Short T.A. Electric Power Distribution Handbook

Подождите немного. Документ загружается.

608 Electric Power Distribution Handbook

12.4.2 Housings

Polymer materials such as silicone or EPR compounds are used for arrester

housings. Porcelain housings are available, and all arresters up until the

1980s were porcelain housed. The main advantage of polymer over the

porcelain is a safer failure mode. An internal fault in a porcelain arrester can

explode the housing and expel porcelain like shrapnel. The polymer housing

will split from the pressures of an internal fault. Arc energies may still crack

apart and eject portions of the blocks; but overall, the failure is less dangerous

than in a porcelain housing. Polymer-housed arresters can pass the ANSI

housing test where a speciﬁed fault current is applied for at least 0.1 sec,

and all components of the arrester must remain conﬁned within the enclo-

sure. Polymer-housed arresters typically can pass these criteria for a 5-kA

fault current for 1/2 s (ÚIdt = I◊t = 2500 C). Porcelain-housed arresters fail

violently for these levels of current. Lat and Kortschinski (1981) ruptured

porcelain arresters with fault currents totaling as low as 75 C (6.8 kA for 11

msec, another failed at 2.7 kA for 50 msec). Expulsion fuses may prevent

explosions in porcelain arresters at low short-circuit currents but not at high

currents (current-limiting fuses are needed at high currents).

A polymer housing keeps water out better than a porcelain housing (water

initiates many failure modes). Polymer-housed arresters — although better

than porcelain-housed arresters — are not immune from moisture ingress.

Moisture can enter through leaks in ﬁttings or by diffusing through the

TABLE 12.5

Commonly Applied Arrester Duty-Cycle Ratings

Nominal

System Voltage

(V)

Four-Wire

Multigrounded

Neutral Wye

Three-Wire

Low Impedance

Grounded

Three-Wire High

Impedance

Grounded

2400 3 (2.55)

4160Y/2400 3 (2.55) 6 (5.1) 6 (5.1)

4260 6 (5.1)

4800 6 (5.1)

6900 9 (7.65)

8320Y/4800 6 (5.1) 9 (7.65)

12000Y/6930 9 (7.65) 12 (10.2)

12470Y/7200 9 (7.65) or 10 (8.4) 15 (12.7)

13200Y/7620 10 (8.4) 15 (12.7)

13800Y/7970 10 (8.4) or 12 (10.1) 15 (12.7)

13800 18 (15.3)

20780Y/12000 15 (12.7) 21 (17.0)

22860Y/13200 18 (15.3) 24 (19.5)

23000 30 (24.4)

24940Y/14400 18 (15.3) 27 (22.0)

27600Y/15930 21 (17.0) 30 (24.4)

34500Y/19920 27 (22.0) 36 (29.0)

Note: MCOV ratings are shown in parenthesis.

1791_book.fm Page 608 Monday, August 4, 2003 3:20 PM

Lightning Protection 609

polymer material. One utility in a high-lightning area has found polymer-

housed arresters with moisture ingress degradation. These show up on ther-

mal imaging systems where an arrester with moisture ingress may be 50∞C

hotter than adjacent arresters. The utility replaces any arresters with signif-

icant thermovision temperature discrepancies. Lahti et. al. (1998, 1999) per-

formed high humidity tests and hot-water immersion tests on polymer-

housed arresters. Arresters with the housing directly molded onto the

arrester body performed well (no failures). Those with housings pressed on

and ﬁtted with caps at the ends often allowed moisture in through leaks in

the end caps.

12.4.3 Other Technologies

Many older arrester technologies are still in place on distribution circuits.

The most prevalent is the gapped silicon-carbide arrester. Silicon-carbide is

a nonlinear resistive material, but it is not as nonlinear as metal oxide. It

requires a gap to isolate the arrester under normal operating voltage. When

an impulse sparks the gap, the resistance of the silicon-carbide drops, con-

ducting the impulse current to ground. With the gap sparked over, the

arrester continues to conduct 100 to 300 A of power follow current until the

gap clears. If the gap fails to clear, the arrester will fail. Annual failure rates

have been about 1% with moisture ingress into the housing causing most

failures of these arresters [an Ontario-Hydro survey found 86% of failures

were from moisture (Lat and Kortschinski, 1981)]. Moisture degrades the

gap, failing it outright or preventing it from clearing a surge properly. Dar-

veniza et al. (1996) recommended that gapped silicon-carbide arresters older

than 15 years be progressively replaced with metal-oxide arresters. Their

examinations and tests found a signiﬁcant portion of silicon-carbide arresters

had serious deterioration with a pronounced upturn after about 13 years.

Under-oil arresters are another protection option. Metal-oxide blocks,

mounted inside of transformers, protect the windings with almost no lead

length. Eliminating the external housing eliminates animal faults across

arrester bushings; on padmounted transformers, connector space is freed.

Since the arrester blocks are immersed in the oil, thermal characteristics are

excellent, which improves temporary overvoltage characteristics [Walling

(1994) shows tests where external arresters failed thermally during ferrores-

onance but under-oil arresters survived]. The main disadvantages of under-

oil arresters are related to failures. If the blocks fail, either due to lightning

or other reason, the energy in the long fault arc around or through the arrester

blocks makes transformer tank failures more likely (Henning et al., 1989).

Even if the transformer tank does not fail, since the arrester is a fundamental

component of the transformer, an arrester failure is a transformer failure: a

crew must take the whole transformer back to the shop.

Most metal-oxide arresters are gapless; but a gapped metal-oxide arrester

is available. A portion of the metal-oxide blocks that would normally be

used are replaced with a gap. This improves the protection. When the gap

1791_book.fm Page 609 Monday, August 4, 2003 3:20 PM

610 Electric Power Distribution Handbook

sparks over, the surge current creates less voltage because there is less metal

oxide. Once the overvoltage clears, the metal oxide provides considerable

impedance to clear the gap with very little power follow current. The gapped

arrester also has good temporary overvoltage capabilities if the gap does

not spark over. If a lightning stroke or ferroresonance sparks the gap, tem-

porary overvoltages can fail the arrester more quickly. Another disadvantage

is the gap itself, which was the most unreliable part of the gapped silicon-

carbide arrester.

12.4.4 Isolators

Distribution arresters have isolators that remove failed arresters from the

circuit. The isolator has an explosive cartridge that blows the end off of a

failed arrester, which provides an external indication of failure. The isolator

itself is not designed to clear the fault. An upstream protective device nor-

mally must clear the fault (although, in a few cases, the isolator may clear

the fault on its own, depending on the available short-circuit current and

other parameters).

Crews should take care with the end lead that attaches to the bottom of

the arrester. It should not be mounted so the isolator could swing the lead

into an energized conductor if the isolator operates. The proper lead size

should also be used (make sure it is not too stiff, which might prevent the

lead from dropping).

Occasionally, lightning can operate an isolator in cases where the arrester

is still functional. One study found that 53% of arresters removed due to

operation of the isolator were because the isolator operated improperly

(Campos et al., 1997) (this ratio is much higher than normal).

Arrester isolators have a small explosive charge similar to a 0.22 caliber

cartridge. The ﬁring mechanism is a gap in parallel with a resistor. Under

high currents, the gap ﬂashes over. The arc through the gap attaches to the

shell and generates heat that ﬁres the shell. The heat from an arc is roughly

a function of ÚI dt, which is the total charge that passes. Most isolators

operate for a total charge of between 5 and 30 C as shown in Figure 12.14.

A signiﬁcant portion of lightning ﬂashes contains charge reaching these

amounts. Even though only a portion of the lightning current normally

ﬂows into arresters, a signiﬁcant number of events will still cause isolator

operation.

Isolators should coordinate with upstream protective devices. Some small

fuses can blow before the isolator clears (the melting time of fuses varies as

ÚI

dt, but the isolators vary as ÚI dt, so there is some crossover point). Most

small transformer fuses may clear in less than a half cycle for faults above

1000 A (the clearing time depends on the timing of the fault relative to the

next zero crossing). A 1/2-cycle fault of 1 kA is 8.3 C, which may not ﬁre an

isolator. This is a safety issue since line crews may replace a blown fuse right

next to a failed arrester that was not isolated properly.

1791_book.fm Page 610 Monday, August 4, 2003 3:20 PM

Lightning Protection 611

12.4.5 Arrester Reliability and Failures

Modern arresters are reliable components with failure rates much less than

1% annually. They do have somewhat of a bad reputation though, probably

because there were bad products produced by some manufacturers during

the early years of metal-oxide arresters. And their tendency to fail violently

did not help.

One manufacturer states a failure rate less than 0.05% for polymer-

housed arresters. This is primarily based on returned arresters, which are

FIGURE 12.14

Operation time and charge required to operate isolators for different current levels (from three

manufacturers).

0 200 400 600 800 1000

0.0

0.2

0.4

0.6

0.8

1.0

0 200 400 600 800 1000

Current, A

Operation time, seconds

Current, A

Charge, coulombs

1791_book.fm Page 611 Monday, August 4, 2003 3:20 PM

612 Electric Power Distribution Handbook

a low estimate of the true failure rate; arresters are almost a throw-away

commodity; not all failures are returned. Arresters may also fail without

utilities noticing; if the isolator operates and a circuit breaker or recloser

closes back in, the failed arrester is left with a dangling lead that may not

be found for some time. Ontario Hydro (CEA 160 D 597, 1998) has recorded

failure rates averaging about 0.15% in a moderately low-lightning area

with about one ﬂash/km

/year. About 40% of their arrester failures

occurred during storm periods.

Lightning arresters fail for a variety of reasons. Moisture ingress, failure

due to lightning, and temporary overvoltages beyond arrester capability

are some of the possibilities. Early in my career, I was involved with lab

tests of arresters and EMTP modeling to evaluate system conditions such

as ferroresonance, presence of distributed generation, and regulation volt-

ages (Short et al., 1994). The main conclusion was that well-made arresters

should perform well. More problems were likely for tightly applied arrest-

ers (such as using a 9-kV duty-cycle arrester with a 7.65-kV MCOV on a

12.47/7.2-kV system), so avoid tightly applying arresters under normal

circumstances.

Darveniza et al. (2000) inspected several arresters damaged in service in

Australia. Of the gapped metal-oxide arresters inspected, both polymer and

porcelain-housed units, moisture ingress caused most failures. Of the gapless

metal-oxide arresters inspected, several were damaged under conditions that

were likely ferroresonance — arresters on riser poles with single-pole switch-

ing or with a cable fault. Another portion were likely from severe lightning,

most likely from multiple strokes. Most of the failures occurred along the

outer surface of the blocks. A small number of metal-oxide arresters, both

polymer and porcelain, showed signs of moisture ingress damage, but over-

all, the metal-oxide arresters had fewer problems with moisture ingress than

might be expected.

Lightning causes some arrester failures. Figure 12.15 shows an example.

The standard test waves (4/10 msec or 8/20 msec) do not replicate lightning

very well but are assumed to test an arrester well enough to verify ﬁeld

performance. The energy of standard test waves along with the charge is

shown in Table 12.6. Charge corresponds well with arrester energy input

since the arrester discharge voltage stays fairly constant with current.

Studies of surge currents through arresters show that individual arresters

only conduct a portion of the lightning current. Normally, the lightning

current takes more than one path to ground. Often that path is a ﬂashover

caused by the lightning; the ﬂashover is a low-impedance path that “pro-

tects” the arrester. The largest stroke through an arrester measured during

an EPRI study with more than 200 arrester years of monitoring using light-

ning transient recorders measured 28 kA (Barker et al., 1993). During the

study, 2% of arresters discharged more than 20 kA annually. The largest

energy event was 10.2 kJ/kV of MCOV rating (the arrester did not fail, but

larger than normal 4.7-cm diameter blocks were used).

1791_book.fm Page 612 Monday, August 4, 2003 3:20 PM

Lightning Protection 613

Manufacturers often cite 2.2 kJ/kV of MCOV rating as the energy capa-

bility of heavy-duty arresters. The actual capability is probably higher than

this. If test results from station-class arresters (Ringler et al., 1997) are trans-

lated to equivalent distribution size blocks, heavy-duty arresters should

withstand 6 to 15 kJ/kV of MCOV (the 100-kA test produces about 4 to 7

kJ/kV of MCOV). Metal-oxide blocks exhibit high variability in energy capa-

bility, even in the same manufacturing batch.

FIGURE 12.15

Lightning ﬂash that failed an arrester. The circuit in the foreground curves around to the right

about a mile beyond the visible poles. Flashovers occurred at poles on either side of the strike

ration. Reprinted with permission.)

TABLE 12.6

Approximate Energy and Charge in ANSI/IEEE

Arrester Test Current Waves

(ANSI/IEEE C62.11-1987)

Test Wave

Energy,

kJ/kV of MCOV Charge, C

100 kA, 4/10 msec 4.5 1.0

40 kA, 8/20 msec 3.6 0.8

10 kA, 8/20 msec 0.9 0.2

1791_book.fm Page 613 Monday, August 4, 2003 3:20 PM

614 Electric Power Distribution Handbook

Darveniza et al. (1994, 1997) found that multiple strokes are more damag-

ing to arresters. Less energy is needed to fail the arrester. The failures occur

as ﬂashovers along the surface of the metal-oxide blocks. The block surface

coating plays an important role; contamination and moisture increased the

probability of failure. Five 8/20-msec test waves were applied within an

interval of 40 msec between each. A set of ﬁve 10-kA impulses applies about

4 kJ/kV of MCOV. Many of the arrester designs failed these tests. The best

design had a housing directly molded onto the blocks.

Longer-duration surges also fail arresters at lower energy levels. Tests by

Kannus et al. (1999) on polymer-housed arresters found many failures with

2.5/70-msec test waves with a peak value of about 2 kA. As with multiple

impulse tests, the outside of the blocks ﬂashed over. Some arresters failed

with as little as 0.5 kJ/kV of MCOV. Moisture ingress increased the failure

probability for some arrester designs (Lahti et al., 2001).

This brings us to a common but difﬁcult to answer question, should we

use normal-duty or heavy-duty arresters? Heavy-duty arresters have more

energy capability. This reduces the chance of failure due to severe lightning

or temporary overvoltages such as ferroresonance. They also have lower

discharge voltages (although the extra margin is not usually needed for

protecting overhead equipment). Normal duty arresters cost less, so the

question remains as to whether the extra capability of the larger arrester is

worth it. For most locations, heavy-duty arresters are appropriate, except in

low lightning areas with less than 0.5 ﬂashes/km

/year. However, this

depends on the environmental exposure (how many trees are around), utility

stocking and standardization considerations (for example, are transformer

arresters the same as arresters at riser poles?), and cost considerations.

12.5 Equipment Protection

Insulation coordination of distribution systems involves some simple steps:

1. Choose the arrester rating based on the nominal system voltage and

the grounding conﬁguration.

2. Find the discharge voltage characteristics of the arrester.

3. Find the voltage that may be impressed on the insulation, consider-

ing any lead length and for cables, any traveling wave effects.

4. Finally, ensure that the impressed voltage is less than the equipment

insulation capability, including a safety margin.

As with many aspects of distribution engineering, standards help simplify

applications. Normally, utilities can standardize equipment protection for a

particular voltage including arrester rating, class, and location, and excep-

1791_book.fm Page 614 Monday, August 4, 2003 3:20 PM

Lightning Protection 615

tions are rarely needed. Construction framing drawings specify how to apply

arresters to minimize lead length.

12.5.1 Equipment Insulation

Distribution insulation performance is characterized with the Basic Lightning

Impulse Insulation Level (BIL). BIL is the peak withstand voltage for a 1.2/50-

msec impulse wave. IEEE Std. 4-1995 deﬁnes voltage impulse test waves as

msec where t

is the equivalent time to crest based on the time taken to

rise from 30 to 90% of the crest (see Figure 12.16). The time to half value t

is the time between the origin of the 30 to 90% virtual front and the point

where it drops to half value. Equipment must withstand a certain number

of applications of the test wave under speciﬁed conditions.

Standard BIL ratings for distribution equipment are 30, 45, 60, 75, 95, 125,

150, 200, 250, and 350 kV. Most equipment has the BILs given in Table 12.7.

Some insulation is also tested with a chopped wave. A chopped wave has

the same characteristic as the 1.2/50-msec wave, but the waveshape is

chopped off after 2 or 3 msec. Because the voltage stress does not last as long,

with most equipment, the chopped wave withstand (CWW) is higher than

the BIL. The CWW values given in Table 12.7 are for transformers (IEEE

FIGURE 12.16

reserved.)

TABLE 12.7

Distribution Equipment Insulation

Impulse Levels (BIL and CWW)

Voltage Class, kV BIL, kV CWW, kV

56069

15 95 110

25 125 145

35 150 175

For transformers: other equipment may

have different CWW.

1.0

0.9

0.5

0.3

1791_book.fm Page 615 Monday, August 4, 2003 3:20 PM

616 Electric Power Distribution Handbook

C57.12.00-2000). For transformers and other oil-ﬁlled equipment and insu-

lators and other air mediums, assume CWW = 1.15 · BIL. For cable insulation,

assume they are equal; CWW = BIL.

Insulation withstands higher voltages when the voltage is applied for

shorter periods. The volt-time or time-lag characteristic of insulation shows

this effect with the time to failure of a given crest magnitude. The volt-time

characteristic of air insulation turns up signiﬁcantly; oil-ﬁlled insulation

turns up less, and solid insulation, very little. For solid insulations like EPR

and XLPE, the small turnup is why the chopped wave withstand is the same

as the BIL.

12.5.2 Protective Margin

Insulation coordination for distribution systems involves checking to see if

there is enough margin between the voltage across the insulation and the

insulation’s capability. A protective margin quantiﬁes the margin between

the voltage the surge arrester allows (protective level) and the insulation

withstand. For overhead transformers and other overhead equipment, we

evaluate two protective margins both in percent, one for the chopped wave

and one for the full wave (IEEE Std. C62.22-1997):

where

CWW = Chopped wave withstand, kV

FOW = Front-of-wave protective level, kV

BIL = Basic lightning impulse insulation level, kV

LPL = Lightning impulse protective level, kV

Ldi/dt = the lead wire voltage due to the rate of change of the current, kV

Both margins should be over 20% (IEEE Std. C62.22-1997); and for various

reasons, we are safer having at least 50% margin. The chopped-wave pro-

tective margin includes voltage due to the rate of change of current through

the arrester leads.

Protective margins for common voltages are shown in Table 12.8 and Table

12.9. The BIL margin is high in all cases; but with long lead lengths, the

CWW margin is low enough to pose increased risk of failure.

The lead length component is very important; the lead voltage can con-

tribute as much as the arrester protective level for long lengths. The arrester

lead inductance is approximately 0.4 mH/ft (1.3 mH/m). Commonly, a rate

of current change is assumed to be 20 kA/msec. Together, this is 8 kV/ft of

Ldi dt

CWW

BIL

CWW

FOW

BIL

LPL

◊

1 100

1791_book.fm Page 616 Monday, August 4, 2003 3:20 PM

Lightning Protection 617

lead length (26 kV/m). This is not an unreasonable rate of rise to use in the

calculation; 20 kA/msec is about the median value for subsequent strokes

during the rise from 30 to 90%.

Lead lengths less than 3 ft (1 m) are necessary to achieve a 50% margin

for protecting overhead equipment. The easiest approach is to tank mount

arresters. Pole or crossarm mounting makes it harder to keep reasonable

lead lengths. It is important to remember that lead length includes the

ground lead as well as the phase wire lead, and the lead-length path is along

the path that the lightning current ﬂows from the phase wire to ground (see

Figure 12.17). Some important, but obvious directions for arrester applica-

tion are:

1. Do not coil leads — While this may look tidy, the inductance is very

high.

TABLE 12.8

Typical BIL Protective Margins for Common

Distribution Voltages (Using the 10-kA Lightning

Protective Level)

System Voltage

(kV)

Arrester MCOV

(kV)

BIL

(kV)

LPL

(kV) PM

BIL

12.47Y/7.2 8.4 95 32 197%

24.9Y/14.4 15.3 125 58 116%

34.5Y/19.9 22 150 87 72%

TABLE 12.9

Typical CWW Protective Margins for Common Distribution Voltages

(Assuming 8 kV/ft, 26 kV/m of Lead Length)

System Voltage

(kV)

Arrester MCOV

(kV)

CWW

(kV)

FOW

(kV)

Ldi/dt

(kV) PM

CWW

No Lead Length

12.47Y/7.2 8.4 110 36 0 206%

24.9Y/14.4 15.3 145 65 0 123%

34.5Y/19.9 22 175 98 0 79%

3-ft (0.9-m) Lead Length

12.47Y/7.2 8.4 110 36 24 83%

24.9Y/14.4 15.3 145 65 24 63%

34.5Y/19.9 22 175 98 24 43%

6-ft (1.8-m) Lead Length

12.47Y/7.2 8.4 110 36 48 31%

24.9Y/14.4 15.3 145 65 48 28%

34.5Y/19.9 22 175 98 48 20%

1791_book.fm Page 617 Monday, August 4, 2003 3:20 PM