Beruflich Dokumente
Kultur Dokumente
September 1991
529
M. Taleb
Electrical a n d C o m p u t e r Engineering D e p a r t m e n t
Clarkson University
Potsdam. New York 13699
Abstract - Structural dynamics of small and large induction machines are investigated using integral manifolds and
are shown t o be markedly different. Whereas the dominant
behavior in small machines is characterized by a well-known
first-order speed model, the corresponding behavior in large
machines is characterized by a new first-order voltage model.
Nonlinear simulations and eigenvalue analyses illustrate the
validity of the structural models developed.
Key Words: Induction Machine, Reduced Order Modeling,
Integral Manifold, Eigenvalue Analysis, Nonlinear Simulation.
1. I n t r o d u c t i o n
However, this overly simplified speed model does not accurately predict the dynamic performance of relatively large induction motors as reported [1],[5]. A simplified second-order
model in the rotor flux magnitude and the shaft speed is presented in [4] as an alternative t o the third-order model.
Recent studies have implicated the induction machine and
other regulating equipment in certain mechanisms of voltage
instabilities and voltage collapse [6]-[lo]. In [6], the authors
use a detailed fifth-order model t o capture an aperiodic voltage instability attributed t o the dynamics of the induction
machine terminal voltage. Using a first-order slip model, the
authors in [7] show that voltage collapse can occur before
the load flow Jacobian becomes singular if load dynamics are
taken into account. In [8], the authors also use slip models t o
analyze the phenomenon of cascaded voltage collapse and, in
193, they study the reverse action of tap-changing-under-load
(TCUL) transformers using the same type of dynamic loads.
In [lo], the authors use a second-order model in the terminal
voltage and angle, obtained originally in [2] from field tests, to
analyze voltage collapse based on a center manifold model describing the system dynamics after a saddle-node bifurcation
of the solutions.
This paper addresses the issue of adequacy of certain reduced order models of induction machines in describing system phenomena such as voltage instabilities and voltage collapse in general. T h e approach taken here uses integral manifold theory [11]-[12] t o extract structural models describing
the dominant dynamic behavior of small and large induction
machines. It is shown that these models are radically different
implying that the mechanisms and dynamics governing these
instabilities are different for each type of machine.
2. Standard A n a l y s i s o f a F i f t h - O r d e r Model
0885-8969/91/WoO-0529$01.ooO1991 IEEE
530
0
= V d r = &idr
0 = vqr = Rriqr
1 d$dr
-w s dt
-U)
+qr
(Ws
(3)
ws
1 d+qr
+ -w s dt
I (us - U )
ws
d'dr
(4)
(5)
(6)
lltqs
=
=
$dr
Xlridr
'$ds
Xlsiqs
Xiriqr
$qr
+
+
+
+
+
+
+ X m ( i q s+ &)
Xm(ids
idr)
(7)
Xm(4qs
iqr)
(8)
Xm(ids
idr)
(9)
=x
=x
x' = xi
s = XIS + x m
X r = Xlr Xm
= x; = x , - x;
Xr
=x
(11)
(12)
(13)
O-
(14)
wsRr
EA = -6
Xm
xr $qr
E: = X
(15)
$ d r
--1
dd'ds
ws
dt
-ws dt
d*qa
-I-$ q s
-c
R
= -*d*
,,/,tq,
+F E : +
vds
-F
R s ~+:
vqs
(17)
- (w" - w)TAEA
(18)
+ (ws -
(19)
I
I
(16)
x-x'
XI EA7
3. Exact Stator M a n i f o l d s
R s
+ Fx -$x 'd
dE' = --X
TA--$
XIE:
TAdE:
dt
- -$*ds
3rd-~rder
500 h p
-41.8 f j373.8
-15.4 f j41.5
-27.5
-15.6 f j41.3
-27.3
(10)
For a smooth-air-gap machine, the direct- and quadratureaxis reactances are equal so that
5'h-order
50 h p
-49.5 f j355.9
-142.5 f j42.5
-14.0
-134.1% j41.1
-14.4
v,
*qs
=0
(22)
53 1
If the initid conditions of $ds and $ q s do not lie on the
manifolds ( 2 1 ) - ( 2 2 ) , then we define the off-manifold variables
Tds
dds
vqs
$qs
-v
(23)
(24)
which are the deviations of $ds and ?,bps from the manifolds
( 2 1 ) - ( 2 2 ) . Substitution of ( 2 3 ) - ( 2 4 ) into (16)-(17) yields the
decoupled set of differential equations
Eh2
Eh1
2R
+ +)
= A c o s ( w s t + +)
Ods(t) = A sin(wst
vqs(t)
(27)
(28)
where A = J&(O)
&(O) and t a n + = 7 1 ~ ~ ( 0 ) / 7 1 ~T~h(e0 ) .
complete solutions for the stator transients are
+v
$ds(t)
vd!s(t)
(29)
$qs(t)
= vqs(t)
(30)
(E; - ( X 2X
x )V )Z
Ed
= (X-XZV2
4x?
(34)
X T,
E& = V
w-ws
1 EA
---
T E&
E q u i l i b r i u m P o i n t s and S e n s i t i v i t y A n a l y s i s
Neglecting stator resistance, the simplified third-order
model of an induction motor connected to an infinite bus is
dE;
Ti- d t
dE&
TA-
at
M -dw
dt
X
= -FE;
x - X
+ (7+V
+C(W- w s ) E &
x
-FE& - T ~ ( wwS)E:
-V-E&
T,
X
(31)
(32)
(33)
W - W s
--1
(39)
532
Table 2
Participation matrices
5 0 - h ~Machine
Eigenvalues
I E& I
X1,z = - 1 3 6 . 2 f j20.7
0.604 I
XB = -15.8
0.136
5 0 0 - h ~Machine
Eigenvalues
EA
X ~ , Z= -15.6f j41.9
0.539
XJ = -27.9
0.023
- I
~~
~~~
I
I
E:,
0.539
0.006
operation ranges between 0 and -45 and unstable operation between -45 and -90.
1
0.065
1.130
Et
0.018
1.005
0.526
0.018
It is observed that the speed is the dominant variable in characterizing the real eigenvalue in the small machine whereas
the voltage variable E; is the dominant variable in characterizing the real eigenvalue in the large machine. This real
eigenvalue, negative for stable operation and positive for unstable operation, is exactly zero for maximum steady-state
torque [l, p. 3161. This important observation leads us t o
believe that small and large induction machines are characterized by different dynamical structures.
Jm
(40)
dE
TL-
=
dt
d6
-d t =
dw
M - dt
X
--E+(X
w-ws-
x-X
)Vcos6
X
X - X Vsin6
Table 3
(-7)x
VE sin 6
--X
Tm
(44)
50
500
0.00044
0.00047
(3)
(3)
0.0210
0.0217
TI2
E=hi(R,m)
0.1557
0.0069
I 0.7834 I 0.0339 I
TI2
6=hz(R,- M X )
(45)
TL
JMX
(42)
(43)
MX
(2)
P
(hp)
tan-(R - a,)
(47)
These approximations can be viewed as the solutions t o (42)(43) taken as algebraic equations although this may not be
true in general. T h e slow model governing the dominant dynamics of small induction machines on these manifolds is
MXdR
TI dt
--v
X-X
(0-0,)
(- x ) 1 (n- R,)2
(48)
E&
(+qV(R
1
+ (0-
- R,)
Rs)2
(50)
533
We can also derive a second-order model in the fast deviations of E and 6 from the manifolds (46)-(47). However, this
model does not present any interest as the corresponding offmanifold variables decay exponentially fast towards the rotor
manifolds. Hence, after a short time following an initial transient, these variables reach the manifolds (46)-(47) and remain
there. T h e remainder of the dynamics of the induction machine is governed by the speed model (48) which becomes the
dominant behavior.
m,
The rotor integral manifolds of both machines are illustrated in this section using both a linear and a nonlinear analysis. Table 4 summarizes the main diffefences in the linear
analysis of small and large induction machines. In the case
of the 50-hp machine, the real eigenvalue is attributed t o the
shaft speed whereas the well-damped complex mode corresponds to the rotor electrical transients described by E and
8. T h e real eigenvalue in the first-order model is obtained by
linearizing the speed model (48) around the riominal speed of
operation. This traditional order reduction procedure is erroneous for the large 500-hp machine and the dominant variable
is the voltage E. The real eigenvalue in the first-order model
is obtained by linearizing the voltage model (56) and it approximates very well the real eigenvalue in the corresponding
third-order model. This clarifies the large discrepancy observed in Table 1. Table 4 also shows a second-order model
for each machine and whose eigenvalues are close t o the fast
complex modes in the third-order models. These models describe the off-manifold variables of each machine [11]-[12].
m / T .
Table 4
I
I
I E,6 I
6 = sin-(-=)
XT,
(52)
a-n, = 0
(53)
Equation (53) does not mean that w - u s = 0 which is clearly
incorrect. It should be remembered that we are performing
an asymptotic analysis where m / T is an infinitesimal
parameter that is capable of assuming arbitrary small values
in the limiting_process.
By virtue of (531,
.
,. the R-manifold
has a nontrivial first-order approximation in the parameter
m / T given by [12]
~
rn
R - 0, g --
XT,
J ( V E ) 2 - (XTm)2
(54)
which describes the dominant behavior of large induction machines. This new result has an important bearing on the corresponding dynamic behavior of large and small induction
motors. Clearly, the dominant behaviors are quite different
and are governed by the reduced models (48) and (56), respectively. Further analytical work is needed t o understand
the mechanisms of voltage instabilities associated with large
induction motors.
I E I
-28.5
I
I
The rotor integral manifolds are best illustrated by a nonlinear simulation as shown in Figures 2 and 3. T h e disturbance considered is a 0.3 pu voltage reduction in the infinite bus voltage lasting 100 ms and followed by a sinusoidal
oscillation of amplitude 0.1 pu and frequency 2 Hz. After
the voltage dip, the stator manifolds are approximated by
$& g 1 0.1 s i n 4 r ( t - 0.2) and $ q s Z 0 in the third-order
models. T h e solid lines represent the exact solutions for E, 6
and w - ws as obtained from the third-order models whereas
the dashed lines correspond t o first-order models and various
approximations of the rotor manifolds.
In the case of the 50-hp machine, the approximate dynamics are simulated using the well-known speed model (48) and
the zeroth-order approximations of the E- and &manifolds.
As expected, the speed variable is approximated uniformly
during the simulation period. There is also a good agreement
between the variables E and 6 obtained from the third-order
model and the manifold approximations (46)-(47). For small
speed deviations, E and 6 are respectively proportional to
terminal voltage V and speed deviation w - ws.
In the case of the 500-hp mkchine, the approximate dynamics are simulated using the new voltage model ( 5 6 ) and
various approximations of the rotor manifolds (51). Here, i t is
the voltage variable E which is approximated uniformly during the simulation. The influence of manifold approximation
is shown in the plot of 6 where the two dashed curves repre-
534
1.20
1.20
0.60
0.00
1 .oo
0.50
0.60
0.00
1.50
0.50
1.08
1.08
I
I
1
I
,
1
I
I
1,
0.92 -
1.50
0.92
Q
1 .oo
time (s)
time (s)
0.76
third-order model
zeroth-order E'-manifold
0.60
0.00
0.50
1 .oo
0.60
0.00
1.50
0.50
1 .oo
1.50
time (s)
time (s)
-O.O*
n
U-0.13
Lo
-0.18
-"
It
'
8
0.00
_____
third-order model
zeroth-order 6-manifold
second-order 6 -manifold
third-order model
zeroth-order &-manifold
0.50
1 .oo
1.50
-0.40
1 .oo
1.50
time (s)
\
U
0.50
0.00
time (s)
<
-3
-15-
3"
I
-third-order
first-order
-25
0.50
model
model
I I
1 .oo
I I 1 1 1 1 I I I 1 1 1 I I I I I I I
0.00
3*
I
3
I I I I I I I I I I
1.50
time (s)
Figure 2: Solutions of t h e third- and first-order models
for t h e 50-hp induction machine.
third-order model
first-order w-manifold
-13
1 1 1 1 I I I I I I 1 1 1 1 I I I I I ~ I II II I I I I I I
0.00
0.50
1 .oo
1.50
time ( s )
Figure 3: Solutions of t h e third- and first-order models
for t h e 500-hp induction machine.
535
sent the zeroth- and the second-order approximations of the 6manifold, the first-order term being identically zero. T h e plot
of w - w 3 is represented using the first-order approximation
(55) since the zeroth- and second-order terms are identically
zero. Of course, i t cannot be expected for these approximations to yield an exact duplicate of the solid lines since the
off-manifolds may have been excited in the process. However, since they are well-damped, their effects should only be
significant during early transients before they disappear.
6. Conclusions
The structural dynamics of small and large induction machines have been investigated using integral manifold theory
and have been shown t o be quite different. T h e results obtained in this paper are significant for voltage studies since
they propose a new first-order voltage model for large induction machines which describes their dominant behavior. The
paper also identifies key parameters for induction machines
which classify them according t o their dominant behavior.
Acknowledgements
T h e authors would like t o thank Prof. P.W. Sauer of
the University of Illinois at Urbana-Champaign and Mr.
W.W. Price of the General Electric Company for their valuable discussions and continued interest in the subject. This
research was supported in part by the National Science Foundation Grant ECS-9058174.
[7] R. J. Thomas and A. Tiranuchit, Dynamic Voltage Instability, Proceedings of the 26th Conference on Decision and Control, vol. 1, pp. 53-58, Los Angeles, CA,
December 1987.
[9]H. Ohtsuki, A. Yokoyama, and Y. Sekine, Reverse Action of on Load-Tap Changer in Association with Voltage
Collapse, Paper 90 WM 095-0 PWRS, IEEE/PES Winter Power Meeting, Atlanta, Georgia, February 1990.
[lo] H.D. Chiang, I. Dobson, R. J. Thomas, J. S. Thorp and
L. Fekih-Ahmed, On Voltage Collapse in Electric Power
Systems, IEEE Transactions on Power Systems, vol. 5,
no. 2, pp. 601-611, May 1990.
References
Biographies
Severe Disturbances, CIGRE Paper 38-18, International Conference on Large High Voltage Electric Systems, Paris, August/September 1986.
[3] D. S. Brereton, D. G. Lewis and C. C. Young, Representation of Induction-Motor Loads During Power-System
Stability Studies, A IEE Transactions, vol. 76, pp. 451461, August 1957.
[4]G. G. Richards and 0. T. Tan, Simplified Models for Induction Machine Transients under Balanced and Unbalanced Conditions, IEEE Transactions on Industry Applications, vol. IA-17, no. 1, pp. 15-21, January/February
1981.
[5] E. Drennan, S. Ahmed-Zaid and P. W. Sauer, Invariant Manifolds and Start-up Dynamics of Induction Machines, Proceedings of the 2 l s t Annual North-American
Power Symposium, University of Missouri-Rolla, Rolla,
Missouri, Oct. 9-10, 1989.
Maamar Taleb was born in Chlef, Algeria in 1958. He received his B.E. degree in Electrical Engineering from Universit4 des Sciences et de la Technologie dOran , Algeria in 1983,
his M.S. degree in Electric Power Engineering from Rensselaer Polytechnic Institute in 1986, and his Ph.D. degree in
Electrical Engineering from Clarkson University in 1990.
He is currently a Research Associate in the department of
Electrical and Computer Engineering at Clarkson University.
His research interests include power system modeling and harmonic analysis of distribution and HVDC systems.