Professional Documents
Culture Documents
Abstract-Temperature, especially the hot spot capacity of existing transformers is very important to
temperature (HST) and top oil temperature (TOT), reduce the costs. To predict the loading capability of
has played the important rule and became the most transformers can be used top oil temperature (TOT)
efective factor in determination of the insulation life and hot spot temperature (HST) as two criteria for
of the transformer. The prediction of HST and TOT is loading these important facilities. Estimate the TOT
very importnt for estimating the loss of life (LOL) of and HST accurately is depended on recognizing the
the transformer in power system. Therefore, an parameters can effect on the transformer's
accurate technique is needed for solving the thermal temperatures such as ambient temperature, loading
models. This paper presents a numerical method curent, cooling modes, oil's viscosity, and other
which can accurately calculate the HST and thus vriable factors. Therefore, to predict HST and TOT,
provides an efective evaluation of LOL of the many models have been proposed such as the top oil
transformer. An altenative solution for solving the rise, top-oil, semi-physical model and thermal
thermal model is proposed in this work and results electrical models[3, 4].
are compared with the actual temperature, measured
To solve the classic thermal model have been
by iber optic sensors and placed on the 30MVA
suggested several methods, such as linear and non
power transformer.
linear regression as statistical methods and the other
one is the exponential method that was offered by
IEEE standard. The purpose of this research offers an
Kwords:Hot Spot Temperature;Loss of Lfe of
accurate technique to determine the TOT and HST
the Transformer; Electrical transformer
due to evaluate the loss of life (LOL) of transformers.
This technique is based on numerical methods and
I. INTRODUCTION solving thermal models by an improved numerical
Overheated in the transformer's winding can be method that is called Runge-Kutta method.
occurred because of overloading for short duration This method has been implemented on 30MVA
time or installing the additional transformer for a transformer, and MATLAB was used in this research
long time. In this case thermal conditions of the for programming the classic thermal model, because
transformer have to predict, whether the transformer of the reliability of this sotware for programming
can tolerate this overheated depend on loading, and plot the results to indicate and compare the
ambient temperature, and cooling systems or not[I]. measured and simulated results.
Forecasting transformer's temperature accurately
makes it possible to ully utilize the capacity of II. CLASSIC THERMAL MODEL
existing transformer. Hence, utilities could save
Much work has been done on developing classic
millions of dollars by using a model that allows them
thermal models. The trend of development of this
to predict these values accurately enough to achieve
thermal model for calculating the TOT has been
provided n section 2.1, and 2.2. In the way to
even a 2% to 3% increase over limits established
using traditional methods. In addition, prediction of
improve the classic thermal model has been
the temperature of the transformer successully and
introduced new parameters. Thus it can improve the
accurately can be used for maximum transformer's
previous models.
loading under diferent loading conditions, ambient
temperature, and various type of cooling[2].
A. Top Oil Rise Model
Electrical transformers include the huge amount
This model is presented in IEEE Loading Guide
of main investment in transmission and distribution
of the IEEE C57.91. standard which is a unction of
electrical systems. Therefore, ully utilization of the
the load and load loss ratios[5]. In this model the
thermal behavior of transformer is estimated as a irst the top-oil-rise model by incorporating ambient
order model, in which the top-oil-temperature rise temperature variations into this model. This
over ambient temperature is govened by the improvement is based on the diferential equation
diferential equation as in equation (1) [3, 5-7]. given by (6).
(1)
(6)
Which solution is an exponential unction of, This has the solution as follows [l3],
- (..)
Bo = ( Bu- ; )( 1-e To )+B; (2)
(7)
Where
III. HOT-SPOT TEMPERATURE FOR CLASSIC
80 Top-oil rise over ambient temperature (DC), THERMAL MODEL
8u Ultimate top-oil rise for load L (DC), The model for the hot spot temperature rise over
top oil is depending on the fact that an increase in the
8j Initial top-oil rise for t=O (DC), loading (crrent) causes an increase in the losses
within transformer windings. Therefore, there is an
To Time constant (hrs), increase in the temperature. Transient winding hottest
spot temperature rise over top oil temperature can be
t duration of load(hrs), calculated as follows [3]:
For
[I�R+l]n [
1BH = (1BH,u -1BH,;) l-exp � +1BH'i l
R+l
_
(3)
Ju -Ojl - (8)
That the initial value for hot-spot temperature rise
(4) over top-oil temperature is given by:
1 1
pu
=__
30
2012 IEEE International Power Engineering and Optimization Conference (PEOC02012), Melaka, Malaysia: 6-7 June 2012
..)HR rated hot spot rise over top oil FEQA is equivalent aging factor for the total time
, temperature (C). period,
A. Euler's Method
IV. CALCULATING THE LosS OF LIFE OF THE
Euler's method is an example of a ixed-step
TRANSFORMER
solver. It means that by choosing the discrete set of
The life of transformers is drectly proportional to values of the independent variable in order that they
the quality of insulation of transformers such as oil and divide the interval of interest into N equal subintervals.
paper, that the conditions of paper play the main rule It has been done by setting the step size h=(b-a)/N .
in the life of the transformers. Moisture, heat and That to=a ,tJ=to+h=a+h and in general, tj=tj_J+h=a+jh .
oxygen are the most important parameters that rom
these parameters the hottest temperature in transformer The last point of the discretization is tN=a+h=b. The
determines and these parameters cause deterioration of values of the dependent variable y will be chosen
paper and decrease the life of the electrical iteratively. At each step, the key mathematical idea is
transformer. the approximation of the graph of the unknown
The following equation could be used to determine unction yet) by its tangent line. The tangent line at
the equivalent aging of the transformer with regard to the point (t ,y(t » is the graph of the unction[13].
o o
the reference temperature for hottest spot isllO DC [5,
14]: (15)
31
2012 IEEE International Power Engineering and Optimization Conference (PEOC02012), Melaka, Malaysia: 6-7 June 2012
K\ =f(to,yo),
h h (22)
K, =f(to+2'Yo+2K\),
h h
K3 =f(to+2'Yo+2K,),
K4 =f(to+h 'Yo+hK3),
With these slopes the next values of the dependent
variable can be calculated as,
NO
h KI + 2K2 + 2K3 + K4
Yk = Yk-I +
(23)
6
is IN � a + h � b •
The values of the dependent variable y will be Figure 2. Algorithm for Runge- Kutta's method
chosen iteratively as in Euler's method. However, it is
not as easy to ind a geometry interpretation for
Runge-Kutta methods as it is for Euler's method[13,
16]. The second-order R-K method has been known as
the improved Euler's method, that starting rom the
32
2012 IEEE International Power Engineering and Optimization Conference (PEOC02012), Melaka, Malaysia: 6-7 June 2012
VI. THERMAL TESTS FOR CASE OF STUDY TOT for 30MVA transformer
100
//
'
transformer and cooling system is with extenal
cooling system, ONAF. This transformer has 3
70 "If
thermometers to indicate the top oil temperature and
RK4 method
winding temperature for high and low voltage RK2 method
60
windings. For gathering required data such as TOT and Euler
Measured
ambient temperature, to validate proposed technique Exponential
lO
has been ofered in this research, using the
thermometers of this transformer and connecting to
401
data logger to collect top oil temperature has been 0 10 12
100
or
90
TepUQlUe 80
indicotor
/' - RK4 method
(Top oi) RK2 method
70
Euler
] 60
Measured
Exponential
1
40 �--�-�--�-�--�-�
�0A
o 10 12
OUT'UT
Time (hrs)
JO MVA ONAF
Figure 5. Hot spot temperature of 30 MYA transformer based
on classic thermal model
'Ml8
FAA for 30MVA transformer
./
and numerical methods (include K4, and Euler's
O.l
method) of the hot spot temperatures of the 30 MVA
transformer shows in Figure 5 together with aging OA -�_�_�_�_�_�_��
40 70 ]20
acceleration factor which is shown in Figure 6. It
should be pointed out that the results are shown
graphically based on the classic thermal models.
Figure 6. Aging acceleration factor of 30 MYA transformer
base on classic thermal model
33
2012 IEEE International Power Engineering and Optimization Conference (PEOC02012), Melaka, Malaysia: 6-7 June 2012
shows the absolute diference error between the [2] D. Karsai, D. Kerenyi, and D. Kiss, Large power
measured value, and numerical methods and also transformers: Elsevier Science Pub. Co. Inc., New York,
NY,1987.
compares with exponential method.
[3] L. Jauregui-Rivera and D. Tylavsky, "Acceptability of
TABLE II.a. Results of classic thermal model of 30 MVA
Four Transformer Top-Oil Thermal Models-Part I:
transformer
Deining Metrics," IEEE Transactions on Power
Delivey, vol. 23,pp. 860-865,2008.
Loss Loss
Loss of Loss of Loss of Loss of
of of [4] K. Takami and 1. Mhmoudi, "Numerical Modelling of
life-hrs Iife% Iife-hrs life-hrs
Iife% life% Heat Generation and Distribution in the Core and
(RK4) (EULER) (EULER) (IEEE)
(RK4) (IEEE) Winding of Power Transformers," International Journal
of Emerging Electric Power Systems, vol. 9,p. 7,2008.
0.0056 10.1352 0.0055 9.9330 0.0061 11.0576 [5] IEEE, "Guide for Loading Mineral-Oil-Immersed Power
Transformers-C57.91-1995.," 1995.
TABLE II.b. Error of Loss of Ltfe [8] Z. Radakovic and K. Feser, "A new method for the
calculation of the hot-spot temperature in power
Ie LOLRk 4-LOMeasured)1 Ie LOLEuler-LOMeasured)1 transformers with ONAN cooling," IEEE Transactions
on Power Delivey, vol. 18,pp. 1284-1292,2003.
min min
[9] S. Tojo, "Proposition of individual loading guide for
0.3352 0.8905
power transformers," IEEE Transactions on Power
Ie LOLIEEE-LOMeasured)1 I<WLk2-WLesed)1 Delivey, vol. 21,pp. 1383-1389,2006.
facilities. Therefore, to increase the eiciency of the [15] 1. Glyn, Moden Engineering Mathematics: Addison
transformer and using instrument based on its rated Wesley,1992.
life, it is necessry to survey the current conditions of C. Edwards and D. Penney, Dfferential equations and
[16]
these expensive facilities. Accurate HST measures bounday value problems: Brooks/Cole Pub Co,2004.
have been given in this study. Later, evaluations of
the loss of life of the transformer according to
numerical analysis methods were presented. IEEE
standard thermal model (Classic thermal model) has
been selected as an acceptable and reliable thermal
model.
34