and Multi-Neutral
Fangxing
Li
ECPE Department Virginia Tech Blacksburg, VA 24060 Abstractdistribution conductors considered. This paper addresses impedance
where the engineer wishes to assume that all of the return current flows through the neutral conductors. Figure 2 provides Figure 2 illustrates conductors numbered a schematic view of the conductors of Fig. 1. 13 current paths consisting of six phase 1-6, six concentric neutrals numbered 7-12,
networks. Two sets of paralleled, three-phase with concentric neutrals and also a separate neutral are An approximate method to calculate the impedance the impedance for is to calculate for one three-phase
this configuration
set of conductors, and then to divide this result by two. This paper compares this common practice with a more exact calculation that considers conductors. significant. impedances. the mutual coupling between two sets of three-phase Errors in the current practice are shown to be Comparisons are made for both phase and sequence Also, a common configuration of parallel conductors
and the separate neutral numbered 13. As indicated in Fig. 2, at each manhole like phase conductors are tied together, such as conductor 1 and 4 are tied together. Also, at each manhole all neutrals, Counting conductors 7-13, are tied together and grounded. the earth return, there are actually 14 current paths.
of different types is studied. It is shown that the approximate method may have up to 30/0 errors in the positive sequence impedance and power flows, Keywords: impedance matrix, distribution concentric neutral conductor, parallel circuits.
789
10
11
12
networks,
I. INTRODUCTION
networks old.
in many location
cities of
States are of
relatively
Due to increased load growth and also the load a number into re-design of these networks.
geographical
As part of evaluating there-design, power flow analysis is required. The results of the analysis depend upon the accuracy of the impedances used in the model. In distribution parallel, phases ~ impedance where networks a circuit as many as ten circuits consists Current in parallel practice (where may be run in making up the Figure Separate Two Circuits with of three-conductors
13
Cables
in
B, and C. of n circuits
are the In the next section Carsons equations the Modified is considered. methods common to the impedance calculation
same) by calculating the result by n. This paper will configuration conductors a circuit. with
This results in a 13x13 impedance in Fig. 1. The Kron and Neutral After the calculation using
shown
in Fig.
sets of three-phase in two adjacent sets is referred to as placed in a duct be analyzed using a that will
concentric
are placed
to equivalent method,
round ducts. Each set of three-phase conductor In addition, below the circuits. in this paper. The impedance common practice. This is the configuration for these circuits will
to Fig. 1. Section III presents the Approximate IV presents errors in the Approximate method.
1
be calculated will
be referred to
as the Approximate method. More exact calculations of the impedance of the two circuits will then be performed. This method will be referred to as the Exact method. Error comparisons between the Approximate and Exact methods are made. Both calculations use the modified Carsons equations. In deriving equivalent phase and sequence impedance matrices for both the Exact and Approximate methods, two different matrix reduction approaches are used, which are the Kron and the Neutral Return Current (NRC) reduction methods [1]. The Kron method should be used where the engineer expects significant earth return currents exist, such as where physical deterioration of the neutral has occurred. The Neutral Return Current method should be used Figure
\
7 , 8 N 13 -
/ 1 . . .
Shown
in Figure
898
II. EXACT
CALCULATION Equations (1) and (2) are used to calculate primitive matrix, impedance matrix. also symmetric, the elements Inverting of the 13x 13 symmetric the this matrix
The Exact method is performed Step 1: Use the Modified element multiplies to current j. [1,2] Step 2: neutral return currentj elements of the 13x13 impedance
steps: impedance
Carsons toobtain
to calculate
gives an admittance
Reduce
return
current.
reduce These
both Kronand
Neutral
Return Current
reduction
)
methods tothe4x4 impedance matrix obtained 3x3 equivalent phase impedance matrices. Step4: Transform the 3x3 phase impedance matrices. in Step 3 to sequence impedance
(3)
A. Primitive
Impedance Matrix
of the self-impedance Applying and mutualthe Modified
B. Admittance
=( fi+O.0953
Matrix
(3) will
Reduction:
+JO.12134
Inti
[1+7934021(Qme)()
now be reduced
Zi
=0.0953
+J0,12134
where
r,= conductor
resistance
[1
!I
be performed
by adding appropriate
ln~+7.93402
(Qlmde)
(2)
use of voltage drops that are equal in paralleled From Fig. 2 we have
la=l, +[4 +15
i i and j. for in
lb =[2
Note that the earth return path has already been accounted
IC =13+16 13
to be determined
ln = ~~7 k
inoneoffour
AVa . A~ . AV4 AVb . AV2 = AV5 AVC . AV3 = AV6 vn=Av7=Av8=L AV13
Making
Y aa ba
ab bb cb
ac
an ]
[Ava
A Vb
AVC
bc cc n.
Ybn Y cn
Figure
of
Two
Phase
Conductors
with
(12)
Y Ca
nb ln [Yna
nn]lAvn
conductor
Dq =
Y aa =q,
+Y,4+Y4\+r44
L
For a phase conductor neutral, Do
Yab = Yba = Y12 + 35 +Y42 45 YGC= YCO= ~3 + 36 + Y43 + Y46
.=
(4) For a concentric neutral to an adjacent concentric neutral,
DY=L
Ybb = Y22 + Y25 + Y52 + Y55
where
concentric neutral strands L. = center-to-center distance between phase conductors c = number of concentric neutral strands.
Yfjfj
899
13 Y= an ma k~7(yl,k+y4,k )
It should be noted that the above reductions only apply to wye connected systems. Both reductions result in a 3x3 equivalent phase impedance matrix which is symmetric, given by
Yb = Yflb =
:7
(~2,k +y5,k )
z; = z,
Ycn = Ync = & ~3,k+y6,k)
[1
z, z,
Zm (16)
z.
z,
Zm Zm z,
Ynn =
$ g(Yk,,)
for all three phases,
Applying gives
the symmetrical
components
transformation
to (16)
Ya6 = YaC= Yba = YbC= YCa= Ych because the in Fig. I is symmetric
Zooo Z;=oz+
[1o The notation 0
o
z.
(17)
be used in Section IV
C. Impedance Matrix
Rewriting
Reduction:
(12) in impedance
(13)
In calculating the impedance for the configuration shown in Fig. 1, the Approximate method assumes that there is no coupling between the two sets of conductors in the separate ducts. Hence, this practice only considers one set of conductors and the independent neutral return, i.e., conductors 1,2,3, 7, 8, 9, and 13 in
Also, Having
matrix
shown
Fig. 1. The steps used in the Approximate Step 1: Use the Modified elements to currentj. element multiplies [1,2] Carsons matrix,
are: the
we can now use two different an equivalent the Kron reduction. current return reduction The Kron itself flows and the
reduction Return
methods Current
to calculate
impedance
divides current
Return Current
Step 2: Reduce the neutral return currents of conductors 7, 8, 9, and 13 to an equivalent neutral return current. These reductions result in a 4x4 impedance matrix. Step 3: Apply both Kron and Neutral Return Current reduction methods to the 4x4 impedance matrix obtained in Step 2 to obtain 3x3 equivalent phase impedance matrices. Step 4: Transform The matrix the 3x3 phase impedance matrices. by 2 to get the final of Step 4 is divided matrices obtained in Step 3 to sequence impedance Step 5: result.
Applying these assumptions results in the following transformation for Kron reduction
(14) where Z;= (ij) element in equivalent element 3x3 impedance matrix.
Similarly,
the matrix
transformation
Return Current method is given by This appears to be a good engineering approximation because the two circuits are in parallel. However, the next section shows that this approximation may result in significant impedance errors, compared with the Exact method presented in Section II.
(15)
1 Sequence (f2/1000ft) R+
0.1752
Impedance
Results
with
Kron
Reduction
??o
Approximate x+
0.0982
Method
(Q/1000ft) x+
0.0983
Error R+
0.00
Xo
0.1762
RO
0.5409
Xo
0.2222
R+
0.1752
RO
-7.21
Xo
26.11
x+
0.10
AA AA AA
900
Table 2
Sequence
Results with Neutral Approximate x+ 0.0982 0.0943 0.0896 0.0853 RO 0.5615 0.4287 0.2884 0.2308 xl)
Return Method
Current (Q/1000ft)
(NRC) Reduction
0/0Error Rt) -5.52 -4.07 -2.44 -1.87 X(l 28.52 24.56 16.43 12.77 R+ -0.06 0.08 0.00 0.14 x+
Exact Method RO 350 MCM 500 MCM 750 MCM 1000 MCM AA AA AA AA 0.5943 0.4469 0.2956 0.2352 Xo 0.1420 0.1075 0.0767 0.0658
0.00
0.00 0.00 0.00
Table 3 Conductor Type GMR of Phase Conductor (inch) 350 MCM 750 MCM 1000 MCM 4/0 Cu AA AA AA 1 2 3 4 Independent Neutral 0.262 0.312 0.385 0.445 0.200
Conductor
Configuration Number Neutral ) Strands 14 12 12 12 .of Resistance of Concentric Neutral (Q/mile 0.9821 0.7339 0.4646 0.3696 -) GlvfR of Concentric Neutral (inch)
Concentric
0.590
0.665 0.770 0.845 --
500 MCM AA
IV.
COMPARING
APPROXIMATE METHOD
AND
EXACT
The errors in the zero sequence quantities significant. The zero sequence reactance ranging fiom approximately 15/0 for the almost 29/0 for the smallest conductor.
shown in tables 1-2 are has the largest errors, largest conductor to
Considering the impedance of the conductor system shown in Fig. 1, results from the Approximate method presented in Section III are to be compared with the results from the Exact method be used presented in Section II. in the comparisons. Tables concentric 1-2 present the comparisons for four representative neutral conductors with an independent neutral. Results approach is used are presented in Table Return Current approach. The notation of ( 16) and (17) will
error in X. for both the Kron and Neutral Return Current methods, where the conductor type is shown in Table 3. In both cases larger conductors result in smaller errors. But even for the largest conductor, 1000 MCM AA, the errors are significant. approximated low, whereas X. is always app~oximated RO is always high.
Table 3 presents parameters for the conductors of tables 1-2. The last conductor in Table 3 is the independent neutral return used for each of the calculations [3]. The value shown in the Conductor Type column of Table 3 is used to identifi the conductors. From, tables 1 and 2 it may be noted that there is essentially no error in the positive been anticipated sequence quantities. This result could have from the theory of balanced three-phase circuits. Error in Zero
Sequence
Kron --*--
Figure 5 Percentage Error of XO Approximate Calculation (NRC) Reduction Both Kron and Neutral Return Current Figure 6 plots ~
for
Conductor
Type
and Neutral Return Current reduction methods. Because the curves are so close together the plots appear to lie on top of one another in Fig. 6. Likewise, Fig. 7 plots X. in ohms for the Exact method for both the Kron and Neutral Return Current methods. This plot Calculation Reduction for presents a range of values for XO which fall between assumptions and the Neutral Return Current assumptions. the Kron
Figure 4 Percentage Error of RO Approximate Both Kron and Neutral Return Current (NRC)
901
1600+j900
kVA
Figure
8 Voltage
Drop
Study
for Parallel
Circuits
with
Different
Cable Types
NOW let us consider voltage drops predicted by impedances of Table 4 with the Exact and Approximate methods with Kron reduction. Figure 6 Results for ~ Yeutral Return Current by Exact Method (NRC) Reductions for Both Kron and The system to be considered is shown in Fig. 8. The feeder is 3,000 1600kW+900kVAR the feeder. I
Zero Sequence Reactance
feet long and a customer load of for each phase is connected at the end point of
The results of the load flow are shown in Table 5. The voltage at the end point of the feeder is 109.6V with the actual impedance, while it is 105.6V with the approximate impedance. That is, the voltage drops are 10.4V and 14.4V, respectively. error in this case is 38 .6/0. Table 5 Power Exact Approx. Flow Results I (A) 725 753 for a Feeder in Figure AV (V) 10.4 14.4 8 The voltage drop
v (v)
109.6 105.6
The above results show that the Exact method allows the feeder to carry method. method. Figure Neutral 7 Results Return for XO by Exact Current (NRC) method for Both Kron and Approximate Reductions heavier Another method loads than that predicted aspect to be by the Approximate here is that the That is, the capacity of the system is larger with the Exact considered a voltage limits. may predict that is out of limits,
be within
V.
PARALLEL
CONDUCTORS TYPES
OF DIFFERENT
VI.
CONCLUSIONS
types
of
conductors
may
be placed
in
the
same CU 500
This paper has compared approximate impedance used by utilities in distribution networks with calculations. Both errors in impedances observations and voltage been shown to be significant the comparisons
q
For instance, a set of three-phase 350 MCM is placed in one duct, while a set of three-phase
method.
MCM AA conductors is placed in a second duct. Also, a 4/0 Cu conductor as an independent neutral return is placed in a third duct. In this configuration, neglecting the mutual coupling causes considerable error. Calculation results for the impedances are shown in Table 4.
the following
may be made: for balanced power are of the same errors errors become are
method is satisfactory
invoIved
method method
would would
the errors
A combination
of different
types of conductors
may cause
]NRC
In the above table, the Exact method is to calculate the full impedance matrix and then crunch it to 3x3 with Kron or NRC reduction. The Approximate method is to calculate the impedance for two conductors separately, and then calculate the overall impedance by considering them as parallel impedances.
errors in the positive sequence impedance with the Approximate method. Hence, errors may be produced even in balanced power flow studies.
q
Voltage
method predict
may be system
unacceptable.
q
902
This work has shown that more accurate impedance power systems. flow calculations are needed in complex,
VII. REFERENCE
[l] Power Distribution 6-PWR, 1992. [2]] Wave Propagation Planning, IEEE Publication Wires with 92 EHO 361Return, Vol. 5,
in Overhead
Ground
Reference
VIII.
Fangxing
BIOGRAPHIES
in 1994 and 1997 in Department of Tech.
engineering from Southeast University, China, respectively. He is currently a PhD candidate Electrical Robert software applied engineering operations. Al Sargent is a Senior Engineer and Computer Broadwater engineering software Engineering at Virginia
(M 1971) is a Professor in power systems and at Virginia engineering works distribution Tech where he teaches courses in and large-scale software design, and in the area computer-aided system analysis,
development.
He received his
903