Power and Instrument Transformer Failures Root Causes and Modern Methods Used to Mitigate Risk. IEEE-NCS,IAS/PES (March 2017)

Transcription

1 Power and Instrument Transformer Failures Root Causes and Modern Methods Used to Mitigate Risk Tamer Abdelazim, PhD, PEng Senior Member, IEEE Eaton Corporation 2611 Hopewell Pl NE Calgary, AB T1Y 3Z8 Scott Basinger, P.Eng. Senior Member, IEEE Eaton Corporation Street NW Edmonton, AB T5V 1E4 IEEE-NCS,IAS/PES (March 2017) Outline Introduction Power Transformer & VT Failure Underlying Concepts Baseline Testing Solution Design Solution Testing Conclusions 2

2 Introduction Over the past few years there have been a number of high profile failures of Power Transformers and Voltage Transformers in medium voltage (5-38kV) applications. These failures have become more frequent because of changes in equipment selection and design in associated power systems. For MV Power Transformers, failures have generally been associated with circuit breaker transient switching: Modern MV Breakers chop current on opening, and pre-strike current on closing. High di/dt interacts with the circuit system characteristic producing a transient voltage of high frequency and high voltage, sometimes exceeding the TRV / RRRV withstand of the breaker, and the BIL rating of the transformer. Dry-type or Low BIL more susceptable, but oil-filled not immune. Short cable or bus connection to transformer. 3 Introduction Most applications do not require mitigation (RC snubbers), but some design choices will make the issue more likely: higher DOE energy-efficient transformers, shorter distance between MV breaker and transformer, dry-type MV transformers with lower BIL designs, MV breaker Vacuum Interrupters designed with higher (>5A) current chop. Switching Transient Studies can help identify the risk before it becomes a problem. 4

3 Introduction For PTs/VTs (Potential Transformers / Voltage Transformers), switching transients can lead to failure as well, although based on our experience, this is the least common mode of failure. The mode of failure we`ve started to see as a more common mode of failure in the field is due to ferroresonance. IEEE Standard defines ferroresonance as A phenomenon usually characterized by overvoltages and very irregular wave shapes, and associated with the excitation of one or more saturable inductors in series with capacitance. In a ferroresonant circuit, the capacitance Xc can be the capacitance of the cable, overhead lines, or stray capacitance of transformer windings or bushings. Under normal circumstances Xc is smaller than XL. In a switching event, voltage can increase, driving the transformer core into saturation and XL is lowered. This `jolts` the transformer XL into a lower saturated value where XL = Xc and ferroresonance starts. 5 Introduction After the upstream breaker opens, there will be a DC trapped voltage on the open line. This trapped voltage can be high enough to saturate the PT and `ring` with a square waveform. This phenomenon is made worse on an under-loaded PT. Modern protective relays have very low burdens compared to electromechanical. The combination of the two design characteristics can lead to some fairly costly failures. Study is required to determine when, but failures observed in the field are typically associated with longer lines and low burdens on switchgear PTs. Can mitigate by several methods to burden the PTs such as loading resistors, saturable reactor (new approach), or a frequency relay tied to a resistor and a timer. 6

4 Transient Overvoltages: 3 Problems Transient Overvoltage (TOV and/or dv/dt) Condition (magnitude and frequency) on Power Transformers leading to Failures. Transient Recovery Voltage (TRV), Rate of Rise of Recovery Voltage (RRRV) important to proper Vacuum Circuit Breaker interruption. PT Ferroresonance due to saturation in VCB switching operations. 7 Introduction Forensic evidence and history of failures Underlying concepts Predicting performance with simulations Mitigating the transients with snubbers Custom designing the snubber Snubber performance measurements 8

5 Unique Case Forensic Evidence Four electricians simultaneously opened four 26kV VCBs simulate utility outage systems transferred to standby generation loud pop in Sub Rm B the relay for VCB feeding transformer TB3 signaled trip Minutes later, two electricians simultaneously closed two 26kV VCBs breakers to Sub Rm A transformer TA3 failed catastrophically 9 Transformer Failure #1 De-energization Examination of primary windings Flash and burn marks on b-phase at bottom & middle Bottom - Indicate a coil-to-coil flashover (high dv/dt) Middle cable used to make delta swung free (lack of support) Transformer passed BIL test at 150kV but failed at 162kV Both failed units: 40 feet of cable High efficiency design VCB switching 10

6 Transformer Failure #2 - Energization Examination of primary windings Coil-to-coil tap burn off Winding showed an upward twist Burn marks from the initial blast Transient on first turns of windings 11 History of Failures Forensic Review Circuit Transformer*** Case Facility Voltage Cable Feet Bil Type Arrester Failure Mode Vacuum Breaker Vendor Switching 1* Hydro Dam Dry No 1st turn A Close 2 Hospital Dry No 1st turn A Close 3 Railroad Liquid N/A middle A Open 4 Data Center Cast coil Yes 1st turn B Close/Open Cast coil Yes None B Close 5 Oil Field Dry No 1st turn C Close 6** Oil Drill Ship <30 75 Cast coil Yes 1st turn C Close Notes: * = 40-50yrs. old with new breaker. ** = 2 yrs. old. All others new. *** = All transformers unloaded or lightly loaded when switched. 12

7 Common Parameters Rules of Thumb to screen applications: Generally, short distance between circuit breaker and transformer about 200 feet or less Dry-type transformer oil filled and cast coil not immune and low BIL Inductive load being switched transformer, motor, etc. (light load or no load) Circuit breaker switching characteristics: chop (vacuum or SF6) or restrike (vacuum) 13 Underlying Concepts - Current Chop VCB opens, arc burns metal vapor Heat supplied by current As current goes to zero, metal vapor ceases Arc ceases or chops All breaker chop current low end 3 5A high end 21A Contact Material Average (A) Maximum (A) CU Ag 4 7 Cr 7 16 W Cr-Cu (75 wt %) Modern VI 3 5 Cr-Cu-Bi (5 wt %) 1 3 Cr-Cu-Sb (9 wt %) 4 11 Cu-Bi (0.15 wt %) Older VI 6 21 WC-Ag (50 wt %) W-Cu (30 wt %) 5 10 Co-AG-SE Cu-Bi-Pb 1 9 6A current chop 14

8 Reignition and Voltage Escalation Current chop plus system C and L imposes high frequency TRV on VCB contacts If TRV exceeds breaker rated TRV, then reignition occurs VCB closes and then opens high frequency current Multiple reignitions lead to voltage escalation 15 Switching inductive circuits Current cannot change instantaneously in an inductor (conservation of energy) Energy Equation ½ LI 2 = ½ CV 2 or V = I L/C Vtransient = Venergy + Vdc + Vosc Venergy is from the Energy Equation Vdc = DC Off-set due to system X/R Vosc = the Oscillatory Ring Wave 16

9 Predicting Performance EMTP Simulations For purposes of screening applications for damaging TOVs Source, breaker, cable and transformer modeled Breaker models for current chop and re-ignition U R UTIL X UTIL UT C1 L CABLE R CABLE C2 T1 L TRAN R TRAN N:1 T2 UN V UTIL C/2 C/2 CH CL R LRG SYSTEM SOURCE AT 13.8 KV VCB BKR CABLE 13.8KV TRANSFORMER 17 Matching model to measurements 3 x 2865KW Gens Vacuum Breaker Short Cable 30KV BIL 1185KVA 630A 1865KW motors V max of 4.96kV < 30kV BIL Oscillation of 20.2kHz 18

10 Transient Mitigation Surge Arrester Overvoltage protection (magnitude only) Surge Arrester + Surge Capacitor Overvoltage protection Slows down rate-of-rise Surge Arrester + RC Snubber Overvoltage protection Slows down rate-of-rise Reduces DC offset and provides damping 19 Breaker opening followed by reignition C TRV exceeds limit C R = 40ohm C = 0.5uF TRV within limit 20

11 A Borderline Case Tier III Data Center 2x 24.8kV lines 2 x 12.5MVA service 13.2kV ring bus 2 x 2250KW generators 6 x 3750KVA cast-coil transformers 90kV BIL VCBs on primary side ft. cables 123kV 969Hz R = 30ohm C = 0.25uF 38.6kV 215Hz 21 Switching a Highly Inductive Circuit UTILITY 4713MVA 3PH SC 9.26 X/R 138KV 1600A 13OHM 50/66/83MVA 135.3/26.4KV 7.5%Z 27KV SF-6 BREAKER 2000A ALUMINUM IPS BUS 53FEET 386kV 1217Hz Vacuum Breaker Short Bus 200KV BIL AUTO LTC 56MVA 27-10KV 3.3%Z VACUUM BREAKER 1200A HEAVY DUTY COPPER PIPE 28FEET LMF XFMR 50/56MVA 25/.53KV 2.5%Z LMF 20MW 56.4kV 200Hz R = 100ohm C = 0.25uF 22

12 Concerns for the Industry VCB retrofit for primary load break switch (LBS) Units subs with LBS and no secondary main Arc flash issues on sec main (no room to install secondary main breaker) Retrofit VCB in LBS box solves AF issue VCB for rectifier (or isolation) transformer DC drives for feed water pumps VCB on primary Short run of cable to transformer (often dry type) New unit sub with primary VCB Metal enclosed vacuum switchgear 7500KVA transformer for gen boilers to meet EPA requirement 5 feet of bus 23 Test without snubber & with snubber Energize without snubber Oscillation continues beyond ¼ cycle Energize with snubber Oscillation well damped Transient followed by high frequency ring Transient near normal crest 24

13 VT failures Circuit configuration A site with a total of 25 SWGRs at 24.9kV A simplified one line diagram Each SWGR has two line-end VTs, one bus VT, and one bus CPT. The UTS breakers are VCBs The UPM devices are manually operated fused padmounted switches Failed VTs during commissioning Performed a comprehensive transient study analyzing transient overvoltage and ferroresonance coupled with test measurements 25 Line-end VT failures Two line-end VTs at 24.9 kv in the switchgear failed on Nov. 18, 2013 which initiated our investigation 26

14 Line-end VT failures When the dielectric rating of the insulation is exceeded, adjacent windings short together, heat and pressure build-up, and the VT housing ruptures. Could take days to weeks before the external evidence occurs. 27 Bus VT failures During the investigation, the bus VT for the same gear failed on Dec. 28,

15 VT Failures Ferroresonance and transient overvoltage Switching transients associated with circuit breakers observed for many years Breaker opening/closing interacts with the circuit elements producing ferroresonance or a transient overvoltage VTs and CPTs applied in system designs not seen 10 years ago We have investigated nearly 150 VT and CPT failures in past 3 years alone Failures may take hours, days or weeks to be detected Across all voltages 5kV, 15kV, 25kV and 35kV Failures not unique to any one manufacturer Mitigating the ferroresonance with damping resistors (conventional solution) with saturable reactors (specialized, custom solution) Mitigating the transient overvoltage RC snubber in combination with surge arrester 29 Underlying Concepts - Ferroresonance phenomenon and the VT When the source of current is interrupted, the trapped dc charge on the cables (capacitance) discharges into the VT VT Nonlinear inductance Dc trapped charge 30

16 Underlying Concepts - Simplified diagram The primary system circuit can be further simplified by lumping the system capacitance into a single representative capacitor and representing the inductance of the transformer as multiple smaller inductors in series. The over voltage across the primary winding is not equally distributed, resulting in high stress voltage differences across some of the windings. If the voltage difference between winding is in excess of their dielectric/insulation limits. VT Nonlinear inductance Dc trapped charge High stress 31 Underlying Concepts VT excitation curve Referring to the excitation diagram, one can see that it takes very little overvoltage to drive significantly higher excitation current through the thin primary conductors. With a turns ratio of 208/1, for only a 50% increase in voltage (180 V), the primary excitation current alone is 0.06 A (i.e / 208) or the full load rating of the primary winding. Amperes volts % 475% 4700 % 32

17 Baseline Testing - Instrumentation Test equipment includes voltage dividers and a transient recording device Voltage Dividers (Capacitive and resistive elements, 10MHz frequency response, SF6 insulated) Three-Phase Power Quality Recorder (Transient voltage waveshape sampling) 200 nsec sample resolution 5 Mhz sampling 33 Baseline Testing - UTS breaker opening Recorded Volts/Amps/Hz Zoomed Detail: 02/25/ :50:36-02/25/ :50: Measurement Volts Squarewaves with 17Hz :50: :50: :50: :50: :50: :50: HH:MM:SS.mmm V Waveform Avg AG V Waveform Avg BG V Waveform Avg CG [kv ] Simulation Squarewaves with 17-25Hz [s ] ase3.pl4; v :S1C :S1C 03B 03C (file C x-v ar t) 03A v v :S1C 34

18 Baseline Testing - UTS breaker opening Recorded Volts/Amps/Hz Zoomed Detail: 02/25/ :30:05-02/25/ :30: Measurement Volts Squarewaves with 10Hz :30: :30: :30: HH:MM:SS.m m m V Waveform Avg AN V Waveform Avg BN V Waveform Avg CN Squarewaves with 10Hz 35 Baseline Testing - UPM switch closing Recorded Volts/Amps/Hz Zoomed Detail: 02/27/ :39:33-02/27/ :39:33 300kV Measurement Volts :39: :39: :39: :39: :39: HH:MM:SS.mmm V Waveform Avg AN V Waveform Avg BN V Waveform Avg CN Recorded Volts/Amps/Hz Zoomed Detail: 02/27/ :26:18-02/27/ :26:18 430kV Measurement Volts :26: :26: :26: :26: :26: :26: HH:MM:SS.mmm V Waveform Avg AN V Waveform Avg BN V Waveform Avg CN 36

19 Solution Design Saturable Reactor + RC Snubbers + Surge Arresters Saturable Reactor (1) RC Snubber + Surge Arrester (2) VT Ohm Noninductive 22 kvpeak KJoules μ F at 27 kv SA Solution Design - Saturable Reactor Circuit The reactor is sized such that as Vs increases, the reactor saturates just before the VT. It is designed to absorb the high excitation current and act as a switch to insert its internal resistance. In addition to matching the magnetic characteristics to the specific VT, the reactor s internal resistance is tuned to the system parameters for critical damping. Since the reactor saturates before the VT and inserts its resistance into the secondary circuit, the VT never goes into saturation. Our modeling optimized the saturation characteristic and the resistance value. The high secondary current is reflected back into the primary, but it is such a short duration that it does not overheat the primary windings. IP Ip Is Xs Rp Rs Xp Is Ie R VP Xm Rm Vp Vs 38

20 Solution Design Saturable Reactor Desired Solution - only apply increased loading on the VT when it experiences ferroresonance. The saturable reactor on the secondary side of the VT only conducts current when the voltage begins to increase above its saturation voltage but below that of the VT s saturation voltage. An off the shelf reactor was initial used for testing, but produced worse results since it saturated below the secondary operating voltage of the VT premature saturation. Since the reactor saturates before the VT and inserts its resistance into the secondary circuit, the VT never goes into saturation. 39 Solution Design RC Snubber 27kV typical snubber & arrester voltage transformer protection non-inductive ceramic resistor 100 ohms Typically 25 to 100 ohms surge capacitor 0.25 μf - Typically 0.15 μf to 0.35 μf 3-phase 24.9kV solidly ground Surge Arrester Resistor 100Ohm Noninductive 22kVpeak 17.5KJoules 0.25μF at 27kV SA VT Surge Cap 40

21 Solution Design Damping Resistor A conventional solution was developed using damping resistors of 250W per phase Easily available Provided damping in 1200ms Can affect VT accuracy Will consume high power during steady state Requires switching circuit to be active only during ferroresonance * (file CaseRp.pl4; x-var t) v:s1c03a v:s1c03b v:s1c03c VT Voltage VT Power (file CaseRp.pl4; x-var t) v:r - 41 Solution Design Saturable Reactor The special design using saturable reactors Custom design Requires extensive simulations Provided damping in 250ms No affect on VT accuracy Will consume very small power during steady state No additional switching circuits required to make it work * (file CaseLp.pl4; x-var t) v:s1c03a v:s1c03b v:s1c03c VT Voltage VT Power (file CaseLp.pl4; x-var t) v:rl - 42

22 Solution Testing - UTS breaker opening Recorded Volts/Amps/Hz Zoomed Detail: 09/03/ :05:55-09/03/ :05: Measurement Volts [kv] :05: :05: HH:MM:SS.mmm V Waveform Avg AG V Waveform Avg BG V Waveform Avg CG Only 250ms Simulation [s] 0.25 (file Case_H_DesignE-Final1.pl4; x-var t) v:s1c03b v:s1c03a v:s1c03c 1 1 = saturable reactor 2 = snubber & arrester 43 Solution Testing - UTS breaker opening 100 Recorded Volts/Amps/Hz Zoomed Detail: 09/03/ :05:55-09/03/ :05: Measurement Volts [V] 75 14:05: :05: :05: HH:MM:SS.mmm V Waveform Avg AG V Waveform Avg BG V Waveform Avg CG 1 Only 250ms Simulation [s] 0.35 (file Cell3_inductor_H_Design.pl4; x-var t) v:tx0001- v:tx0003- v:tx = saturable reactor 2 = snubber & arrester 44

23 Solution Testing - UPM switch closing Recorded Volts/Amps/Hz Zoomed Detail: 09/03/ :42:15-09/03/ :42:15 220V Measurement Volts 0-50 F F Magnitude and Frequency are 14:42: :42: :42: HH:MM:SS.mmm V Waveform Avg AG V Waveform Avg BG V Waveform Avg CG well damped 150 [V] 125V Simulation [ms] 30 (file NM-1-Close-PD.pl4; x-var t) v:tx0001- v:tx0003- v:tx = saturable reactor 2 = snubber & arrester 45 Conclusions Failures due to MV breaker switching transients is on the increase due to higher efficiency transformers, installations with shorter cable lengths (~200ft or less), or VI s with high current chops (>5A). These failures can be mitigated by `hardened` transient resistant transformer designs, or RC snubbers a study is needed to identify the risk. System characteristics and nature of the switching can cause VT ferroresonance Damping resistors or saturable reactors both are able to mitigate ferroresonance 46