1, March 1993
ANALYSIS A N D A V E R A G E V A L U E MODELING OF L I N E C O M M U T A T E D
C O N V E R T E R  SYNCHRONOUS MACHINE SYSTEMS
Abstract Analytical relationships are established which can converter system if the commutating reactance is set equal to
be used to predict the steadystate characteristics of line the subtransient reactance of the machine [Z]. However, the
commutated converter  synchronous machine systems. In subtransient reactances of synchronous machines are, in gen
particular, basic relationships are established in which the eral, different for the q and daxes and it is not clear as to
average dc voltage and the average electromagnetic torque are which of the two reactances, or whether some weighted aver
related to the converter firing delay angle. It is shown that age of the two reactances, should be used in the classical
these averagevalue relationships predict the steadystate per averagevalue equations. In [Z], empirical justification is given
formance with significantly higher accuracy than the classical for modeling the alternator by its daxis subtransient reac
converter averagevalue equations in which the daxis sub tance during commutation.
transient reactance is used as the commutating reactance. In this paper, averagevalue relationships are derived for
the general case where the subtransient reactances are
Key Words: linecommutated converter, synchronous different along the q and daxes. The resulting expression for
machine, electric drives, inverter  motor, the average de voltage is similar to the classical expression
alternator  rectifier. with the exception that the equivalent commutating reactance
is a function of both the q and daxis subtransient reactances
INTRODUCTION and of the converter firing angle. In addition, basic relation
ships are established in which the average dc voltage and the
Linecommutated acdc converters connected to synchro
average electromagnetic torque are related to the converter
nous machines are used in a wide variety of applications. For
firing delay angle. The results are applicable either for
example, alternator  rectifier systems are used in aircraft
alternatorrectifier or invertermotor operation. It is shown
power systems [l],in highpower dc supplies [2], and in excita
that these averagevalue relationships are significantly more
tion systems of large electric generators [3] and [4]. In these
accurate than the classical converter averagevalue equations
applications, the converter may consist of diodes connected as
in which the daxis subtransient reactance is used as the com
a 6pulse bridge rectifier (uncontrolled) or, if rapid control of
mutating reactance.
the output (dc) voltage is desired, the diodes may be replaced
by thyristor valves whereupon phase control can be used to MATHEMATICAL D E V E L O P M E N T
regulate the average output voltage. If a thyristor bridge is
used, the converter can also be controlled so that the direction In order to set forth basic definitions and to define
of power transfer is from the dc system to the synchronous nomenclature, it is useful to first examine the idealized source
machine whereupon the converter  machine operates as an  linecommutated converter system shown in Fig. 1.
inverter  motor. Thus, linecommutated converter  syn Therein, the source impedance is purely inductive. The
chronous machine systems can be used in variablespeed elec threephase source voltages are given by
tric drive systems [5]. eag = E cos 8, (1)
The fundamentalfrequency or averagevalue characteris
tics of linecommutated converters are generally established ebg = E cos(8,  2)
by assuming that the converter is connected to an idealized ac
system in which the opencircuit ac voltages form a balanced ecg = E cos(8, + 2)
threephase set and the source impedance is primarily induc
tive. Under these conditions, relationships between the aver where E is the peak amplitude and 0, denotes electrical posi .
age dc voltage and current and the fundamental components tion. The inductance between the source and converter is
of the ac voltages and currents are readily derived (61 and (71. referred to as the commutating inductance L,. The indivi
The series reactance of the ac source is commonly referred to dual bridge valves are numbered 16.
as the commutating reactance and is generally assumed to be Idc
constant when deriving the averagevalue relationships.
If the ac side of the converter is connected to a synchro
nous machine rather than to an idealized ac source, the con
ventional averagevalue relationships may be used to approxi
mate the salient characteristics of the synchronous machine 
+
0, sin($,  2") sin(@, $)
I
In (8), er represents the rotor position of the synchronous
(8)
In per unit, the electromagnetic torque can be expressed as [9]
T e = &si&  %siis (24)
With the selected reference direction for stator currents, posi
tive electromagnetic torque corresponds to motor action.
Equations (31) and (32) are similar in form to (10) and (11) if To evaluate (41), it is necessary to find As  '$cs at each
integration limit. In order to determine this, it is first neces
E = v(E;)' + (E:)' (35) sary to specify the stator currents at each limit of integration.
In particular,
and
q5 = atan(E: ,E;) (36)
where atan(y,x) is the arc tangent function which includEs
quadrant information. If r, can be neglected and X i = X ,
then (31) and (32) are identical in form to (10) and ( 1 4 . L J
Given the rotor flux linkages (thus E:, E:), the dc current
and the firing control parameters ( cy or p ), the average dc
voltage may be computed using (46).
It is interesting to examine the variation of commutating
reactance as a function p, as depicted in Fig. 3, for
X i = 0.328 pu and X i = 0.172 pu. At p = 0 or T , the com
mutating reactance is equal to the daxis subtransient reac
tance. When p = 7r/2, the commutating ;eacta?,ce is equal to
the qaxis subtransient reactance. For Xq > X d the commu
tating reactance takes ?? a minimum at p = 7r/6 where it is (53)
equal to 3Xi/2 X,/2, and a maximum value of
3X:/2  X i / 2 at p = 2 ~ / 3 . In the case that X i < X i , the
roles of maximum and minimum are interchanged. It is
important to note that the commutating reactance can be %+a
much greater than either the daxis subtransient reactance or ;r 3 3
the average of the q and daxis subtransient reactances, Ids,cond = iis(er) der (54)
T ,A+ @ + U
which are used to approximate the commutating reactance of
the synchronous machine.
In (49)(54), U is the commutation angle, the subscript "com"
designates the commutation interval in which valves 1, 2 and
3 are conducting, and "cond" designates the conduction inter
val which occurs after commutation from valves 1 to 3 has
taken place and only valves 2 and 3 are conducting. During
the conduction period, the stator currents are given by (43).
Transforming these currents into the rotor reference frame
and substituting into (53) and (54) yields, after simplification
0 1 I 1
0 7l)6 7112 243 7r During the commutation interval, the stator currents may be
expressed as
P, rad
Fig. 3 Variation in commutating reactance as a function of p.
(57)
Calculation of Stator Currents
It is useful to relate the average q and daxis stator
currents to the dc current. This can be accomplished using Transforming these currents into the rotor reference frame
(4) and equating the instantaneous real power on the ac and yields
dc sides of the converter as described in [7]. However, an
alternative approach is given here.
The q and daxis currents are periodic with a period of
6 4 . The average over the 60' switching interval beginning
when valve 3 is fired may be written
,+2rr
a
7 3 During the commutation interval, the a and b phases are
lqs = J i'ps(er1der (47) connected together. Neglecting stator resistance,
?r 4 + a
2rr
+a
I 3 3
ids = J iis(4) der (48)
where K is a constant. Now, '$= and As
can be expressed in
" A
+a terms of i,, Idc, $;, and '$i
by transforming (29) and (30)
The previous expressions can be broken down into two com into abc variables and using (57) to eliminate ibs and i,.
ponents corresponding to the commutation and conduction Substituting the resulting expressions into (61) and solving for
intervals ias yields
7 7
1qs = lqs,com + c s , c o n d
7 9 7
(49)
1,s =
Kfi[$i cos(^,
( X i  X i ) C0S(2Or
+ $1 + 4; sin(er + +)
+ +) + (Xh' + X i )
I
Ids = lds,com + lds,cond
where
96
[(xb'Xf)cos(261 + $) + +(Xh' + X'd)]IdC angle. Since Bmin represents the earliest instant a valve can
be turned on, the firing delay angle & may be defined as
 (62)
(Xi  xf ) C O S ( 2 4+ I)+ (Xi + X'd) & = P  Pmin (68)
where Here, & represents the delay in firing of a given valve meas
ured relative to the instant its anodetocathode voltage
K = fi(+isin /3 + +;cos p) becomes positive.
For motor operation, it is important that the delay angle
not exceed a maximum value, p, to allow sufficient time
for commutation to take place. If p>pm,, a converter misfire
Equation (62) represents an explicit expression for the stator will occur requiring a momentary shutdown of the dc source.
current during the commutation interval. Commutation is Thus, if the given system is used in a variablespeed electric
complete when iaS(OI) = 0. Setting the lefthand side of (62)
to zero and substituting + +
P U for 6, on the righthand drive, it is important to predict whether the commanded
delay angle p exceeds the maximum allowable value, P.,
side yields an implicit expression for the commutation angle. This is accomplished using (64). In particular, if for a given
After simplification, value of p, (64) yields a positive value for U less than 60',
then commutation will occur normally. On the other hand, if
(64) fails to provide a positive solution for U, then p exceeds
PmaXand a smaller value of P must be used.
MODEL V E R I F I C A T I O N
Equation (64) may be solved numerically for the commutation
angle U. Once the commutation angle is known, (62) can be The validity of the averagevalue relationships set forth
substituted into (58) asd (59) whi$ are then substituted into in the preceding section is established by comparing the
(51) and (52) to find iqs,eomand id,,". Unfortunately, the steadystate performance of a synchronous machine  con
resulting integral is difficult to evaluate in closed form. Alter verter system, as predicted using these averagevalue relation
natively, using Simpson's method, with four integration inter ships, with that of a detailed valvebyvalve simulation. It is
vals, yields shown that, by using these averagevalue relationships, it is
possible to predict the steadystate performance with
significantly higher accuracy than by using the classical rela
tionships in which the daxis subtransient reactance is used as
the commutating reactance.
The machine parameters are summarized in Table 1. In
the following studies, it is assumed that the dc current Idc is
equal to 0.9 pu and that the field excitation
(exfd=vfdXmd/rfd) iS equal to 2.15 pu. The firing delay angle
p is varied from zero to a maximum value which is deter
mined by the commutation limit. The resulting steadystate
characteristics are Gown in Figs. 46. The average magnetiz
ing or airgap flux Gm is shown in Fig. 4 (the magnetizing flux
@ defined as q!Jm = v), the average dc voltage
v d c is shown in Fig. 5, and the average electromagnetic
torque T, is shown in Fig. 6. The solid lines depict the
The values of i& and izs on the righthand sides of (65) and steadystate characteristics calculated using the averagevalue
(66) are obtained by evaluating i,, using (62) and substituting relations set forth in this paper, the points marked "+"
the result into (58) and (59). represent values established using a detailed representation of
Equations (65) and (66) represent closedform expressions the synchronous machine  converter system, and the dotted
for the average stator currents corresponding to the commu lines portray the steadystate characteristics calculated using
tation interval while (55) and (56) are closedform expressions the classical averagevalue equations in which the daxis sub
corresponding to the the conduction interval. Summing these transient reactance is used as the commutating reactance.
components as indicated in (49) and (50) yields the q and d
axis components of the stator currents averaged over a com T a b l e 1. Machine P a r a m e t e r s
plete 60' switching interval.
r
97
becomes positive corresponding to motor operation. The
I........................
n..
maximum delay angle /3 as predicted using the method
described in this paper is 105" which is within a degree of the 2
value established using the detailed representation. The max 
imum delay angle predicted using the classical approach is
112.5'. Va,
As shown in Figs. 46, the averagevalue relations set p.u.
forth in this paper portray the steadystate characteristics of Yy.,
the synchronous machine  converter system with a much P
higher degree of accuracy than the classical averagevalue 0
relations. In fact, it is difficult to distinguish the steadystate
performance calculated using the relations set forth in this
paper with that of the detailed valvebyvalve representation.
This improvement in accuracy compared with the classical
relationships is further illustrated in Figs. 7 and 8 in which + Detailed Simulation
the instantaneous values of the stator and dc voltages, the  ProposedModel
stator current, and electromagnetic torque are plotted. Fig I .e Classical Model
ure 7 corresponds to generatorrectifier operation while Fig. 8
depicts motorinverter operation. Superimposed upon these 2 
waveforms are the average values of these waveforms as
predicted using the previously described method (dashed line)
and the classical averagevalue equations in which the daxis
subtransient reactance is used as the commutating reactance Fig. 5 Average dc voltage versus firing angle p.
(dotted line). Using the classical method, the error in average
torque is approximately 4.2% in Fig. 7, and the error in aver 1.5
age dc voltage is approximately 3.8%. In Fig. 8, @ is set to
105' which represents the maximum permissible value
without causing a commutation failure. Using the classical 
1 + Detailed Simulation
method, the error in average torque is approximately 24.4%
and the error in average voltage is approximately 16%. This
Te
p.u.
 ProposedModel
e.. Classical Model
inaccuracy of the classical model can lead to an overly conser
vative or an inadequate system design. It has been observed
that some improvement in accuracy for motor operation can
be achieved by using the arithmetic mean of the q and daxis 0
subtransient reactances as the commutating reactance; how
ever, the resulting model less accurate when used to predict
steadystate response for generator operation. On the other
hand, the average values calculated using the method set
forth in this paper are nearly identical, in all cases, to the
values established using the detailed valvebyvalve represen
tation. Clearly, the averagevalue relations presented in this
paper are significantly more accurate with only a minor
increase in complexity. 1.5 
I
Fig. 6 Average electromagnetic torque versus firing angle p.

1ym are related to the converter firing delay angle and dc current.
p.u. It is shown that these relationships accurately predict the
steadystate characteristics of synchronous machine  con
verter systems for either alternator  rectifier or inverter 
motor operation with a significantly higher degree of accuracy
than the classical averagevalue equations in which the daxis
subtransient reactance is used as the commutating reactance.
This improvement in accuracy is achieved with only a minor
increase in complexity.
+ Detailed Simulation
i
ACKNOWLEDGMENTS
 ProposedModel
Classical Model This work was performed as part of contract N6153389
D0037 DO5 "Electric Driver Modeling and Performance Stu
0  O0 60' 1200 dies" with David Taylor Research Center, Annapolis, Mary
land.
PJ.
1.2
I

"as,
p.u.
1.2p\
0 .\ &fl
1.2 J
 ..........ss
U
.#'
 .
...
.. .:
.1
:
1.L 2
0
Te
p.u.
9 T,, 2
........................................................
p.u.
1.2 
2 0 ,
p.u.
01 I I I I I I
0 60 120 180 240 300 360
e,, deg
Fig. 7 Steadystate operation as a generatorrectifier; solid Fig. 8 Steadystate operation as a motorinverter at max
line  instantaneous variable; dashedline  average imum permissible delay angle; solid line  instantane
value established using method presented in paper; ous variable; dashedline  average value established
dotted line  average value established using classical using method presented in paper; dotted line  aver
approach. age value established using classical approach.
99
APPENDIX A List of Symbols
vdc dc voltage
hc dc current
x, commutating reactance
a firing delay relative to subtransient voltage
P firing delay relative to rotor position
4 angle of subtransient voltage relative to rotor
U commutation angle
vag,vbs, vcs stator voltages
vis ,vss q and daxis components of stator voltages
in the rotor reference frame
i,,, ibs, i, stator currents
,iis q and daxis components of stator currents
in the rotor reference frame
Gag, As, $stator flux linkages
PqS,@ds q and daxis components of stator flux linkages
in the rotor reference frame
E; ,E: q and daxis subtransient voltages
E magnitude of subtransient voltage
X i ,Xi q and daxis subtransient reactances
h q qaxis damper flux linkage
h d daxis damper flux linkage
hd field winding flux linkage
*ms qaxis magnetizing flux linkage
*md daxis magnetizing flux linkage
4m net magnetizing flux linkage
tkq qaxis damper current
lkd daxis damper current
ifd field winding current
Te electromagnetic torque