transformer evalution overloading conditions

7/31/2019 Transformer Evalution Overloading Conditions

1/6


2/6

n is the empirical exponentTOis the oil time constantt is the timef is a function of: TO,U, TO,i [1]

___________________________________________

_________Jos A. Jardini and Luiz C. Magrini are with SoPaulo University,So Paulo, BrazilJos L. P. Brittes and Marco A. Bini are with CPFLCompanhia

Paulista de Fora e Luz, Campinas, Brazil

sensors on which pertinent test to determine itsparameters, were carried on. The results obtainedin such an evaluation will be discussed in thispaper.

II. TESTS PERFORMED

During the construction of the above mentioned138/13.8 kV, 30/40 MVA, ONAN/ONAF typetransformer a number of 8 embedded temperaturesensors with fiber optics cables, were installed,aside to the top oil temperature, thermal image,radiator input and output sensors.The additional sensors location in the transformer,can be seen in Fig. 1.

Fig. 1 - Temperature sensors Iocation

The sensors are: 1-LV (low voltage) top winding,second layer; 2-LV top oil duct; 3-LV top winding,

third layer; 4-LV bottom oil; 5-HV (high voltage)top winding; 6-HV top winding; 7- HV top oil duct;8- HV bottom oil

The following tests were carried on:

II a) (ONAF) 100%, 70%, 116.5% (of ratedload,

40MVA),II b) (ONAF) load profile values varying from

30% to100% on hourly basis. In the 11 th step

the load wasraised to 160% for 30 mm, and in the21st step to170% for 15 mm (see Fig. 2)

II c) Load losses at rated load and no loadlosses evaluation

III. RESULTS

A.Losses Evaluation (item II c above)

The values recorded, were:

No load losses = 17.8 kWRated load losses = 244.9 kWR = 13.72

B. Ultimate Temperature (item II a)

Table 1 shows the ultimate temperature rises aboveambient for the above (II a) test.

Load % DTOTop oil

DLVLow voltage

winding70 26.5 37.5100 41.4 57.0

116.5 50.9 69.4Table 1 - Ultimate Temperature Rise above ambient ( C)Note: TO = 2 hours, at 100% load

W= 10 mm (hot resistance measurement)

C. Load Profile (item II b)

During this test, each loading step was applied for a 1

hour period hence the time for temperature stabilizationwas not enough. Consequently, in order to obtainultimate values of temperature rise an exponentialtemperature curve was fitted to values recorded, andextrapolated. Therefore, an error may have beenappeared during this process. Results corresponding tothis test are shown in Fig. 2.


3/6

Fig. 2 - Load profile test result

The ultimate values obtained from tests II a and IIb, can be seen in Fig. 3 and 4, next:

Figure 3 - Ultimate top oil temperature rise (DTO)above ambient

Figure 4 - Hottest spot temperature rise (DLV)

above ambient, Low Voltage winding.

Figure 5 - lhe 100% loading test results

From Fig. 5, it can be verified that: the top oil ducttemperature (DOD) follows the winding temperature(see: DLV - DOD); the oil duct minus the top oiltemperature, is initially high but converging latter to atemperature of 4C (see: DOD - DTO).

Such results in Fig. 5 duly converted into equationsare intended to be used in an real time system forevaluating, the transformer temperatures thus itsadequacy for temporary overloading.To define a calculation method to be utilized, thefollowing attempt and respective results, have beencarried on:

Procedure 1: Calculation of the top oil and hottest spottemperatures, as indicated in items I, II, equations (6),(2), (3), (7), and (5) the latter modified in which H =H,U. The errors pertaining to this evaluation(measured value minus calculated value) can be seen

in Fig. 6.


4/6

Figure 4 Hottest spot temperatura rise (DLV)above ambient, Low Voltage winding.

Note: The upper curve has a margin of 5C to thelower.

The equations for fihied curves in the abovefigures are:

DTO = 0,00105 p2 + 0,3043 p + 0,0082(6)DLV = 0,00127 p2 + 0,4429 p + 0,0307(7)

p: loading in percent of 40 MVA

The correspondent values of m, n could beevaluated, however, that will not be necessarybecause equations (5), (6) above can be used in

substitution to (1) and (4).

IV. ANALYSES OF RESULTS

Fig. 5 shows the measured temperature riseabove ambient during the 100% loading step

Procedure 2: Calculation using the alternativemethod presented in [1]. The errors in this case areshown in Fig. 7.

Procedure 3: Calculation of top oil temperature, usingthe method presented in [4]. lhe errors in this case are

shown in Fig. 8.

Figure 6 - Errors when applying Procedure 1(measured minus calculated values)

In the Fig. 6 (Procedure 1) as for the top oiltemperature, the errors are less than 3C except

for cases with high power steps in which errorsare less than 8C. Most ofthe errors are negativewhich shows that the calcuLations may lead tovalues on safety side.The hottest spot temperature errors are less than5C, except aiso for high power steps for whichthe values obtained were less than l5C.

Figure 7 - Errors when applying Procedure 2

It can be seen in Fig. 7 (Procedure 2) that thismethodology presents good results with errorsless than 3C and 5C respectively for the top ouand hottest spot; except again for high powersteps (5C and 10C).

Figure 8 - Top oil temperature errors when applyingProcedure 3

In this approach (Fig. 8), the resulting valuesdepend on the adopted values for n and TO. In theabove figure, TO= l20min, and with it the less errorvalues are obtained when n = 1. It could also beused another value of n, less than the latter, toobtain a similar result, however the value ofTO willneed to be increased.For all the analysis conducted (Procedures 1,2,3),high errors appear at the beginning of the powervariation. To identify the previous fact, Fig. 9displays the results for 100% loading case along

with the stated calculation Procedures and themeasured values.


5/6

As can be observed, all the approachesconsidered lead to adequate vatues whilecalculating the top oil temperature. In relation tothe hottest spot temperature, Procedure 2

presents better results.To improve the values obtained in Procedure 1, itwould be necessary to add in the equations afunction that may represent the correction curveshown in Fig. 9. Such a function being:

Values corresponding Tc, TO should bedetermined in a future anaiys is.

V. CONCLUSIONS

The calculation methodologies of top oiltemperatures so far presented, have shownsatisfactory results.However, the equations to obtain hottest spottemperatures will need to be improved. Thismainly referred to short intervals of high power,which is of our concern, and where the ultimatevalues are not achieved, so far.The use of Annex G from the IEEE Std C57.91-1995, provided results close to those measured inthe transformers, therefore, it can be used to

obtain quite a reliable results.VI. REFERENCES

[3] PC 57.119 Draft Recommended Practicefor Perforrning

temperature Rise Tests on Oil ImrnersedPower Transformers at Loads BeyondNameplate Rating, IEEE, October 1996

[4] Swift G, Molinski T. S., Lehn W. AFundamental Approach to TlransformerTherrnal Modeling - Part I: Theory andEquivalent Circuit, IEEE Trans. On PowerDelivery, vol. 16, April 2001, pp 171-175.

VII. BIOGRAPHIES

Jos Antonio Jardini (M1966, SM 1978, F 1990) wasborn in So Paulo, Brazil, onMarch 27th, 1941. He graduatedfrorn Escola Politcnica da

Universidade de So Paulo in1963 (Electrical Engineering).Fron the same institution hereceived the MSc, PhD,Associate Professor and Head

Professor degrees in 1971, 1973, 1991 and 1999,respectively. For 25 years he worked for ThemagEngenharia Ltda., a leading consulting cornpanyin Brazil, where he conducted many powersysterns studies and participated in major powersystem projects such as the Itaipu hydro plant. Heis currently Head Professor at Escola Politcnicada Universidade de So Paulo, where he teaches

power systern analysis and digital automation,and where he leads the GAGTD group, which is


6/6

[1] IEEE std C57.91-1995 IEEE Guide for

Loading Mineral - Oil - ImmersedTransformer, 1995.

[2] IEC 354 International Standard LoadingGuide for Oil Immersed PowerTransforrners, 1991-09

responsible for the study and development ofautornation systerns in the fields ofgeneration,transrnission and distribution of electricity. Herepresented Brasil at SC-38 of CIGR and is aDistinguished Lecturer of IAS/IEEE.

Luiz Carlos Magrini was bomin So Paulo, Brasil, on May3th, 1954. He graduated frornEscola Politcnica daUniversidade de So Paulo in1977 (Electrical Engineering).From the same institution hereceived the MSc and PhDdegrees in 1995 and 1999,respectively.

For 17 years he worked for Themag EngenhariaLtda, a leading consulting cornpany in Brasil. He

is currently a researcher at Escola Politcnica daUniversidade de So Paulo GAGTD group.

transformer evalution overloading conditions

Documents