Вы находитесь на странице: 1из 6

Power System Stabilization

Using GPS and Second-order Eigenvalue Sensitivity


Tomoya Hsegawa, Keita h a s h i m a * , Toshiya Ohtaka, Shinichi Iwamoto.
Ikpmmcnt ofEla.tncal. Eltilrunicr and Coinpuling Engmmng. Waudi Unirr.nily. ToA)o. JAPAN

Abslracl --As the electric power demand increases, power monitoring power system^.'^"^' It is reported that it is
system have become larger and more complex. The possible to calculate phase angles on real time by bus
dependence on the electric power in snciely tends lo increase voltage wave forms of buses, hecause GPS has a function
such as information products prevail. On the other hand, giving accurate synchronization signals.
recently studies on applications of Global Positioning In this paper, we use on line GPS phase angle
System (GPS) to power system monitoring are being
conducted. In this paper, at first we propose a GPS measurements to monitor a power system. And we
placement method with a viewpoint of eNectively propose a GPS placement method with a viewpoint of
monitoring the power system stabilily. This method uses the effectively monitoring the power system stability. This
lint order eigenvalue sensitivities for phase angles from method uses the first-order eigenvalue sensitivities for
relationship between eigenvalues and phase angles. Next, we phase angles from the relationship between important
propose a method for delermining more appropriate eigenvalues and phase angles. Moreover, we propose a
generation dispatch rescheduling to improve the power method for determining a more proper generation
system stability when the power syslem stability deceases. dispatch rescheduling for improving the power system
At that time we use lint and second order eigenvalue
stability when the power system stability deceases. At
sensitivities. The proposed method is demonstrated using a
sample 4 machine 10 bus system and we examine first we use first-order eigenvalue sensitivities because
eNectivenesr of the second order eigenvalue sensitivities eigenvalues are important for the dynamic stability
compared with the first order ones. analysis and phase angles are related to generator powers.
Next we use more proper second-order eigenvalue
Index Ienns -.Power system dynamic stability, Global sensitivities to nonlinear changes of eigenvalues. The
Posilioning System Eigenvalue, Eigenvalue Sensitivity proposed method is demonstrated using a sample 4
machine IO-bus system.
I. INTRODUCTION
11. GPS
As the electric power demand increases and IPP
(Independent Power Producer) participates for power Global Positioning System (GPS) is a system which
industry deregulation, power systems have become larger uses the satellites of USA. It is well know for being used
and more complex. The dependence on the electric power in navigation systems. GPS has a function of giving
in society tends to increase such as information accurate synchronization signals. By this function,
technology products (Personal Computer etc) prevail. simultaneous measurements are possible at some remote
The problem for the electric power demand will have sub-stations. I n the power system field, many studies
more social and/or economic influence. It is important to using this function have been reported. A point of the
keep the reliability and stahility of supply grasping common fact among them is to develop techniques for
operating condition of the power system. measuring phase angles on real time.
. The power system analysis considering the dynamic An actual scenario is shown as follows. GPS receivers
stability, which is one of the power system stabilities, is and measuring instruments for bus voltage wave forms
becoming more important. The eigenvalue method, the are set up at remote sub-stations. Synchronizing signals
frequency response method and the root locus method from GPS and bus voltage wave forms are measured and
have been studied for the analysis and control of the recorded simultancousl y. The information of sub-stations
dynamic stability. Especially the eigenvalue method is by these systems is sent to the main station through
usually used. This method uses the eigenvalues of the digital circuits. Phase angles among sub stations can be
system matrix by linearizing the dynamics of the system calculated on real time by measuring the time differences
at an operating point. It is possible to monitor the power of zero cross points of bus voltage wave forms using
system by monitoring the eigenvalues. However the synchronizing signals. It is possible to measure the bus
calculation of the eigenvalues is difficult and voltage wave forms in 3200 Hz and transmit the
time-consuming. Fast methods calculating the information (synchronizing signals, bus voltage wave
eigenvalues and also decentralized hierarchical forms) of sub-station to the main-station every 200 ms.
eigenvalue control methods have been already published Therefore phase angles can be calculated on real time.
The sensitivities of the eigenvalue parameters are The outline of phase angles measuring system using GPS
usually used for the eigenvalue c~ntrol.‘’~[~’ This is shown in Fig. I .
sensitivities are considered as effective means, as they
are easy to be used to adjust parameters for the power
system stability.
On the other hand, recently studies on Global
Positioning System (GPS) are being conducted for

0-7803-7525-4/02/$17.00 0 2002 IEEE 1670


aA
-U +A-
an
au =-ut au
1- (3)
as, as, as, as,
And multiply (3) by left eigenvector vT of the
I eigenvalue I o f the system matrix A from left.

As vT i s the left eigenvector o f the eigenvalue I of the


system matrix A , (5) is represented as follows.
v r A = v'1 (5)
d u l i l n l It",,
Bv w i n e ( 5 ) . (4) can be rewritten as follows.
,n-au
I

i C I , S ,,..,.,wrj v ' -JA


u+v'R-=v'u-+v au an (6)
i I . i l V l l ~ i l lYl I I I I !
as, as, as, as,
vTu i s 1 from the property o f the eigenvector. As a
Fig. I Phase angle measuring system using GPS result, eigenvalue sensitivities for phase angles o f the
eigenvalue I in (7) are shown as follows by (6).
111. GPS PLACEMENTMETHOD
PROPOSED
(7)
The real parts of bad damping eigenvalues are
important for the eigenvalue analysis o f the dynamic Eigenvalue sensitivities for phase angles i n (7)
stability. I t i s necessary to monitor these eigenvalues for represent how the eigenvalues change for small changes
the power system stability. I t is difficult to always o f phase angles. We propose an index S by (8) which is
monitor eigenvalues, hecause it i s difiicult to calculate eigenvalue sensitivities o f phase angles to the
eigenvalues for large power systems. So we consider eigenvalue A,,, as a GPS placement method considering
monitoring changes o f eigenvalues by monitoring the dynamic stability.

I
changes of phase angles relating to eigenvalues i n place
o f eigenvalues. GPS can measure phase angles. That is. it S, = Re ~

is possible to monitor power systems by placing GPS


measuring phase angles on real time. where, I,,, is the crucial eigenvalue
I n this chapter, we propose a method for determining GPS placement ranking i s decided with the philosophy
most effective GPS placement buses for the dynamic that the index S is the highest i s most important. As real
stability. GPS i s placed not at all buses but at some buses part of eigenvalues influence the dynamic stability, this
o f the power system. As a GPS placement method, we index S uses only real parts o f eigenvalue sensitivities.
use first-order eigenvalue sensitivities for phase angles, As it i s necessary to monitor the generator having the
which called the GPS first-order eigenvalue sensitivities, highest influence to power system regardless o f the sign,
because eigenvalues are important for the eigenvalue direction o f the eigenvalue sensitivities, this index S uses
analysis o f the dynamic stability and GPS can calculate the absolute values of the eigenvalue sensitivities.
phase angles on real time.
IV. METHOD
RESCHEDULINGDISPATCH
The linearized state equation of ( I ) i s obtained by
linearizing the dynamic equation near an operating point A. Ourline
in the power systems
X=Ax (1) I t i s possible to control real part of eigenvalues by
where, readjusting parameters in the power system equation. I n
x : Generator state variable vector this chapter. we propose a generation dispatch
A : System matrix rescheduling method using eigenvalue sensitivities for
It is possible to evaluate the dynamic stability, which is phase angles as the method for shifting the eigenvalue
one of the power system stabilities, by evaluating from the unstable area to the stable area.
eigenvalues o f the system matrix A o f (I).Eigenvalues o f We consider the generation dispatch rescheduling
the system matrix A exist to the number of state considering the dynamic stability as an optimization
variables. However important eigenvalues having bad problem. Bad damping eigenvalues, whose real parts are
influence in the power system are only bad damping located near the positive area, influence stability o f the
eigenvalues. I n this chapter, we pay attention to the power system. Therefore the objective i s an optimization
eigenvalue influencing the power system stability. o f the real part of the minimum eigenvalue.
GPS first-order eigenvalue sensitivities are derived as
follows. (2) i s represented by right eigenvector U of the B. Generation dispatch rescheduling using firsf-order
eigenvalue h o f the system matrix A . eigenvalue sensitivity
Au=h (2)
To obtain eigenvalue sensitivities for phase angles o f the B y using eigenvalue sensitivities, the eigenvalue i s
shown as follows.
eigenvalue 2 , we carry out differentiation o f (3) in terms
o f phase angles. AA,mn= -As, a?","
, as,
Therefore the optimizing objective function uses the real

1671
part of the minimum eigenvalue as follows. A is eigenvalue, U is right eigenvector and v r is left
f =Re(A?,,,) (10) eigenvector for the eigenvalue. Then (1.1) and (14) is
Here variables are generation powers. However formulated.
variables of (IO) are phase angles. Thus we use phase Au=h (13)
angles ohtained by the relationship between generation vfA= v'l (14)
powers and phase angles in the objective function (IO). (15) and (16) are given hy differentiations of (13) and
We change values of generator powers which have (14) in terms of phase angles.
upper and lower limit constrains and the sum of changes
is zero. We need to impress constrains on other
eigenvalues to prevent the real part of the eigenvalue
except the minimum eigenvalue from hecoming bigger
hy the generation dispatch rescheduling.
where, I is a unit matrix having the same
Formulation is as follows that dimension of the system matrix
( i ) Objective function : min f As the sensitivity ?!5 of U is orthogonal to v ' , (17)
( ii ) Equality constrains : a 8,
generator e;,= 4 . v,; = v, is written as

load P,, =5 % Q,, =Q,


generator change value
zw;,= 0 As the sensitivity d.I of v' is orthogonal to U, (18)
as,
(iii) Inequality constrains is written as
-
generator - 5 e;, 4, 5 K. -
Q,, 5 Q,,, 5 Q ~ ; , aV'
U-=o (18)
voltage
145 Iv, I 5 a 8,
(19)isobtainedby(15),(16),(17)and(18).
eigenvalue WL"+AA""") < A,,,
where,
e. .Q,
: active, reactive power
Therefore the GPS second-order eigenvalue sensitivities
v, : voltage which are the second-order eigenvalue sensitivities are
written a s (20) by using (19) to (12).
4 : changes of generation power

A", : reference of real part of eigen value


-
: upper limit, lower limit We propose generation rescheduling dispatch method
- using the GPS second-order eigenvalue sensitivities. The
II : absolute value eigenvalue using the GPS second-order eigenvalue
sensitivities for the objective function is written as
C. Generation dispatch rescheduling using
secofid-order eigenvalue sensiriviry

The generation dispatch rescheduling using GPS Minimizing the objective function we use the real part of
first-order eigenvalue sensitivities might not calculate an the eigenvalue.
optimal power, because changes of eigenvalues for f = Reb?,,,. (22)
changes of system state are large. Thus we propose a We carry out the same process as the generation dispatch
generation dispatch rescheduling method using the rescheduling using the GPS first-order sensitivities.
second-order eigenvalue sensitivities for phase angles Procedure of the proposed method is shown as Fig.2.
(GPS second-order eigenvalue sensitivities). This method
can determine more proper generation powers for
changes of system states than the GPS first-order
eigenvalue sensitivities.
The GPS second-order eigenvalue sensitivities which
are second-order eigenvalue sensitivities for phase angles

I '
I
are derived as follows. The GPS first-order eigenvalue
sensitivities are expressed as in ( I 1) by Chapter 3. Check ofthe p w e r system stability
31
-= y' -U
aA
(11) I
as, as, Rescheduling Generation dispatch
The GPS second-order eigenvalue sensitivities (12) are using GPS sensitivities
obtained by differentiating ( I I ) in terms of phase angles.

. .
A is the system matrix of the linearized state equation. Rg.2 Flawchan of generationrescheduling dispatch

I672
V. SIMULATION
I Fist-order GPS Index S
A. Example with 4-machine IO-BusSystem GI I 3.887+3.600i I 3.887
~

Simulations are carried out for a 4 machine IO bus G2 -5.317fS.284i 5.317


system"', shown in Fig.3 G3 1.678+1.119i I 1.678

The index S of G2 (generator 2) i s the maximum as


seen from Tahel.3. Thus placing GPS at generator 2 i s
most effective for monitoring the power system. Thus we
consider that GPS placement following the ranking of
index S i s more effective.
We compute changes of the real part of the eigenvalue
Ln to confirm whether the values o f index S are true.
Fig.4 shows the relationship between the changes o f the
generation powers and the real part o f the eigenvalue
knwhen each generation power changes 10% from the
initial state.

Fig.3 4 machine 10 bus system

The slack bus i s bus 0,


and the system base capacity
i s IOOO[MVA]. T h e generator parameters are shown i n
Table. I .

Tahle I Generator Panmeters


Capacity M Xd Xd' Xq
I
-lo"

[MVAI [pul [PUI IPUI LPUI


GI 4800 8.0 1.7 0.297 0.32 Rg.4 Relationship between generation power and eigenvalue
G2 2400 8.0 1.7 0.297 0.32
From the slope o f each generation corresponding to the
G7 2400 8.0 1.7 0.297 0.32
eigenvalue sensitivity i n Fig.4. we find the calculation of
Index S is appropriate and Generation 2 i s most
important for monitoring power system stability because
the slope of the generation 2 is steepest.
B. GPS Placement
C. Generation rescheduling dispatch
We calculate eigenvalues at thc operating point. Three
eigenvalues which have much influence to the power The initial state i s the same, and the upper and lower
system are shown i n Tahle.2. limits for optimal power flow calculation i s &IO%, and
the eigenvalues and eigenvalue sensitivities are the same
Tahle Z Eigenvalues influencing power system as in Table.2 and 3.
E!@L."".%l"e We change the power o f all generations by calculating
their participations. When we carry out the optimal
Eigenvalue1 -I,064f2.232i
power llow calculation using these values, the optimal
rescheduling generation dispatch satisfying constrains are
Eigenvalue3 -I,084i2.588i ohtained as in Tahle.4.

Table 4 Generator rescheduling dispatch


(GPS first-ordereigenvalue sensitivity)
Generator I Beforelllul I After loul
GI 3.500 I 3.584

G3 II 1.702 II 1.7oh

Tahle.5 shows the eigenvalue calculated hefore and


after the generation dispatch rescheduling. The shifts o f
three important eigenvalues computed hefore and after
rescheduling generation dispatch are shown i n Fig.5. And

1673
Fig.6 shows the change of generator 2 phase angle before Table 7 Eigenvalue before and after rescheduling dispatch
(GPS second-order eigenvalue sensitivity)
and after the rescheduling generation dispatch.
Eigenvalue
Table 5 Eigenvalue before and after rescheduling dispatch
(GPS firsborder eigenvalue sensitivity)
Before I -1.O64*2.2321

Eisenvaluel After -1.069f2.26li

After -1.066t2.243i

0 9 t
f
-1101 -,m, -101, ID,, IO.,

rc.1 part t
CJ -e AA
e e ~'
eigm value2 + eigen "al."c,l
sigen valve3

sigeen valve3
Rg.7 Changes of eigenvalues before and after rescheduling dispatch
(GPS second-ordereigenvalue sensitivity)
Rg.5 Changes of eigenvalues before and after rescheduling dispatch
(GPS firs-order eigenvalue sensitivity)

Rg.8 Generator 2 phase angle before and after rescheduling dispatch


(GPS second-order eigenvalue sensitivity)
Rg.6 Generator 2 phase angle before and after rescheduling dispatch
(GPS first-order eigenvalue sensitivity) From Fig.7, eigenvalues shift to the stable side. Fig.8
shows the changes of generator 2 phase angle before and
From Fig.5, the eigenvalue Ln shifts to the safer area after the generation dispatch rescheduling. The damping
and the second minimum eigenvalue a, becomes after the generation dispatch rescheduling is smaller than
bigger by using the proposed method. This i s the reason before. That is. the generation rescheduling dispatch
why the eigenvalue & , differs from the minimum using the GPS second-order eigenvalue sensitivities is
eigenvalue in the influence to the change of generation effective in improving the power system stability.
powers. However it is no problem for the power system The changes of Generation 2 phase angle by the GPS
because there are eigenvalues in the safe area and same first and second-order eigenvalue sensitivities are shown
result is obtained in other case. Therefore we found that it in Fig.9.
is possible to shift eigenvalues to the safe area by using
this method.
Next we carry out the optimal power flow calculation
using the GPS second-order sensitivities and the
constrains which are the same as in the case of the GPS
first-order sensitivities. The generation power before and 1
3
after the generation dispatch rescheduling are presented
in Table.6. Similarly the eigenvalue L,before and
after the generalion dispatch rescheduling is presented in
Table.7. *t
Table 6 Generator rescheduling dispatch
(GPS second-order eigenvalue sensitivity)
Generator Before [pul After [pu] Rg.9 Generator 2 phase angle after rescheduling dispatch
(GPS fin1 and second order eigenvalue sensitivity)
GI 3.5w I 3.500
G2 I.7w
I 1.530 From Fig.9, the damping by the GPS second-order
sensitivities is better than by the GPS first-order ones.
These results show that the method using the GPS
1674
Small Signal Stability Analysis of Large Power Systems". IEEE
Summer Meeting. 2000
[lo] T.hasegawa, T,Ohtaka, S.lwamoto :"GPS Placement and
Application for Power System Stability Analysis". Power System
Symposium of IEE of Japan, PEW-73
1111Y.Sekine :"Transient Analysis Theory in Power System (in
Japanese)". Ohm Co., 1984

VIII. BIOGRAPHIES

Generator Before [pul After [pu] Tomoya Hasegawa was born in 1976. He received his B.E.,M.E.
.from Waseda University, Tokyo. Japan in 1999. 2001 respectively. He
GI 3.500 3.502 is presently with IBM. His research interest is mainly eigenvalue
sensitivity and state estimation.
GZ 1.700 1.530
Keita I m a s h i m was ham in 1978. He received his B.E. from
Waseda University, Tokyo, Japan in 2001 respectively. His research
interest i s mainly eigenvalue sensitivity and state estimation.
From Table.6 and 8, the rescheduling generation dispatch
by using true eigenvalues i s almost equal to the one by Toshiya Ohtaka was born in 1976. He received his B.E. and M.E.
using the GPS second-order sensitivities. This confirms degree5 from Waseda University. Tokyo, Japan in 1998, 2000,
proposed method can determine proper generation respectively. He i s presently a student of W.D. c o m e at Waseda
University. His research interests are mainly load flow computation.
powers speedily and accurately. voltage stability analysis. harmonic circuit analysis. and energy stonge
system applications to power systems.
VII. CONCLUSION
Shiniehi l w a m t o was born in 1948. He received his B.E.. M.E.. and
In this paper, we have proposed a GPS placement Ph.D. degrees from Waseda University, Tokyo. Japan in 1971. 1975,
1978. respectively. Fmm 1972 10 1974. he was at Clarkson University,
method with a viewpoint o f effectively monitoring the U S . . and received his M.E. degree. He is presently a full professor at
power system stability. Next we have proposed a Waseda University. From 1992 to 1993. he was at University of
rescheduling generation dispatch method to move Washington as a visiting professor. His research interests are inaioly
eigenvalues at the operating point to a more stable area load flow computation. voltage stability analysis. transient stability
analysis and GPS applications to power systems.
by using the first-order eigenvalue sensitivities.
Furthermore we have proposed rescheduling generation
dispatch using second-order eigenvalue sensitivities for
the changes o f eigenvalues. We could determine a GPS
placement bus more effectively and make sure
effectiveness of the proposed method by simulations.
Also, power system stability i s improved by the
rescheduling generation dispatch using the second-order
eigenvalue sensitivities. The proposed method i s founded
to he effective for monitoring and controlling power
system stability.

VIII. REFERENCES

111 HSasaki. F.lshikawa,A.Trrasako.J.Kuboltawa :"Steady-state


stability analysis o f large-scale electric power system using
modified Arnold method.". Power Systeni Symposium of IEE of
Japan. PE9S-91
121 M.Nagata. A.Yokoyama :"[Xcentralized and Hierarchical
Eigenvalue Control Using Network lkwinposition". IEE of Japan.
PE-96.109
[31 F.lrhikawa. H.Sasaki :"An Efficient Algorithm for Evaluating the
Eigenvalue Sensitivity in Large Power Systems". Power System
Symposium of IEE of Japan. PE96-62
141 L.Xu. S.Ahmed-&id :'Tuning of Power System Controllers
Using Symbolic Eigcnsensitivity Analysis and Linrar Programing".
IEEE trans. on power systems. vol.lO. no. I, February, 1995
151 0,takahashi. Y.sato.H.tokita :"System oscillation recording
analysis system.". Power System Symposium of IEE of Japan,
PE94-166
IS1 'A.GPhadke :"Synchroniled Phasor Measuremena in Power
".
Systems IEEE Computer Application in Power, pp.10-15. April
I993
[71 A.kurita :"FACTS Vision proposed by EPRI-Fundamental Survey
and Numerical Study-". CRIEPI. 792020, 1993
[SI H.K.Nam, Y.K.Kim, K.S.Shim. K.Y.Lee :"A New
Eigen-Sensitivity Theory of Augmented Matrix and its Applications
to Power System Stahilily", IEEE inns. on power systems, ~01.15.
"0.1, February, 2000
191 K.S.Shirn, H.K.Nam, S.GSong, Y.GKirn. K.Y.Lee :"Application
Results of the Eigen-Sensitivity Theory of Augmented Matrix to

1675

Вам также может понравиться