Sie sind auf Seite 1von 7

1594

IEEE Transactions on Power Delivery, Vol. 11, No. 3 , J ~ i l y 1996

ENT SHEATH OVERVOLTAGES IN ARMOURED POWER CABLES


B. GUStaVSen, member IEEE Norwegian Electric Power Research Institute Trondheim, Norway
Abstract This paper is concerned with methods of ithg the build-up of transient voltages between sheath and armour in long armoured power cables. Calculations by a frequency dependent cable model demonstrate that this voltage can be efficiently h i t i t e d to an acceptable level by introducing sheath-armour bondings at regular intervals, or by using a semiconductive sheath-armour interlayer. The paper investigates the required minimum length between bondhgs, as well as the required conductivity of the sheatharmour interlayer if the use of bondings is to be avoided.
1. INTRODUCTION 'In normal cable constructions the lead sheath and the armour are separated by an insulating or semiconductive layer. Armour and sheath are usually connected and grounded at the cable terminals. If the cable is exposed to a lightning overvoltage, a voltage can develop between armour and sheath which increases with the distance from the cable terminal. For long cables this voltage may reach values that can cause breakdown of the sheath-armour insulation. Measures must therefore be provided to limit the voltage. In the case of an insulating interlayer, the sheath voltage is limited by introducing sheath-armour bondings at regular intervals. An approximate formula by Rusck and Uhlman [I] is often used to calculate the sheath voltage along the cable. This formula gives the maximum sheath voltage along a semiinfinite cable exposed to a step voltage between the core and the sheath (ground). The distance at which the sheath voltage exceeds a permissible level is taken as the the maximum permissible distance between sheath-armour bondings. Wedepohl and Wilcox [2] have shown that the sheath voltage calculated by this formula is too high. As lightning overvoltages have a limited duration it may be assumed that rhe sheath voltages are even lower than those calculated in [2] for step voltages.
95 SM 622-1 PWRD A paper recommended and approved by the IEEE Transmission and Distribution Committee of the IEEE Power Engineering Society f o r presentation at the 1995 IEEEjPES Summer Meeting, July 23-27, 1995, Portland, OR. Manuscript submitted December 28, 1994; made available for printing June 1 3 , 1995.

J. Sletbak, member IEEE The Norwegian Institute of Technology Trondheim, Norway


As bondings are a complication, their number should be limited as much as possible or be completely avoided. latter is possible if the interlayer between sheath and armour is made of a semiconductive material. In order to limit the risk of galvanic corrosion, the conductivity of the interlayer sho'uld not be higher than necessary.
In this paper sheath voltages are calculated on a 16.8 km Long submarine cable, both with insulating and semicontductive interlayer. Accurate calculations were performed using the travelling wave method in the phase domain. The method is described in [3].

2. CABLE DATA The analysis in this work is made as a case study on a specific 66 kV submarine cable. The maximum permissible sheath-armour voltage for this cable is 33kV and must not be exceeded when the cable is subjected to a 325 kV 1/50 p impulse voltage. Some cable data used in the analysis are given in fig. 2.1.

a= 6 . 3 5 "
b=1 4 . 5 0 " c=2 2 . 2 0 " d=2 4 . 7 0 " e=2 8 . 8 5 " f = 34.4!5mm

Fig. 2.1 Cable design


Further data used are : Copper core conductivity (75C) : 4.488 lo7S/m Lead sheath conductivity (75C) : 0.388 lo7S/m Steel armour conductivity (75C): 0.798 IO7 S/m Core-sheath capacitance : C,, = 0.554 fF/m Sheath-armour capacitance : C,, = 1.153 nF/m

The stranded armour was modelled as a tubular conductor

w i t h arelative permeability equal to 3 and conductivity equal to that of solid steel. It should be noted that this representation is
not very accurate as steel has a non-linear permeability and because the stranding reduces the effective conductivity as

0885-8977/96/$05.00 0 1995 IEEE

Authorized licensed use limited to: Christian REMY. Downloaded on December 1, 2008 at 08:05 from IEEE Xplore. Restrictions apply.

1595

compared to that of solid steel. The effect of the armour conductivity on the sheath voltages is subjected to a brief investigation in section 5.1. The cable system was assumed to consist of 3 cables located 50 m below the surface of the sea having resistivity equal to 0.5 ohm . n as shown in fig. 2.2. However, only one of the cables was considered in the calculations, as explained in section 5.1.

(3)

where
Z(w) = R (0) +joL(a) ;seriesimpedance

(4) (5)

air
~

Y(o)= G(w) + j w C ( o ) ;shuntadmittance

sea - - - -

-t
--I

Z(w) was calculatedusing the ATP Cable Constants routine [4]

_ _ - -

0 --

0)

0.5m 0.5m
Fig. 2.2. Cable location

in which skin effect and ground conduction effects are taken into account. The imaginary part of Y(o) was assembled from the core-sheath and sheath-armour capacitances. The real part of Y(o) is zero when the sheath-armour interlayer is nonconductive. When conductive, three elements in G becomes nonzero. These elements are equal and were calculated as :

G = 2.n. o/ [In( e / d ) ]

(6)

3. THE RUSCK - UHLMAN FORMULA


The Rusck-Uhlman formula [l] gives the max. sheath voltage between sheath and armour as a function of the distance from the termhal of a semiinfinite cable when a unit step voltage is applied between conductor and sheath/armour:
fl

where o is the conductivity, and d and e are radii defined in fig. 2.1. When calculating Z and Y it was assumed that the armourground voltage is zero along the cable. This will be the case when the protective layer outside the armour is soaked by sea water. 4.2. Calculation of transient voltages

v23

=
2 1 '

12
a

[l-exp(-p.x)]

I '23

The sheath voltage (step response) in the case of semiinfinite cable length was calculated directly from the phase domain propagation matrix of the cable :

The formula is based on the following assumptions : -losses in conductor and armour are neglected -losses in the sheath are calculated assuming DC value for the sheath resistance -The velocity, v, is calculated assuming infinite conductivity of conductor, sheath and armour. The permittivity is assumed equal for the core-sheath and sheath-armourinsulations. From the cable data in section 2 we find the DC-resistance of the sheath to be Rsheath=0.700 Ohmflan, and the lossless velocity of the core-sheath loop to be 1.456 10' m/s. 4. CALCULATION PROCEDURE 4.1. Calculation of cable parameters The cable parameters Z and Y for a system of parallel conductors are in the frequency domain described by the well known matrix differential equations :

where F-' denotes the inverse Fourier Transform. The sheath voltage at position x from the energized cable end then corresponds to aparticular element in h*.(t). The response when subjecting the cable to a 1/50 ps lightning impulse was calculated in the time domain as the convolution between the lightning impulse and the impulse response h,(t) corresponding to h,,(t). The lightning impulse was defined to have a 1 ps linear front and an exponential t a i lwith 50 ps time to half value. In the following the calculation procedure by (7) is termed "Fourier Method". The transient voltages for a cable with a finite length were calculated in the time domain using a phase domain, frequency dependent cable model based on the travelling wave method [3]. (The fitting of the step response with piecewise linear segments described in [3] was, however, not applied.)

Authorized licensed use limited to: Christian REMY. Downloaded on December 1, 2008 at 08:05 from IEEE Xplore. Restrictions apply.

1596

5. CALCULAmD RESULTS 5.1 Sheath voltage along semiinfinite cable Because the armour is thick it may be expected that the magnetic flux penetrating the armour will be small when a bransient voltage is applied between the core and sheath/armow. The sheath voltage at x 4 k m along the mid-cable was calculated by the Fourier Method when applying a step voltage to the core of this cable at the cable terminal. All sheaths and the cores of the two other phases were grounded at this end. The sheath voltage was then recalculated with the two outer phases (cables) neglected in the calculation of cable parameters (Z and Y). The relative deviation between the sheath voltage of the two calculations was found to be less than 0.1% throughout the time span 8-5OOps. Both calculations were recalculated when reducing the conductivity of the armow by 50% to take into account the stranding. The maximum relative deviation then increased to 0.5%. It can thus be concluded that the inductive coupling between phases can be neglected when calculating uansienr sheath voltages. Therefore, only one phase is considered in the remainder of the paper. When considering only one cable, the effect of reducing the armour conductivity by 50% was a 5% decrease of the peak value of the sheath voltage. The voltage along the cable sheath was calculated by the Rusck-Uhlman formula and the Fourier Method when applying a 325 kV step voltage to the core with the sheath grounded at this end. The sheath voltage was also calculated by the Fourier Method when applytng a 325 kV 1/50 ps voltage. Fig. 5.1. shows the maximum sheath voltage for these cases. It is seen that the overestimate according to the RusckUhlman formula is considerable and increases along the cable. The maximum alowable sheath voltage (33 kV) is exceeded at x 4 . 2 km by the Rusck-Uhlman formula, while the Fourier Method predicts about x=6.5 km for the step voltage and x=9.5km for the 1/50 ps impulse voltage.
[kV] 60
50

The voltage along the cable sheath was recalculated by the Founier Method for the step response case when assuming a nonzero conductivity of the layer between sheath and armour. Fig. 5.2. shows the maximum sheath voltage for different values of the conductivity. It is seen that the continuous increase in maximum sheath voltage with distance is reduced when the sheath-armour interlayer is conductive. The reduction is negligible at o = ~ O -S/m. ~ As the conductivity is increased to 0=10-~ S/m, the voltage is limited to 21 kV at x=lOkm. As the u r t h e rincreased, the max. sheath voltage will no conductivity is f longer increase monotonously along the cable. At a=10 S/m the sheath voltage attains its maximum value of 4.2kV at x=3km. [kV] Max. sheath voltage

I O x[kml

Fig. 5.2. Maximum sheath voltage along cable for conducdive sheath-armour insulation (step response)
Fig. 5.3 shows the sheath voltage (step response) at x 4 k m as function of time for the conductivities in Fig. 5.2. It is seen that the tail portion of the response is more influenced by the conductivity than the front portion.

I-----,-- i,-.-Max. sheath_voltage ------~

25

7 Sheath voltage [kV]


-

20 -

_ - - _ _ - - _ _ _ _ _ L _ - _ _ _ ~ - _ _ _ -L----_~

15

Rusck-Uhlman

0
0
e

1
0.1 0.2

1
0.3
0.4

,
0.5

t [ms]

i o xkml

Fig. 5.3. Sheath voltage (step resonse) at x = 4 km.

Fig. 5.1. Maximum sheath voltage along cable.

Authorized licensed use limited to: Christian REMY. Downloaded on December 1, 2008 at 08:05 from IEEE Xplore. Restrictions apply.

1597

5.2. Voltages on a 16.8 km long cable


The sheath voltage dong the cable will now be calculated when the cable length is 16.8 km and the sheath is earthed at both ends of the cable. According to the Rusck-Uhlman formula, 3 bonding points 4.2 h apart should be introduced, as shown in fig. 5.4. In the calculations, the cable was split into 8 parts having length 2.1 km, and a cable model 131 was set up for each part.
section 1
.4-

--

section 2

- _section 3 - -

section 4

Fig. 5.4. Bonding sheath and armour at points 4.2 km cepaa

6
Voltaves in the absence of bondings The core and sheath voltages at points 1-9 (fig. 5.4) along the cable were first calculated when the sheath was connected to the armour at the cable ends only. A zero conductivity for the sheath-armour insulation was assumed. Figs. 5.5 and 5.6 show the core and sheath voltages when applying a 325 kV step voltage to the cable core at point 1 in fig. 5.4. The corresponding voltages when applying a 325 kV 1/50 ps lightning impulse voltage are shown in fig. 5.7. and 5.8. It is seen that in both cases the maximum sheath voltage appears near the cable end opposite to the energized end. It is also seen that the maximum sheath voltage along the cable is lower in case of the 1/50 ps impulse voltage, as w a s also the case for the semiinfinitecable in fig. 5.1.

0.1

0 . 2

0 . 3

0 . 4

0.5 t [ms]

Fig. 5.6. Sheath voltages (325 kVstep response)

4oo

Core voltage [kVl

6oo

fp Core voltage [kV]

0.1

0 . 2

0 . 3

0.4

0.5 t [ms]

Fig. 5.7 Core voltages (U50p s lightning response)

50

Sheath

[kV]

200
100

0O 1 P 0
0.1

0.2

0.3

0.4

0.5 t [msl

Fig. 5.5 Core voltages ( 325 kV step response)

*
0.1

0.2

0.3

0.4

0 . 5 t [msl

Fig. 5.8 Sheath voltages (U50LIS lightning response)

Authorized licensed use limited to: Christian REMY. Downloaded on December 1, 2008 at 08:05 from IEEE Xplore. Restrictions apply.

1598

Sheath voI u s when bondinp the sheath and armour The sheath and armour were bonded at positions 3 , 5 and 7 in fig. 5.4. Each bonding was in the calculations represented by a 1 psZ resistor. Figs. 5.9 and 5.10 shows the sheath voltage calculated at the 4 midpoints (points 2, 4, 6 and 8) when subjecting the cable to a 325 kV 1/50 ps lightning impulse voltage. The maximum sheath voltage at the four midpoints was 12.3, 10.3, 8.7 and 7.5 kV, repectively. This implies that the maximum sheath voltage can in general be expected to appear between the energized cable end and the first midpoint.

Sheath vo ltap - e when using semiconductive interlaver. The calculated sheath voltage along the semiinfinite cable (fig. 5.2) showed that the maximum sheath voltage is limited to values well below 33 kV if the sheath-armour insulation has a S/m. conductivity of With this conductivity the sheath voltage along the cable w a srecalculated when the sheath was connected to the armour at the cable ends only. Figs. 5.11 and 5.12 show the sheath voltage at points 2-8 (fig. 5.4) along the cable, when subjecting the cable to a 325 kV step voltage and 1/50 ~.ls lightning impulse voltage, respectively. The maximum sheath voltage in case of the step response is seen to be l i m i t e d to slightly above 20 kV, as was also the case for the semiinfinite case in Fig. 5.2. It is also seen that the maximum sheath voltage is somewhat lower in case of the lightning response than for the step response.

first midpoint
~

'
I

I
,

- - - - - - - second midpoint

-lo

1
I

W
P

0.1

0 . 2

0 . 3

0.4

0 . 5

t [ms]

Fig. 5.9. Sheath voltage atfirst and second midpoint

-10
third midpoint
10

i
0
0.1

0.2

0.3

0 . 4

0.5 t [ms]

- - - _ - _ _ fourth midpoint
~

Fig. 5.11. Sheath voltages (step response)


25 ]Sheath voltage [kV]

-10

!
0

0.1

0.2

0.3

0 . 4

* 0.5 t [msl

Fig. 5.10. Sheath voltage at third and fourth midpoint


The sheath voltages at the midpoints (points 2,4, 6 and 8) were also calculated when applying a 325 kV step voltage to the core. The maximum sheath voltages at the four midpoints were then found to be 14.0,12.1,10.6 and 9.2 kV, respectively. These values are approximately 2 kV higher than for the previous case of a 325 kV 1/50 ps lightning voltage.

.in ! 0

."

7 -

0.1

0.2

0.3

0 . 4

0.5 t [ms]

Fig. 5.12. Sheath voltages (U50,us lightning response)

Authorized licensed use limited to: Christian REMY. Downloaded on December 1, 2008 at 08:05 from IEEE Xplore. Restrictions apply.

1599

6. DISCUSSION
The results obtained in section 5.1. show that in case of a semiinfinite cable with non-conductive interlayer and sheath grounded at the cable end, the sheath voltage increases with distance from the cable end when subjected to a steep fronted voltage. For a cable of fmite length (16.8 km) and with the sheath grounded at the cable ends only (section 5.2), the maximum sheath voltage along the cable is practically unaffected by the reflections at the far end and thus by the termination of the cable. The reason for the latter can be found by studying the wave shapes in figs. 5.5-5.8. The sheath voltage propagating from the energized end is seen to have a very steep wave front, and the maximum sheath voltage at a given point occurs shortly after arrival of the surge. Since the front time is only a fraction of the travelling time between the cable ends, the reflected sheath voltage will not be able to reduce the peak value of the sheath voltage. It should be noted that the step response of the sheath voltage is in general much steeper for an armoured cable than for cables without armour, as has previously been explained in [2]. The front time of the sheath voltage at a given point can be expected to be about equal to the difference in travel times of the coaxial and sheath-armour wave from the energized end. This difference is small because the difference in velocity of the coaxial and sheath-armourmodes of propagation is small at high frequencies. For the cable in question the two velocities at high frequencies approach 1.46 10' and 1.67 10' m / s , respectively.

calculating transient sheath voltages.

- The Rusck-Uhlman formula gives a significant overestimate of the maximum sheath voltage for a semiinfinite cable when the cable is subjected to a step voltage. The overestimate is even higher when compared to a cable subjected to a 1/50 ps impulse voltage.
When the cable has finite length and is subjected to a step voltage, the maximum sheath voltage occurring along the cable is not affected by the sheath grounding at the far end, except very close to that end.
~

- When the sheath is bonded to the armour at regular intervals, the maximum sheath voltage occurs close to the first bonding.
- When using a semiconductive interlayer between sheath and miourV, the maximum sheath voltage is reduced along the cable.

The reduction increases with increasing conductivity. A conductivity of S/m is sufficient for the cable used in the case study. 8. REFERENCES

e11 S. Rusck and E. Uhlman, "Transient voltages between the lead sheath and armouring of long submarine cables", Direct Current, 1962,7, pp. 225-227.
621 L. M. Wedepohl and D. J. Wilcox, "Estimation of transient sheath overvoltages in power-cable transmission systems", Proc. TEE, vol. 120, no. 8, August 1973, pp. 877-882.

When bonding the sheath to the armour at regdar intervals (section 5.2), the maximum sheath voltage appears between the energized cable end and the first bonding point. The maximum sheath voltage occuring at the midpoint between these points was, however, unaffected by the bondings. The reason for this is that the wave reflected at the bonding point arrives at the midpoint after the peak value has occurred. Thus, the distance between bondings could for the cable in question be selected without considering the reflected wave from the bondings, provided the front time of the applied voltage is small compared to the travel time between bondings. When using a semiconductive insulation between sheath and armour instead of bondings, the sheath voltage was reduced dong the cable. For the cable in question, a conductivity of S/m or higher reduced the maximum sheath voltage to values well below the maximum allowable voltage. For comparison, a conductivity of the order of 10" S/m is common for the semiconductive screens used in XLPE coaxial cables.

[31 B. Gustavsen, J. Sletbak and T. Henriksen, "Calculation of electromagnetic transients i n transmission cables and lines taking frequency dependent effects accurately into account", paper no 94 SM 466-3 PWRD, presented at the 1994 PES Summer Meeting in San Francisco.
[41 Electromagnetic Transients Program reference manual (EMTP Theory Book), prepared by H. W. Dommel for Bonneville Power Administration, P.O. Box 3621, Portland Oregon, 97208 U.S.A., August 1986. 9. BIOGRAPHIES

7. CONCLUSIONS
Measures of limiting transient sheath voltages on a specific 66 kV armoured cable has been investigated. The main conclusions are :

Bjmn Gustavsen was born in Harstad, Norway in 1965. He received the M.Sc. degree in Electrical Engineering from The Norwegian Institute of Technology in 1989, and the Dr. Ing. degree in 1993. He is presently working at the Norwegian Electric Power Research Institute with transient studies.
Jarle Sletbak was born in Namdalseid, Norway, in 1928. He received the M.Sc. degree in Electrical Engineering from The Norwegian Institute of Technology in 1953. He is a professor in High Voltage Technology.His main field of interest is insulating materials and their application in high voltage equipment.

- Coupling to neighbour phases can be neglected when

Authorized licensed use limited to: Christian REMY. Downloaded on December 1, 2008 at 08:05 from IEEE Xplore. Restrictions apply.

1600
Discussion A.Ametani (Doshisha Univ., Kyoto 610-03, Japan): The authors should be commended for presenting an interesting observation of a sheath overvoltage on an armoured cable. A proposed countermeasure to reduce the sheath overvoltage is physically understandable, and is expected to be a valuable contribution to a cable engineering. I would appreciate a further elaboration by the authors to the following questions. (1) The resistivity of a sea water is often taken to be 0.2 n m ,while it is 0.5 D m in the paper. This might result in a less current in the sea water, and the circulating current between the core, sheath and amor is estimated to increase. The authors' comment on the effect of the sea water resistiivity on their work is welcome. Also, if a cable is underground, but not submarine, are the results in the paper stiff applicable? (2) In Sec. 4.2, it is said that the response for a 1/50 P s lightning impulse was calculated in the time domain as the convolution. The approach of the time domain convolution is tedious and inefficient in this case. Having applied Fourier transform method, the impulse response h 6(t) is directly calculated fiom eq. (7) by replacing 10 U for a unit step function by k( l/( a+ju)-Z/( B+j U ) } for an impulse wave k{exp(- a t>exp(- B f ) } , Why the authors adopted such the inefficient approach of the time convolution. (3) Eq. (7) is not clear as a phase domain solution of a sheath Wansient voltage. A bdef explanation of 2 and Y in the equation might help to understand the equation.

2) The reason for adopting a numerical convolution when calculating the 1150ps lightning response in the semi-infinite case (fig. 5.1) was that it was found convenient for a singletime application. However, if this type of calculation was to be performed on a regular basis we would prefer the procedure suggested by the discussor, i.e. to calculate the lightning response in the frequency domain prior to inversion by the Fourier Transform. 3) Z and Y in eq. (7) are matrices in the phase domain, i.e. they are of dimension NxN where N is the number of conductors of the cable system. Therefore, the matrix

is also in the phase domain. The actual calculation of Hstep is carried out as follows :

a) Diagonalize the matrix product Z . Y :

Manuscript received August 15, 1995.

Z.Y =
b)

s
--

. Gustavsen and J. Sletbak : We would like to thank prof. Ametani for his questions and interest taken in the paper.
1) For the cable system investigated in this paper we found the conductivity of the sea water to have a negligible effect on the sheath overvoltages. This is mainly due to the fact that only a small amount of magnetic flux will penetrate the thick armour into the sea. We expect that in case of armoured underground cables the armour may be considerably thinner than that of the cable in this case study. The conductivity of the soil and in particular the presence of the neighbour cables may then be of importance, although we have not investigated this. The results in the paper do not apply to cables without an armour [2].

Calculate Hstep in the phase domain :


r

e "_I

Let i denote the sheath conductor of interest. Element Hslep(i, j ) then represents the sheath voltage when energizing conductor j with a step voltage. (Note that any artificial eigenvector switchover occurring in the diagonalization process will not affect the calculated value for
Hstep

1.

Manuscript received October 16, 1995.

Authorized licensed use limited to: Christian REMY. Downloaded on December 1, 2008 at 08:05 from IEEE Xplore. Restrictions apply.