Geagre
International Conference on Large High Voltage Electric Systems:
112, boulevard Wa
1986 Session — 27th August-4th September
23-09
seman — 15008 Pais
°
Pref, Subj. 2
IMPROVEMENT OF INSULATION COORDINATION
FOR 420 kv GAS INSULATED
M. LATORRE, M. JIMENEZ-PELEGRI*
Empresa National de Electricidad S.A.
(Spain)
Sonmary
Netal-clad SF,-insulated substations have to meet
a high etandafd of reliability in operation, Thts
mmeans that the risk of failure due to uvervoltages
must be minintzed,
For 4 420-KY SF,-insulated substation, connected to
foverhoad Lines,® the influence of the parameters of
the substation and the Lines are investigated under
the aspect of the risk of Lightning.
Parameters of influence ace the extension of the
substation, number, type and Location of arrester
fand the Lightning parameters of the lines, these
fare nunber” and arrangement of earth wires and
Impulse characteristic of tover grounding.
In this paper design exiteria for Lightning protec-
tion are derived, which are valld for many. sub:
Stations, taking into account the probability of
Lightning. all eriterta have consequently been
realized in the 420 RY substation under investl-
Keywords
Inswlation coordination - lightaing - gas insula
ted substacion
Gas insulated substations (G1S) have nowadays found
broad application in high and extra high voltage
networks. Their advantages in eize, reliability and
Security are well known and described in wany
papers. The scope of this study ie to coordinate
Ingulation depending on the nominal voltage level
and the type of existing trananiesion network.
When connecting 2 CIS to an overhead Tine network
Lightning performance tnvestigations will be neces
sary. AU the highest voltage levels additional. work
has to be done to verity that switching overvolta=
gee do not exceed ineulation levels.
In this study for @ 420-KV GIS, connected to the
Spanish tranamlesion network, Insulation coordina
thon for lightning was performed by computer, Simic
lation.
SUBSTATION BY SIMULATION
by
G. BRAUNER, N. SOMMERFELD
AEG Aktiongesollechatt
(Federal Republic of Germany)
The insulation of metalelad substations can be as
aumed to be predominantly self restoring. The aoml~
nal switching and Lightning withstand voltages are
tested by the so called "fifteen inpulse withstand
Fiftetn impulses of doth polarities of rated vith=
stand voltage and standard shape vaveforn are a
Lied for the test
IE the nunber of disruptive discharges sn the self-
restoring insulation does not exceed to, while in
the nonvself-restoring part of the insulation no
lischarge occurs, che equipment ts considered to
have passed the test successfully.
For the standard insulation level of 420 HV, consi-
dered here, according to the 1EC=recoamandation 1
NY attferenc insulating levels can be chosen #e~
cording to highest possible switching or lightning
overvoltaze:
The environnental influences upon the sneulation of
GIS" 1s extremely lov in comparison with en oper
te equipment. Therefore no decreasing of the di-
Glectrie Insulation and ao ageing caused by ambient
influences will happen. Beesure of thix fact the
insulation levels for CIS chosen {rom IEC 71 given
in {HG 517 are lover than for open air appar
ratus. Thus the recommended lightning level for =
420 KV GIS 16 1300 kV. For epectal purposes higher
values are possible 62g. 1550 WV, which willbe
nore expensive
1S ean be single ph
¢ oF three phase encapsula-
ted, For single phare insvlation only overvoltages
between inner conductor and grounded shield are
Possible. For three phase encapsulation in addl-
tion the dielectric strength Between phase haw
to be dimensioned sufficiently, Az switching over
voltages are normally higher between phases. than
Detween phase to earth, this hav to be considered
in netvorke with high switching overvoltagess
‘The 420-KV GIS investigated here, 18 of single
phase encapsulated type
*
.
Principe ds Vergara 187, 20022 MAORI
Spain23.09 a
Yo minimize the risk of failure by overvoltages, 4
sufficiene margin between the rated withstand vol-
tages and the maxim overvoltages i necessary
Ihe ratio between both quantities gives the safety
factor. {f the statintical distribution of both is
known, this ratio 1a called the statistical sufety
factor. If the avervoltages are obtained by oxpo-
Hence 1t 1s called conventional safety Factor.
here only the conventional safety factor for light
ning 18 considered, which depends on the system
hhoninal voltage and should be at Tease 1.25 for
20 ev
The actual Iightntng safety factor can differ com
siderably from the proposed value:
= The withscund voltage of insulation depends on
polarity and duration of overvoltage. For a
Short duration below about one microsecond
Gnd @ given overvoltage anplitude the safety
factor will be higher than for standard shape
test voltage.
= tf arresters are in service, the safety factor
te replaced by the protection factor, as the
peek value of overvoltage is Limited’ to the
protective level. The nominal procective Level
fs determined by mesna of 4 defined waveform.
In reality this’ Level depends on che front
tine of # surge, eapectally If arresters with
faps are weed. fieause of the Inductivity of
Srrester connection end the residual earthing
Inpedance, higher overvoltages are co be ex
peeted espectally on steep wave fronts.
= tue to reflections in extended G13, the highest
overvoltage within the GIS can differ signifi~
Cantly from the peak value at the arrester
Tocation, Thus the actual safety factor 1s not
ye remote strokes
Lightning striking an overheed Tine of # distance
of about nore than’ I km 5 less severe than
Stcokes in the vicinity of @ GIS because of the
following reasons
= skin-effect and corona, result in front times
~ refractions on the point of intersection of
overhead line to GIS and superposition of
Feflected wavex in GIS reault in further
Ainintshed front tines:
= the amplitude of incoming waves cannot exceed
the insulation level of the incoming overhead
Line.
Normally extended GIS are protected by arresters
te feeders. Av most economic rolution conventional
Spon air afresters with gaps can be used. The most
Cepenoive but most effective rolution 1s encapsu-
Tated, ZOnarvesters located at eritical points
Tt only an arrester at the point of intersection
Of overhead Line to encaprulated bus duct is in
Service, the peatective zone may be Limited
[As a worst case condition, a metalclad feeder, dis
Connected at the end fron’ the busbar, can be con
Sidered (fig. 1).
For remote stroke condition the permissible Length
to be protected by one arrester 1s found under the
Condition, thet the overvoltage at the end Le equal
to the Lightning smpulee withstand voltage.
‘a
UplUp, ———
a
jew fw pw
T= [ws 0] se
zp | so [os [ur
T[ eo [was [coo | us
t[a_[ sso] woe
ao [ies [ano [28
Permivetble Length 1 of metalclad
feeders, connected to overhead Line for
remote stroke
arrester protection level
Tightning.inpulse withstand voltage
highest overvoltage on open end
Ujyyly pretectton factor
amplitude of incoming surge, Limited
to withstand voltage of overhead Tine
the protection factor p at the arrester locarton
fe the ratio Setwoon the nominal Lightalng tnpulse
Withstand voltage Dyyyand the protection level of
the arrester J, of “ars.
‘The moximm permigeible feeder length 1s given Lf
the conventional safety factor at the end reaches
A further parameter 4x the highest possible ampli-
tude U, of the incoming surge which cannot ex:
eed th® Lightning impulse witherand voltage of
the overhead line.
1f for exemple bocaure of pollution the insulation
level of the everhead Tine i= dimensioned 20%
higher than that of the substation (Ug/Uyyy = 1.2)
then for a piven protection factor of p *'"1.25
the permisaible feeder length will be 25-2.
Yor the 420-KV substation described in section 3
the protective factor te. p= 1550/910 ~ 1.7
hich gives on infinite protected Length.4, Close range’ stroke
Here two mechantons have to be distinguished
beck Machover
Lighteings of Mish auplitude tn current end of
short front tine, striking the top of @ tower,
can result in high voltage amplitudes due to
the impedance of tower and earthing. If the
overveltage on the tower exceeds the impulse
withstand voltage, a back flashover to the
Phase wire will occur, resulting in surges of
very short front times,
epending on amplitude end front time of Light=
fing, short protective rones will be possible
To evaluste the risk of failure in e420 KY metal-
clad substation, conputer simulations were perfor
ned with the help of = eravelling wave program,
The simulation included the effect af
= overheadlines, esrthwires, earthin
dinensions, insulation flashever
depending On polarity and voltage rise
arrosters and thelr sparkover-characteristic
depending on front tine and polarity,
fand thelr nonlinear u,t-characterseeie,
= probsbsLity function of Lightning current
amplitude and front cine
probability function of direct strokes duc
fo number and arrangenent of earth wires,
23.09
fig. 2: 420-KY GIS Almeria
HW achenatic dtagean23.09
5. Surweture of
A20-AY-GIS_ investigated
After ENDESA had decided to set up a power station,
2x 550 IW in Almeria, on the south east coast of
Spein, a 420 bY substation was built, for connec=
tion with the Spanish high voltage network.
Because of the pollutant environment, = fully me~
alcclad Si, substation vas selected. An overall
overvoltage*protection had to be developed in order
to minimize the risk of failure,
‘The Aineria, substation Ls arranged according to the
scheme of Ti clreult breakers with three feeders to
overhead 1ine and to feeders to generator trans-
forners ({4g, 2). Two incoming overhead lines are
in parallel of one tower the third 19 on separate
towers but runt parallel to che others for 1 km.
‘The double Line systen was int tially equipped with
only one earth wire, but in the last span with
three, Ihe single Line had tho earth wires for full
system length, not shown here.
hecause of the length of St,-tubes of 43 m for
overhead Line feeders and efpectally of 90 m for
busbar to generator feeders (Fig. 2), emphasis had
to he laid upon finding the necessary number and
the best location for srresters, To reduce costs,
Hvar desirable to use conventional arresters and
it had to be snvestigated whether or not they would
be suf fictent
Data of Almeria
‘ingle phase encapsulated surge impedance
z= "85 chm
rated Lightning impulse utthetand voltage
Uw 2 1330 KV
arrester sparkover voltage (12/50 us)
1050 Kv
arrester residual voltage (10 kA)
‘10k
ws ee wee
me
Correlation between amplitude 1 and
mnean front steepness ai/at of Tight
fing strokes and necessary values for
back flashover in a 420 kV tower.
tower hight 40 m, upper phase conductor
Som, span width’ 500m
+ positive first strokes n= 20
X negative firet strokes n= 9!
subsequent strokes nla
a
8 Results
6.1 Back flashover
For a realtetie design of overvaltege protection,
some statistical data on the correlation of front
Steepness ond amplitude were used. The statistical
data are baseé on mearurenents by Berger” on Monte
San Salvatore between 1963 and 1977 /2/ and Italian
nenturenents between 196 and 1978 /3/,/4/.
‘These experimental data show that negative first
stroker have anplituder op to 100 kA evest but are
Limited in steepness to about 50 kA/us (£ige 3).
Subsequent negative strokes are Linited fn ampli
tide £0.50 KA and in steeprers to 100 kA/ya- Aa &
worst case condition, 100 kt crest end 50 ka/us can
be ausuned. ‘The probability can be extinated to be
far bolow |X of all strokes, For positive strokes
a Hinilar eetination leads to a worst caze condi~
Thon of | 200 kA crest and 25 KA/us
The voltage distribution on the tower (fig.d and
£5. 5) shove, that a back flashover is possible,
‘indy the condition of 10 Ohm earthing resistance
of over
fie 4:
Overvoltage on tover after vorst case
positive stroke
200 ta, 25 kA/us for earthing resistance
of 10 Ohm
overvoltage on tover after vorst case
negative stroke
100 KA 50 kA/ue for sarthing reesetance
of 10 Obn
The tisk of back flashover follows from che distri-
bution function of lightning anplitede and front
tine (€ig. 3).
In this diagram current amplitude and steepness ae-
ceanacy for back flashover are shown under the re~
Sumption af a tower height of WO my 4 phase con
ductor ats height of 30m and x tower earthing
reeistance of | 3 oF 10 Oe“55 23.09
‘As follows from this diagram for a 5 Ohm earthtns,
Tightning enptitedes above 200 ka are necessary. for
back flashover. Considering the influence of the
systen operating voltage the amplitude of current
for back Flashover cannot be defined by a definite
ine but by a range (Peak to peak) shown in ft
by a shaded area, For low steepness. di/dt. this
region has a vertical lapse, because here only the
earthing resistance Ky of tovers are of influsnce,
For higher steepness fower amplitudes are necessary
because of inuctiviey of tower Ly and earthing Ly
3
Wee Mgt th eed
Imax “rae tte ae * Me
From this investigation £ follows, that it is ne
cessary to keep earthing resistances below 5 Ohm
aiving a risk of back flashover which is nogltai-
ble. Tower earthing of 10 Ohn would eecult ts
hack flashover Beginning from about half the
current amplitude as befsre,
6.2 Direct stroke
For vertical Lightning strokes there exists a cri
tical distance between leader tip and conductor. tn
relation to the distance between tip and earth,
Which will cause a direct stroke to a conductor
Golde /5/ suggested a correlation between this
critical distance and the return current empli tude,
This can bo uoed for @ simplified graphical evaluee
tion to determine the area and the highest possible
current anplitade fore direct stroke (fig. @)-
Determination of the area and possible
amplitudes of strikes hiding earthvire
or pharewires pi, p2
fig. 6
Knowing the LsoKeraunic level in strokes per
Square ke and year and che dieteibution function of
Tightning amplitudes, the probability of shielding
Fatluves can be estinated. Additionslly the risk of
direct strokes into the earth vires and the risk of
back flashover can be determined.
by simulation st could be verSfied, that @ shicl-
Ging angle of 38 degrees, ss originally present
rerulted in poseible direct stroker up to. 32 kas
For this worrt cage the overveltage in GIS is
show in fig. 7-
According to table 1 and fig. 24 the insulation
evel 16 axcaadec on busnar, ‘point 25"
7”
(—
Overvoltage in GIS after direct stroke
32 kA, 50 kalus
Overhead Line shielding by one earthing
wire, shtelding angle 35"
1 aide max. oveevottage ink
ka kus AD BBE TT
325013371377 1588 1438
25 1356 1434 TaTS 142
1513331345 12861352
101265 126411951279,
15 5010381097 12431254
2 10311221210 1298
13 1oas 114511991307
10 501059 1082 10451345
It Maximun overvol tages on feederbushingt
My "Bt, busbar end BBE and
transformer TI for direct stroke.
Although the simulation program is able to take
into account corona danping and skin-effect, helt
influence was neglected, because only the moat
critical Lightning strokes in the vicinity oF the
To reduce the maximum anplitudes of overvoltages
without Installing mecalelad arresters, the shiel~
ding of the incoming overhead Lines on’ the Last
four spans was improved by installation of «ae
cond earth wire as shows In. fig,E i
Inprovenent of shielding by double oarth-
ites on last 1.5 km of cverhead Line
fig. 8
The shielding angle decreased to 13 degrees, resul~
ting ins asxtnum current amplitude of stroke below
15'ka (fig. 8). By this measure also the number of
about 40% of the original value,
2400
wv
soo:
3 5 e
oe
fig. 9: Overvoltage tn GIS after direct stroke
15 ka, 50 kAlus
Overhead line shielding by two earthing
wires, shielding engle 13"
7. Gonctusien
For the tmprovenent of overvoltage protection of
GIS three lightning mechanisms have to be consi-
dered
= remote stroke in phase conductor
2 rtroke Into tower and subsequent back flashover
2 Girect stroke to phase conductor
by computer simulation insulation cue to Lightning
wae coordinated.
For renote stroke the GIS was dimensioned for infi~
nite protection length of conventional arresters.
To minimise the risk of back flashover to an ex
treme Low probability it would be shown, thar it is
iuffictent to inprove the earthing systent of the
Inet towers to a value not exceeding 5 ohn. This
could be verified for the distribution funetion of
Lightning amplitudes and front steepness. The risk
of overvaleages due to direct strokes could be
Binimized by shielding wires on the lest kilometer
CE the overhead Lines
AIL measures together allow overvoltage protection
of GIS by conventional arresters.
Reference
ny tee 11-1
12) K. Borger: Methoden und Resultate der BLite-
Forschung auf den Monte San Salvatore bef
Lugano in den Jahren 1963-71
Bulletin’ SEV (1972)
13) &, Garbagnatt; E. Gutdies; G.8.L. Pipero
Neesung von Blitcstrénen in italien =
Ergebnisse oiner statéstischen Ausvertung
cteea 99 (1978), 654-658
fo], Garbagnatt; 6.8.1. Piparo!
Parameter von Blitzecrénen
cetera 103 (1982), 61-65
15 Ret. Golde:
Lightning, Academie Pres 1977
Insulation coordination (1976)
xcept rom Ineraonal Confrence on Large High Voto Elec Systeme
hte Senior