Ferroresonance Overvoltages and its MitigationIn PEA Distribution Network

Cunyi Yu, Nit Petcharaks, Preecha Sakarung Pradit Fuangfoo Arwut TakkabutraFaculty of Engineering

Dhurakijpundit University10210 Bangkok, Thailand

Research DivisionPEA

Bangkok

Surge Protection DivisionPrecise Co. Ltd.

BangkokTel. (662) 9547300(EXT:585) Fax 9547356 Tel.5905701, Fax:5894859 Tel.9614510 Fax: 9614500

Email: yuc@ait.ac.th , nitp@dpu.ac.th and preecha@dpu.ac.th E-mail: mgrrsd@pea.or.th E-mail: arwut<kandam@mozart.inet.co.th

Abstract

This paper presents the main results related to theFerroresonance Overvoltages and its Mitigation inPEA Distribution Network. The study was carried outwith two steps: First to find the factors that couldaffect on the ferroresonance overvoltages. Second, tofind the methods which can mitigate theFerroresonance Overvoltages. Two methodologieshave been adopted: Field test and Simulation.Research results indicated that the FerroresonancePhenomena are existed on the PEA distributionnetwork. The Peak Value of FerroresonanceOvervoltages could reach about 2.2 p.u. for 22 kVnetwork (with Arrester). These FerroresonanceOvervoltages could also last a longer time period thatdepended on the switching time duration. Thedifferent switching phase angle can obtain thedifferent Ferroresonance Overvoltage for energization.However the Ferroresonance Overvoltage of de-energization is nothing to do with the switching phaseangle, but depends on the charged capacitance(including the Transformer Size and Line length,secondary capacitor bank). So many Transformers ona feeder it will be easier to cause the Ferroresonance.Many methods have been investigated andrecommended to mitigate the FerroresonanceOvervoltages. Study also indicated that ZnO gaplessArrester can effectively clamp and damp theFerroresonance Overvoltages. However from the longrun point of view a special specification of Arrestershould be considered.KeywordsFerroresonance, Overvoltage, Energization, De-energization, Surge arrester.

1. INTRODUCTION

Ferroresonance has been a long-standing concern ofdistribution engineers, first with the introduction of 25kV-class overhead rural systems in some countries inthe 1950’s. [1] The factors which caused theFerroresonance Overvoltage and the methods tomitigate the Ferroresonance Overvoltage have beeninvestigated. However it still has been investigated inrecent years. [2, 3, 4] This may be in recent years thedistribution network has been rapidly developed inmany developing countries. The economics and thus

the reliability of distribution network are getting moreand more important. In addition every Power Utilityhas own particular case therefore the investigation isstill necessary.

In recent years there were a lot of events related toSurge Arresters failure occurred in PEA distributionnetwork. It is suspected that these events are related tothe Arrester failure under the FerroresonanceOvervoltages that was either the Surge ArresterQuality Problems or the inherent network Problems.A research Project has been established fromSeptember 1999 to investigate the causes of theseevents and the methods to mitigate the events. Theresearch team was consisted of three organizations:The Power System Engineering Research Center,Dhurakijpundit University; The Research Division,Provincial Electricity Authority; The Surge ProtectionDivision, Precise Electric MFG. Co., Ltd.

The study was divided in two steps: First to find thefactors that could affect on the ferroresonanceovervoltages. Second, to find the methods which canmitigate the Ferroresonance Overvoltages. Twomethods have been adopted: Field test and Simulation.

Two field tests have been carried out in PEAdistribution network: First Field Test was carried outat Phuthamonthon, Nakorn Pathom in 30 October1999. The Second Field Test was carried out atSanpatong, Chiang Mai in 30 March 2000.

MT (Micro Tran) version of EMTP has been used forsimulation.

Field Tests and Simulation results confirmed that theFerroresonance Phenomena exists in the PEADistribution Network. It is easy to cause theFerroresonance Overvoltages under certain conditions(such as: Single phase switching, certain length ofoverhead line, and delta connection of Transformer’sprimary winding). Results also indicated that the ZnOgapless arresters are effectively to clamp and dampthe Ferroresonance Overvoltages.

Many methods have been investigated to mitigate theferroresonance Overvoltages. Based on the study therecommendations for the present and future have beenmade.

2. FUNDAMENTALS

Ferroresonance is a Resonance Phenomena between acapacitance and an iron core, and thus nonlinearinductor. In a power system the iron core inductor isusually the magnetizing branch of a transformer. Thecapacitance could be the natural capacitance of theoverhead line (or under-ground cable).

In power systems, ferroresonance usually refers to aseries resonant condition. The series circuit isestablished during single phase switching condition(including the Energization and De-Energizationswitching) and its occurrence is dependent on theconfiguration of the circuit, such as the deltaconnection of a transformer winding and with certainlength of lines. A system may also experienceferroresonance overvoltages if some abnormal systemcondition, such as a single line fault, cutout Fuseblowing. Figure 1. shown the diagram of seriesferroresonance.

where L – the magnetizing inductance of transformeriron core;

C – the capacitance of line or cable;E – the voltage source.

Under usual condition:

Where ω =2πf ,However, when

Ferroresonance occurred.

In power systems following circuits are susceptible tocause the Ferroresonance:

This circuit can be simplified as following seriesresonance circuits:

This diagrams can be summarized as followingdiagram:

Under certain conditions the ferroresonanceovervoltages can be caused. In the diagram the CdA isthe capacitance of switching line (say, phase A) toground. It is no affected on the ferroresonance ofenergization due to the voltage of CdA is forced byvoltage source. Therefore it can be neglected in theEnergization. However the CdA will joint theferroresonance while de-energization (see Fig. 9).

CL

ωω 1>

CL

ωω 1=

Figure 2. Diagram to cause the Ferroresonancewhile Single-phase Switching.

Figure 3. Diagram for SimplifiedFerroresonance Circuit.

L

CdCC

L

Figure 4. Diagram of Simplified

L

EA

CdCdA

Figure 1. Diagram of Series Ferroresonance

CE

L

I

3. FIELD TEST

Two Field Tests have been carried out. The First oneis the case of only one Transformer with 100m linelength. The Second Field Test covered fourTransformers with different sizes on a feeder. Figure5.shows the Diagram of First Field Test.

Table 1 is the Results of First Field Test.

Note: Energi. – Energization;De-Energi. – De-Energization;Y – with device; N – without device;CB - Capacitor Bank; Ar – Arrester.

As the study is on the distribution network, thereforethe p.u. value is defined as:

From the First Field Tests following conclusions canbe drawn:1. All case results indicated that there was noFerroresonance Overvoltage occurred forEnergization. This may be due to the random offerroresonance by closing phase angle for energization(see Simulation). But the FerroresonanceOvervoltages can reach about 2.26 p.u. for De-Energization.

2. For the case of Transformer with capacitor bank insecondary winding (According to the PEA Regulation,30% of transformer size is taken), the Ferroresonanceof Energization will rarely be caused. However the

De-energization Overvoltages are higher. This is trueeither with surge arrester or without arrester.

3. The highest value of Ferroresonance Overvoltageoccurred at the case of without Surge Arrester but withCapacitor Bank when single phase been de-energized.The Overvoltage peak value could reach 2.26 p.u.However when with Arrester the highest value wasonly 1.7 p.u.

4. The Ferroresonance Overvoltage Value was slightlyless when the switch location changed from SW11 toSW22. This may be the distance different was onlyabout 100 m (SW22 is located 2 m from Transformer).

Figure 6. is the diagram of Second Field Test.

In second field test total 7 cases have been tested. Itincluded: cases with one, two, three and fourTransformers, Energization and De-Energization,different line lengths, with and without CapacitorBank.

Table 2. is a part Results of Second Field Test.

Figure 6. Diagram of second Field Test

G

10.1km

Ar8.2

T7 : 50 kVA*(15.8 km)

T5 : 100 kVA(10.5 km)

T6 : 50 kVA(13.2 km)

T4 : 500 kVA*(10.6 km)

Table 1. Summary of First Field Test ResultsSwitching atCB Ar Power

SW11 SW22N Y Energi. 0.91 1.00N N Energi. 0.92 1.10N Y De-Energi. 1.26 -

0.96N N De-Energi. 1.371.10

Y Y Energi. 1.0 -Y N Energi. 1.0 1.0Y Y De-Energi. 1.70 -

2.06Y N De-Energi. 2.262.19

Figure 5. The Diagram of First Field Test

SW11 SW22

CBank

40 kVAR

160 kVA

∆∆∆∆/Y

Table 2. Sample of Second Field Test Results(Case of Two Transformers in parallel) (p.u.)

Case No. Case 4.Transformer (kVA) 500 + 100Line Length (km) 11Capacitor Bank NPower Phase A B CEnergi. A 1.0 2.20 2.20Energi. B 1.0 1.0 2.20Energi. C 1.0 1.0 1.0De-Energi. C 1.0 1.0 2.20De-Energi. B 1.0 2.20 2.10De-Energi. A 0 0 0

Note: Energi.: Energization De-Energi.: De-Energization.

kV..p.u1 189629173

222 ≈=×=

The Second Field Test Results can be summarized asfollowing:1). From Results we observed that the FerroresonanceOvervoltages for the energization can be caused insome cases only. The case of Multi-Transformer inFeeder seems easier to occur. However, theFerroresonance often occurred when De-Energizationby cutout Fuse. The Peak value reached about 2.20 pu.At the same time the Hum noise can be heard from therelative Transformer. Even for the Transformer with500kVA size (22 kV voltage level) the FerroresonanceOvervoltage could be occurred when the line length isabout 10.6 km.

2). All results also clearly indicated that for theEnergization (A – B – C) after phase A closed,Ferroresonance Overvoltages appeared at both phaseB and phase C. These Overvoltages will be maintaineduntil relevant phase closed. The Phase C experiencedthe longest Overvoltages (last closed phase). Howeverfor the case of De-Energization the FerroresonanceOvervoltages occurred at each opened phase until thelast phase (Phase A) opened.

3). During second field test all Arresters on the feederencountered many times Ferroresonance Overvoltagesin the half a day (including Energization and De-Energization). Majorities of Arresters successfullywithstood the applied Ferroresonance Overvoltages,and clamped it to a lower value (2.2 p.u.). OnlyArrester (at Phase C of Ar8.2, see Figure 7.) wasfailure in case 5. This Arrester has been used for manyyears.

4. SIMULATION

MT (Micro Tran) of EMTP version has been used tosimulate the Ferroresonance phenomena inDistribution Network.The main items simulated are:Effect of energization phase angle;Effect of de-energization phase angle;Single Transformer and Multi Transformers,Effect of with and without Capacitor Bank inSecondary.

Results from energization with different phase anglesshown that there is different Overvoltage value indifferent closing phase angle. The highest value couldreach 2.23 p.u. However for the de-energization thereis no any differences between different switching offangle. In all simulation we always take the worstphase angle for the Energization.

To confirm the results from simulation the comparisonhas been made. Table 3.shown the comparison.

The results shown that the peak values ofFerroresonance Overvoltages calculated by simulationquite well agree with those of Field tests. TheWaveform of Ferroresonance Overvoltages fromSimulation also well agree with Field Test. Figure 7shown the recorded Ferroresonance Overvoltagewaveform from Field Test. Figure 8 is the simulatedFerroresonance Overvoltage waveform (Energization).

Table 3. Comparison between Field Test andSimulation

(Line length: 10 km, With Arresters, Without CBank) (p.u.)Transformer 100 kVA

Phase A B C

Field Test Energi. 1.0 1.0 2.22(Second) De-Energi. 2.2 2.2Simulation Energi. 1.0 1.0 2.2

De-Energi. 2.2 2.2Note: Energization: A – B – C;

De-Energization: C – B – A.

Figure 8. Simulated Waveform of Energization

Figure 7. Recorded Field Test Waveform ofEnergization

To find methods mitigating the FerroresonanceOvervoltages some simulations have been carried out.That included changing the transformer primarywinding connection (from delta to wye connection),multi-phases switching synchronously, effect of linelength, effect of switching with certain loads, etc.

As an example, Table 4. shown the Effect of switchingwith certain load.

All results with load for the FerroresonanceOvervoltages shown that only a few oscillations(overshoots), and 5% of load is proper for mitigatingFerroresonance Overvoltages.

5. DISCUSSION

From the Field Tests and Simulations somephenomena can be explained as following:

As above mentioned in second Field Test. Accordingto the Regulation of PEA the normal operationsequence is: B – A – C for Energization, and C – A –B for De-Energization. Therefore, Phase C alwaysencountered the longest period of FerroresonanceOvervoltage. This may be the reason why most failureArresters was on the Phase C.

Meanwhile from these facts the conclusion can bedrawn that the switch (either energization or de-energization) time duration should be as short aspossible to reduce the maintaining time of theFerroresonance Overvoltages, thus to reduce thepossibility of arrester damaged.

When the phase A of transformer (or say Terminal A)be energized through a line, the voltage on CdA, thusterminal A is fixed by voltage source EA (see Fig.4).Therefore the CdA could be neglected in theenergization equivalent circuit diagram. However,voltage to ground at Terminal B and Terminal C (orvoltage on CdB and CdC) are the same each other dueto the symmetrical structure of transformer. Thereforethe equivalent circuit will be as Figure 4. This is thesimilar with Figure 1. When the ferroresonanceoccurred, the overvoltage on CdB and CdC will rise tothe same value. The ferroresonance overvoltage willbe remained until Phase B and Phase C be energizedrespectively. For the existed operation procedure of

PEA (B – A – C for Energization) the Arrester atPhase C is always stressed the longest time duration ofFerroresonance Overvoltage. Therefore Arrester atPhase C has the highest possibility to be damaged.

For the case of De-Energization (C – B – A), theequivalent diagram should be as following (Figure 9):

For the de-energization (opening sequence: C – B –A), after phase C (Terminal C) opened, phase A and B(or Terminal A and Terminal B) are still forced byvoltage source EA and EB, respectively (Figure 9).Therefore the CdA and CdB could be neglected in thede-energization equivalent circuit diagram. Theferroresonance will be occurred between phase Cinductance (LCA and LCB) and capacitance CdC. Figure10. is the de-energization equivalent circuit diagram.This is similar with Figure 1. The Overvoltageappeared only on Terminal C. Next, after Terminal Bopened, then the Ferroresonance Overvoltageappeared on both Terminal C and Terminal B.Terminal A will still keep the normal operationvoltage. These overvoltages (voltage on Terminal Cand Terminal B) will be remained until Terminal Aopened.

Figure 10. Diagram of Series Ferroresonancefor De-Energization.

CdCE

L

Figure 9. Diagram of Ferroresonance Circuit forDe-Energization

EC

CdC

LCA LCB

LAB

C

BA

EA EB

CdA CdB

Table 4. Effect of Switching with certain Load(Transformer: 160 kVA, with Arresters, no CBank,Switching at SW11, See Figure 4-1.) (p.u.)

Phase

A B CNo Load 0 2.00 2.00160 with 1 % Load 0 1.00 1.00160 with 5 % Load 0 1.00 1.00160 with 10 % Load 0 1.00 1.00

Note: De-Energization (C – B – A);

6. CONCLUSIONS

After theoretical study, field tests and simulations theconclusions from the research can be drawn asfollowings:(1). It is very high possibility that the FerroresonancePhenomena occurred on the PEA DistributionNetwork (22 kV system) due to the all existed primarywinding connection of distribution transformer (Deltaconnection), longer line length and switching by asingle phase cutout Fuse. When the Ferroresonanceoccurred the voltage could rise to 2.0 – 2.2 p.u. (WithSurge Arresters) and also can keep a longer timeduration. At the meantime a hum noise could also beheard from the Transformer.

This is probably the major reason why so many eventsoccurred and a lot of Surge Arresters damaged.According to the PEA operation regulation (B – A – Cfor Energization and C – B – A for De-Energization)Arrester at Phase C always encountered the longesttime duration of Ferroresonance Overvoltages. Thismay be the reason why most Arresters failed on PhaseC.

(2). The Ferroresonance Overvoltages can be causedby single phase Energization and also by single phaseDe-Energization (including the event of a single-phaseground). Ferroresonance Overvoltages can be occurrednot only on the small size Transformer (say 160 kVA),but also the large size Transformer (say 500 kVA)while the line length is about 10 km.

(3). Gapless ZnO Surge Arresters can effectivelyclamp and damp the Ferroresonance Overvoltages.However multi number, particular the longer timeduration of Ferroresonance Overvoltages applying tothe Gapless Arrester could result in it damaged. As thePeak Value of Ferroresonance Overvoltages usuallyless than the breakdown value of gap the gappedArresters could have less opportunity to be damaged.But the insulation of distribution transformer andunderground cable may be decayed.

(4). Simulation Results quite well agree with theresults from Field Test. There are many possiblemethods to mitigate the Ferroresonance andFerroresonance Overvoltages. For instance, as theprimary winding connection of distributiontransformer can’t be changed at moment, use thethree-phase synchronously switching, such asReclosure, to replace the cutout Fuse in someparticular place; Constraining the total switching timewhile the cutout Fuse is used; Switching with certainpercent load (about 5%) and so on can effectivelymitigate the Ferroresonance Overvoltages.

ACKNOWLEDGEMENTS

The Research Team that consists of theDhurakijpundit University (DPU), the ProvincialElectricity Authority (PEA) and Precise Electric MFG.Co., Ltd. The Research Team has carried out thisProject from September 1999 to June 2000. We wouldlike to express their sincere appreciated to the peoplewho made contributions to the Research Work, andfield test particularly, to Mr. Pongsak Harnboonyanon,the Manager of Research Division, PEA; Mr. EkaratKandam, the Manager of Surge Protection Division,PRECISE Co. for their interesting, favorablediscussion, and strong support. It would be impossibleto complete the Research Project without theirsupport.

REFERENCES

[1] R.A.Walling and et al. “FERRORESONANCEOVERVOLTAGES IN GROUNDED WYE-WYEPADMOUNT TRANSFORMERS WITH LOW-LOSS SILICON-STEEL CORES”, IEEE Trans. PD,Vol. 8, No. 3, July 1993, pp.1647 – 1660[2] Thomas A. Short, “APPLICATION OF MOVsIN DISTRIBUTION ENVIRONMENT”, IEEE Trans.PD, Vol. 9, No. 1, January 1994, pp.293 – 305[3] R.A. Walling, R.K.Hartana and et al.“PERFORMANCE OF METAL-OXIDE ARRESTEREXPOSED TO FERRORESONANCE INPADMOUNT TRANSFORMERS”, IEEE Trans. PD,Vol. 9, No. 2, April 1994, pp.788 – 795[4] S. Lam Du and et al “OVERVOLTAGES ONDISTRIBUTION SYSTEMS”, IEEE Trans. PD,1998