1

Grid Frequency and Voltage Support using PV

Systems with Energy Storage
Ravi Bhatt, Student Member IEEE and Badrul Chowdhury, Senior Member IEEE

energy storage techniques, effective two way communications

Abstract—An optimized operating scheme for a utility connected and a robust demand response program, a grid-tied PV system
community based photovoltaic (PV) system is described. The can create additional value, primarily by enabling increased
system can participate in grid ancillary services like frequency PV participation in grid support functions, such as frequency
and voltage regulation functions based on the Smart Grid
and voltage regulation.
framework. The proposed model comprises of a PV plant with
Li-ion batteries coupled to the grid with the help of a three phase PV generation has increased on a large scale especially in
inverter. A two-way communication between the PV plant and the US and Europe, with the largest plant rated at 60MW at
the grid is assumed. The PV/storage plant provides constant Olmedilla in Spain [2]. If proper control technique is
updates on its current kW/kVar capability and the grid transmits available, such large plants can work hand in hand with the
the demand for specific amounts of power and for specific lengths utility to increase system reliability. The ultimate goal of any
of time. The battery charging energy can originate from either
power system is to maintain a balance between demand and
the PV system or the grid depending on the prevailing energy
prices. The batteries are discharged when two conditions are supply of active and reactive power at any given point in time.
met: the grid requests energy from the community-based PV Any difference between the demanded and the generated
system and if the PV system itself fails to meet the requested kW quantities directly impacts the system frequency and voltage.
or kVar demand. The PV plant and the battery storage are Traditional approach to achieve frequency regulation is to use
integrated with the grid with the help of dc-dc and dc-ac fossil fuel and hydro reserves to generate electrical energy [3];
converters in such a way that bi-directional flow of active and
while the general approach to achieve voltage regulation is to
reactive powers can be achieved. Controllers integrating energy
sources respond to the received signals and attempt to fulfill the use generator var reserves, shunt capacitors, and var
grid demand. The system response is almost instantaneous and compensators.
thus can be very helpful in frequency and voltage regulation. This paper discusses a method to provide active and reactive
power compensation to the grid to participate in frequency and
Index Terms—Frequency regulation, voltage regulation, delta voltage regulation functions. Multifunction approach increases
modulation, dc link voltage control, smart grid, maximum power the utilization factor and improves the quality of the grid [4].
point tracking.
Energy storage systems are promising technologies which may
work in conjunction with PV systems to regulate frequency
I. INTRODUCTION
and voltage.

I n today’s dynamic world, electric utilities are facing

challenges like rising energy demand, increasing fuel costs,
aging assets, pressure to adopt renewable portfolio standards,
II. OBJECTIVES
In conventional power systems, the utility provides ancillary
etc. [1]. Much of this can be tackled without compromising
services such as frequency regulation, voltage support,
the overall performance and service quality of the utility
spinning reserves etc. These services are typically not free,
system. In recent years, the presence of photovoltaic (PV)
i.e., generators are paid for ancillary support in regions which
generations on the utility grid is on the rise. With increase in
have such a market [5]. As shown in Fig. 1, the objective of
PV penetration and the progress of the global PV market, there
this paper is to propose a power management scheme which
is a need to enable the PV systems with features which make
would tap the future energy markets for regulation and create
them smart to create an effective business model. Further, by
an enhanced value for PV systems.
leveraging the Smart Grid technologies, and taking advantage
of the distributed nature of PV, new opportunities to unlock
value can be created. With the implementation of advanced

This work was supported in part by the Intelligent Systems Center of

Missouri University of Science & Technology (Missouri S&T).
R. Bhatt is with the Department of Electrical & Computer Engineering,
Missouri S&T, Rolla, MO 65409 USA (e-mail: [email protected]). Fig.1. Grid Connected PV Systems
B. H. Chowdhury is with the Department of Electrical & Computer
Engineering, Missouri S&T, Rolla, MO 65409 USA (e-mail: In this paper, control over the active and reactive power
[email protected]).
output of a PV system is proposed. With the suggested
techniques, PV systems can deliver a variable amount of
 2

power based on the amount demanded from the grid. This operated in a four-quadrant mode. One can see from Fig. 2
enables load following capability along with frequency that if P1 is the amount of DC power available at the input,
regulation. Thus, voltage and frequency management may be then by introducing the proper phase to the output current, the
possible even under islanded operating conditions. inverter can deliver 9 kVar of leading or lagging reactive
power. If the power at the dc bus (Pdc) is equal to S, then
III. P-Q CAPABILITY OF THE PV SYSTEM inverter loses its reactive power capability; on the other hand
if Pdc is zero, then inverter capacity can be dedicated to
The P-Q capability is the ability of the system to provide provide reactive power only. However there are some losses
active and reactive power instantaneously. An intelligent associated with the inverter, and so, Pdc cannot be zero. The
controller implemented here is responsible for proper inverter size can be selected in order to provide a
generating/absorbing the maximum active power (±P) and specific amount of voltage regulation. The normal practice is
generating/absorbing the maximum reactive power (±Q) that to select an inverter with S = Pdcmax where Pdcmax is the
the system can handle. It measures the power output of the PV maximum active power that the inverter injects into the grid.
system and state of charge of the battery and decides how Over-sizing the inverter just by 5% provide 32% additional
much energy can be delivered/absorbed at a given moment. capacity of reactive power as shown in Fig. 3.
The nominal voltage of the battery is 240 volts and its capacity
is 11 Ah. Assuming that the state of charge of the battery is
50%., the amount of energy that can be absorbed or released
from the battery is 240*5.5Ah = 1.32 kWh; depending on the
charge/discharge rate, its capability can be defined as 15.84
kW for 5 minutes, or 7.92 kW for 10 minutes or 5.28 kW for
15 minutes. System’s capability to absorb energy can further Fig. 3 Effect of Inverter size on the reactive power capability.
be increased by tracking pseudo power point on the PV curve
[6]. The method discussed in [6] shows that active power IV. SYSTEM DESCRIPTION
injected from the PV can be effectively controlled from
A. Overview
maximum power to almost zero. It means power can be
absorbed even if battery is fully charged. Currently, standards A block diagram of the system model is shown in Fig 4. The
such as IEEE 1547 and UL1741 state that the PV inverter system is implemented in a two-stage power conversion, dc-dc
cannot participate in regulating voltage at the PCC [7]. and then dc-ac. It has three independent control loops for the
Therefore PV inverters are forced to operate at unity power dc-dc converters and the three-phase dc-ac inverter. The
factor. However inverters are capable of providing reactive system comprises of:
power along with the active power. The apparent power rating - PV panels with MPPT/pseudo tracking
(S) of the inverter can be resolved into components of P and Q - Batteries with controlled bi-directional dc-dc converter
as: - Three phase inverter with delta modulation scheme

ܵ ൌ ඥܲଶ ൅ ܳଶ (1)

Depending on the inverter rating and the amount of DC power

available at the input, the P-Q capability of the inverter can be
determined as shown in Fig. 2.

Fig. 4. Grid connected PV system with energy storage

For the system shown in Fig. 4, a boost converter is used to

track the maximum power from PV panels; a bi-directional dc-
dc converter connects batteries to the dc link and provides dc
link voltage regulation, and an inverter, controlled by delta
modulation, injects power into the grid at a desired power
Fig. 2 The P-Q capability of a 10 kVA PV inverter.
factor. This topology can provide almost instantaneous power
The circle plotted in Fig. 2 is the boundary of the 10 kVA transfer capability.
inverter. If proper switching is provided, the inverter can be
 3

B. PV Array D. Battery and Bi-directional DC-DC converter

As shown in Fig. 5, a single diode model of the PV cell is In the proposed system, batteries are used to help in
used to simulate PV characteristics in Matlab [8]. The effect of decreasing the impact of intermittency of the PV source.
temperature on panel voltage is taken care of in the model. Batteries possess high energy density and provide power at
almost constant voltage if charging/discharging cycles are
properly controlled. Fig. 7 shows the simple battery model
implemented in Matlab. It is a controlled voltage source in
series with resistance whose control input E is a function of
the amplitude of the current and actual charge (it) of the
battery is a function of the state of charge. If a characteristic of
Fig. 5 Single diode model of PV cell a battery cell is known which is generally given by the
manufacturer, then finding the parameters such as A, B, K and
Current Iph is a direct function of irradiance and temperature E0 is simple and is given in [11].
and the diode exhibits the cell’s p-n junction characteristics. A
GE-PV 200W solar panel is used to build a 2 kW PV array
[9]. To build a 2 kW array, five panels are connected in series
in order to reach the desired voltage (26.3*5=131.5 V), and
two such strings are connected in parallel to deliver the
desired amount of current (7.6*2 = 15.2 A).
C. MPPT Controller
Due to the moment to moment variation of insolation,
temperature and cell characteristics, there is a need to track the
maximum power from the PV array. The dp/dv method, shown
in Fig. 6, is used to track the maximum power output [10]. The
method is simple and independent of panel parameters. The
Fig. 7 Battery model [11].
operating current and voltage of the array are constantly
monitored and the difference in these variables is checked at The parameters of a single Li-ion cell are taken from [11] and
every instant of time. The difference between the current and used to build a battery with the capacity of 11 Ah and nominal
the previous values of power is checked to determine the sign voltage of 240V. Fig. 8 shows the discharge curve and it can
of the slope on the power characteristic curve. The slope thus be seen that after the exponential zone battery voltage is
calculated is used to generate the reference voltage. A PI merely constant. Battery is discharged at 1C rate i.e. at 11A.
controller uses this reference voltage to generate the switching
pulses for the boost converter. A new reference value indicates
that the operating point is moving towards the MPP. This
process is repeated until the slope is zero, i.e., the MPP is
%

tracked.
Volts
Amps

Time (sec)
Fig. 8. Battery characteristics of a 240V, 11 Ah Li-ion battery

The battery thus modeled is integrated into the DC link via the
bi-directional dc-dc converter as shown in Fig. 9. This
converter is responsible for charging or discharging the battery
and regulating the dc link voltage to 300 volts. Therefore,
Fig. 6. MPPT algorithm - the dp/dv method
 4

when the battery is injecting power into the grid, the converter the PV power available at the dc bus to the grid.
will operate in the boost mode, and, when the battery is
absorbing power from the grid or the PV panels, the converter V. SIMULATION RESULTS
will operate in the buck mode. An analysis and study of the complete system is done under
various scenarios and presented as two cases. Each case
simulation is done in Matlab/Simulink R2010a and PLECS
[13].
Case 1: Down regulation with bi-directional reactive power
flow.
Down regulation means that the grid has excess active power
than demanded, and as a result, the system has frequency
increased to some value higher than 60 Hz. The PV/battery
system is simulated under this scenario where there is a need
Fig. 9. Bi-directional dc-dc converter. to absorb a certain amount of active power from the grid. The
system is also simulated for its voltage regulation capability.
E. Control Approach The bottom half of Fig. 10 represents the signal received from
The three converters in the system are controlled by three the utility operator commanding the PV/battery system to
independent control loops. The first control loop is associated switch modes in order to fulfill its demand. The command
with the PV system and responsible for tracking the maximum signal indicates that there is a need for different levels of
power output of the array. As explained earlier, a dp/dv active power with the inverter operating as either an inductor
algorithm outputs a voltage value associated with the or a capacitor. The top half of Fig. 10 is the power observed at
maximum power. A simple PI loop compares this value with the grid side. It can be seen that the controller is very fast in
the measured input voltage and generates a duty ratio. The following the commanded signal.
controlling action is to change the duty ratio of the boost
converter in order to change the input voltage, which is Vmpp.
The controller works effectively if the DC link voltage is
constant. The second control loop regulates the dc link voltage
kW / kVar

by changing the duty ratio of the bi-directional dc-dc

converter. It measures the dc link voltage and compares with
the reference value and processes the error through a voltage
mode compensator in order to generate the required duty ratio.
This control strategy is independent of the direction of power
flow and generates switching signals for both the buck and the
boost modes. An intelligent controller selects the mode of Time (sec)
operation and passes the pulses to a designated semiconductor
switch. The decision whether the converter operates in the
buck mode or the boost mode is based on the command signal
received from the grid. If a regulation signal is not received
and the battery needs to be charged, then the battery SOC is
the variable used in deciding whether the converter needs to
be operated in the buck mode. The third controller is
responsible for converting the dc power from the PV/battery
system into ac for output to the grid at the desired power Fig. 10. Power measured on the AC side
factor. It also acts to convert the ac power from the grid to dc
for charging the batteries. The three phase bridge is controlled Fig. 11 shows the power on the dc side of the system. The
by the delta modulation scheme [12]. To have the desired three phase bridge is operating as an active rectifier and the bi-
reference AC currents at the output, the magnitude and phase directional dc-dc converter connected between the dc bus and
the battery is operating in the buck mode. Power coming from
information is essential. The command signal received from
the PV and the grid is being absorbed by the battery. At t = 1
the grid can be used to generate the magnitude and a phase
sec, there appears a change in insolation from 1000 W/m2 to
locked loop provides the voltage phase angle information. By
800 W/m2 and one can see that the converter injects less power
introducing an additional angle theta, the power factor of the into the battery. At t = 1.4 sec, there is a command signal to
bridge can be controlled. If the sign of the current is changed, absorb reactive power from the grid and at t = 1.8 sec, the
then the bridge operates as an active rectifier and absorbs the inverter injects reactive power into the grid. One interesting
power from the grid. If a regulation signal is not received, the point to note is that although the inverter is injecting reactive
battery is disconnected from the rest of the system. The power, it has no effect on the power at the dc side because of
inverter takes control over the DC link voltage and transfers
 5

the presence of the capacitor at the input of the inverter. The

average dc current going into the battery is only a function of
the active power. The DC link capacitor deals with the
reactive power requirement.
Watts

Time (sec)
Time (sec) Fig. 13 Transient response to perturbation in large signal

Fig. 11 Power at the DC side

Fig. 12 shows the dc link voltage throughout the operation. At

t = 0.25 sec, there is a large perturbation in active power
demand from 1000 W to 5000 W. The spike in the dc link
voltage is a result of this perturbation in demand. One can see
that as soon as the system starts process the reactive power,
the voltage ripple at the dc bus increases because the capacitor
is providing the instantaneous component of the current. Time (sec)
Fig. 14 AC currents at the output

Fig. 14 shows the currents observed on the grid side.

Harmonics in the current are a function of filter response and
switching frequency of the inverter. For low power
applications, MOSFETs are available which can be used at
high switching frequencies. IGBTs, on the other hand, limit
the switching frequency. In the delta modulation scheme, the
switching frequency is variable but the maximum switching
frequency can be controlled.

Case 2: UP regulation with bi-directional reactive power

flow.
During the period of UP regulation, the PV/battery system is
Time (sec) responsible for injecting active power into the grid. As shown
Fig. 12 DC link voltage in Fig. 15, the system responds very quickly to the command
signal received from the utility. The amount of regulation that
Fig. 13 shows the transient response of the dc link voltage the system is injecting is just a part of the overall frequency
under the perturbation of P from 1 kW to 5 kW. The amount and voltage regulation. It is assumed that there are other DGs
of overshoot observed is 11.67 % which is a reasonable which are also under frequency and voltage control and
margin since the usual practice is to keep the overshoot under respond to utility commands in order to provide a viable
20%. The settling time is observed to be 0.13 sec. These regulation amount that the utility might need.
parameters are a result of the system bandwidth and phase The system is capable of operating under a few other scenarios
margin of the plant. By proper tuning of the compensators, the and these are also listed in Table 1.
desired performance can be achieved.
 6

VII. ACKNOWLEDGMENT
The authors are thankful to Dr. Jonathan W. Kimball of
the ECE Department for his support and helpful discussions
during the study.
Watts

VIII. REFERENCE
[1] J. G. Cupp, M. E. Beehler “Implementing Smartgrid communications,”
Burns and McDonnell Tech brief 2008 [Online] Available:
http://www.smartgridnews.com/artman/uploads/1/article-smartgrid-
part2-084.pdf, [Accessed: Apr 20,2011].
[2] D. Lenardic, “Large Scale Photovoltaic Power plants”, May 31,2010
[Online] Available: http://www.pvresources.com/en/top50pv.php,
[Accessed: June 06,2010].
[3] M. L. Lazarewicz, A. Rojas,"Grid frequency regulation by recycling
electrical energy in flywheels," IEEE Power Engineering Society
Time (sec) General Meeting, pp.2038-2042 Vol.2, 10-10 June 2004.
[4] H. Yu, J. Pan, and A. Xiang, “A multi-function grid-connected PV
system with reactive power compensation for the grid,” Solar energy,
79, (2005), pp101-106.
[5] B. J. Kirby, “Frequency Regulation Basics and Trends”, Oak Ridge
National Laboratory Report ORNL/TM-2004/291, December 2004.
[6] V. A. K. Pappu, B. H. Chowdhury, R. Bhatt, "Implementing frequency
regulation capability in a solar photovoltaic power plant," 2010 North
American Power Symposium (NAPS), pp.1-6, 26-28 Sept. 2010.
[7] E. Liu, J. Bebic, “Distribution system voltage performance analysis for
Fig. 15 Up regulation with reactive power flow high-penetration photovoltaics,” NREL/SR-581-42298, Tech. Rep.,
2008.[Online].Available:http://www1.eere.energy.gov/solar/pdfs/42298.
pdf, [Accessed: Mar 15,2011].
By means of proper switching, all the energy sources can be [8] M. Francisco, L. González, “Model of Photovoltaic Module in Matlab,”
isolated from each other. All modes of operation, except Mode Latin American Congress of Students of Electrical Engineering,
1, are possible even when PV is not available. All the results Electronic and Computation, 2005.
shown earlier are based on the assumption that grid is [9] GE PV 200W Solar module characteristics, Available:
http://www.powerupco.com/panels/ge.php, [Accessed: MAR 01,2011].
commanding power only if the PV/battery system is capable [10] T. Esram, P. L. Chapman, "Comparison of Photovoltaic Array
of providing it. Maximum Power Point Tracking Techniques," IEEE Transactions on
TABLE I Energy Conversion, vol.22, no.2, pp.439-449, June 2007.
POWER FLOW UNDER VARIOUS SCENARIOS [11] O. Tremblay, L. A. Dessaint, A. -I Dekkiche, "A Generic Battery Model
for the Dynamic Simulation of Hybrid Electric Vehicles," IEEE Vehicle
Power and Propulsion Conference, VPPC 2007. 9-12 Sept. 2007, pp.
Mode of operation Power Flow 284-289.
1. No regulation signal PV to Grid [12] P. C. Krause, O. Wasynczuk and S. D. Sudhoff, Analysis of Electric
Machinery and Drive Systems, 2nd ed., Wiley-IEEE Press, 2002, p.512.
2. Battery Charging PV to Battery ( Day time) [13] PLECS Blockset - http://www.plexim.com/, [Accessed: APR 23,2011].
Grid to Battery (Night time)
3. UP regulation ( Batt + PV ) to Grid IX. BIOGRAPHIES
4. Down regulation ( PV + Grid ) to Battery
Ravi Bhatt was born in Mumbai, India on June 18,
1986. He received his Bachelor's degree in Electrical
VI. CONCLUSION AND FUTURE WORK Engineering from Mumbai University, India. He is
currently pursuing his Master's degree in Electrical
A new technique to implement frequency and voltage Engineering at Missouri University of Science &
regulation capability in solar photovoltaic power plant in the Technology, MO, USA. Prior to joining the Master's
framework of the Smart Grid is discussed. This method allows program he has worked as an Engineering Intern at
effective control over active and reactive power available from TATA Power Co. Ltd. and SIEMENS Ltd. His
research interests include renewable energy sources
the system. A system comprising of a 2 kW PV array, 2.64 and its integration.
kWh battery with bi-directional dc-dc converter, three phase
inverter and the grid was modeled, and simulated in Badrul H. Chowdhury (M’83, SM’93) obtained
MATLAB. A battery increases the reliability and flexibility of B.S degree from Bangladesh University of
Engineering & Technology in 1981, his M.S. and
the system. Results show that based on the commanded signal, Ph.D. degrees from Virginia Tech, Blacksburg, VA
the PV plant can respond quickly and can participate in in 1983 and 1987 respectively, all in Electrical
regulation. The control techniques adopted are simple, Engineering. He is currently a Professor in the
autonomous and easy to implement. The most important factor Electrical & Computer Engineering department of
the Missouri University of Science and Technology,
in regulation is the size of the system. Depending on the size formerly known as the University of Missouri-Rolla.
of the feeder, market economics, geographical conditions etc. Dr. Chowdhury’s research interests are in power
PV and batteries can be properly sized. The system proposed system modeling, analysis and control, and
is very useful if used as a community based PV system. renewable and sustainable energy integration. He is
a Senior Member of IEEE.
Since the simulation looks promising, the next step is to
check its feasibility on hardware and to incorporate other
energy storage options as well.