Académique Documents
Professionnel Documents
Culture Documents
Michel Duval
TSUG 2018
A Brief History of Transformers
-1887: Nikola Tesla
-1890: electric chair
-1892: George Westinghouse and Thomas Edison
-1893: General Electric
-1986: GE Transf. US shuts down
-1998: Westinghouse Transf. US goes bankrupt
2
Analysis of Gases in Transformers
-1917: first 213 kV transformer built in the US.
-1921: Buchholz gas alarm relays, to detect gases
formed by faults in transformers.
-1950s: -observation that gases formed in gas relays
are “combustible”.
-combustible gas detectors, borrowed from
the mining industry, to detect and measure
them.
-2018: -limits of “combustible gas concentrations”
(TDGC) still used in the IEEE Gas Guide
today.
3
Analysis of Gases in Transformers
-1960s: development of laboratory methods to extract gases dissolved in oil
itself, using:
-either vacuum and mercury: Toepler Pump (Central Europe), Partial
Degassing (Canada, ASTM Method A). Still today the most accurate methods.
-or bubbling of a gas through the oil (France, ASTM Method B).
4
Analysis of Gases in Transformers
- -then, the ”Head-Space (HS) extraction” method in 1989 (Perkin Elmer),
using the partition of gases between an oil and a gas phase in a glass vial.
6
Analysis of Gases in Transformers
- -gases formed and dissolved in oil were first obtained from oil
samples taken from transformers at regular intervals, e.g., every
year, then analyzed by the laboratory.
-today, more than a dozen on-line multi-and single-gas monitors are available
commercially, and are under evaluation by CIGRE.
8
Analysis of Gases in Transformers
(Serveron)
9
Failures in Service
-the failure rate of power transformers in service (internal failures needing
repairs) typically is 0.3% per year.
10
Failures in Service
-20 to 40 will have quick-developing gas formation and faults detectable only
with on-line gas monitors.
11
Gas Formation in Transformers
Hydrogen H2
Methane CH4
Ethane C2H6
Ethylene C2H4
Acetylene C2H2
Carbon monoxide CO
Carbon dioxide CO2
Oxygen O2
Nitrogen N2
12
Gas Formation in Transformers
13
Gas Formation in Transformers
14
Gas Formation Patterns
-are related only to the materials used and faults involved
-are the same in all equipment where these materials are used:
-power transformers, sealed or air-breathing, core or shell-type
-reactors
-instrument transformers
-bushings
-LTCs
-cables
15
The 6 Basic Types of Faults Detectable by DGA
-PD: corona partial discharges (in a gas phase)
-D1: low-energy discharges (including sparking PDs)
-D2: high-energy discharges (Buccholz alarm)
-T1: thermal faults, T<300°C (overheating)
-T2: thermal faults, 300°C<T<700°C (hot spots)
-T3: thermal faults, T>700°C (very hot spots, core)
16
The 4 Sub-Types of Thermal Faults (since 2008)
-S: stray gassing (overheating) of oil, T<200°C (chemical instability of
oils)
-O: overheating <250°C (no paper carbonization)
-T3-H: hot spots in oil only
-C: carbonization of paper >300°C
17
Occurrence of Faults in Service (CIGRE WG47)
18
Fault Identification Methods
-are all using hydrocarbon gases and H2 :
-one gas at a time (“Key Gas Method”)
-two gases (“Rogers Ratios”, “IEC Ratios”)
-three gases (“Duval Triangles 1 to 7”)
-five gases (“Duval Pentagons 1 to 3”)
-CO, CO2 and furans are then used to confirm if the fault involves
paper or not.
19
Fault Identification Methods - Limitations
-Key Gas: ~50% of wrong diagnosis
-Rogers, IEC: 30% to 20% of cases with no diagnosis.
-Triangle 1: always provides a diagnosis, mostly accurate, allows to
visually follow the evolution of faults.
Sometimes difficult with Triangle 1 to distinguish between faults PD et
S (around zone PD).
-Triangles 4 and 5 developed to detect faults S, O, PD and T3-H.
-combinations of Triangles and Pentagons to identify mixtures of
faults.
20
Fault Identification Methods (Triangle 1)
Triangle 1
21
Fault Identification Methods (Triangle 4)
Triangle 4
22
Fault Identification Methods (Triangle 5)
Triangle 5
23
Fault Identification Methods (Pentagon 1)
Pentagon 1
24
Fault Identification Methods (Pentagon 2)
Pentagon 2
25
Minimum Gas Levels for Using Fault Identification
Methods
-at least one of the gas concentration values in ppm should meet the
requirement of IEC for accuracy (± 15% above 10 ppm of gas).
-if the laboratory cannot certify the accuracy of its ppm values, the
recommendation of IEC is that at least one concentration value in
ppm is above typical 90% concentration values.
26
Mixtures of Faults
27
Detection of Mixtures of Faults in Transformers
-First clue is if there is a change in gas pattern. Second clue is if Triangles 1,4,
5 and Pentagons 1, 2 do not provide the same diagnosis. This may be an
indication of multiple faults.
28
When to Use the Triangles and Pentagons ?
-if interested only by the 6 basic types of faults and by single faults,
display the evolution of faults in Triangle 1 and Pentagon 1.
-if also interested by the 4 sub-types of thermal faults (S, O, T3-H, C),
switch to Triangles 4 or 5 and to Pentagon 2.
29
When to Use the Triangles and Pentagons ?
-if interested also to detect mixtures of faults, compare the diagnosis
provided by Triangles 1, 4 or 5 and Pentagons 1 or 2.
30
Fault Identification Methods (esters)
31
Fault Identification Methods (esters)
33
IEEE 90% percentile Condition 1 Concentration Values
(ppm)
-IEEE C57.104
(vs MVA, age, % O2
2017)
34
IEEE Status 1 (90% percentile) Concentration Values (ppm)
-IEEE C57.104
(2018)
35
IEEE Status 2 (95% percentile) Concentration Values (ppm)
-IEEE C57.104
(2018)
36
CIGRE Influence of O2/N2 Ratio on Condition 1
Concentration Values (ppm) and Occurrence of Faults
-CIGRE WG47 (2018)
37
IEEE Status 1 Maximum Change in Concentration Values
(ppm)
-IEEE C57.104
(2018)
38
IEEE Status 1 Gassing Rate Values (ppm/ month)
-IEEE C57.104
(2018)
39
CIGRE Gas Concentration Levels above Typical (Condition
1) Values (ppm)
-CIGRE TB # 443
PFS = probability of having a
failure-related event in service
in %, vs concentration of
gases in ppm
P = prefailure value
T = typical (condition 1) 90%
percentile value
40
CIGRE Gas Concentration Levels above Typical (Condition
1) Values (ppm)
-CIGRE TB # 443 (2010) / WG 47 (2018)
-for a transformer to move into higher condition level, only one gas
needs to exceed its gas level.
CIGRE Table 1
41
CIGRE Gassing Rates Levels above Typical (Condition 1)
Values
-CIGRE TB # 443 (2010) / Rev WG 47 (2018)
-Manual sampling; with on-line monitors, multiply values by 10.
-Typical rates in ppm/year; other rates in ppm/month
CIGRE Table 2
42
CIGRE Recommended Actions on the Equipment vs Gas
Levels
-CIGRE TB # 443 (2010) / Rev WG 47 (2018)
43
CIGRE 90% percentile Typical (Condition 1) Concentration
Values (ppm) vs type of fault
-CIGRE WG47 (2018), calculated from WG DGA database.
-gases not indicated in Table 3 are not affected by type of fault.
CIGRE Table 3
44
CIGRE Concentration and Rate Levels above Typical
Values vs Type of Fault
-the influence of type of fault can reasonably be assumed to be the
same for intermediate and PF values as for typical values:
45
CIGRE Concentration and Rate Levels vs Specific Locations
of Faults
-in case of faults D1 in oil only, high levels of C2H2 (e.g., 1400 ppm) are
observed without failure yet.
-therefore prefailure values, as well as typical and intermediate values
of C2H2 in CIGRE Tables 1 and 2, can be multiplied by 1400/450 (with
450 = PF value of C2H2 in Table 1 when the location of fault is not
known). The corresponding typical value of C2H2 becomes 8 (1400/45)
= 25 ppm
47
CIGRE Typical and PF Values vs Type and Location of Fault
48
Exceptional Gas Values in Transformers
-CIGRE cases of exceptionally high values of gases without failure,
observed in less than 1% of DGA cases:
49
Exceptional Gas Values in Transformers
-CIGRE cases of exceptionally low values of gases followed by failure,
probably because it occurred suddenly between two oil samplings.
Observed in less than 0.5% of DGA cases.
50
Exemples of Stray Gassing in Oil S
-sealed transformer with coated paper
52
Exemples of Stray Gassing in Oil S
-faults S (160 C) in Korea (H.Seo)
53
Exemples of Stray Gassing in Oil S
-90% typical values in wind farm transformers (CIGRE WG47)
54
Exemples of Stray Gassing in Oil S
-90% typical values in bushings (CIGRE WG47), with signatures of faults S
55
Exemples of Corona DPs in Bushings
-very high values of H2 in bushings have been reported in India without
failure (CIGRE A2-208, 2018), between 15,700 and 76,052 ppm, due mostly to
corona PDs, sometimes in combination with stray gassing S.
56
Distinction between Stray Gassing in Oil S and Corona PD
-Stray gases formed by some oils at 120 °C are sometimes very close to the
boundary between zones S (stray gassing) and PD (corona partial discharges)
of Duval Triangles 4 and Pentagons 1-2.
-A PD test on the whole transformer may also confirm or not the presence of
corona PDs.
57
Distinction between Stray Gassing in Oil S and Corona PD
(S.Eeckoudt)
58
Distinction between Stray Gassing in Oil S and Corona PD
(T.Bucchaz)
59
Examples of Overheating Faults O < 250°C
(E.Alzieu)
60
Examples of Overheating Faults O < 250°C
-fault O in 345 kV, 460 MVA GSU transformer, with 1216 ppm C2H6, 416 ppm CH4,
no failure (W. Johnson)
61
Examples of T3 Faults in Oil Only (T3-H)
-rectifier
- 20kV, 20 MVA
(A.Nunez)
63
Examples of T3 Faults in Oil Only (T3-H)
(A.Constant)
64
Examples of Faults C in Leads
-transformer 24kV, 40 MVA
65
Examples of Faults C in Leads
(A.O’Malley)
66
Examples of T3 Faults C in Leads and T3-H
(A.O’Malley)
67
Examples of Faults T3-H and C in Leads
(O.Amirouche)
68
Examples of Faults C in Leads and T2-H
(M.Foata)
69
Examples of Faults C in Leads
(S.Dorieux)
70
Examples of Faults C in Leads
(Serveron)
71
Examples of Faults C in Pressboard
(S.Dorieux)
72
Examples of Faults C in Windings
73
Examples of Faults C in Windings
74
Examples of Faults C in Windings
-voltage transformer 400kV
-burnt paper in Faraday cage and upper part of windings found by inspection (C)
(M.Martins)
75
Examples of Faults C in Windings and Leads
(A.M.Dale)
76
Example of Fault C in Winding Turn
(A.Fieldsen-Roxborough)
77
Example of Fault C in Winding Turn
-fault due to the carbonization of paper on middle winding turn of reactor, because of
wrongly placed washer in oil cooling duct (design problem), followed by arcing.
-fault wrongly attributed to low DP of paper (200) on winding turn.
78
Examples of Faults D1 in Oil
79
Examples of Faults D1 in Oil
80
(A.Constant)
Exemples of faults D1 in Paper
-Bushing 230 kV (Omicron)
81
Exemples of faults D1 in Paper
-Bushing 230 kV
82
Exemples of Faults D1 in Paper
-transformer 400 kV (E.Alzieu)
83
Exemples of Faults D1 in Oil
84
(S.Spremic)
Exemples of faults D1 in Paper
-overheating of leads and windings and arcing D1 in windings found by inspection (S.Spremic)
85
Exemples of faults D2 in Paper
(O.Amirouche)
87
Example of Fast Occurring Fault C in Paper
-fast occurring fault C in paper confirmed by
inspection in bottom of LV windings (L.Paulhiac).
88
Example of Fast Occurring D2 Fault
Day 2 – 16:00
Day 3 – 12:00
Day 2 – 12:00
Day 3 – 04:00
Day 3 – 00:00
Day 2 – 20:00
Day 3 – 16:00
Day 3 – 08:00
Day 23 – 04:00 to
Day 24 – 08:00
Followed by
transformer failure
(Serveron)
89
Example of Fast Occurring D1 Fault
(Serveron)
90
Advantages/ Limitations of On-Line Gas Monitors
-will catch abnormal formation of gases and quick-developing faults occurring
suddenly between two oil samplings, whatever the time interval (year or week).
91
General Types of On-Line Gas Monitors
(CIGRE WG47)
92
Faults Detectable by On-Line Gas Monitors
93 (CIGRE WG47)
Which Monitors for Which Transformers?
(CIGRE WG47)
94
Advantages/ Limitations of Multi-Gas Monitors
-Multi-gas monitors are able to detect all types of faults, even in their early
stages at condition 1, and without false alarms since they provide DGA
diagnosis on-line.
95
Advantages/ Limitations of Hydrogen Monitors
-Hydrogen monitors are less expensive than multi-gas monitors.
-however, they do not provide DGA diagnosis on-line and will not detect arcing
faults D1 and thermal faults T3, T2, C, O in their early stages, only in their late
stages, sometimes too late.
96
Limitations of Hydrogen Monitors
-Hydrogen monitors are very sensitive to faults S and corona PDs, which are of
relatively minor concern in power transformers.
-in case of faults T1, O, C, T2, T3, T3-H, much lower levels of H2 are formed than of
the other hydrocarbons gases.
-in case of faults D1, D2, most hydrogen monitors cannot detect the low H2 levels (~
6 ppm) associated with the typical C2H2 formations (~ 2 ppm) to detect.
-choosing threshold values of H2 in H2 monitors may therefore be a challenge.
97
Limitations of CO Monitors
-interpretation of CO readings is the main challenge of monitors M2.
-e.g., sealed transformers may have high concentrations of CO not due to a fault in
paper, but to oil oxidation under conditions of low O2 supply.
-in case of faults involving a small amount of paper (see examples above), CO
values often are not be high enough to be detectable, and require knowledge of
hydrocarbon gases.
-finally, in order to confirm a fault in paper, CO only is usually not enough, and
requires knowledge of the CO2/CO ratio.
98
Cost/ Benefit of DGA Monitoring Techniques
-based on a population of 2000 power transformers with a failure rate of 0.3%
per year, i.e,
-1800 of them with normal gas formation,
-200 abnormal gas formations,
-20 to 40 fast occurring gas/ fault formations,
-6 failures
-less than 1 catastrophic failure.
99
Cost/ Benefit of DGA Monitoring Techniques
-cost of not monitoring at all by DGA (including laboratory DGA): 35 M$
100
Fault Identification Methods for LTCs
101
Examples of Faults in Compartment-Type LTCs
102
Example of Fault in an MR-LTC
104
Accuracy of DGA Analysis and Diagnosis
-Method C1 = Head Space of IEC 60599
105
Accuracy of DGA Monitors
-CIGRE WG47
106
Accuracy of DGA Analysis and Diagnosis
107