Sunteți pe pagina 1din 4

2008 International Conference on Condition Monitoring and Diagnosis, Beijing, China, April 21-24, 2008

Transformer Hotspot Temperature Calculation


using IEEE Loading Guide
Mohd Taufiq Ishak and Zhongdong Wang*
1
School of Electrical and Electronic Engineering, University of Manchester, M60 1QD, UK
*E-mail: zhongdong.wang@manchester.ac.uk

 A basic thermal model for power transformers is given in


Abstract— The life of a transformer depends mainly on the figure 1, where it is assumed that the oil temperature inside the
life of its insulation system. Ageing is a process mainly defined by windings increases linearly from bottom to top, the
the chemical reaction which primarily depends on temperature temperature difference between the duct oil and winding
and time. The hotspot temperature is usually taken into
conductor is constant throughout the winding. The hotspot
consideration in investigating the remaining life of transformer.
There are two models available in IEEE Loading Guide in temperature rise is higher than the temperature rise of
calculating the hotspot temperature known as Clause 7 and conductor at the top of the winding as shown in figure 1 when
Annex G methods. This paper presents a comparative study considering the increases due to stray losses.
between the two methods, through calculating hotspot
?HS
temperatures of the four transformers using ONAN, ONAF, Top of Top oil Hot spot
OFAF and ODAF cooling modes under conditions of constant the temperature temperature
winding
load, daily load and ambient temperature profile, and a short-
period overload. The loss of life is also estimated.
g
Average Average winding
Index Terms— Thermal Ageing, Hotspot Temperature, IEEE oil temperature
Loading Guide, Loss of Life, Overloading, and Power temperat
ure
Transformers.

Bottom of
I. INTRODUCTION the Bottom
oil
winding

H
temperat Temperature
otspot temperature is one of the most important ure rises(C)
parameters when defining thermal condition and
Figure 1: Basic thermal model
overloading capability of a transformer. Thermal ageing and
The basic calculation method, given as an example in (1)
degradation of the insulation affects the transformer lifetime,
for an ON transformer, relies on the measured top oil
since cellulose degradation caused by overheating is
temperature and the approximation of the temperature
irreversible. It is widely accepted that the insulation
difference between the hotspot temperature and the top oil
degradation approximately doubles with every 6°C increase of
temperature.
temperature [1]. Therefore an accurate prediction of hotspot 2m
§ I · (1)
temperature is important for both manufacturers and utilities. 4 HS 4 TO  ' 4 HR ¨ ¸
There are a few methods to measure the hotspot temperature, © IR ¹
one of which is using fibre optical temperature sensors where HS is the hotspot temperature (°C), TO is the
positioned at the predicted hotspot of the windings. The measured top-oil temperature (°C) and HR is the rated
thermal sensors, attached to the end optical fibre, are usually hotspot temperature rise over top-oil temperature rise (°C)
placed between the insulated conductor and spacer, and their obtained during the heat-run test, and m is the winding
signals via optical fibre transmitted out of the tank. Significant exponent – an empirical parameter recommended by the
improvement over the years overcomes the fibre optic being loading guides.
too fragile and needing delicate handling. By using this However, this traditional hotspot temperature calculation
method, accurate measurement of hotspot temperature can be method has been shown inadequacy when compared to the
done. actual data measured by fibre optic sensors[4, 5]. It has been
IEEE and IEC Loading Guides [2, 3] have been used to noticed by [6, 7] that during overload conditions there is a
calculate hotspot temperature, by using data obtained through time lag between the top-oil temperature rise and the duct oil
heat-run test and empirical parameters. temperature rise. This phenomenon results in winding hotspot
temperature greater than predicted by the IEEE Loading
Mr. Ishak appreciates the sponsorship given by University Tun Hussein Guide Clause 7 method [3]. In [6] equations were proposed to
Onn, Malaysia to allow him studying at the University of Manchester for the account for the duct oil temperature, the change of winding
PhD degree in Electrical Power Engineering. resistance with temperature, the change of oil viscosity with

978-1-4244-1622-6/08/$25.00 ©2007 IEEE


temperature, the effect of tap position and also the variation of previous calculation at time t1 are used to compute the
ambient temperature during the load cycle, which have been temperatures at the next instant of time t1+t. At each time
ignored in the Clause 7 method. These improved equations step the losses, the winding resistance and the change of fluid
were included in the IEEE Loading Guide [3] as an alternative viscosity are recalculated for the load.
known as the Annex G method for the calculation of hotspot
temperature. IV. RESULTS AND DISCUSSIONS
In this paper, a comparative study has been made between Three case studies were made in order to compare the
the two methods available in the IEEE Loading Guide to hotspot temperature calculation methods under 1) constant
calculate the hotspot temperature [3]. Four transformers with load condition, 2) typical daily load and ambient temperature
different cooling modes of ONAN, ONAF, OFAF and ODAF, profiles, and 3) four hours overload condition. The loss-of-life
of which the data are given in appendix, were used to assess under these conditions was also calculated.
the two methods under three case studies of constant load,
daily load profile and short-period overload conditions. ONAF A. Case 1: constant load condition
and OFAF calculation cases were verified by the thermal tests The hotspot temperature is allowed to reach its steady state
and measurements done by [6, 7]. Hotspot temperature and the value, under five constant load conditions of 0.8, 1.0, 1.2, 1.4
consequential loss-of-life were studied using a MATLAB and 1.6 per unit load. The ambient temperature is set constant
simulation program. as 30°C. The calculated hotspot temperatures for the four
transformers under ONAN, ONAF, OFAF and ODAF are
II. CLAUSE 7 CALCULATION METHOD shown in figure 2.
The Clause 7 equations are based on the assumption that ONAN Example, Steady state, Tamb=30 deg C ONAF Example, Steady State, Tamb=30 deg C

the temperature of the oil exiting the winding ducts is the 180 180

160 160

140 140

same as the top-oil temperature. The hotspot temperature can

Hot Spot temp -deg C


Hot spot Temp-deg C
120 120

100 100

be calculated as: 80 80

TH T A  ' T TO  ' T H
60 60

(2) 40

20
40

20

where LJH is the hotspot temperature (°C), LJA is the ambient


0 0
0.8 1 1.2 1.4 1.6 0.8 1 1.2 1.4 1.6

Load (p.u) Load ( p.u)

temperature (°C), LJTO is the top-oil temperature rise over the OFAF Example, Steady state, Tamb=30 deg C ODAF Example, Steady state, Tamb=30 deg C

180 180

ambient (°C) and LJH is the winding hotspot temperature rise 160 160

140 140

over the top-oil (°C).

Hot spot Temp- deg C


Hot spot Temp- deg C

120 120

100 100

The detailed calculation methods for each temperature rise 80

60
80

60

component are given in [3]. The ambient temperature is 40

20
40

20

assumed to be constant at 30°C. The equations use an 0


0.8 1 1.2
Load (p.u)
1.4 1.6
0
0.8 1 1.2
Load (p.u)
1.4 1.6

exponential expression containing a time constant to Clause 7 Annex G


determine the variation of the top-oil temperature from the
Figure 2: Hotspot temperatures under steady state loads
initial to the final steady state conditions. The time constant
changes with load and must be corrected for the load change
at each calculation step. As shown in figure 2, for the ONAN and ODAF cooled
transformer, the calculated hotspot temperature by the two
III. ANNEX G CALCULATION METHOD methods show a good agreement up to 1.0 pu of load, whereas
for the ONAF and OFAF cooled transformer a good
The Annex G equations are based on the fluid flow
agreement up to 1.4 pu of load. In the graphs for both the
conditions occurring in the transformer during transient
ONAF and ODAF transformer, hotspot temperature exceed
conditions. The hotspot temperature is made up of the
180 °C under 1.6 pu of load and the transformer thermal
following components:
overload protection will trip before this temperature is reached.
TH T A  T B O  ' T W O / B O  ' T H /W O (3) It is shown that the Clause 7 method significantly
where  BO is the bottom oil temperature rise over the ambient underestimates the temperature rises for ODAF cooled
(°C) which can be measured during heat-run test, WO/BO is transformer as compared with the Annex G method. In Clause
the oil temperature rise over the bottom oil at the hotspot 7 method, the increase in winding resistance with temperature
location (°C), and H/WO is the winding hotspot temperature and the decrease in oil viscosity with temperature are assumed
rise over the oil at the hotspot location (°C).The detailed to offset each other. This assumption may be true for the other
calculation methods for each temperature rise component are cases except for directed oil situation, as oil velocity flowing
given in annex G of [3]. through winding ducts is controlled mostly by the pump, and
For overload conditions, when the duct oil temperature is heat transfer rate is not dominated by the decreased viscosity
less than the top oil temperature, WO/BO is assumed to be of oil. This is the reason why the largest discrepancy occurs
equal to the top-oil temperature rise. with the directed oil case.
The temperature rise in (3) is obtained for the conservation
of energy during a small time internal t. The equations are
formulated so that the temperatures obtained from the
B. Case 2: typical daily load & ambient temperature profiles implication is that under overload condition the loss of life for
Typical daily load curve and ambient temperature profile a transformer may be higher than expected. This will lead to
were given as in figure 3. The load had a maximum of 1.1 pu an underestimate of the risk of overheating and bubble
for 2 hours. The hotspot temperatures for the four formation.
ONAN Cooling ONAF Cooling
160 160
transformers are calculated using the two methods and shown 150 150

in figure 4. 140 140

130 130

Hot Spot temper ature , C

Hot Spot temper ature , C


120 120

110 110
1.5
100 100

90 90
40
80 80

70 70
AmbientTemperature

Load current(per unit)


60 60
600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800
Time (minutes) Time (minutes)
1
OFAF Cooling ODAF Cooling
160 180
30
150
160
140

130

H ot Spot temper ature, C

H ot Spot temper ature, C


140
120

110 120
20 0.5
0 5 10 15 20 25
100
Time (h)
100
90

Figure 3: Typical daily load and ambient temperature profiles 80


80
ONAN Co o li n g ONAF Co o li n g 70
120 120

110 60 60
110 500 1000 1500 2000 2500 3000 500 1000 1500 2000 2500
Time (minutes) Time ( minutes)
100 100
Clause 7 Annex G
Ho t Sp o t te mp e ra tu re , C

Ho t Sp o t te m pe ra ture , C

90 90

80 80 Figure 5: Hotspot temperatures for four hours overload


70

60
70

60
condition
50 50

40 40
25 30 35
Tim e( h o u rs)
OFAF Co o li n g
40 45 50 25 30 35
Ti m e(h o u rs)
40 45 50
V. CALCULATED LOSS- OF -LIFE FOR TRANSFORMERS
ODAF Co o li n g
120 140

110
130

120
Experimental evidence indicates that the relationship
100
110
between insulation degradation to time and temperature
t m p era tu re, C

H ot Sp o t te mp e ra tu re , C

90 100

90 follows the Arrhenius reaction rate theory.


Ho t Sp o t e

80
80

70 70 IEEE Loading Guide [3] defined the aging equation as:


60
60
50
18 ­ 15000 ½ (4)
50 40 Per unit life =9.80 u10 exp ® ¾
¯ 4 H  273 ¿
30 35 40 45 30 35 40 45
Ti m e (h o u rs) Ti m e(h o u rs)

Clause 7 Annex G

Figure 4: Hotspot temperatures under daily load condition Normal life for most transmission transformers is
considered to be around 40 to 50 years, due to the
For typical daily load and ambient temperature profiles, conservative way of operating transformers by the utilities.
there results obtained show that the Annex G method predicts The hotspot temperature for thermally upgraded cellulose to
higher hotspot temperature than the clause 7 for all the cases attain a one per unit life is 110°C based on an assumed
except for OFAF cooled transformer. The hotspot ambient temperature of 30°C plus 65°C average winding
temperatures calculated by both methods show very close to gradient over ambient plus a 15°C allowance for the hot spot
each others. gradient over average winding temperature. Using hotspot
temperature of 110°C yields a per unit life=1, and lower than
C. Case 3: four hours overload condition 110°C yields a more than 1 per unit life, indicating an
In this case study, it is assumed that four hours overload of extended lifetime.
1.4 pu occurs on a transformer, with a constant loading Instead of using the per unit life, the IEEE Loading
condition of 0.7 pu before and after this overload period. The Guide[3] introduced the idea of relative aging rate. The
ambient temperature is assumed constant at 30°C. The hotspot accelerating aging factor in (5) can be derived and is defined
temperatures for the four transformers were calculated using as:
the two methods and the results are shown in figure 5. ª B B º
«  »
As shown in figure 5, the Clause 7 method produces lower ¬« ( 4  273) (4  2 7 3 ) ¼»
FA A = e H ,R H
(5)
hotspot temperature than the Annex G method, except for the
where FAA is insulation aging acceleration factor, B is an
OFAF cooled transformer. The test data for this OFAF
ageing rate constant, typically 15000, and H,R is winding
transformer reported in [7], indicated that the hotspot
hotspot temperature at rated load (95°C if W/A,R=55°C and
temperature was 9.1 to 13.5°C lower than what was calculated
110°C if W/A,R=65° C )
by the Clause 7 method. It seems that for this case that the
If the value of FAA is greater than 1, the hotspot temperature
Annex G method agrees with the test result. Once again, for
is greater than 110°C, suggesting accelerated loss of life.
the ODAF transformer, the Clause 7 method underestimates
Since insulation ageing is a cumulative process, an
the hotspot temperature with a big difference as compared to
equivalent ageing acceleration factor can be expressed as a
the Annex G method. This comparison indicates the
summation of (5) over discrete time intervals, averaged over
limitation of the Clause 7 approach in its use to estimate the
the total time elapsed [3]:
hotspot temperature under overload conditions. The
N
VII. REFERENCES
¦F
n 1
AAn 'tn (6) [1] McNutt, W.J., Insulation Thermal Life Considerations for
FEQA N
Trasnformer Loading Guides. IEEE Transactions on Power
¦ 't
n 1
n Delivery, 1992. vol.7(no.1): p. 392-401.
[2] IEC 354:1991, Guide To Loading of Oil-Immersed Power
where FEQA is the equivalent aging factor for the total time Transformers. 1991.
period,t, 'tn is the time interval in hour, N is the total [3] C57.92-1995, I., IEEE Guide for Loading Mineral-Oil-Immersed
Transformers. IEEE Standard C57.92-1995, 1995. Institute of
number of time intervals, FAA,n is the ageing acceleration Electrical and Electronic Engineers (New York).
factor for the temperature which exist during the time interval [4] CIGRE WG 12-09, Direct Measurements of the Hot Spot
Temperature of Transformer. Electra 1990, 1990. Vol.129: p.
'tn . pp47-51.
The calculation of loss-of-life for case study 2 and case [5] Nordman H., Hironnimi E., Pesonen A.J., Determination of Hot
Spot Temperature Rise at Rated Load and at Overload. CIGRE
study 3 are shown as in table 1 and table 2. In table 1, the Paper, 1990. 12-103.
estimated loss-of-life over a daily load and ambient [6] Pierce, L.W., Predicting Liquid Filled Trasnformer Loading
temperature profile for all the four transformers are not greater Capability. IEEE Transactions on Industry Applications, 1994.
than 1 pu. Vol.30 (1): p. 170-178.
[7] . Pierce, L.W., An Investigation of The Thermal Performance of An
Table 1: Calculated loss-of-life under daily load Oil Filled Transformer Winding. IEEE Transactions on Power
Delivery, 1992. Vol.7 (3): p. 1347-1358.
Loss of ONAN
life, (hours) ONAF OFAF ODAF
Clause 7 0.20 1.00 0.16 0.38 APPENDIX: TRANSFORMER DATA
Annex G 0.15 0.84 0.18 0.21

The two methods show a reasonable agreement since the Unit ONAN ONAF OFAF ODAF
MVA Base for
loss-of-life is cumulative over a typical daily load and ambient Loss data MVA 80 250 605 200
temperature where the overloading condition does not occur Temperature
base for loss data °C 75 75 85 75
frequently. Winding I2R
Table 2 shows the loss-of-life for the four transformers Losses W 166434 411780 929800 525072
under a four hours overloading condition. The two methods Winding Eddy
Losses W 17340 29469 285000 0
yield quite different results since the ageing rate is sensitive to
Stray Losses W 24796 43391 71000 0
the hotspot temperature and the largest difference occur for
Core Losses W 0 0 0 54560
the ODAF transformer since the Clause 7 method produces
lower hotspot temperature than the Annex G method. Cooling mode
type 1 2 3 4
Nameplate MVA 80 500 605 200
Table 2: Calculated loss-of-life during 4 hours overload Rated Average
winding rise °C 65 65 65 65
ONAN ONAF OFAF ODAF Test average
Clause 7 2.25 16.76 5.42 24.84 winding rise °C 42.66 41.7 44.7 50.6
Annex G 3.34 55.53 2.48 121.87 Rated hotspot rise °C 55.2 58.6 56.9 62.2
Rated top oil rise °C 34.2 38.3 33.4 32.2
VI. CONCLUSIONS Rated bottom oil
rise °C 19.7 16 24.6 29.2
The winding hotspot temperatures obtained from the Clause Rated ambient
temperature °C 30 30 30 30
7 method has been compared to Annex G method. It shown
that the Annex G method yields higher hotspot temperatures Winding
compared to the Clause 7 method, and these higher hotspot conductor 2 2 2 2
temperature agreed with the test data reported. This shows that Per unit eddy loss
at winding hotspot 0.5 0.4 0.71 0
inadequate estimation methodology and underestimation of Winding time
hotspot temperature in the Clause 7 method. However, the constant min 8 6 10 5
Per unit winding
Annex G method is more complicated as it requires a height to hotspot 1 1 1 1
measured top oil temperature rise and bottom oil temperature
rise, which are normally not supplied by the manufacturer Weight of core
during the heat-run test. To accurate estimate hotspot and coil kg 66690 153606 195152 225500
Weight of tank
temperature and the related loss-of-life, an improved and fittings kg 19833 44096 50153 102600
specification on thermal aspects of a power transformer is Fluid type 1 1 1 1
needed. Weight of oil kg 34800 73887 79746 21696

S-ar putea să vă placă și