IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO.

5, OCTOBER 2015 2303

Cable Sheath Loss Reduction Strategy Research

Based on the Coupled Line Model
Yong Lin and Zheng Xu, Member, IEEE

Abstract—Power cables are playing an increasingly important systems is even comparable to the loss in the cable cores, which
role in the power grid. The sheath loss is one of the main short- will add to the thermal problems and limit the ampacity of the
comings of the power cable which will cause thermal problems cable system [6]–[8].
and limit the cable's ampacity. We carry out a systematic re-
search on the sheath loss of cables with different arrangements, The sheath loss problem has drawn the interest of many re-
including single bonding, solid bonding, and cross bonding. For searchers since it was raised in the 1920s [9], [10]. Some strate-
cross-bonded cables, we study how the cross-bonding joints gies have been proposed to reduce the sheath loss, such as trans-
number and segment length deviation influence the sheath loss. posing or cable sheathes cross bonding [11]; raising sheath resis-
To reduce the sheath loss, this paper studies a promising measure tance [7], [12]; cable compensation [13]; adopting power-elec-
using connecting impedance in the cross-bonding joints, and
the characteristic of the connecting impedance is fully revealed. tronic equipment to reduce the operating frequency [14]; and
The computer efficient algorithm is established to fulfill the loss connecting the resistor and reactor in the cross-bonding joint
calculation task, which is based on the improved coupled line [15]–[17]. Transposing and cable compensation are two effec-
model. The key of the coupled line model is to use series expansion tive ways to limit sheath loss; however, they are limited by high
to calculate the admittance matrix without diagonalizing the cost and the space restrictions. The cost of power-electronic
complex propagation matrix of the cable line. The dimension of
the nodal equation is also greatly reduced. equipment is also unbearable. Raising the resistance may have
cost and implementation problems as well. Connecting the re-
Index Terms—Connecting reactor, coupled line model, loss re- sistor and reactor in the cross-bonding joint is a promising way
duction strategy, power cable, sheath loss.
and has already been used in a project in South Korea [16], [17].
The Hybrid Firefly Algorithm is used to optimize the arrange-
I. INTRODUCTION ment of reactors in [15]. However, apart from the deficiency in
the loss calculating method, the characteristic of the connected
resistor and reactor has not been fully explored in the existing

P OWER CABLE has developed quickly in recent decades

and there are more underground transmission systems
being considered when expanding city grids. Basically, there
literature. In this paper, we will carry out systematic research
on the loss reduction strategy and reveal the characteristic of
the connecting sheath resistor and reactor.
are three main reasons for the soar of power cables [1]–[3]: first, Basically, there are two main analytical methods to calculate
the rising environmental pressure limits the usage of overhead sheath loss apart from simulation and experiment methods.
transmission lines in cities; second, the need to transfer power 1) The simplified loss calculation method [15], [18], [19]: It is
over sea area is also creating a market for the submarine cables a very simple and effective way to assess sheath loss. How-
because overhead transmission line is not economical in that ever, this method is not accurate enough because it regards
case; third, the technological progress also made insulated the sheath current as constant along the cable. Actually, the
cables more competitive. sheath current varies along the cable line.
In a single-core cable, the sheath is adopted as a coat to pro- 2) The multiconductor method introduced in [1], [20], and
tect the core from ingress of moisture and possible mechanical [21] can calculate the sheath loss accurately if the nodal
damage. It also serves as an electrostatic shield and acts as a re- equation is built at an increment which is small enough.
turn path for fault current and capacitive charging currents [4]. However, there are still some problems in the conventional
When main current is flowing in the cable core, induced voltage multiconductor method: first, the nodal equation has a large
appears in the cable sheath. And if the cable sheath is grounded dimension and it is time consuming to solve if is small.
at both ends, the circulating current will flow in the sheath and Second, to obtain the admittance matrix of a cable line,
cause considerable sheath loss. The sheath loss of some cable we need to diagonalize the propagation matrix (
and are the admittance matrix and the impedance matrix
Manuscript received November 03, 2014; revised February 13, 2015; ac- per unit length.)
cepted March 15, 2015. Date of publication March 18, 2015; date of current However, there is no mathematical guarantee that the propaga-
version September 21, 2015. This work was supported by The National High
tion matrix can always be diagonalized [22], [23]. The diagonal-
Technology Research and Development Program of China under (863 Program)
(2011AA05A119). Paper no. TPWRD-01333-2014. izing procedure may fail when the frequency-dependent propa-
The authors are with Zhejiang University, Hangzhou, Zhejiang 310000, gation matrix has repeated eigenvalues. In this paper, efforts are
China (e-mail: [email protected]; [email protected]).
made to develop an accurate and computer-efficient algorithm.
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. The proposed algorithm is based on the modified coupled line
Digital Object Identifier 10.1109/TPWRD.2015.2414655 model. The coupled line model was first put forward in [22]

0885-8977 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
2304 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO. 5, OCTOBER 2015

and developed in [23], which use series expansion to avoid the

difficulty in diagonalizing the propagation matrix. In our algo-
rithm, the nodal equation is reduced in dimension because we
only need build the nodal equation at each cable segment (not
at every increment of a cable segment).
With the help of the proposed loss calculation algorithm, we
study and compare the sheath losses of cable systems with dif-
ferent arrangements, including transposed, cross bonded, two-
ends solid bonded, and single-end-bonded cable systems. Com-
paratively systematic research is carried on the proper sheath-
loss reduction strategies, which is helpful in practical applica-
tions. The method of installing sheath connecting impedance is
comprehensively studied, which proves to be effective to reduce
the loss in cable sheaths.
In this paper, we are only concerned about the induced
circulating current on the cable sheath. We also assume that
the impedance and admittance matrix per-unit length along the
cable line are as constant as [1], [15], [20], and [21] are.
The outline of this paper is as follows. Section II introduces
the coupled line model for cable lines and establishes the sheath
loss calculation algorithm. In Section III, we study the sheath
loss of cables with different cable arrangements and discuss
the method of installing sheath connecting impedance to reduce
sheath loss. The paper ends with a conclusion in Section IV.

II. COMPUTER-EFFICIENT ALGORITHM TO

CALCULATE SHEATH LOSS
A three-phase single-core cable system without armor is Fig. 1. Basic model of a three-phase (single-core) cable system. (a) Cable line
shown in Fig. 1 [1]. In the single-end bonded cable system, composed of cross-bonding joints and segments. (b) Cross-bonding
joint. (c) Transposing joint. (d) th cable segment.
the entire cable line has only one segment, and only single-end
sheaths of the entire cable line are bonded together and
grounded. In the solid-bonded cable system, the entire cable
line also has only one segment, but the two end sheathes of the A. Coupled Line Model for Cable Lines
entire cable line are bonded, respectively, and solidly grounded, Let , , , and represent the voltage and cur-
respectively. In the cross-bonded (or transposed) cable system, rent vector on the sending end and receiving end of the th cable
the entire cable line is composed of separate cable segment as shown in Fig. 1(d). is from 1 to . Let
segments and cross bonding (or transposing) joints, and the and represent the voltage and current vector at the location
two end sheathes of the entire cable line are bonded, respec- away from the sending end. For a small increment of length
tively, and solidly grounded, respectively. In Fig. 1, at location , the variation of voltage and current vector can
cross-bonding joints ( is a positive integer), such as 2 and 20 be expressed as
joints, can form complete cross-bonding sections. In other
words, three segments can form a complete cross-bonding sec-
tion, as shown in Fig. 1(a). In the following discussion, cables (1)
with cross-bonding joints are denoted by “ CBJ” for short.
Cross bonding and transposing joints are shown in Fig. 1(b) and (2)
(c), respectively. Fig. 1(d) shows the structure of the th cable
segment. There are six coupled parallel conductors. Three of where and are the impedance matrix and the admittance
them are cable cores which are numbered from 1 to 3 and the matrix of the coupled lines per unit length, respectively. Since
remaining three are cable sheathes which are numbered 4 to 6, the length of the cable segment is , we can solve the differential
respectively. (1) and (2) for , , , and with the boundary
The cable system composed of six coupled conductors is not condition at both ends: , ,
a totally symmetric multiconductor system, which cannot be , and . The relationship between , ,
studied in detail by a simplified equivalent circuit [20]. In other , and can be expressed as (3)[22]
words, we need to adopt the multiconductor method to calcu-
late the sheath circulating current. In this section, we establish
the computer-efficient algorithm based on the modified coupled
line model. Efforts are also made to reduce the dimension of the
(3)
nodal equation.
LIN AND XU: CABLE SHEATH LOSS REDUCTION STRATEGY RESEARCH BASED ON THE COUPLED LINE MODEL 2305

where , which is called the propagation matrix. Equation (7) can be transformed to
and are the self and mutual admittance matrix of
the cable segment, respectively. To avoid diagonalizing , the
coupled line model uses series expansion to calculate and
. The series expansions of hyperbolic function (9)
and are shown as

where

(4)

and are the coefficient matrices

which can be calculated form , and the Bernoulli number
(5) . The coefficient matrices are same for all cable segments of
any length because and are constant along the cable line.
When the coefficient matrices are calculated and stored at the
where is the Bernoulli number and the convergent domain
beginning, we only need to change the in (8) and (9) to obtain
is .
and for different cable segments.
and can be represented as
Now that we have obtained the admittance matrix of each
cable segment, the entire nodal equation for nodes on both ends
of each cable segment can be built after combining the admit-
tance matrix of cross-bonding joints and boundary conditions,
which has been clearly explained in [1], [20], and [21]. Solving
(6) the nodal equation, we can obtain the voltage and current on
both ends of each cable segment. Note that we build the nodal
equation for each cable segment rather than every small incre-
ment, so the dimension of the nodal equation is ,
which is much smaller than that of the conventional multicon-
ductor method. The current along the cable segment is calcu-
(7) lated by the method introduced in part B.

B. Division of Cable Segment and Sheath Loss Calculation

After the series expansion of the hyperbolic function,
and can be obtained by simple calculation steps. Since We have already obtained the current and voltage vectors on
cable lines in service are mostly within 60 km [1], much shorter the sending and receiving ends of each cable segment. To cal-
than overhead transmission lines, the problem of exceeding the culate the sheath loss, we also need to know the sheath current
convergent domain does not exist here. Thus we do not need along the cable segment. In principle, we need to divide the
to adopt the recursive method as [23] does. The and cable segment into small parts so that the current and voltage
series converge quickly and it is enough to calculate the first ten along each small part can be regarded as constant.
terms for computational accuracy [23]. Let , , , and represent the voltage and
There are many occasions where we need to calculate the self current vectors on both ends of the th division part of the th
and mutual admittance matrix of cable segments. The length of cable segment. is from 1 to and is the number of cross-
the cable segments may differ from one another; however, the bonding joints. is from 1 to and is the total division
matrix and are constant for all cable segments. So in order number of the cable segment. The current and voltage vectors
to improve the computational efficiency for the self and mutual have the relationship as shown
admittance matrices of various cable segments, the coefficient
matrices and are introduced.
Equation (6) can be transformed to (8)

(10)

(8) where is the length of the division part of the th cable

segment. and are the self and mutual admittance
where matrix of this cable part. and can be easily cal-
culated by (8) and (9) after substituting with . When we
divide the cable segment equally, is equal to and the
self and mutual admittance matrix of each part is the same.
2306 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO. 5, OCTOBER 2015

TABLE I
BASIC PARAMETERS OF THE UGC SYSTEM

Fig. 2. Sheath voltage of single-end bonded cable (SiB), two ends solid bonded
cable (SoB), and one sections cross-bonded cable (CB).

Solving (10), we can obtain (11), as shown at the bottom of

the page.
The matrix is the same for each part of the cable segment
and it needs to be calculated only once. The method to calcu-
late the voltage and current along the th cable segment can be
concluded as follows:

(12)

Fig. 3. Sheath current of single-end bonded cable (SiB), two ends solid-bonded
cable (SoB), and one sections cross-bonded cable (CB).
(13)

where is the total division number of each cable segment. bonding (denoted by CB) arrangements. Generally, the solid
and are the current and voltage vector at the sending bonded system has the smallest sheath voltage and the largest
end of the th cable segment, which we have already obtained in circulating current, where the maximum sheath voltage along
part A. Sheath voltage and current along the entire line can be the cable line is near 10 V and the sheath circulating current is
obtained after calculating the sheath voltage and current along larger than 1500 A, which is close to the current carried by the
each cable segment. cable core. The sheath current of the single-bonded system is
much less than that of the solid-bonded system; however, the
C. Test System and Illustration sheath voltage is too high, which is more than 2000 V as shown
An underground cable (UGC) system is chosen as a typical in Fig. 2, and far exceeds the safe limit. So the single-bonded
case to illustrate the calculation results [20]. The UGC system system is not suitable for the high-voltage (HV) transmission
operates at a frequency of 50 Hz. The sending-end source system either. The cross-bonding arrangement is an effective
voltage is 400 2.5 kV and the receiving-end source voltage is way to limit the sheath voltage and circulating current, which
400 kV, and the transmitted power is about 650 MW. The can be easily deduced from Fig. 2. However, it may be costly
impedances of the sources on two ends are to perform cross bonding because we need to cut the sheath off
for each phase. The other basic parameters of the cable system and cross bond them at the joints.
are shown in Table I. If the division part is small enough, the current along the di-
The proposed method can draw the sheath current and voltage vision part can be regarded as constant. Let represent the
curves along the cable line, which is helpful to understand the sheath current on the th cable of the th division part of the
cable arrangements. Each segment is divided into 100 parts here th segment, where is from 1 to 3 for three-phase cable. The
to display the curve in detail. Figs. 2 and 3 show the sheath three-phase sheath loss of the considered part can be expressed
voltage and current of the cable system with single-bonding as (14). There are segments in the entire cable line and
(denoted by SiB), solid-bonding (denoted by SoB), and cross- each cable segment is divided into division parts. The total

(11)
LIN AND XU: CABLE SHEATH LOSS REDUCTION STRATEGY RESEARCH BASED ON THE COUPLED LINE MODEL 2307

TABLE II
COMPARISON BETWEEN THREE METHODS

For incomplete cross-bonded cable systems, we find that the

relative error is within 0.5% when is less than 200 m. The
Fig. 4. Relationship between relative error and division length . sheath circulating current is composed of longitudinal (due to
inductive coupling) and transverse (due to capacitive coupling)
induced currents. The capacitive current causes the variation of
sheath loss can be calculated by summing the loss of all division circulating current along the cable line. Since the longitudinal
parts of each cross-bonding segment, which is shown in (15) induced current is larger in incompletely cross-bonded cable
systems while the capacitive current is the same, the total cir-
(14) culating current along the cable is much smoother. That is why
larger can be adopted here for the same accuracy demand.

(15) D. Computer-Efficient Algorithm for Sheath Loss Calculation

The algorithm for the calculation of cable sheath loss can be
concluded as follows:
Here, is the sheath resistance per length and is assumed con-
Step 1) Form impedance and admittance matrix per-unit
stant along the cable. The total cable length is , and the seg-
length, and , based on the basic cable parame-
ment length is for the cable with equal distance
ters. (There are well-proved methods available and
joints. Let represent the length of each division part and it can
we did not discuss it in detail in this paper.) Calcu-
be shown as
late and store the coefficient matrices
(16) and by , and Bernoulli numbers.
Step 2) Calculate the admittance matrix of each cable seg-
ment based on the coefficient matrices and segments
It is not hard to derive (17) after combining (15) and (16)
length according to (8) and (9).
(17) Step 3) Build the nodal equation and solve the voltage and
current vectors on both ends of each cable segment.
where represents the average of the square of sheath current Step 4) Calculate the sheath current along each cable seg-
along the cable line. ment according to (12) and (13).
Generally, the sheath loss can be calculated more accurately Step 5) Calculate the total sheath loss by (15).
when we divide the cable segment into smaller parts, which, To illustrate the efficiency and accuracy of the proposed
however, will also require more computational effort. The algorithm, a comparison is made between the proposed method
proper division length is discussed in this part. (denoted by “P”); the conventional multiconductor method
First, we need to make the accuracy demand for the loss cal- [20], [21] (denoted by “M”); and the simplified loss calculation
culation algorithm clear. Since the resistance of the cable sheath method which regards the sheath current to be constant [15]
may vary slightly with the working temperature and manufac- (denoted by “S”). In the method “P” and “M”, the division
turing deviation, the loss calculation method within 1% error length is 40 m for complete cross-bonded cable systems (14
limit can meet the accuracy demand in most cases [19]. The and 20 CBJ) and 200 m for incomplete cross-bonded cable
0.5% error limit is a more rigorous standard and can be used in systems (19 CBJ). The sparse matrix is used in method “P” as
more severe cases. recommended in [20] and [21].
We find that the sheath loss calculation results vary within We find that the calculation results of the proposed method
0.01% when is smaller than 0.1 m, so the sheath loss cal- and the conventional multiconductor method are in very good
culated by dividing the cable into 0.1-m parts is chosen as the agreement and the relative error between the two methods is
reference and the relative error is calculated based on it. Fig. 4 within 0.01% if they have the same division length. The relative
shows how the calculation error increases with for the test errors and time consumption of the three methods are shown in
system with four different completely cross-bonded arrange- Table II.
ments. It can be seen that the relative error is almost directly It is shown that the simplified loss calculation method has
proportional to . The four curves are in high proximity, which a much larger error compared with the other two methods,
means that the chosen for certain accuracy demand is not in- especially for completely cross-bonded cable systems. The ac-
fluenced by the CBJ number for completely cross-bonded ar- curacy of the proposed method is the same as the conventional
rangements. So the recommended division length for the 0.5% multiconductor method while the calculation time decreases
error limit is 40 m. dramatically.
2308 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO. 5, OCTOBER 2015

Fig. 6. and of cables with a different number of cross-bonding joints.

Fig. 5. and of single-end bonded cable (SiB), two ends solid-bonded

cable (SoB) and one section's cross-bonded cable (CB).

III. RESEARCH ON CABLE SHEATH LOSS

REDUCTION STRATEGY

A. Sheath Loss of Cables With Different Arrangements

Equation (17) shows that the total sheath loss is proportional
to the average of the square of the sheath current , the cable
sheath resistance per-unit length , and the total cable length Fig. 7. of the test system with different segment length deviation.
. We are not interested in or because they cannot be
changed easily. So we mainly focus on in the following dis-
cussion. The maximum current along the cable sheath is another lengths, which are denoted by , , and , respectively. , ,
concern for us, which is also one of the main factors that limit and are represented by the following equations when the test
the ampacity. In the following discussion, the maximum of the system has complete cross-bonding sections
square of the sheath current, denoted by , is also discussed.
Fig. 5 shows and of the test system with single end
bonded, two ends solid bonded, and one section's completely
cross-bonded arrangements. The sheath loss of the two ends where is the total length of the test system. is the deviation
solid-bonded arrangement is almost 20 times larger than that index of the segment length. We calculated the sheath losses
of the single-end bonded arrangement. And the of the two of the test system with changed from 0.2 to 0.2. The re-
ends solid-bonded arrangement is nearly 10 times larger than sults are shown in Fig. 7. It is clear that the sheath loss increases
that of the single-end bonded arrangement. The enormous heat quickly when becomes larger. In real projects, efforts should
produced by the sheath circulating current limits the usage of the be made to ensure the segment length constant. The cable com-
two ends solid-bonding arrangement in bulk power transmis- pensation strategy is an appropriate way to reduce . However,
sion. The and of the one section's completely cross- the cost and the restriction in space should also be taken into
bonded arrangement are less than the other two in an order of consideration.
magnitude. It can be clearly seen that cross bonding can reduce Based on the analysis from before, we can draw the following
the sheath loss and maximum sheath current significantly. conclusions: 1) The two ends solid-bonded cable system and
Reference [20] indicated that in a real project, the cable line the single-end bonded cable system are defective in practical
is composed of seven sections of complete cross bonding and projects because of their great sheath loss or high sheath voltage.
each segment is about 570 m long. To conduct comprehen- 2) The cable sheath loss decreases quickly with the number of
sive research on how the cross-bonding joints number influ- cross-bonding joints. 3) The cable system with cross
ence the sheath loss, we calculate the sheath losses based on bonding joints can form complete cross-bonding sections, the
the test system with 14–29 cross-bonding joints. The segment sheath loss of which is much less than that of the system with
length here varies from about 400 to 800 m. The calculation or cross-bonding joints. 4) For a cable system
results are shown in Fig. 6. It can be clearly concluded that if composed of complete cross-bonding sections, the deviation in
the cross-bonding joints form complete cross-bonding sections segments length can enlarge the sheath loss dramatically.
(such as 14, 17, 20, 23, 26, and 29 joints), the sheath loss is much
less than that of the incomplete cross-bonding sections (such as B. Adopt Sheath Connecting Impedance to Reduce Sheath Loss
15, 16, 18, 19, 21, 22, 24, 25, 27, and 28 joints), and to com- References [7] and [12] propose that the circulating current
plete or incomplete cross-bonding sections, the sheath loss will can be reduced by increasing the resistance of the sheath.
decrease quickly with the number of cross-bonding joints. However, there are some problems in that method: 1) Although
Still using the test system above, we assume that the three increasing resistance attenuates the circulating current, the in-
segments in each complete cross-bonding section have different creased resistance itself backs off the effect of reducing sheath
LIN AND XU: CABLE SHEATH LOSS REDUCTION STRATEGY RESEARCH BASED ON THE COUPLED LINE MODEL 2309

Fig. 8. Installation of the connecting impedance.

Fig. 9. Influence of the connected impedance with a different phase angle on

, and .
loss since sheath loss is also proportional to the resistance
if the current is maintained. 2) An additional procedure and
material should be added when manufacturing the cable sheath, have better sheath loss reduction and sheath voltage limitation.
which will increase the cost. 3) It is also difficult to raise the Please note that the connected reactor itself will not generate
sheath resistance of a cable system already in service. The loss, which is also an advantage over the resistor. So in the fol-
method of installing sheath connecting impedance is proposed lowing discussion, we only discuss the connecting reactor be-
in [15]–[17] to limit the sheath loss, which is shown to work tween sheaths.
well by the calculation as well as experimental results. In this 2) Determine the Amplitude of : In this part, the proper
part, we carry out a comprehensive research on the proper reactance of the connected reactor is discussed. Based on the
impedance to reduce the sheath loss. We study and compare test system, the studied situations include 0, 2, 6, 7, and 20 CBJ,
the influence of the impedance with different phase angle and which have covered two ends solid bonded, completely and in-
magnitude. The characteristic of the connected impedance is completely cross-bonded arrangements. Fig. 10 shows how the
fully shown in this part. reactance of the reactor influences the sheath loss and sheath
The connecting impedance is installed in the link between voltage. The reactance here varies from 0.0 to 0.5 . The mea-
two sheaths at the cross-bonding joints or in the sheath surement 0.0 corresponds to the situation that not one con-
grounding line, which is shown in Fig. 8. The loss generated necting reactor is installed in the cable sheath. The cable sheath
by the connected impedance and the consequent heating will loss decreases quickly with the reactance of the connected re-
not limit the cable ampacity because the impedance heat can be actor. And if the reactance of the reactor is larger than 0.5 ,
easily handled by an additional ventilating device at the joint and drop more than 50%, which is quite considerable.
spot. So in the following discussion, we only consider the loss When the reactance is larger than 0.5 , the curve for 6 and 7
produced by the cable sheath and focus on and . The CBJ cables in Fig. 10(a) and (b) is quite flat, which indicates
proper sheath connecting impedance should not increase the that it is not efficient to adopt a reactor larger than 0.5 . So we
sheath voltage beyond the standard while, of course, notably recommend connecting 0.5 in the 6 and 7 CBJ arrangements.
decreasing the sheath loss. So we are also concerned about the Similarly, it is not hard to find that the proper reactances for “2
maximum sheath voltage along the cable line, which is denoted CBJ” and “20 CBJ” are 0.4 and 0.3 , respectively.
by for short. From Fig. 10(c), we know that the sheath voltage does
After connecting the impedance, we need to modify not increase obviously with the reactance for cross-bonded
the boundary conditions and the admittance matrix of the arrangements. However, the sheath voltage of the two ends
cross-bonding joint accordingly. Let represent the con- solid-bonded arrangement increases with the connected re-
nected impedance. The origin connecting resistance is neglected actance. It is not suitable for the two ends solid-bonded
because it is very small compared with . arrangement to install a reactor whose reactance is larger than
We need to determine the amplitude and phase angle of . 0.1 in this case. However, the sheath loss can be reduced by
1) Determine the Phase Angle of : To study how the nearly 20%, even only 0.1- reactance is installed, which will
phase angle of influences the sheath loss, we carried out not increase the sheath voltage over the limit for safety.
research on the test system with one complete cross-bonding The recommended reactances for the considered arrange-
section. The calculating result is shown in Fig. 9. The vertical ments are listed in Table III. A comparison of sheath loss and
axis shows the ratio of , , and after adopting con- between arrangements with and without rectors is also
nection impedance to that before adopting the impedance. In presented. It is shown that installing a reactor in the cable
this part, the amplitude of is fixed to be 0.2 . sheath is an efficient way to reduce the sheath loss, which is
It can be clearly seen from Fig. 9 that the effect of the con- also not costly.
nected impedance becomes better when the phase angle be- The curves shown in Figs. 9 and 10 are instructive and helpful
comes larger. Both the sheath loss and the maximum sheath to find proper reactance. For a new cable system, similar re-
voltage decrease with lager impedance phase angle. Connecting search should also be carried out and comprehensive calculation
capacitance into the sheath joint will add the sheath loss as on sheath loss with different reactances is needed. If modern
well as increase the sheath voltage. The calculation results of optimization algorithms (such as the genetic algorithm and hy-
larger impedance magnitude and more joints are similar to those brid firefly algorithm [15]) are taken to further optimize the re-
shown in Fig. 9. It is shown that we should adopt a reactor to actor arrangement, the calculation is even larger. The proposed
2310 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 30, NO. 5, OCTOBER 2015

The transient characteristic of cables with a connected reactor

needs further investigation, which is left to future work.

REFERENCES
[1] A. Paolucci, EHV AC Undergrounding Electrical Power, A. Paolucci
and R. Benato, Eds. New York, USA: Springer, 2010.
[2] J.-R. R. Ruiz and A. M. Xavier, “Effects of the circulating sheath cur-
rents in the magnetic field generated by an underground power line,”
in Proc. Int. Conf. Renew. Energy Power Qual., 2006, pp. 1–6.
[3] J. R. Ruiz, G. Antoni, and A. M. Xavier, “Circulating sheath currents
in flat formation underground power lines,” in Proc. Int. Conf. Renew.
Energies Power Qual., 2007, pp. 1–5.
[4] M. Abdel-Slam, H. Anis, A. El-Morshedy, and R. Radwan, High
Voltage Engineering Theory and Practice. New York, USA: Marcel
Dekker, 2000.
[5] J. S. Barrett and G. J. Anders, “Circulating current and hysteresis losses
in screens, sheaths and armour of electric power cables—mathematical
models and comparison with IEC standard 287,” Proc. Inst. Elect. Eng.,
Sci., Meas. Technol., vol. 144, no. 3, pp. 101–110, 1997.
[6] O. E. Gouda, A. Z. E. Dein, and G. M. Amer, “Effect of the formation
of the dry zone around underground power cables on their ratings,”
IEEE Trans. Power Del., vol. 26, no. 2, pp. 972–978, Apr. 2011.
[7] O. E. Gouda and A. A. Farag, “Factors affecting the sheath losses
in single-core underground power cables with two-points bonding
Method J,” Int. J. Elect. Comput. Eng., vol. 2, no. 1, pp. 7–16, 2011.
[8] I. Sarajcev, M. Majstrovic, and I. Medic, Sarajcev, M. Majstrovic: Eihp.
hr “Calculation of losses in electric power cable as the base for cable
temperature analysis/I,” Wessex Inst. Technol., 2000.
[9] F. W. Carter, “Note on losses in cable sheaths,” in Proc. Math. Cam-
bridge Philosoph. Soc. Cambridge, U.K.: Cambridge Univ. Press,
1928, vol. 24.
[10] W. Cramp and G. M. Harvey, “Note on sheath losses in single-core
single-phase cables,” J. Inst. Elect. Eng., vol. 67, no. 387, pp. 335–336,
1929.
[11] IEEE Guide for the Application of Sheath-Bonding Methods for Single-
Conductor Cables and the Calculation of Induced Voltages and Cur-
rents in Cable Sheaths, IEEE Standard 575, 1988.
[12] B. Novák, L. Koller, and I. Berta, “Loss reduction in cable sheathing,”
presented at the ICREPQ, Granada, Spain, Mar. 23–25, 2010.
Fig. 10. Influence of the connected reactance on and . (a) The
[13] P. Cheng, H. M. Huang, and S. J. Shia, “Analysis and precautions of
influence of the connected reactance on . (b) The influence of the connected
abnormal ring current flowed in the metal sheath of single-core cable,”
reactance on . (c) The influence of the connected reactance on .
Distrib. Utiliz., vol. 24, no. 5, pp. 50–51, 2007.
[14] K. Steinbrich, “Reduction of operating frequency for power cables,”
Eur. Trans. Elect. Power, vol. 12, no. 3, pp. 235–238, 2007.
TABLE III [15] S.-J. Huang, X.-Z. Liu, W.-F. Su, and S.-H. Yang, “Application of
RECOMMENDED REACTOR AND CALCULATION RESULTS hybrid Firefly algorithm for sheath loss reduction of underground
transmission systems,” IEEE Trans. Power Del., vol. 28, no. 4, pp.
2085–2092, Oct. 2013.
[16] J. S. Kim and K. H. Suk, “Development and efficacy of sheath current
reduction devices on underground power cable systems,” in Proc. Int.
Conf. Power Insul. Cables, 2007, pp. 1–6.
[17] C. K. Jung, J. B. Lee, J. W. Kang, X. H. Wang, and Y. H. Song, “Sheath
current characteristic and its reduction on underground power cable
systems,” in Proc. IEEE Power Eng. Soc. Gen. Meeting, 2005, vol. 3,
pp. 1–8.
[18] H. B. Dwight, “Proximity effect in wires and thin tubes,” J. Amer. Inst.
Elect. Eng., vol. 43, p. 366, 1923.
[19] A. H. M. Arnold, “The theory of sheath losses in single-conductor lead-
sheath loss algorithm can do a good job in this steady-state loss covered cables,” J. Inst. Elect. Eng., vol. 67, no. 384, pp. 69–89, 1928.
analysis. [20] R. Benato, “Multiconductor analysis of underground power transmis-
sion systems: EHV AC cables,” Elect. Power Syst. Res., vol. 79, no. 1,
pp. 27–38, 2009.
IV. CONCLUSION
[21] R. Benato et al., “Gas insulated transmission lines in railway galleries,”
This paper presents a complete method to analyze the cable IEEE Trans. Power Del., vol. 20, no. 2, pt. 1, pp. 704–709, Apr. 2005.
sheath loss. The method guarantees efficiency as well as accu- [22] Z. Xu and H. Liu, “The harmonic model and its algorithm for coupled
multi-phase transmission line,” in Proc. 9th Int. Conf. Harmonics Qual.
racy in the calculation. A complete cross-bonding arrangement Power, 2000, vol. 3, pp. 938–941.
can reduce sheath loss effectively, and installing a proper sheath [23] H. Weng and Z. Xu, “Numerical harmonic modeling of long coupled
connecting reactor in cross-bonding joints is also a cost-effec- transmission lines using matrix series theory and recursive approach,”
Int. J. Numer. Model., vol. 26, no. 3, pp. 225–237, 2013.
tive measure to reduce the sheath loss. Please note that the pro- [24] G. H. Golub and A. V. d. V. Henk, “Eigenvalue computation in the 20th
posed algorithm only focuses on the steady-state calculation. century,” J. Comput. Appl. Math., vol. 123, no. 1, pp. 35–65, 2000.
LIN AND XU: CABLE SHEATH LOSS REDUCTION STRATEGY RESEARCH BASED ON THE COUPLED LINE MODEL 2311

Yong Lin was born in Zhejiang, China, in September 1991. He received the B.S. Zheng Xu (M’00) was born in Zhejiang, China, in September 1962. He received
degree in electrical engineering from Zhejiang University, Hangzhou, China, in the B.S., M.S., and Ph.D. degrees in electrical engineering from Zhejiang Uni-
2013, where he is currently pursuing the M.S. degree in electrical engineering. versity, Hangzhou, China, in 1983, 1986, and 1993, respectively.
His main field of interest is high-voltage underground systems. He has been with the Department of Electrical Engineering, Zhejiang Univer-
sity, since 1986 and has been a Professor there since 1998. His research areas
include HVDC, flexible ac transmission systems, and grid integration of renew-
able energy.