Você está na página 1de 8

IEEE Transactions on Power Apparatus and Systems, Vol. PAS-100, No.

11 November 1981 A SIMPLE AND EFFICIENT AC-DC LOAD-FLOW METHOD FOR MULTITERMINAL DC SYSTEMS
H. Fudeh, Student, IEEE

4389

C. M. Ong, Senior Member, IEEE

School of Electrical Engineering Purdue University West Lafayette, Indiana


Abstract ac-dc Load-flow method for multiterminal dcthat uses a novel approach to solve the equations of the multiterminal dc system is described. A simpLified version of the ac-dc load-flow method with all of the capabilities of estabLished Newton's methods is shown to have the best qualities of simultaneous and sequential solution ac-dc methods combined. Next, important features such as discrete voltage and taps, distinction between scheduled scheduled angle controls, fixed taps and extended phase controls, and current limits are incorporated to enhance the versatility of the method, especially for transient stability studies. Numerical examples are used to ilLustrate the various features of the method. Its performance is compared with those of an established Newton's method and a recent method. The method is simpLe to program, economical and fast.
A
new ac

systems

INTRODUCTION
Most ac-dc load-flow methods [1-9] can be cLassified broadLy into two main categories: the simultaneIn simulous and the sequential soLution methods. taneous solution methods, the equations describing the various dc system's components - the dc network, the dc terminals and their controls - are incorporated with the equations of the ac system; the collective set of equations of the ac-dc systems is solved, ususolualLy by a Newton's method. But the sequential tion methods maintain and solve the equations of the dc system separately from those of the ac system by considering the real and reactive powers and the ac voltages at the converter buses as interface conditions that can be iterated upon until these conditions in both ac and dc solutions match. Often, both dc and ac solutions have to be repeated a few times. For existing ac-dc systems with few dc terminals, the computing effort and time for the dc solution are only small fractions of those for the ac soLution; the efficiencies of sequential solution methods are, therefore, diluted mainly by the extra effort of having to On the other hand, the repeat the ac solution. sequential solution methods are simpler to program and will adapt much more easily than simultaneous solution methods to future development of ac load-flow technique and to new forms of dc system controls. Depending on which representation is used [5,2), the total number of variabLes in the equations With describing a bipolar terminal can be 11 or 15. these many variables per bipolar terminal, the computing times and storage requirements of conventional Newton's methods wilL increase rapidly if the number of terminals in the dc system is increased.

An important part of dc load-flows which has received considerable attention is the representation or the implementation of the conditions established by the controls of the converter and its transformer. Scheduled current, scheduled power, and scheduled voLtage with a certain minimum angle (amin for a rectifier or for a rectifier) are the types of converter controls that have been considered with transformer taps that are assumed to be continuous. Very little effort [9] has been made to distinguish between scheduled voLtage (constant voltage obtained with variable extinction or ignition angle characteristics) and scheduled angle (constant extinction or ignition angle) controls that are used on the voltage This distinction is not only controlling terminal. necessary for handling a new power factor control [10] but also essential for representing the different transient behaviors of these two controls. It is also proper that discrete tap be considered along with these controls, because the approximation of the discrete tap by a continuous tap can introduce as much as 9% error in the control angle alone. Ac load-flow methods play a useful role in transient stability studies of ac systems. For an ac-dc load-flow method to have the same useful role in transient stability studies of ac-dc systems, it should be capable of handling the conditions of fixed taps and extended phase controLs during a transient: the taps often remain unchanged for considerable Length of time of the transient because of the built-in delays in the tap controLs, but the responses of the phase controls of the converters are prompt. Furthermore, the dc currents of the terminals with scheduled power controls and the dc current of the voltage controlling, or slack terminal, can change significantly during a transient, the current limits at these terminals should be considered. All these conditions of transient operations have received very little attention. A new ac-dc load-flow method for multiterminal dcac systems that uses a novel approach to solve the equations of the multiterminal dc system is described. This paper is organized into three main sections. First, a simplified version of the ac-dc load-flow method with all of the capabilities of established Newton's methods is shown to have the best qualities of simuLtaneous and sequential solution ac-dc methods combined. Next, important features such as discrete and voltage taps, distinction between scheduled scheduled angle controls, fixed taps and extended phase controls, and current limits are incorporated to enhance the versatility of the method, especially for transient stability studies. FinalLy, numerical exampLes are given to illustrate the various features of the method. Its performance is compared with those of an established Newton's method and a recent method.

ymin

A SIMPLE VERSION OF THE METHOD


A paper recommended and approved by the 81 sl 301-1 IEEE Power System Engineering Committee of the IEEE Power Engineering Society for presentation at the IEEE PES Summer Meeting, Portland, Oregon, July 26-31, 1981. Manuscirpt submitted August 29, 1980; made available for printing March 23, 1981.

The simplicity and the economical computationaL requirements of the new approach are best ilLustrated by This load-flow. a simplified version of the ac-dc simple version, nevertheless, has all of the capabiLities of established load-flow methods. For this simplified version of the ac-dc load-flow, the usual assumptions of continuous converter transformer tap,

1981 IEEE

4390

scheduled voLtage control with a certain minimum controL angle, and fixed voltage margins at those terminals with a scheduled current or power control, as in established methods, are made.

Converter controls equations


A practical operating scheme for multiterminal dc system using local terminal controls is to have the dc system voltage determined at one terminal - the volThe other terminals are tage controlling terminal. provided with scheduled power or current settings. To keep the reactive power consumptions of the converters and the losses in the snubber circuits low, the control angles should be small. But to maintain phase control and reliable commutation, a minimum control angle should be maintained. Typical values of the minimum ignition angle a min range from 5 0 to 7 0 and those of the minimum extinction angle range from 150 to 200. In most load-flow methods, the voltage controlLing terminal that is operating at the scheduled voltage sch Vd is also assumed to operate with a certain minimum

Basic Equations
The basic equations describing the converter with its firing angle and tap controls and the dc network are summarized. The equations and assumptions are similar to those given in references [5).
Converter equations
The converter model is based on the relationship between the ripple-free average dc quantities and the fundamental frequency ac quantities. Based on the per unit system given in Appendix III, the following equations can be written for every converter terminal. For the kth converter, its dc voltage equation in terms of its tap ak, ac voltage Vk control angle 0k' commutation resistance Rck, and the dc current Idk is

control angle Omin Thus if the mth terminal is the voltage controlling terminal, its dc voltage and control angle are
V

Vdk = akVkcos k

Rck dk

(1)
and

dm
=

Vsch d
0min m

(5) (6)

Its dc power equation is

em

Pdk =VdkIdk
Neglecting the losses in the kth converter and its transformer and equating the expressions for powers on the ac side and dc side, the equation obtained for its power factor angle (* -Ck) is V =~ ~~

Vdk

akVkcos(*k3k))

For the simple circuit representation of the converter transformer shown in Fig. 1, the equation for the reactive power flowing from the ac bus into the kth converter terminal is

Qk Pdktan(C1k-k)

(4)

But with the more elaborate representation of the transformer and auxiliary equipment shown in Fig. 7 of Appendix I, the real and reactive powers on the ac bus side of the transformer are no longer given by Eqs. (2) and (4) respectively; they can, however, be determined by the procedure outlined in Appendix I, using the known values of Pdk' Qk' Idk and akVk.
+

For the terminal with a scheduled current or power control, it is common practice to coordinate the tap control with the phase control so that the terminal will operate at some dc voltage below its own minimum ignition or extinction angle characteristic to avoid frequent mode shifts from occurring with normal ac voLtage fluctuations. Typically, a 3% voltage margin is provided; with the average a given or y above, typical values of the control angles a and y are 150 and 200 respectively for those dc terminals with a scheduled current or power controL. This typical voltage margin of 3% in practice can be considered in the load-flow computation by modifying the dc voltage equations for such terminals with a coefficient of K = 0.97 [53. Thus if the kth terminal has a scheduled current control, its dc current is equal to the scheduled
current Idk h that is an d dksch is v

and its dc voltage equation is


Vdk
=

+zo+
.9.

Idk
-0
--

6+

Kk [akVkcosOk

RCkIdk

(8)

POSITIVE POLE

Similarly, if the kth terminal has a scheduled power control, its dc power is equal to the scheduled power

pd

sch

that is p

psch (9) dk dk and its dc voltage equation is also given by Eq. (8).
Dc network equations

V_
Fig. 1. Equivalent representation

NEGATIVE POLE

dk
of a

bipolar

station

The equations for the dc network can be formulated suit the procedure that is used to solve them. Since multiterminal dc networks in the near future are unlikely to have greater than 30 buses, the present choice is the R bus Gauss-Seidel method. Although the algorithm is applicable to a general bipolar network (Appendix II), there is no loss in generality by considering a symmetrical m-terminal bipolar system that can be economically represented by
to
an

equivalent m-terminal monopolar system.

4391

If the buses are numbered so that the mth terminal is the voltage controlling terminal and its network terminal is also the reference bus for the Rbus' the voltage equations for the dc network of the equivalent m-terminal monopolar system can be written as

vdk ik =1
where

m-1
k

Id di

+ Vd, dm

k = 1,

...,

(m-1)

(10)

rki 's

are elements of the dc network's Rbus with the terminaL of the mth terminal as its reference.

Note that Vdm is the dc voltage at the terminal of the voltage controlling terminal; Vdm is equal to the

limits, the dc system's voltage at the voltage controlling terminal is normally rescheduled and the whole procedure has to be repeated. A way of rescheduling the dc system's voLtage has been described in reference [5). Briefly, the tap which exceeds its limit by the largest amount is first identified, say ak. If ak is greater than amax, the scheduled dc voltage at the voltage controlling termia nal is decreased by the ratio max And if a is less k ak than amin, the scheduled dc voltage is increased by a the ratio min ak Upper and lower tap limits can be separately exceeded at different terminals, for example, when there is an unusually steep voltage gradient in the dc network. However, this condition is not common in normal steady-state operations and has been excluded in certain methods by stopping the computation whenever such a condition is detected. A flowchart of the simple version of the ac-dc load-flow method for ac-dc systems with multiterminal dc networks is given in Fig. 2. From this description of the method of solution, many desirable features of the method become apparent:

scheduled voltage V sch of that terminal.


Method of Solution

The method of solution is simple but different from those of previous ac-dc load-flow methods. First, the voltage equations in Eq. (10) are solved = d by using Vdmdm Vsch Idi = Isch for the terminals with di

equal network solution establishes the values of dc current, dc voltage, and dc power at every dc terminal of the m-terminal network. Next, the product of the tap a and the ac voltage V, or aV, at every terminal is determined individually: For the voltage controlling terminal, this is determined by substituting its values of Vd and 0 from Eqs. (5) and (6) and its value of Id from the dc network soLution into Eq. (1). For the other terminals with scheduled current or power controls, their aV's are determined individually by substituting their

scheduled current settings, and Idi = Pdch /Vdi for those with scheduled power settings in the GaussSeidel iterative procedure of the Vd's that starts s to Vd ch . This dc set with all Vd's

(1)

The new ac-dc method can adopt any method for the ac solution.

ac

load-flow

initially

(2)

It is an economical method; both dc and ac solutions need only be computed once to obtain the ac-dc solution with taps that are within their limits.
Its storage requirement for the dc solution is mainly for the Rbus Gauss-Seidel dc network solution; this is minimal compared to that for the Jacobian in most Newton's method. It is a simple approach to understand and to program.

(3)

(4)

values of Vd and Id from the dc network solution and the value of K into Eq. (8). Note that at this point, the V's at the converter buses are still unknown quantities. With the knowledge of the aV's and the Vd's, the power factor angLes (*-U)'s can now be determined from Eq. (3). With the simple transformer representation shown in Fig. 1, the real and reactive powers flowing from the ac bus to the converter terminal are given directly by But with the more elaborate Eqs. (2) and (4). transformer representation of Fig. 7 in Appendix I, the real and reactive powers flowing from the ac bus have to be determined iteratively using the known values of Pd' Q, Id, and aV at that converter terminal. In any case, the real and reactive powers flowing from the ac buses into every converter terminal, or the whole dc system, can now be obtained. Knowing the real and reactive powers fLowing from the ac buses into the dc system complete the description of all the real and reactive loads on the ac system; the ac load-flow can be determined. The ac load-flow provides the values of ac voLtages at those ac buses connected to the dc system; knowing these ac voltages V's, together with the aV's obtained previously, the tap a of every converter transformer can be determined. And if these taps are within their upper and lower limits, the complete ac-dc solution is obtained. However, if any of these taps exceeds its

2
3

1.
2.

3.
4
5 6 8L

4e
6.
8.

5,

7.

9.
YES

ENTER WITH DATA AND SET INITIAL CONDITIONS SOLVE DC NETWORK'S EQUATIONS FOR V 's$ d I's AND P d's d DETERMINE THE aV PRODUCTS AND POWER FACTOR ANGLES (X - E) S DETERMINE REAL AND REACTIVE POWER LOADINGS OF DC SYSTEM AT AC BUSES AC LOAD-FLOW DETERMINE TAPS a's FROM AC VOLTAGES V's AND aV PRODUCTS TAP LIMITS EXCEEDED? RESCHEDULE DC VOLTAGE PRINT RESULT

Fig. 2. Flowchart for the simple version of the ac-dc load- flow

4392

Capabilities of the simple version of the method


In spite of its simplicity, the simpLe version of the method has many capabilities: It wilL handle general bipoLar dc systems with symmetrical or asymmetricaL network configurations. For asymmetricaL network configurations, the network equations for both positive and negative poles (an example of the positive pole equations is given in Appendix II) can be solved simultaneously by the Rbus Gauss-Seidel iterative method using the scheduled values of the converter terminals in both poles. Back to back 'systems, and also pole-paralleling operations, with zero resistance network connections between adjacent terminals present no problem to the Rbus formulation used. Voltage control at a remote dc bus, or line voltage drop bias, can simply be accommodated by rearranging the dc' network equations so that the specified voltage Vdv at the remote bus appears on the right hand side of the network equations. This is accomplished by expressing Vdv in terms of Vdm using Eq. (8) and then substituting back to eliminate Vdm; Eq. (8) becomes

minor calculations have been added to the simple version of the ac-dc load-flow that is given in Fig. 2. Limits on the dc currents of the terminals can be conveniently applied during the Gauss-Seidel iterative solution of the dc network equations.
For scheduled angLe with discrete tap and fixed tap with extended phase controL, iterations between dc and ac soLutions are required. But these are special features which neither 'simuLtaneous nor sequential soLution Newton's methods has been shown to be capable of providing. Besides, the numericaL examples will later show that for conditions requiring these features the numbers of iterations between ac and dc solutions of the method are about that taken by an established Newton's method to handle conditions that the simple version of this method can solve with just one dc soLution and one ac solution.

Vdk dk =

m-1 2 (rk

i=1

V i - r.)Id i

V, d

k=1,v k*v

...

(11)

SPECIAL FEATURES ADDED


Features such as discrete tap, distinction between scheduled voltage and angle controls, fixed taps and extended phase controls, and converter current limits that are important for transient stability studies of ac-dc systems can all be easily incorporated with the efficient method of solution above to form a truly versatile ac-dc load-flow method. The numerous but simple steps taken to incorporate these features are best described by flowcharts. First, a simplified flowchart of the ac-dc load-flow is given in Fig. 3. A number of blocks containing

The main steps in the dc load-flow are shown in Fig. 4. Many of these steps involve very little computing effort. Because the terminal with a scheduled angle control and discrete tap can have only discrete step changes in its open-circuit voltage, its condition is better handled by manipulating directly with its open-circuit voltage es (or aVcossc h) than with its terminal voltage Vd. Thus for the terminal with a scheduled angle control, an internal bus with open-circuit voltage es is defined, and at the same time its commutating resistance is incorporated into the Rbus of the dc network.

1
NO

W1
1 3o5 ENTER

|4 {5x---

I. 2. YES A{6>
\ NO -

ENTER

3. READ DATA AND SET INITIAL CONDITIONS 4. FORM R-BUS MATRIX


5.

FIRST DC LOAD-FLOW?

6. 8.

<4 )

YES

4. UPPER TAP LIMIT EXCEEDED?


5.
RESCHEDULE DC VOLTAGE
6.

2. 3,

DC LOAD-FLOW

R<

YES

7.

FIXED TAP? SCHEDULED VOLTAGE CONTROL?

10

9.
10.

AC LOAD-FLOW

tYES

\6
<8 ANO

L >

7.

9.

8.

CONTINUOUS TAP? PRINT RESULT FIXED TAP?

11.
12

12.

10 |

11. 12. 13.

10.

l11 1
ES

14.

15.

SCHEDULED ANGLE CONTROL AND DISCRETE TAP? DETERMINE OPEN-CIRCUIT VOLTAGE OF TERMINAL JWITH SCHEDULED ANGLE CONTROL USING THE UPDATED AC VOLTAGES AC LOAD-FLOW
CHECK CONVERGENCE OF AC VOLTAGES OF AC

cos' s ASSUMING FIXED acose PRODUCTS DETERMINE cosO's USING UPDATED AC VOLTAGES UPDATE O CONSUMPTIONS

DETERMINE NEAREST TAP POSITIONS AND

/13 14
1/ 10
| 1fiM

13.

NO

NO
NO
17

19 20
,

NO

17.
21

14. 15. 16.


18. 19.

SCHEDULED ANGLE CONTROL? IF DISCRETE TAP, SET ITS OPEN-CIRCUIT e TO CORRESPOND TO THE SCHEDULED ANGLE AND TAP ADD R OF SCHEDULED ANGLE CONTROL TERMINAL R-BUS ~~~~~~~~~~TO UPDATE DC VOLTAGES DETERMINE DC CURRENITS OF SCHEDULED POWER lCONTROL TERMINALS, SET TO MAXIMUM VALUES IF THESE LIMITS ARE EXCEEDED CHECK CONVERGENCE OF DC VOLTAGES DETERMINE CURRENT, POWER AND VOLTAGE OF SLACK TERMINAL SCHEDULED VOLTASE CONTROL? FIXED TAP? SET V = Om AND CALCULATE aV OF BUS AT 5 mn SLACK TERMINAL DETERMINE 0 FROM DC VOLTAGE EQUATION

DETERMINE

sYES ______16.

BUSES WITH DC TERMINALS PRINT RESULT

20.
, l

DETERMINE 0
-

FIORD TAP?e FROM


sch
s

cos

21

oH5

S/

SaV

L22. 23.
25.

<22

YES

24.

26.

TAP? DETERMINE cosO FROM DC VOLTAGE EQUATION WITH KNOWN DC CURRENT, VOLTAGE AND aV PRODUCT DETERMINE aV AND cos9 FOR NON-SLACK TERMINAL DETERMINE P AND Q LOADINGS ON AC SYSTEM RETURN

FIXED

Fig. 3. Flowchart for the ac-dc load-flow

Fig. 4. FLowchart for the dc Load-fLow

4393
TEST

RESULTS

all

of the features mentioned above are given. The ac-dc load-flow program that is used has all of the features mentioned above. The fast decoupled ac load-flow method of Stott and Alsac C133 has been used to solve the ac load-flows in all these numerical exTable 1 gives a summary of the test cases of the two systems solved by the new ac-dc load-flow method. It gives the types and scheduled values of the controls on the converter terminals, the size of the tap steps, and the current limits.
Bd
B~ ~ ~ ~ ~
LL

In t-his section, numerical examples of two sample systems solved by the ac-dc load-flow method that has

amples.

quirements

ac test system C14]. The ac interconnections between buses 2 and 5, 2 and 4, 4 and 5 have been replaced by a 3-terminal dc system using the same ac lines. At the three converter terminals the reactive repower

System A shown in Fig. 5 is the asymmetrical biposystem of Braunagel, Kraft and Whysong; the data of this system is given in reference System B C6]. shown in Fig. 6 is an ac-dc system similar to that used in reference E8], it is based on the AEP 14 bus

lar

are

tive powers with the ac

compensated flowing from the

locally so
ac

connections before.

buses

are

that the reacabout those

Bdc

Bd

l Ed

JD

ac)~~~~~~~~~~~~~~~~~~~~~~~~~~~~a~

d 8

r
ac2

~~~.

Bdc9

B
ac5

Fig. 5. Sample bipolar system

Fig. 6. Sample ac-dc system

Table 1.
Isysteml
case

A summary of test cases


C5

types of converter control

Cl
4.90

C2
.980

C3
4.90

C4
.980

C6

special constraints
continuous TAP continuous

iteration /

timetsec)

II(sch)IV(sch)IItsch)IV(sch)IP(sch)lI(sch)I
4.90

I1
A

1-4.90

I"2"dc 1.001

|I(sch)l

2
A

IV(sch)l
.980 4.90 4.90
.980

4.90

II(sch)IV(sch)lP(sch)ll(sch)l
4.90 .975 4.90 1-4.90

TAP,remote controL of 1.001 10's voltage C

I"2"dc

II(sch)|Y(sch)tI(sch)IA(sch)|P(sch)II(sch)I
I
4.90
.980
4.90 1-4.90

3
A
4

I step=.015
discrete TAP

discrete TAP

|"2"dc
1.002

II(sch)It(sch)II(sch)1A(sch)IP(sch)I(sch)I
18.0
4.90
18.0
4.90 1-4.90

1"2"dc
1.002

step=.015

5
B

II(sch)| V I1(sch)I A IP(sch)tI(sch)l fixed I"2"dc TAP 4.90 1-4.9 4.90 1.8912 4.90 1.8855 1(max)=5.0 1.002

I1 1
B
2

IP(sch)IP(sch)IV(sch)l I.976451.211351.9800
IP(sch)IP(sch)| 1.976451.211351 IV(sch)l
.980

continuous TAP continuous

I"2"dc,(16A,12V)ac
1.001 + .26

TAP,remote

IC2's voltage
I step=.015
discrete TAP

control of

I"2"dc,(160,12V)ac 1.001 + .26

B 3
B

IP(sch)|P(sch)I|(sch)l |.976451.211351.9800

1"2"dc,(16A,12f)ac,(1tl1V)ac +.03 1.001 + .26

4
|8 5

IP(sch)IP(sch)10(sch)l | .976451.211351 18.0 1


IP(sch)IP(sch)l A 1.976451.211351.8751
I

discrete

TAP step=.015
fixed TAP

I"2"dc,(160,12f)ac,(10,2V)ac,"2"dc,(1f,1V)acI .04 + .001+ .03 + 1.001 + .26


1.001 + .26

I1(max)=.80

1"2"dc,t(16,12V)ac,t(1,2V)ac,"2"dc,t(1,1V)acl + .001+ .03 + .04

The number in quotes are the Gauss-Seidel iterations for the dc network solution in block 2 of Fig. 2 or bLocks 11, 12 and 13 of Fig. 4.

4394

In Table 1, and also in Table 4, the numbers of iterations in the dc and ac solutions are given along with the recorded computing times taken on a CDC 6500 computer. The numbers enclosed within the double quotation marks are the Gauss-Seidel iterations for the dc network solution. The notations of reference C13] have been used to present the recorded numbers of solutions for EAe) and [AV] in each ac load-flow. With continuous taps, the method took just one dc soLution and one ac solution to solve the ac-dc system. If the taps were discrete, onLy a second dc solution was required. However, for scheduled angle control with discrete taps, and also fixed taps, not only a second dc solution but also a third ac solution was required. The resuLts of the tests on each of the two sample systems are given separately in TabLes 2 and 3. The asterisk placed after a value of power, voltage or current indicates the type and the scheduled vaLue of the converter control used. In the comparative study, method A is the method of Reeve, Fahmy and Stott E5), which is an established Newton's method with many of the capabilities mentioned earlier. Method B is the sequential method of Ong and Hanzei-nejad E93, which has all of the capabilities of the new method. The convergence tolerances used in methods A and B were 0.001pu for the residuals of dc and ac quantities in the ac and dc Load-flows and 0.01 pu for the residuaLs of the reaL and reactive powers that were exchanged between ac and dc load-fLows. In aLL the tests, the convergence tolerances used in the new method were 0.00001 pu for the Gauss-Seidel dc network soLution and 0.001 pu for the ac Load-fLow.

The comparison of performance given in Table 4 shows the merits of the new method. The new method required only one dc solution and one ac solution, the other methods required more for the same ac-dc solution. The new method is also a faster method than the

others.

CONCLUSIONS

This paper has described a simple method for ing the equations of the multiterminal dc system ac-dc load-flow. The method

solv-

in

an

simplified version of the ac-dc load-flow which has all of the capabilities of estabLished Newton's methods is shown to have the best qualities of simultaneous and sequentiaL solution acdc methods combined: It is an economical method; both dc and ac solutions need only be computed once to obtain the ac-dc solution with taps that are within their limits. It can adopt any ac load-flow method for the ac solution. Furthermore, its storage requirement for the dc solution is minimal compared to that for the Jacobian in Newton's conventional methods.
The versatility of the ac-dc method is enhanced by the incorporation of features that are important for transient stability studies, such as discrete tap, distinction between scheduled voltage and scheduled Its angle controls, and converter current limits. versatility is illustrated by the variety of dc controls and system conditions of the numerical examples that have been solved by the method.
The comparison of performance with those of an established Newton's method and another recent method showed that the new method is economical and fast.

TabLe 2.
case

NumericaL results
voLtage I current
Vd Id P

on sample
converter

bipolar systems.
tap
controL
Ti

Iconvertericonvertericonverterl
number

Load

setting
1.0289
1.0353 1.0521 1.0498

|(degrees)l
15.605 15.604 15.605
21.126

angle

voltage
Vn

neutral

(1)

C1 C5 C6 C2 I C3 C4
ci C5 C c C6 C2 C3 C4
C5

.9900 4.900 *| 4.851 4.954 I 4.900 * .9891 .9695 -4.900 o| -4.751 -4.855 .9800 o| -4.954 .9897 4.900 of 4.850 .9800 of -4.900 -4.802

1.580 1.600 2.015

1.580 1.598 1.566


1.616

1.618

1.0285 1.0433

16.000*1
16.000*1
15.604 21.124

0 0 .0098 .0001

0 .0001 0 .0098 .0005


0

Table 3.
I

Numerical results

on sample

ac-dc system.
control
AC

(2)

I .98000 *v 4.900 *o 4.802 5.152 I 4.900 * .95110 .93150 -4.900 of -4.564 .98930 -5.152 I -5.097 .98430 4.900 vf 4.823 .97500 of -4.900 I -4.777 .9900 4.900 4.954 .9891 .9695 I -4.900 .9800 of -4.954 .9897 4.900 .9800 of -4.900 I .9332 .9329 I .9133 .9221 .9378 .9287 .8957 .8949 .8753 .8855 .9008 .8912 *o

1.943 1.705 1.572


1.591

I I

1.0186

1.0113 1.0602
1.0230 1.0380
1.0450
1.0600

.9965 I

15.560

16.000*1
15.604

16.000*1
15.829
17.412

0 .0005

case

fconverterlconverterlconverterl
number

voLtage
Vd

converter

current

Id

C1

(3)

C6 C2 C3 C4
C1 C5 C6

o| -4.751

4.851 4.900 *

load
* * * *

setting
Ti

tap

1.598
1.749

I -4.855 of 4.850 I -4.802

2.112
1.603 1.624

1.769

|* f* |* f* fo fo

1.0300

0 .0098

|(degrees)lmagnitudel
1.045 .9941
1.045 .9941 1.024

angle

voltage

1.0600
1.0300 1.0450 I

22.206

17.825 15.892 16.327

.0001 0 .0001 0 0 .0098

(1)

C2 C3 C1 C2 C3

C1

.99778 f .9786 .98465 f .2146 .98000 *1-1.1933

1-1.1694

.97645 .21135

.4245 .0678 .4975 .4257 .0679 .4987 .7072 .5067 .4820 .7485 .5602
.4384 .1067 .5287

f f

1.0411 l 15.381 1.0402 15.614 16.000 *| 1.0405


1.0368 1.0354 1.0360

(4)

C2 C3 C4
C5

f 4.900 *o 4.573 I 5.252 I 4.900 * -4.900 of -4.475 -5.252 f -4.843 I 4.900 o 4.595 f -4.900 f -4.551 4.900 f* 5.000 I -4.900 -5.000 4.900 -4.900 *o f of f of 4.389 4.474 -4.289 -4.427 4.414 -4.367

1.717 1.738 2.007

1.804 1.656 1.680 2.308 2.287 2.380 2.146 2.259 2.112

|* 0.9850 f* 0.9850 I f* 1.0000

f* 0.9850 f* 1.0000 a 0 0 f*

1* 1.0000

18.364 16.994 2 2.293 2

(2) I
(3)

.99319 f .9831 .98000 o| .2157 .97533 -1.199


.9978 .9846 .9800 .9230 .9090

1-1.1692

.97645 .21135

16.000 *l

15.378 15.614

1.024 1.045 .9936 1.023 1.045 .9931 1.017 1.045 .9850 1.012

18.000*|
25.382 27.472 f 24.049 25.467 24.049

17.508

18.000*1
26.138

.0007 0 .0007

C1 C2
C3 C C2 C3 C C2

.9786 .97645 .2146 f .21135 *f -1.193 1-1.1694

f f

* *
* *

.4343

1.0450 16.133 1.0450 I 16.436 I 1.0450 16.562


0.9850 0.9700 0.9850 18.102 17.204 I 18.000 *| 26.927 25.121 25.519

C1

(5)

C6 C2 C3 C4

1.0000 1.0000 1.0000 1.0000 |* 1.0000 fo 1.0000

0 0 .0098 .0003 0 .0003

I (4)

.9038
.8898 .8793 .8751

1-1.2905
|* .8000

1.0579 .2326

1-1.1663

.97645 .21135

(5)

C3

1-1.0404

.2404

1-.91039

.71185 .21135 *

f*

l* 1.0000

1.0000

|* 1.0000

Table 4.

Comparative performance
method
A

on a

CDC 6500 computer.


method B new method

Isysteml
case

I-

I.-.-. --------------lI_I_1.001 .01 | time(sec) | .85 | 1 . ______- -.-------------------------------------- --------------------------------------------_ B iteration ld|c,(160,12V)ac,ldc,(10,2V)ac,ldcI 2dc,(160,12V)ac,3dcl"2"dc,(160,12V)acI ----I.-------Il time(sec) 1.095+ .26 +.095 + .04 +.0951.005+ .26 +.008 1.001 + .26 1
.
___________
-----------------------------------

I A

iteration

2dc

2dc

I"2"dc

--_I-.--

--

--

--

--

---

--

4395

ACKNOWLEDGEMENT

This work was sponsored by the Electric Energy Systems Division of the U.S. Department of Energy under the contract ET-78-S-01-3396.

I :a
..

P Q'
aV
X
p

xs

REFERENCES

Vt
Xf

1.

2.

3.

4.
5.

6.

7.

8.

9. 10.

11.
I

12.
13.

14.

1978. J. Arrillaga and P. Bodger, "AC-DC Load-flow with reaListic representation of the converter plant," Proc. IEE, Vol. 125, No. 1, 1978, pp. 41-46. J. Arrillaga and P. Bodger, "Integration of HVDC links with fast-decoupled load-flow solutions," Proc. IEE, Vol. 124, No. 5, 1977, pp. 463-468. J. Reeve, G. Fahmy and B. Stott, "Versatile load flow method for multiterminal HVDC systems," IEEE Trans. Power Apparatus and Systems, VoL. PAS-96, No. 3, 1977, pp. 925-933. D. A. Braunagel, L. A. Kraft and J. L. Whysong, "Inclusion of dc converter and transmission equations directly in a Newton power flow," IEEE Trans. Power Apparatus and Systems, Vol. PAS, No. 1, 1976, pp. 76-88. G. B. ShebLe and G. T. Heydt, "Power flow studies for systems with HVDC transmission," Proc. Power 1975, pp. Industry Computer Applications, 223-228. H. Sato and J. ArrilLaga, "Improved load-fLow techniques for integrated ac-dc systems," Proc. of IEE, Vol. 116, No. 4, 1969, pp. 525-532. C. M. Ong and A. Hamzei-nejad, "A general-purpose multiterminal dc load-flow," IEEE 1981 Winter Meeting, Paper 81 WM 0207 F. Nishimura, A. Watanabe, N. Fujii and F. Ogata, "Constant power factor control system for hvdc transmission," IEEE Trans. Power Apparatus and Systems, Vol. PAS-95, No. 6, pp. 1845-1853. E. W. Kimbark, Direct current transmission, Vol. 1, John WiLey & Sons, Inc., New York, 1971. E. UhLmann, Power transmission by direct current, Springer-Verlag, Berlin Heidelberg, New York 1975. B. Stott and 0. Alsac, "Fast decoupled load flow," IEEE Trans. Power Apparatus and Systems, Vol. PAS-93, May/June 1974, pp. 859-869. IEEE Computer Applications Sub-Committee Standard Test System, American Electric Power Service Corporation, 1962.
APPENDIX I

M. M. El-Marsafawy and R. M. Mathur, "A new, fast technique for load-flow solution of integrated multi-terminal dc/ac systems," IEEE Trans. Power Apparatus and Systems, Vol. PAS-99, No. 1, 1980, pp. 246-253. J. Arrillaga, B. J. Harker, and P. Bodger, "Fast decoupled load flow algorithms for ac-dc systems," IEEE Paper A78 555-5, Presented at the PES Summer Meeting, Los Angeles, CA, July 16-21,

Fig. 7. Equivalent representation of a 3-winding transformer with auxiliary equipment.


To determine the real and reactive powers fLowing from the ac bus into transformer, a simple load-flow of the T network has to be solved using bus 1 as the reference bus with a voltage magnitude aV, bus 3 as a P , IVtI bus, and bus 2 as a Pd' Q' bus. At bus 2 the real power Pd is the dc power obtained from the dc network solution and reactive power Q' is calculated from the expression

Q'

Pd

tan

(4-O)

(d)_2>

(12)

APPENDIX II
GENERAL BIPOLAR NETWORK EQUATIONS In a general bipolar system, the dc network configuration, and also the number of converters, of the positive pole can be different from those of the negaAt a terminal the dc currents of the two tive poLe. poLes need not be the same, in fact, one of these For such asymmetricaL currents may even be zero. operation, the equation for the positive pole dc terminal voltage at the kth terminaL of the general mterminal bipolar system is
+ = Vd dk

in-1

X= k~idi

+ - r(I dk -I dk ) + rgk

dm

(13) dm gm Cdm where rki are the eLements of the positive pole network Rbus matrix with the mth terminal, the voltage controlLing, chosen as reference. The dc currents flowing into the positive pole dc network should sum to zero. Thus,

THREE-WINDING TRANSFORMER REPRESENTATION

E i =1

Idi

=0

(14)

ary

Many 3-winding converter transformers with auxiliequipment can be represented by the equivalent circuit given in Fig. 7 [see discussion of [6]].

Similar equations can be written for the negative pole dc terminal voltage and currents.

4396
APPENDIX III

Vac base = VMline to line, rms value).

List of symbols V/ - ac bus voltage and phase angle I/_ - r.m.s alternating current and phase angle a - transformer tap ratio a - control angle, ignition or extinction angle e - open-circuit voltage, Vacose Vd - direct voLtage Id- direct current r.. - (i,j)th element of the resistance bus ma1J trix rg - resistance of ground connection X - reactance Rc - equivalent commutating resistance nb - number of series-connected bridges in a terminal superscripts: + for positive poLe quantities - for negative pole quantities max for maximum value min for minimum value sch for scheduled value
Per unit system A common base power dc systems.

Zac base =
ac base Zhoe

Vacbase base ac b =K ac base

,v3 V ac base

abase

vac

Choose V dc base

K V ac base

ir where K = 3V4 nb
then

Idc

base

Zdc

basez =K

K~~2

ac base

Zac

base

If Pbase is on one converter transformer rating and Xc is the per unit commutating reactance, the equivalent commutating resistance in per unit is given by
Rc

Pbase

is chosen for both

ac

and

6n b X c

Você também pode gostar