1

Analysis of Voltage Stability and Optimal

Wind Power Penetration Limits for a Non-radial
Network with an Energy Storage System
Ha Thu Le, Member Surya Santoso, Senior Member, IEEE

more wind power into power grids. However, wind power

Abstract-- Increasing wind power penetration into power grids plants (i.e. wind farms) possess two major adverse
is desirable because of its environmental and other benefits. characteristics, namely, (i) the high demand for reactive
Unfortunately, wind farms could cause serious voltage stability power, and (ii) the variability and intermittency in the plant
problems due to their output variability and intermittency. It is
power output. If left untreated, these characteristics could
important to minimize impacts on voltage stability. We propose
to use energy storage systems (ESS) to control the intermittency cause voltage stability problems [6-9], namely, (a) voltage
of wind power. The proposed solution, developed and evaluated fluctuation at the point of connection with the power grid, (b)
in connection with the grid voltage stability, comprises three voltage drop (sag) or rise, and (c) voltage asymmetry and
steps: (i) Quantification of wind farm impacts on grid voltage flicker, and even (d) voltage collapse. Given the increasingly-
stability, (ii) Determination of acceptable wind power penetration higher wind power penetration levels, the impacts could be
levels, and (iii) Using ESS to facilitate wind power integration.
more profound. The problem of reactive power compensation
The ultimate goal is to increase wind power penetration while
maintaining and/or improving grid voltage stability. for wind farms has been dealt with by investigators [10-15].
Meanwhile, the problem of wind power variability and
Index Terms— Energy storage, steady state stability, voltage intermittency remains largely unsolved and therefore requires
stability, wind farm, wind turbine, wind power integration, wind special attention.
power penetration limit Using energy storage systems (ESS) to deal with the output
variability and intermittency of wind farms is a logical idea.
I. INTRODUCTION The concept is simple: surplus wind power is stored in the
In the past few years, the wind power industry continues to storage and is released appropriately to compensate for drops
grow at a rapid pace. The world wind power grew at the 5- in the wind power plant output. Though, the implementation
year (2001-2005) average rate of 20.5%, reaching 59,000- of this concept is not simple. Some doubt its technical and/or
MW capacity by the end of 2005. The US wind generation economic viability. Though, few recent studies [16-20] show
capacity now stands at over 9,000 MW with the recent that ESS may be used to reduce variability and intermittency
average growth rate of 21.8%. As of July 31, 2006, Texas, in wind turbine/wind farm power output, to maintain
with a cumulative installed wind power capacity of 2,400 acceptable voltages at the point of connection, and to improve
MW, overtook California (2,323 MW) as the number-one US transient stability of the wind power plants. The ESS under
wind state. Wind farms continue to expand in size and in consideration includes battery, supercapacitor, and
capacity. As a result, the wind power penetration level superconducting magnetic energy storage (SMES) systems.
becomes increasingly higher in many power systems. By the The investigated wind turbine or wind farm is of small size
end of 2005, wind power plants supply about 5.5% of (around 10 MW or less) and connected to a simple or
Germany total electricity consumption, 8.25% of Spain, and simplified power system. The work focus on using the ESS to
21% of Denmark. The wind power penetration levels are “smooth” the wind power plant output and regulate the
expected to continue to increase in the coming years. For voltage at the point of connection, but mostly neglect the
example, a study carried out by the German Energy Agency effect of the combination (i.e. wind farm and ESS) on the
(DENA) in 2005 concludes that, by 2015, wind power would overall grid condition, such as grid voltage stability or
supply around 14% of German electricity demand. In Texas, reliability.
it is expected that between 8% and 11% of the state electricity We believe that an ESS-based solution for controlling a
will come from wind resources by 2015 [1-5]. wind farm output fluctuation is more effective if it is
The environmental benefits of wind power and the needs to developed and evaluated in connection with the grid
diversify power supply sources make it desirable to channel condition. In this way the wind farm, the ESS and the grid are

H.T. Le and S. Santoso are with the Department of Electrical and Computer
Engineering, The University of Texas, Austin, TX 78712.

1-4244-1298-6/07/$25.00 ©2007 IEEE.

 2

treated as a combination, making the solution comprehensive. Case study 1: Voltage Stability analysis of the wind farm
Therefore, in our research, we develop a systematic ESS- impacts on the 6-bus power system
based solution Case study 2: Determination of maximum-possible wind
power penetration level(s)
for improving wind power integration in connection with the Case study 3: Application of an energy storage system for
grid voltage stability. improving wind power integration
Case study 1 aims at identifying and quantifying the wind
II. APPROACH, OBJECTIVES AND IMPLEMENTATION OUTLINES farm impacts on system voltage stability. The results help
clarify and justify the need for installing an ESS. Case study 2
A. General approach determines how much wind power could be integrated to
As mentioned above, our ESS-based solution is developed maintain an acceptable voltage stability condition for the
and evaluated while taking into account the voltage stability system. The information obtained in the first two case studies
characteristics of the power system of interest. We start with is used to determine the size and develop ESS charge-
evaluation of the grid stability condition under the impacts of discharge scheme. Finally, the operation and the
a large wind farm and determination of acceptable wind effectiveness of the ESS are investigated in Case study 3.
power penetration levels for the system. The information
obtained is used in the design and operation of the ESS. III. STUDY ENVIRONMENT AND GENERAL CONDITIONS
Finally, the ESS operation and effectiveness are evaluated
with respect to the amount of integrated wind power and the A. Description of the 6-bus power system
system voltage stability. In this way, the solution comprises The 6-bus power system (Fig. 1) is a strong network with
three steps: (i) Quantification of wind farm impacts on grid two generators which supply power to the loads through three
voltage stability, (ii) Determination of acceptable wind power 138-kV lines. The total line length is 125 km and total load is
penetration levels, and (iii) Using ESS to facilitate wind 203.7 MW and 82.1 MVAR (219.6 MVA). The total
power integration. generation capacity is 270 MVA. The difference between the
Both static and dynamic voltage stability analyses will be total generation and total load (i.e. the spinning reserve) is
used for the solution development. The analyses are initially 18.7%. Generator Swing1 represents a stiff supply grid with
carried out on a small 6-bus power system, then on a realistic the sending end voltage (V1) maintained at 1.0∠00 pu. The
transmission system. In this paper, we present the operational real power output of generator Gen6 is 43 MW
development and evaluation of the three-step solution by and the reactive power limits are ±25 MVAR. The operational
static voltage stability analyses of the small power system. In voltage at Bus 6 is set to be 1.0 pu. No transformer taps are
the next research phases, the solution is refined using steady- set to raise the secondary-side voltages. The system major
state (or static) and dynamic analyses of the real-world load (Load 3) is connected to the main power source (i.e.
transmission system. The 6-bus system is custom-designed Swing1) via a short and strong line (Line 23) which has small
based on the real-world transmission system in such a way so impedance. Load 4 is located near Gen6 which feeds it. Load
that its major properties are analogous to those of the real- 5 is connected at Bus 5 and is fed through a relatively-weak
world system. Therefore, despite its modest size, the 6-bus and long line. This line has significantly higher impedance
system is also realistic and can be used to provide insights to with the length more than twice that of Line 23 and Line 34
actual conditions. The main advantage of using this simple combined. When there is no wind farm or ESS operational,
system is that the system voltage stability characteristics can the system is in normal operation condition with the voltage
be determined analytically (i.e. using by-hand solutions). profile within +/-5% limits (i.e. within the range of
B. Objectives 95%÷105% of the nominal values). This condition is used as
the base case (called Base case 1) for comparison purposes.
The objectives of this paper’s study are:
When the wind farm and/or ESS are concerned, they are
(i) To evaluate voltage stability impacts of a large wind
connected at Bus 5.
farm (140 MW) on the 6-bus power system. The
system real power loss is used as additional
information of the wind farm impacts.
(ii) To determine maximum-possible penetration level(s) of
wind power for the 6-bus system
(iii) To investigate the possibility of using an energy storage
system to increase wind power integration into the 6-
bus system
Fig. 1. Schematic diagram of the 6-bus power system with a wind farm
C. Implementation outlines
Power system data
Three case studies are carried out to meet the three
Generator Swing1: 220 MVA, 18 kV, R= 0.0008713 Ω, Xd = 2 pu
objectives, specifically Generator Gen6: 50 MVA, 18 kV, R= 0.000052 Ω, Xd = 2 pu
 3

Transformer TRF12: 300 MVA, 18 kV/138 kV, GY-GY, R= 0.154%, X=6.0304% output levels. For identification purpose, each of this power
Transformer TRF46: 55 MVA, 18 kV/138 kV, GY-GY, R= 0.3%, X= 10.05%
Line 23: 15 km, R=0.0409132 Ω/km, X=0.375125 Ω/km, B=4.43 microS/km level is associated with a condition called a “case”. For
Line 34: 25 km, R=0.0409132 Ω/km, X=0.375125 Ω/km, B=4.43 microS/km example, in Case 1, the wind farm outputs 5 MW, in Case 28
Line 45: 85 km, R= 0.135343 Ω/km, X= 0.483124 Ω/km, B= 3.425 microS/km
Load 3: 153.5 + j63.9 MVA, PF= 0.923 lagging
the wind farm outputs 140 MW and so on. Note that, apart
Load 4: 41.3 + j15.5 MVA, PF = 0.936 lagging from WF output level, each case has other properties such as
Load 5: 8.9 + j2.7 MVA, PF = 0.957 lagging
Wind farm: 140 MW, PF=1.0
power loss and voltage stability characteristics. Base case 1 is
used for comparison purposes.
B. Modeling of wind farm and ESS
Power loss profile
This study is concerned primarily with static voltage
The system total real power losses (or P losses) for Base
stability analyses. To serve the analyses, the wind farm and
case 1 (i.e. the network without the wind farm) and the cases
ESS are modeled as negative loads with appropriate power
with the wind farm operational are shown in Fig. 2. The wind
profiles. This means that the wind farm is operated at or very
farm first helps reduce the total P losses at its lower power
close to the unity power factor. Today, this is a requirement
production but increases the losses at higher output levels.
(i.e. grid code) and/or a common practice [21-24].
14
C. Simulation package
P loss = 12.703 MW
Power system simulation package named Neplan is used in 12
this study. This simulations package is capable of performing
static and dynamic stability analyses. Matlab is used where 10

Real power loss, MW

necessary. For further information of the software, see [25]
8
and [26]. Wind farm

D. Criteria for evaluating system voltage stability condition 6

A combination of three voltage stability techniques, 4

namely, V-Q curve, V-Q sensitivity, and V-Q modal analysis
Base case, no Wind farm
[27] are used in this study. The system voltage stability 2 P loss = 1.365 MW
condition is evaluated using two main criteria: (i) reactive
reserve margin and (ii) four smallest eigenvalues. The reactive 0
0 20 40 60 80 100 120 140
reserve margin, or Q-reserve margin, is the var-distance Wind farm power output profile, MW

measured from the point on the V-Q curve of interest where Fig. 2. System real power losses under various wind power generation.
∂V / ∂Q = 0 to the zero-Q line (see Fig. 3 for illustration of Q-
Compared to Base case 1, the loss reduction is achieved
reserve margin for Case 28). The Q-reserve margin chosen to
for the wind farm production range from 5 to 30 MW. The
represent the reactive power condition of the system is the
highest loss reduction is 15.8%, obtained when the wind farm
margin calculated at the weakest bus. To determine the
outputs 15 MW. Visibly, the main reason for the loss
weakest bus, a V-Q sensitivity simulation is performed on the
reduction is that the wind farm supplies power to the local
6-bus system under the condition of Base case 1. Bus 5
load (8.9 MW for Load 5), leading to the smaller power flow
exhibits the largest V-Q sensitivity, indicating that it is most
from Swing1 (see Fig. 1 for the 6-bus system component
sensitive to any changes in reactive power (either injection or
layout). With the wind farm operational, most or all feeding
absorption) in the system. Therefore, it is chosen as the
power to Load 5 does not have to go through Line 45, which
indicator of the system reactive power reserve status. For this
is a long line with high impedance (i.e. higher loss).
6-bus system, there are four eigenvalues in total. They are
However, when the wind farm power production becomes
calculated following V-Q modal analysis and are used in
high, the losses rise dramatically, up to 830.6% (12.703 MW).
parallel with the Q-reserve margin to judge the system
The obvious reason is that the wind power now goes in the
stability condition.
reverse direction to the system major load (Load 3) through
the high-loss Line 45. The power flow over Line 34 also
IV. CASE STUDY RESULTS
becomes stronger, causing additional losses on the same line.
A. Case study 1: Analysis of the wind farm voltage stability Evaluation of system voltage stability condition
impacts on the 6-bus power system We now evaluate the system voltage stability condition by
This case study aims at identifying and quantifying the comparing the system reactive reserve margins for four
wind farm impacts on system voltage stability. The results selected cases (i.e. four power output levels) with that of Base
would help clarify and justify the need for mitigation case 1 (69.36 MVAR). Fig. 3 shows that the system has more
measures (i.e. using an ESS). Q-reserve (71.64 MVAR) when the wind farm power
Setting: The 140-MW wind farm (WF) is connected at Bus production is relatively low, such as in Case 6 (WF P=30
5 (Fig. 1). Its power output is gradually increased from 5 MW MW). However, the wind farm quickly exhausts the system
to 140 MW in 5-MW increments. This results in 28 power reactive reserve as more and more of its power is pushed into
 4

the system. The Q-reserve margins for Case 20, 24, and 28 are power losses and two voltage-stability criteria, namely, system
56.19 MVAR, 47.24 MVAR, and 36.54 MVAR respectively. reactive reserve margins and eigenvalues.
The Q-reserve for Case 28 is reduced by 47.4% compared to This finding is an interesting one. To date, wind farm high
that of the base case. This means that the system voltage demand for reactive power is often considered as the main
stability condition significantly deteriorates as the wind farm factor that affects the stability condition of power networks.
output becomes high. The V-Q curves are also narrower, This case study reveals that, even with full reactive
indicating that the system is more vulnerable to voltage compensation, a strong (or excessive) reverse power flow
instability. from the wind farm could also worsen the system stability
condition (by exhausting the network reactive power reserve
120
and increasing power losses).
100

80
B. Case Study 2: Determination of maximum-possible wind
Base case 1
power penetration levels
Reactive power, MVAR

60
Case 28
Case study 2 determines how much wind power should or
40 Q-reserve margin
Case 24 for Case 28 could be integrated while maintaining an acceptable voltage
20
stability condition for the system. The results are used
0 together with the results of Case study 1 in operating the ESS
-20 in Case study 3.
-40 Setting: The wind farm is connected at Bus 5 (Fig. 1). Its
-60 Case 6
power production is gradually increased from 5 MW in 5-MW
-80 Case 20 increments until the system power flow calculation fails to
40 50 60 70 80 90 100 110 120
converge (i.e. the system is close to voltage-collapse and V-Q
Bus voltage, % curve calculation cannot be performed). This means that the
wind farm output is allowed to exceed its original rating (i.e.
Fig. 3. Selected V-Q curves calculated at Bus 5
140 MW). The amount of integrated wind power (i.e.
Note: Case 6: WF P = 30 MW; Case 20: WF P=120 MW penetration level) is determined with respect to different
Case 24: WF P=140 MW; Case 28: WF P=140 MW
voltage stability conditions (i.e. limits) which are represented
TABLE I by the system Q-reserve margins and four smallest
SYSTEM EIGENVALUES FOR FOUR SELECTED CASES eigenvalues. Base case 1 is used for comparison purposes.
Note that, since the system is at peak load and peak
Base case 1 Case 6 Case 20 Case 24 Case 28
WF=0 MW WF=30 MW WF=70 MW WF=100 MW WF=140 MW generation in the base case, this is a lowest voltage stability
3.0513 3.1001 2.9092 2.1708 1.9541 condition without the wind farm (i.e. a conservative situation
13.6490 13.7142 13.5613 11.0932 10.8662 for voltage stability evaluation).
45.5067 45.6042 45.2777 42.8893 42.4347
104.6754 104.7861 104.4714 104.0193 103.5058 Results
The main results obtained for this case study are shown in
Table I shows the system eigenvalues for the above four Fig. 4 and Table II. Fig. 4 illustrates the amount of wind
selected cases. For the 6-bus system, there are four power that can be integrated into the 6-bus system versus the
eigenvalues for a system condition (i.e. Base case 1, Case 6, respective reactive reserve margin. Table II contains the
and so on). The system eigenvalues for the selected cases system eigenvalues for the conditions (i.e. different wind
confirm the findings using the V-Q curves. The eigenvalues power injection levels, or cases) illustrated in Fig. 4. It also
are improved (i.e. larger) for Case 6 where the V-Q curve shows how the eigenvalues change compared to those of Base
indicates better Q-reserve margin, meaning that the system is case 1.
more voltage-stable. All the eigenvalues for Case 20 through
Case 28 are worse than the corresponding values of Base case 200 190

1, indicating that the voltage stability condition is worsen in 180

175
160
these cases. This result agrees with what shown by the
Integrated wind power, MW

160
140
respective V-Q curves. 140
120
120 110
100
Discussion for Case Study 1 100

The wind farm has positive impacts on the network when 80 70

60
its power production is relatively comparable with the local
40
load. It helps reduce real power losses and improve the
20
voltage stability condition. However, as the wind farm power 0
becomes high compared to the local power consumption, the 66.00 56.19 51.94 47.24 36.54 24.21 13.95 2.88
Q-reserve margins, MVAR
strong reverse power flow increases the losses and lowers the
system stability. This judgment is based on the system real Fig. 4. Amount of integrated wind power versus Q-reserve margin
 5

The wind power injection levels shown in Fig. 4 and Table power penetration level for the 6-bus system is around 190
II correspond to the numbered cases, as following: Case 14 MW. Other lower penetration levels may also be defined in
(70 MW), Case 20 (100 MW), Case 22 (110 MW), Case 24 observance of the respective voltage stability limits, but the
(120 MW), Case 28 (140 MW), Case 32 (160 MW), Case 35 three identified penetration levels correspond to three most
(175 MW), Case 38 (190 MW). The eigenvalues for Case 20 important voltage stability conditions of the 6-bus system.
are not shown in Table II due to space limitation.
Discussion for Case study 2
The selected three voltage stability techniques (i.e. V-Q
TABLE II curve, V-Q sensitivity and V-Q modal analysis) can be used
SYSTEM EIGENVALUES FOR DIFFERENT WIND POWER INJECTION LEVELS effectively to determine different penetration levels of wind
power for a power network.
System eigenvalues, MVAR/% For the 6-bus system, three most important penetration
Base levels are identified. Penetration level 1 is 70 MW of wind
case 1 Case 14 Case 22 Case 24 Case 28 Case 32 Case 35 Case 38
1 2 3 4 5 6 7 8
power, or 34.36 % of the system demand. At this penetration
3.0513 3.0427 2.2553 2.1708 1.9541 1.6439 1.2952 0.6330 level, the static voltage stability condition of the original
13.6490 13.6790 11.1799 11.0932 10.8662 10.5378 10.1725 9.5072 system is maintained. At Penetration level 2 (110 MW, or
45.5067 45.5123 43.0661 42.8893 42.4347 41.7903 41.0862 39.8321 54%), the system stability condition is reduced by around
104.675 104.703 104.22 104.019 103.506 102.784 102.004 100.636
25%, but its normal operation is still possible. Penetration
Change in system eigenvalues compared to Base case 1, %
level 3 (MW, or 93.27%) corresponds to a lowest stability
Base
case 1 Case 14 Case 22 Case 24 Case 28 Case 32 Case 35 Case 38 condition of the system. Beyond this limit, the system enters
9 10 11 12 13 14 15 16 voltage-unstable region that could lead to voltage collapse.
0.00 -0.28 -26.09 -28.86 -35.96 -46.12 -57.55 -79.25 The knowledge of these limits (i.e. penetration levels) is
0.00 0.22 -18.09 -18.73 -20.39 -22.79 -25.47 -30.35
useful since they indicate the boundaries for the system safe
0.00 0.01 -5.36 -5.75 -6.75 -8.17 -9.71 -12.47
0.00 0.03 -0.44 -0.63 -1.12 -1.81 -2.55 -3.86
operation. As shown in Case study 3, the limits help
Note: For Column 10-16, plus sign (+) = increase and minus (−) = decrease
determine a reference point for efficient operation of the
energy storage system.
The analysis of all the obtained results reveals three major
C. Case Study 3: Application of an energy storage system for
wind power penetration levels with respect to three critical
improving wind power integration
voltage stability limits for the 6-bus system, as following.
Penetration level 1: 70 MW of wind power, or 34.36 % of In this case study, the operation and the effectiveness of
the system demand (based on the total real power demand of the ESS are investigated.
203.7 MW). The voltage stability condition (i.e. Case 14) is Setting: The wind farm is connected to Bus 5 (Fig. 1). Its
essentially the same as that of the original system (i.e. Base power production for 8 hours is shown in Fig. 5 and Column 2
case 1). The Q-reserve margin is 66 MVAR (Fig. 4), which is of Table III. An energy storage system (ESS), modeled as a
less than that of Base case 1 (69.36 MVAR) by just 4.85%. negative load, is connected at the same bus as the wind farm.
The system eigenvalues are close to those of Base case 1 The storage system is controlled to store and release wind
(Column 1, 2 and 10, Table II). energy according to the schemes developed in Table III where
Penetration level 2: 110 MW, or 54% of the system Column 3 is the charge and Column 4 is the discharge
demand. The voltage stability limit is reduced by around 25% scheme. Base case 1 is used for comparison purposes.
compared to that of the original system (i.e. Base case 1). The Assumptions: (a) It is assumed that the 6-bus system
figure of 25% is based on the fact that the system Q-reserve demand (i.e. total load) and the matching generation stay at
margin for Case 22 (51.94 MVAR, Fig. 4) is reduced by their peaks (i.e. the condition in Base case 1) for the
25.12% and the system smallest eigenvalue is decreased by considered 8 hours. This is a conservative scenario for voltage
26.09% (Column 11, Table II). However, the system normal stability evaluation since the system stability is at a lowest
operation is still possible with the voltage profile within +/- level under the peak-load condition. (b) For each hour of
5% limits. There is no line or generator overload. consideration, the wind farm power output is unchanged. In
Penetration level 3: 190 MW, or 93.27% of the system practice, a wind farm power output changes every minute or
demand. The system reactive reserve is almost exhausted even every second. In this case study, hourly-averaged wind
(2.88 MVAR, Fig. 4) and the eigenvalues are reduced by up to power production is considered. (c) Wind power forecast is
79.25% compared to those of Base case 1 (Column 16, Table 100-percent correct i.e. the power production for any hour is
II). A penetration level higher than this causes the system to predicted with 100-percent accuracy. In practice, wind power
become voltage-unstable and can lead to voltage collapse. forecast inaccuracy may range from 15% to 50%. In the next
Simulation shows that, when the wind farm power is increased research phases, the accuracy of less than 100% will be
to 194 MW, the system power flow fails to converge. considered. (d) The ESS capacity is sufficiently large to store
Calculation of V-Q curve is impossible because V-Q all wind energy as desired. For each hour of consideration, the
sensitivity is too large. This means that the maximum wind ESS charging rate is fast enough to store all surplus energy
 6

during that 1-hour period. power to the system at all times. Certainly, if the wind farm
power production is low, an output of less than 60 MW is
Selection of charge-discharge reference point for ESS
acceptable since the amount does not affect the voltage
The reference point is essential for the ESS operation since
stability condition of the original system.
it indicates when the ESS is to start to store wind power or to
release it. In case a control system for the ESS is modeled Calculation of power charge-discharge profiles for ESS
explicitly (likely in dynamic simulation), this point would be a Given the wind farm 8-hour power profile and the
reference point for the control system. reference point of 60 MW, it is possible to calculate the
It is not straightforward to set this reference point. If it is amount of wind power that should be stored in the ESS and
too low and the wind farm power output is high, a large the amount that the ESS should release during each hour. The
amount of wind power would have to be diverted into the result is shown in Table III. Column 3 shows the amount of
ESS. This means that the ESS size would have to be very wind power to be stored in the ESS for every hour while
large (i.e. higher ESS capital cost). In addition, less wind Column 4 indicates the hourly power that the ESS should
power is channeled into the power system in this situation. If discharge. Column 5 shows the cumulative power in the ESS
the reference is too high, the system stability may be affected i.e. the energy available in the ESS during each hour. The
because of the excessive wind power flow. In [16, 17] the combined hourly power output from the wind farm and the
wind power plant power output of 0.9 pu is the control storage system is calculated and shown in Column 6.
objective. In [19] the real power reference is chosen to be the
TABLE III
wind farm average real power. This study uses a grid-based
CALCULATION OF POWER CHARGE-DISCHARGE PROFILE FOR ESS
approach. It selects the reference point based on the overall Cumulative Combined
voltage stability status of the power network. The core idea is Time WF Power ESS energy in output of
that the reference is chosen so as to meet two criteria Interval, power, to ESS, output, ESS, WF & ESS,
hour MW MW MW MWh MW
simultaneously: (i) to maintain the network voltage stability
1 2 3 4 5 6
and (ii) to channel a maximum-possible amount of wind
1 75 15 0 15 60
power into the system. 2 105 45 0 60 60
3 70 10 0 70 60
120 4 15 0 45 25 60
110
105 5 110 50 0 75 60
100 6 45 0 15 60 60
Wind farm power, MW

7 25 0 35 25 60
80 75
70 8 35 0 25 0 60
60 60 60 60 60 60 60 60
60
45 The wind farm power profile and the calculated charge-
40 35 discharge profile for the ESS are based on the above
25
15
simplified conditions (i.e. the assumptions). They are
20
sufficient for this pilot investigation of the ESS-based solution
0 viability and effectiveness. In the next studies, more realistic
1 2 3 4 5 6 7 8 conditions will be considered, such as different wind farm
Time interval, hour
power output profiles, ESS charging rate, and/or size.
Fig. 5. Uncontrolled WF output and target power profile for 8 hours
Results
Note: Fig. 5 shows the uncontrolled 140-MW wind farm power The main simulation results are plotted in Fig. 6 and Fig. 7.
production for 8 hours (i.e. without the ESS) and the desired power Without the ESS, the results show that the system voltage
profile (i.e. with the ESS). The set of dark-filled bars is the wind farm
power profile and the set of stripe-bars is the target power profile. stability condition varies dramatically as the wind farm power
production goes up and down. At times the system Q-reserve
The results of Case study 1 and Case study 2 (Section A margin is improved by around 3% while at other times the
and B) show that channeling around 70 MW of wind power margin drops by up to 25.13% (Fig. 6, Time interval 5). The
(i.e. 34.36 % of the system demand, or Penetration level 1) system eigenvalues confirm this situation. The smallest
into the 6-bus system could maintain the original stability eigenvalue is reduced by up to 26.09% (Fig. 7, Time interval
condition (i.e. Base case 1). Therefore, to be more 5) as the wind farm outputs up to 110 MW into the system.
conservative, the reference point of 60 MW is chosen for the Operation of the 6-bus system under this condition is clearly
ESS. For example, if the wind farm outputs 75 MW, 60 MW undesirable since the stability condition is deteriorated
is allowed to go freely into the grid and 15 MW is diverted to significantly every time the wind power flow becomes high
the ESS for storing. In the opposite situation, if the wind farm (e.g. 105 MW at Time interval 2, 110 MW at Time interval 5).
outputs less than 60 MW, say 50 MW, the ESS releases 10 With the ESS operational, a steady wind power flow of 60
MW (provided the power is available in the storage) to the MW is maintained for the entire 8-hour period. The system
system. The final goal is to ensure around 60 MW of wind reactive reserve margin stays close to that of Base case 1 with
 7

a slight decrease of 1.54% (Fig. 6). On the other hand, the voltage stability condition slightly. One more advantage of the
system eigenvalues are improved (Fig. 7). This means that the ESS is that it increases the system spinning reserve, thanks to
ESS helps increase the voltage stability limit for the 6-bus the energy available in the storage at certain times. For
system. All the wind power produced by the wind farm is instance, at the 5th hour, the stored energy (i.e. not yet
integrated into the system. released) in the ESS is 75 MWh (Column 5, Table 4.7). This
For the 8-hour period, the ESS releases a total amount of means that, during this hour, the original system spinning
120 MW of wind power (sum of Column 4, Table III). reserve (50.4 MVA, or 18.7%) becomes 125.4 MVA, or
Without the ESS, either this amount of wind power has to be 46.4%. Larger spinning reserve implies better voltage stability
thrown away, or be integrated into the system at the cost of since the system has more power to support voltage and
lowering stability. The maximum amount of energy that the frequency during the time interval.
ESS has to carry is 75 MWh (during the 5th hour, Column 5, By using static analyses, some useful additional
Table III). This means that the ESS would need to be sized at information is also obtained. It includes the total energy
around 75 MWh to meet the storage requirement for the 8 released by the ESS during the 8-hour period and the
hours of interest. This could be an economic size given the maximum energy that the ESS is to carry during an hour-time.
ESS total energy release of 120-MW during the same period. This information may be used to determine an appropriate
The wind farm capacity is 140 MW, so the ESS size is about size for the ESS and in ESS economic analyses.
half of the wind farm rating.
V. CONCLUSION
10
In this paper, a three-step solution using an energy storage
5 Base case 1
system for improving wind power integration is developed. Its
effectiveness is evaluated on a small 6-bus non-radial power
Change in Q-reserve margin, %

0
system that hosts a large 140-MW wind farm. The goal is to
-5
With ESS
integrate a maximum-possible amount of wind power into the
system without compromising the system voltage stability.
-10
The results obtained from the three case studies show that
-15 goal has been achieved and the solution is effective. The
Without ESS
-20
solution will be refined using static and dynamic analyses of a
realistic transmission system and the results will be reported
-25
in the future.
-30
1 2 3 4 5 6 7 8
Time interval, hour VI. REFERENCES
Fig. 6. Change in Q-reserve margin with and without ESS
[1] GWEC, "GWEC Global Wind 2005 Report," Global Wind Energy
10 Council, http://www.gwec.net/fileadmin/documents/Publications/GWEC-
Global_Wind_05_Report_low_res_01.pdf, current 25 Sept 2005.
5 With ESS [2] AWEA, "Wind power outlook 2006," American Wind Energy
Association, http://www.awea.org/pubs/documents/Outlook_2006.pdf,
Change in smallest eigenvalue, %

0 current Sept. 25, 2006.

[3] AWEA, " Wind Energy Projects Throughout the United States of
-5 Base case 1
America," American Wind Energy Association
http://www.awea.org/projects/, current Oct 02, 2006.
-10
[4] DSIRE, The Database of State Incentives for Renewable Energy
-15
(DSIRE), http://www.dsireusa.org/index.cfm?EE=0&RE=1, current
Sept. 25, 2006.
Without ESS [5] "2003 State of the Martket Report for the ERCOT Wholesale Electricity
-20
Market," Potomac Economics, Ltd. 2004.
-25 [6] S. Heier, Grid Integration of Wind Energy Conversion Systems.
Chichester: John Wiley & Sons Ltd., 1998.
-30 [7] T. Ackermann (Ed.). Wind Power in Power Systems. Chichester: John
1 2 3 4 5 6 7 8
Time interval, hour Wiley and Sons, Ltd., 2005.
[8] M. Bruntt, J. Havsager, and H. Knudsen, "Incorporation of wind power in
Fig. 7. Change in system smallest eigenvalue with and without ESS the east Danish power system," in Proc. 1999 International Conference
on Electric Power Engineering (PowerTech).
Discussion for Case Study 3 [9] P. Fairley, "Steady as She Blows," IEEE Spectrum, vol. 40, pp. 35-39,
2003.
Overall, the results obtained in this case study show that [10] Ha Thu Le and T. K. Saha, "Investigation of power loss and voltage
using an energy storage system to increase wind power stability limits for large wind farm connections to a subtransmission
penetration is effective. The ESS helps integrate all the power network," in Proc. 2004 IEEE Power Engineering Society General
Meeting, pp. 2251-2256 vol. 2.
produced by the wind farm during the 8-hour period of
[11] Z. Saad-Saoud, M. L. Lisboa, J. B. Ekanayake, N. Jenkins, and G.
consideration. In addition, the ESS improves the system Strbac, "Application of STATCOMs to wind farms," in Proc. 1998 IEEE
 8

conference on Generation, Transmission and Distribution, pp. 511-516 BIOGRAPHIES

vol. 145.
[12] S. K. Salman and A. L. J. Teo, "Improvement of fault clearing time of
wind farm using reactive power compensation," in Proc. 2001 IEEE Surya Santoso (SM '2002) has been Assistant Professor with Department of
Porto Power Tech vol. 2. Electrical and Computer Engineering, The University of Texas at Austin since
[13] N. Dizdarevic and M. Majstrovic, "FACTS-based reactive power 2003. He was a Sr. Power Systems/Consulting Engineer with Electrotek
compensation of wind energy conversion system," in Proc. 2003 IEEE Concepts, Knoxville, TN between 1997 and 2003. He earned the BSEE degree
Bologna Power Tech Conference vol. 2. from Satya Wacana Christian University, Indonesia in 1992, and the MSEE and
[14] C. Abbey and G. Joos, "Optimal reactive power allocation in a wind Ph.D degrees from the University of Texas at Austin in 1994 and 1996,
powered doubly-fed induction generator," in Proc. 2004 IEEE Power respectively. His research interests include power system analysis, modeling and
Engineering Society General Meeting, pp. 1491 - 1495 vol. 2. simulation of power system phenomena, the development of intelligent systems
for electric power system applications, and wind power integration. He is
[15] S. Kalsi, D. Madura, R. Howard, G. Snitchler, T. MacDonald, D.
coauthor of Electrical Power Systems Quality published by McGraw-Hill, now
Bradshaw, I. Grant, and M. Ingram, "Superconducting dynamic
in its 2nd edition. He chairs a task force on Intelligent System Applications to
synchronous condenser for improved grid voltage support," in Proc. 2003
Data Mining and Analysis, and a member of the IEEE PES Power Systems
IEEE PES Transmission and Distribution Conference and Exposition,
Analysis, Computing, and Economics Committee
pp. 742- 747 vol. 2.
[16] C. Abbey and G. Joos, "A doubly-fed induction machine and energy
storage system for wind power generation," in Proc. 2004 Canadian Ha Thu Le (M '2002) received her BSEE from Odessa College (Odessa city,
Conference on Electrical and Computer Engineering, pp. 1059-1062 Ukraine) and MEEE from the University of Queensland (Brisbane city,
vol. 2. Australia) in 1994 and 2003, respectively. Presently, she is pursuing a PhD
degree at the University of Texas at Austin. Before going back to graduate
[17] C. Abbey and G. Joos, "Integration of energy storage with a doubly-fed
study, she worked in Hanoi-Vietnam, her home country, for over six years. Her
induction machine for wind power applications," in Proc. 2004 IEEE
employment experience includes the Vietnam National Standard Institute (1995-
35th Annual Power Electronics Specialists Conference, pp. 1964-1968
2000) where she involved in standardization work and EAN Vietnam (2000-
vol. 3.
2002) where she provided consulting service to businesses on automatic
[18] C. Banos, M. Aten, P. Cartwright, and T. C. Green, "Benefits and control
identification for supply chain management. She assisted the Transmission
of STATCOM with energy storage in wind power generation," in Proc.
Group in Vietnam-Japan project on training of EVN (Vietnam Power
2006 The 8th IEE International Conference on AC and DC Power
Corporation) engineers in 2004. Her special interests include wind power and
Transmission, pp. 230-235.
renewable energy, wind power integration, power system simulation, and power
[19] F. Zhou, G. Joos, C. Abbey, L. Jiao, and B. T. Ooi, "Use of large capacity quality.
SMES to improve the power quality and stability of wind farms," in Proc.
2004 2004 IEEE Power Engineering Society General Meeting, pp.
2025-2030 vol. 2.
[20] T. Senjyu, T. Kinjyo, K. Uezato, and H. Fujita, "Analysis of Terminal
Voltage and Output Power Control of Wind Turbine Generator by Series
and Parallel Compensation using SMES " in Proc. 2004 35th Annual
IEEE Power Electronics Specialists Conference, pp. 4278-4284.
[21] I. Erlich and U. Bachmann, "2005 Grid Code Requirements Concerning
Connection and Operation of Wind Turbines in Germany," in Proc. 2005.
[22] R. M. Zavadil and J. C. Smith, "Status of Wind-Related US National and
Regional Grid Code Activities," in Proc. 2005 IEEE Power Engineering
Society General Meeting.
[23] E. Fagan, S. Grimes, J. McArdle, P. Smith, and M. Stronge, "Grid code
provisions for wind generators in Ireland," in Proc. 2005 IEEE Power
Engineering Society General Meeting, pp. 1241-1247 Vol. 2.
[24] L. McGrow, "Beat the system (Great Britain Grid Code)," Power
Engineer [see also Power Engineering Journal], vol. 20, pp. 28-30,
2006.
[25] BCP, "NEPLAN Electricity," BCP Busarello+Cott+Partner Inc,
http://www.neplan.ch/sites/en/neplan_elec.asp, current Oct. 18, 2006.
[26] "Matlab and Simulink Featured Products," The Matworks, Inc.,
http://www.mathworks.com/, current Sept 25, 2006.
[27] P. Kundur, Power System Stability and Control. New York: McGraw-
Hill, Inc., 1994.

VII.