non-sustained disruptive discharges: test experiences, standardization status and network...

194 Smeets and Lathouwers: Non-Sustained Disruptive Discharges: Test Experiences, Standardization Status

Non-Sustained Disruptive Discharges: Test Experiences, Standardization Status and Network Consequences

R. P. P. Smeets and A. G. A. Lathouwers KEMA High-Power Laboratory

Utrechtseweg 310,6812 AR Amhem, The Netherlands

ABSTRACT In the test practice of KEMA 32% of the vacuum circuit breakers offered for testing in 1999 have shown non-sustained disruptive discharges (NSDD), mostly within 300 ms after interruption. This implies, that although current has been interrupted, very brief periods of conduction occur, which seems to be an accepted inherent property of vacuum switching devices. NSDDs occurred in the complete range of rated voltages (12 kV-50 kV) and short-circuit breaking current (12 kA-40 kA) of the tested devices. No clear correlation of NSDD occurrence with arcing current could be established. In 79% of the cases, less than four NSDD occurrences per test report were counted. It is demonstrated from short-circuit current test data that during the occurrence of NSDD, a much larger part of the circuit is contributing to transient phenomena than just the “local” parasitic in the immediate vicinity of the breaker. This is derived from the measured duration of the conducting period during NSDD and contradicts the general notion that NSDD is just a local phenomenon, the duration of which is entirely determined by the parameters of the switching device itself. The new standard IEC 62271-100 for circuit breakers makes no distinction between (a) capacitive re-strike (late breakdown leading to polarity reversal of the load capacitor) and (b) NSDD in the case of a vacuum switching device breaking down more than half a power frequency cycle after capacitive current interruption. I t is demonstrated that over-voltages that accompany such “capacitive NSDDs” certainly can not be neglected.

1 INTRODUCTION CCASIONALLY, vacuum interrupters break down 0 relatively long (up to 1 s) after the interruption of

current and restore insulation very soon thereafter. These “late” breakdowns are commonly named “ Non-Sustained Disruptive Discharges” (NSDD) because of the inherent characteristic of vacuum interrupters to restore the insu- lating state almost immediately after the start of the NSDD. Such breakdowns are thought to be initiated by mechanical vibrations, leading to the release of macro- particles [1] or to a sudden increase of field emission level leading to breakdown [2].

In testing, the problem with NSDD is the following: The International Electrotechnical Commission (IEC) and the IEEE/American National Standards Institute (IEEE/ANSI) both describe most tests in a three-phase test circuit. If in such a circuit, a breaker shows weakness (because of poor design) in one phase after all phases ceased to conduct, no current results because of absence of a ground return current path. In that case it is not always possible to show by the recovery voltage that either

one phase was weak-indicating a design flaw, or an NSDD occurred.

The discussion on NSDD as a phenomenon that might affect users of switchgear started in the mid-eighties. The “International Council on Large Electric Systems” (CIGRE) was not intending to further study this phenomenon due to their idea that it might be a problem for the end-users, for instance because of the resulting prob- lems with protection systems, which is out of their scope. The test laboratories (both independent and those operated by manufacturers), however, faced the problem of how to interpret NSDD. Therefore STL (Short-circuit Testing Liaison, an organisation of independent and man- ufacturer-operated test laboratories) continued discussing on the interpretation of NSDD.

Already in 1985 STL stated that “Certification will not be allowed if there are four or more occurrences of the NSDD-phenomenon during a complete test series as re- quired by the standard, including any test duties which are repeated or re-started for whatever reason”. This statement has been further evolved over the last 10 years.

1070-9878/1/$17.00 0 2002 IEEE

IEEE Transactions on Dielectrics and Electrical Insulation

18

As a result, in certification tests in accordance with the widely used circuit breaker standard IEC 62271-100 (formerly IEC 60056), the absence of multiple ( > 3) NSDD’s must be verified by subjecting the interrupter to the rated power frequency voltage for at least 300 ms after interruption (in contrast to 100 ms for other breakers). If there are more than three occurrences of NSDD during the entire series of test-duties, certification will not be possible. Restoration of power frequency current is never allowed.

These requirements, originally formulated by STL [3] as an amendment to the 1987 edition of IEC 60056, is now fully integrated in the revised edition IEC 62271-100, ef- fective from 2001. It has already been integrated into the IEC standard for switches IEC 60265-1 (1998). The commission entrusted with the revision of the US circuit breaker standard IEEE Std C37.09-1999 has chosen not to address the issue of NSDD.

6 9 11 1

2 EXPERIENCE WITH NSDD DURING TESTING

A systematic investigation is made of the occurrence of NSDDs during one full year (1999) of short-circuit testing of vacuum circuit breakers (VCBs) at KEMA High-Power Laboratories. In that year, 133 test reports were produced on different VCBs that underwent current interruption tests. In 43 reports (32% of total), 134 occurrences of NSDD were mentioned. The test-objects covering these 43 reports, originated from 10 different manufacturers (Europe, Asia and North-America). In 9 reports, more than 3 NSDD occurrences were reported during the entire test series, 4 cases of which showed a variety of prob- lems with current interruption and 5 were refused certification purely on NSDD grounds (> 3 occurrences) and further testing was aborted. This experience is outlined in Figure 1. The maximum number of NSDD after a single current interruption was 21, during the observation window lasting from 1/4 cycle of power frequency to 300 ms

I K E M 1994 133 test reports on vacuumsw Shgear I 43 reports w IVI USDD

90 test reports wimoui NSW

VOI. 9, NO. 2 , 4 r i i 2002 195

I I 1 .o

L 3 9 * 0.6

3 E ; 0.4 E

;; 0.2

U

. 2 0

8 + i capacitor bank current ~

0.0 10 ’ 2 0 30 . 40 50

rated voltage (kV)

Figure 2. Ratio of test current/rated short circuit breaking current vs. rated voltage of tests with NSDDs.

after the last poles have cleared. The rated voltage of the population that exhibited NSDD was between 12-50 kV, the rated short circuit current range was 12-40 kA. In Figure 2, the value of the relative breaking current (= actual breaking current/rated short circuit breaking current) of the test duties in which NSDDs were observed, is plotted against the VCB’s rated voltage.

Since testing is focussed on short-circuit breaking performance, no probabilistic interpretation of this graph is possible. Most points in Figure 2 represent multiple NSDD occurrences. This is because rated current of VCBs is in a limited range (40 kA, 31.5 kA, 25 kA, 20 kA, 16 kA etc.) and the test current is always a fixed fraction of the values in this range (60%, 30%, 10%). Therefore, the number of points in Figure 2 is significantly smaller than the number of NSDD occurrences.

From Figure 2, it must be concluded that NSDD is not necessarily a high-current phenomenon. This confirms the findings reported in [4], a study that shows NSDD do ac- tually occur rather frequently even with a negligible current. Keeping this in mind, it is not surprising, that also during capacitor bank test duties, (400 A test current), several occurrences of late re-strikes were observed. From the results, it can not be concluded that NSDD is associated with VCBs in the higher rated voltage range only. During all the tests, the moment of occurrence of the NSDDs was recorded. In Figure 3, the histogram of this distribution is shown, illustrating a sharp decay of NSDD occurrence within a post;current zero interval of 300 ms. In interpreting this graph, it must be realized that the observation window (period of applied recovery voltage) normally was 300 ms so that the data beyond 300 ms cannot be used probabilistically. Breakdowns occurring shorter than 5 ms (50 Hz) or 4.2 ms (60 Hz) after complete interruption of all the three phases are considered to be part of the interruption process [3] and are not counted in the present survey.

.


50 ,

4 duratan of ac recovery vonage applicatan in IEC certdlcation tests c- $ 4 0 - :

c I , I I

I , I , t

I

I

I I I 1 I I I

0

c 8 3 0 ~ c

.I

100 200 300 400 500 600 700 >700 time (after interruption) of occurrence (ms)

Figure 3. Frequency of occurrence (%) of NSDD as a function of time after complete interruption.

Regarding the maximum allowed number of 3 NSDD, this is an arbitrary choice, rather based on a “political” compromise than on technical grounds. Looking at the test experience, a distribution of the number of NSDD per report as outlined in Figure 4 arises. It can be observed that in the overwhelming number of cases, less than 4 NSDD occur. Such data must be handled with care, since not in all cases, the same number of tests were performed for every report.

The numerical data show that approx. 25% of the certified circuit breakers have exhibited NSDD at tests, which implies-together with the findings of the study by Schlaug et al. [4], that also in service conditions, NSDD must be a very common phenomenon. No correlation between NSDD and performance can be established based on the limited statistics reported here.

,

3 HIGH-FREQUENCY ANALYSIS OF TEST RESULTS

In the standard IEC 62271-100, NSDD is defined as “a disruptive discharge between the contacts of a vacuum cir-

Lf LS VCB Lt LI

3rdpar osc

Figure 5. Generalized circuit diagram and H F oscillation paths for current arising as a result of VCB breakdown.

cuit breaker during the power frequency recovery voltage period resulting in a high-frequency current flow which is related to stray capacitance local to the interrupter”. In the following, it will be attempted to quantify the terms “stray capacitance local to the interrupter”.

At breakdown of the vacuum interrupter, a number of high-frequency (HF) oscillations will be excited in the circuit. The circuit parts contributing to these oscillations are indicated in the circuit diagram of Figure 5: a generalized HF representation of any (supply-/test-) circuit. These circuits can be reduced to a single phase diagram (with a proper choice of components) because only one vacuum gap is conducting during NSDD.

At a breakdown of the VCB, the following currents through the VCB in principle arise in a generalized circuit

first parallel oscillation through parasitic inductance Lp and capacitance Cp of the breaker itself;

second parallel oscillation, through parasitic (cable, feedthrough) capacitances Cs, Ct between the VCB and ground;

third parallel oscillation, through source side capacitance Cs and load impedance Z1;

fourth parallel oscillation, through load side capacitance, source inductance Lf and -capacitance Cg;

main circuit oscillation, through both source and load.

Depending on the circuit, each oscillation path is char- acterized by its frequency (determines the frequency of the oscillating current) and surge impedance (determines the amplitude of the oscillating current). The duration (DN) of the NSDD is a complicated function of the circuit parameters, the voltage across the gap at the moment of breakdown and the gap ability to interrupt HF current. It is thus virtually impossible to predict this duration with a given circuit and breaker. The other approach, of deter- mining from measurement what oscillations are excited and thus what parts of the circuit are contributing to NSDD current given a certain NSDD duration and a set of circuit parameters, is possible. This approach will be used here to quantify the spatial extent of NSDD activity in a circuit.

[51:


- 10'

9 0

5 102 9

1

first NSDD in phase 1

e dildt at NSDD current zero .... ~ ......... ~~ ............ ~ ........ ~~~~ ...... I.-..- .

*. . . ~. . . . . .: _-!. . . . . . . ~. . . . . ~ ~. . . . . ~. . . . .. . ~ - . . . . . ~. . . . . . . . .

time 10 usldiv

second NSDD

Figure 6. Measured voltage across VCB at the occurrence of NSDD in phase 1 (upper) and phase 3 (lower).

Data are used from high power tests of commercially developed vacuum circuit breakers, mostly offered for certification in accordance with IEC 62271-100. One measured example is given in Figure 6, where two NSDD"s (the first in phase 1 and the second 600 ps later in phase 3) are shown. In this example, the duration of the first NSDD (identified by a period of zero voltage across the VCB) DNl = 20 ps and of the second: DN2 = 70 ps. The data were recorded during a short circuit test with 100% of the (symmetrical) rated breaking current, with the short circuit point solidly connected to ground (Z1= 0 in Figure 4). The test-circuit's HF characteristics are given in Table 1.

Analysis with an advanced circuit-analysis package (MatLab Power System Blockset) of the HF phenomena in the complete three-phase circuit including the parasitic components of the test circuit shows the simulated voltages and NSDD current as plotted in Figure 7. As can be seen, the NSDD current has a clear 40 kHz component, proving that the main circuit oscillation is essentially tak- ing part in the events during NSDD.

Table 1. HF characteristics of generalized circuit. Surge

Oscillation Frequency impedance (see text) (WZ) (a )

First par. osc. (estimated) > 10000 32 Second par. osc. (estimated) 410 90 Third parallel oscillation 260 70 Fourth parallel oscillation 51 750 Main circuit oscillation 40 200

Vol. 9, No. 2, April 2002 197

voltages across VCB (50 kVldiv)

NSDD current

time 100 usldiv _ _ _ _ _ _ _ _ _ _ -

Figure 7. Upper part: simulated voltages across VCB, lower part: current through VCB of first and third phase.

In terms of the di/dt at current zero (the value of which is indicative of whether the interrupter is able to interrupt), it can be concluded that (in this case) di/dt has to decrease to a value of approx. 50 A/ps, before this breaker is able to interrupt the HF current, both in phase 1 and phase 3. This is made clear in Figure 8, an, enlargement of the reconstructed NSDD current in phase 3 (from Figure 6). In the lower part of Figure 8, the value of di/dt at current zero is indicated by dots. It can be seen that three value ranges of di/dt can be distinguished, each corresponding again with excitations of different portions of the circuit. The first current zero crossings occur with values > lo5 A / p s (corresponding with the first parallel, purely local oscillation), the second group is between 100 and

1 / I NSDD current 200Aldiv j IO!

1

3.909 3.91 3.911 3.912 3.913 3.914 3.915 3.916 3.917 3.918 3.919

10' lime (us)

Figure 8. Enlarged reconstructed NSDD current (phase 3, DN2) (upper) and di/dt at NSDD current zero (lower).

198 Smeets and Lathou wers: Non-Sustained Disruptive Discharges: Test Experiences, Standardization Status

Table 2. Assessment of capacitive breakdown in standard IEC62271-100

Time of Assessment Acceptability

(cycles) event certification occurrence of for

t < 1/4 .. re-ignition allowed 1/4 2 t 5 1/2 restrike not allowed

t > 1 / 2 NSDD up to 3 allowed

10,000 A/ps whereas the third group (associated with the main-circuit oscillation) has values of di/dt < 100 A/ps . Only these values of di/dt are within the “interruptible range” of this particular interrupter. This tendency was found in other tests as well. From this, it can be concluded that the NSDD current is not necessarily localized purely to parasitic capacitances (Cp, Cs, Ct) in the vicinity of the interrupter, and that NSDD is not necessarily a local phenomenon and may therefore interact with more distant circuit components. General rules cannot be given.

4 NSDD IN CAPACITIVE CIRCUITS An important change in the new (2001) version of the

IEC standard IEC 62271-100 with respect to the 1987 version is the assessment of late breakdowns in VCBs at the interruption of capacitive current.

Starting from a three-phase test circuit with isolated neutral, where a distinction between NSDD and re-strike can not be made, the new IEC 60056 makes a distinction based on the moment of occurrence (expressed in power f requency cycles a f te r clearance of the last phase), see Table 2. Because no distinction is made between consequences of the breakdown, the distinction based on the moment of occurrence may lead to peculiar (and some- times unacceptable) situations as illustrated below.

As an example, Figure 9 shows a very short (20 ps) lasting conducting period, with the appearance of NSDD in phase 1, which in IEC certification tests should be assessed as a re-strike because of its moment of occurrence, and would therefore lead to rejection of certification. Al- ternatively, also the opposite happens, as shown in Figure 10, resulting from a back-to-back capacitor bank test. Here, a late breakdown in phase 1, initially leads to a dra- matic load-side over-voltage (out of measured scale) and results in a breakdown in the test circuit affecting the voltage in the other phases as shown. In conformity with the new IEC 62271-100, this severe event must be assessed as an NSDD, and is therefore formally acceptable (up to 3 times). However, interpretation of this event as a (capacitive) re-strike-polarity reversal of load capacitance-is technically more appropriate.

Therefore STL has prescribed the following procedure which enables to make a distinction between re-strike and NSDD. By grounding both the supply and the load side or by grounding the supply side and connecting the load side

I

Figure 9. Measured short breakdown with the appearance of NSDD that is assessed in IEC 62271-100 as a restrike. Time, 5 ms/div; voltage, 20 kV/div.

capacitors (CI) neutral to ground with a capacitor having a value Cn = 0.1 C1, a re-strike can be recognized by any current flow between circuit breaker and load side capacitance. Because this is not easy to measure the following results of this current flow are indicative:

-ill- voltage at load side of VCB ’

I / Figure 10. Breakdowns with appearance of a restrike causing large over-voltages, assessed in IEC 62271-100 as an NSDD. Time, 20 ms/div; voltage, 10 kV/div.


changing of polarity of load side voltage in case of

decrease of voltage on load side by more than 5%,

voltage escalation in case of multiple breakdowns Since NSDD is a phenomenon by definition only influ-

encing local capacitances around the VCB of,one phase, without causing any current to flow in the main capacitive circuit, none of the above mentioned three points may occur in case of NSDD. Practical experience has to be gath- ered yet regarding the suitability of the methods and the optimum choice of the value of neutral capacitor and acceptable voltage decrease.

odd number of current loops

due to energy loss caused by re-strike current

5 NETWORK CONSEQUENCES-OF NSDD

In the majority of cases, NSDD is a harmless curiosity, inherent to vacuum interrupting devices, leading to very short periods of conductivity. The length of this period is the result of a complicated interaction between interrupter and circuit, the circuit being not just the parasitic circuitry of the breaker itself (first parallel oscillation). Would the VCB had been able to interrupt the very high

voi. 9, NO. 2 , 4 r i i 2002 199

restrike initiated by NSDD t c

_ _ _ _ _ . _ . . _ . Figure 11. Voltage across VCB. Power frequency current in two phases initiated by NSDD. Time 2 ms/div; voltage, 50 kV/div.

large capacitance close to the ‘vacuum gap, these capacitances will discharge when the NSDD occurs not followed by immediate high-frequency quenching. It is clear that such a condition is relatively far away from the standard test-circuit. Therefore, in the future, test-circuits should be defined adequately to simulate the results of high fre-. quency phenomena like NSDD as in service conditions like in the case of small-inductive current interruption testing.

6 CONCLUSIONS frequency current related to this oscillation, then the occurrence of NSDD would have remained hidden alto- gether in test laboratories where voltages are always measured at some distance (few m) from the breaker. Because of the dependence on the circuit beyond the VCB itself, in some cases, however, unexpected consequences of NSDD must be recognized. Some cases are illustrated below.

1. During current interruption tests, NSDDs occur

2. Of all circuit breakers, certified by KEMA, approx.

3. The majority of NSDD (> 85%) occur within 300

rather frequently (in > 30% of the VCB test series).

25% exhibited NSDD in testing.

ms after current interruption. As a result of NSDD current in one phase, flowing

through more distant (non-local) circuit components, rather severe discontinuities in voltage across the VCB arise in all three phases as shown in Figure 6. It is well known that in the majority of cases, such discontinuities are associated with NSDD: they even serve as an identi- fier of NSDD occurrences during testing [6]. Since most test circuits have one grounding only (either on load side or on source side), a power frequency current by NSDD in one phase cannot occur.

Several cases have been observed in which the increased dielectric stress induces breakdown in a neighboring phase, causing power frequency current in these two phases (Figure 11). Such a current, though interrupted at the next current zero may interfere adversely with the protection system. In grounded power systems, or in the presence of phase-to-ground faults in ungrounded systems, NSDD may lead to resumption of (one loop of) short circuit current.

A low surge impedance of the second parallel oscillation path may be a situation that deserves attention. This is the case in extended cable networks, or any circuit with considerable zero sequence impedance. Having relatively

4. Approx. 80% of the NSDD-containing reports shows less than 4 occurrences of NSDD, which is no objection for certification in conformity with IEC 62271-100.

5. The occurrence of NSDD has been confirmed in the total tested range of rated short-circuit breaking current and rated voltage and was not confined exclusively to high-current tests or VCBs of higher voltage rating.

6. Many cases have been observed in which the complete test circuit was involved in the oscillations excited by NSDD. Defining NSDD as a purely local phenomenon is thus incorrect, and any re-strike that results in high- frequency current not developing into power frequency current after short-circuit current interruption could be termed NSDD.

7. In cases where a significant amount of capacitive energy is present in the circuit (such as in the switching of capacitor banks), late breakdown can lead to excessive over-voltages. Acceptance of such an event as NSDD is hazardous and technically incorrect.

8. Voltage surges, associated with NSDD can induce NSDD in a neighboring phase, resulting in power frequency current.


REFERENCES [l] R. Gebel and D. Falkenberg, “Mechanical Shocks as Cause of

Late Discharges in Vacuum Circuit Breakers”, IEEE Trans. Electr. Insul., Vol. 28, pp. 468-472, 1993.

[2] B. Juttner, M. Lindmayer and G. Duning, “Instabilities of Pre- breakdown Currents in Vacuum I: Late Breakdowns”, J. Phys.

[3] STL Guide to the Interpretation of IEC Publication 56: 4th ed. 1987, 1988.

[41 M. Schlaug and L.T. Falkingham, “Non-Sustained Disruptive Discharges (NSDD)-A New Investigation Method Leading To

D: Appl. Phys., Vol. 32, pp. 2537-2543, 1999.

Increased Understanding of this Phenomenon”, XIXth Int. Symp. on Disch. and Elec. Insul. in Vac., Xi’an, China, pp.

[SI W. M. C. van den Heuvel, “Interruption of Small Inductive Cur- rents in AC circuits”, Ph.D. thesis, Eindhoven University, The Netherlands, 1966.

[61 R. P. P. Smeets and A.G.A. Lathouwers, “Current-Interruption Testing of Vacuum Sitching Devices,” lEEE, Trans. Diel. and Elec. Ins., Vol. 6 , No. 4, pp. 394-399, 1999.

This paper is based on a presentation giwn at the 19th ISDEIV Conference, Xi’an, China, 18 -22 September 2000. Manuscript receiwd on 4 May, 2001, in final form 12 October, 2001.

490-493, 2000.

non-sustained disruptive discharges: test experiences, standardization status and network...

Documents