You are on page 1of 5

1

Synchronous Machine Stability Model, an Update


to IEEE Std 1110-2002 Data Translation Technique
A, Moeini, I. Kamwa Fellow, IEEE, P. Brunelle, G. Sybille

technique transforms one model to another. IEEE Std 1110-


Abstract—IEEE Std 1110-2002 comprises a variety of 2002 covers data translation practices, models, and their
synchronous machine models and standard methods determining applications in power system stability analysis [8]. The
model parameters from standard tests. It also covers the so-called recognized mathematical models are based on a number of
data translation method which links (1)Time-constant-based,
operational-inductance and (2)Equivalent-circuit-based models.
resistances, inductances, and time constants, either extracted
In this paper, data translation technique for direct axis of well- from e.g. short circuit or open circuit tests. Among stablished
known second-order synchronous machine dynamic model is models, the second-order model has been coded in many
reviewed and re-formulated. A new approach based on field educational and industrial stability analysis software. IEEE
resistance measurement is proposed to calculate precisely model Std 115-2009 also comprises test procedures, acceptance, and
parameters in the presence of so-called Canay inductance. parameter determination for dynamic analysis of synchronous
Index Terms--Canay Inductance, Damper winding, IEEE Std machines [9].
1110-2002, Power system stability, Second-Order dynamic model,
Conventional theory of the Synchronous Machine modeling
Synchronous Machine, MATLAB, EMTP, PSS/E
for stability analysis is based on the assumption that the
I. NOMENCLATURE
mutual inductances between armature, dampers, and field on
d &q Subscripts for D-axis & Q-axis parameters direct-axis windings are identical [10-11]. Generally, damper
ψ d ( s),ψ q ( s), e fd ( s) Stator flux linkages and field voltage windings are near air-gap and as a result the flux linking
id ( s), iq ( s) Stator currents at operating point damper circuits is almost equal to flux linking armature.
Without losing generality, this hypothesis leads to acceptable
Ld ( s), Lq ( s) Operational inductances outcomes for wide range of stability studies especially those
G ( s) Armature flux to field transfer function of network-side. However, when it comes to field current
Rd , Rq , Lad , Laq DC resistances and magnetizing inductances studies, there is a considerable error. In some case, field
current measurements in transient and sub-transient modes
Ll Stator leakage inductance have shown higher alternating amplitude [12]. Traditional
L1d , R1d Damper winding inductance and resistance two-axis model describes only the armature circuit correctly.
An additional inductance needs to be introduced representing
L fd , R fd Field winding inductance and resistance
the difference between field-damper and field-armature
L f 1d Canay inductance mutual inductances on D-axis [12-14] and [15]. This
Ld , L'd , L''d Steady, transient, and sub-transient inductances inductance, mostly called Canay inductance, has been shown
' ''
in Fig. 1 and labeled as Xc. In fact, Canay inductance
T , T Open-circuit transient and sub-transient time constants
do do corresponds to the leakage flux ΦC and is interpreted as a
Td' , Td'' Short-circuit transient and sub-transient time constants corrective element in equivalent model which may acquire
T1d Damper winding time constant even negative value.
As a common practice, when Canay inductance is omitted,
II. INTRODUCTION
field winding base current and armature leakage inductance

S ynchronous machine modeling is the start point of any


power system stability studies. Generally, realized models
from commissioned test data are given in the forms of
are considered free parameters when calculating equivalent-
circuit models. Based on this assumption, remaining
parameters are computed such that the model reflects same
operational inductances or equivalent circuits. This procedure behavior from terminal point of view. In this case, the leakage
has been defined in literature based on two following inductance would not be equal to the value supplied by the
methodologies: manufacturer. In some cases, a negative value may be
1. Fitting operational inductances, model with dynamic attributed to armature leakage inductance. Moreover, few
reactances and time constants, to measurements [1-4] approximations are also considered in the process of data
2. Fitting equivalent circuits, model with constant translation when computing specific parameters, such as
impedances parameters to measurements[5-7] damper winding time constant on D-axis Further
Regardless the fitted model, the so-called data translation approximation is again required calculating open-circuit time
constants from short-circuit time constants or vice versa [8].
This work was supported by Hydro-Québec/IREQ. In this paper, taking into account the so-called Canay
A. Moeini, I. Kamwa, P. Brunelle, and G. Sybille are with Hydro- inductance, the IEEE Std 1110-2002 data translation method is
Québec/IREQ, Power System and Mathematics, Varennes QC, Canada J3X
1S1. (e-mails: moeini.ali@ireq.ca, kamwa.innocent@ireq.ca, reformulated to retain the value of armature leakage
brunelle.patrice@ireq.ca, sybille.gilbert@ireq.ca). inductance and avoid any approximation in process of

978-1-5386-7703-2/18/$31.00 ©2018 IEEE


2

calculating fundamental parameters from standard parameters specified rotor circuit numbers. For example, the well-known
and vice versa for widely-used second-order model. The second-order dynamic model has order of two for both
proposed mathematical method computes damper winding numerator and denominator of Ld(s) and Lq(s) as:
time constants and open-circuit/short circuit time constants
from commonly given data and offers a unique D-axis Ld ( s) = Ld
(1 + T s )(1 + T s )
d
'
d
''

(2)
equivalent circuit. Analytical and simulation studies have (1 + T s )(1 + T s )
'
do
''
do

proven the precision and superiority of proposed model over


previously developed models. Lq ( s) = Lq
(1 + T s )(1 + T s )
q
'
q
''

(3)
(1 + T s )(1 + T s )
'
qo
''
qo

Фl IV. TEST PROCEDURES AND DATA TRANSLATION


Armature
Among commonly commissioned tests, the StandStill
air gap Frequency Response (SSFR) is reviewed here [9].

Фmd A. StandStill Frequency Response (SSFR) Test


A very well stablished practical method to measure stability
Damper Ф1d
parameters is SSFR test. IEEE Std 115-2009 comprises an
extensive clause on this topic. A number of tests are
commissioned and operational quantities are measured as:
Фc
Direct-axis operational impedance, Zd(s), defined by (4) as:
Field Δe ( s)
Фf Z d ( s) = Rd + sLd ( s) = − d (4)
Δid ( s) Δe = 0
fd

Fig. 1. Mutual and leakage fluxes on d-axis Suppose Ld(s) is available from fitting techniques. When
expanded into partial fractions, it becomes:
III. SYNCHRONOUS MACHINE DYNAMIC MODEL 1 1 ⎛ 1 1 ⎞ sTd' ⎛ 1 1 ⎞ sTd''
= +⎜ ' − ⎟ + ⎜ − ⎟ (5)
IEEE Std 115-2009 comprises a variety of practical accepted Ld ( s ) Ld ⎝ Ld Ld ⎠ 1 + sTd' ⎝ L''d L'd ⎠ 1 + sTd''
tests for determination of synchronous machine dynamic Where for second-order dynamic model
model parameters. Each test has its own advantage and
⎡ ⎛ Tdo' ⎞⎛ Tdo'' ⎞ ⎤
disadvantages while interpretation of results requires ⎜ 1− ' ⎟⎜ 1 −
1 1 ⎢
1 ⎝ Td ⎠⎝ Td' ⎟⎠ ⎥
additional knowledge about synchronous machine’s complex = − ⎢ ⎥ (6)
nature. However, machine model data should finally be L'd Ld Ld ⎢ ⎛ Td'' ⎞ ⎥
⎢ ⎜1 − T ' ⎟ ⎥
defined in following formats in accordance with stability study ⎣ ⎝ d ⎠ ⎦
programs: Equivalent Circuit and Operational Inductances. ⎡ ⎛ Tdo' ⎞ ⎛ Tdo'' ⎞ ⎤
⎜ 1− '' ⎟ ⎜ 1 −
A. Equivalent Circuit: 1 1 ⎢
1 ⎝ Td ⎠ ⎝ Td'' ⎟⎠ ⎥
= − ⎢ ⎥ (7)
Under this classification the objective is to find an equivalent L''d L'd Ld ⎢ ⎛ Td' ⎞ ⎥
electrical circuit including a number of resistances and ⎢ ⎜1 − T '' ⎟ ⎥
⎣ ⎝ d ⎠ ⎦
inductances, each represents either physical or mutual linking
between the elements. Several equivalent circuits are proposed B. Data translation technique
during last decades [1][16]. Of those, the structure with one IEEE Std 1110-2002 defines data translation as the
damper on D-axis and two dampers on Q-axis, the so-called approach which (1) extract equivalent circuit model from
second-order model, have been shown to be generally less operational inductance model (2) vice versa. In other words,
complex but adequate for stability studies. Fig. 2 shows starting from one given machine model, data translation
equivalent circuit of both direct and Quadratic axes of second- calculates other corresponding model. Note that current work
order model. The Lf1d represents Canay inductance. focuses solely on D-axis. The procedure for Q-axis equivalent
circuit is similar to Annex E of IEEE Std 1110-2002.
B. Operational Inductances:
1) From operational Inductances to equivalent circuits:
The operational parameter has been approved as an effective
Suppose Ld , L'd , L''d , Td' , Td'' are at least available:
approach for identifying synchronous machine dynamics
[17][9]. In this approach a transfer function is developed for This set of data is usually obtained from standard sudden
direct and quadrature axes according to small perturbations at three-phase short-circuit test. To avoid data inconsistencies,
machine terminals as follows: open-circuit time constants are calculated from (6) and (7) as:
Δψ d ( s ) = G ( s )Δe fd ( s ) − Ld ( s )id ( s ) Td' Ld L''d + Λ1 + Td'' ( Ld L'd + L'd L''d − Ld L''d )
(1) Td'o = (8)
Δψ q ( s ) = − Lq ( s )iq ( s ) 2 L'd L'd'
As a matter of fact, (1) satisfies the condition for any number Td' Ld L''d − Λ1 + Td'' ( Ld L'd + L'd L''d − Ld L''d )
of rotor circuits. The order of developed polynomial Td''o = (9)
2 L'd L'd'
operational inductances on both D-axis and Q-axis are equal to
3

Λ1 = Ld 2 L''d 2 (Td' − Td'' ) 2 + 2Td' Td'' Ld L'd L''d ( Ld − 2 L'd + L''d ) − 2Td'' Ld L'd L''d ( Ld − L'd + L''d ) + Td'' 2 L''d 2 ( Ld 2 + L''d 2 ) (10)
id Ll Lf1d Rfd Lfd
ifd
IEEE Std 1110-2002, it is computed approximately as:
b
i1d T1d = (25)
c
L1d
Ψd Lad efd with
Lad ( L"d − Ll )
R1d b=a
aLd + L l Lad − ( a + Lad ) L"d
(a) (26)
iq Ll
1 ⎛
c = " ⎜b +
aLl Lad ⎞ ( L'd − Ll )
⎟ a= L
i1q i2q Td ⎝ L l Lad + aLd ⎠ ( Ld − L'd ) ad
Ψq Laq L1q L2q In this work, a new method is proposed for exact
determination of the T1d. The idea was inspired from the fact
R1q R2q that the Rfd is commonly available from commissioned tests or
(b) provided by manufacturer. Hence, the following alternative
Fig. 2. Synchronous machine Equivalent circuit (a) D-axis, (b) Q-axis expression of (19) is given:
It should be noted that IEEE Std 1110-2002 recommends L K
K 2 = R fd d 21 (27)
measured Tdo' should be used when available instead of (8). Lad
Consequently, the Tdo'' is calculated as follows: Supposing that Tdo' , Tdo'' , Td' , Td'' are available, the K1 is
calculated from (20). By substitution of K1 in (27) the K2 is
Ld Td' Td''
Tdo'' = (11) directly computed. Consequently, the T1d is the exact root of
L'd' Tdo' following univariate quadratic function:
By any means, if open- circuit data set is given, AT1d 2 + BT1d + C = 0 (28)
Ld , L'd , L''d , Tdo' , Tdo'' , short-circuit time constants are computed where
from (6) and (7) as: A = ( Tdo' + Tdo'' − Td' − Td'' ) B = −2 (Tdo' Tdo'' − Td' Td'' )
L' (T ' + T '' ) − Λ 2 (29)
Td' = d do ' do (12) C = Tdo' Tdo'' (Td' + Td'' ) − Td' Td'' (Tdo' + Tdo'' ) − K 2
2 Ld
T1d is supposed to be the smallest non-negative root of (28).
L'd L''d (Tdo' + Tdo'' ) − L''d Λ 2 Regarding T1d, it should be mentioned that D-axis sub-
Td'' = (13)
2( Ld L'd + L'd L''d − Ld L''d ) transient time constant must be very small since related
current damps right after few cycles when disturbance stresses
Λ 2 = L'd 2 (Tdo' + Tdo'' )2 − 4Tdo' Td''o ( Ld L'd + L'd L''d − Ld L''d ) (14)
synchronous machine. A sanity check is also recommended by
At this point operational parameter data set, transient and sub- calculation of originally-given data from resulted parameters.
transient inductances and time constants are given. According Having T1d, K3, K4, and K5 are computed from (22)-(24)
to [16] equivalent circuit elements are calculated as follows: respectively. Finally, the rest of equivalent circuit parameters
L2 K K as well as Lf1d can be simply calculated from (15)-(19).
L f 1d = ad 3 − Lad , L fd = R fd 5 (15-16) 2) From Equivalent circuit to operational inductance
Ld K 4 K4
Similar to SSFR concept, the operational parameters can be
K1 K1 Lad 2 K 2 expressed from given equivalent circuit elements. According
L1d = T1d R fd − , R1d = R fd , R fd = (17-19)
K 42 K 42 Ld K1 to Fig. 2, the D-axis flux linkages are given by:
where: ⎡ψ d ⎤ ⎡ L d M d 1d M fd ⎤ ⎡ −id ⎤
⎢ ⎥ ⎢ ⎥⎢ ⎥
K1 = (Tdo' + Tdo'' − Td' − Td'' ) ⎡⎣Tdo' Tdo'' (Td' + Td'' ) − Td' Td'' (Tdo' + Tdo'' )⎤⎦ ⎢ψ 1d ⎥ = ⎢ M d 1d L 11d M f 1d ⎥ ⎢ i1d ⎥ (30)
(20) ⎢ψ fd ⎥ ⎢ M fd M f 1d L ffd ⎥ ⎢ i fd ⎥
− (Tdo' Tdo'' − Td' Td'' )
2
⎣ ⎦ ⎣ ⎦⎣ ⎦
where
K 2 = Tdo' Tdo'' (Td' + Td'' ) − Td' Td'' (Tdo' + Tdo'' ) Ld = Ll + Lad , L11d = L1d + L f 1d + Lad , M d 1d = Lad
(21) (31)
+ T1d 2 (Tdo' + Tdo'' − Td' − Td'' ) −2T1d (Tdo' Tdo'' − Td' Td'' ) L ffd = L fd + L f 1d + Lad , M fd = Lad , M f 1d = L f 1d + Lad
K 3 = Tdo' Tdo'' − T1d (Tdo' + Tdo'' − T1d ) (22) The per unit rotor circuit voltage equations are also defined as:
dψ fd dψ 1d
K 4 = Tdo' Tdo'' − Td' Td'' − T1d (Tdo' + Tdo'' − Td' − Td'' ) (23) e fd = + R fd i fd , 0 = + R1d i1d (32-33)
dt dt
K 5 = Tdo' Tdo'' ( Td' + Td'' ) − Td' Td'' (Tdo' + Tdo'' ) − T1d ( Tdo' Tdo'' − Td' Td'' ) (24) Equations (32) and (33) are substituted in (30) and all
The only remaining parameter is the T1d, the time constant of variables are expressed by their incremental form while
D-axis damping winding. In [15], a method has been proposed corresponding initial values are eliminated. The procedure
to capture T1d from field current oscillation. According to leads to accurate time constants which yield in equivalent
4

circuit elements. For the sake of comparison, the operational EMTP show again same response (Figs. 5 and 6).
parameters are expressed similar to [18]:
Tdo' + Tdo'' = T1d + T2 d , Tdo' Tdo'' = T1d T3 d (34)
Td' + Td'' = T4 d + T5 d , Td' Td'' = T4 d T6 d (35)
L1d
T1d = (36)
R1d

Pt (p.u)
where
1 1
T1d =
R fd
( L fd + Lad + L f 1d ) , T2 d =
R1d
( L1d + Lad + L f 1d )
1 ⎛ L fd ( Lad + L f 1d ) ⎞
T3d = ⎜⎜ L1d + ⎟
R1d ⎝ ( L fd + Lad + L f 1d ) ⎟⎠
1 ⎛ Lad Ll ⎞ Fig. 3. During and after contingency active power variations
T4 d = ⎜ L fd + L f 1d + ⎟ (37)
R fd ⎝ Lad + Ll ⎠
1 ⎛ Lad Ll ⎞
T5d = ⎜ L1d + L f 1d + ⎟
R1d ⎝ Lad + Ll ⎠

1 ⎛ L fd ( Lad Ll + L f 1d ( Lad + Ll ) ) ⎞

Vt (p.u)
T5d = ⎜ L1d + L f 1d + ⎟
R1d ⎜ ( L fd + L f 1d )( Lad + Ll ) + Lad Ll ⎟
⎝ ⎠
Similarly, the transient and sub-transient inductances can
be calculated from (6) and (7).
By recognizing the similarity of D-axis and Q-axis models,
when Lf1d is omitted, or the so-called dynamic model with
equal mutual inductances, the Q-axis time constants are Fig. 4. During and after contingency terminal voltage
similar to equations given in [18] for D-axis.

V. SIMULATION RESULTS
The proposed method has been implemented on Simscape
Speed Deviation (p.u)

Power Systems synchronous machine model using both


fundamental and standard parameters. Four simple
benchmarks are then developed in different environments:
(1)MATLAB/SPS (2)Electro Magnetic Transient Program
(EMTP-RV) (3)PSSE with GENROU machine model
(4)PSSE with CGEN1 machine model. The studied systems
are identical and include one synchronous machine connected
directly to a load. In order to examine the performance of data
translation technique behind studied software, each machine is Fig. 5. Synchronous machine speed deviations over 10 seconds
initialized by operational parameter taken from [1]. Three
phase fault is applied to each machine’s terminal at t=1/60 sec
with duration of 4 cycles. Figs. 3-4 show machine active
Speed Deviation (p.u)

output powers and terminal voltages. During the fault, PSSE


modeled machines have no oscillations and remain zero. The
EMTP and SPS modeled machines have identical oscillations
which prove the fact that damping torque generated by
damping windings are considered. The post contingency
recovery of PSSE synchronous machine type CGEN1 shows a
slight variation. However, the GENROU model has
oscillation. Apart from initial differences, all models reach
same steady-state values after 2 seconds.
Fig. 6. During and after contingency speed deviations
Similarly, same responses have been recorded for speed
deviation: PSSE GENROU model shows no oscillation and Implementation of Canay inductance is mostly understood
reduces exponentially after fault clearance; PSSE CGEN1 to impact rotor-side variables such as field current. Although
seems to have logic but weakly damped response; SPS and EMTP and SPS machine models have slightly different post
5

contingency responses, SPS machine model has more VIII. REFERENCES


oscillation than EMTP machine as shown in Fig. 7. The reason [1] I. M. Canay, “Determination of the Model Parameters of Machines from
behind this dissimilarity is the simplifications made in the the Reactance Operators xd(p), xq(p), (Evaluation of Standstill Frequency
process of current EMTP machine modeling. Response Test),” IEEE Transactions, EC-8(2), pp. 272–279, June 1993.
[2] EPRI Report, EL-1424, Determination of Synchronous Machine
Finally, it should be noted that Canay inductance will be Stability Study Constants, vol. 2, Ontario Hydro, 1980.
implemented in next MATLAB/SIMULINK version. As of [3] G. Shackshaft, and P. B. Henser, “Model of Generator Saturation for
today, an agreement has been made with EMTP developer Use in Power System Studies,” Proceedings IEE, vol. 126, pp. 759–763,
1979.
team to upgrade also current synchronous machine model. [4] J. Verbeeck, R. Pintelon, and P. Guillaume, “Determination of
Synchronous Machine Parameters Using Network Synthesis
Techniques,” IEEE Transactions, EC-14(3), pp. 310–314, September
1999.
[5] I. Kamwa, M. Pilote, P. Viarouge, B. Mpanda-Mabwe, M. Crappe, and
R. Mahfoudi, “Experience with Computer-Aided Graphical Analysis of
Sudden-Short-Circuit Oscillograms of Large Synchronous Machines,”
IEEE Transactions, EC-10(3), pp. 407–414, September 1995.
If (p.u)

[6] I. Kamwa, P. Viarouge, H. Le Huy, and J. Dickinson, “The Three-


Transfer-Function Approach for Building Phenomenological Models of
Synchronous Machines,” IEE Proc. Gener. Transm. Distrib., 141(2), pp.
89–98, 1994.
[7] H. Tsai, A. Keyhani, J. Demcko, and R. G. Farmer, “On-Line
Synchronous Machine Parameter Estimation from Small Disturbance
Operating Data,” IEEE Transactions, EC-10(1), pp. 25–36, March 1995.
[8] IEEE Guide for Synchronous Generator Modeling Practices and
Applications in Power System Stability Analyses, IEEE Std 1110-2002
Fig. 7. Field current oscillation (Revision of IEEE Std 1110-1991) , pp.0_1-72, 2003
[9] IEEE Guide for Test Procedures for Synchronous Machines Part I:
VI. CONCLUTION Acceptance and Performance Testing, Part II: Test Procedures and
Parameter Determination for Dynamic Analysis, IEEE Std 115-2009
In this paper the data translation method of IEEE Std 1110- (Revision of IEEE Std 115-1995), pp.1-219, May 7 2010.
2002 has been reviewed. A number of simplifications are [10] B. Adkins, "The General Theory of Electrical Machines." London,
Chapman and Hall, 1957.
considered in the process of computing time-constant-based [11] K. Bonfert, "Betriebsverhalten der Synchronmaschine." Berlin,
model operational-inductance from equivalent-circuit-based Springer, 1962.
model and vise-versa. While taking on to account the so-called [12] I. M. Canay, "Causes of Discrepancies on Calculation of Rotor
Quantities and Exact Equivalent Diagrams of the Synchronous
Canay inductance, data translation method has been re- Machine," in IEEE Transactions on Power Apparatus and Systems,
formulated recognizing the fact that field resistance is PAS-88, no. 7, pp. 1114-1120, July 1969.
commonly accessible. This parameter is provided by [13] I.M. Canay, "Investigation of the rotor overvoltage in synchronous
machines with rectifier excitation, " Electric Machines & Power
manufacture and is measurable by SSFR test. Having the
Systems, 3:1, 21-38, (1978)
value field resistance, d-axis damper time constant is [14] I.M. Canay, "Extended synchronous-machine model for the calculation
computed without any simplification. The new formulated of transient processes and stability, " Electric Machines & Power
data translation method has been coded in Systems, vol. 1, no 2, PP. 137-150, 1977
[15] Y. Takeda and B. Adkins, "Determination of synchronous machine
MATLAB/SimPowerSystem. Identical benchmarks, including parameters allowing for unequal mutual ·inductances", Electrical
one synchronous machine connected directly to a load, has Engineers, Proceedings of the Institution of, vol. 121, no. 12, pp. 1501-
been developed on three well-known software, 1504, December 1974.
[16] L. Salvatore & M. Savino, "Exact relationships between parameters and
(1)MATLAB/SPS, (2)EMTP-RV, and (3)PSSE. The results test data for models of synchronous machines," Electric Machines &
show how accurate implementation of Canay inductance and Power Systems, vol. 8, no. 3, pp. 169-184, 1983.
damper parameters may impact machine variable, specially [17] IEEE Committee Report (P. L. Dandeno, Chair) "Supplementary
Definitions and Associated Test Methods for Obtaining Parameters for
filed current oscillations. Synchronous Machine Stability Study Simulations." IEEE Transactions
on Power Apparatus and Systems, vol. PAS-99, no. 4, pp. 1625–1633,
VII. APPENDIX July–Aug. 1980. 13
[18] Kundur, P., Power System Stability and Control, New York: McGraw
Hill, 1994.
General Parameters of studied benchmark are from [18],
Examples 3.1, 3.2, 8.1 (pgs. 91,102, 345)
Pn= 722.2e6; % three-phase nominal power of machine (VA)
Vn= 26e3; % nominal line to line voltage (Vrms)
fn= 60; % nominal frequency (Hz)
Ifbase= 3060; % nominal field current (A)
H= 3.7; % sec
Load= 555e6 MW
The synchronous Machine dynamic data are from [1].

You might also like