Professional Documents
Culture Documents
2, Apnl1991 479
Abstract - Generating station electrical system analysis may involve GLOSSARY OF SYMBOLS
evaluation of generator leads and the station service transformer secon-
dary leads to carry increased load currents. These leads usually consist of
several, large single-conductor power cables per phase. Phase currents Source voltage - phase A, line-neutral, volts
may not divide equally among the conductors in each phase. This paper Self reactance, conductor A l . ohms
presents a matrix solution method that enables the engineer to determine
the phase current distribution among the individual conductors. Use of a Mutual reactance between conductor A1 and A2, ohms
commercial mathematics software package on a desktop MS-DOS per-
AC resistance of conductor at temperature Tc , ohms
sonal computer to solve the resulting matrices is illustrated. Results of
field measurements on a six-conductor-per-phase cable installation are Current in conductor A l , amperes
compared with calculation results. Application of the results to cable as-
sessment activities is discussed, including ampacity verification for in- Matrix of source voltages, volts
creased loads, cable insulation life assessment studies, and cable imped-
Matrix of conductor current, amperes
ance input into station service system analysis.
Matrix of impedances, ohms
INTRODUCTION
DC resistance of conductor at temperature Tc , ohms
An integral part of generating station electrical system analysis is the per foot, for Eqs. 10 and 12, expressed in wn/ft
evaluation of electrical equipment capability. This includes power cables Length of conductor A l , feet
used for generator leads and the secondary leads of station service trans-
formers and unit auxiliaries transformers. In most cases, multiple single- Conductor skin effect, unitless
conductor cables per phase are used for these large loads. Unless the
Conductor proximity effect, unitless
single conductors are arranged and connected in a specific configuration
(e.g., flat arrangement - ABCCBA), the phase current does not divide Shield eddy-current loss, unitless
equally among the individual conductors. Determining this division in
phase current is necessary to ensure cable ampacity is not exceeded on Shield circulating-current loss, unitless
any one conductor, especially in the case of the larger loads that may Inverse of Z matrix
result from the enhancement of the electrical system. For insulation life
assessment studies, the temperature of the insulation, which is a function Conductor temperature, OC
of the current in the conductor, is required.
Impedance of load, ohms per phase
In 1949, F. H. Buller presented an iterative solution technique to
solve for the reactance of each conductor and the resulting current divi- Constants used in Eqs. 10 and 12
sion [ 11. Subsequent work by A. Y. Wu confirmed that the Buller tech-
nique does not converge for all cable configurations [2]. Both Buller and Radius of conductor, inches
Wu present the current distribution for several configurations, though not
A stranding correction factor in Eq. 13
necessarily all that will be encountered in typical plant design.
This paper presents a matrix algebra method for solving the current Center-to-center distance between conductor A 1 and
distribution among phase conductors. For most problems the matrix size A2, inches
becomes too large to calculate the necessary inverse matrix by hand. Ap-
Spacing between conductors, used in Eq. 11, inches
plication of a commercial math software package to produce the matrices
and calculate the inverse matrices is described. Diameter of conductor, inches
Test data are presented for a six-cable-per-phase application, and
the results compared with calculated values. D~
Application of the results of the current distribution calculation in
several areas of work on cable assessments is discussed, including cable When multiple conductors per p h s e are used to supply a load, cur-
ampacity verification, insulation life assessment, and determination of cir- rent division among the conductors in each phase will be in inverse pro-
cuit impedance for use in short circuit and voltage profile studies. portion to their individual impedances. Figure 1 presents a typical dia-
gram of the connections to the load supplied by two conductors per
phase.
The cable impedance is the combination of the conductor's ac resis-
tance R U C ,self reactance X A ~ and
, mutual reactance X A I- AZ from the
other conductors. Mutual reactance is a function of the distance between
89 SM 6 4 1 - 2 PWRD A p a p e r recommended a n d approved the conductors. The voltage drop due to mutual reactance depends on
by t h e I E E E I n s u l a t e d Conductors Committee of t h e I E E E the magnitude and phase of the current in the other conductors. If the
Power E n g i n e e r i n g S o c i e t y f o r p r e s e n t a t i o n a t t h e I E E E / conductors are not arranged so that the mutual reactance of the conduc-
PES 1989 Summer Meeting, Long Beach, C a l i f o r n i a , J u l y 9 tor to the other phase conductors is equal to that of its companion con-
- 1 4 , 1 9 8 9 . Manuscript s u b m i t t e d J a n u a r y 25, 1989; ductor in the same phase, then a nonsymmetrical configuration will exist.
made a v a i l a b l e f o r p r i n t i n g June 7 , 1 9 8 9 . The load current then will not divide evenly, resulting in a current imbal-
ance among the phase conductors.
480
Phase A
-
Conductors
A. r - - - - - - - - -1
I
a
I
I
XL ~
where: all elements of matrix are expressed in SZ, if matrix E is in
volts.
Phase B I
Source IC
I
Personal computer (PC)-based software can be used to determine
R, I the inverse of the Z matrix. Z inverse Z -
times the E matrix provides
I the current in each conductor as follows:
I
XL 1
Phase C I I = Z-lE (6)
I
I
L---------! The development of terms in the Z matrix and in the E matrix are
Balanced
three-phase load discussed below.
234.5 + Tc
Rdc 63 Tc = (,,,., + 25) Rdc @ Z5"C (8)
The exact temperature of the conductors can not be known until the
conductor distribution among the conductors is determined, and the am-
E = Z I pacity of each conductor is solved. As a first-order approximation, the
temperature rating of the insulation should be used for all of the conduc-
tors. Successive iterations using conductor temperature based on calcu-
[]
lated current distribution are not necessary, as the resistance in large con-
ductors is small compared to the self reactance of the conductor.
Conductor skin effect Ycs is the phenomenon of unequal current
E = (3) distribution in a conductor that is caused by the pulsating magnetic field
set up by the current in the conductor itself. For a 60 H z system with
EC concentric or compact stranded conductors, Yc, is found as follows [3]:
ductors. Y c ~varies with conductor spacing s and conductor diameter E Voltaee Matrix
D, . For a 60 Hz system with concentric or compact stranded conductors,
The magnitude of eich b l p x n t of the E matrix is the line to neutral
(%r
YCP is found as follows [5]:
value of the source voltage. Balanced source voltages are assumed, which
Concentric round None, Tin, or Alloy 1.0 1.0 Ebase = Source voltage (line -neutral) v
Compact round None 1.0 0.6
7.5 in
Imuedance of Load
Measurements of the individual conductor currents as well as the tively. For the A2 conductor, the calculation predicted the conductor cur-
total phase current were taken at a generator output of 27.5 MVA, rent within 1.5 percent of actual measurements. The author recognizes
14.4 kV, as shown in Table 11. To determine the degree of accuracy that that correlation for other conductors is not as close. Use of as-built data
can be expected using the matrix solution method, the generator leads (which was unavailable) instead of design data would, in the author’s
were modeled. Balanced load conditions were used in the model, even opinion, improve the model.
though actual measurements show some phase imbalance existed.
The calculation also accounted for the expected difference in APPLICATION OF RESULTS
lengths of each conductor for the turns made in the cables to exit the
building to the unit station transformer. Table I11 presents the calculation The results of the current division calculation may be used to con-
results. firm the cable ampacity for cable insulation life assessment studies, and to
In Tables I1 and 111, conductor currents are given in amperes and determine the cable feeder impedance for input to voltage drop and fault
also expressed as percent of total phase current. Because of phase imbal- current studies of the electrical system.
ance, as indicated in measured phase current, the percent values are used In many cases the electrical systems or generator are being enhanced
to correlate measured values with modeling results. as part of life extension activities. This may require the feeder cables to
Calculation results correctly indicate the conductors with the maxi- carry a new higher load. The actual current in each phase conductor must
mum and minimum percent of total phase current: A2 and C4. respec- be compared directly against the cable ampacity. The maximum current
483
value can be determined directly from the I matrix. Discussion of cable CONCLUSION
ampacity development is outside the scope of this paper. The reader is
referred to Reference 7 for cable ampacity. If ramp is set equal to the Multiple single-conductor power cables per phase are used as gen-
ampacity of the single conductor cable, and max ( I - %) equal to the erator leads and as secondary leads of station service and unit auxiliaries
maximum conductor current split in percentage of the total, then the transformers in generating stations. Unless a specific cable arrangement
feeder ampacity is: exists, phase current will not divide equally among the conductors. The
matrix solution method presented herein for determining the expected
ramp current division is recommended. This method provides the information
Feeder ampacity = A
max ( I - % ) necessary to perform cable assessment activities, including cable ampacity
verification and insulation life assessment.
Cable insulation aging is accelerated by high insulation temperature
and by effects of moisture, ozone, and unequal electric-stress distribu-
tion. Once the cable ampacity is determined for the particular installa- ACKNOWLEDGMENT
tion, then the cable insulation temperature for the actual cable loading
can be determined for the life assessment as follows: The author would like to acknowledge K. Hicks for developing the
matrix solution method in 1985, Northern States Power for making the
test data available, and A. W. Goldman for his review and encourage-
ment of this paper.
REFERENCES
Tl, Tal = Conductor (insulation) temperature Power Cable Ampacities. New York: Republished by Institute
and ambient temperature for I’ of Electrical and Electronic Engineers, IPCEA, P-46-426
conditions, “C (ICEAOEEE S-135-1-62, S-135-2-62), 1962.
The conductor temperature Tc‘ is compared against the insulation Ethylene Propylene Rubber-Insulated Wire and Cable for the
temperature rating of the cable. Insulation life increases by a factor of 2 Transmission and Distribution of Electrical Energy. Washing-
with every decrease of 8 ” to 10°C in insulation temperature [6]. A rough ton, D.C.: National Electrical Manufacturers Association,
estimate of the possible life extension of the cable can then be made in ICEA S-68-516 (NEMA WC 8-1983), 1983.
concert with evaluation of the other aging mechanisms.
J. H. Neher and M. H. McGrath, “The Calculation of the
When the utilities system increases its available fault MVA system
Temperature Rise and Load Capability of Cable Systems.”
contribution, reanalysis of the station service switchgear rating may be
AIEE Transactions, Part I11 PAS, vol. 76, pp. 752-72,
necessary. As additional loads are added, a voltage study of the electrical
October 1957.
system also may be required. In both cases the cable impedance of the
feeder cable is required as input into these calculations. The cable imped- K. A. Petty, Power Plant Electrical Reference Series, Vol. 4 :
ance of each phase can be determined by dividing the voltage drop across Wire and Cable. Palo Alto, CA: Electric Power Research Insti-
the cables by the sum of the phase current of that phase. For most volt- tute, (EL-5036 Vol. 4), 1987.
age and fault calculations the average cable impedance is used. The end
of Appendix A illustrates the equations to determine the cable Mathsoft, User’s Manual - MATHCAD. Version 2.0.
impedance Cambridge, MA: Mathsoft, 1987.
APPENDIX A. Author-produced TemplatelResults for Current Imbalance Calculations
j := $I;
a := -0.5 + -. j
2 NOTE: NO FURTHER INPIPP DATA BEYOND THIS POINT
484
The following produces the E matrix THE FOLLOWING EQUATION EXPaESSES THE IDAD AS A CONSTANT IMPEDANCE
Zload := pf + sin(acos(pf)).j
N
nPH := - this is the number of conductors per phase Zload = 0.8,+ 0.6i per unit
3
PRODUCES THE DIAGONAL ELEMENTS FOR Z MATRIX: RESISTANCE AND SELF
q := 1 ..nm E := 1 A phase source voltage REACTANCE OF CONWCTOR, PER UNIT.
q
tempx := 1 ..N tempy := 1 ..N
q := (nPH + 1) ..2-nPH 2
E := a B phase source voltage b := 0
¶ tmpxI tmPY
k := 1 ..N
a := 0
t m p x I temPY
The Z matrix is the sum of "a" matrix, which is the mutual reactance,
"b" matrix, which is the Rac and the self reactance, and "c"matrix,
which is the Zload portion of the equations THE FOLIOWING PRODUCES Zload IMPEDANCE MATRIX:
The following produces a Position matrix by arranging the x-y
corrdinate of each conductor by phase. C := 0
tempx. tMlPY
a := P PX := x PY := y q := 1 ..nPH r := 1 ..m
1 l a l a
a := P PX := x PY := y c := Zload
2 2 a 2 a
a := P Px := x PY := Y q,r
3 3 a 3 a q := (nPH + 1) ..2.nPH r := (nPH + 1) ..2.nPH
a := P PX := x PY := y
4 4 a 4 a c := Zload
a := P m : = x PY:=y
5 5 a 5 a q,r
q := (2 nPH + 1) ..3.nPH r := (2 nPH + 1) ..3.nPH
a := P PX := x PY := y
6 6 a 6 a c := Zload
9,r
Z : = a + b + c
Zbase := - V
R per unit
Iload $;
-
I 1
0.351 0.227i
0.429 - 0.37i
-0.467 0.153i-
I = -0.443 0.223i-
0.014 + 0.551i
10.117 + 0.426il 1-0.5 + 0.866ij
D
P.q
:= lp - PXd2 + pp- PYA2
DISTANCE,CENTER TO
cENTER,BElwEEN
EACH CONDUCTOR,in.
IMAG
t.1
:=
I1t.ll
k := 1 ..N
D := 1 This makes is possible to add a mutual reactance
k.k matrix to the self reactance matrix.
10.4421
1000
1mpedPHB.Zbase -
L
= 0.008 + 0.028i ohms/lOOO ft
q := (2.nPH + 1) ..3,npH r := 3 . n p H
Vdrop := E
r
- Zload.>: I
¶
¶
Vdrop
ImpedWC := -
1000
1mpdHC~Zbdse~- = 0 . 0 0 5 + 0.029i ohms per phase
L
Discussion the cable system. How does the author account for the
effects of an unbalanced system?
THOMAS L. JONES, American Electric Power Service
Corporation, Columbus, Ohio: The need to consider The author is to be commended for his recognition
both the thermal and electrical circuits in analyzing of the need to integrate both the thermal and electri-
the ampacities of paralleled cables installed in trays cal circuits into the solution of the ampacity of a
is recognized by the author. A method which considers cable system. If the author would answer the questions
both circuits has been developed for cables installed posed, it would help to clarify his approach.
underground [l]. However, for cables in trays, a
thermal circuit model which considers the heat trans- 111 T. L. Jones, "The Calculation of Cable Parameters
fer between each cable and tray is needed, as well as Using Combined Thermal and Electrical Circuit
mutual thermal coupling. A l s o , for metallic trays, an Models," IEEE Transactions on Power Delivery,
electrical representation of the tray is needed to V o l . 4 , No. 3, July, 1989.
account for circulating currents, eddy currents, and
the distortion of the magnetic fields caused by the [2] T. L. Jones, "Neher-McGrath Revisited," Insulated
presence of a high permeability material. Is the Conductor Committee Minutes, 82nd Meeting,
author aware of thermal and electrical models that Columbus, Ohio, April 17-20, 1988.
take into account these factors?
The author mentions that using electrical resis- Manuscript received July 25, 1989.
tances calculated at the rated insulation temperature
is satisfactory from a cable impedance viewpoint.
However, when calculating conductor and shield losses,
this seems to be a very conservative assumption. Does I would like to thank Mr. Jones for his interesting and
the author adjust conductor and shield resistances for valuable contribution to the discussion of the paper. The thrust of the
their calculated operating temperatures when determin- paper was not meant to cover cable ampacity, but instead to present a
ing losses? simplified approach in determining current imbalance in single conductor
power cable for generating station applications and support it by test data.
The author provides a method to account for This has been done.
shield losses. The presence of shields that are In discussion with W. Rueger [l], he indicates that the approach
multigrounded can significantly alter the current presented in his paper [2] could be used to account for the effect of the
distribution in a paralleled cable grouping. For the tray on the distribution of the current. The tray would be modeled as a
example given in Figure 2, assuming equal cable series of conductors in the shape of the tray. In review of the field data
lengths, the following current distributions were presented in my paper (see Table II in the body of the paper), the effect of
calculated using the method in [l] assuming 1100 amps the tray does not appear to be significant enough to warrant the additional
per phase. The currents without parentheses are complexity associated with the calculation. In regards to the thermal
without shields and those in parentheses are with influence from the tray, typically the cables are spaced away from the
21-#16 copper wire shields. How does the author siderail of the tray a minimum of one cable diameter. This ensures that if
consider this in his calculations with multigrounded significant currents were induced in the tray the resultant heat from the
shields? tray would not significantly influence the phase conductors.
In calculating losses in the conductor and shields for ampacity consider-
Phase A Phase B Phase C ation, I agree that resistance of these components should be based on
Conductors Conductors Conductors actual temperature. However, for calculating the current distribution this
is not warranted, since the reactance is significantly greater than the
143 259 265 245 260 138 resistance.
(144) (244) (189) (282) (256) (146) It is recognized that the grounding of the shields at both ends, thus
permitting circulating currents in the shields, will produce a balancing
91 216 244 224 219 81 effect on the distribution of the currents in the phase conductors. Neglect-
(106) (216) (166) (261) (213) (101) ing shield circulating current in the model as presented does yield conser-
vative results, however, without significant loss of accuracy. I should
143 259 265 245 260 139 point out that the shields used in the medium voltage cables in generating
(144) (244) (189) (282) (256) (146) stations are typically one 20% overlapped 5 or IO mil copper tape, not
21-16 AWG copper wire shields, so that the degree of balancing would
The author defines the term I as the ampacity not be as great as you have shown, i.e., 6% reduction in imbalance.
of the single conductor which &pies the highest In my paper, I, would be based on the stated ampacity at rated
current. In underground systems, the hottest conduc- conductor temperature not necessarily actual conductor temperature. In
tor is not necessarily the one with the highest duct bank applications the hottest conductor may not be the conductor with
current [21. Has the author considered this possibil- the largest current imbalance. In those cases the approach outlined in your
ity? Also, it is not clear what ampacity is assigned paper might offer some benefits. In duct bank or direct burial applications,
to I Is it that of a single cable operating by I would calculate the current distribution as outlined in my paper based on
itsef%?'or that of a cable when all other cables have actual load conditions. I would then input these currents into an ampacity
the same current, or that of a cable when the other model based on reference [3], which can calculate the conductor tempera-
cables have currents in accordance with their current ture for unequally loaded conductors. The resulting conductor temperature
distribution? If it is the latter, how is this will be compared against the insulation temperature rating. No successive
distribution incorporated into the thermal model to iterations is necessary, except to arrive at the conductor temperature.
obtain the desired solution? I am not aware of thermal models which account for unequal loading of
cables in trays. As indicated above, the normal installation method assures
The electrical model that is used by the author little thermal interaction between the cables.
is valid for balanced three phase systems. Does the In generating stations, the large loads which require the use of multiple
author have a method to account for single phase conductors per phase are balanced three phase loads, e.g. generators,
cables or neutral conductors? Also, does the author transformers, and motors. Line-to-ground faults are temporary conditions,
consider unbalanced faults, such as a line-to-ground which do not impact the thermal design of the cable. Also, these large
fault, when designing and operating a system? For loads typically do not have neutral conductors routed with the phase
unbalanced systems, the zero sequence impedance plays conductors. It is for these reasons that effects of unbalanced systems have
a significant role in determining current flows, which not been included in the paper. I offer no method to calculate current
ultimately affect the ampacity and overall design of distribution for your application of unbalanced conditions.
481