DIgSILENT PowerFactory

Application Example

Battery Energy Storing Systems

BESS
 DIgSILENT GmbH

Heinrich-Hertz-Str. 9
72810 - Gomaringen
Germany

T: +49 7072 9168 00

F: +49 7072 9168 88

http://www.digsilent.de
[email protected]

Copyright ©2010, DIgSILENT GmbH. Copyright of this document belongs to DIgSILENT GmbH.
No part of this document may be reproduced, copied, or transmitted in any form, by any means
electronic or mechanical, without the prior written permission of DIgSILENT GmbH.

Battery Energy Storing Systems (BESS) 1

Contents

Contents

1 Introduction 3

2 BESS Simulation Model 3

2.1 Battery Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.2 Voltage Sourced Converter (VSC) . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2.1 Short Circuit Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.3 The BESS Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.4 Testing the BESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

3 Network Model 15

4 Application Case 1: Frequency Regulation 18

4.1 Replacement of Primary Control Energy (Load Change) . . . . . . . . . . . . . . 18

4.2 Replacement of Primary Control Energy (Generator Outage) . . . . . . . . . . . 22

5 Application Case 2: FRT Simulation 25

References 28

Battery Energy Storing Systems (BESS) 2

2 BESS Simulation Model

1 Introduction

Large Battery Energy Storage Systems (BESS) are being increasingly used in Flexible AC
Transmission Systems (FACTS) applications as a way to improve the voltage, frequency, oscil-
latory and/or transient stability of the system and hence enhance the reliability of power supply.
These applications include different FACT controllers, where the storage devices are interfaced
with the power system through either shunt- or series-connected voltage sourced converters
(VSC).

This document describes the assessment of a BESS to replace the spinning reserve/primary re-
serve in a test network. The document gives the reader guidance on how to perform the analysis
in the simulation tool PowerFactory (PDF). This BESS application becomes significantly impor-
tant in small or island power system, with rather low spinning reserve, where load perturbations
have a considerable effect on the network frequency.

In the first part of this application guide describes the relevant characteristics of the test network
and discusses all modelling and project organisational issues that must be taken into account
when setting up the model in PDF. Secondly, all required dynamic models, including the battery
and the shunt VSC controller models, are presented and verified. Finally, different cases are
simulated and results discussed.

For a better usability are the names of the study cases of the project as well as the important
steps in PowerFactory given. This parts are marked in italic. So the reader could use the
document in parallel with the PowerFactory project of the BESS.

2 BESS Simulation Model

A battery energy storing system (BESS) consists of two parts. Firstly a storing part that could
store/restore energy in an electrochemical process. Secondly a rectifier/inverter that could
transform the DC-voltage from the storing part to the AC-voltage needed for the grid and vice
versa. The rectifier/inverter is normally based on a voltage sourced converter (VSC) with a
pulse width modulation (PWM). This element is well known and available in PowerFactory . The
storing part in this case a rechargeable battery is an element that depends on the actual appli-
cation. The problem with batteries is the huge diversity of the technologies and also the variety
inside one technology. So there is no easy, accurate model, valid for all batteries [4][3][2][5].

2.1 Battery Model

There are two main challenges with battery models. The first problem is to get a model that
is not too complex but accurate enough. The second problem is to get the parameters from
manufacturers or own measurements needed for the model. Only a model with appropriate
parameters could deliver good results.

Common batteries in the industry are often lead-acid batteries. But there are also a lot of other
types like nickelcadmium (NiCd), nickel-metal hybrid (NiMH) and several lithium-ion types. Each
type has its own assets and drawbacks.

To design a good model for a battery is a science on itself. Different approaches could be found
in [3], [2] and [5]. A battery model should depict the terminal voltage and the internal resistance
which are a function of several intern-related variables such as the Battery State of Charge
(SOC), the age and temperature of the battery [5]. The State of Charge corresponds to the
current loading state of the battery. The battery is fully loaded if the SOC is one and it is zero if

Battery Energy Storing Systems (BESS) 3

2 BESS Simulation Model

the battery is empty.

A simple electric equivalent for a battery together with a more complex equivalent with parasitic
path is shown in Figure 2.1. It depends on the application case which model will be used. There
is an important equation for the capacity of all lead-acid batteries, called ”Peukert’s law” (1). It
expresses the capacity of a battery in terms of the rate at which it is discharged. As the rate
increases, the battery’s available capacity decreases.

Cp = I k t (1)

Cp is the capacity according to Peukert, at a one-ampere discharge rate, expressed in Ah.

I is the discharge current, expressed in A.
k is the Peukert constant, dimensionless.
t is the time of discharge, expressed in h.

For a lead-acid battery, the value of k is typically between 1.1 and 1.3. The problem with
Peukert.s law is, that at a discharge current of zero, the available capacity becomes infinity.

Figure 2.1: Battery Electric Equivalent, (a) simple, (b) with parasitic reaction [5]

The discharge voltage of a battery is also dependent on the SOC as exemplarily shown in
the picture below. It is obvious, that the battery voltage is nonlinearly dependent on the SOC,
especially for 0 < SOC < 0.2.

Battery Energy Storing Systems (BESS) 4

2 BESS Simulation Model

3 2,5
Internal Resistance
Discharge Voltage
2,5

2
Resistance [ W ]

Voltage [V]
1,5 2

0,5

0 1,5
0 0,2 0,4 0,6 0,8 1
State of Charge (SOC)

Figure 2.2: Typical Discharge Profile of a Lead-Acid Battery [1]

Also the internal resistance is dependent on the SOC (Figure 2.2). Again the nonlinearity is
obvious, especially for low values of the SOC.

There are no general empirical equations available to describe the behaviour of arbitrary batter-
ies. But if there is a detailed datasheet for one specific battery it is possible to use that data in a
PowerFactory DSL model. The data could be stored in a one- or two dimensional array and then
used in a DSL model. Such a battery model would be more accurate for one specific battery.
But it is not easy to obtain the needed manufacturer data.

In the following a simplified battery model for a showcase BESS is presented. Some assump-
tions are made to get a simple, but functional model. At first, it is assumed, that the battery is
only discharged down to 20%. In that case the voltage could be assumed as linearly dependent
on the SOC. Additionally the internal resistance is assumed as constant because it has to be
anyway very small due to the high current application. The battery capacity is assumed as con-
stant, this is valid if the discharge current is known in advance. So the expected capacity could
be calculated and inserted in the data of the model. If the battery is discharged at other rates
than assumed before the capacity has to be recalculated and changed in the battery model
parameters.

With the assumptions above the battery model will be similar to the equivalent simple model
in Figure 2.1 with a constant inner resistance Z(s,SOC)=Z and a voltage source with a voltage
dependent on the SOC. This is detailed enough for an application case but for a detailed study
is also a detailed battery model needed. The equation for the model is given in 2. The SOC is
calculated with an integrator, counting the current of the battery.

UDC = Umax · SOC + Umin · (1 − SOC) − I · Zi (2)

Battery Energy Storing Systems (BESS) 5

2 BESS Simulation Model

The implementation of the battery model is done in DSL (Figure 2.3). Only the DC-current is
needed as input signal to the model. The model gives out the outer DC-voltage, the SOC and
also the DC-cell-voltage which is may be needed for a charge controller.

Variable Unit Description

u min V Cell voltage of discharged cell.
u max V Cell voltage of fully loaded cell.
CellCapacity Ah Capacity of one cell for the used discharge current (see Peukert’s law)
CellsParallel Int Amount of parallel connected cells (increases the total capacity).
CellsInRow Int Amount of serial connected cells (increases the output voltage).
RiCell Ohm Internal resistance per cell (average value).
Unom kV Voltage rating of the connected DC-busbar.
SOC0 Int State of Charge at the beginning of the Simulation.

DIgSILENT
SimpleBattery:

kA to A

K K u_Ri - Lim >min Ucell

0
RiCell 1000

0
u_intern

Normalize Vterm
1
CellsInRow ,Unom

I 1/K 1/sT 1/K

CellsParallel -3.6 CellCapacity internVoltage
u_min,u_max
kA

kA to A -> Ah

SOC - State of Charge

2

ICell
3

Figure 2.3: Simplified Battery Model in DSL (∗ .BlkDef)

The integrator (1/sT) that counts the current in- and output of the battery needs an initial condi-
tion (SOC0). This value is part of the model parameters. During the initialisation process it is
assumed, that the battery is not being charged or discharged. So the active power exchange of
the BESS is zero for the load flow.

The battery appears in PowerFactory in the single line diagram as a DC-voltage source. To apply
the voltage output of the battery model (Figure 2.3) to the DC-voltage source is a DSL-frame
needed (Figure 2.4).

Battery Energy Storing Systems (BESS) 6

2 BESS Simulation Model

Vterm DC-Voltage Source

0
ElmDcu

I Battery_Model Ucell
DC-Current Measurement ElmDSL 1 0
StaImea*
SOC
2 1

ICell
3 2

Figure 2.4: Frame for the Battery Model in DSL (∗ .BlkDef)

2.2 Voltage Sourced Converter (VSC)

The VSC converts the DC-voltage from the battery to an AC-voltage through fast switching of
IGBT-valves.

Figure 2.5: Equivalent Circuit of a PWM Converter

The switching signal, also called amplification factor (Pm), comes from the control system of the
VSC. This could also be expressed for a fundamental frequency model as:

UACr = K0 P mr UDC (3)

UACi = K0 P mi UDC (4)

UACr and UACi is the real, respectively imaginary part of the AC-voltage. Pm is the already
mentioned amplification factor and UDC is the DC-voltage of the battery. K0 is a constant
factor, depending on the used PWM-method. For sinusoidal modulation:

√
3
K0 = √ (5)
2 3

For |P m| ≤ 1 the equations above are valid. If Pm becomes bigger than one there are problems
with harmonics. So it has to be secured, that the DC-voltage is always over a certain level:

√
2 2
UDC ≥ √ | UAC | (6)
3

Battery Energy Storing Systems (BESS) 7

2 BESS Simulation Model

This could also be showed in Figure 2.6.

Figure 2.6: Saturation of a VSC with Sinusoidal PWM

The PowerFactory PWM-converter model has different possible input combinations. For that
case the ”id ref” and ”iq ref” input is used. This is the current phasor of the PWM-converter
transformed in a dq-reference frame. The advantage of the transformation is that ”id ref”
equates to the active power output of the converter and ”iq ref” equates to the reactive power
of the PWM-converter. So it is easy to implement a controller for the BESS. The PWM-model
uses an internal current controller to generate the actual signal for the modulation factor.

2.2.1 Short Circuit Calculation

Short circuit calculation according to IEC60909 / VDE0102 is not possible with a BESS because
devices with power electronics like the PWM-converter are not considered by the norm. But
PowerFactory offers also the possibility to calculate a short circuit according to the so called
complete method. The complete method takes also the load flow into account. Furthermore it
is possible to configure the PWM-converter model for the complete short circuit method. The
two options are constant current or constant voltage. For an IGBT-based converter would be
”constant I”. the right choice because the valves of the converter are only designed for a certain
maximum current. To obtain the maximum current with the complete short circuit method the
PWM-converter has to be configured to full active power on the load flow page. Better results
could be achieved with a RMS simulation this is described in chapter 5 (only symmetrical cases).

2.3 The BESS Controller

Before a controller could be implemented it has to be clear what the task of the controller is. In
the case of the BESS with an IGBT-based converter there are two current parameters to control,

Battery Energy Storing Systems (BESS) 8

2 BESS Simulation Model

one in the d-axis and one in the q-axis. This equates to the real and reactive power. With the
real power output it is possible to control the frequency of a grid with synchronous generators.
With the reactive power could the AC-voltage be controlled. The reactive power output of the
BESS is not bound to the battery capacity, so the AC-voltage could be controlled continuously.
But the apparent power of the PWM-converter is limited and so the reactive power (7).

p
S= P 2 + Q2 (7)

There are more boundary conditions that have to be considered:

• The BESS could only consume active power if the battery is not fully loaded (SOC < 1).

• The BESS could only supply active power if the battery is not discharged (SOC > 0).
• The battery should be recharged if the SOC is below a certain level. So that the BESS
has always the ability to control the active power in both ways.
• The total output of active and reactive power should together not be bigger than the ap-
parent rated power, according to (7). Hence a priority for the active or reactive power is
needed.

All the conditions could be fulfilled from a charge controller. For the showcase it was assumed,
that the SOC is available as a signal. In a real application the SOC has to be calculated from
the battery current and voltage.

When all tasks and conditions for the controller are known the design in DSL could be started.
For a better overview it is recommended to divide the whole controller in smaller parts:

• Frequency controller (Frequency Control)

• Active/reactive power controller (PQ-Control)
• Charging controller (Charge Control)

Together with the needed measurement devices we get the structure showed in Figure 2.8.
This structure is called a frame in PowerFactory . A frame and also a DSL model (like the
battery model showed in Figure 2.3) are only the definition (∗ .BlkDef), like a type of a line or
a transformer. The entity of a frame is a composite model (∗ .ElmComp). The entity of a DSL
model is a common model (∗ .ElmDsl).

Battery Energy Storing Systems (BESS) 9

2 BESS Simulation Model

fgrid Frequ. Pref

Controller

Pmeas
Umeas PQ
Controller

idref iqref

Charge SOC
Bat-Model
Controller
UDC
Id-limref Iq-limref
IDC

Figure 2.7: Structure of the BESS

AC-Voltage
StaVmea PQ-Measurement
*
p
u

Frquency Measurement
ElmPhi*
0
fmeas

1
id_ref1;iqref1 id_ref;iq_ref
0 0 PWM-Converter
PQ-Control ElmVsc*
Frequency Control dpref ElmDsl*
2
ElmDsl*
dv ref 3
4

Charge Control
ElmDSL*
Ucell
0 1

Battery Model SOC

1 2
ElmComp,ElmDSL
1
Icell
2 3

deltai

Figure 2.8: Frame for the BESS-Controller (∗ .BlkDef)

The frequency controller is a simple proportional controller with a small dead band. In a real

Battery Energy Storing Systems (BESS) 10

2 BESS Simulation Model

power system it is important, that there is only one integrator that controls the frequency. Other-
wise there could be problems with oscillations. The model of the frequency controller is showed
below. The droop defines how much active power is activated in a case of a frequency devia-
tion. Granted that K = 0.04 then the full active power of the BESS is activated if the frequency
deviation is equal or greater than 2 Hz (in a 50 Hz system). All values are per unit values. The
droop should be coordinated to fulfil grid code requirements in a certain application case.

The command variable is f0, during the initialisation process this value will be fixed to frq (nor-
mally 1) with the command inc(f0) = frq. The block ”offset” with the output ”p 0” is used to
compensate ”dpref” if that value is not equal to zero after the load flow (because ”p order frequ”
is always zero after the initialisation).

offset
f0
0

p_0
1

frq - df Dband(K) fdead 1/K p_order_frequ dpref

1
db droop

-1

Figure 2.9: Model of the Frequency Controller (∗ .BlkDef)

The controller for the active and reactive power is a little bit more complex. It is showed in Figure
2.10. The control deviation is filtered with a PT1 element. After that the signals are used as
a input to a PI-controller. In the active path is the signal ”delta i” added. In Figure 2.8 could
be seen, that ”delta i” comes from the charge controller. The voltage (or Q) controller has a
very slow I-controller for set point tracing and a slope with a dead band for proportional voltage
support.

deltai
0 id_max

pin - dp (1/(1+sT)) yi1 yi2 {K+1/sT)} id_ref

1 0
Tr Kp,Tip

dpref
2 id_min

iq_max
vref
3

vin - dv yi o11 iq_ref

4
(1/(1+sT)) Deadband_Offset_Lim 1
Trq AC_deadband,Kq
o1

iq_min
iq_max

{1/sT}
Tiq

iq_min

Figure 2.10: Model of the PQ-Controller (∗ .BlkDef)

The charge controller consists of two parts (Figure 2.11). One charging logic to achieve the

Battery Energy Storing Systems (BESS) 11

2 BESS Simulation Model

already described boundary conditions and one block that limits the absolute value of the current
order according to (7). The active current (d-axis) has always the higher priority than the reactive
current (q-axis).

There are only four parameters needed. The charging current (ChargingCur), the minimal SOC
(minSOC), the maximum SOC (maxSOC) and the maximal absolute current (maxAbsCur). The
signal ”deltai” is the difference of the reference d-current from the PQ-controller and the modified
d-current from the charging logic. The feedback of that signal to the PQ-controller prevents a
windup of the PI-controller.

- deltai
0

maxAbsCur

id_ref_in
0 0

ChargeCtrl idin id_ref_out

0 0 1
ChargeCur,minSOC,maxSOC
SOC
1 1

Current Limiter

iq_ref_in iq_ref_out
2 1 1 2

Figure 2.11: Model of the Charge Controller (∗ .BlkDef)

To get a working model of the controller in PowerFactory it is important to use the right names
for the signals in the frame definition. Only signals with the same names in the frame and in the
model, inserted in the frame are connected. Furthermore it is important to understand how the
initialisation process works. The first step of a simulation is the initialisation. PowerFactory uses
for that a load flow. If the load flow converges the results are used inside the DSL models to set
all integrator outputs. A starting point of a simulation should always be a balanced state. So all
derivatives have to be zero, i.e. all inputs of integrators have to be zero too. The inputs, known
from the load flow results are in the composite model of the controller (Figure 2.8):

• id ref/iq ref - (the currents in the dq-frame in p.u.) known from the PWM-converter

• fmeas - (the frequency in p.u.) known from a PLL measurement device

• u - (the absolute AC-voltage in p.u.) known from a voltage measurement device

• p - (the active power in p.u.) known from a power measurement device

The slot ”Battery Model” is filled with the composite model of the battery (Figure 2.4). From
there is after the initialisation load flow also the current known.

After designing the frames and the models (i.e. the types) the composite and common models
could be created. There are two composite models, one for the battery (Figure 2.12) and one
for the BESS controller (Figure 2.13). The slots are filled with the appropriate entities of the
DSL models. The measurement devices have to be created and connected properly. A voltage

Battery Energy Storing Systems (BESS) 12

2 BESS Simulation Model

measurement device has to be connected only to the proper bus. But for a current or a power
measurement device it is also important to select the proper cubicle on the bus. Otherwise the
prefix of the measured signal could be wrong and so the whole control strategy.

Figure 2.12: Composite Model of the Battery (∗ .ElmComp)

Figure 2.13: Composite Model of the BESS Controller (∗ .ElmComp)

Battery Energy Storing Systems (BESS) 13

2 BESS Simulation Model

PWM Converter

Bus 4
DC-Terminal

V
Battery

Figure 2.14: The BESS in the Single Line Diagram

The model for the BESS is now complete; only the parameters and the configuration of the
elements in the single line diagram are missing. All elements in the single line diagram have
different tabs for the basic data, the load flow etc. For the DC-voltage source only the Basic
Data has to be configured. Only the value of the nominal voltage (Unom) is of interest. In the
showcase is the value 0.8 kV. The data for the battery model should be chosen so that equation
(6) is fulfilled for all charging states. Otherwise the PWM converter saturates and the control of
the whole BESS will not work properly.

The PWM-converter has also to be configured. On the basic data tab the AC- and DC-voltage
(Unom/Undomdc) and the rated power (Snom) have to be inserted. The data on the load flow
page is important for the initialisation load flow. One control condition should be P = 0 as already
mentioned (this is the condition for the d-axis). The other condition could be a certain value for
the reactive power or an AC-voltage on a selected node (q-axis). In case of the PWM-converter
the data on the RMS-Simulation tab is important too. There should be the ”Use Integrated
Current Controller” activated. The gain is 1 for both axis; the integrator time constants are both
1 ms, a very fast inner control loop for the current.

The parameters of the frequency, PQ and charge controller depends of the requirements of
the application case. The battery parameters depend on the selected battery and on the used
model for that battery.

It is important to consider the limitations of the battery and the PWM-converter during operation.
For that the single controllers are equipped with limiters. For realistic simulation results these
limits should be set right. Some batteries could be discharged with a higher current than they
could be charged. For that case the PQ-controller needs other limits for the positive d-current
than for the negative d-current.

In some cases it may be also high active negative control power needed. In that case a copper
inside the DC circuit could consume that power. But normally a BESS is needed to prevent a
frequency under-run. Then a copper is not needed.

2.4 Testing the BESS

Study Case: Testing Case

Battery Energy Storing Systems (BESS) 14

3 Network Model

Before the model could be used in a bigger simulation grid it should be tested in a small testing
environment. That testing grid should be simple so that the right functionality of the BESS could
be verified. The grid consists of three loads, one synchronous machine and one BESS. ”Load
1” with 110MW active and 10MVar reactive power and ”Load 2” with 30MW active and 10MVar
reactive power are connected to the terminal. After initialisation of a RMS simulation (root mean
square based values) the synchronous machine supplies 140MW active power to the grid. The
BESS supplies no active power. After 120 s the load ”Load Step” is connected to the grid.
This is equal to a power plant outage with 25MW active power. Hence the turbine governor
is configured as very slow the BESS supplies the missing power. After 540 s the ”Load 2” is
reduced to zero with a slope within 60 s, this could be another power plant that is activated. The
BESS starts loading after that.

The first step of a dynamical simulation is the initialisation in PowerFactory . For that the ”Calcu-
late Initial Conditions” button ( ) is used. If all elements are initialized (no errors in the output
window) then the simulation could be started with the ”Start Simulation” button ( ). The simu-
lation results could then be found in the virtual instrument panel named ”Testresults” in the lower
left area of the PowerFactory window.
Load 1

Bus 3

T3 Coal-Fired Power Plant

G3 ..
G
~
Load 2

T4
Load Step

PWM Converter
Bus 4

DC-Terminal
Terminal

V
Battery

Figure 2.15: Small Testing Grid for the BESS

3 Network Model

The investigated network consists of three power plants; one coal-fired power plant, one fast gas
turbine and one medium fast hydropower plant (Figure 3.1). The three plants are connected on
the 230 kV level via a closed ring with overhead lines. The loads are aggregated to three loads
(Load A to C), also on the 230 kV level.

Battery Energy Storing Systems (BESS) 15

3 Network Model

With PowerFactory it is easy to get a full report of all installed active/reactive power, the in-
stalled capacity and the spinning reserve in the model (Output Calculation Analysis -¿ Grid
Summary).

The maximum load is assumed to 345 MW, for that all three power plants are needed. The
minimum load case is 2/3 of the maximum load. During the low load period the power of the gas
turbine is reduced, because the gas turbine is normally the most expensive plant. For different
load scenarios Operation Scenarios are used in PowerFactory .

Load C
Line 3 Line 4
T3
Coal-Fired Pow er Plant

G3 (coal)
G
~
Bus 3
T2
G2 (gas)

G
~

Bus 8
Line 2

Gas Turbine
Bus 2

Bus 7

Line 5

Bus 9
Bus 5 Bus 6
Line 1

Line 6

Load A Load B Big Industry

Bus 10
T1

Bus 1

Hydropow er Plant
G
~
G 1 (hydro)

Figure 3.1: Single Line Diagram

For a steady state load flow analysis ( ) are no further dynamic models of the power plants
needed. Only the generator types of the power plants and the information on the load flow page
has to be configured. In that information is also the capability limit curve for reactive power
included. It is important to have one reference machine in the grid. The generator from the
hydropower plant is the reference in the sample grid. The other generators are configured as
PV-nodes, they control their active power output and the local voltage.

If the load flow is calculated with the option ”Active Power Control according Secondary Con-
trol” then it is possible to configure the participation of the plant on the compensation of a load
change with the variable ”Primary Frequency Bias” (Kpf). An outage of a generator is for exam-
ple a very big load change in the network. With the configuration of Kpf is it possible to define a
maximum frequency deviation. If an outage of the gas turbine (60 MW) should be compensated
from the coal power plant and hydropower plant within a frequency deviation of 1,5 Hz, then the
Kpf-factor has to be 20 MW/Hz (=60 MW / 2 / 1,5 Hz).

The transformers have automatic tap changers to control the voltage on the high voltage side.
To make sure, that the tap changers work automatically and that the generators are inside their
reactive power limits the options ”Automatic Tap Adjust of Transformers” and ”Consider Reactive

Battery Energy Storing Systems (BESS) 16

3 Network Model

Power Limits” have to be activated in the load flow calculation dialogue.

It is important, that the original grid is configured right. Otherwise it is later together with the
BESS more complicate to find errors or problems. So the first step should be a Data Verification
( ). If there are no errors or warnings, then as a next step a load flow with the original setup
is recommended. If the option ”Show Verification Report” on the ”Output” tab of the load flow
command is checked, then there are only overloaded (more than 80% loading) elements printed
in the output window. It is also important, that there are no warnings in the output window after
the load flow.

For a planned expansion of the grid, like in this case with a BESS it is recommended to use
the variations in PowerFactory . So the original grid could be easily restored and fast switching
between different expansion stages is possible.

To simulate different cases (generator outage, short circuits etc.) in the time domain, different
study cases are used. So each case could be treated separately with a different set of simulation
events, result plots, active operation scenarios and variations. It is generally recommended
to work with study cases, operation scenarios and variations. Operation scenarios are used
to simulate different operating points with the same grid setup. So the original setup is not
changed since all parameters are stored in the appropriate scenario. To manage changes in
the grid setup are variations used. With variations could be different expansion stages of the
original grid managed without a change on the original grid. In PowerFactory the variations are
connected to a certain (user defined) date and time. Each study case has its own time also
set by the user. With the study case time could be different expansion stages of one variation
handled.

Up to that point only a static load flow was considered. For a RMS-simulation in the time domain
are however further detailed models of the power plants needed. The models are implemented
in DSL (DIgSILENT Simulation Language). For the different power plants are pre defined and
standardized models used. The average voltage controller (vcolIEEET1), the turbine governor
(pcuIEEG1) and the synchronous machine (G3) are connected in a composite model (Figure
3.2). A power system stabilizer (PSS) is not used in this case; therefore the pss-slot is empty.

Figure 3.2: Composite Model - Power Plant Control (∗ .ElmComp)

Battery Energy Storing Systems (BESS) 17

4 Application Case 1: Frequency Regulation

The turbine governor controls the speed of the turbine and so, via the synchronous machine the
electric frequency. In case of a fast load change, for example the outage of a power plant, the
frequency rapidly drops due to the lack of active power in the grid. To prevent a slowdown of the
whole system the turbine governor has to activate more power. This is called primary frequency
control or also spinning reserve, activation has to be instantaneous. This means in reality that
a coal-fired power plant has all the time a loss of energy due to the permanent choking on the
turbine valve. So it would be more economic if another device could deliver the primary control
energy until the coal power plant has increased the active power. This is called secondary
frequency control; activation has to occur within 15 minutes after the frequency deviation. To fill
that gap a battery energy storing system (BESS) could be a solution.

4 Application Case 1: Frequency Regulation

Application case 1 consists of two parts. The first part is a simulated load change for that are two
study cases used in the PowerFactory project; one with and one without a BESS. The second
part is a simulated generator outage of the fast gas turbine. For that are also two study cases
used.

4.1 Replacement of Primary Control Energy (Load Change)

Study Case: Industry Load without BESS; Industry Load with BESS

Primary control energy is needed in a grid to control the frequency within a very small gap
between an upper and a lower frequency limit. In the UCTE-grid this margin is only 200 mHz in
the positive and negative direction. The worst expected case in the UCTE is the double outage
of a nuclear power plant. This would lead to an active power gap of 3 GW. The 3 GW have to
be activated within 30 seconds and the power has to be continued for at least 15 minutes. But
there are a lot of power plants who do that primary control, so the control power of each plant
is not so big. The demand from the UCTE is that each grid operator has to deliver 2% of his
actual produced active power as primary control power. This primary control power could also
come from a BESS.

If a BESS is used for primary control, then it should be known how much active power is needed.
With the known active power for the worst case and the time (as mentioned 15 min in the UCTE)
is it possible to calculate the needed capacity of the battery. Normally one single battery cell has
a DC-voltage between 1.2 and 3.4 V dependent on the battery technology. The capacity per cell
varies from a few mAh up to several 100 Ah. So a battery of a BESS is always a combination
of parallel and serial connected single cells. The more batteries are connected in parallel, the
higher is the total capacity, the total inner resistance will be lower. The current per cell will also
be lower. But there are also serial connected cells needed to get a higher total DC-voltage. With
the DC-voltage is also the internal resistance increased.

As described in (6) the DC-voltage shouldn’t under-run a certain level. So there should be
enough cells in series to ensure this. The AC-voltage level depends on the case and so does
the DC-voltage level.

If the BESS is used for primary control, then the battery technology should support constant
loading/unloading cycles without a (fast) degeneration in terms of a so called memory effect.
The control power is activated if the frequency deviation is more than 10 mHz or 0.0002 pu in
a 50 Hz system. So the dead band showed in Figure 2.9 has to be very small. The variable
”droop” has to be 0.004, i.e. activation of the full active power (of the PWM converter) within
200 mHz frequency deviation.

Battery Energy Storing Systems (BESS) 18

4 Application Case 1: Frequency Regulation

For the following investigation is the already described grid used. The PWM-converter has a
rated power of 30 MVA. This corresponds with the already mentioned frequency control bias to
a frequency bias of 30 MW/0.2 Hz =150 MW/Hz. To test the BESS a sudden load change of 30
MW active power is simulated (simple connection of a additional load).

During the simulation without the BESS (study case: ”Industry load without BESS”) all governor
controllers are tuned with standard values. It is obvious, that a load change of nearly 10% of
the total grid load causes a big frequency under-run (red curves in Figure 4.2 and Figure 4.3).
The gas turbine is important for the grid because it could increase the active power output much
faster than the hydropower plant or the coal power plant. A under frequency of nearly 49 Hz
would lead in the UCTE grid to severe action.

In the second simulation (study case: ”Industry load with BESS”) the BESS replaces the primary
control power of the coal-fired power plant. The turbine governor control of the coal-fired power
plant is modified for that. The controller gain is reduced from the standard value (K = 25) to K
= 1. The participation of G3 on the primary control is now reduced to nearly zero. With the fast
BESS the frequency under-run is stopped and stabilized within a range of 200 mHz. 7 Minutes
after the load step the active power of the coal-fired power plant is increased due to a dispatch
command. The power is ramped up from 148 MW to 170 MW in five minutes. This equates
to a gradient of only 2.6%/min. With the BESS all power plants are unburdened from fast load
changes.

The dispatch command is set via a parameter event in PowerFactory . To obtain a slope with
a certain gradient and no step the model of the governor was expanded with a gradient limiter
block (Figure 4.1).

With PowerFactory is it possible to display results from different simulations within one plot. For
that are different result files used. In Figure 4.2 and Figure 4.3 the results are from two different
study cases. But it is also possible to use several result files within one study case. The
advantage of one study case/variation per result file is that the results and the used parameters
for the simulation match together.

0.0005

dispatc.. Gradient Limi..

0
1
o113

-0.0005 ptHP

pturbhp pt
0.0005 0
Pt/Pturb
0

cosnhp
1
psetphp
1

dispatch
K1

K3

K5

K7

Gradient Limiter
K

K
K

0
o114

1
Uo Pmax

w dw K (1+sTb)/(1+sTa) - 1/K Limit.. {1/s} 1/(1+sT) 1/(1+sT) 1/(1+sT) 1/(1+sT)

o113

-
K T1,T2
-
T3 T4 T5
-0.0005 T6 T7

wref
3 x0 Uc Pmin x1 x2 x3 x4 x5
K2

K4

K6

K8
K

K
K

o112
o14

o18

psetphp pturblp ptlp

1 0 1
Pt/Pturb x L1

cosnlp 1

cosn
4

cosnlp
5

Figure 4.1: Edited Governor Controller for Dispatch Command (via event)

Battery Energy Storing Systems (BESS) 19

4 Application Case 1: Frequency Regulation

DIgSILENT
130,00

125,00

120,00

115,00

110,00

105,00

100,00
60,00 70,00 80,00 90,00 [s] 100,00
G1 (hydro): Active Pow er in MW - w ithout BESS
G1 (hydro): Active Pow er in MW - w ith BESS

81,02

77,24

73,45

69,67

65,88

62,09
60,00 70,00 80,00 90,00 [s] 100,00
G2 (gas): Active Pow er in MW - w ithout BESS
G2 (gas): Active Pow er in MW - w ith BESS

170,00

165,00

160,00

155,00

150,00

145,00
60,00 70,00 80,00 90,00 [s] 100,00
G3 (coal): Active Pow er in MW - w ithout BESS
G3 (coal): Active Pow er in MW - w ith BESS

1,001

0,998

0,995

0,992

0,989

0,987
60,00 70,00 80,00 90,00 [s] 100,00
G1 (hydro): Speed in p.u. - w ithout BESS
G1 (hydro): Speed in p.u. - w ith BESS

Figure 4.2: Simulation Results of a 30 MW Load Step (detail)

Battery Energy Storing Systems (BESS) 20

4 Application Case 1: Frequency Regulation

DIgSILENT
130,00

125,00

120,00

115,00

110,00

105,00

100,00
0,000 3,000 6,000 9,000 12,00 [min] 15,00
G1 (hydro): Active Pow er in MW - w ithout BESS
G1 (hydro): Active Pow er in MW - w ith BESS

81,02

77,24

73,45

69,67

65,88

62,09
0,000 3,000 6,000 9,000 12,00 [min] 15,00
G2 (gas): Active Pow er in MW - w ithout BESS
G2 (gas): Active Pow er in MW - w ith BESS

180,00

170,00

160,00

150,00

140,00
0,000 3,000 6,000 9,000 12,00 [min] 15,00
G3 (coal): Active Pow er in MW - w ithout BESS
G3 (coal): Active Pow er in MW - w ith BESS

1,001

0,998

0,995

0,992

0,989

0,987
0,000 3,000 6,000 9,000 12,00 [min] 15,00
G1 (hydro): Speed in p.u. - w ithout BESS
G1 (hydro): Speed in p.u. - w ith BESS

Figure 4.3: Simulation Results of a 30 MW Load Step (full)

Battery Energy Storing Systems (BESS) 21

4 Application Case 1: Frequency Regulation

4.2 Replacement of Primary Control Energy (Generator Outage)

Study Case: Gas Outage without BESS; Gas Outage with BESS

The worst case for a generator outage in the small showcase grid is the outage of the gas
turbine. The gas turbine is the fastest power plant in the grid. Without that fast controllable plant
would it be much harder to control the frequency after a disturbance. To study the outage with
more realistic results a low load scenario is assumed with only 80% of the original loads. For
that is an operation scenario used in PowerFactory . With an operation scenario is it possible
to change and store operational data for different cases, like a low load and a high load case
without the need for an additional variation.

The outage of the power plant with the gas turbine is simply simulated by opening the switch
on the HV-side of the main transformer. The following frequency under-run is without the BESS
more distinct than in the load change chase. This comes from the lack of a fast controllable
power plant (Picture x red curves).

Again the BESS is used to improve the behaviour of the grid performance and to replace the
primary control power of the coal-fired power plant. The BESS could control the active power
output much faster than a power plant because there are no mechanical inertia and delays.
Therefore the frequency deviation is smaller.

Battery Energy Storing Systems (BESS) 22

4 Application Case 1: Frequency Regulation

DIgSILENT
124,6

116,6

108,6

100,7

92,68

84,71
60,00 70,00 80,00 90,00 [s] 100,00
G1 (hydro): Active Pow er in MW - w ithout BESS
G1 (hydro): Active Pow er in MW - w ith BESS

170,00

160,00

150,00

140,00

130,00

120,00
60,00 70,00 80,00 90,00 [s] 100,00
G3 (coal): Active Pow er in MW - w ithout BESS
G3 (coal): Active Pow er in MW - w ith BESS

1,005

0,994

0,983

0,973

0,962

0,951
60,00 70,00 80,00 90,00 [s] 100,00
G1 (hydro): Speed in p.u. - w ithout BESS
G1 (hydro): Speed in p.u. - w ith BESS

1,378

1,247

1,115

0,984

0,853

0,721
60,00 70,00 80,00 90,00 [s] 100,00
SimpleBattery: Ucell in p.u. (Basis: 10,00 )
SimpleBattery: SOC

Figure 4.4: Simulation Results of Gas Turbine Outage (detail)

Battery Energy Storing Systems (BESS) 23

4 Application Case 1: Frequency Regulation

DIgSILENT
124,6

116,6

108,6

100,7

92,68

84,71
1,00 6,00 11,00 16,00 21,00 [min] 26,00
G1 (hydro): Active Pow er in MW - w ithout BESS
G1 (hydro): Active Pow er in MW - w ith BESS

180,00

170,00

160,00

150,00

140,00

130,00

120,00
1,00 6,00 11,00 16,00 21,00 [min] 26,00
G3 (coal): Active Pow er in MW - w ithout BESS
G3 (coal): Active Pow er in MW - w ith BESS

1,005

0,994

0,983

0,973

0,962

0,951
1,00 6,00 11,00 16,00 21,00 [min] 26,00
G1 (hydro): Speed in p.u. - w ithout BESS
G1 (hydro): Speed in p.u. - w ith BESS

1,415

1,121

0,827

0,533

0,239

-0,055
1,00 6,00 11,00 16,00 21,00 [min] 26,00
SimpleBattery: Ucell in p.u. (Basis: 10,00 )
SimpleBattery: SOC

Figure 4.5: Simulation Results of Gas Turbine Outage (full)

Battery Energy Storing Systems (BESS) 24

5 Application Case 2: FRT Simulation

5 Application Case 2: FRT Simulation

Study Case: FRT without BESS; FRT with BESS

Together with the different renewable energy sources, such as wind turbines and photovoltaic,
the ride through of voltage dips becomes more and more an important topic. So it is also
important to have a BESS model that could be used in a fault ride through (FTR) study. Since
the FRT is a very fast but also short action the model could be simplified in some parts. It is
very unlikely that the SOC of the battery changes significantly within one or two seconds. Also
the battery DC-voltage could be assumed as constant during that time. The important branch of
the BESS model is then the PV/PQ- and the current controller. To obtain that the whole battery-
system frame is out of service in the used study cases. So the SOC in the charge controller is
fixed to a constant value (0.8 pu). The DC-voltage is also constant fixed to 1 pu. To improve the
behaviour of the BESS the charge controller is expanded with a fault modus. If the AC-voltage
deviates more than 0.1 pu from its nominal value the charge controller is switched to reactive
power preference.

The example grid is supplemented by a 8 MW asynchronous motor. The BESS is to the motor
bus connected for that simulation as shown in Figure 5.2. During a voltage dip an asynchronous
motor delivers reactive power to the grid. But after the voltage dip an asynchronous motor needs
a lot of reactive power to re-magnetize the rotor, then during the re-magnetization the voltage
on the motor bus is low. Using the BESS connected to the Motor Bus the voltage recovers faster
due to the effect of the BESS reactive power control.

- deltai
0

maxAbsCur

id_ref_in
0

ChargeCtrl idin id_ref_out

0 0 1
ChargeCur,minSOC,maxSOC
SOC
1

Current Limiter
deltaU
iq_ref_in
1

uac iq_ref_out
2 1 2

Figure 5.1: Charge Controller with switchable Current Limiter (ElmDsl)

Battery Energy Storing Systems (BESS) 25

5 Application Case 2: FRT Simulation

Bus 8
Battery Energy Storage System

Line 5

Bus 9
PWM Converter

Bus 4
DC-Terminal
Bus 6

2-Winding Transformer
Line 6

Battery

T4
Load B

Motor Bus
T1

M
~
Asynchronous Machine

Asynchonous Machine
Bus 1

Figure 5.2: Changed Grid Structure for FRT-Study

In the results (Figure 5.3) is it obvious, that the recovery of the system with the BESS is faster
than the recovery of the system without the BESS.

Battery Energy Storing Systems (BESS) 26

5 Application Case 2: FRT Simulation

Figure 5.3: Simulation Results of a Short Circuit on Bus 6

Battery Energy Storing Systems (BESS) 27

References

References

[1] Rechargable Batteries Applications Handbook. Technical report, Technical Marketing Staff
of Gates Energy Products Inc. Butterworth-Heinemann, 1992.
[2] S. Barsali and M. Ceraolo. Dynamical Models of Lead-Acid Batteries: Implementation Is-
sues. IEEE Transactions on on Energy Conversion, 17(1), March 2002.

[3] M. Ceraolo. New Dynamical Models of Lead-Acid Batteries. IEEE Transactions on Power
Systems,, 15(4), November 2000.
[4] D. Kottick, M. Blau, and D. Edelstein. Battery Energy Storage for Frequency Regulation in
an Island Power System. IEEE Transactions on Energy Conversion,, 8(3), September 1993.
[5] N. K. Medora and A. Kusko. Dynamic Battery Modeling of Lead-Acid Batteries using Man-
ufacturer’s Data. Twenty-Seventh International Telecommunications Conference, INTELEC
’05., 2005.

Battery Energy Storing Systems (BESS) 28