Impacts of Distributed Generation on

Power System Transient Stability

J.G. Slootweg W.L. Kling
Member IEEE Member IEEE

Electrical Power Systems, Faculty of Information Technology and Systems, Delft University of Technology
P.O. Box 5031, 2600 GA Delft, The Netherlands
Phone: +31 15 278 6219, Fax: +31 15 278 1182, e-mail: [email protected]

Abstract - It is expected that increasing amounts of new generation A further difference between these new technologies and
technologies will be connected to electrical power systems in the conventional means of electrical power generation is that
near future. Most of these technologies are of considerably smaller many of them are of considerably smaller size than
scale than conventional synchronous generators and are therefore
conventional thermal, nuclear and hydro units that up to this
connected to distribution grids. Further, many are based on
technologies different from the synchronous generator, such as the
moment deliver the majority of the electrical power consumed
squirrel cage induction generator and high or low speed generators worldwide. Therefore, they are often connected to low and
that are grid coupled through a power electronic converter. medium voltage grids and not to the high voltage
When connected in small amounts, the impact of distributed transmission grid, resulting in longer electrical distances.
generation on power system transient stability will be negligible. As long as the penetration level of these new technologies in
However, if its penetration level becomes higher, distributed power systems is still low and they only cover a minor
generation may start to influence the dynamic behavior of the power fraction of the system load, they hardly impact the dynamic
system as a whole. In this paper, the impact of distributed generation behavior of a power system. Therefore, in power system
technology and penetration level on the dynamics of a test system is
dynamics and transient stability studies, they are normally
investigated. It is found that the effects of distributed generation on
the dynamics of a power system strongly depend on the technology
considered as negative load and their intrinsic dynamics and
of the distributed generators. their controllers, if present, are not taken into account.
However, if the amount of new generation technology
Keywords: distributed/dispersed generation, transient stability, introduced in a power system becomes substantial, it may
power system dynamics, modeling, simulation start to influence the overall behavior of this system [1,2]. Up
to this moment, it is unknown at which penetration level and
I. INTRODUCTION in which way distributed generation influences power system
dynamic behavior, which is partly caused by the lack of
It is expected that increasing amounts of new technologies for adequate models of the new technologies.
electrical power generation will be introduced in electrical The goal of the research presented here is to acquire insight in
power systems in the near future. A number of reasons for the impact of distributed generation on power system
this development exist. The emphasis on power generation transient stability. To this end, various technologies and
from renewable sources in order to reduce the environmental penetration levels are studied. The paper is structured as
impact of power generation leads to the development and the follows. First, the research approach is described and the
use of technologies for renewable power generation, such as preparation of the test system and the modeling of the various
solar panels, wind turbines and wave power plants. The goals distributed generation technologies is commented upon. Then
of decreasing the cost associated with electrical power the results of the investigations are presented and discussed.
transmission in liberalized markets and increasing the
efficiency of primary fuel use by using combined heat and II. RESEARCH APPROACH
power generation (CHP) lead to the installation of generation
equipment at consumer sites. For the research, a well known dynamics test system was
Many of these new technologies do not use a conventional chosen and adapted to the needs of this study. In the base
grid coupled synchronous generator to convert primary case, this test system does not contain any distributed
energy into electricity. Instead, they use a squirrel cage or generation. The whole load is covered by the ten large scale
doubly fed induction generator (some wind turbine concepts) synchronous generators present in the system. This system
or a synchronous or squirrel cage induction generator that is has been widely used for various kinds of power system
grid connected through a power electronic converter (other dynamics studies. The considerations leading to the choice of
wind turbine concepts, small scale high speed gas turbines). this test system and the development of the models used in
In the case of solar panels and fuel cells, it is not even the research will be commented upon in more detail in the
mechanical power that is converted into electricity. next paragraph, in which the preparation of the test system is
described.

0-7803-7519-X/02/$17.00 © 2002 IEEE

862
Five different distributed generation technologies are studied
below, namely:
& squirrel cage induction generator
& uncontrolled synchronous generator
& synchronous generator with grid voltage and frequency
control
& uncontrolled power electronic converter
& power electronic converter with grid voltage and Fig. 1. Transient stability indicators: maximum rotor speed deviation and
frequency control oscillation duration
The penetration level of each of these technologies was
varied in the following way: fault, namely:
& A distributed generator was connected to each load & maximum rotor speed deviation
bus by an impedance of 0.05j p.u. on the 100 MVA & oscillation duration
system base. The meaning of the maximum rotor speed deviation is
& Both real and reactive power of the loads were obvious. It is the maximum rotor speed deviation reached by
increased in steps of five per cent until the load was the generator during or after the fault. The oscillation
increased to one and a half times that of the base case. duration is defined as follows:
& The distributed generator connected to each of the
load buses was adjusted in order to generate an The oscillation duration is equal to the time interval between
amount of active power equal to the increase of the the application of the fault and the moment after which the
real power consumed by the load at that bus. rotor speed stays within a bandwidth of 110-4 p.u. during a
Thus, the active power generated by the large scale generators time interval longer than 2.5 seconds.
was not changed, except for the active power of generator
number 2, which acts as the swing bus or slack node. Its Thus, the oscillation duration is a measure for the time span
power is changed during the solution of the load flow to make that is needed to reach a new equilibrium after a disturbance.
up for changes in the network losses, if any. Note that a load Both measures are depicted in figure 1. The lower the value
increase of 50 per cent that is covered by distributed of each of these indicators for a given case, the better the
generation corresponds with a distributed generation transient stability is.
penetration level of 33 per cent.
It is not considered realistic to further increase the distributed III. TEST SYSTEM PREPARATION
generation penetration level. Higher penetration levels might
give rise to the necessity to reconsider the contemporary To investigate the impact of distributed generation on power
electrical power systems concept and compare it with other system transient stability, a model of a power system of a
solutions, such as DC links or (semi-)autonomous systems. certain scale is necessary in order to arrive at results that can
Therefore, it seems not useful to investigate the impact of be considered representative. On the other hand, the system
such high penetration levels of distributed generation on the should not be too large, in order to reduce the computation
dynamics of a classic interconnected AC power system. time and to limit the number of possible scenarios.
It was assumed that the uncontrolled distributed generation Keeping this in mind, the widely used New England test
technologies operated at unity power factor, whereas the system has been chosen as the test system to be used here [3].
reactive power of the controlled technologies was able to vary The system is depicted in figure 2 and some of its
between zero and half the active power (cos 1±0.9) and characteristics are given in table 1. To arrive at a full dynamic
controlled in such a way that the terminal voltage was kept as model of the system, representative values for the parameters
near to its nominal value as possible without exceeding the of the generators and the exciters and governors have been
reactive power limits of the generators. taken from other sources [4,5]. The loads are equally divided
To investigate the transient stability of the test system, a fault in constant impedance, constant power and constant current.
was applied to the transmission line between the buses 15 and As already discussed above, in this paper five types of
16 of the test system, which was carrying 315 MW and 150 distributed generation are distinguished. For the synchronous
MVAr in the pre fault base case scenario. After 150 ms, the and squirrel cage induction generator, as well as for the
fault was cleared by tripping the faulted line. It is assumed governors and exciters of the distributed generation formed
that none of the centralized and distributed generators based by controlled synchronous generators, standard models for
on synchronous or asynchronous machines are disconnected use in power system dynamics simulations can be found in the
during the applied fault. literature [4,5].
To assess the impact of distributed generation technology and The power electronic converter is modeled as a source of
penetration level on the transient stability, an indicator for the active power (P) and reactive power (Q). This is necessary
transient stability is needed. Here, two indicators have been because the grid representation in power system dynamics
applied to the rotor speed oscillations of the large scale simulation software and the typical time step used do not
generators in the test system that occur after application of the allow detailed modeling of power electronics and its

863
 Table 2. Results for the maximum rotor speed deviation as stability indicator
for various distributed generation technologies and penetration levels. The
number of the generator at which the value occurs is indicated in
parenthesis.
Dist. Maximum rotor speed deviation [10-3 p.u.] and relative
pen. change [%]
level
[%] ASM SM SMC PE PEC

0 6.4 (6) 6.4 (6) 6.4 (6) 6.4 (6) 6.4 (6)

0 0 0 0 0

9 6.1 (6) 6.1 (6) 6.1 (6) 5.9 (6) 5.9 (6)

-4.7 -4.7 -4.7 -7.8 -7.8

Fig. 2. One line diagram of the New England test system [3]
17 6.1 (1) 5.9 (6) 6.0 (6) 5.5 (6) 5.5 (6)
Table 1. Characteristics of the New England test system
-4.7 -7.8 -6.3 -14.1 -14.1
System characteristic Value
23 6.1 (1) 5.8 (6) 5.8 (6) 5.1 (6) 5.1 (6)
# of buses 39
-4.7 -9.4 -9.4 -20.3 -20.3
# of generators 10
29 6.1 (1) 5.6 (6) 5.7 (6) 4.7 (6) 4.7 (6)
# of loads 19
-4.7 -12.5 -10.9 -26.6 -26.6
# of transmission lines 46
33 6.3 (10) 5.5 (6) 5.5 (6) 4.3 (6) 4.3 (6)
Total generation 6140.7 MW / 1264.3 MVAr
-1.6 -14.1 -14.1 -32.8 -32.8
Total reactive power compensation 1408.9 MVAr
Table 3. Results for the oscillation duration as stability indicator for various
Total load 6097.1 MW / 1408.7 MVAr
distributed generation technologies and penetration levels. The number of
controllers for both theoretical and practical reasons [4, 6-8]. the generator at which the value occurs is indicated in parenthesis.
In PSS/E, no standard model for representing power Dist. Oscillation duration [s] and relative change [%]
electronics is available. Therefore, a so-called user-written pen.
model of a power electronic converter was developed and level
ASM SM SMC PE PEC
integrated into this program [5]. [%]
The prime mover of the synchronous and asynchronous 0 5.8 (5) 5.8 (5) 5.8 (5) 5.8 (5) 5.8 (5)
generators and the primary side of the generators grid coupled
through a power electronic converter is not taken into 0 0 0 0 0
account, because they are not of major importance in this 9 5.9 (5) 5.7 (5) 6.7 (9) 5.6 (9) 5.5 (9)
study. Details on how to model wind turbines, one of the most
frequently occurring distributed generation technologies +1.7 -1.7 +15.5 -3.4 -5.2
including their prime mover, can be found in [9, 10].
17 5.9 (9) 6.1 (7) 6.2 (7) 6.2 (9) 5.1 (2)
IV. RESULTS +1.7 +5.2 +6.9 +6.9 -12.1

A. Overview of results 23 6.5 (3) 5.9 (9) 6.1 (7) 7.4 (9) 5.1 (9)
In tables 2 and 3, the simulation results are summarized. In +12.1 +1.7 +5.2 +27.6 -12.1
the first row for each of the penetration levels studied, the
maximum, and thus worst values of the two stability 29 6.1 (9) 6.9 (9) 6.2 (7) 8.4 (9) 5.1 (9)
indicators used, namely the maximum rotor speed deviation
+5.2 +19.0 +6.9 +44.8 -12.1
and the oscillation duration, are displayed, together with the
numbers of the generators at which the given values occur in 33 5.7 (3) 7.4 (10) 6.3 (2/3) 8.5 (9) 5.1 (9)
parenthesis. These numbers correspond to figure 2. In each
-1.7 +27.6 +8.6 +46.6 -12.1
second row, the relative change compared to the base case is
given in per cent.
In the left column, the distributed generation penetration level induction generator, SM synchronous machine, PE stands for
is indicated. In the second row, the distributed technologies distributed generation which is grid connected through a
are indicated. ASM means asynchronous squirrel cage power electronics interface and if a C is added to the

864
acronym, the distributed generators are equipped with voltage which in turn slows down the speeding up of the synchronous
and frequency control. generators. When the asynchronous generator is at a larger
From tables 2 and 3, it can be concluded that it is difficult to distance and more weakly coupled to the synchronous
derive an overall conclusion with respect to the impact of generator, its speeding up during the fault will result in an
distributed generation technology and penetration level on increasing reactive power demand. This will result in a lower
power system stability, because the results are quite mixed. terminal voltage at the remote synchronous generator and thus
Therefore, each of the three main technologies, namely the in a decrease of synchronizing torque and a faster increase in
squirrel cage induction generator, the synchronous generator rotor speed.
and the power electronics interface will separately be In the cases studied here, both the synchronous generators
commented upon. and the distributed asynchronous generators are spread
through the system. The finding that distributed generation
B. Asynchronous generator based on squirrel cage induction generators seems not too
From tables 2 and 3, it can be seen that distributed squirrel have much effect on the transient stability of power systems
cage induction generators have an ambiguous influence on the can therefore probably be explained by noticing that both
two stability indicators. Both the maximum rotor speed effects occur simultaneously and counterbalance.
deviation and the oscillation duration are not very much
effected and the maximum value occurs at various generators C. Synchronous generator
when the distributed generation penetration level changes. It can be concluded from table 2 that an increasing
The shape of the oscillation is not very much influenced as penetration level of distributed generation based on
well, as can be seen from figure 3. In this figure, the rotor synchronous generators either with or without voltage control
speed deviation of the generators 1 and 6 is depicted for the leads to a decrease in the overspeeding of the synchronous
base case and for a distributed generation penetration level of generators during the fault. This result can be explained by
20 and 33 per cent respectively. By inspection, it can be seen realizing that the distributed synchronous generators are
from this figure that the oscillation duration is in the range of equipped with an excitation winding on the rotor, which
6 seconds. This is in the same range as the longest oscillation keeps the generators excited during the fault. As a result of
duration which occurs at generators 5, 3 and again 3 this, the distributed generators supply a fault current and the
respectively for the depicted distributed generation voltage during a fault does not drop as far as with no
penetration levels, according to table 3. These observations distributed generators present. As a result, the synchronizing
lead to the conclusion that distributed squirrel cage generators torque at the generators remains higher.
do not have very much influence on the transient stability of This reasoning is supported by simulation results depicted in
an electrical power system. figure 4. In this figure, the terminal voltage of generator 6,
This result can be explained as follows. As discussed in [11], which has the highest overspeeding during the fault as can be
the effect of squirrel cage induction generators on power concluded from table 2, is given for a penetration level of 0,
system stability depends on their distance to the synchronous 20 and 33 per cent of both uncontrolled and controlled
generators. If they are located near synchronous generators synchronous generators. It can be seen that the higher the
and the latter speed up during a fault, the stator frequency of penetration level of distributed synchronous generators, the
the asynchronous generators increases. This leads to a higher the terminal voltage of generator 6. The response is
decrease in the slip frequency and thus in generated power,

Fig. 4. Terminal voltage of generator 6 when varying amounts of distributed

Fig. 3. Rotor speed deviation of generators 1 (upper graph) and 6 (lower generation based on uncontrolled (upper graph) or controlled (lower graph)
graph). The solid line corresponds to the base case, the dashed and dotted synchronous generators are connected to the system. The solid line
lines to a distributed generation penetration of 20 and 33 per cent corresponds to the base case, the dashed and dotted lines to a 20 and 33 per
respectively. cent distributed generation penetration respectively.

865
Fig. 5. Rotor speed deviation of generator 10. The solid line corresponds to
the base case, the dotted line to a 33 per cent penetration level of
uncontrolled synchronous generators.

similar for distributed synchronous generators without and

with voltage and frequency control, which was to be
expected, because both have a similar effect on the amount of
overspeeding. Fig. 6. Node voltage of buses 4 (upper graph)and 8 (lower graph). The solid
line corresponds to the base case, the dashed and dotted lines to a 20 and 33
With respect to the oscillation duration, it can be concluded
per cent distributed generation penetration respectively.
from table 3, that it is increased when synchronous distributed
generation is connected to the system and that, in opposition
to what applies to the maximum rotor speed deviation, there Although the observed reduction in rotor speed deviations is
is a difference in the oscillation duration in the case of as such desirable, the price paid for it is a large voltage drop
controlled and uncontrolled synchronous generators. at some nodes due to the disconnection of the distributed
In figure 5, the rotor speed deviation of generator 10 with a generators when the fault occurs. This can be seen in figure 6,
33 per cent penetration level of uncontrolled synchronous in which the voltage of the buses 4 and 8 is depicted for the
generators is depicted together with the base case. It can be base case and for a 20 and 33 per cent penetration level of
seen that the damping of the oscillation is reduced by the uncontrolled power electronics. It can be seen that the voltage
distributed synchronous generators. This observation might drop becomes increasingly severe when more distributed
be caused by the interarea oscillation phenomenon [4, 12]. generation is connected to the system. This may lead to other
However, this conjecture can not be confirmed or denied by problems, such as tripping of generators and/or loads.
these results alone and requires more research. Further, it was Therefore, it might be questionable if the current operating
difficult to conclude whether this result occurs only in the test practice, in which power electronic converters are
system used or in general. Therefore, further research on the disconnected in case of a voltage drop in order to protect the
impact of synchronous distributed generators on the damping semiconductor switches and to fulfill utility requirements, is
of power system oscillations is necessary. optimal in the long run if the penetration level of this
technology will increase.
D. Power electronic converter The oscillation duration time shows substantial differences
As can be seen in table 2, the connection of distributed between uncontrolled and controlled power electronics
generation grid coupled through a power electronics interface interfaces. This might be caused by the governor models
results in a decrease in the maximum rotor speed deviation of fitted to the synchronous generators, as can be concluded
the synchronous generators. This is caused by the fact that from figure 7. In this figure, the rotor speed deviation of
distributed generators are quickly disconnected by their generator 9 is depicted for the base case and for a 20 and 33
power electronic converters when a fault occurs, because of per cent penetration of both uncontrolled and controlled
the terminal voltage drop. Thus, a lot of generation is lost in distributed generation based on power electronic converters.
case of a fault, reducing the acceleration of the rotors of the It can be seen that in case of uncontrolled distributed
synchronous generators. This also explains why the generation, it takes a long time until the rotor speed stays
overspeeding is further reduced, when more distributed within a bandwith of 110-4 p.u. as required according to the
generation is connected: the amount of generation lost at the definition. This is not only caused by the oscillations, but as
fault increases, resulting in a decreasing rotor acceleration. well by the long time it takes until a new equilibrium has been
The observation that the maximum rotor speed deviation is reached. This may be caused by the synchronous generator
the same in the cases of controlled and uncontrolled governor model and its parameters and will perhaps be
distributed generators can be explained as follows. The different when governor models that are able to change the
maximum rotor speed deviation is reached during the fault. mechanical power more rapidly are used. Further research
However, during the fault, both uncontrolled and controlled must therefore be devoted to the impact of uncontrolled
distributed generators grid coupled through a power power electronics on power system transient stability.
electronics interface are disconnected. Therefore, neither the The figure also shows that when controlled power electronics
uncontrolled nor the controlled generators are on line during is connected, the oscillations damp out quickly, which is in
the fault and as a result, the maximum rotor speed deviation is agreement with results presented in the literature [2].
the same in both cases.

866
 ACKNOWLEDGEMENTS

We would like to thank our colleague Robert H.A.M.

Reijntjes for developing the software to create and manipulate
load flow and dynamics files for PSS/E. Timothy F.
Laskowski from PTI, Schenectady, NY, is acknowledged for
his valuable help in interfacing the user made models of the
uncontrolled and controlled distributed generators grid
coupled through power electronics with PSS/E. The financial
support by the Netherlands Organization for Scientific
Research (NWO) is greatly acknowledged.

REFERENCES

Fig. 7. Rotor speed deviation of generator 9 after connection of uncontrolled [1] M.K. Donnelly, J.E. Dagle, D.J. Trudnowski, G.J. Rogers, “Impacts of
(upper graph) and controlled (lower graph) distributed generation. The solid the distributed utility on transmission system stability”, IEEE Transactions
line corresponds to the base case, the dashed and dotted lines to a distributed on Power Systems, v.11, n.2, p.741-746, 1996.
generation penetration of 20 and 33 per cent respectively. [2] F. de Leon, B.-T. Ooi, “Damping power oscillations by unidirectional
control of alternative power generation plants”, IEEE PES Winter Meeting
V. CONCLUSIONS AND DISCUSSION 2001, Columbus, US, Jan. 28th-Feb. 1st, 2001.
[3] M.A. Pai, Energy function analsysis for power system stability, Boston,
US: Kluwer Academic Publishers, 1989.
In the paper, the results of an investigation of the impact of [4] P. Kundur, Power system stability and control, New York, US: McGraw-
distributed generation technology and penetration level on Hill, Inc., 1994.
power system transient stability are presented. Starting from a [5] PSS/E 25, On-line Documentation, Schenectady, US: Power
Technologies, Inc., December 1997.
base case, the New England test system, the load was [6] Dispersed System Impacts: Survey and Requirements Study, EPRI TR-
increased in a number of steps and the load increases were 103337, Palo Alto, EPRI, July 1994.
covered by various distributed generation technologies. Then, [7] N. Hatziargyriou (ed.), Modeling new forms of generation and storage,
for each case a line fault was applied and the rotor speed CIGRE TF 38.01.10, Paris, November 2000.
[8] J. Arrillaga, N.R. Watson, Computer modelling of electrical power
deviation and oscillation duration of the large scale systems, J. Wiley & Sons, Ltd.:Chicester, UK, 2nd edition, 2001.
synchronous generators were studied. It was concluded that [9] J.G. Slootweg, H. Polinder, W.L. Kling, “Dynamic Modelling of a Wind
the impact of distributed generation on power system Turbine with Doubly Fed Induction Generator”, 2001 IEEE Power
transient stability depends both on the penetration level and Engineering Society Summer Meeting, July 15-19, 2001.
[10] J.G. Slootweg, H. Polinder, W.L. Kling, “Dynamic Modelling of a
the technology of the distributed generators. Wind Turbine with Direct Drive Synchronous Generator and Back to back
Distributed generation based on asynchronous generators Voltage Source Converter and its Controls”, 2001 European Wind Energy
does not have much impact on the transient stability, probably Conference and Exhibition, Copenhagen, Denmark, July 2-6, 2001.
because the opposite effects of near and remote generators [11] “Load representation for dynamic performance analysis”, IEEE
Transactions on Power Systems, v.8, n.2, 1993, p. 472-482.
counterbalance. Distributed generation based on synchronous [12] M. Klein, G.J. Rogers, P. Kundur, “A fundamental study of interarea
generators decreases the overspeeding of the large scale oscillations”, IEEE Transactions on Power Systems, v.6, n.3, 1991, p.914-
generators, but seems to decrease the transient stability by 921.
increasing the oscillation duration. The latter might be caused BIOGRAPHIES
by the interarea oscillation phenomenon. Distributed
J.G. Slootweg received his MSc degree in electrical
generation based on power electronics decreases the
engineering from Delft University of Technology on
overspeeding of generators, because it is disconnected during September 23rd, 1998. During his education he stayed in
a fault. However, at increasing penetration levels, this results Berlin for six months, to hear lectures at TU Berlin and to
in large voltage drops at some nodes, due to the loss of large conduct research at the Dynamowerk of Siemens AG. He
amounts of generation. The impact on the duration of the is currently working towards a PhD on the effects of large
scale integration of new technology on power system
rotor speed oscillations depends on whether the power dynamics. The research is carried out at the Electrical
electronic converters are equipped with voltage and Power Systems Laboratory of Delft UT.
frequency control or not.
In this paper, only one test system has been analyzed. It W.L. Kling received his MSc degree in electrical
engineering from the Technical University of Eindhoven
proved possible to explain most of the observations using in 1978. Currently he is a part time professor at the
generally accepted theoretical insights obtained from the Electric Power Systems Laboratory of Delft UT. His
literature. Therefore, it can be expected that in other test experience lies in the area of planning and operation of
systems similar results will be obtained. However, some power systems. He is involved in scientific organizations
such as Cigré and IEEE. He is the Dutch representative in
results could not be fully clarified. Especially these need the Cigré Study Committee 37 Planning and
further research, using other test systems and methods of Development of Power Systems.
investigation.

867