The Modeling and Analysis of Parallel

W. H. Kersting, Life Fellow IEEE

Distribution Lines

 IS1  IR1 
Abstract— It is fairly common in a distribution system to +  v1  _
+ +
find instances where distribution lines are “physically”
 I1
parallel. The parallel combination may have both  VS1  Y1 
2
 Y1 
2
 VR1
distribution lines constructed on the same pole or the two    
_
lines may run in parallel on separate poles but on the same  ISC1  IRC1
right-of-way. For example, two different feeders leaving a
substation may share a common pole or right-of-way before
they branch out to their own service area. It is also possible  IS2   IR2 
that two feeders may converge and run in parallel until +  v2  _
+ +
again they branch out into their own services areas. The  Y2   I2   Y2 
lines could also be underground circuits sharing a common  VS 2   2 2  VR2 
   
trench. In all of the cases the question becomes how should  ISC 2  IRC2  _
the parallel lines be modeled and analyzed?

The purpose of this paper is to develop the models for the Figure 1 – Equivalent Pi Circuits
parallel lines and demonstrate how they can be analyzed
using the new models. In Figure 2 the definition of the matrices are as
follows:
Index Terms – Distribution systems, parallel lines,
overhead lines, underground lines, models, analysis
 ISn  = Sending end line currents in line n
I. INTRODUCTION  IRn  = Receiving end line currents in line n
This paper develops models of overhead and underground  VSn  = Sending end LN voltages for line n
distribution lines that are physically and/or electrically
parallel. The parallel lines are modeled utilizing the abcd  VRn  = Receiving end LN voltages for line n
matrices and the A and B matrices that are developed in  ISCn  = Sending end charging current in line n
Reference [1] for standard components of a distribution
system. The models are used to analyze the effects of mutual  IRCn  = Receiving end charging current in line n
coupling (inductive and capacitive) between the parallel lines.  vn  = Series voltage drop in line n
Examples for various configurations of overhead and
underground lines are included in the paper.
In order to compute the series voltage drops and the
charging currents, it is necessary to compute the “phase
II.EQUIVALENT PI CIRCUITS impedance” matrix and “shunt admittance” matrix for each
The equivalent Pi circuits for two parallel lines are line. These matrices must take into account the close
shown in Figure 1. coupling with the adjacent parallel lines. Carson’s equations
are used to compute the “primitive” impedance matrix
followed by the Kron reduction to reduce the phase
impedance matrix to a 6x6 matrix. This procedure can be
found in Reference [1]. The computation of the shunt
admittance matrix is similar to the computation of the phase
impedance matrix. In this case the primitive admittance
matrix for overhead lines incorporates the potential
coefficient matrix for overhead lines followed by the Kron
reduction. Underground phase admittance matrices are
computed directly as outlined in Reference [1].

III. OVERHEAD LINE PHASE IMPEDANCE AND

SHUNT ADMITTANCE MATRICES

Manuscript received December 10, 2004. This work was supported in part
by Milsoft Utility Solutions, Inc. Two overhead lines physically parallel on a common
W. H. Kersting is a retired professor of Electrical Engineering at New pole are shown in Figure 2.
Mexico State University. He currently is serving as a consultant to Milsoft
Utility Solutions, Inc.

.
 2.5' 4.5'
a b c The development of the shunt admittance matrix is similar
Line 1 to the steps taken to create the phase impedance matrix. The
2.0' 3.0'
numbering of the conductors must be the same as was used in
Line 2 a b c
6.0'
developing the phase impedance matrix. To develop the
shunt admittance matrix for overhead lines it is necessary to
know the distance from each conductor to ground and it will
be necessary to know the radius in ft. for each conductor.
n
The first step is to create the primitive potential coefficient
matrix. This will be an ncond x ncond matrix where ncond is
30.0' the total number of phase and ground conductors. For the
lines of Figure 2 ncond will be 7, for two lines each with its
own grounded neutral ncond will be 8.

The elements of the primitive potential coefficient matrix are

Figure 2 – Parallel Lines on a Common Pole given by:

The first step in the computation of the primitive S

impedance matrix for the lines in Figure 2 is to number the P ln ii
�ii = 11.17689 �
RDi
phase and neutral conductors according to the following: mile/μF (3)
Sij
�ij = 11.17689 �
P ln
Position 1 2 3 4 5 6 7-
Dij
Line- 1-a 1-b 1-c 2-a 2-b 2-c neutral
Phase
where: Sii distance in ft. from a conductor to its image
With the phases numbered the 7x7 primitive impedance below ground
matrix for one mile can be computed using the modified Sij = distance in ft. from a conductor to the image of an
Carson’s equations [1]. It should be pointed out that if the adjacent conductor
two parallel lines are on different poles, most likely each pole Dij = distance in ft. between two overhead conductors
will have a grounded neutral conductor. In this case there RDi = radius in ft. of conductor i
will be 8 positions and position 8 will correspond to the
neutral on line 2. An 8x8 primitive impedance matrix will be The last one or two rows and columns of the primitive
developed for this case. The Kron reduction [1] will reduce potential coefficient matrix are eliminated by using Kron
the matrix to a 6x6 phase impedance matrix. With reference reduction. The resulting voltage equation is:
to Figure 2, the “phase impedance matrix” in Ohms/mile is:
V 1ag �
�
� � �P11aa P11ab P11ac P12 aa P12ab P12 ac ��q1a �
�z11aa z11ab z11ac z12aa z12ab z12ac � V 1bg � �
�z11
� P11 P11bb P11bc P12ba P12bb P12bc ��
��q1b �
�
� ba z11bb z11bc z12ba z12bb z12bc ��
� � � ba
V 1cg � �P11ca
� P11cb P11cc P12ca P12cb P12cc ��q1c �
�z11ca z11cb z11cc z12ca z12cb z12cc � � =� � �
��
�
z phase � V 2 � P 21 P 21ab P 21ac P 22aa P 22ab P 22ac ��q 2a �
� �= �z 21aa z 21ab z 21ac z 22aa z 22ab
� (1)
z 22ac � � ag � � aa
� V2 � �
� P 21ba P 21bb P 21bc P 22ba P 22bb P 22bc ��q 2b �
�z 21ba z 21bb z 21bc z 22ba z 22bb z 22bc � � bg � � P 21ca P 21cb P 21cc P 22ca P 22cb
�� �
P 22cc ��q 2c �
� � �
V 2cg � �
�z 21ca z 21cb z 21cc z 22ca z 22cb z 22cc � � �
(4)
When Equation 1 is partitioned between the third and
fourth rows and columns the line voltage drops for one mile In short hand form Equation 4 is:
of lines are given by:
 VLG  =  P  �
 q (5)
� v1 � � z11  z12 �� I1 �
 v =  z  �
 I  = �v 2 �= �z 21 z 22 �� � � (2)
�  � �    �� I 2 � The shunt capacitance matrix is determined by:

 q  =  P  -1 �
�va �
Where:  vn  = �vb �
�  VLG  =  C  �
 VLG  (6)
� series voltage drops in line n
�
�c �
v � The resulting capacitance matrix is partitioned between the
third and fourth rows and columns.
�Ia �
 I n  = �Ib �
�
� phase currents in line n
� C11  C12 �
 C  =  P  -1 = �C 21 C 22 � (7)
�
�c �
I � �   �
The shunt admittance matrix in partitioned form is given by: d - ds
R = od ft. (10)
24
 y11
�
10-6 = �
 y12 �
 y =  C �
jw � S (8)
 y 21
�  y 22 �
�
The equivalent resistance of a concentric neutral is:

r
where: w = 2 �
p �frequency rcn = s /mile (11)
k
The charging currents for one mile of line at the receiving
ends of the lines are given by: The Kron reduction is used to reduce the primitive
impedance matrix down to a 6 x 6 phase impedance matrix.
 IRC1 �= � y11
�  y12  �� VR1 � The series voltage drops for each line is then given by
� � A (9) Equation 2.
 IRC 2 �
�
�
� � y 21   �� VR 2 �
y 22
��
�
The phase shunt admittance matrix for the parallel
A similar matrix equation will compute the charging currents concentric neutral lines is straight forward. The concentric
at the sending end of the lines. neutrals shield adjacent cables from the electro-static lines so
that there is no capacitive coupling between cables. Cross-
Linked Polyethlyene is a very popular insulation material. If
III. CONCENTRIC NEUTRAL LINE PHASE AND SHUNT the minimum value of relative permittivity is assumed (2.3),
ADMITTANCE MATRICES the equation for the shunt admittance of the concentric
neutral cable is given by[1]:
Two concentric neutral cable lines in parallel are shown in
77.3619
Figure 3. yii = 0 + j
Rb 1 k � RDs S/mile (12)
ln - ln
RDc k Rb
D1-13
D1-3
D1-2 D2-3 D3-13 Where: RDc = radius in ft. of the center phase conductor
7 8 9
RDs = radius in ft. of a neutral strand
13
1 2 3 The phase admittance matrix is a diagonal matrix with
D2-13 each diagonal term computed according to Equation 12. The
Dnm
charging currents are given by Equation 9.
D4-14
D4-6 IV. COMPUTATION OF THE MATRICES
D4-5 D5-6 D6-14
10 11 12 Reference [1] develops the abcd and A and B matrices for
14
three phase distribution overhead and underground lines.
4 5
The matrices developed are for individual lines, that is, not in
6
parallel with other lines. The derivation of the matrices for
D5-14 parallel lines follows exactly the procedure in the reference.
For that reason, the derivation of the matrices for parallel
Figure 3 – Parallel Concentric Neutral Cable Lines lines will not be presented in this paper. Instead, the results
from Reference [1] will be expanded to include parallel lines.
In Figure 3 note the numbering of the center phase The equations and definitions for the single lines from
conductors, the concentric neutrals and the additional Reference [1] are in terms of the sending end voltages and
neutrals for each line. For this configuration applying currents as a function of the receiving end voltages and
Carson’s equations will result in a 14 x 14 primitive currents. The defining equations for sending end voltages
impedance matrix. Before applying Carson’s equations, the and currents are:
equivalent GMR and resistance of the concentric neutrals for
each cable are computed as [1]:  VS  =  a  �
 VR  +  b  �
 IR 
(13)
k Rbk -1
GMRcn = k GMRs �� ft. (9)
 IS  =  c �
 VR  +  d  �
 IR 
Where: Rb = radius of a circle passing through the In Equation 13 the matrices are defined as:
center of the concentric neutral strands.
k = number of strands
GMRs =GMR of a strand
 1 VS IR1 VR
 a =  U  + �
 Z �
Y  IS1
2 IS IR
 b =  Z 
1 (14)
 c =  Y  + �
Y  �
Z�
Y 
4 IS2 IR2
1
 d  = U  + � Y  �
Z
2 Figure 4 – Electrically Parallel Lines

 Z  = 3x3 Phase Impedance Matrix In the ladder iterative process the receiving end voltages
will either have been assumed or computed and the total
Where:  Y  = 3x3 Phase Admittance Matrix phase currents [IR] will be known. With [VR] and [VS]
 U  = 3x3 Unity Matrix common to both lines, the first step must be to determine how
much of the total current [IR] flows on each line. Since the
lines are electrically parallel, the voltage part of Equation 15
For two three-phase lines in parallel the phase
can be modified to reflect this condition:
impedance, shunt admittance and unity matrices will be 6x6
so that the abcd matrices will also be 6x6. Equation 13 is
used to model the parallel lines with the understanding that  VS  � � a11
�  a12  �� VR  �+ � b11  b12  �� IR1 �
� �= � � �
all of the matrices are of the order 6. In order to maintain the  VS  � � a21
�  22  ��  � � 21  22  �� IR 2 �
a
��
VR
��
b b
��
�
identity of each line, The voltage and current equations for
the parallel lines are represented in partitioned form with the
(18)
partitions between rows 3 and 4 and columns 3 and 4. In
partitioned form, the equations for the parallel lines are:
The current in line 2 as a function of the total current and
the current in line 1 is given by:
 VS1 �= � a11
�  a12  ��
�
 VR1 �+ � b11  b12  ���
 IR1 �
�
 VS 2 �
� ��
�
 a21   a22  �� VR 2 � � b21   b22  �� IR 2 �
�� �� ��
�  IR 2 =  IR  -  IR1 (19)

� IS1 �= � c11  c12  ��

�
 VR1 �+ � d11  d12  ���
 IR1 � Substitute Equation 19 into Equation 18:
�
 IS 2 �
� � �
�
 c21   c22  �� VR 2 � � d21   d22  �� IR2  �
�� �� ��
�
(15)  VS  � � a11
�  a12  �� VR  �+ � b11  b12  ��  IR1 �
� �= � � �
 VS  � � a21
�  a22  �� VR  � � b21  b22  �� IR  - [ IR1�
�� �� ��
�
In the ladder iterative procedure, Equations 15 are used for
the “forward” sweep. (20)

In Reference [1] the equation for the “backward” sweep is: Since the sending end voltages are equal, Equation 20 is
modified to reflect this:
 VR  =  A �
 VS  -  B  �
 IR  (16)

 A =  a  -1
(  a11 +  a12 ) � VR  + (  b11 -  b12 ) � IR1 +  b12 � IR 
Where: = (  a 21 +  a 22 ) �
 VR  + (  b21 -  b22 ) �  IR1 +  b22 � IR 
 B  =  a  -1 �
 b
(21)
As before, Equation 16 can be expanded for the case of two
parallel lines. Equation 16 in partitioned form for the Collect terms in Equation 21:
parallel lines is:

 VR1 �= � A11
�  A12  ��
�
 VS1 �- � B11  B12  ��
�
 IR1 � (  a11 +  a12 -  a 21 -  a 22 ) �
 VR  + (  b12 -  b22 ) �
 IR 
�
 VR 2 � �
 A21  22  ��  � � 21  22  �� IR 2 �
A
��
VS 2
��
B B
��
� �� �
= (  b 21 -  b 22  -  b11 +  b12 ) �
 IR1
(17)

V. SPECIAL CASE OF ELECTRICALLY PARALLEL LINES (22)

This is the special case where both the sending end and the Equation 22 is in the form of:
receiving end of the lines are common as shown in Figure 3.
 Aa �
 VR  +  Bb �
 IR  =  Cb �
 IR1 (23)
  Aa  =  a11 +  a12 -  a 21 -  a 22 1.476 + j 5.560 1.519 + j 4.848 1.496 + j3.931�
�
�
1.519 + j 4.848 1.545 + j5.336 1.521 + j 4.323 �
where:  Bb  =  b12 -  b 22 (24)  b12 = � �
1.496 + j 3.931 1.521 + j 4.323 1.507 + j5.460 �
�
 Cc  =  b21 -  b22 -  b11 +  b12 � �

Equation 23 is used to solve for the receiving end current in  b21 =  b12
line 1:
�5.616 + j11.211 1.560 + j5.017 1.535 + j3.849 �
 IR1 =  Cc  -1 �(  Aa  �  IR  )
 VR  +  Bb  � (25)  b22 = �1.560 + j5.017 5.706 + j10.913 1.580 + j 4.237 �
�
�
�1.535 + j 3.849 1.580 + j 4.237 5.655 + j11.082 �
� �
Equation 19 is used to compute the receiving end current in � 0 0 0� �
�
line 2.
 c11 =  c12 =  c 21 = �c 22 = �0 0 0 ��
� � �
�
With the two receiving end line currents known, Equations � �0 0 0 ��
� � ��
15 and 17 are used until convergence of the voltages occurs.
When applying these equations it is good to observe that the
sending end voltages using Equation 15 will be the same for  d11 =  a11
both lines. The backward sweep using Equation 17 will also
show the receiving end voltages to be the same.
 d12 =  a12
 d 21 =  a 21
VII. OVERHEAD EXAMPLES
 d 22 =  a 22
Two parallel distribution lines on the same pole are
constructed as shown in Figure 2. The lines are in parallel for In order to demonstrate the effect of the parallel lines, the
10 miles. Data for the conductors are: following data is used:

Line #1 Data: Balanced sending end voltages of 24.9 kV LL

Conductors: 336,400 26/7 ACSR Unbalanced loading:
GMR = 0.0244 ft., R = 0.306 Ω/mile, Dia = 0.721 Load #1: Phase a – 1440 kVA at 0.95 lag
inch Phase b – 1150 kVA at 0.9 lag
Line #1 Data: Phase c – 1720 kVA at 0.85 lag
Conductors: 250,000 AA Load #2: Phase a – 1300 kVA at 0.9 lag
GMR = 0.0171 ft., R = 0.410 Ω/mile, Dia = 0.574 Phase b – 1720 kVA t 0.85 lag
inch Phase c – 1370 kVA t 0.95 lag
Neutral Conductor Data :
Conductor: 4/0 6/1 ACSR For the case of two lines in parallel on the same pole,
GMR = 0.00814 ft., R=0.592 Ω/mile, Dia = 0.563 two cases were studied:
inch
Case 1: Lines are physically parallel with a common
Using the methods presented in the paper, the abcd sending end node and separate receiving end nodes.
matrices in partitioned form are computed to be: Case 2: Lines are electrically parallel with common
sending and receiving end nodes and the same total load.
0.9998 + j 0
� 0 0 � With reference to Figure 1, the results of these cases are
� �
 a11 =  a22 = � 0 0.9998 + j 0 0 � summarized in Table 1 where voltages are in per-unit based
� 0 0 0.9998 + j 0 �
� upon 24.9 kV line-to-line.
�

�0 0 0� Table 1
 a12 =  a 21 = �0 0 0 �
�
� Parallel Lines on the Same Pole
�
�0 0 0��
Case 1 Case 2
VR1-a 0.9639 0.9595
�4.502 + j11.029 1.464 + j 5.334 1.452 + j 4.126 �
�1.464 + j 5.334 4.548 + j10.873 1.475 + j 4.584 � VR1-b 0.0684 0.9555
  �
b11 = � VR1-c 0.9364 0.9408
�
�1.452 + j 4.126 1.475 + j 4.584 4.523 + j10.956 �
� VR2-a 0.9557 0.9595
VR2-b 0.9400 0.9555
VR2-c 0.9448 0.9408
IS1-a 103.47 106.30
 IS1-b 82.27 111.29 �6.268 + j 4.128 1.539 + j0.467 0.914 - j0.153�
IS1-c 127.77 121.08
IR1-a 103.91 106.82  b11 = �1.539 + j0.467 5.962 + j3.756 1.196 + j0.331�
�
�
IR1-b 82.61 111.71 �
�0.914 - j 0.153 1.196 + j0.331 5.623 + j3.710 �
�
IR1-c 127.77 121.38
IC1-a 0.871 0.869 1.700 + j 0.708 1.372 + j0.187 0.914 - j0.153�
�
�
1.372 + j 0.187 1.393 + j 0.348 1.033 + j 0.041 �
IC1-b 0.970 0.963  b12 = � �
IC1-c 0.729 0.732
�
0.914
� - j 0.153 1.033 + j 0.041 1.058 + j 0.298 �
�
IS2-a 94.12 91.15
IS2-b 126.89 96.82  b21 =  b12
IS2-c 100.70 105.60
IR2-a 94.62 91.58
IR2-b 127.28 97.15 �6.218 + j 4.299 1.511 + j 0.632 0.976 + j0.008 �
IR2-c 100.87 105.81  b22 = �1.511 + j 0.632 5.930 + j3.930 1.181 + j0.468 �
�
�
IC2-a 0.8188 0.8209
IC2-b 0.9033 0.9115
�
�0.976 + j 0.008 1.181 + j 0.468 5.597 + j3.831��
IC2-c 0.6979 0.6956 �j 0.001 0 0 �
In Table 1 IC1 and IC2 represent the total charging  c11 =  c 22 = �
� 0 j 0.001 0 � �
current in the line. It should be noted that the effect of shunt �
� 0 0 j 0.001�
�
capacitance on the lines is very small. Very little error is
made in neglecting the shunt capacitance.
�0 0 0�
VIII. UNDERGROUND CABLE EXAMPLES  c12 =  c21 = �0 0 0 �
�
�
Two parallel underground concentric neutral cable lines are �
�0 0 0 �
�
shown in Figure 3. For these lines the concentric neutral
cables are 250,000 CM, AL 1/3 neutral with the following
data:
 d11 =  a11
 d12 =  a12
Phase conductor: 250,000 CM AL
GMR = 0.0171 ft., R = 0.41 Ω/mile, Dia = 0.567 inch  d 21 =  a 21
Concentric neutral: 13 - #14 Copper strands
GMR = 0.00208 ft., R = 13.8722 Ω/mile, Dia = 0.0641  d 22 =  a22
inch
Grounded neutral: 4/0 Copper For the case of two underground lines in parallel, two cases
GMR = 0.01579 ft., R = 0.303 Ω/mile, Dia 0.522 inch were studied:

For this configuration the primitive impedance matrix will Case 1: Lines are physically parallel with a common
be 14x14 and the Kron reduction will reduce it to a 6x6 phase sending end node and separate receiving end nodes.
impedance matrix. Because of the concentric neutrals on the Case 2: Lines are electrically parallel with common
cables, the 6x6 shunt admittance matrix is computed directly sending and receiving end nodes and the same total load.
using Equation 12. The abcd matrices for the 10 miles of
parallel lines are as follows: With reference to Figure 1, the results of these cases are
summarized in Table 2.
�.9980 + j.003 -.0002 - j.0007 .0001 + j 0005 �
Table 2
  �
a11 = �-.0002 - j.0007 .9982 + j 0.0029 -.0002 + j.0006 �
� Parallel Concentric Neutral Lines
�.0001 + j 0005
� -.0002 + j.0006 .9982 + j.0027 � �
Case 1 Case 2
�-.0003 + j.0008 -.0001 + j.0007 .0001 + j.0004 � VR1-a 0.9509 0.9524
� .0000 + j.0005 �
  �-.0001 + j.0007
a12 = -.0002 + j.0007 � VR1-b 0.9702 0.9612
VR1-c 0.9493 0.9561
�.0001 + j.0004
� .0000 + j.0005 -.0001 + j.0005�
�
VR2-a 0.9538 0.9524
 a 21 =  a12 VR2-b 0.9517 0.9612
VR2-c 0.9632 0.9561
�.9979 + j.0030 -.0003 + j.0007 .0000 + j.0005 � IS1-a 101.81 95.64
IS1-b 77.41 97.54
 a 22 = �
�-.0003 + j.0007 .9981 + j.0029 -.0002 + j.0006 �
� IS1-c 119.03 106.15
�.0000 + j.0005 -.0002 + j.0006 .9981 + j.0027 �
� � IR1-a 105.34 99.89
IR1-b 82.45 103.56
 IR1-c 126.03 111.44
IC1-a 13.547 13.56 W. H. Kersting (SM’64, F’89) was born in Santa Fe, NM.
IC1-b 13.692 13.62 He received the BSEE degree from New Mexico State
IC1-c 13.736 13.58 University, Las Cruces, and the MSEE degree from Illinois
IS2-a 89.98 95.44 Institute of Technology. He joined the faculty at New Mexico
IS2-b 119.07 97.85 State University in 1962 and served as Professor of Electrical
IS2-c 95.42 106.48 Engineering and Director of the Electric Utility Management
IR2-a 94.83 99.80 Program until his retirement in 2002. He is currently a
IR2-b 125.71 103.87 consultant for Milsoft Utility Solutions. He is also a partner
in WH Power Consultants, Las Cruces, NM.
IR2-c 98.94 111.80
IC2-a 13.5663 13.56
IC2-b 13.5532 13.62
IC2-c 13.6326 13.58

As in Table 1, IC1 and IC2 represent the total charging

current in the two lines. It is noted that with the underground
cable that the charging current now must be taken into
consideration.

IX. CONCLUSIONS

In this paper Carson’s equations and the Kron reduction

were used to compute the phase impedance matrix for parallel
overhead and underground lines. The shunt admittance for
overhead lines incorporated the use of the potential coefficient
matrix and the Kron reduction. For underground cables the
computation of the shunt admittance matrix is computed
directly. The phase impedance matrices and the shunt
admittance matrices are used to compute the generalized abcd
matrices used in the analysis of lines that are physically
parallel and a different set of matrices for the lines when they
are electrically parallel.
Examples of parallel overhead lines on one pole show that
the effect of the shunt capacitance is very small and can be
ignored. However, for parallel underground lines the effect of
the shunt capacitance is significant and must be included in
all calculations.

REFERENCE
1. W. H. Kersting, Distribution System Modeling and
Analysis, CRC Press, 2001.