LIST OF PRINCIPAL SYMBOLS ^d> k m d, L mdr etc.as defined here in their application to
the d axis.
Variables and matrices
v = vector of axis voltages
i = vector of axis currents 1 INTRODUCTION
Te = electromagnetic torque One of the principal considerations in supplying the excita-
u)T = rotor speed tion requirements of synchronous generators from a.c. ex-
ws = departure from nominal speed citers operating in conjunction with rectifier systems is
6 = rotor angle that of the peak-inverse voltage that rectifiers are required
L = inductance matrix to withstand in service. Of particular concern are operating
G = torque matrix conditions in which, if the supply were of a bipolarity form,
R = diagonal matrix of winding resistances the field current would reverse; for this is precluded by
Inductance coefficients the essentially unidirectional form of supply that an a.c.
exciter and a rectifier system provide. If, for example,
Ld = stator total inductance in d axis generator operating conditions cause a progressively de-
L
md = magnetising inductance in d axis creasing field current, and a point is reached at which, with a
L
mdr = mutual inductancen between d axis stator or field d.c. exciter, the field current would reverse, then, in the
winding and r* equivalent d axis eddy-current case of an a.c. exciter and a rectifier system, this reversal
circuit in field current would be blocked. A momentary disturbance
tn
L
kdr = total inductance of r d axis equivalent eddy- to the m.m.f.balance between rotor and stator arising in
current circuit this way causes a rapid rise in rotor flux linkages and a
tn tn
L
idr = mutual inductance between the i and r equiva- transient pulse of voltage in the field system. It is this
lent eddy-current circuits in d axis induced voltage, when the rectifiers are in a nonconducting
mode, that makes particular demands on the peak-inverse-
Resistance coefficients voltage rating of rectifiers.
= stator-winding resistance Site tests 1 have been carried out to assess overvoltage
a = field-winding resistance magnitudes that field-current blocking by rectifiers might
Rfd = resistance of r t n d axis equivalent eddy-current in practice cause, and computer studies of the site-test con-
Rdr
circuit ditions have been made. In predicting the greatest over-
voltages likely to arise, it appears generally to be the view
Field solution from Green's function that the most onerous conditions are those of generator
c = semipole pitch pole slipping. There is no doubt that, should pole slipping
g = airgap length, m occur, conditions can arise that lead to severe field-system
Hr = relative permeability of rotor steel overvoltages, and this leads to the question of whether there
o = conductivity of rotor steel fi-m may be other, perhaps more generally encountered, modes
of operation, that can in some measure reproduce the con-
Subscripts s and r identify stator and rotor circuits, res- ditions of pole slipping in so far as they influence field-
pectively. The term 'total inductance' is used to denote the system overvoltages. Depending on the precise circum-
sum of the magnetising and leakage inductances. Q-axis stances, this can be the case during synchronising, in that
inductance coefficients Lq, L m q, L m q r etc. correspond to the incoming unit may momentarily operate in either motor-
ing or generating modes. Accepting this, a series of studies
has recently been carried out specifically to assess the
extent to which there may be the hazard of field-system
Paper 7173 P, first received 1st November 1973 and in overvoltages when synchronising from different conditions.
revised form 18th March 1974 The studies confirm that, when synchronising within the
Dr.Humpage is with the Power Systems Laboratory, Univer- limits typical of British practice, no adverse effects are
sity of Manchester Institute of Science & Technology ,PO likely to arise for a wide range of units of differing ratings
Box 88, Sackville Street, Manchester M60 1QD, England. and construction. At the same time, the constantly repeat-
Dr. Smith and Mr. Rogers are with the Department of Elec- ing nature of generator synchronisation suggests extending
trical Engineering, University of Southampton, Southampton, the study to locate, in representative cases, the combinations
Hants. SO9 5NH, England of circumstances during synchronising, where field-system
PROC. IEE, Vol. 121, No. 7, JULY 1974 647
overvoltages could be generated. The paper seeks to report Approaches to this have previously been given,2 together
investigatory work in this area, together with the computer with the curve-fitting techniques3 for incorporating para-
simulation methods on which it has been based. meter updating into computer solution.
Beyond this, the mechanical axes of the turbogenerator must
of course be included, together with the interaction between
2 SYNCHRONISING
the electrical and mechanical axes; and the account of the
There can be few more basic aspects of system operation model on which the studies of the present paper have been
than that of synchronising an incoming generator unit to the based is confined principally to the electrical and mechani-
system with which it is to operate in parallel. Automatic cal axes. The study application is one that makes no special
methods of checking that conditions are favourable for syn- demands on prime-mover, governor and excitation-con-
chronisation have found increasing application, but, irrespec- troller representations, and those of the kind previously
tive of the precise mode of working, the essential require- reported should meet the requirements of the present study.
ment is one of securing small, and preferably diminishing,
discrepancies between the phase, frequency and magnitude 3.2 Electrical axes
of the voltage phasors, of the incoming machine and of the
parent system. Any mismatch between the two sets of pha- Expressed in the 2-axis rotor frame of reference, the volt-
sors gives rise to a transient period of operation immedia- age equations in the electrical axes may be convieniently
tely following the instant at which electrical connection used in the vector-matrix form:
between the machine and the system is made, and bounds
are imposed on these discrepancies to limit the transients v = Lpi + Gcori + Ri (1)
that might in this way be generated. Typical limits in prac-
tice for phase, frequency and voltage errors are given in Partitioning eqn.l to separate the stator and rotor variables
Table 1. then gives:
Mismatch Limit Re
(2)
Phase displacement ±5°
Frequency ±0-75% In this form
Voltage ±6% (3)
and
No tendency towards the generation of field-voltage tran-
sients has been found within this spectrum of synchronising ikqi> (4)
conditions: it remains to locate the limits for which over-
voltages can arise. In one aspect, this will allow the mar- where the rotor eddy-current paths are represented by n
gins associated with particular limits used to be quantified equivalent circuits in each of the d and q axes. Each equi-
with respect to conditions where field-circuit overvoltages valent circuit is short-circuited, and so the only nonzero
may constitute an unacceptable hazard. Equally, it will allow entry in the vector of rotor voltages, v r , is the first entry
potential overvoltage conditions to be evaluated where, for corresponding to the field-winding voltage. The vector of
one reason or another, synchronising limits less exacting stator voltages, v s , is given by:
than those of Table 1 are considered. In this latter case, the
possibility of field-circuit overvoltages may be a principal v s = v d ,v q (5)
consideration in choosing safe limits. Computer simulation
methods provide flexible and reliable means of undertaking 4
In recent work, rotor eddy-current-path representations
definitive studies in this area. have been sought from applying Green's function to a simpli-
fied electromagnetic model for the solid rotor of a uniform
airgap machine. Correlations between site test and com-
COMPUTER SIMULATION puter results in the recent work4 support the general validity
of the approach, the outcome of which can perhaps be best
3.1 Particular requirements summarised in the explicit forms of the coefficient matrix
partitions of eqn. 2.
Turbogenerator models for use in dynamic-analysis studies
have been widely developed, 2 ' 3 ' 4 and are now in general For the stator-rotor partition of the generator inductance
application, but it is worth referring to the aspects of models matrix,
that become emphasised in their application to overvoltage
\ fd kdl kql kdn kqn
calculations. As field-system overvoltages arise from field-
current blocking, and as the field system cannot contribute L
to maintaining the rotor-stator balance when this occurs, Lmd mdi Lmdn
the representation of the remaining rotor circuits has a Lsr —
special importance. Depending on the rotor construction, the
rotor circuits other than the field winding are those of the L ^mqn
mqi
amortisseurs, where separately provided, or, in the case of (6)
uniform airgap machines, the eddy-current paths in the The mutual coefficients in eqn. 6 are given by:
solidly forged rotor body. The present work has been con-
fined to solid-rotor machines, but, whatever the rotor con-
L
struction, the calculation of rotor overvoltages is a particu- mdr = L md (7)
lar study application for which a careful representation of
the rotor paths other than that of the field winding is re- and
quired.
As rotor overvoltages arise from a disturbance to the nor- — L mq (8)
mal m.m.f. balance between rotor and stator, the stator
operating level at the time of the disturbance strongly in- for
fluences the magnitude of subsequent field-system over- (9)
voltages. This focuses attention on the mutual coefficients r=l,2.
that link the rotor and stator equation sets, and by which
stator operating levels are reflected into the rotor circuits. and where
A judicious choice of parameter values might be attempted (10)
for different operating conditions, but perhaps the most Cr= -
satisfactory arrangement here is one of continuously re-
calculating coefficient values as the solution proceeds.
648 PROC. IEE, Vol. 121 ,No.7, JULY 1974
and and
{(2r- k)
J (25)
mq
T e = itGwpi (27)
kdn Lmdn L
ndi L
kdn
i The instantaneous departure from fundamental speed w
kqn L
nqi L
kqn and the rotor angle 6 are then available from:
(13)
= ^ [Ti - T e ] (28)
For the diagonal elements of the rotor submatrix L r r other M
than that corresponding to the field circuit,
and
L
kdr = md L (14) (29)
(1 + k) 2 (15)
k
O
for
r=l,2, (16) b
For the offdiagonal elements of L r r other than those relat-
ing to the field circuit, if 200
o >
a
L
idr — L md (17) i ioo
k
1+k (18)
Liqr — 1-25
for >< 1°
r=l,2, (19)
&1 ° 7 5
and 9 3 o-5
i = 1, 2, (20) ^ £ 0 25
The stator partition of the inductance matrix is the orthodox
submatrix of d-q models:
S1 300
qs _^
200
ds o
(21) 100
b
qs
1 0
6
The G matrix is formed from the inductance coefficients, and
the explicit form of this matrix is given in Appendix 10.1. 4
Finally, in summarising the coefficient matrices in the
electrical axes, 2
0
R s = diag (Ra, Ra) (22)
The equations in the mechanical axes then complete the • instant of switching
formulation:
pw s = — (31)
M
and
p6 = (32)
— instant of synchronising
40 -
0P 10
-field current } ||-field voltage
•I 'I
0 0-1 0-2
time, s
Fig. 4
Field-system overvoltages: Goldington machine
Mismatch: parent system 12° in advance of incoming genera-
tor unit; system-voltage magnitude 12°/0 greater Hum genera-
tor voltage
7 CONCLUSIONS
The studies that the paper seeks to report confirm that,
synchronising within close limits, such as those of Table 1,
avoids altogether the possibility of unexpected induced-
voltage pulses in the field system of generator units excited
from a.c. exciters through rectifier systems. They also
indicate the areas of hazard where, for one reason or another,
extended synchronising, or resynchronising, limits might
arise. More generally, the computer simulation methods on
which the paper has drawn allow the margins of safe syn-
chronising with respect to field-system overvoltages to be
quantified. They provide means of carrying out a compre-
hensive search for those combinations of conditions for
which field-system overvoltages might arise.
O 01 0-2
time, s Although study results have been reported here for turbo-
generator units of significantly differing sets of parameters,
Fig. 3 there may now be scope for seeking to correlate overvoltage
Field-system overvoltages: Belvedere machine magnitudes with machine ratings and constructions. Care is
Mismatch: parent system 12° in advance of incoming genera- needed in this owing to the essentially discrete groupings
tor unit; system-voltage magnitude 12% greater than genera- of parameters for individual machines, but useful patterns
tor voltage might well emerge from a series of wide-ranging studies.
PROC. IEE, Vol. 121,No. 7, JULY 1974 651
Apart from this, any particular case, or set of conditions, and
may be very directly evaluated using the analysis methods
given in the paper. fd kdl kql kdn kqn
ds ~ L mqi
8 ACKNOWLEDGMENTS
Gsr =
qs L
_ L m d |L m di mdn
The authors are grateful to Prof. L.M.Wedepohl, Head of
the Department of Electrical Engineering & Electronics, (41)
University of Manchester Institute of Science & Technology,
for his support and encouragement in this work and for the 10.2 Principal data of generator unit in
facilities in the UMIST Power System Laboratories. J.R. site tests at Belvedere power station
Smith and G.J.Rogers are similarly grateful to Prof.P.
Hammond, Head of the Department of Electrical Engineering Data item Value
at the University of Southampton.
Rating 120 MW, 13-8 kV
09 pF (lag), 3000 rev/min
REFERENCES
Ld 3-97 x 10"3
HUMPHRIES, H. J., and FAIRNEY,W.: 'Excitation recti-
fier schemes for large generators', Proc.IEE., 1972,
L
q 3-62 x 10-3
L
md 3-58 x 10-3
119, (6), pp. 661-671
SHACKSHAFT, G.: 'General-purpose turboalternator Lmq 3-24 x 10-3
Lfd 3-97 x 10"3
model', ibid., 1963,110, (4), pp. 703-713
HUMPAGE, W.D.,and SAHA, T.N.: 'Digital-computer Rfd 0-000862
Ra 0-002
methods in dynamic-response analyses of turbogenera- M
tor units', ibid., 1967,114, (8), pp. 1115-1130 0-0246
ROGERS, G. J., and SMITH, J. R.: 'Synchronous-machine w0 314 rad/s
Mr 100
model including eddy currents', ibid., 1973,120, (4),
pp.461-468 a 28 x 10-8 nm
Ld 5-77 x 10-3
G= sr (39) 5-23 x 10-3
Lq
L
md 5-33 x 10-3
Lmq 4-79 x 10-3
The explicit form of the stator partition G s s and the stator- '"4
5-77 x 10-3
rotor partition G s r are then: Lfd 1-07 x 10-3
Rfd
Ra 2-10-3
M 0-0288
w0 314 rad/s
Mr 100
(40)
CT 28 x 10" 8 fim