Você está na página 1de 6

Implementation and Validation of the Nordic Test

System in DIgSILENT PowerFactory

Luis David Pabon Ospina Gustav Lammert
Andres Felipe Correa Department of Energy Management and
Department of Power System Control and Dynamics Power System Operation
Fraunhofer IWES University of Kassel
Kassel, Germany Kassel, Germany
luis.david.pabon.ospina@iwes.fraunhofer.de gustav.lammert@uni-kassel.de

AbstractThis paper presents the implementation of the Nordic32 system which includes induction motor loads is used
Nordic test system in DIgSILENT PowerFactory. The imple- to investigate the delayed voltage recovery after system faults.
mented system is a variant of the Nordic32 network and is In [4], a coordinated decentralized emergency voltage and
based on the technical report Test Systems for Voltage Stability
Analysis and Security Assessment prepared by the IEEE Power reactive power control to prevent long-term voltage instability
System Dynamic Performance Committee, the Power System is proposed and its effectiveness is proven on the Nordic32
Stability Subcommittee, and the Test Systems for Voltage Stability test system. In [5], an algorithm which prevents undesirable
and Security Assessment Task Force. This work deems all of distance protection operations that might result during voltage
the implemented controllers and validates the PowerFactory instability periods is tested on the Nordic32 power system. In
implementation against the original model presented in the afore-
mentioned technical report. It is shown through RMS simulations [6], the Nordic32 system is simplied in a 5 busbar test system
that the action of load tap changers and over excitation limiters named N5area in which the key voltage collapse characteristics
are leading forces through a long-term voltage collapse and of the Nordic32 are reected in order to test in a less complex
therefore, their accurate modeling is crucial. Furthermore, the network so the actions of load recovery and excitation current
implemented system can be whether downloaded from the IEEE limiters are easier to simulate. In [7], small-signal stability is
PES Power System Dynamic Performance Committee website or
requested directly to the authors. assessed using the Nordic32. In [8], a procedure to identify the
onset of long-term voltage instability from the time evolution
Index TermsDynamic security assessment, long-term voltage of the distribution voltages controlled by LTCs is proposed
stability, Nordic 32 test system, voltage collapse. and demonstrated using the Nordic32 test system.
Due to its particular characteristics, the Nordic test system
I. I NTRODUCTION has shown a growing demand for different kinds of power
The Nordic test system is a variation of the so-called system stability studies. Therefore, the IEEE Power System
Nordic32 system, which was created in order to illustrate the Dynamic Performance Committee provides in [9] input data
voltage collapse in Sweden that happened in the year 1983 for different software tools including RAMSES [10], PSS/E,
[1]. It has been proposed by the CIGRE Task Force 38-02- and ANATEM. DIgSILENT PowerFactory is a widely used
08 in 1995 [2] and it is a ctional approximation of the commercial power system analysis software, especially in
Swedish system that along with other network models was Europe and South America. Numerous transmission and dis-
proposed in order to assess the performance of simulation tribution system operators as well as research institutes and
tools and provide researchers with benchmarks to develop the universities use this software simulation tool [11]. In order to
Long-Term Dynamics eld. Nevertheless, the system is not serve the need of academia and industry, the new contribution
limited to long-term phenomena, its applicability to short- of this paper is the implementation and validation of the
term dynamics such as transient stability and small-signal Nordic test system in DIgSILENT PowerFactory. Furthermore,
oscillatory angle stability is recognized. From its very rst the implemented system can be whether downloaded from the
appearance on, the Nordic test system has been used not only IEEE PES Power System Dynamic Performance Committee
for software testing purposes but also for other sort of dynamic website or requested directly to the authors.
studies such as short-term and long-term voltage stability,
power system restoration, system protection and small-signal II. S YSTEM DESCRIPTION
stability studies among many others. In [2], it is shown through The Nordic test system depicted in Fig. 1 consists of 4 areas:
RMS simulations carried out using PSS/E and EUROSTAG an equivalent simplied network that has the biggest genera-
the impact of Over Excitation Limiters (OELs) and Load Tap tors and therefore serves usually as the system reference, the
Changers (LTCs) on the long-term voltage collapse. In [1], northern region with few load and more generation, a central
the Nordic32 test system is used in order to investigate power area with more load than generation and a southern region
system restoration issues. In [3], a modied version of the loosely connected to the rest of the system. The test network
g19 g9 g2  
4071 4011 1011 1013 P = P0 (1)
71 11 12 13  
4012 1012 1014 Q = Q0 (2)
g10 g1
g3 being = 1 (constant current) and = 2 (constant
g5 NORTH g11
72 g20 400 kV impedance). V0 assumes the voltage at initial conditions of
1021 1022 4022 4021 220 kV
130 kV
the bus to which the load is connected.
CS condenser The Nordic test system makes use of hydro and thermal
g4 2032 2031 22 4031 4032
generation. Out of the total of 20 generators, the Equivalent
and the North networks have hydro generation that is modeled
32 31
g8 42 by salient-pole machines. The two biggest generators belong to
4042 the Equivalent grid while the North area has 10 generators. On
4041 CENTRAL the other hand, there are seven thermal generators modeled by
41 CS
g14 round-rotor machines, two of them belong to the small South
g13 g7 4 4044 4043 4046
3 region and the remaining ve to the Central one that also
1043 g6 has a synchronous capacitor that is modeled as a salient-pole
1044 43 46
61 1042 The models for the above listed power system elements
62 1041 1045 g15
are detailed enough for long-term as well as for short-term
g17 2 stability studies. A complete system description can be found
4062 1 g16
5 51 in [12].
4063 4051
g18 63
SOUTH All the dynamic models such as Automatic Voltage Regu-
lators (AVRs), Power System Stabilizers (PSSs), Over Excita-
Fig. 1. Nordic test system - single line diagram [12]. tion Limiters (OELs), Load Tap Changers (LTCs) and speed
governors are described in detail in [12]. This section focuses
specially on those which implementation can vary from one
is composed by 32 transmission buses with nominal voltages software tool to another.
of 400 kV, 220 kV and 130 kV. Additionally, 22 buses are at
a distribution level with a voltage of 20 kV and 20 generator A. Automatic voltage regulator, over excitation limiter and
nodes with a voltage of 15 kV. power system stabilizer
There are basically three kinds of transformers involved in The model depicted in Fig. 2 is used to represent the AVR,
this power system, the 20 step-up transformers convert the the PSS and the OEL. The PSS is present in all of the
generation voltage of 15 kV into the transmission voltage that generators with the exception of the two big generators of
can be 400 kV, 220 kV or 130 kV according to the region the Equivalent grid and the synchronous capacitor. A rst-
level. Eight transformers convert the voltage among the three order system represents the exciter and the AVR also includes
different already mentioned transmission voltage levels. The a transient gain reduction chosen to limit the overshoot in
22 step-down transformers are placed for distribution duties terminal voltage following a step change in voltage reference
and hence their function is to convert the transmission voltage when the generator operates in open circuit. The PSS is formed
level into the distribution level for the loads at 20 kV. All of by a washout lter with two cascaded leads which provide
the transformers neglect the copper losses and the magnetizing damping for oscillation frequencies from 0.2 Hz to more than
susceptance. 1 Hz.
The reactive power management is helped by a total of 11 There is more than 3 GW of power transfer from North to
shunt elements out of which nine are capacitors and two are Central area and the maximum power that can be delivered
reactors. The Equivalent grid has one shunt reactor, while the depends mostly on the reactive power capabilities of the
North region has one reactor and one capacitor, the South Central area. The reactive power limits are dened by the
region doesnt have any sort of shunt compensation at all, OELs. This system presents long-term voltage instability given
leaving the Central region which has to cope with signicant that the maximum power that can be delivered is smaller
power transfer from the North with the eight remaining shunt than the one that LTCs aim to restore in some contingency
capacitors. scenarios.
All loads are connected to 20 kV buses and are represented The OELs are one of the leading forces through a long-
by an exponential model which behavior is described as term voltage collapse. In this particular case their effect can be
follows: easily observed when one of the transmission lines connecting

f r Tm
timer 0.1 0 q H 1 1
+ 1 y
y 0 ( )2 q
z + Tw s Pm
if d s 1
y 0
1 L2
ilim 2
fd L1 3 Hs 1
G(1+ sTa) + vf d
Min 10 1
1+ sTb s
+ gain reduction 0 Fig. 4. Model of hydro turbine [12].
Kps 1+ sT1 1+ sT1 C. Load tap changer
1+ sTw 1+ sT2 1+ sT2
power system stabilizer C All distribution transformers are equipped with LTCs keep-
ing the distribution voltage in the deadband [0.99 - 1.01] p.u.
Fig. 2. Model of Exciter, AVR, PSS and OEL [12]. The LTCs adjust the transformer ratios in the range [0.88 -
1.20] over 33 positions (thus from one position to the next,
the ratio varies by 0.01) [12].
the North zone with the Central zone is tripped. After such The LTCs have intentional delays. When the distribution
event, LTCs act to restore distribution voltages and hence voltage leaves the above deadband at time t0 , the rst tap
load consumption. The generators are forced to increase their change takes place at time t0 + 1 and the subsequent changes
reactive power injection in order to keep the voltage at the at times t0 + 1 + k2 (k = 1, 2, ...). The delay is reset to 1
set-point value. When the eld current exceeds for a certain after the controlled voltage has reentered to (or jumped from
time the current limit denoted as ilim
f d and equal to 105% of the one side to the other of) the deadband [12]. The values of 1
rated eld current irated
fd , the OEL sets the eld current to ilim
fd . and 2 are given in [12] and differ from one transformer to
The above mentioned time depends on the parameters L1, f another in order to avoid unrealistic tap synchronization.
and r in Fig. 2. Without an accurate modeling of the OEL the The implementation of the LTC control was addressed in
long-term voltage collapse to be described in Section V will this work with a simple state machine shown in Fig. 5, which
signicantly differ from the one in [12]. control logic is explained in Table I.
D. Saturation
B. Speed governor and hydro turbine The generators main-ux saturation is another issue to be
considered. Small mismatches on the saturation characteristic
The nominal frequency of the system is 50 Hz and is lead to events shifted in time in a long-term voltage collapse
controlled by the speed governors (see Fig. 3) of the hydro when compared with the original system. The main reason is
generators, the thermal generators of the Central and South that the OELs will limit the eld current at different times if
area are not involved in frequency control and for the explained saturation is not considered or if it differs from the one in
reasons in [12], constant mechanical torque is assumed for the [12].
machines of thermal plants. The model of the speed governor
includes a simple power measurement, a PI control and a
servomotor represented by a rst-order system with a time
constant of 0.2 s and non-windup limits on z which represents z00
the gate opening. z00
The speed governor gives the gate opening to the hydro
turbine model depicted in Fig. 4, which is represented by a
simple lossless model with a water time constant Tw of 1 s. e2 e1
In this model q represents the water ow, H the head, Pm the z12

mechanical power and Tm the mechanical torque.


Fig. 5. State machine for the implementation of the LTC control.

PI control 0.1
+ + + 1 z TABLE I
2.0 5.0
+ + 0.1
1 0
1 s Transition Condition Action
z00 V returns to or jumps the dead band e=0
P +
1 Po z01 e=0 and V is out of dead band for 1 s e=1 and tap step
1+ s2
z12 e=1 and V is out of dead band for 2 s e=2 and tap step
Fig. 3. Model of speed governor [12]. z21 e=2 and V is out of dead band for 2 s e=1 and tap step
In this work, the saturation characteristic is as it is depicted The corrective control takes place after the lowest transmis-
in Fig. 6. sion voltage reaches 0.9 p.u. which is at 100 s of simulation.
Two different variants have been implemented:
Vnl [p.u.]
Air-gap line Scenario 2a: reducing the set point of the ve LTCs

1.2 A B C controlling the distribution buses 1, 2, 3, 4 and 5.

Scenario 2b: reducing the set point of the 11 LTCs
controlling the distribution buses 1 to 5 in addition to
the buses 41, 42, 43, 46, 47 and 51.


A. Scenario 1
1.0 i1.0 i1.2 if d [p.u.]
The results of the Scenario 1 are depicted from Fig. 7
Fig. 6. Implemented saturation characteristic.
to Fig. 10. For all the gures in this paper, the original
results taken from [12] correspond to solid lines, while the
According to [12], the saturation characteristic of all gen-
PowerFactory implementation ones correspond to the dashed
erators in the Nordic test system is given by:
| AC | The chain of events resulting in a voltage collapse runs
k= = 1 + m(Vnl )n (3)
| AB | as follows: a solid three phase fault is applied to the bus
4032. The line 4032-4044 is tripped 0.1 s later. The system is
in which, for Vnl = 1.0 p.u. k = 1.1 and hence m = 0.1, and
short-term stable, and it settles to a new equilibrium in 30 s.
for Vnl = 1.2 p.u. k = 1.3 and hence n = 6.0257.
At around 35 s, the LTCs start acting attempting to restore
From (3), it can be derived that i1.0 = 1.1 p.u. and
the distribution voltage and hence the load consumption. The
i1.2 = 1.56 p.u.
action of the LTCs forces the generators to increase their
The implemented saturation characteristic is an exponential
reactive power injection and therefore their eld current as
function with the following input parameters:
it can be seen in Fig. 8. Due to the actions of the LTCs, the
i1.0 OELs of the seven generators in Fig. 8 limit the eld current.
SG10 = 1 (4)
i0 Some other generators as the ones in Fig. 9 are not limited.
It can be seen in Fig. 10 how the limited generators, as in the
SG12 = 1 (5) case of g6 and g7, loose their capability to control voltage and
1.2 i0 their terminal voltage drops until the system nally collapses.
where i0 = 1.0 and hence SG10 = 0.1 and SG12 = 0.3. More
details about the saturation model can be found in [13]. B. Scenario 2
IV. S TUDY CASES Fig. 11 shows the evolution of the voltage magnitude of
the transmission bus 1041 which is the one with the lowest
In order to validate the implemented system and compare it
voltage. The action of reducing the voltage set-point of the
against the results in [12], two different scenarios have been
LTCs controlling distribution voltages can be observed. The
rst variation (Scenario 2a) of the post disturbance control
A. Scenario 1 consisting in reducing the above mentioned voltage set-point
A solid three phase fault is applied to the bus 4032 with by 0.05 p.u. shows that although the control action succeeds
a subsequent opening of the line 4032-4044 without further in reducing the power consumption, it is still more than what
re-connection. As stated in [12], the initial fault is simulated can be provided and the system nally collapses shortly before
just to be more realistic, but it is actually the line outage that 200 s.
causes the long-term voltage collapse. The long-term behavior On the other hand, the second variation (Scenario 2b), shows
would be very similar without the initial three phase fault. to be an effective emergency control as it is able to prevent
the load restoration in a way that the voltage of transmission
B. Scenario 2 buses is recovered even to a higher value than the one prior
The same disturbance as in Scenario 1 is applied. In this the disturbance. Due to the LTCs delays, the action of this
case, a corrective post disturbance control is implemented. post disturbance control is slow but it succeeds restoring the
This control is a typical System Integrity Protection Scheme transmission voltage before 600 s without any under-voltage
(SIPS), which consist in decreasing the voltage set-point of load shedding.
LTCs controlling the distribution voltage by 0.05 p.u. Due to Note that the transmission voltages are recovered to values
the load characteristics explained in (1) and (2), by decreasing even higher than those before the disturbance. In fact, this
0.05 p.u. the voltage set-point, the active power consumption emergency control is possible due to the regaining voltage
is expected to reduce 5%, while the reactive power approxi- control of the synchronous generators. As shown in Fig. 12, the
mately 10%. generators which were limited in Fig. 8, recover their voltage
Voltage Magnitude [p.u.]

Voltage Magnitude [p.u.]


Bus 1041 0.9

Bus 1042 No action on LTCs
Bus 4012 0.8 Voltage setpoint reduction on 5 LTCs
Bus 4062 Voltage setpoint reduction on 11 LTCs
0.7 0.7
0 20 40 60 80 100 120 140 160 180 0 100 200 300 400 500 600
time [s] time [s]

Fig. 7. Voltage magnitude of affected buses Original test system (solid Fig. 11. Voltage magnitude at bus 1041 Original test system (solid lines)
lines) PowerFactory results (dashed lines). PowerFactory results (dashed lines).

4 4

Field Current [p.u.]

Field Current [p.u.]

3 g11 3
2 2
g06 g11 g14 g16
g07 g12 g15
1 1
0 20 40 60 80 100 120 140 160 180 0 100 200 300 400 500 600
time [s] time [s]

Fig. 8. Field currents of seven limited generators Original test system (solid Fig. 12. Generators eld currents with the implemented post disturbance
lines) PowerFactory results (dashed lines). emergency control.

Voltage Stability and Security Assessment Task Force from
Field Current [p.u.]

2.5 g08 the IEEE Power System Dynamic Performance Committee,

g18 was implemented in DIgSILENT PowerFactory. The results
2.0 of the implemented system are validated against those in
the aforementioned report through RMS simulations. Results
1.5 show that the long-term voltage collapse phenomena can be
0 20 40 60 80 100 120 140 160 180 observed mainly because of the action of LTCs and OELs after
time [s] a system disturbance. A corrective post disturbance emergency
control was implemented and validated against the original
Fig. 9. Field currents of three non limited generators Original test system
(solid lines) PowerFactory results (dashed lines). system, showing that it can avoid the long-term voltage
collapse. Furthermore, the implemented system can be whether
downloaded from the IEEE PES Power System Dynamic
Voltage Magnitude [p.u.]

Performance Committee website or requested directly to the
The implemented Nordic test system model in DIgSI-
LENT PowerFactory can be whether downloaded from the
0 20 40 60 80 100 120 140 160 180 IEEE PES Power System Dynamic Performance Committee
time [s] website: http://ewh.ieee.org/soc/pes/psdpc/PSDP benchmark
systems.htm or requested directly to the authors.
Fig. 10. Terminal voltage of two limited generators Original test system
(solid lines) PowerFactory results (dashed lines). The authors request, that the publications derived from the
use of the downloaded system explicitly acknowledge that fact
by citing this paper.
control after some time avoiding thereby several over-voltages
in the transmission system. ACKNOWLEDGMENT
Due to their helpful feedback during the development of
VI. C ONCLUSIONS this work [14], [15], the authors would like to thank Thierry
In this work, the Nordic test system described in the tech- Van Cutsem from the Department of Electrical Engineering
nical document Test Systems for Voltage Stability Analysis and Computer Science at University of Liege - Belgium
and Security Assessment prepared by the Test Systems for and Past-Chair of the Power System Dynamic Performance
Committee, and Costas Vournas from the School of Electrical
and Computer Engineering at the National Technical Univer-
sity of Athens and Chairman of the IEEE Dynamic Security
Assessment working group.
This research work was supported by the German Federal
Ministry for Economic Affairs and Energy within the frame-
work of the project Netz:Kraft (FKZ: 0325776A).
[1] O. Samuelsson, L. Lindgren, and B. Eliasson, Simulated power sys-
tem restoration, in Universities Power Engineering Conference, 2008.
UPEC 2008. 43rd International, Sept 2008, pp. 15.
[2] CIGRE - Task Force C38-02-08, Long-Term Dynamics Summary Part
II - A practical Assessment of Simulation Tools, Tech. Rep., 1995.
[3] H. Khoshkhoo and S. M. Shahrtash, Fast Online Dynamic Voltage In-
stability Prediction and Voltage Stability Classication, IET Generation,
Transmission Distribution, vol. 8, no. 5, pp. 957965, May 2014.
[4] S. R. Islam, D. Sutanto, and K. M. Muttaqi, Coordinated Decentralized
Emergency Voltage and Reactive Power Control to Prevent Long-Term
Voltage Instability in a Power System, IEEE Transactions on Power
Systems, vol. 30, no. 5, pp. 25912603, Sept 2015.
[5] M. Jonsson and J. E. Daalder, An Adaptive Scheme to Prevent
Undesirable Distance Protection Operation During Voltage Instability,
IEEE Transactions on Power Delivery, vol. 18, no. 4, pp. 11741180,
Oct 2003.
[6] M. R. S. Tirtashi, O. Samuelsson, and J. Svensson, Long-term Voltage
Collapse Analysis on a Reduced Order Nordic System Model, in Power
Engineering Conference (UPEC), 2014 49th International Universities,
Sept 2014, pp. 16.
[7] F. R. S. Sevilla and L. Vanfretti, A Small-signal Stability Index for
Power System Dynamic Impact Assessment Using Time-domain Simu-
lations, in 2014 IEEE PES General Meeting Conference Exposition,
July 2014, pp. 15.
[8] C. D. Vournas and T. Van Cutsem, Local Identication of Voltage
Emergency Situations, IEEE Transactions on Power Systems, vol. 23,
no. 3, pp. 12391248, Aug 2008.
[9] IEEE Power & Energy Society Power System Dynamic Performance
Committee. Benchmark Systems. [Online]. Available: http://ewh.ieee.
org/soc/pes/psdpc/PSDP benchmark systems.htm
[10] P. Aristidou, D. Fabozzi, and T. Van Cutsem, Dynamic Simulation of
Large-Scale Power Systems Using a Parallel Schur-Complement-Based
Decomposition Method, IEEE Transactions on Parallel and Distributed
Systems, vol. 25, no. 10, pp. 25612570, Oct 2014.
[11] G. Lammert, L. D. Pabon Ospina, P. Pourbeik, D. Fetzer, and M. Braun,
Implementation and Validation of WECC Generic Photovoltaic System
Models in DIgSILENT PowerFactory, in 2016 IEEE Power & Energy
Society General Meeting, Boston, July 2016, pp. 15.
[12] IEEE Task Force on Test Systems for Voltage Stability and Security
Assessment, Test systems for voltage stability analysis and security
assessment, PES-TR19, Tech. Rep., August 2015.
[13] DIgSILENT GmbH, DIgSILENT PowerFactory Synchronous Machine
Technical Reference Documentation, Tech. Rep., 2016.
[14] T. Van Cutsem, Private letter with the authors, Aug. 2016.
[15] C. D. Vournas, Private letter with the authors, Sep. 2016.