regulated FSWG
The extracted power fromthe wind of a stall regulated FSWG
(SR-FSWG) is calculated by using (15)(17). Clearly, it is a
function of the rotor angular speed among other variables [11]:
(15)
(16)
(17)
where is the mechanical power (W), is the air density
( ), is the swept area of the blades ( ), is the wind
speed (m/s), is the radius of the rotor (m), is the gearbox
ratio, is the pitch angle (degrees), is the mechanical an-
gular speed of the generator (rad/s) and the constants to
are the parameters of the wind turbine design. In this case, the
only unknown variable is the mechanical angular speed of the
generator, as seen from (15). Consequently, this parameter acts
as the state variable within the iterative process which enables
to achieve the internal equilibrium point in the SR-FSWG given
CASTRO et al.: SOLUTION OF POWER FLOW WITH AUTOMATIC LOAD-FREQUENCY CONTROL DEVICES INCLUDING WIND FARMS 2189
by the conversion process of mechanical power into electrical
power.
In the framework of the PFALFC method, the set of
power ow mismatch equations that has to be solved when a
SR-FSWG is connected at the node of the system is
(18)
(19)
(20)
where and represent the active and reactive powers
drawn by the load at bus , respectively, and and
represent the powers injected at bus . They are given by
(21)
(22)
Lastly, the set of linearized equations (23) is integrated into
the set of linearized power mismatch equations of the whole net-
work (1) to compute a new equilibrium point through a power
ow solution. In this case, the generator angular speed is up-
dated after the th iteration as given by (24):
(23)
(24)
B. Pitch Regulated FSWG
A pitch regulated FSWG (PR-FSWG) has practically the
same mechanical construction as the SR-FSWG; however, the
PR-FSWG has a blade pitch angle mechanism which operates
for wind speeds beyond a rated value to limit the power ex-
tracted from the wind [12]. A typical pitch angle controller is
shown in Fig. 4, which takes the power output of the wind
generator as the input signal and regulates the pitch angle
through a PID controller.
For an operating condition where the generated active power
is below the specied value , the mathematical equations
representing the SR-FSWG model are also valid to represent
the steady-state operation of a PR-FSWG. On the other hand,
Fig. 4. Typical blade pitch angle controller.
when the output power surpasses , the blade pitch angle
controller is activated. In this case, the controller sets at the
constant value , whereas the reactive power has to be
calculated within the iterative process. Thus, by neglecting the
core losses in the induction machine, the mechanical power can
be computed by
(25)
where and are the three-phase stator and rotor
power losses, respectively.
Assuming that the PR-FSWG is connected at node , the
values of the network state variables and the wind generator
that satisfy the mismatch (26)(28) are obtained by solving (29).
This last equation has to be accommodated with the rest of the
linearized power mismatch equations of the network for a uni-
ed power ow solution:
(26)
(27)
(28)
(29)
where the state variable is updated after each iteration ac-
cording to (24), and represents the power balance
inside the induction machine obtained by equating (10) and
(25). In summary, the next conditions must be considered when
a PR-FSWG is included into the PFALFC program: if
, then (15)(24) must be used. If , then (21),
(22), and (24)(29) must be used instead, setting .
From the software design perspective, the simulation starts
by employing (15)(24). The active power generated by the
PR-FSWG is checked at the end of the second iteration because
the rst iteration generally provides an inaccurate approxima-
tion to the power ow solution. If happens to be equal or
2190 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 27, NO. 4, NOVEMBER 2012
Fig. 5. Automatic load-frequency control of a DFIG.
greater than , the next iteration is executed using (21),
(22), and (24)(29). Note that even when a high wind speed is
selected and was meant to surpass its maximum limit, the
drop in the systemfrequency can force to be less than .
In such a case the (15)(24) will continue being used during the
iterative process.
C. DFIG With Primary Frequency Regulation
The deloading of the wind turbine allows a DFIG to provide
frequency support. Its response to frequency deviations will be
dependent on its speed droop characteristic and its power gen-
eration (see Fig. 5) [7], which is calculated according to
the level of deloading of its wind turbine, .
Hence, the active power generated by the DFIG is computed
by using (30)(31), where is the initial active power ob-
tained from the power curve of the wind generator for any given
wind speed:
(30)
(31)
One of the most attractive features of this kind of wind gen-
erator is its control capability of reactive power. The control
strategy adopted to operate a DFIGdenes if it will be absorbing
or injecting reactive power. Therefore, depending on the reac-
tive power control mode, a different set of power ow equations
must be accommodated with the rest of the equations.
1) Fixed Power Factor Control Mode: Given that the power
factor is specied, the reactive power generated by the wind
generator is computed according to ,
where stands for the power factor angle. However, care
should be taken when choosing the value of the power factor so
as not to exceed the reactive power limits, and ,
imposed by the power electronic converter. Therefore, for a
DFIG connected at node , the power ow mismatch equations
that must be solved simultaneously with the ones corresponding
to the rest of the network are
(32)
(33)
(34)
2) Constant Voltage Magnitude Control Mode: In this con-
trol strategy, the voltage magnitude at the generators termi-
nals is specied, and its active power output is obtained from
(30)(31). Therefore, the PFALFC formulation for a DFIG con-
trolling the voltage magnitude at its terminal is given by
(35)
Fig. 6. PMSG-based wind generator model for power ow studies.
(36)
Even though the reactive power mismatch equation of the
voltage magnitude controlled node is not considered in the for-
mulation, it is solved at each iterative step to assess whether or
not the generator reactive power is within limits. If a limit vio-
lation occurs then the generated reactive power is xed at that
limit, and the voltage magnitude is freed. In this case, this model
is converted into the DFIG model operating with a xed power
factor.
Note that the impact of the systems frequency on the stator
electrical variables is considered to be negligible for both
DFIG models. This is a reasonable assumption if we regard the
electrical and mechanical rotor frequencies are decoupled: the
voltage source converter connected to the slip-rings of the rotor
injects a current with a variable frequency to compensate the
difference between the mechanical and electrical frequencies
[11].
D. PMSG-Based Wind Generator With Primary Frequency
Regulation
Presently there is great interest in assessing how these wind
generators can participate in the primary frequency control since
they have been gaining prominence in power systems [13]. Un-
like a DFIG, the PMSG-based wind generator supplies all its
generated active power through its full-scale converter. Based
on the electric circuit shown in Fig. 6, a suitable model for
power ow studies can be derived by explicitly including the
representation of the wind generator step-up transformer, where
stands for the output power obtained from the wind
generator power curve for a given wind speed, and
are the voltages at the machine-side converter and grid-side con-
verter terminal, respectively, and is the step-up transformer
impedance.
Note that this model will also allow for direct voltage magni-
tude control at the transformers high-voltage side [14] for the
cases where this reactive power control mode is selected.
Assuming that the PMSG-based wind generator has a primary
frequency controller represented by Fig. 5, its generated active
power will be determined by
(37)
(38)
CASTRO et al.: SOLUTION OF POWER FLOW WITH AUTOMATIC LOAD-FREQUENCY CONTROL DEVICES INCLUDING WIND FARMS 2191
On the other hand, assuming that , the
active and reactive powers owing from the grid-side converter
terminal to the th bus are
(39)
(40)
For the active and reactive powers owing from bus to the
grid-side converter terminal, the subscripts and are ex-
changed in (39) and (40).
Regarding the adopted reactive power control strategy, a dif-
ferent set of power ow equations must be integrated with the
rest of the network equations as shown next.
1) Fixed Power Factor Control Mode: For a given power
factor angle , the reactive power generated by this type of wind
generator is calculated as . Hence, having
dened both generated powers, the set of power ow mismatch
equations that must be solved when the generator is connected
at node is given by (41)(45) shown at the bottom of the page.
2) Constant Voltage Magnitude Control Mode: Under this
control strategy, the nodal voltage magnitude at the trans-
formers high-voltage side is maintained at a constant value by
adjusting the generators reactive power. In this case, the initial
output of active power is computed using (37) and (38). Thus,
the NR-based power ow formulation for this control strategy
is given by
(46)
(47)
(48)
(49)
Lastly, if the generator cannot provide the reactive power
support to constrain its terminal voltage magnitude at a spec-
ied value, the reactive power is xed at the offending limit. In
this case, the control strategy is changed to operate with a xed
power factor.
V. INITIALIZATION OF WIND GENERATORS
When solving power ows by using Newtons method, ade-
quate initial conditions must be specied in order to achieve an
iterative solution with quadratic convergence. In this context,
the proposals to initialize the wind generator state variables are
stated below.
A good initial value of to execute power ow studies
with a network having SR-FSWGs and PR-FSWGs is given by
, since the operating point of these generators is
normally found close to the nominal mechanical angular speed.
The adopted strategy control must be considered rst when
initializing VSWGs. If a constant power factor is specied, then
the reactive power is computed using the initial active power ob-
tained from the power curve and is kept constant during the iter-
ative process. On the other hand, if a constant voltage magnitude
is selected, the output reactive power is calculated within the it-
erative process as stated in Section IV-CII. Lastly, the voltage
magnitudes are initialized at 1 p.u. at all uncontrolled voltage
magnitude nodes, while the controlled PV nodes are initialized
at specied values that remain constant throughout the iterative
solution if no generator reactive power limits are violated. The
initial voltage phase angles are selected to be 0 at all buses.
VI. CASE STUDIES
The suitability of the proposed approach to conduct a power
ow analysis with automatic load-frequency control is tested
on a three-machine, eight-bus system and the IEEE 14-bus test
system. Data for both systems are given in the Appendix A. For
(41)
(42)
(43)
(44)
(45)
2192 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 27, NO. 4, NOVEMBER 2012
Fig. 7. Power system used to accommodate a PR-FSWG.
the former case, only the PR-FSWG model is examined given
that the analysis of the SR-FSWG was addressed in [2]. The
design of the case studies is given below.
A. Example I
As stated in Section IV-B, the blade pitch angle regulator acts
when the generated active power rises above . In this con-
text, because the power output of the FSWG is frequency de-
pendent as given in (12), the blade angle regulator can be ac-
tivated or deactivated depending on the frequency deviations
in the network, and not only because of wind speeds beyond a
specied rated value. To validate this statement, this example
concerns the performance of the PR-FSWG model when using
the PFALFC algorithm. The eight-bus test system, comprising
two synchronous generators (SG) and one PR-FSWG as shown
in Fig. 7, is used to execute this study. The SGconnected at node
1 is considered the reference generator with automatic load-fre-
quency control capability while the one connected at node 3 acts
as a conventional PV generator controlling its terminal voltage
at 1 p.u.
The disturbances that deviate the frequency from its
nominal value are simulated by altering the active and reac-
tive power demanded by all the systems loads assuming a
constant power factor as shown in Table I. The base case corre-
sponds to the power ow solution obtained by considering that
. For all simulations whose power ow solutions are
reported in Table I for a power mismatch tolerance of ,
the rated values of the active power and the wind speed of the
wind generator are 2 MW and 15 m/s, respectively.
Three key aspects can be drawn from the results presented
in Table I. Firstly, when simulating the PR-FSWG with a wind
speed of 14 m/s, the generated active and reactive powers are
different for each value of . This is primarily because of
the frequency deviations since they force the wind generator to
nd another equilibrium point. Secondly, the blade pitch angle
mechanism is activated when simulating a at
rated wind speed. In this case, the generated active power hits
because of the increment in the systems frequency.
Lastly, when simulating the scenario where the wind speed is
above the rated one, i.e., 16 m/s, a deactivation of the pitch
angle mechanism occurs because of the frequency drop caused
by the increase of 10% in the total system load.
The above conclusions can be physically inferred: a decrease
in the mechanical speed of the induction machine will lead to a
decrement in the output power of the wind generator and vice
versa.
TABLE I
ELECTRICAL RESPONSE OF A PR-FSWG FOR DIFFERENT VALUES
Fig. 8. Modied IEEE 14-bus test system.
The number of iterations required to obtain the solution in
each case is reported in Table I. Note that the proposed ap-
proach arrives at the solution with local quadratic convergence
except when the pitch angle mechanismis activated as described
above; in this case, one or two additional iterations are needed
to obtain the power ow solution.
B. Example II
In order to examine the joint operation of the wind gener-
ator models proposed in the previous section, the IEEE 14-bus
test system [15] is modied according to Fig. 8 to accommo-
date four wind farms. Wind farms I and II are composed of
ve SR-FSWGs and ve PR-FSWGs, respectively, whereas the
remaining are VSWG-based wind farms comprised of fteen
DFIGs and fteen PMSGs, respectively. Generators connected
at buses 1 and 2 are assumed to possess automatic load-fre-
quency control capability. Lastly, the generator 1 is considered
as the reference generator.
In this example, the active and reactive powers generated by
each wind farm as well as the system frequency are reported for
two different load perturbations in the system, that is,
and . In addition, two scenarios were simulated
to demonstrate the power regulation capability of the VSWG-
based wind farms by assuming a deloading of 5% and 10% for
CASTRO et al.: SOLUTION OF POWER FLOW WITH AUTOMATIC LOAD-FREQUENCY CONTROL DEVICES INCLUDING WIND FARMS 2193
TABLE II
POWER GENERATION, VOLTAGES OF WIND FARMS AND SYSTEM FREQUENCY FOR DIFFERENT VALUES OF AND
each wind generator composing these wind farms. The power
ow results of these simulations are given in Table II.
For the case of the FSWG-based wind farms, simulations
show that their active and reactive powers are different from
one scenario to another because of their frequency and voltage
dependency. Additionally, the PMSG-based wind farm controls
its terminal voltage at 1 p.u., as specied in all simulations, by
injecting into the grid the required reactive power.
The results presented in Table II also reveal that a load in-
crease of 15%leads to a systemfrequency beyond 49.5 Hz when
the VSWG-based wind farms are not providing frequency sup-
port. However, when they are participating in the frequency con-
trol, all wind turbines use all their spinning reserve regardless of
the level of deloading, resulting in a systemfrequency of 49.526
Hz and 49.563 Hz for and 10%, respectively. In this
case, the wind power penetration accounts for 30% of the total
system load. The VSWG-based wind farms contribution to the
frequency regulation corresponds to 3 MW and 6 MW of active
power that was used as spinning reserve for and 10%,
respectively.
If frequency response is provided by wind farms then the
power system is clearly strengthened and acquires more ex-
ibility when facing abnormal conditions. The importance of the
deloading of wind generators lies in the fact that the system will
count on other distributed slack generators which will help to
mitigate active power imbalances.
VII. CONCLUSIONS
This paper has proposed the mathematical modeling of
several types of wind generators taking into account their
dependence with respect to system frequency deviations. These
models were implemented in a Newton-based power ow
algorithm with frequency control devices to estimate their
electrical response after the action of the primary frequency
regulation. Through numerical simulations, the PFALFC algo-
rithm developed is shown to maintain its quadratic convergence
characteristic, and the solutions obtained illustrate the contri-
butions to the frequency support of VSWG-based wind farms.
APPENDIX A
Parameters of the SR-FSWG: Generator data [16]:
, , , ,
; 2 MW, 690 V, 50 Hz, 2 pole pairs. Fixed capac-
itor of 0.6 MVAr. Wind turbine data [16]: , ,
; , , , ,
, , , , , .
Parameters of the PR-FSWG: Generator data [16]: the
same as the SR-FSWG. Fixed capacitor of 0.6 MVAr. Wind
turbine data [17]: , , ;
, , , , ,
, , , .
Parameters of the Test System Used in Example I on a
Base Power of 100 MVA: TL1, TL2, and TL3:
, . TL4: .
TR1 and TR2: . TR3: . TR4:
. PL1: . PL2: . The
coefcients for reactive power generation of the generator reg-
ulator are and ; , .
Modied Parameters of the IEEE 14-Bus Test System Used
in Example II on a Base Power of 100 MVA: Step-up trans-
former impedance for each wind farm: .
The transformer impedance for each wind generator is
. The governor droop characteristic for each
DFIG and PMSG is 0.05 p.u. The DFIG and PMSG power
curve is shown in Fig. 9 [18]. The coefcients for reactive
power generation of the generator regulators are (see also
Table III) and . The loads connected at nodes
10, 11 and 12 are dependent on frequency and voltage with
the following parameters: , , ,
, , , and , whereas
the rest are constant.
APPENDIX B
The relevant Jacobian elements which are created and accom-
modated with the original algorithm are the following:
; ; ;
; ; ;
; ; ; ;
2194 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 27, NO. 4, NOVEMBER 2012
Fig. 9. DFIG and PMSG power curve.
TABLE III
GENERATOR CHARACTERISTICS ACCORDING TO THE DELOADING
OF THE VSWG-BASED WIND FARMS
; ;
; ;
; ;
; ;
; ; ;
;
; .
Stall Regulated FSWG: See equation (B1)(B11) at the
bottom of the page.
Pitch Regulated FSWG: See equation (B12)(B14) at the
bottom of the next page.
DFIG:
(B15)
PMSG-Based Wind Generator:
(B16)
(B17)
(B18)
(B19)
(B1)
(B2)
(B3)
(B4)
(B5)
(B6)
(B7)
(B8)
(B9)
(B10)
(B11)
CASTRO et al.: SOLUTION OF POWER FLOW WITH AUTOMATIC LOAD-FREQUENCY CONTROL DEVICES INCLUDING WIND FARMS 2195
(B20)
(B21)
(B22)
(B23)
(B24)
REFERENCES
[1] W. Okamura, Y. Oumar, S. Hayashi, K. Vemura, and F. Ishiguro, A
new power ow model and solution method, IEEE Trans. Power App.
Syst., vol. PAS-94, pp. 10421050, May 1975.
[2] L. M. Castro, C. R. Fuerte-Esquivel, and J. H. Tovar-Hernndez, As-
sessing the steady state post-disturbance condition of power systems
including xed-speed and doubly-fed induction generators, in Proc.
2011 IEEE North American Power Symp. (NAPS), pp. 17.
[3] L. M. Castro, C. R. Fuerte-Esquivel, and J. H. Tovar-Hernndez, A
unied approach for the solution of power ows in electric power sys-
tems including wind farms, Elect. Power Syst. Res., vol. 81, no. 10,
pp. 18591865, Oct. 2011.
[4] G. Ramtharan, J. B. Ekanayake, and N. Jenkins, Frequency support
fromdoubly fed induction generator wind turbines, IET Renew. Power
Gen., vol. 1, no. 1, pp. 39, Mar. 2007.
[5] B. H. Chowdhury and H. T. Ma, Frequency regulation with wind
power plants, in Proc. 2008 IEEE Power Eng. Soc. General Meeting,
pp. 15.
[6] J. Morren, J. Pierik, and S. W. H. de Haan, Inertial response of vari-
able speed wind turbines, Elect. Power Syst. Res., vol. 76, no. 11, pp.
980987, Jul. 2006.
[7] P. Moutis, E. Loukarakis, S. Papathanasiou, and N. D. Hatziargyriou,
Primary load-frequency control from pitch-controlled wind turbines,
in Proc. 2009 IEEE PowerTech, Bucharest, Romania, pp. 17.
[8] A. Teninge, C. Jecu, D. Roye, S. Bacha, J. Duval, and R. Belhomme,
Contribution to frequency control through wind turbine inertial energy
storage, IET Renew. Power Gen., vol. 3, no. 3, pp. 358370, Mar.
2009.
[9] R. G. de Almeida and J. A. Peas Lopes, Participation of doubly
fed induction wind generators in system frequency regulation, IEEE
Trans. Power Syst., vol. 22, no. 3, pp. 944950, Aug. 2007.
[10] R. G. de Almeida, E. D. Castronuovo, and J. A. Peas Lopes, Op-
timum generation control in wind parks when carrying out system op-
erator requests, IEEE Trans. Power Syst., vol. 21, no. 2, pp. 718725,
May 2006.
[11] T. Ackerman, Wind Power in Power Systems. New York: Wiley,
2005.
[12] F. D. Bianchi, H. De Battista, and R. J. Mantz, Wind Turbine Control
Systems Principles, Modelling and Gain Scheduling Design. New
York: Springer, 2006.
[13] J. F. Conroy and R. Watson, Frequency response capability of full
converter wind turbine generators in comparison to conventional gen-
eration, IEEE Trans. Power Syst., vol. 23, no. 2, pp. 649656, May
2008.
[14] L. M. Castro, C. R. Fuerte-Esquivel, E. Barrios, and C. Angeles-Ca-
macho, An integrated power owsolution of exible ACtransmission
systems containing wind energy conversion systems, in Proc. 2011
IET Renewable Power Generation Conf., pp. 16.
[15] Power Systems Test Case Archive, Univ. Washington. [Online]. Avail-
able: http://www.ee.washington.edu/research/pstca/,
[16] V. Akhmatov, Analysis of dynamic behaviour of electric power sys-
tems with large amount of wind power, Ph.D. dissertation, Technical
Univ. Denmark, Electric Power Engineering, Orsted-DTU, Lyngby,
2003.
[17] J. G. Slootweg, H. Polinder, and W. L. Kling, Representing wind tur-
bine electrical generating systems in fundamental frequency simula-
tions, IEEE Trans. Energy Convers., vol. 18, no. 4, pp. 516524, Dec.
2003.
[18] S. Li and T. A. Haskew, Energy capture conversion and control study
of DFIG wind turbine under Weibull wind distribution, in Proc. 2009
IEEE Power Eng. Soc. General Meeting, pp. 19.
Luis M. Castro received the B.Eng. degree from the Instituto Tecnolgico
de Morelia, Morelia, Mxico, in 2006 and the M.S. degree from the Instituto
Tecnolgico de Morelia, Morelia, Mxico, in 2008. He is currently pursuing
the Ph.D. degree at the Universidad Michoacana de San Nicols de Hidalgo
(UMSNH) in the area of wind power integration in power systems.
Claudio R. Fuerte-Esquivel (SM07) received the B.Eng. (Hons.) degree from
the Instituto Tecnolgico de Morelia, Morelia, Mxico, in 1990, the M.S. degree
(summa cum laude) from the Instituto Politcnico Nacional, Mxico, in 1993,
and the Ph.D. degree from the University of Glasgow, Glasgow, U.K., in 1997.
Currently, he is a Professor at the Universidad Michoacana de San Nicols
de Hidalgo (UMSNH), Morelia, where his research interests lie in the dynamic
and steady-state analysis of FACTS.
J. Horacio Tovar-Hernndez received the B.Eng. degree from the Instituto
Tecnolgico de Morelia, Morelia, Mxico, in 1984 and the M.S. and Ph.D. de-
grees from the Instituto Politcnico Nacional, Mxico, in 1989 and 1995, re-
spectively.
Currently, he is a Professor at the Instituto Tecnolgico de Morelia, where his
research interests lie in power systems and electricity markets.
(B12)
(B13)
(B14)