Investigation Report

Power Transformer Tank Rupture and Mitigation

- Current State of Practice and Knowledge

by the Task Force of

IEEE Power Transformer Subcommittee

March 09, 2010

Houston, TX.
CONTENT

1.Investigation and Analysis -10 minutes

2.Utility Experience (15 minutes)
3.Manufacturer’s Perspective (25 minutes)
4.PRD (10 minutes)
5.Conclusions (10 minutes )
6.Q&A (5 minutes)
Analysis of Transformer
Tank Ruptures

Electric Power Research Institute

Wayne Johnson
[email protected]

Nick Abi-Samra
[email protected]
Survey On Transformer Rupture

Objectives
• Develop an understanding of the tank rupture process
associated with internal faults.
• Develop tools, and methods, for evaluating the influence
of tank designs on rupture characteristics

2
 Background

• 42 transformer failures (tanks

ruptured or deformed without
rupture)
• 22 utilities
• 10-year period (1980-1990)
• 7 transformer manufacturers
• Different voltage levels
• Different designs: (GSU’s, Auto,
Phase Shifters, etc.)

3
Conclusions

• The arc energy is the critical rupture parameter

• Differences in transformer design and application are not major
discriminators in the tank rupture
• The fault energy capacity of a tank can be increased by increasing
the tank rupture pressure limit and tank flexibility.
– The pressure at which tanks rupture can be increased by local
strengthening of weak points, while the tank flexibility can be
increased by replacing large beams with a number of smaller
beams (which permit greater deflection at a given stress level).
• Venting to conservators or to auxiliary tanks was not found to be
effective for heavy faults (those with arc power greater than 300 MW)

4
Conclusions

• A long arcing time is not

necessary for rupture:
– About 75% of the cases
occurred with arcing times
less than 4.5 cycles.
• Since arc energy is
proportional to I2t, where the t
is duration in seconds, the
parameter which offers the
most opportunity for control of
the risk of rupture is the
magnitude of the current, and Graph of Tank Deformation
Rupture as a function of fault
specifically, the peak crest current and fault duration.
value of the first half cycle of
the fault current (and the
associated X/R of the circuit).

5
Power Transformer
Tank Rupture
A Utility's Experience

Marc Foata
IEEE Transformers Committee
Houston, March 2010
 Presentation
¾ Assessment of the risk
z Statistics
z Arc energy
¾ Prevention of tank rupture
z Venting
z Containment
¾ Specification of a Tank pressure withstand
requirement

2
Assessing the risk - Statistics

3
Assessing the risk - Statistics

4
Assessing the risk - Statistics

5
Arc energy - 4 MJ

6
Arc Energy - 8 MJ

7
Arc Energy - 12 MJ

8
Arc Energy - 14 MJ

9
Arc energy vs Damage

10
 Arc Energy - Calculation

Earc = 0.9 V I t
¾ I – Arc current: Evaluated from short-circuit level
¾ T – Fault clearing time: Depends mainly on
protection
¾ V – Arc voltage: Very difficult to evaluate
¾ 0.9 – Factor introduced for square waveform of V

11
Arc Energy - Recordings
A 100
Unité Support et analyses - DESTT

E BN TT MAIS MICOUA
80
E BN L-7019 MICOUA
60 E BN L-7011 MICOUA
E BN L-7027 MICOUA
40
20

(kV)
0
-20

-40
-60
-80
205 210 215 220 225 230 235 240 245 250 255
(ms rel. à 05:34:32.2870)
2008-09-12 05:34:32.499 MICOUA 735-02 TENSION RÉSIDUELLE DÉFAUT TRANSFO T8-B

12
11
10
9
8

(MJoules)
7
6
5 ÉNTERGIE TOTALE (TT MAIS)
4 ÉNTERGIE TOTALE (TTC 7019)
3 ÉNTERGIE TOTALE (TTC 7011)
2 ÉNTERGIE TOTALE (TTC 7027)
1
0
-1
205 210 215 220 225 230 235 240 245 250 255
(ms rel. à 05:34:32.2870)
2008-09-12 05:34:32.499 MICOUA 735-02 ESTIMÉ DE L'ÉNERGIE TOTALE DANS LE DÉFAUT TRANSFO T8-B

20000

15000

10000

5000
(A)

-5000

-10000
COURANTS CÔTÉ 735 kV
-15000 COURANTS APROX. CÔTÉ 315 kV

-20000
190 200 210 220 230 240 250 260 270
(ms rel. à 05:34:32.2870)
2008-09-12 05:34:32.499 MICOUA 735-02 COURANTS DE DÉFAUT TRANSFO T8-B

Lundi 15 Septembre, 2008

12
\\VECTEUR\DONNEES\PARTAGE\@AUTOMATISMES\COMPORTEMENT\RAPPORT\RAPP_2008\TOSC_ANA\20080912_053432499_ENERGIE_T8-B_MICOUA.TOS
 Arc Energy - Evaluation of Arc
Voltage
¾ 60-100 V/cm range is often referred
¾ For 40 kV, this means an arc length of
more than 4 m !!

13
 Arc Energy - Pressure Effect

V = 55L P
¾ Constant 55 V/cm
¾ L is arc length (m)
¾ P is absolute pressure (atm)
¾ Pressure in the gas bubble at arc ignition
can reach extremely high values
14
Prevention of tank rupture –
Venting Simulations

15
Prevention of tank rupture –
Venting Simulations

16
Prevention of tank rupture – Vented
vs Non-vented Example

17
 Prevention of tank rupture –
Conclusions on venting

¾ Pressure reduction from a single 25 cm

aperture is low and becomes negligible
when the arc is more than 1 meter away.
¾ An effective pressure venting strategy
would require either a very large venting
duct or numerous small apertures in the
close vicinity of the arc.

18
Prevention of tank rupture -
Containment

19
 Prevention of tank rupture -
Containment

¾ Present design can contain up to 10 MJ

for the largest tanks (735 kV)
¾ More resistant tank design can be
achieved
¾ Need to implement specifications with
minimum energy requirement to meet.
¾ Energy requirement will be a compromise
between the feasibility and the likelihood.

20
 New Specification - Philosophy
¾ Priority is given to the protection of the
workers
¾ Worst energy levels may not always be
containable by the tank
¾ First rupture point must be the cover
¾ Required calculation tools must be
accessible to transformer designers
¾ Must take into account the highly dynamic
phenomena involved
¾ Must be easily verified

21
 New specification - Formula

⎡ ⎤
Ps= F
1 kE
⎢100 + − 50⎥
⎣ 4 100C ⎦

¾ Ps – Calculated tank pressure withstand

¾ F – Dynamic (time & location) amplification
¾ E – Fault energy level to withstand
¾ K – Arc energy conversion factor
¾ C – Tank expansion coefficient
22
 New specifications – Dynamic
factor
3

2,5

2
F

1,5

1
0 20 40 60 80 100 120 140 160 180 200

C/V (x 10 -5 kPa-1)

¾ Time related dynamic factor (pressure and

deformation)
¾ Proximity related dynamic factor
¾ Takes into account tank volume
23
New specification : Hydro-Québec's
Energy Containment Requirements

Voltage Class Arc Energy

24
New specification - Implementation

¾ All transformer suppliers since 1992 have shown

adequate tank withstand calculation capabilities.
¾ Since implementing energy level requirements
(2006), manufacturers have been forced to
improve their tank design.
¾ Detailed analysis by a number of manufacturers
confirmed that all the specified energy
requirements can be met.

25
 Transformer Tanks
- Some Factors Related to Rupture

A Manufacturers Perspective

IEEE Transformers Committee

Houston, March 9, 2010

by Bill Darovny, P.Eng.

Siemens Canada
 Facts about Liquid Filled Transformer Tanks

• C57.12.00 and C57.12.10 define the operating

pressures for transformer tanks
– full vacuum = -101.4 kPa (-14.7 psig)
– pressure 25% above the normal operating
pressure

• Transformer tanks are not pressure vessels

– are not required to be designed to the ASME
Boiler and Pressure Vessel Code.
• ASME code is mandatory when operating
pressure exceeds 2 atmospheres (203 kPa).
 Facts about Liquid Filled Transformer Tanks

• C57.12.10 requires a pressure relief device to

be mounted on the tank cover
– Typically activate at 34.5 to 69 kPa (5 to 10 psig)

• Most rectangular tanks will sustain internal

pressures of 140 to 210 kPa (20 to 30 psig)
before rupture

• A tank on its own cannot be made strong

enough to resist all magnitudes of internal
pressure
Transformers have been saved from
rupture by common protection devices

• Pressure Relief Devices

• Gas Detector Relays
• Rapid Pressure Rise Relays
• Real time Gas Monitors

Provided the alarm signals are quickly

recognized and the transformer is
de-energized
Pressure Relief Device in Action
 Not Every Tank Rupture Results in a Fire
Unit gassing, GDR alarmed, long delay in response to alarm,
ambient -30°C, reaction force to rupture shifted unit on the pad.
Some Units Burn After Tank Rupture- 750 MVA 500kV
Same Rating Different Supplier
 The Location Where a Tank will Rupture is a
Function of:

1. the rate of change of the pressure increase

Slow rate:
• Pressure has time to distribute throughout the tank
Fast rate:
• Results in a pressure build-up at the source

2. the co-ordinates of the pressure source

inside the tank

3. the closest weak spot to those co-ordinates

 Slow Rate of Pressure Increase
- Failed at 2 locations on the Cover
- Tensile end reactions tore the welds at the ends of the stiffeners
- Cover end angle rotated and the weld to tank flange cracked
 Moderate Rate of Pressure Increase

• Tank failed at
the cover joint
• Some stiffeners
on the tank wall
were permanently
deformed
• There were no
cracks in the oil
containment
welds in the tank
body
 Rapid Rate of Pressure Increase

• Co-ordinates of
pressure source was
about 1/3 tank height

• Wall plate fracture

started at the corner
welds and ran almost
full height of the tank

• Tank wall to bottom

weld joint also failed
 Rapid Rate of Pressure Increase

• Cover weld did not fail

• Tank failed at the high

stress points in welds at
the tank corners and
wall penetrations then
propagated through the
wall plates

• Unit was returned to

the factory and rebuilt
 Weakest Points of a Rectangular Tank

• Main cover to tank wall flange - weld joint

• Tank wall corners - weld joint

• Tank wall to base plate - weld joint

• High stress points at throats and large

penetrations through the tank plates

• High stress points at ends of stiffeners where the

end reaction force is transmitted to the tank plate
Main Cover to Tank Wall Flange
Mode of Weld Failure

Typical Cover Welded Joint

Upward and outward forces

due to internal pressures

Stress is concentrated at the

weakest point - the root of
the weld.

Weld crack progresses

outward from the root
Typical Rectangular Tank Corner Joints

Stresses are on the face of

the weld.

The face of the weld is

stronger than the root of
the weld.

Adding corner gussets will

reinforce this joint
 Cylindrical Tanks are Inherently Stronger
than Rectangular Tanks

• 110MVAR 735 kV
shunt reactor
• The tank wall is
stressed in hoop
tension
• Typically the
cylinder walls can
sustain pressures
> 350 kPa (50 psig)
• Weakest point is
the cover weld
 Design to Help Reduce Tank Rupture

• Design in service systems to detect faults

early & de-energize quickly

• Use detection & relief accessories

– Real time gas monitors
– Gas detector relays - ensure gas collection
system / piping functions as intended
– Rapid pressure relays
– Pressure relief devices
• To be effective, relief devices must be located close to
the pressure source
• Standard size relief devices may not prevent all tank
ruptures
 Design to Help Reduce Tank Rupture

• The best location for a tank to fail is at the welded

cover joint as this will minimize fluid loss

• Strengthen the tank below the cover joint

– Reinforce tank corners and wall to bottom joints
with plates / gussets

– Distribute stiffener end reaction forces with

reinforcements or by connecting to stiffeners on
adjacent walls

– Reinforce around wall penetrations to reduce the

high stress points
 Transformer Tank Rupture and Mitigation
Tutorial
March 9, 2010

Mitigation Research and Example Techniques

Presented by Craig Swinderman
Mitsubishi Electric Power Products, Inc.
 Research on Transformer Tank Rupture
Mitigation
• Joint research performed in the mid-1980’s by three large Electric
Utilities in Japan, Tokyo University, and several transformer
manufacturers.

• Goal was to study ways of reducing the risk of transformer tank

explosion in urban substations/ underground substations.

• Full scale model testing performed.

• Arc energies calculated and gas generation rate observed.

• Pressure rise models developed.

• Tank construction countermeasures developed.

 Summary of Research

Tank Explosion Study Process for Internal Fault Verification

Process Tests
Estimation of Fault Condition
Internal Fault Arc Test in Oil Tank
① Arc Current ② Arc Voltage ③ Time
Dissolved Gas
Generation ④ Dissolved Gas Generation Volume

Internal Pressure Pressure Rise Analysis Pressure Rise Test

Increase
Dynamic Analysis considering Oil Motion using Full Scale Tx

Internal Pressure Comparison between Internal

Pressure and Tank Strength
> Tank Strength

Countermeasure
Tank
① Tank Strength Improvement ② Protection Relay Improvement
Explosion
③ System Improvement ④ Pressure Restrained Structure
 Dynamics of Internal Fault (1)
10000
Arc
ArcTest
Test in
inOil
OilTank
Tank

Arc Voltage (V)

1000
Electrode

100
Arc 10 100 1000
Arc Length (mm)
Oil Gas Volume
1

Gas Volume (m3)

Arc Current : 1.3 to 40.9 kA
Time : 3Cycle (0.05sec)
0.1
Arc Energy : 0.11 to 2.64 MJ
Gap Length : 100 to 300 mm

*Reference – T. KAWAMURA, M. UEDA, K. ANDO, Y. MAEDA, Y. ABIRU, M. WATANABE, K.

MORITSU “Prevention of Tank Rupture Due to Internal Fault of Oil Filled Transformers”, 0.01
CIGRE, 12-02, 1988.
0.1 1 10
Arc Energy (MJ)
 Dynamics of Internal Fault (2)
Pressure
Pressure Rise
Rise Test
Test using
using Full
Full Scale
Scale model
model for
for verifying
verifying pressure
pressure rise
rise
- Pressure rise at internal fault can be simulated by powder combustion, considering nozzle area of
container and powder amount.
- Dead space (steel tank) was set for simulating internal parts(Core,Coil).in the tested tank.
Measured Results
Transformer : 275kV 300MVA
Analytical Results
Arc Energy : 142,000kW、 Time : 80msec

Pressure
(Single Line-Ground Fault at Upper Tank) Upper Tank

Time
Combustion Container
Φ300,L570 Middle Tank

Pressure
Upper Tank
Gas Outlet
(Φ12*72)
Time

Lower Tank

Pressure
Time
Middle Tank *Reference – T. KAWAMURA, M. UEDA, K. ANDO, Y. MAEDA, Y.
Lower Tank
Cartridge increment ABIRU, M. WATANABE, K. MORITSU “Prevention of Tank
(Max.7500g) Rupture Due to Internal Fault of Oil Filled Transformers”, CIGRE,
12-02, 1988.
 Results of Analysis
• Decomposed gas generation calculated to be around
0.5L/kW sec for larger 275 kV class transformers at HV lead.*

• Dynamic oscillation of fault pressure wave (kinetic energy)

has a considerable influence on the pressure rise. (Dynamic
Load Factor approx. 1.3 was recommended)*

• Tank expansion characteristics and tank strength are

important in determining the transformer’s capability to
resist rupture.

• Reinforcements can be made at the joining flange between

the tank and cover (or flange between upper and lower tank
for shell-form) to significantly improve the tank strength
against rupture.

*Reference – T. KAWAMURA, M. UEDA, K. ANDO, Y. MAEDA, Y. ABIRU, M. WATANABE, K. MORITSU

“Prevention of Tank Rupture Due to Internal Fault of Oil Filled Transformers”, CIGRE, 12-02, 1988.
 Example of Tank Strength Improvement

Conventional Type Improved Type

Connecting Part
Alleviation of Stress Improvement of Connecting Part
Concentration Strength by Tie Reinforcement
 Pressure Reducing Space
• Diaphragm type Conservator Tanks can be used as
effective pressure reducing space if the connection duct to
the main tank is short, the cross sectional area of the duct
is large (approx. 1.4 m dia.), and the air space in the
conservator diaphragm is adequate

Diaphragm (bladder) Conservator Tank

Connection duct

Transformer main tank

•Tests on 300 MVA, 230 kV units have verified ability to

withstand 15,000 MVA short circuit capacity without
rupture of the tank.
•Still requires operation of protective relays and circuit
breakers to clear fault within approx. 60 - 80 ms.
 Gas Insulated Power Transformers

•Use SF6 Gas as the insulating and cooling medium

instead of insulating oil.

•First units produced in 1967.

•Several thousand units of various sizes now in

service worldwide, several manufacturers.

•Transformer applications: From Distribution class

units up to 400 MVA, 345 kV ratings.

•Primarily used in substations located in urban

areas (including inside buildings, underground) due
to safety benefits.
 Features of Gas Insulated Transformers

• Use SF6 Gas as insulating and cooling

medium, instead of oil.

• Typically use special internal insulation

materials such as plastics, special paper,
and pressboard.

• SF6 has excellent dielectric properties, but

not as good for heat transfer.
 Benefit of Gas Insulated Transformers
SF6 Gas Insulation: non-flammable, compressible gas

Pressure rise during an internal fault is slower than oil-immersed (non-

compressible fluid), thus SF6 gas reduces the chances of tank explosion
Pressure Rise (%)

100
100 Tank Strength
8080

6060
Oil-Immersed
Transformer
4040 Gas Insulated
2020 Transformer
00
00 0.0.22 0.0.44 0.0.66 0.0.88 11
Fault Time (sec)
 Application Example

15MVA, Three phase, 50Hz,

Continuous Rating,
Core-Form,
Forced-Gas Natual Air,
with On-Load Tap Changer
GNAN/GFAN

H.V. 64.5kV +10/-10% Star

L.V. 6.6kV Delta

For underground substation

beneath office building
A VERY BRIEF HISTORY:

PRESSURE RELIEF DEVICES

(PRDs) AND THEIR USE ON
POWER AND DISTRIBUTION
TRANSFORMERS

Transformer Tank Rupture and

Mitigation - March 9, 2010 J.Herz,
Qualitrol
BEFORE THERE WERE RE-SEALABLE PRDs…
Rupture Discs

„ GOOSENECK CAN ADD TO BACK PRESSURE

„ HAD TO BE REPLACED AFTER OPERATION - LEFT
THE TRANSFORMER OPEN TO ATMOSPHERIC
MOISTURE IN THE INTERIM
„ TYPICALLY WITH NO ALARM

More Recently
„ THEY HAVE BEEN USED IN MULTIPLE SETS ALONG
THE TOP AND BOTTOM OF A SINGLE TRANSFORMER
TO MAXIMIZE THE PRESSURE RELIEF AREA AND TO
BE LOCATED AS CLOSE AS POSSIBLE TO ANY
POTENTIAL FAULT LOCATION.
„ THEY ARE THE RELIEF MECHANISM FOR
COMBINATION PRESSURE RELIEF/FIRE
SUPPRESSION SYSTEMS
THERE WERE SEVERAL DIFFERENT
DESIGN APPROACHES TO RE-SEALING
PRDs MADE BY VARIOUS
MANUFACTURERS

IN THE LATE 1950’s THE INDUSTRY

SETTLED OVERWHELMINGLY ON ONE
DESIGN.
THE DESIGN IS STILL IN FREQUENT
USE TODAY. IT OFFERS SIMPLICITY
AND DURABILITY IN ADDITION TO
RE-SEALABILITY.

REFINEMENTS SINCE HAVE TO DO

WITH IMPROVING THE SEAL,
SHIELDING AND PROTECTION,
SWITCH CAPACITY, CORROSION
RESISTANCE, ETC
 WITH OPERATING PRESSURE
WHEN OPERATING PRESSURE ACTING ON THE LARGER AREA
IS REACHED, TOP SEAL CIRCUMSCRIBED BY THE SIDE
OPENS WHILE SIDE SEAL SEAL, THE SPRING IS RAPIDLY
REMAINS BRIEFLY CLOSED COMPRESSED AND THE VALVE
EXHAUSTS QUICKLY
8400 SCFM AT 50% OVERPRESSURE ON A 10 PSI PRD WAS
TYPICAL. NOW THERE ARE DEVICES WHICH GO TO 12,600 SCFM.
Most frequent question: Will it protect?

Most frequent answer is: Depends.

• location of fault
• magnitude of fault
• duration of fault
TEST 1: 1958 AT GE SCHENECTADY

„ The test tank was approximately 6 ft in diameter and 4 ft deep, with the
PRD mounted in the center of the 6 ft. diameter cover.
„ The gas space was 10 inches below the cover which resulted in about 700
gallons of oil and 23 cu. ft. of gas (air).
„ Ball nosed copper electrodes (2) were threaded with a small copper wire to
trigger the arc, the highest of which was 25K amps and 20K volts. Oil,
smoke, mist, spray blasted out of the relief device over a radius of about 40
feet.

TEST 2: 1958 AT GE SCHENECTADY

„ In tests performed by Jim Barr on a transformer with NO COVER , the fault

was introduced near the bottom of the tank and the bottom of the tank BLEW
OUT, at 10K amps and 10K volts.
There are thousands of events where rupture
discs and re-sealable PRDs have successfully
protected transformers:

“Internal arcing, breaker insulation break

down, load tap changer problems, phase
angle regulator problems, and internal
winding problems” are some of the more
common.

Transformer Tank Rupture and

Mitigation - March 9, 2010 J.Herz,
Qualitrol
 CONCLUSIONS

9 TTR is a complex problem. The severity is a function of arc location, arc I and T
as well as oil volume and tank expansion characteristics.

9 It’s possible to reduce the risk of TTR by performing modifications to the tank.

9 PRDs help to protect the tank against low energy internal arcing faults.

9 Fluids with high fire point will reduce the consequences of a tank rupture;
however it is not yet proven if these fluids will prevent tank rupture.

9 GITs will eliminate the risk of tank rupture.

9 Improved electrical protection and electrical system design can also help
prevent TTR.

9 The IEEE currently has no standards that provide guidance on TTR mitigation.
IEEE TRANSACTIONS ON POWER DELIVERY
Volume: 24 Issue: 4 Date: Oct. 2009
Page(s): 1959 - 1967

Power Transformer Tank Rupture and Mitigation

- A Summary of Current State of Practice and Knowledge
by the Task Force of IEEE Power Transformer Subcommittee
Nick Abi-Samra, Javier Arteaga, Bill Darovny, Marc Foata, Joshua Herz, Terence Lee, Van Nhi Nguyen,
Guillaume Perigaud, Craig Swinderman, Robert Thompson, Ge (Jim) Zhang, and Peter D. Zhao
 Action Next –
Planning to Generate an IEEE Std
You are all welcome to join