IEEE Transaction on Power Apparatus and Systems, Vol. PAS-103, No.

7, July 1984
A METHOD TO TEACH THE DESIGN AND OPERATION QF A DISTRIBUTION SYSTEM
W. H. Kersting Senior Member

1945

New Mexico State University Las Cruces, New Mexico 88003 INTRODUCTION The undergraduate "power option" program at New Mexico State University consists of five courses. Of these five, one is required of all electrical engineering majors, one is a "block elective" and three are senior electives. The "block elective" is a method by which an electrical engineering major must select three courses from a list of six. The list includes one course in each of the six "option" areas offered by the department. The five power courses are:
1.

has been found that this course provides a great mechanism for review of some basics of power systems theory and a marvelous way in which to develop and expand upon the basics.
The goals of the
course are

straight forward:

1. 2. 3. 4. 5.
6.

Define the basic components starting with the bulk power substation and terminating at the
Define and discuss this thing called "load". Develop typical substation layouts. Develop typical layouts of substransmission and distribution feeder systems. Define acceptable steady-state operating conditions. Develop methods of analyzing the steady-state operating conditions of unbalanced three-phase
systems.

customer's meter.

Electric

Machinery ---

required of all

2. 3.
4.

Power Transmission Lines "block elective" Power Systems Analysis ---- senior elective
Distribution

majors
senior

5.

Power System Protection

Systems -

elective

---

senior elective

The first three courses are offered every semester and are taken sequentially by the students. The last two courses are offered in alternate semesters and may be taken along with the Power Systems Analysis
course.

The purpose of this paper is to discuss the Distribution Systems course. In particular a computer program will be described that has been developed to help demonstrate some design and operating characteristics of a distribution system.

inherently unbalanced three-phase distribution system. The review process is what makes this a good terminal couirse for graduating seniors or a good first semester g--raduate course for students returning to school after working in induistry or for new graduate students who did their undergraduate work at some other university.
One of the difficulties of teaching the Distribution Systems course, and most power systems analysis course's, is not being able to have a real system to experiment with in the laboratory. Without such a real system to experiment with some simulation means is necessary. As is true in most cases like this, the digital computer has become a very popular device used to simulate real systems.

In the process of achieving the goals it is necessary for the students to review material from previous power courses and-then to expand upon these basics in order to understand the design and operation of an

THE DISTRIBUTION SYSTEMS COURSE

at New Mexico State is the industry advisory committee that is an integral part of the industry sponsored

One of the positive things about the

power program

graduate program. One important function of the advisory committee is to annually review the power curriculum (undergraduate and graduate) and make recommendations for changes, additions and/or deletions. It-was this advisory committee that made the recommendation nearly fifteen years ago for the creation of a course in Distribution Systems. It was further recommended that the course should be offered as a senior elective for terminating students or as a first semester graduate course for students starting the graduate program. The reason for the need of a Distribution Systems course was simply that the majority of new electrical engineers going to work for a utility company are first assigned to work in the distribution department. On the surface it may appear that this is a much too specialized topic to appear as a senior elective let alone as a first semester graduate course. However, it

During the last ten years a "Radial Three-Phase Power Flow Program" (RADFLOW) has been developed to simulate the design and operation of a distribution system. This program has become a very effective tool used by the students to solve regular homework problems and assist in the semester long project of the design and analysis of a distribution system.
THE RADIAL THREE-PHASE POWER FLOW PROGRAM
years ago

A paper recommended and approved 84 WM 179-8 by the IEEE Power Engineering Education Committee of the IEEE Power Engineering Society for presentation at the IEEE/PES 1984 Winter Meeting, Dallas, Texas, January 29 - February 3, 1984. Manuscript submitted August 12, 1.983; made available for printing November 11, 1983.

The first version of RADFLOW was developed ten and was a basic Gauss-Seidel program expanded to model all three phases of a distribution system. 1] Unfortunately this first version was handicapped by the very poor convergence characteristics of GaussSeidel on radial systems. The students spent more time fighting the convergence problems than they did learning about distributions sy,stems. The first version was replaced by the-present RADFLOW program which utilizes a modified ladder network theory in the iterative routine. This method displays outstanding convergence characteristics for radial systems. [ 2] Before describing the modified ladder theory, it is necessary to develop the line section model and the load model used by the program.

0018-9510/84/0700-1945$01.00

1984 IEEE

1946 Line Section Model

In the prerequisite course (Power Transmission Lines) the students have been introduced to the calculation of transmission line impedance. In that course the text Elements of Power Systems Analysis is used. 3] Assumptions are made in that text in order to reduce the complexity of the calculation of line inductance. The two fundamental assumptions are:

Vag
Vbg
Iv

- VagI
-

Zaa
=

zI ab
Zb

ac

zt

an
x

I1a

Vb,g
V v
c
I

Zba
ZI ca

Zb'c Zb'n
cc cn zt zt

cg
-

cb
zt

Zna
L

nb

nc

nn

LI

II ici
n

I bI

(1)

In a grounded neutral system:

V

1. 2.

Balanced three-phase currents Complete transposition of conductors

ng -V, ng

(2)

With these two assumptions a simple equation is derived for phase inductance which includes the self inductance and the effect of the mutual inductance with the other two phases. For balanced three-phase system analysis this is a viable method and the one most commonly used in industry for the determination of impedance values to use in power flow studies of transmission systems. However, a distribution system does not lend itself to either of the two assumptions. Because of the dominance of single-phase loads the assumption of balanced three-phase currents is not justified. Distribution lines are never transposed nor can it be assumed that the conductor configuration is an equilateral triangle. When the two assumptions are thrown out it is necessary to introduce the students to a more sophisticated and accurate method of calculating line impedances. The method presently being taught is the application of Carson's equations to a three-phase grounded four wire wye system. [4 ] Carson's equations are not derived in total, but it is possible to develop them in form based upon the derivation of inductance as developed in the previous course. This development is an example of what was previously described as a nice feature of the course. The students are required to reviewthe derivation of inductance and then to see that method expanded to a very accurate, though computationally difficult, method of calculating line impedances.

By substituting Equation 2 into Equation 1, the 4 x 4 matrix can be reduced to a 3 x 3 matrix as shown in Equation 3.

aa' Vbb?

ab ac a ag -aa Vbg - Vbg = Zba Zbb Zbc 9

Ib
cj

(3)

V LcciL'cg

CT

ca

cb

cc

The relationship between the Carson equation impedances of Equation 1 and the final impedances of Equation 3 is given by:

. Zij
Z

Z.
=

zij

in nnj
t I

Z,
I

(4)

nn

Figure 2 represents the final model of a three-phase, four wire grounded wye line as defined by Equation 3.

Carson's equations allow for the computation of conductor self impedance and the mutual impedances between conductors for any number of conductors above ground. For a three-phase, four wire grounded wye system a simple model is shown in Figure 1.

x \ % N \ k \ \ \

% -,

9I

Fig. 2.
a

Final Three-Phase Line Model

b~~~~~~~~a
b
C

_I
pIn
I
\

1_~
!.
\\ \\\s

The same methodology is used to model double-phase and single-phase line sections. In the case of a double-phase line involving phases a and c, for example, Carson's equations lead to a 3 x 3 matrix (a,c,n) which is then reduced to a 2 x 2 matrix'.ising Equation 4. The 2 x 2 matrix is then expanded to a 3 x 3 by placing zeros in the row and column of the missing phase (b in this example). The resulting matrix equation is shown as Equation 5.
aa

9\

ag

ag
=

aa

ac

Fig. 1.

Three-Phase, Four Wire Line Model

Vbb
cc

0
ca
cc

(5)
I

For this four wire system, Carson's equations will lead to the development of a 4 x 4 impedance matrix. This matrix is used to calculate conductor voltage drops as shown in Equation 1.

Vcg - cg

For a single-phase line, Carson's equations result in a 2 x 2 matrix. At this point the 2 x 2 could be reduced to a single term using Equation 4. However, research at the Public Service Company of New Mexico has shown that in a single-phase line nearly all of the phase current returns through the neutral conductor. [ 5] With the return current through the neutral, a different procedure is followed in reducing the 2 x 2-matrix.

1947
A single-phase line consisting of a phase a and grounded neutral n is shown in Figure 3.
1aa
I

zI

a'

an

z--II nf
g
\\ \N N N \ \\ N N N,N \N

J--

Ng \\\\\

Fig. 3.

Single Phase Line Section

Fig. 4. Load Model
In Figure 4 the three-phase load does not have to be balanced. That is Sa, Sb and Sc can be different values or even zero. In fact, double-phase and single-phase loads are modeled by setting the values of the complex power to zero for the non-exist phases. Figure 4 then represents the load model for all loads. In Figure 4 it is assumed that the complex powers of each phase are known and that the line-to-neutral voltages have been specified. (More on this later in the paper). With this known, the load currents are determined by:
I a = (S ai an a a)
- Z'

From Figure 3 the equation for voltage drop on phase a can be written as:
V

aa

=V

-Vain an

I - Z' I =Z' I + Zt I n - Z' na a nn n an aa a

(6)

Because all of the phase current returns in the neutral conductor:

In = - I a

(7)

Substituting Equation 7 into Equation 6 leads to:

V

aa ,

V an

-Va a'n
+ Z'

(Z' aa

nn

an

- Z'

na )

(8)
(9)

Ib
I c

(Sb/Vbn)*

(11)

Single-phase impedance is now defined:

= (S /V )* c cm

l4

+ zt z' nn aa

zs an

zt na

THE LADDER THEORY ITERATIVE METHOD

Time and space do not permit a detailed explanation of the iterative technique used in the program. A complete development and description can be found in Reference 2 . A general outline will have to suffice for this paper. The iterative technique used is an adaptation of the Ladder Network Theory of linear systems as taught in In that basic sophomore level circuit courses. method a linear ladder network, as shown in Figure 5, is analyzed by first assuming a voltage-at node n. Current, In, is calculated using Equation 12.
I n = Y n Vn

Because the program to be described assumes that all line sections are modeled by a 3 x 3 matrix, the single-phase line (phase a for example) is modeled by the matrix shown in Equation 10 where the missing phases (b and c) are represented by zeros in the appropriate rows and columns.

vaa'
bb|
=

va -ai V
=

z1

0 0

Ia
x

(10)

(12)

The end result is that all line sections, be they three-phase, double-phase or single-phase can use Figure 2 as the model. Equation 3 represents the general equation used to calculate voltage drop on a line segment. In cases of double-phase and single-phase lines, only the physically present phases will have current flow and voltage drops. This is accomplished by setting to zero the appropriate terms for the missing phases. Load Model

All loads are assumed to be fixed complex power P + jQ) by phase. It is further assumed that all three-phase loads are wye connected and all doublephase and single-phase loads are connected line to grounded neutral. The three-phase load model is shown in Figure 4.

(S

Fig. 5.

Linear Ladder Network

this is added to the previous line current so the next line section voltage drop can be calculated. This process is repeated until the voltage at the source (V1)

(n-l to n) can be calculated and in turn Vn.l is calcuated. Now the load current at n-l can be determined and

With In known, the voltage drop in the line section

1948 has been calculated. The calculated V1 is compared to the known input voltage (Es) and a "correction factor" (CF) derived by:
CF
=

S3
V3
-

E s/V

(13)
V1
S2
V4

Because the network is linear all calculated voltages and currents are multiplied by this correction factor

and

final solution is achieved.

A distribution system is non-linear because loads assumed to be constant complex power. The exact ladder theory method can not be applied. However, a modification to the procedure just described can be applied. Figure 6 shows a simple non-linear ladder network. Note the only difference between this and the linear network is that loads are constant complex power rather than constant admittance as before.
are

Fig. 7.

Main Feeder With A Tap

Fig. 6.

Non-Linear Ladder Network

a

To get started a voltage Vn is assumed at node n. On single-phase basis the load current In is calculated:
n=

(Sn/Vn)

(14)

In such a system V4 would first be assumed and V2 calculated. Because bus 2 is a "tap bus", the calculated V2 now becomes a "known voltage". The procedure at this point is to assume a voltage V3 at bus 3. As before a voltage V' is calculated. Now the same procedure is used as is used at the source bus. That is, V2 calculated is compared to the previously calculated V2. If the two are not within tolerance, a "delta voltage" is calculated and applied to V3 and a new value of V' is calculated. This continues until V2 and V2 are within tolerance. When V2 and V2 are within tolerance the solution continues back to the source bus as before. Once again the procedure of comparing the calculated voltage to the known source voltage is repeated. If these two voltages are not within tolerance, the "delta voltage" is applied to the most remote bus voltage and the total procedure is repeated. Each calculation of a source voltage is the completion of a "major" iteration while the calculation of a tap voltage is known as a "sub-iteration". As such, one major iteration may have several hundred sub-iterations; depending upon the complexity of the system.

As before, the system can be analyzed working back towards the source using Equation 14 to calculate all load currents. In time V1 will be calculated and compared to the known source voltage Es. Because the system is non-linear a simple correction factor can not be determined and then applied to all other calculated currents and voltages. Instead a "delta voltage" is determined by:
AV
=

The present version of RADFLOW is dimensioned for 400 buses. Input consists of line data cards which will contain the "from" and "to" bus numbers, the distance of the line and a configuration code number that is used by the computer to assign the proper impedance matrix to the line section. Bus data cards contain the complex powers by phase and the value of shunt capacitors, if present, at each bus.
program has been modified many times over the make it as user friendly as possible and to add additional system models. The most recent additions have been the inclusion of step voltage regulators and line transformers. The line transformer addition was installed because some systems being studied by the students will change voltage levels on long rural feeders. Each year brings more modifications that add to the complexity of the program but also adds to the usefulness in helping the students better understand the design and operation of a distribution system.

V1

(15)

The

The calculated delta voltage is applied to the assumed voltage at Vn using Equation 16.

years to

v.n
n

new

= V

n old

AV

(16)

With the new Vn the system is analyzed by working from node back to the source node. This procedure is continued until the calculated voltage V1 is within a specified tolerance of the known source voltage Es.

The basic procedure is expanded to a three-phase analysis by using Equation 11 to calculate phase load currents and using Equation 3 to calculate three-phase line voltage drops. Recall that in the development of Equations 3 and 11 that there were no restrictions placed on the lines and/or loads being balanced. Remember also that Equations 3 and 11 can be used to model double-phase and single-phase lines and loads.
A typical distribution system is rarely a "pure" ladder network. That is, typical distribution systems will have lateral taps off the "primary main" feeder and the lateral taps themselves will also be tapped. This complicates the solution technique using the ladder theory, but does not make it impossible. Figure 7 shows a section of a main feeder with a lateral tap.

RADFLOW APPLICATIONS
RADFLOW can be and is used in a variety of ways in the Distribution Systems course. Figure 8 shows a oneline diagram of a small 4.16 kV distribution system. This test system will be used to demonstrate some of the applications of RADFLOW.

1949
The 3 x 3 impedance matrices are first determined as outlined previously. RADFLOW requires that each conductor size and configuration be assigned a specific code. Table I gives this breakdown.

Conductor

Configuration Codes
Phases
Code 1 2 3 4 5 6 7 8

Table I

Conductor Size

336,400 1/0 1/0 1/0 1/0 1/0 1/0 1/0

a,n

a,b,c,n a,b,c,n

c,n a,c,n

b,n

a,b,n b,c,n

Using Table I explains the numbers underlined

on the one-line diagram of Figure 8. For example, the line section from bus 170 to bus 150 is a code 8 which means it is a double-phase line consisting of phases b and c and neutral.

After the data has been entered and the program run, attention is turned to the output. The output from RADFLOW is broken into four parts.
1.

2.

Fig. 8.

One-Line Diagram of Test System

In Figure 8 the distances between buses is given in feet. The values of the loads are given as either three-phase or single-phase kVA. Base Case
The first study performed will be referred to as the "base case". If this was a new system it would be necessary to select conductor sizes and configurations and to determine the t"phasing" of loads. For the purposes of this paper it will be assumed that this is an existing system with all of those decisions already made. However, in the Distribution Systems course leaving these decisions up to the students is an excellent exercise to help the students develop an understanding and appreciation of what is involved in laying
out a new system.

3.

4.

matrix for each line section. Voltage Profile -- This is a quick summary of the per-unit voltage magnitude and angle in degrees by phase for each bus and the value of the voltage unbalance at the bus. Voltage unbalance is defined as the ratio of negative sequence voltage to positive sequence voltage. Radial Load Flow Solution -- This is a typical output giving all essential bus information and the line flows into and out of the bus. The program has the option to either output line flow in kW and kVAR or current magnitude in amps and angle in degrees. Summary -- This is the final ouput and includes: a. Required number of iterations for solution b. Voltage tolerances used c. Total system input by phase d. Total system load by phase e. Total system losses by phase f. List of capacitors by bus and phase g. List of overloaded lines

Listing of line and load data. The line data output gives the per-unit 3 x 3 impedance

For the test system it will be primary main conductors are 336,400 laterals (three-phase, double-phase are 1/0 ACSR. The configuration of the poles are shown in Figure 9.
4.5' :-2. 1
t.-

Figure 10 shows the Voltage Profile for the first test system run, Figure 11 gives a sample of the Load Flow Solution output and Figure 12 shows the Summary page.
Armed with all of this output, what does the student do? The first concern is the voltage profile. Prior to making the first run some general guidelines have been established in class as to what will constitute an acceptable voltage profile. The two basic require-

assumed that the 26/7 ACSR and all and single-phase) the conductors on

ments are:
71

b
4' 41

f
n

cl~ L
n

1.

A~b~
nT
24'

_+
2.
24'

No more than a 3% voltage drop from the substation to the last bus. With a 1.0 per-unit voltage at the substation, -- this means that the minimum acceptable per-unit voltage is 0.97. Voltage unbalance can not exceed 2%.

24'

As can be seen from Figure 10, the voltage drop criteria has not been satisfied but the voltage unbalance is acceptable. In order to solve the voltage drop problem, shunt capacitors need to be installed. The voltage drops below 0.97 per-unit starting at bus 20, therefore, that

Three-Phase

Double-Phase

Single-Phase

Fig. 9.

Pole Configuration

1950
OL r A G E
P R OF I L

BUS

NAME

PHASE A AMNLE MAGNITUDE 1.000000

0.986461 0.984246 0.982054

MAGNITUDE
1.000000 0.990152 0.987878 0.985395

PHASE B ANGLE

PHASE C

MAGNITUDE
1.000000 0.988011
0.985703 0.982275

ANGLE
120.000000

UNA LA ACE
0.0001

500

SUBSTA
TAP

0.000000
-0.553486
-0.647968

350
110

TAP
BUS 140
BUS 40

-120.000000 -120.451416 -120.560425

-120.564896

119 .506439
119.385895

0.1t108
0.1 167

140

-0.675814

119.332184
119. 292465

0.1260
0.129 1
0.1094

40
80

0.978601

BUS

80

0.976762
0.980679

119.272461
119. 341202

90
30
220

BUS 90
BUS 30

-0.688453
-0.697955

0.984346 -120.589264

0.982477 0.984897

0.9825S9 -120.608994
0.982878 0.981336
0.981336

rAP BUS 170

0.986542 -120.634399

119.342407

170
150

-0.708954 -0.708954

0.984661 -120.661285
0.982556
-

0.984238 0.983517

119.333771
119.313278
119. 327545

BUS ISO BUS 120

END

120.692596

0.1 089 0.0675

0.1 138

120 100

0.981689
0

130
160
20 400 70 60

BUS 130 BUS 160 BUS 20

TIE

.976 170 .967078

.963414

-0.711455 -0.644149
-0.657873 -0.722773 -0.722773

0 .978887 -120.732315 0.985366 -120.644135

0.982984 -12 0.462830

0.981806 -120.472549

0.983828 .980199
0.973463 0.973463

119.461304
119. 446335

0.2229
0-2282 0.3329

0.974972
0

0.979125

PT.

0.967078
0

BUS 70
BUS 60

-0.725213 -0.725367
-0.800105

0.976994 -120.464584 0.976994 -120.464584 0.972654 -120.414749

0.970246 -120.407364
0.975347 -120.S15991

119 .430542

119.430542
119 .360168

0.332 9
0.3331
0.3442 0.4636

0.966481
0.965086

119.344803

10
200

BUS 10
BUS 200 BUS
180

0.960933 0.961655 0.958180 0.956170 0.954760

0.964234
0.973712

119.354752 119.469498
119.493271

I80
210

-0.83974t
-0.872335

0.973353 -120.566269

0.974083
0.974083

0.5419
0.6018

BUS 210
BUS 190

190
50

0.972865 -120.582382 0.971941 -120.592499

119.493271

BUS

50

-0.888023

Fig. 10.

Test System Voltage Profile

-_--_ --R A D I AL LO D PL O w

r T SO L U

N O

TEsr
BUS

rsrTM FOR TBH

PHASE A

1983 SUMMER POWER MEErING PAPER

PLEASE B
PHASE C

NAME
SUBSTA

(UNITS)

500

P.U. VOLTAGE: LOCAL LOAD: STATIC CAP: TO BUS 350:

P.U.

1.000007
0.000

0.000019 0.000
-29.476

0.000

1.000004 0.000 240.752 0.990152 0.000 240.752 -105.976 135.064

275.133
0.986461 0.000

-119-9996 0.000 0.000 -149.660

1.000000 0.000

252.409
0.988011 0.000 252.409 106.107 146.569

120-0000 0.000 0.000 90.368

119.5064 0.000 0.000 90.368 93.466 88.126

(MAG ANG) (EM XVAR)

EVAR) IAMP MG)

350

TAP

VOLTAGE:

-0.553486
0.000

-120.4514
0.000 0.000

SHUNT CAP.:
FROM BUS 500: TO Bus 110: To HUS 130:

LOCAL LOAD:

0.000

(AG ANG) (EM KVAR)

1

EVAR)

275.133 95.453

179.871

-29.476 -26.540 -31.034

-149.660 -146.499
-120.5604
0.000

-152.140

(AMP DEG (AMP DEGC (AMP DEG)

110

TAP

P.U. VOLTAGE:

0.984246
0.000

-0.647968

0.987878
0.000 105.976 42.315 63.661

0.985703
0.000

LOCAL LOAD: SUNT CAP.: FROM BUS 350: 140

BUS 140 P.U. VOLTAGE:

0-000

0.000

0.000

119.3859 0.000

TO BUS 140: TO BUS 220:

95.453 42.426 53.027

-26.540 -26.524 -26.552

-146.499 -146.429 -146.545

106.107

63.780 42.327

0.000 93.466 93.457

(NAG A1G) (Of EVAR)

EVAR)

93.478

(AYP DEG) (AMP DEG) (AMP DEG) {(AG TV )

1

0.982054
45.000
42 .426
2 1.228

-0.675814
21 *794 0-000
-26.524 -26.530

0.98S395
45*000
42.315

-120.5649
21 o794 0.000
-146.429 -146.451

0.982275
45.000 63.780 42.586 0.000

119.3322
21.794 0.000
93.457 93.441 0.000

LOCAL LOAD: SHUT. CAP.:

FROM BUS 110: TO BUS 40: To BUS 90:

KVAR)

KVAR)

21.188

(AMP DEG) AMP DEG)

t(AP DEG)

Fig. 11.

Test System Sample Load Flow Solution

__

__SUMMARY__
REQUIRED)

ITERATIONS REQUZRED

( 267 SUB-ITERATIONS FOR CONVERGENCE VOLTAGE TOLERANCE: REAL: 0.000010

ZMAG: 0.000010
ScW

KVAR
325.157

KVA
578.234 606.229
639.713 564.819 589.781 660.812

PP
0.870563 0.868975

TOTAL SYSTEM INPUT:

PHASE A: PHASE B: PHASE C:

PHASE

502.471 526.945 493.599 516.099

561.099

575.279

286.142 299.739

0.869217
0.877110

rorAL SYSTEM LOAD:

A: PHASE B: PHASE C: A:

307.248

274.556 285.453
0.000
0.000

0.873906
0.875069

TOTAL SiUNT CAP.:

PHASE

PHASE B: PHASE C: PHASE B: PHASE C:

PHASE

0.000 0.000
14.179

0.000

0.000

0.000 0.000

0.000

0.000000

0.000000 0.000000
0.620729 0.607963 0.604690

TOTAL SYSTEM LOSSES:

A:

10.846

8.872

17.910

11586 14.286

22.843 14.593 17.936

Fig. 12.

Test System Summary

1951
is a logical location to install a three-phase shunt capacitor bank. The next question is what size? From previous courses the students have been taught that a good starting point is to install capacitor banks such that the total reactive'power flow into the bus is supplied by the capacitor bank. The output at bus 20 (not shown here) shows that the complex power-flow into bus is 315.9 kW and 189.0 kVAR. 20 from bus 130 on phhse Since capacitors are normally rated in increments of 25 kVAR, a 200 kVAR per phase capacitor bank will be installed at bus 20.
a

voltage unbalance is 0.942 at bus 210 and total system

shows that the minimum voltage is 0.9403 at bus 50,

losses are 77.90 kW. The summary shows that no lines are overloaded. This study indicates that the system can experience such a growth rate and the only serious problem will be the low voltages which can be corrected by the additionof shunt capacitors or, perhaps, the installation of a voltage regulator.

Voltage Regulators
Step voltage regulators are work horses on most distribution feeders. Regulators are found in the substations in the form of load tap changing (LTC) transformers or separate three-phase or single-phase units. Step regulators often are found out in the system. This is particularly true on long rural feeders.
A step voltage regulator is a dynamic device in that the amount of "bucking" or boosting" of voltage is a function of the load. The intelligence for the step regulator is provided by the line drop compensator and the associated control circuit. It is far beyond the scope of this paper to develop the theory and operation of step voltage regulators. However, in the Distribution Systems course quite a few lectures are presented starting with a review of auto-trdnsformers and concluding with the theory of the determination of R and X settings for the line drop compensator. RADFLOW allows for the modeling of up to seven step-regulators. These regulators may be in a series along the main feeder and/or placed on laterals. Required input data for each regulator consists of:

When the program is run with the 200 kVAR per phase capacitor bank all voltages are above 0.97 except for the phase a voltage from bus 180 down to bus 50. When a 50 kVAR capacitor bank is installed at bus 50 on phase a, the voltage at bus 50 becomes 0.971 per-unit.

The system with the capacitor banks installed at buses 20 and 50 constitutes what is now referred to as the "final base case". It is interesting to compare the summary of losses for the initial run and this final base case. Table II gives this breakdown:
Table II Power Loss Comparison
Phase a Phase b Phase c Total

Initial 14.18 kW 8.87 kW

10.85 kw

33.90 kW

Final 11.01 7.30 8.96 27.27

kW kW kW kW

Obviously there is a lesson to be learned in observing how the losses by phase vary and how the total losses are reduced by nearly 20% with the installation of the capacitor banks. A nice assignment at this time is to get the students involved with the economics and determine the pay-back period for the capacitor costs based upon the loss savings. Contingency Study
Once the final base case is established it is interesting to determine how the system will'perform should the substation be out of service. In such an instance, it would be necessary to feed the system through a tie point to an adjacent feeder. In the test system such a tie point exists at Bus 400.

1. 2. 3.

4. 5. 6.

Regulator location (bus number) Potential and current transformer ratios Location of the "load center" (bus number) Voltage level desired at the load center Bandwidth setting Option desired Option 1: Load center voltage is held constant. RADFLOW adjusts taps and calculates the required R and X settings for the compensator. Option 2: R and X settings are specified. RADFLOW adjusts taps and calculates. the resulting load center voltage.

RADFLOW has the option to re-define the source point once the base case has been established. In this case the system is run again only with bus 400 defined as the source point. An analysis of the output shows the lowest voltage to be 0.9615 per-unit at bus 120, maximun voltage unbalance is 0.423% at bus 210 and total system losses to be 33.61 kW. This would be an acceptable operating condition for a contingency operation. It is interesting that the maximum voltage unbalance is reduced from the 0.552% value at bus 210 in the final base case.
Load Growth

In the previous study of the test system it was found, in the load growth study, low voltages occurred starting at bus 130 and continuing to the end of the various laterals. For demonstration purposes three single-phase step regulators are installed at bus 130. Bus 70 is defined as the "load center". Option 1 is specified and the voltage level set to 122 volts (1.0167 per unit) with a 2 volt bandwidth.

IDuTPur
rEsr SYSTEM POR

rTE

1983 SUEMMER PRWER MEErTIN

PAPEa

US_

Another feature of the program is the ability to automatically increase the system loads. At input time each bus is defined to.be in an area. The program allows for a maximum of ten areas. Once the base case has been run, the user has the option to specify a percentage growth rate for each area and the number of The program years such a growth rate is anticipated. will increase the-loads and the run the study.
For the test system three areas were defined and, for simplicity, it was assumed each area would grow at a rate of 5% for ten years. Output for the study

PROM

TO

CoLPE sARo YMPEDANCE IIN VOLTS) ESiSTASNCE REACTACE

0.000
0.000

130 4130

UR rUiMMAs VOLTAGE IVOLTS rREANSPONMER LEVEL EANGWIO7 POrENrIAL CUARENr OPTZON 100. 70 2400. 1 122. 2.

CE.NTR

LOAD

&US

PROM

tO

vA_c RArJxG-E
A

-jrA LoJADimGC0
A
C

COMPENSArtOR OPE'RAr Nc IMPEDANCE SIN VOLTS) RESISTANCE REACTANCE

1.053

POSITION
A 0

rAP

130 4130

250. 250.

250.

001.

448.

489.

0.561

10

10

Fig. 13. Voltage Regulator Summary Report

The "Voltage Regulator Summary" report shown in Figure 13 gives a summary of the input data and the

1952
operating conditions of the regulators consisting of the kVA load for each phase and the tap position of each phase regulator. Included are the R and X settings in volts for the line drop compensator. The voltage profile summary for this study shows the regulator output voltage on phase a is 1.0315 per-unit and the voltage at bus 70 on phase a is 1.0155 per-unit or 121.87 volts on the regulator base.
Extreme Unbalanced Loading
A distribution system with extreme unbalanced loading will display some unusual operating characteristics. To demonstrate two seemingly strange phenomena, a 5000 ft. 1/0 three-phase line is connected between bus 100 and a new bus 111 in the test system. At bus 111 a severely unbalanced load of 500 kVA on phase a, and 50 kVA on phases b and c is connected. Power factor is assumed to be 0.9 lagging. Table III summarizes the two end bus voltages.

Not only is the power loss by phase wrong, but the total three-phase power loss is wrong. This is another example of how the students can be reviewed to the fact that I2R only works in very special cases.

Admittedly, the severe unbalance placed on the test system is not very realistic. However, many times in the past the voltage rise and the negative power loss have been observed when studying actual distribution systems.
CONCLUSIONS
RADFLOW has proven to be an invaluable tool in teaching the Distribution Systems course. In using the program the students are forced to review material from previous courses in order to develop the required data for the program. This is one of the goals of the Normal and abnormal course as mentioned previously. operating conditions can be demonstrated without requiring the students to spend hours doing hand calculations or requiring the instructor to derive underivable equations.

Table III Bus Voltages for Unbalanced Loading

Phase-a
Bus 100 Bus 111

0.9404 /-2.5 0.7797 /-5.6 17.7% drop

Phase-b 0.9981 /-120.9 1.0127 /-121.2 1.5% rise

Phase-c

0.9819 /119.6 0.9638 /120.6 1.8% drop

As is expected, phase a has experienced a severe voltage drop. The unexpected is the voltage rise on phase b. Phase c experiences a normal voltage drop. The voltage rise phenomena is interesting and is correct. The reason for the rise is the manner in which the mutual coupling between phases occurs. A very good exercise for the students is to require them to sketch the phasor diagram to help explain how the voltage rise occurs.

Various versions of the program are presently being used by a number of utility companies and an engineering consulting firm. It has been gratifying to know that many times the program has predicted strange operating conditions that have been confirmed by actual field tests. [7 ] A greatly modified version of RADFLOW has recently been developed under a contract to a computer company that provides computer services to electric cooperatives in New Mexico, Arizona and Colorado. [6 ] That program allows a cooperative to run studies based upon actual billing data. Again results of some of these studies have been verified by field tests.
As time goes on, RADFLOW will continue to change in an effort to more closely model actual equipment and operating conditions of a distribution feeder. Without RADFLOW, the teaching of the Distribution Systems course would be boring to the students and impossible for the instructor. REFERENCES

The second strange condition displayed by the severe unbalanced loading is the power flow over the line from bus 100 to bus 111. A summary of the real power flows is given in Table IV.

W. H. Kersting and S. A. Seeker, "A Program to Study the Effects of Mutual Coupling and UnbalPhase-a Phase-b anced Loading on a Distribution System," IEEE Conference Paper C.75 047-6. Presented at the f527.760 kW 44.236 kW Leaving Bus 100 Winter Power Meeting, New York, January 1975. Arriving at Bus 111 d550.000 kW 45.000 kW 12] W. H. Kersting and D. L. Mendive, "An Application of Ladder Network Theory to the Solution of ThreePower Loss 1.234 kW 77.760 kW -0.764 kW Phase Radial Load-Flow Problems," IEEE Conference Total Loss = 78.231 kI Paper. Presented at the Winter Power Meeting, 'New York, January 1976. As might be expected, phase a hasaratherlarg power Elements of Power System loss([3] W. D. Stevenson. loss (14.7%), phase c hasanormal 1982. power lo 4th ed. New York: but phase b has a nega [4] J. R. Carson, "Wave Propagation in Overhead Wires stu t bright young engineer Bell System Technical Journal, and Ground theVo.5196 once power loss has to be iLn error.

Table IV ReaUl rower Flows

[1]

Phase-c 46.234 kW 45.000 kW

accurate modeling of tthe mutual coupling has surfaced an unusual actual conclition. Power has been transferred from.phase a tc phase b without the benefit of a solid connection in mtich the same manner that power is transferred across a ttransformer.
One last observati Lon on the power loss. Most students firmly belieste that power loss can always be determined by taking I E2R each phase. When this is for done for the unbalancead condition the following losses are determined:

sss(-1.7) (2.7Analysis, Ortivenpoe lnoss (-1.7%).gany Return," onerror However, knowsheronceagativhe again
[5]

McGraw-Hill,

tion Power Flow Program," Master of Science Technical Report, New Mexico State University, 1982. [7] R. W. Fischer, "Voltage Unbalance on Three-Phase Distribution Systems," Southeastern Electric E ~~~~~~~Exchange, 1977 Annual Conference.

[6] T. A. Carlson, "Development of a Radial Distribu-

Io.5 96 R. E. McCotter, "The Accurate Determination of Distribution Line Impedances," Master of Science Thesis, New Mexico State University, 1982.

Phase a: Phase b: Phase c: Total

75.612 0.448 0.495 76.555

kW kW kW kW