ISTANBUL TECHNICAL UNIVERSITY  GRADUATE SCHOOL OF SCIENCE

ENGINEERING AND TECHNOLOGY

DETERMINATION OF HOT-SPOT
TEMPERATURE FOR ONAN DISTRIBUTION
TRANSFORMERS WITH DYNAMIC THERMAL MODELLING

M.Sc. THESIS

Oluş SÖNMEZ

Department Of Electrical Engineering

Electrical Engineering Programme

Anabilim Dalı : Herhangi Mühendislik, Bilim

Programı : Herhangi Program

JUNE 2012
ISTANBUL TECHNICAL UNIVERSITY  GRADUATE SCHOOL OF SCIENCE
ENGINEERING AND TECHNOLOGY

DETERMINATION OF HOT-SPOT
TEMPERATURE FOR ONAN DISTRIBUTION
TRANSFORMERS WITH DYNAMIC THERMAL MODELLING

M.Sc. THESIS

Oluş SÖNMEZ
(504081042)

Department of Electrical Engineering

Electrical Engineering Programme

Thesis Advisor: Ass. Prof. Dr. Güven KÖMÜRGÖZ

Anabilim Dalı : Herhangi Mühendislik, Bilim
Programı : Herhangi Program

JUNE 2012
ĠSTANBUL TEKNĠK ÜNĠVERSĠTESĠ  FEN BĠLĠMLERĠ ENSTĠTÜSÜ

ONAN SOĞUTMALI DAĞITIM

TRANSFORMATÖRLERĠNĠN EN-SICAK NOKTA SICAKLIĞININ
DĠNAMĠK ISIL MODELLEME ĠLE BELĠRLENMESĠ

YÜKSEK LĠSANS TEZĠ

Oluş SÖNMEZ
(504081042)

Elektrik Mühendisliği Anabilim Dalı

Elektrik Mühendisliği Programı

Tez Danışmanı: Yrd. Doç. Dr. Güven KÖMÜRGÖZ

Anabilim Dalı : Herhangi Mühendislik, Bilim
Programı : Herhangi Program

HAZĠRAN 2012
Oluş Sönmez, a M.Sc.student of ITU Graduate School of Science Engineering
and Technology student ID 504081042, successfully defended the thesisentitled
“Determination of Hot-spot Temperature For ONAN Distribitution Transformers
with Dynamic Thermal Modelling ”, which he prepared after fulfilling the
requirements specified in the associated legislations, before the jury whose signatures
are below.

Thesis Advisor : Ass. Prof. Dr. Güven KÖMÜRGÖZ ..............................

İstanbul Technical University

Jury Members : Prof. Dr. Ġbrahim ÖZKOL .............................

İstanbul Technical University

Ass. Prof. Dr. Mustafa BAĞRIYANIK ..............................

İstanbul Technical University

Date of Submission : 04 May 2012

Date of Defense : 05 June 2012

v
vi
 To my grandfather; Mr. Süreyya Sönmez,

vii
viii
FOREWORD

I would like to thank to my supervisor Ass. Prof. Dr. Güven Kömürgöz for showing
such confidence on my abilities. I am very grateful for her continuous support,
encouragement and guidance.

I want to thankto all valuable personnel of Sönmez Transformer company, assisting

me during this study.

I am also grateful to Dr. Bora Alboyacı, for all his contributions to my academic
career.

I want to thank to my mother, father and brother for their continuous support to me
especially to my father Mr. Ümit Sönmez.Thank you for the wisdom and experiences
you shared with me, without which this document would not have been.

I would like to thank to my love, Merve, for supporting me continuously with an

extreme patience from start of this study.

Finally, I would like to thank to my grandfather Mr. Süreyya Sönmez for directing
me to transformer engineering.

May 2012 Oluş SÖNMEZ

Electrical Engineer

ix
x
TABLE OF CONTENTS

Page

FOREWORD ............................................................................................................. ix
TABLE OF CONTENTS .......................................................................................... xi
ABBREVIATIONS AND SYMBOLS ................................................................... xiii
LIST OF TABLES ................................................................................................. xvii
LIST OF FIGURES ................................................................................................ xix
SUMMARY ............................................................................................................. xxi
ÖZET...................................................................................................................... xxiii
1. INTRODUCTION .................................................................................................. 1
2. THERMAL MODELLING OF DISTRIBUTION TRANSFORMERS ........... 5
2.1 IEEE Standard C57.91-1995 Model................................................................... 5
2.1.1 Top-oil temperature rise over ambient ........................................................ 6
2.1.2 Oil time constant ......................................................................................... 7
2.1.3 Winding hot-spot temperature rise over top-oil rise ................................... 7
2.2 IEC Standard 60076-7 Model ............................................................................ 8
2.2.1 Top-oil temperature rise over ambient ........................................................ 8
2.2.2 Hot-spot temperature rise over top-oil temperature .................................. 11
2.2.3 Winding time constant .............................................................................. 12
2.3 D. Susa’s Thermal Model................................................................................. 12
2.3.1 Thermal electrical analogy ........................................................................ 12
2.3.2 Top oil temperature rise over ambient model ........................................... 16
2.3.3 Hot-spot temperature rise over oil temperature model ............................. 18
3. EXPERIMENTAL STUDY................................................................................. 21
3.1 Measurements From Temperature Rise Tests .................................................. 21
3.1.1 Tested unit design parameters ................................................................... 22
3.1.2 Location of thermocouples........................................................................ 26
3.1.3 Applied temperature rise tests ................................................................... 30
3.1.4 Results of tests ......................................................................................... 34
3.1.4.1 Top-oil temperature............................................................................ 34
3.1.4.2 Bottom-oil temperature ...................................................................... 36
3.1.4.3 Hot-spot to top-oil temperature rise ................................................... 38
3.1.4.4 Average temperature rises .................................................................. 41
3.2 Results Obtained With Thermal Models .......................................................... 46
3.3 Comparison Of Measured Values With Results Of Thermal Models .............. 49
4. CONCLUSIONS .................................................................................................. 53
REFERENCES ......................................................................................................... 55
APPENDICES .......................................................................................................... 59
CURRICULUM VITAE .......................................................................................... 69

xi
xii
ABBREVIATIONS AND SYMBOLS

B1 : Mixed bottom oil thermocouple location

B2 : Lowest part of tank thermocouple location
Cel : Electrical capacitance
Cth : Thermal capacitance
Cth-oil : Thermal capacitance of oil
D : Constants for oil physical characteristics
DC : Direct current
f1(t) : Relative increase of the top-oil temperature rise over ambient in per-
unit value
f2(t) : Relative increase of the hot-spot temperature rise to top-oil
temperature in per-unit value
f3(t) : Relative decrease of the top-oil temperature over ambient in per-unit
value
G : Constants for oil physical characteristics
gr : Rated average winding to average oil temperature gradient
H : Hot spot factor
I : Current during that loading time period
Irated : Rated current of transformer for full load
IEC : The Institute of Electrical and Electronics Engineers
i : Electrical Current
j : Index for each time step during temperature rise test
K : Load factor
k : Measured variable
k11 : Thermal Constant
k21 : Thermal Constant
k22 : Thermal Constant
kg : Kilogram
kV : Kilovolt
kVA : Kilovoltampere
LV : Low voltage
m : A constant
min : Minute
Map : Weight of active part ( core and coil assembly)
Mfe : Weight of magnetic steel core
Mw : Weight of windings
Mtank : Weight of tank and other steel parts which have direct contact to oil
Moil : weight of oil
nIEEE : An exponential constant for IEEE model
nSusa : A costant for D. Susa Model
ONAN : Oil natural – air natural cooling
P : Total losses at rated load
PDC : DC losses ( I2R losses )

xiii
PDC,pu : DC losses ( I2R losses ) in per unit
Peddy : Eddy losses
Peddy,pu : Eddy losses in per unit
Pfe : No-load losses
Pl : Total load losses
P(θhs)pu : Load loss varies on temperature change in per unit
Pstray : Stray losses
pu : Per unit
q : Generated heat
qfe : Generated heat by no-load losses
qcu : Generated heat by load losses
qtot : Generated heat by total losses
R : Ratio of load losses at rated power to the no-load losses
Ri : Cold resistance (initial measured value)
Rel : Electrical resistance
Rhv : High voltage winding resistance
Rlv : Low voltage winding resistance
Rth : Thermal resistance
Rth-oil : Non-linear oil to air thermal resistance
Ru : Calculated resistance (ultimate measured value)
sec : Second
t : Time
Temp. : Temperature
T1 : Thermometer pocket location
T2 : Top of the cooling fins thermocouple location
T3 : Outlet of the winding cooling channel (duct ) thermocouple location
u : Electrical voltage
W : Watt
x : An oil thermal constant
y : A thermal constant
z : Vector varies with thermal time constant
μ : Viscosity of oil
μpu : Viscosity of oil in per unit
θ : Temperature
θamb : Ambient Temperature
θbo : Bottom oil temperature
θbo,i : Initial bottom oil temperature
θhs : Hot-spot temperature
θhs,hv : Hot-spot temperature for high voltage winding
θhs,i : Initial hot-spot temperature
θhs,lv : Hot-spot temperature for low voltage winding
θoil : Top-oil temperature
θoil,average : Average oil temperature
θoil,i : Initial top-oil temperature
θoil,bottom-surfece : Oil temperature at the bottom of tank
θoil,top-surfece : Oil temperature at the top of tank
θi : Cold temperature (calculated temperature)
θk : Temperature factor for calculation
θu : Calculated temperature (ultimate temperature)
θw,average : Average winding temperature

xiv
Δθbo,rated : Rated bottom-oil temperature rise at rated power
Δθhs : Hot-spot temperature rise over top-oil temperature
Δθhs,I : Initial winding temperature rise over top-oil temperature
Δθhs,rated : Rated hot-spot temperature rise over top-oil temperature at rated load
Δθhs,U : Ultimate winding temperature rise over top-oil temperature
Δθoil,I : Initial top oil temperature rise over ambient
Δθoil,rated : Rated bottom-oil temperature rise at rated power
Δθoil,U : Ultimate top oil temperature rise over ambient
Δθw/A,R : Average winding temperature rise over ambient at rated load
τoil : Oil time constant
τoil,rated : Rated oil time at rated load
τw : Winding time constant

xv
xvi
LIST OF TABLES

Page

Table 2.1 : Recommended constants for ONAN cooled distribution transformers .... 9
Table 2.2 : Parameters for Thermal and Electrical Analogy . ................................... 13
Table 2.3 :Viscosity calculation constants ................................................................ 15
Table 3.1 : Nameplate data of tested transformer. .................................................... 22
Table 3.2 :Load steps of the second test ................................................................... 32
Table 3.3 : Measured highest top-oil temperatures . ................................................. 36
Table 3.4 : Measured highest bottom-oil temperatures ............................................. 38
Table 3.5 :Hot-spot temperature rises to top oil temperature rises ......................... 40
Table 3.6 : Measured warm resistances. ................................................................... 42
Table 3.7 :Calculated resistance and average winding temperatures ........................ 44
Table 3.8 :Average winding – average oil temperature rise according to IEC. ........ 44
Table 3.9 : Average winding – average oil temperature rise according to IEEE. ..... 45
Table 3.10 :Average winding – average oil temperature rise according to D. Susa 46
Table 3.11 :Data which is used in thermal models ................................................... 47
Table A.1 :Technical datasheet for temperature monitoring and recording devices 61
Table A.2 :Technical datasheet of temperature monitoring device .......................... 62
Table A.3 :CPC 100 Winding resistance measurement test device datasheet .......... 64

xvii
xviii
LIST OF FIGURES

Page

Figure 1.1 : Aging Acceleration rate ........................................................................... 2

Figure 2.1 : Basic electrical RC circuit ..................................................................... 13
Figure 2.2 : Basic thermal RC circuit ....................................................................... 14
Figure 2.3 : Oil physical parameter changes with varying temperature .................. 15
Figure 2.4 : Top-oil temperature rise over ambient model ....................................... 16
Figure 2.5 : Hot-spot temperature circuit model together with top-oil model circuit
............................................................................................................... 18
Figure 3.1 : Active part of transformer ..................................................................... 22
Figure 3.2 : Internal view of tested transformer........................................................ 23
Figure 3.3 : Internal view of tested transformer with thermocouples ....................... 24
Figure 3.4 : Exterior view of tested transformer from HV side ................................ 25
Figure 3.5 : Exterior view of tested transformer from LV side ................................ 25
Figure 3.6 : Positions of thermocouples on phase B winding................................... 27
Figure 3.7 : Top view of transformer showing thermometer pocket......................... 27
Figure 3.8 : Thermocouple in thermometer pocket................................................... 28
Figure 3.9 : Location of thermocouples in transformer ............................................ 28
Figure 3.10 : Location of thermocouples in windings .............................................. 29
Figure 3.11 : Thermocouples for bottom oil temperature. ........................................ 29
Figure 3.12 : Thermocouples located inside of transformer tank ............................. 30
Figure 3.13 : Transformer during constant load test ................................................. 31
Figure 3.14 : Thermal camera view from narrow side .............................................. 33
Figure 3.15 : Thermal camera view from long side .................................................. 33
Figure 3.16 : Measured top-oil temperatures atdifferent locations in first test. ........ 35
Figure 3.17 : Measured top-oil temperatures at different locations in second test. . 35
Figure 3.18 : Measured bottom-oil temperatures at different locations in first test. 37
Figure 3.19 : Measured bottom-oil temperatures at different locations in second test
............................................................................................................... 37
Figure 3.20 : Winding hot spot temperatures in the first test (Constant load test) ... 39
Figure 3.21 : Winding hot spot temperatures in the second test (Varying load test) 40
Figure 3.22 : Hot-spot to top-oil temperature rises ................................................... 41
Figure 3.23 : Warm resistance values for high voltage winding. ............................. 43
Figure 3.24 : Warm resistance values for low voltage winding................................ 43
Figure 3.25 : Top oil temperature for varying load test ............................................ 50
Figure 3.26 : LV hot-spot temperature for varying load test ................................... 51
Figure 3.27 : HV hot-spot temperature for varying load test ................................... 51
Figure A.1 : Temperature monitoring and recording device .................................... 60
Figure A.2: Temperature monitoring and recording devices during ( Test 2) .......... 63
Figure A.3: Thermal camera ..................................................................................... 64
Figure A.4: CPC 100 winding resistance measurement test device view. ................ 65
Figure A.5 : Test room while recording temperature reading during test 1.............. 66

xix
Figure A.6 : Tested transformer view from test laboratory during test .................... 67

xx
 DETERMINATION OF HOT-SPOT TEMPERATURE FOR ONAN
DISTRIBUTION TRANSFORMERS WITH DYNAMIC THERMAL
MODELLING

SUMMARY

Transformers are one of the largest capital investment part of distribution networks.
Transformers’ reliability is very important for electrical networks considering cost
impact of power outages.
High temperature rise problem, one of the most serious problems for transformers.
Loading cycles and ambient conditions of transformers can be different than design
conditions. Transformer top-oil temperature and winding hot spot temperatures are
most critical parameters for transformer because it causes aging and therefore affects
life time of transformer. It is very important to determine hot-spot temperature and
oil temperature accurately which depends on ambient conditions and loading
conditions to avoid loss of life on transformer.
One solution to determine top-oil temperature and winding hot-spot temperature is
to use on-line monitoring devices like fiber optical measurement devices. Other
solution is to use dynamic thermal models to define top-oil temperature and winding
hot-spot temperatures. Considering that fiber optical measurement solutions are very
costly for distribution transformers, generally thermal models are mandatory to
define critical temperatures of distribution transformers.
IEEE C57.91-1995 and IEC 60076-7 standards present thermal models for defining
critical temperatures in transformers. These models are widely used in industry. On
the other hand these methods are solved with exponential equations and they do not
take into account the affects of change on oil temperature. Dynamic thermal model
which is proposed by D. Susa, requires similar input parameters with IEEE C57.91-
1995 model and IEC 60076-7 model, in addition to this, this thermal model takes
into account the effects of change on oil temperature.
Scope of this study is to verify three differenttemperature rise models for distribution
transformers to allow optimum loading of natural oil cooled transformers by using
the data which could be obtained from manufacturer.
In this study, on a 1000 kVA 33/0.4 kV ONAN cooled distribution transformer, two
temperature rise tests are realized. In first test, total losses are injected and necessary
data to determine transformers specific parameters for thermal models obtained. In
the second test, a varying load test is realized. Three different thermal models are
used for same varying loading characteristic and results are compared with
measurement results.
Significant advantage of this study is, necessary parameters for all thermal models
are determined with proper measurements of tested unit, therefore all three models
are compared under same conditions.

xxi
xxii
 ONAN SOĞUTMALI DAĞITIM TRANSFORMATÖRLERĠNĠN EN-SICAK
NOKTA SICAKLIĞININ DĠNAMĠK ISIL MODELLEME ĠLE
BELĠRLENMESĠ

ÖZET

Transformatörler, elektrik dağıtım ve iletim şebekelerindeki en değerli

elemanlardandır. Günümüzde enerji kesintileri, elektrik enerjisi dağıtım firmalarına
büyük maliyetler doğurmaktadır, transformatörlerin güvenilebilir olması, enerji
kesintilerinin asgariye indirilebilmesi için büyük önem taşır.
Transformatör kayıpları yükte kayıplar ve boşta kayıplar olarak ikiye ayrılır. Boşta
kayıplar, transformatörün anma geriliminde ve yüksüz şekilde çalışması durumunda
ortaya çıkan kayıplardır. Yükte kayıplar ise, transformatörün anma gücünde
çalışması durumunda oluşan kayıplardır. Bu kayıplar, transformatörün ısıl
davranışına farklı şekilde etki etmektedirler.
Transformatörlerin aşırı ısınması, transformatörler için en ciddi sorunlardan biridir.
Transformatörler tasarım aşamasında belirli koşullarda çalışmak üzere
tasarlanırlar.Transformatörlerin pratikte çalışma koşulları, tasarlanan çalışma
koşullarından farklı olabilir. Transformatörün çalışma koşulları hem yüklenme
karakteristiğine hem de ortam sıcaklığı gibi ortam şartlarına bağlıdır.
Transformatörlerin aşırı ısınmasının engellenmesi için transformatörlerin tepe yağ
sıcaklığı ve sıcak nokta sıcaklıkları çok önemlidir, çünkü transformatörün
yaşlanmasına doğrudan etki etmektedir. Yağlı tip transformatörlerde yaygın olarak A
sınıfı izolasyon malzemeleri kullanılmaktadır. A sınıfı izolasyon malzemelerinin,
sistem sıcaklığı 105 ºC’dir ve genel olarak selülozik içerikli maddelerden
üretilmektedir. Bu tip izolasyon malzemeleri, sıcaklık 110ºC’yi aştığında çok hızlı
bir şekilde yaşlanmakta ve ömür kaybına uğramaktadır. Transformatörün
planlanandan erken yaşlanması ise öngörülen ömrünün kısalmasına neden olacaktır.
Tepe yağ sıcaklığını ve sargı sıcak nokta sıcaklığını belirlemenin bir yöntemi, sürekli
çevrimiçi ölçüm yapılmasıdır. Fakat sargı sıcak nokta sıcaklığının sürekli çevrimiçi
ölçülebilmesi için sargı içine fiber optik ölçüm cihazlarının monte edilmiş olması
gerekir. Büyük güç transformatörlerinde bu yöntem uygulanabilir olsa da dağıtım
transformatörleri için bu çözüm çok yüksek maliyetli olacaktır. Bu nedenle dağıtım
transformatörlerinin kritik sıcaklıklarının belirlenebilmesi ve öngörülebilmesiiçin
dinamik ısıl modeller gereklidir.
IEEE C57.91-1995 standardı ve IEC 60076-7 standardı, değişken yük durumları için
transformatörlerin tepe yağ ve en sıcak nokta sıcaklarının belirlenmesine yardımcı
olan ısıl modeller sunmaktadır. Bu modeller, transformatör kullanıcıları tarafından
yaygın olarak kullanmaktadır. Böylece transformatörün kritik sıcaklıkları
belirlenmektedir. Öte yandan, bu modeller, yağ sıcaklık değişiminin kayıplara ve
yağın özelliklerine olan etkilerini dikkate almayan göreceli olarak basit modellerdir.
D. Susa tarafından önerilen model ise, temel ısıl prensipler ve temel elektriki

xxiii
prensipler arasındaki benzerliği kullanarak, transformatörlerin ısıl analizi için direnç
ve kapasite içeren bir devre sunar. Bu devrede, direnç, yağ sıcaklığına bağlı olarak
değişkendir. Bu devrenin çözümü ile bir diferansiyel denklem elde edilir. Sonuç
olarak karmaşık ısıl problem, bir elektrik devresine benzetilerek, diferansiyel
denklem ile çözülür. Bu yöntemde, standartlarda önerilen yöntemden farklı olarak,
yağ sıcaklığı değişiminin yükte kayıplara ve yağ viskozitesine etkisi dikkate
alınmıştır.
Bu çalışmanın kapsamı, yukarıda belirtilen üç farklı ısıl modelin, üreticiden
alınabilecek veriler kullanılarak, doğal soğutmalı dağıtım transformatörlerinin
optimum olarak yüklenebilmesinin sağlanması amacı ile doğrulanmasıdır.
Bu çalışmada, 1000 kVA gücünde 33/0.4 kV değiştirme oranına sahip, ONAN
soğutmalı (doğal yağ soğutmalı) bir dağıtım transformatörü kullanılmıştır.
Transformatör ülkemizde yaygın olarak kullanılan aynı güçteki transformatörlere
eşdeğer nitelikte seçilmiştir. Söz konusu transformatörün dışında ve içinde sıcaklık
ölçümü için ısı sensörleri monte edilmiştir. Transformatör içindeki sensörler, her bir
sargının alçak gerilim ve yüksek gerilim kısımlarına konulmuştur. Sargı içindeki
sensörlerin yerlerinin seçiminde, daha önce yapılan çalışmalar ve tecrübeler dikkate
alınarak en sıcak noktaların oluşması beklenen bölgeler seçilmiştir. Özellikle en
sıcak noktanın beklendiği sargılara, ilave olarak ikişer sensör daha konulmuştur.
Sargı içindeki ısı sensörlerine ek olarak, çeşitli noktalardaki yağ sıcaklıklarını
ölçebilmek için, transformatör kapak ve kazanına da ısı sensörleri yerleştirilmiştir.
Bu sensörler ile tepe yağ sıcaklığı üç farklı noktadan ölçülebilmiştir. Dip yağ
sıcaklığı ise iki farklı noktadan sensörlerin yardımı ile ölçülmüştür. Toplamda on beş
sensör kullanılmıştır. Bu sensörler, görüntüleme ve kaydetme cihazlarına
bağlanmıştır. Ölçümler sırasında sürekli olarak ortam sıcaklığı da ölçülerek
kaydedilmiştir. Ortam sıcaklığının değişimi etkisi, transformatörün kritik
sıcaklıklarını doğrudan etkileyeceği için bütün modellerde, ortam sıcaklığı etkisi
dikkate alınmıştır. Testler sırasında ısıl kamera kullanılarak, ısı dağılımı
gözlenmiştir. Isıl kamera ile transformatörün kazanın içinde yağın etkisi ile beklenen
dengeli sıcaklık dağılımı doğrulanmıştır.
Transformatör üzerinde iki farklı sıcaklık artış deneyi uygulanmıştır. Birinci
deneyde, toplam kayıplar transformatör üzerinden geçirilmiştir ve söz konusu
transformatörün ısıl modelleri için gerekli değişkenler ve sabitler elde edilmiştir.
Transformatörün ısıl özelliklerinin belirlenmesi için bu test kullanılmıştır. Sensörler
yardımı ile yapılan ölçümlere ek olarak, transformatörün alçak gerilim ve yüksek
gerilim sargılarının doğru akım dirençleri soğuk ve sıcak durumlarda ölçülerek
ortalama sargı sıcaklık artış değerleri belirlenmiştir.
İkinci test sırasında ise farklı akımlar transformatöre uygulanarak değişken yük
durumu deneyi gerçekleştirilmiştir. Bu test sırasında transformatör üç saat boyunca
anma gücü ile yüklenmiştir ardından 105 dakika boyunca iki katı yük ile
yüklenmiştir.Ardındam transformatör iki katı yüklenmesi kesilerek, son yarım saat
transformatör yüksüz durumda bırakılarak ölçümlere devam edilmiştir.Testler
transformatörün sekonder tarafı kısa devre edilmek sureti ile primerden akım
uygulanması şeklinde gerçekleştirilmiştir. İkinci test sonucunda elde edilen tepe yağ
sıcaklığı ve alçak gerilim ve yüksek gerilim sargıları için en sıcak nokta sıcaklıkları,
matematiksel model ile elde edilen sonuçlarla karşılaştırılmıştır.
Bu çalışmada, üç farklı matematiksel model kullanılmıştır.Bunlardan ilki IEEE
C58.91 standardında belirtilen modeldir, ikincisi IEC 60076-7 standardında belirtilen

xxiv
modeldir. Bu iki modele ek olarak, D. Susa tarafından geliştirilen model
kullanılmıştır. Bu modelde ısıl devre ile elektrik devreleri arasındaki benzerlik
kullanılmıştır. Bu benzerliğe göre ısı kaynağı, akım kaynağı olarak; sıcaklık, gerilim
olarak; ısıl direnç, elektriksel direnç olarak ve ısıl kapasite; elektriksel kapasite
olarak modellenmiştir. Bu şekilde karmaşık ısıl problem, basit bir R-C elektrik
devresine indirgenmiştir. Bu devre, devre analizi yöntemleri ile çözülerek
diferansiyel denklemler elde edilmiştir. Tepe yağ sıcaklığı ve en sıcak nokta
sıcaklıkları ise bu diferansiyel denklemler çözüşerek elde edilmiştir. Bu modelde,
standartlarda verilen modellerden farklı olarak sıcaklıkla, yağın fiziksel
özelliklerindeki değişim de dikkate alınmıştır. Yağın fiziksel değerlerindeki en etkin
değişim viskozite değerinde olduğu için, modelde viskozitenin sıcaklık ile değişim
etkisi kullanılmıştır. Buna ek olarak, transformatör kayıplarının da sıcaklık ile
değişimi modelde dikkate alınmıştır. Standartlarda verilen modeller kullanılırken,
gerekli olan değişkenler, deneyler ile elde edilen veriler ve optimizasyon yöntemleri
kullanılarak belirlenmiştir.
Testte uygulanan ile aynı değişken yük durumu, söz konusu üç ısıl modelde
gerçeklenmiştir. Modeller ile hem tepe yağ sıcaklığı hem de alçak gerilim ve yüksek
gerilim sargıları için sıcak nokta sıcaklıkları hesaplanmıştır. Elde edilen sonuçlar,
ölçüm sonucu ile elde edilen sonuçlar ile karşılaştırılmıştır.
Bu çalışmanın önemli bir üstünlüğü, uygulanacak ısıl modeller için değişkenler,
gerçekleştirilen test sayesinde, doğru bir şekilde elde edilmiştir. Isıl modellerin
karşılaştırılması için gerekli olan ölçümler yapılmıştır. Bu nedenle her üç model de
aynı şartlar altında karşılaştırılabilmiştir.

xxv
xxvi
1. INTRODUCTION

Transformer history starts from the early 1880s. It is based on induction principle
which is discovered by Faraday in 1831 [1]. Modern transformers are widely used in
transmission, distribution systems and also industrial systems. Transformers are one
of the most expensive equipments of electrical systems. Efficiency of transformers
are relatively high compared to different electrical machines. Except no load and
load losses, output power is approximately equal to input power. On the other hand,
these losses causetemperature rises on transformers.

Transformer losses are categorized as no-load losses and load-losses[2]. No-load

losses are losses, which occurs during rated voltage and no-load operation. Load
losses are consists from DC (I2R) losses, stray losses and eddy losses. DC load losses
are calculated from measured winding resistance values. Stray losses and eddy losses
are calculated by subtracting DC losses from measured load losses. Load losses are
measured with short-circuit test method which is described in IEC 60076-1 standard
[3]. No-load losses, DC losses and stray losses have different affects on transformers
thermal behavior.

High temperature rise problem , one of the most serious problems for all kind of
transformers[4]. Loading cycles and ambient conditions of transformers can be
different than design conditions. Transformer top-oil temperature and winding hot
spot temperatures are most critical parameters for transformer because it causes
aging and therefore affects life time of transformer. In oil filled distribution
transformers A class insulation which of system temperature is 105 C is commonly
used [5]. A class insulation contains mainly cellulosic based materials. For cellulosic
insulation materials rate of aging will rapidly increase above 110 C, this affects aging
rate of distribution transformer directly. In Figure 1.1, aging acceleration rate
changes with hot-spot temperature change is plotted according to equations given in
IEEE standard C57.91-1995 [6,7].

1
 1000
Aging acceleration factor, pu

100

10

0,1

0,01

0,001
60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
Winding hot-spot temperature, ̊C

Aging acceleration factor

Figure 1.1 : Aging acceleration rate.

Top-oil temperature is the temperature, measured at top layer of the oil in

transformer. Hot-spot temperature is the hottest point temperature of conductor part
of transformer, which is in contact with oil[6]. Transformers loading beyond
nameplate can be defined by respecting limits of the top-oil and hot-spot
temperatures of transformer. Therefore, it is very important to define hot-spot
temperature properly.

Many on-line and off-line monitoring systems have been developed [8-12]. However
on-line monitoring by direct measurement of winding temperature by using fiber
optic probes, is a very costly method for distribution transformers. Therefore,
thermal models should be used for distribution transformers to define loading
capability.

These thermal models are based on conventional heat transfer theory [13],
transformer thermal tests [14-27], application for the lumped capacitance method,
thermal electrical analogy [27] .In previous studies, it has been noticed that hot-spot
temperature rise over top-oil temperature for load changes is a function which
depends on time [24,27]. Present IEC and IEEE standards recommends thermal
models to determine top-oil temperature and hot-spot temperatures [6-7,28 ]. These
models are widely used in industry to determine temperatures. In addition to this, an

2
investigation performed on a 2500 kVA transformer without external cooling showed
accurate results [24]. Author was highly motivated to make a study to verify this
method with measured values of a smaller distribution transformer and compare
results with IEC and IEEE standards which are very widely used in market [6,28].

Scope of this study is to verify different thermal models for distribution transformers
to allow optimum loading of natural oil cooled transformers by using the data which
could be obtained from manufacturer.In this thesis, two different temperature rise
tests are realized on a 1000 kVA ONAN cooled distribution transformer, installed
with measurement thermocouples. First temperature rise test is constant load test.
From the data obtained from the first test, parameters required for thermal modelsare
determined by using optimization methods.In the second temperature rise test,
varying step loads are applied as given in Table 3.2. Temperature values for top-oil
and hot-spots are determined by three different thermal models for same loading
condition which is realized during Test 2. Results obtained from variable load tests
for tested distribution transformer is compared with calculated temperature values.
For calculations of hot-spot and oil temperatures, three different models are used and
results are compared.

Primary aim of this thesis is to verify different models which could be applied for
distribution transformers by using input data which could be obtained by
manufacturer.Therefore various thermal models are compared with measured values
for a standard ONAN cooled distribution transformer which means without external
cooling.

Three models which will be used in this study is explained in Chapter 2 with details.
Two of the methods are given in generally accepted standards and other method is
developed by D. Susa [24,26]. Experimental study which realized with a 1000 kVA
33/0.4 kV ONAN cooled distribution transformer is explained with details in Chapter
3. In this chapter, measured values and calculated values are plotted and compared.
Conclusions and comments obtained from this study are given in Chapter 4.

3
4
2. THERMAL MODELLING OF DISTRIBUTION TRANSFORMERS

Transformer life is very dependent on condition of insulation materials. High oil

temperature and high hot-spot temperature affects condition of insulation materials.
Therefore it is very important to determine oil temperature and hot-spot temperature
of transformers according to load conditions and ambient conditions. Determinations
of these temperatures are possible with dynamic thermal models. In this study, three
different thermal models are used to determine top-oil temperature and hot-spot
temperature of both high voltage and low voltage windings. Methods which are
given in IEEE Standard C57.91-1995 and in IEC 60076-7, are widely used in
industry to determine top-oil and hot-spot temperatures [6,28]. Affects of
temperature change on oil viscosity and losses are taken into account in dynamic
thermal model which is developed by D. Susa.

In this section, three different method for thermal modeling is described. Definition
of methods are as following;

- IEEE Standard C57.91-1995 ( Cor.2002)[6,7]

- IEC Standard 60076-7 ed1.0[28]

- D. Susa’s thermal model [24]

2.1 IEEE Standard C57.91-1995 Model

IEEE standard C57.91-1995 provides two different methods for predicting thermal
behavior of transformers. One of them is explained in Clause 7 of related standard
and other method is given in Annex G of this standard. The Annex G method,
provides a non-linear model, however it requires, more input data for model, and for
distribution transformers required data for Annex G method could not be easily
found in the market for all distribution transformers[6,30]. Special temperature rise
tests like bottom temperature rise measurement required to obtain requested input

5
data to model. Considering that this study is mainly targeting distribution
transformers, only Clause 7 model is used in this study. Normally, according to
Clause 7 model, variable ambient temperature affects are not considered, however in
this study, affects of ambient changes applied in model to obtain more accurate
results.

According to this thermal model, first top-oil temperature over ambient is calculated.
For this calculations, oil time constant should be defined. Following, winding hot-
spot rise over top-oil rise is calculated.

2.1.1 Top-oil temperature rise over ambient

In standard, top-oil temperature rise at a time after a step load change is given with
following equation [28];

t
τoil
Δθ oil (Δ oil,U Δθ oil,I )(1 exp ) Δθ oil,I (2.1)

Where;

Δθoil,I initial top oil temperature rise over ambient ( in K ) Δθoil,U

ultimate top oil temperature rise over ambient ( in K )τoil oil time
constant ( in hour ).

In case of multi step loading cycles, equation 2.1 is used for each load step and in
each step top-oil rise calculated for the previous load step is used as initial top-oil
rise of next step.

Ultimate top oil temperature rise over ambient is calculated with following equation;

n IEEE
K 2R 1
oil, U oil, rated (2.2)
R 1

Where; K is theratio of load to nominal load ( in pu ), R is the ratio of load losses at

rated power to the no-load losses and nIEEE is exponential constant, which is 0.8 for
ONAN cooled transformers.

6
2.1.2 Oil time constant

In IEEE C57.91 standard, rated oil time constant is given with an equation which
uses thermal capacity value, rated temperature rise value and total losses. Formula
is given in equation 2.3.

C th oil oil, rated

oil, rated (2.3)
P

Thermal capacity value ( Cth-oil ) is calculated with following equoation

C th oil 0.1323 M AP 0.0882 M tan k 0.40M oil (2.4)

Where; MAP is the weight of active part (core and coil assembly) (in kg) , Mtank is
the weight of tank and other steel parts which have direct contact to oil (in kg) and
Moil is weight of oil (in kg).

Top oil time constant for each interval calculated with following equation;

oil, U oil, I
( ) ( )
oil, rated oil, rated
oil oil, rated 1 1 (2.5)
oil, U n IEEE oil, I n IEEE
( ) ( )
oil, rated oil, rated

where

nIEEE is constant exponential defined 0.8 for ONAN cooled transformers [6].

2.1.3 Winding hot-spot temperature rise over top-oil rise

Winding hottest-spot temperature over top-oil rise for each step load cycle is given
by following equation;

hs ( hs, U hs, I )(1 exp w

) hs, I
(2.6)

where;

Δθhs,I initial winding temperature rise over top-oil temperature ( in K )

7
Δθhs,U ultimate winding temperature rise over top-oil temperature ( in K )

τw winding time constant (in min.)

Ultimate winding temperature rise over top-oil temperature calculated with following
equation;

hs, U hs, rated K 2m (2.7)

WhereΔθhs,ratedis rated winding temperature rise over top oil temperature ( in K) and
m is the exponential constant which is 0.8 for ONAN cooled transformers.

2.2 IEC Standard 60076-7 Model

IEC standard provides two thermal models. One of them is given as an exponential
equation solution, this method is more suitable for load cycles which are changing
with steps [28,31]. The other method which is given as a differential equation
solution, is suitable for randomly changing load cycles. Second method is mainly
applicable for on-line monitoring [31]. In this study, exponential equation solution
model is used.

In this model, primarily, top oil temperature rise over ambient is calculated by
defining necessary variables and constants. Afterwards, hot-spot temperature rise
over top-oil temperature rise is calculated by defining necessary variable and
constants.

2.2.1 Top-oil temperature rise over ambient

According to IEC 60076-7 standard, oil temperature rise over ambient temperature is
given with an equation depends on a function. This function includes exponential
constants. IEC 60076-7 has recommended some values for constants [28].

Equation given in 2.8 is applied at load increase. Equation which is given in 2.9 is
applied at load decrease.

oil ( oil ,U oil , I ) f1 (t ) oil , I (2.8)

8
 oil ( oil, I oil, U )f 3 ( t ) oil, U (2.9)

Ultimate temperature rise value is calculated from equation 2.10

x
K 2R 1
oil, U oil, rated (2.10)
R 1

where;x is the oil exponent, f1(t) indicates the relative increase of the top-oil
temperature rise in per-unit value. f3(t) indicates the relative decrease of the top-oil
temperature rise value in per-unit value.f1(t) and f3(t) fuctions are solved with
following equations which are given in equation 2.11 and equation 2.12;

t
k11
f 1 ( t ) (1 exp oil , rated
) (2.11)

t
k11
f 3 ( t ) exp oil , rated (2.12)

where, k11 is the thermal model constant.

It is recommended by IEC Standard to use standard values for thermal constants

[28]. These values are given in related standard. For x value it is recommended to use
0.8 value for ONAN cooled distribution transformers. For k11, it is recommended
use 0.8 for ONAN cooled distribution transformers. Recommended constants are
given in Table 2.1.

Table 2.1 :Recommended constants for ONAN cooled distribution transformers.

Oil exponent x 0.8

Winding exponent y 1.6
Constant k11 1.0
Constant k21 1.0
Constant k22 2.0
Oil time constant τoil 180 minutes
Winding time constant τw 4 minutes

9
However more accurate results had been obtained by estimating this constant from
top-oil temperature rise curve which is plotted by realizing temperature rise test
according to IEC 60076-2 standard [32].

Measured relative increase of the top-oil temperature rise of the steady state value is
calculated as per unit value with following equation [31]

k oilj k ambj
kf 1j (2.13)
oil, rated

where

k measured variable

j number of each step

f1j relative increase of oil temperature rise over ambient for related step

After relative increase of top-oil temperature rise is obtained, non-linear regression

method is used to define the constant to minimize the sum of squares between
differences f1j and kf1j.

Equation is given following;

(2.14)

where;

k measured variable

z vector varies with thermal time constant k11

j index for each time step during temperature rise test

Recommended k11 value in IEC 60076-7 which is given in Table 2.1 is used for
initial value of minimization [28].

10
2.2.2 Hot-spot temperature rise over top-oil temperature

Hot-spot temperature rise over top-oil temperature is calculated with a similar

formula to the top-oil temperature rise.

Equation given in 2.15is applied at load increase.Equation given in 2.16 is applied at

load decrease.

hs ( hs, U hs, I )f 2 ( t ) hs, I (2.15)

hs hs, U (2.16)

Ultimate winding temperature rise over top-oil temperature rise is calculated with
following equation.

hs, U hs, rated Ky (2.17)

Where y is the winding exponent.

f2(t) indicates the relative increase of the hot-spot temperature rise to top-oil
temperature rise value in per-unit value.Equations is shown in Equation 2.18.

t
t oil , rated
k 22 x k 22
f 2 (t) k 21x (1 exp w
) (k 21 1)(1 exp ) (2.18)

Where; k21 is thermal model time constant, k22 is the thermal model time constant
and τw is the winding time constant ( min.).

It is recommended to use defined values for thermal constants by IEC Standard [28].
These values are given inthe related standard. k21 is given as 1.0 for ONAN cooled
distribution transformers. k22is given as 2.0 for ONAN cooled distribution

11
transformers as shown in Table 2.1. For y value it is recommended to use 1.6 for
ONAN cooled distribution transformers.

2.2.3 Winding time constant

According to IEC standard, the winding time constant value which is used in
equation 2.18 is calculated with equation 2.19 which is given below [28].

M w cg r
w
60Pl (2.19)

Where

c specific heat of the conductor material in Ws/kg.K. It is 390 for

copper and 890 for aluminum

gr average winding to average oil temperature gradient in K

Mw weight of the windings in kg

Pl total winding losses at rated load in W

2.3 D. Susa’s Thermal Model

Dejan Susa’s method is based on thermal-electrical analogy [24,27,29].In this study,

top-oil temperature model and hot-spot temperature model is used. According to
developed model’s circuit theory, capacitance value is taken as constant and
resistance value is taken as variable which depends on oil viscosiy changes and loss
value changes with temperature changes. Basically input variables of model are load
pu value, oil viscosity value and ambient temperaeture, and main input constants
are, losses, time constants, rated temperature rise values. As a result of model, output
variables will be top oil temperature and hot-spot temperatures.

2.3.1 Thermal electrical analogy

One of the most used heat transfer law; Fourier theory’s equation form is with the
same form of Ohm’s law which is one of the most-well known electrcal law.
Therefore mathematical solution of a thermal system is usually similar to

12
mathematical soluation of an electrical system. By using this similarity, an equivalent
electrical circuit can be obtained to solve a thermal problem [20]. In Figure 2.1, a
basic electrical RC circuit is given.

u R

i C

Figure 2.1 : Basic electrical RC circuit.

It will be noticed that thermal analogy could be replaced as in Table 2.2 into
electrical analogy parameters based on basic heat transfer principles [13,20,25,27].

Table 2.2 :Parameters for Thermal and Electrical Analogy

Thermal Electrical

Heat Value, q ( W ) Current, i (A)

Temperature, θ ( C ) Voltage, u (V)

Thermal Resistance Rth( C / watt) Electrical Resistance, Rel ( ohm )

Thermal Capacitance Cth ( j / C ) Electrical Capacitance, Cel( f )

Accordingly, new RC circuit is defined for thermal analogy which is given in Figure
2.2;

13
 R th q amb
q

Cth
q

Figure 2.2 : Basic thermal RC circuit.

Thermal capacitance and resistance values can be defined as material’s thermal

characteristics. This is a common method for transformers thermal analysis.
Because, it allows to make mathematical solutions for complex heat transfer
problems.Past studies about lumped capacitance method are given in
[17,20,21,23,33].

In Susa’s method, thermal characteristic’s of materials are not considered as constant

as they were in previous studies. Therefore a non-linear thermal resistance is
introduced[27]. Because transformer oil characteristics like density, specific heat,
thermal conductivity, coefficient of thermal cubic expansion and viscosity values are
strongly dependent to oil temperature [22].

The change of viscosity with oil temperature change is significantly higher than the
change on other parameters [15,22,27]. This is shown in Figure 2.3 by plotting all
major oil physical parameter changes, which depends on temperature. Therefore,
only viscosity value of oil is variable in the model and other parameters are constant.
Oil physical parameters’ equations are taken from [15,27].

14
 18
16

Oil Physical Parameters, pu 14

12
10
8
6
4
2
0
-25 -5 15 35 55 75 95 115
Oil Temperature, ̊C
Viscosity
Density
Specific Heat
Thermal Conductivity
Coefficient for thermal volume expansion

Figure 2.3 : Oil physical parameter changes with varying temperature.

The viscosity of oil is dependent on oil temperature and formula is given in Equation
2.20.

D exp oil 273 (2.20)

D and Gparameters are constants related with transformer oil characteristic and they
are given in IEEE standard which is shown in Table 2.3. In the Table 2.3, viscosity
constants for two different insulation fluids are given[6].

Table 2.3 :Viscosity calculation constants [6,7]

Insulation Fluid D G
Mineral Oil 0.0013573 2797.3
Silicone Fluid 0.12127 1782.3

15
2.3.2 Top oil temperature rise over ambient model

From basic RC circuit which is given in Figure 2.2, thermal circuit is defined for top-
oil temperature as shown in Figure 2.4;[17, 24]. In this model, no-load losses and
load losses are defined as current sources in circuit diagram [25,26]. No-load losses
are constant and load losses are dependent on load factor.Because load losses varies
with load factor.For top oil temperature rise model, both no load and load losses are
heat sources since oil temperature is affected from heat of both losses. Capacitance
of oil is indicated as constant and calculation equation is given in Equation 2.4.
Thermal resistance value is variable of the circuit and depends on oil viscosity
change with temperature. Ambient temperature is indicated as voltage source, since it
is constant and added on top-oil temperature rise value to determine top-oil
temperature.

q tot R th-oil
q oil

q fe qcu C th-oil
q amb

Figure 2.4 : Top-oil temperature rise over ambient model.

Where; qtotis total heat generated with affect of losses (in W), qfeis heat generated
with affect of only no load losses (in W), qcuis heat generated with affect of load
losses (DC losses, eddy and stray losses ) (in W) and Rth-oilis non-linear oil to air
thermal resistance.

This circuit is solved with circuit analysis theory and differential equation which is
given Equation 2.21 is found [6,7,27-28].

1 RK 2 n Susa n Susa d oil ( oil amb ) n Susa 1

1 R
pu oil, rated pu oil, rated
dt n Susa (2.21)
oil, rated

16
Where,

R ratio of load losses to no-load losses

K load factor

μpu oil viscosity ( pu )

θamb ambient temperature ( C )

Δθoil,rated rated top-oil temperature rise over ambient temperature ( K )

τoil,rated rated oil time constant ( minutes )

In Susa’s method, rated oil time constant is calculated according to IEEE Std.
C57.91-1995(Cor.2002)[7]. It is noticed that in IEEE method constant is used in
hours, however in Susa’s method it is used in minutes.

Oil viscosity is calculated from Equation 2.20 for each oil temperature value,
however in model it is used in per unit.

Ratio of load losses to no-load losses is calculated with following equation

Pl
R (2.22)
Pfe

Load factor is calculated with Equation 2.23.

I
K (2.23)
I rated

where;

I current during that loading time period ( A )

Irated rated current of transformer for full load. (A )

17
In this model, different than other two models, which are described in the previous
sections, both thermal resistance and oil time constants are affected from change of
oil viscosity [27].

2.3.3 Hot-spot temperature rise over oil temperature model

R th-hs-oil
q hs

qcu C th-wnd q oil

q tot R th-oil
q oil

q fe qcu C th-oil
q amb

Figure 2.5 : Hot-spot temperature circuit model together with top-oil model circuit.

Where, qcuis heat generated with affect of load losses (DC losses, eddy and stray
losses) ( W ) , Rth-hs-oil isnon-linear hot-spot to oil thermal resistance, θhsis hot-spot
temperature ( C ).

Model for hot-spot temperature rise over oil temperature is based on thermal model
which is using thermal electrical analogy similar to oil temperature rise model [13].

18
RC circuit which is developed for hot-spot temperature is given in Figure 2.5. In
developed thermal model, different than oil temperature rise model, only load losses
are taken into account. Load losses are given as a current source which varies with
load factor since it is heat source. Thermal capacitance of winding is given as
constant. Thermal resistance of winding is a variable value which depends on oil
viscosity. Top oil temperature is given as voltage source and value of top oil
temperature is taken from top-oil temperature model which equation is given in
Equation 2.21.

This circuit is solved with similar way with top-oil model with circuit analysis theory
and differential equation which is given Equation 2.24 is found [6-7, 29].

2 n Susa n Susa d hs ( hs oil ) n Susa 1

K P( )
hs pu pu hs, rated pu w n S usa (2.24)
dt hs, rated

where

Δθhs,rated rated hot-spot temperature rise over top-oil temperature ( K )

τw rated winding time constant ( minutes )

P(θhs)pu load loss pu value varies on temperature change ( pu )

nSusa constant ( given as 0.25 for ONAN transformers )

Load losses’ change depending on temperature is calculated with Equation 2.25.

hs k hs , rated k
P( )
hs pu PDC, pu ( ) Peddy , pu ( ) (2.25)
hs , rated k hs k

PDC,pu DC losses ( I2R losses ) ( in pu )

Peddy,pu eddy losses ( in pu )

θk temperature factor for calculation, equal to 225 for aluminum and 235
for copper;

19
Differential equations which are obtained in Equation 2.21 and 2.24 are solved with a
numerical analysis method to determine the temperature for both oil and hot-spot for
windings.

20
3. EXPERIMENTAL STUDY

An experimental study is realized with a 1000 kVA 33/0.4 kV ONAN cooled

distribution transformer. One goal of this experimental study is determine necessary
parameters for thermal models explained in Section 2. Other goal is to compare and
verify results of thermal models with measured values. During the measurements,
oil temperatures, winding temperatures and ambient temperatures are recorded. Two
different test is done. One of the test is constant load test with the aim of
determination of thermal model parameters. Second test is varying load test to
compare and verify results with dynamic thermal models which are given in Section
2.

Unit, which tests have been performed is a distribution transformer without external
cooling (ONAN cooled). Fifteen Thermocouples has been installed during
manufacturing of transformer. Location of thermocouples have been determined
according to past experiment results to measure hottest points [23-24].
Thermocouples installed to same locations in all three coils to compare temperature
differences between each coil.Measurement devices and thermocouples data is given
in Appendix A. Considering tolerance of thermocouple and monitoring device, total
accuracy of measurements expected ±2%.In this section, experimental study is
explained with three subsections.In first part, realized test unit and realized tests are
explained. In the second part, calculations, which are realized with thermal models,
are given. In the third subsection, measured and calculated values are compared.

3.1 Measurements From Temperature Rise Tests

Two different tests are realized with 1000 kVA distribution transformer. Results of
tests are plotted and necessary parameters for thermal models are determined from
this test. Tested unit details, realized tests and results of tests are given in following
subsections.

21
3.1.1 Tested unit design parameters

Table 3.1 : Nameplate data of tested transformer.

Rated Power 1000 kVA
Primary Voltage 33000 V
Secondary Voltage 400 V
Frequency 50 Hz
Short Circuit Voltage % 6.2 %
Cooling ONAN
Vector Group Dyn11
No-load Losses 1648 W
Load Losses 10177 W
Manufacturing Standard IEC 60076-1

Table 3.1 shows the nameplate data of the tested transformer. These values are taken
from test report of manufacturer.

Figure 3.1 : Active part of transformer.

In Figure 3.1 , basic transformer active part schema is given . There are fifteen
Aluminum foil windings in the secondary winding (LV winding). There are two
cooling channels between the layers, one of them is located between layer five and

22
layer six and the other channel is located between layer ten and layer eleven. The
distance between secondaryconductor to the core yoke is 58 mm from top. It is also
the same for the bottom of winding. 45 mm of this space is pressboard insulation
material.

Primary winding has 14 layers, and total number of turns are 2142.Enameled isolated
aluminum conductor is used as primary winding conductor. There are three cooling
channels in primary winding. First cooling channel is located immediately after
secondary winding and before the start of the primary winding. Second cooling
channel is located between layer 4 and layer 5 and third cooling channel is located
between layer 8 and layer 9. The distance between primary conductor to the core
yoke is 58 mm from top. It is same for bottom. 45 mm of this space is pressboard
insulation material. Internal view of transformer is given in Figure 3.2 and Figure
3.3. In figure 3.3, thermocouples inside of tank are shown.

Figure 3.2 : Internal view of tested transformer.

23
 Figure 3.3 : Internal view of tested transformer with thermocouples.

Transformer tank is corrugated wall design which consists from total of 112 cooling
fins. 25 of this fins are located in each short side of transformer tank and 31 of fins
are located in each long side of transformer. Figure 3.4 and Figure 3.5 shows the
transformer’s exterior view.Quantities of cooling fins and locations are shown in
Figure 3.7.

Transformer is breathing type with conservator. In addition to thermocouples on

transformer, pressure relief device and Buchholz Relay are installed.

In Figure 3.5, thermocouples which are coming inside of tank to outside of tank is
shown. In same figure, monitoring devices for temperature is also shown. During
tests, these monitoring devices are not installed on transformers fins, they are
installed in a separate area.

24
Figure 3.4 : Exterior view of tested transformer from HV side.

Figure 3.5 : Exterior view of tested transformer from LV side.

25
3.1.2 Location of thermocouples

There are total of 15 thermocouples installed on transformer. Locations of

thermocouples are as following:

Five thermocouples are installed in secondary (low voltage) windings. In A and C

coils, two thermocouples are installed. In B coil, three thermocouples installed since
it was expected that hottest temperature occurs in this winding. In figure 3.10,
locations of thermocouples are shown. All thermocouples are installed
approximately 5 mm inside from foils, since it was expected to get hottest
temperature at this location [24]. Thermocouples on windings are shown in Figure
3.6.

Four thermocouples are installed in primary (high voltage) windings. In each coil,
one thermocouple is installed. All thermocouples are installed approximately 5 mm
depth from aluminum conductor material, since it was expected to get hottest
temperature at this location [24]. In addition to this, one thermocouple is installed to
outlet of cooling channel.

One thermocouple is located on the magnetic core of transformer.

One thermocouple is located in oil pocket on transformer cover. One thermocouple is

located on the top of cooling fin Thermocouple on cooling fin is located on the long
side of fins and on the centerline. Locations of thermocouples for oil temperatures
are shown in Figure 3.9.

Transformers top view drawing is given in Figure 3.7. Location of bushings and
thermometer pocket can be seen in that figure. In addition cooling fins of transformer
tank is shown on the drawing which is given in Figure 3.7.

Two thermocouples are installed in the bottom part of the tank. One of them is
located in the bottom part of phase insulation pressboard to measure the mixed
bottom oil temperature and other thermocouple is installed in the lowest part of the
bottom of the tank. Locations of thermocouples are shown in Figure 3.9.

26
 Figure 3.6 : Positions of thermocouples on phase B winding.

Figure 3.7 : Top view of transformer showing thermometer pocket.

27
 Figure 3.8 : Thermocouple in thermometer pocket.

Figure 3.9 : Location of thermocouples in transformer.

28
 Figure 3.10 : Location of thermocouples in windings.

Figure 3.11 : Thermocouples for bottom oil temperature.

29
To measure the ambient temperature during the tests, two opposite side located
thermocouples are installed in the test room with a distance of 1 m from transformer.

During the tests, six of the thermocouples which are installed outtterpary of the tank
are connected to a data logger, which records temperature readings continuously with
15 second intervals. Details of data logger, which designed for measurement and
recording of temperatures are given in Appendix A.

In addition, thermal images of transformer during tests has been taken with thermal
camera which is data is given in Appendix A. However, thermal photos are only
taken to observe distribution of temperature on transformer tank , these data is not
used neither in calculation nor plotting..

Figure 3.12 : Thermocouples located inside of transformer tank.

3.1.3 Applied temperature rise tests

Two different temperature rise tests applied to the transformer. In the first test,
constant nominal load applied. This test realized according to total loss injection
method which is givenin standard IEC 60076-2[32]. According to this method, rated
current and rated voltage is not applied to transformer simultaneously, total losses

30
which are written on routine test report of transformer is injected. Injection method is
to short-circuit transformers secondary side and apply necessary current from
primary side to inject total losses to transformer. This test has two steps. In first step,
total losses ( both no-load and load losses ) are injected to transformer till rate of
change of top oil temperature rise has fallen below 1 K per hour. After this period, in
the second step, only nominal current is applied to transformer from primary side
while secondary is short-circuit condition for 1 hour. At the end of the hour,
connections are removed in shortest time possible and DC resistance measurement
for windings is done. Average winding temperature rise for both high voltage and
low voltage side is determined by using DC resistance measurements after test and
before test in cold condition [32]. Temperature difference is defined by comparing
change on the DC resistance values. DC resistance values are measured with
specially designed test device for this application which applies high currents to
obtain resistance measurement in short time and accurately. Detail of device is given
in Appendix A. In figure 3.13, transformer is shown during first test. As described,
transformer’s secondary is in short-circuit condition and current is applied from
primary side as seen in Figure 3.13.

Figure 3.13 : Transformer during constant load test.

31
In the second test, varying load test is realized. First nominal current (1.00 pu) is
applied for three hours, and following double current ( 2.00 pu) is applied for 1.45
hours. Load steps are given in Table 3.2. this load steps arechosenrandomly by
aiming to obtain a sudden change on loading., and extreme overloading conditions is
applied to transformer during double current step.

Table 3.2 : Load steps of the second test.

Time Period (minutes) Load (pu)
0.0 - 180.0 1.0
180.0 - 285.0 2.0
285.0 - 317.0 0.0

All assembled thermocouple’s measurements have been recorded for both tests.

Both tests are realized in main tap position which is 4th tap with 33000 V rating.

Total fifteen unit thermocouples are used during tests. These thermocouples are three
wire type Pt100 resistance temperature detectors (RTD).Six of these thermocouples
are connected to a continuous data monitoring and recording device and rest of the
thermocouples are connected to three separate Pt100 monitoring device.
thermocouples. Considering tolerance of thermocouple and monitoring device, total
accuracy of measurements expected ±2%. Details of measurement devices are given
in Appendix A.

Continuous data monitoring and recording device has recorded temperature values
for six thermocouples continuously with 15 second intervals. For the other
thermocouples only monitoring devices are used and values are recorded
manually.Taken thermal images during tests are given in Figure 3.14 and Figure
3.15. In Figure 3.14, narrow side of transformer cooling fins are shown during test.
In Figure 3.15, longer side of transformer cooling fins are shown.

32
Figure 3.14 : Thermal camera view from narrow side.

Figure 3.15 : Thermal camera view from long side.

33
3.1.4 Results of tests

Measurements recorded during both tests are analyzed. Top oil temperatures and
bottom oil temperatures which are measured from different locations are compared.
Hot-spot temperatures for both high voltage and low voltage side is determined.
Measured values for both top-oil temperature and winding hot spot temperatures are
plotted. Average winding temperature rise is calculated according to data which is
obtained from temperature rise test. Average oil over ambient temperature rise values
are defined according to different methods which are explained in IEC 60076-2
standard, IEEE C57.12.90 standard and D. Susa’s study [2,24,32]. Results are
compared. Gradients are calculated according to different methods. Results
calculated for gradients are compared.

3.1.4.1 Top-oil temperature

Top oil temperaturesare measured and recorded from three different locations. Most
common measurement method is thermometer pocket measurement since most of the
transformers in market has thermometer pocket and thermometers installed on
existing transformers are assembled in thermometer pockets. In addition to this,
measurements are recorded from thermocouples assembled on top radiator and oil
cooling channel outlet in tank.

Measured values in constant load test (Test 1) are plotted in Figure 3.16. Measured
values in varying load test (Test 2) are plotted in Figure 3.17.For both tests,
measurement results from different top oil temperature locations are given in Figures
3.16 and Figures 3.17. According to measurements, highest top oil temperature is
obtained from the thermocouple, located on outlet of cooling channel.Although
highest top oil temperature is obtained from outlet of cooling channel, these values
are not used in calculations and models. In industry usually no thermocouples are
located inside of distribution transformers’ tank therefore measurement of top oil at
the outlet of cooling channel is not common [24].

34
 80

70

Top-oil Temperatures, ̊C
60

50

40

30

20

10

0
0 100 200 300 400 500 600
Time, minutes

Ambient Temperature T1- Thermometer Pocket

T2 - Top of the cooling fins T3 - Outlet of the Winding Duct

Figure 3.16 : Measured top-oil temperatures at three different locations in first test.

140

120
Top-oil Temperatures, ̊C

100

80

60

40

20

0
0 50 100 150 200 250 300 350
Time, minutes

Ambient Temperature T1- Thermometer Pocket

T2 - Top of the cooling fins T3 - Outlet of the Winding Duct

Figure 3.17 : Measured top-oil temperatures at different locations in second test.

35
In Table 3.3, highest measured values in different locations are given for both test 1
and test 2. As also shown in Figure 3.16 and Figure 3.17,hightest top oil temperature
is obtained from thermocouple, which is located in the outlet of winding,’s cooling
channel. In Table 3.3 T1, T2 and T3 locations are indicating thermocouples at
following locations; T1 is thermometer pocket, T2 is top of the cooling fins and T3
is outlet of the winding cooling channel (duct).

Table 3.3 : Measured highest top-oil temperatures.

Test Load Time , min Location θoil , C
T1 69.0
Test 1
1.0 pu 480 T2 62.5
(Constant Load)
T3 73.0
T1 50.9
1.0 pu 180 T2 49.0
Test 2 T3 54.5
(Variable Load) T1 117,2
2.0 pu 285 T2 111.7
T3 129.0

3.1.4.2 Bottom-oil temperature

Bottom oil temperatures are measured and recorded from two different locations.
One of them is installed inside of the tank to measure mixed bottom oil temperature
and other thermocouple was installed on the lowest bottom part of transformer tank.

36
 60

50
Bottom-oil Temperatures, ̊C
40

30

20

10

0
0 100 200 300 400 500 600
Time, minutes

Ambient Temperature B1- Mixed bottom oil

B2 - Lowest part of tank

Figure 3.18 : Measured bottom-oil temperatures at different locations in first test.

100
90
Bottom-oil Temperatures, ̊C

80
70
60
50
40
30
20
10
0
0 50 100 150 200 250 300 350
Time, minutes

Ambient Temperature B1- Mixed bottom oil temperature

B2 - Lowest part of tank

Figure 3.19 : Measured bottom-oil temperatures at different locations in second test.

37
The Figure 3.18 shows the bottom oil temperatures measured from different
locations in first test. Figure 3.19 shows the top oil temperatures measured from
different locations in test 2.

In Table 3.4 highest measured values of bottom oil temperatures in different

locations are given for both Test 1 and Test 2.

Hottest bottom oil temperature is measured in mixed bottom oil temperature

thermocouple. It has been observed that mixed bottom oil temperature has
responded to load changes analogous with the top oil temperature. In addition,
measurements on thermocouple located in the bottom of the tank has shown that, oil
in the lowest part of tank, has not been changed significantly in both of the tests as
clearly shown in Figure 3.18 and Figure 3.19. This is a result of low natural oil
circulation under active part and under the cooling fins. Therefore, measurement
data, which is received from the thermocouple, located in the lowest part of tank is
not used in calculations and analyzes.

Table 3.4 : Measured highest bottom-oil temperatures

Test Load Time , min Location θbo , C
Test 1 B1 48.6
1.0 pu 480
(Constant Load) B2 25.2
B1 36.5
1.0 pu 180
Test 2 B2 18.7
(Variable Load) B1 86.2
2.0 pu 285
B2 24.7

where;B1 is indicating the location of thermocouple for mixed bottom oil

temperature and B2 is indicating location of thermocouple for lowest part of the
tank inside transformer.

3.1.4.3 Hot-spot to top-oil temperature rise

Hottest spot temperatures are defined by comparing the hottest measurement values.
For both primary (high voltage) and secondary (low voltage) windings, highest
temperature measured in phase B which is the mid located phase. This result is
reasonable since mid located phase windings are affected from both other two

38
windings temperature raises. On the other hand, the difference is not
significant.Measured values in constant load test are plotted in Figure 3.20. In figure
3.21, varying load test results for hot-spot temperatures are given. It also noted that
low voltage winding hot spot temperatures are higher than high voltage winding’s
hot spot temperatures. This result is as expected since low voltage winding is internal
located between magnetic core and primary windings, therefore oil circulation and
heat transfer is expected less than high voltage windings.Location of thermocouples
which measured the hottest points are shown in Figure 3.10.

100
90
80
Hot spot Temperatures, ̊C

70
60
50
40
30
20
10
0
0 100 200 300 400 500 600
Time, minutes

Ambient Temperature Low voltage winding hottest spot

High voltage winding hottest spot

Figure 3.20 : Winding hot spot temperatures in the first test (constant load test)

39
 250

200
Hot spot Temperatures, ̊C

150

100

50

0
0 50 100 150 200 250 300 350
Time, minutes

Ambient Temperature Low voltage winding hottest spot

High voltage winding hottest spot

Figure 3.21 : Winding hot spot temperatures in the second test (varying load test)

Hottest spot rise above top oil temperature is calculated for both high voltage and
low voltage winding. During this calculation top oil temperature has been taken from
different locations and results are compared in Table 3.5.

Table 3.5 : Hot-spottemperatıre rises to top oil temperatures

Δθhs, rated , K
Time, min. Load Location
LV HV
T1 24.3 21
180 1 pu T2 27.3 24
T3 20 17
T1 73 66
285 2 pu T2 79 70
T3 62 56

In this study rated hot-spot to top oil temperature rise is take by using oil temperature
in thermometer pocket. Reason of this is, oil thermometer pocket is most common
measurement location for top oil temperature in industry as already explained in

40
Section 3.1.4.1. Hottest spot rise over top oil pocket temperature is plotted in
Figure3.22. It had been observed that the hot-spot temperature rise over top-oil
temperature had not experienced with overshoot. This were expected since it had
been verified in the past studies that in transformers without external cooling , a
function with overshootshould not be observed [23,28].

30
Hot spot to top oil Temperature rises, K

25

20

15

10

0
0 50 100 150 200 250
Time, minutes

Low voltage winding hottest spot rise over top oil

High voltage winding hottest spot rise over top oil

Figure 3.22 : Hot-spot to top-oil temperature rises

3.1.4.4 Average temperature rises

As a part of constant load test, resistance values are measured immediately after tests
as per related IEC standard [32]. Measured values and cold resistance values are
given Table 3.6 and Table 3.7.

Average winding temperatures are defined with extrapolation graph.Average

winding temperature was defined with following formula from resistance values[32].

Ru
u ( i k ) k (3.1)
Ri

41
Where;θu is calculated temperature(in C);θiiscold temperature (initial measured
value) (in C) ;θkis temperature factor for calculation, equal to 225 for aluminum; 235
for copper;Riis the cold resistance (initial measured value) (in ohm) and Ru is
thecalculated resistance (in ohm).

Before starting temperature rise test, during transformer is in cold condition, initial
resistance values are measured and recorded. These values are given in Table 3.7.
After completing test, during transformer in warm condition, resistance measurement
test of same winding is repeated. Measured values are given in Table 3.6. Since
transformer winding is cooling down during measurement, more than ten measured
resistance values are recorded. It is shown in Table 3.6 that resistance values are
continuously reducing.Measurement is realized with a sensitive resistance
measurement device which is data is given in Appendix A. Considering that, an
amount of time is lose during removing connections of test and connecting test
device for resistance measurement, resistance at the end of the test is determined
with an extrapolation. This extrapolation is realized by plotting measured values and
finding equation of this function from graph.

Table 3.6 : Measured warm resistances

Time Rhv Rlv
sec. ohm μohm
45 11.665 -
60 11.517 1255
75 11.423 1251
90 11.364 1246
105 11.322 1243
120 11.284 1242
135 11.250 1240
150 11.225 1240
165 11.200 1239
180 11.181 1238
195 11.166 1237
210 11.166 1236

42
 14
RHV = -3E-07t3 + 0,0001t2 - 0,02t + 12,364
12
Warm Resistance values, ohm

10

0
0 50 100 150 200 250
Time, sec.

Figure 3.23 : Warm resistance values for high voltage winding.

1320
LV Warm Resistance Values, μohm

1300
RLV = -1E-05t3 + 0,0054t2 - 0,9166t + 1293
1280
1260
1240
1220
1200
1180
1160
1140
1120
0 50 100 150 200 250
Time, sec.

Figure 3.24 : Warm resistance values for low voltage winding.

According to extrapolation, values from plotted graphs which are given in Figure
3.23 for high voltage winding and given in Figure 3.24 for low voltage winding,
warm resistance values were defined. From warm resistance values, average winding
temperatures are calculated by using Equoation 3.1. Calculated values are given in
Table 3.7.

43
 Table 3.7 : Calculated resistance and average winding temperatures
Warm
Cold Resistances θw,average, C Resistances θw,average, C
HV LV HV LV HV LV HV LV
9.92 1028 12.364 1293
17 17
ohm μohm ohm μohm 76.62 79.38

It is very important to define average oil temperature correctly to determine average

winding to average oil temperature rise ( gr) correctly. It is relatively difficult to
define average temperature rise for the transformer without external cooling
compared to transformers with external cooling. Because, in transformer without
external cooling, there is a complex natural oil circulation inside of tank[24].

In this study, average oil temperature rises are defined with different methods and
results are compared.

Gradients calculated according to IEC 60076-2

According to IEC standard, average oil temperature is calculated only using top oil
temperature. Equation is given in Equation 3.2 [32].

oil,average 0.8( oil amb ) amb (3.2)

According to equaiton which is given in Equation 3.2, average oil temperature is

calculated. Calculated average oil temperature’s difference between hottest spot
winding temperatures for both high voltage and low voltage windings are given in
Table 3.8.

Table 3.8 : Average winding – average oil temperature rise according to IEC method
According to IEC 60076-2
θoil, average , gr, K H, pu
Location
C LV HV LV HV
T1 58.66 20.72 17.96 1.17 1.17

H is the hot spot factor and it is calculated with following equoation;

44
 hs
H (3.3)
gr

Gradients calculated according to IEEE C57.12.90

According to IEEE standard, average oil temperature is calculated with following

equotation [2].

oil,average oil 0.5( oil, top surface bottom,surface ) (3.4)

Average winding to average oil tempeatureis calculated with following equotaiton;

gr w , acerage oil, average (3.5)

θw,averageis calculated from resistance measurements with Equation 3.1 and results are
shown inTable 3.7.

H is the hot spot factor and it is calculated with Equoation 3.3.

From this equation, calculated average oil temperature values are given in Table 3.9.

Table 3.9 : Average winding- average oil temperature rise according to IEEE
According to IEEE
θoil, average , gr, K H, pu
Location
C LV HV LV HV
T1 50.35 29.03 26.27 0.84 0.80

Gradients calculated according to bottom oil temperature measured in tank

Considering that , measured top oil temperature and bottom oil temperatures are for
mixed oil temperatures, average mixed oil temperature is calculated with following
equotaiton [24];

oil bo
oil,average (3.6)
2

Top oil temperature is taken 69 C from first test ( constant load test) results which is
shown Table 3.3. Bottom oil temperature is taken 48.6 C from Table 3.4.

45
Average winding to average oil tempeatureis calculated with Equation 3.6.H is the
hot spot factor and it is calculated from equoation 3.3.

Calculated values are given in Table 3.10.

Table 3.10 : Average winding – average oil temperature rise according to D. Susa
According to D. Susa Method
θoil, average , gr, K H, pu
Location
C LV HV LV HV
T1 58.80 20.58 17.82 1.18 1.18

It is observed that the difference between average winding to average oil temperature
rise values calculated according to IEC standard and calculated according to bottom
oil temperature measurements in tank (D.Susa Method) gives similar results. On the
other hand, values had been calculated according to IEEE method, are notably higher
than the calculated values according to bottom oil temperature measurements in tank.

3.2 Results Obtained With Thermal Models

Top oil temperature and hot-spot winding temperatures for both high voltage and low
voltage windings are calculated with three different methods which are explained in
Section 3. Data required for these models are obtained from test 1 which is constant
load test as explained in previous part of study. Optimization is done for defining
some parameters which are required in thermal models. Data which are used for
calculation of thermal models are given in Table 3.11. Average oil temperature rises
and average winding temperature rises are determined according to different methods
which are given in IEC, IEEE standards and Susa’s study [6,7,24,28]. Results are
compared between each other and between measured values during test 2 which is
varying load test.

46
Table 3.11 : Data which is used in thermal models
Windings
Quantity Unit
LV HV
kVA Base 1000 1000 kVA
Temp. Base 75 75 C
PDC 9550 W
Peddy+ Pstray 627 W
Pfe 100 W
pu Base 1000 1000 kVA
Δθw/A,R 62.08 59.32 K
Δθhs,rated 24.3 21 K
Δθoil,rated 51.7 K
Δθbo,rated 31.3 K
H 1.181 1.178
gr 20.58 17.82 K
τoil,rated 199.04 min
τw (IEEE) 5 5 min
τw (IEC) 8.2 8.6 min
k11 0.822
k21 1.0
k22 2.0
x 0.8
y 1.6
nIEEE 0.8
m 0.8
nSusa 0.25
Map 1810 kg
Mfe 1205 kg
Mw 114 192 kg
Mtank 550 kg
Moil 930 kg
θhs,i 24 25 K
θoil,i 20.8 K
θbo,i 15.1 K

47
For all models; power rating base is taken as 1000 kVA since transformer rated
power is 1000 kVA, temperature base is taken as 75 C. Temperature base is
important to define load losses correctly since load losses depends on temperature
changes. It is very important especially in D. Susa method, since affect of
temperature change on losses are considered, base temperature value is very
important.

No-load losses are taken reduced since Test 2 ( varying load test ) is realized with
short-circuit method which is already explained in Section 3.1.3. Other loss values
are obtained from manufacturer’s test report. It is confirmed that, given load losses
on test report are with the same temperature base which is 75 C as given in Table
3.11.

Δθw/A,R which indicated rated average winding temperature increase over ambient is
calculated by using average winding temperature rise values which are given in
Table 3.7 and ambient temperature during test.

Ambient temperature during test is measured averagely 17.3 C as shown in Figure

3.16, Figure 3.16 and Figure 3.18.

Top-oil temperature rise over ambient which is taken from Table 3.3 by using
ambient temperature which is explained in previous paragraph.Δθhs,ratedis taken from
Table 3.5 by using oil temperature in thermometer pocket ( T1 ).

Top oil temperature rise over ambient temperature which is shown with Δθoil,rated in
Table 3.11 is calculated by using top oil temperature value in Table 3.3 and ambient
temperature. Similarly bottom oil temperature rise over ambient (Δθ bo,rated ) is
calculated from Table 3.4 by using ambient temperature.

Hot spot factor ( H ) is calculated from Equation 3.3 by using rated hot-spot
temperature rise over top oil temperature which is given in Table 3.5 and gradient
which is calculated according to in Table 3.10.

Rated oil temperature time constant (τoil,rated) is calculated from Equation 2.3.
Winding time constants for both high voltage and low voltage windings for IEEE
model, are taken as 5 minutes as per recommendation of the standard [6]. Winding

48
time constants for IEC and Susa’s models arecalculated according to equation which
is given in section 2.2.3 , Equation 2.19.

Constants required for IEC thermal model is indicated as k11, k12, k22, x and y as
already explained in Section 2.2. k11 value is determined by realizing optimization
as described in Equation 2.13 and 2.14. Optimization is realized by minimizing
function which is given in Equation 2.14. Other constants are taken from IEC 60076-
7 recommendations which are given in Table 2.1.

Constants for IEEE C57.91-1995 method which are shown as nIEEE and m in Table
3.11 are taken from recomandations of this standard[6].

Weight data of transformer’s various parts are obtained from manufacturer. These
data is indicated as Map, Mwdn, Mtankand Moil in Table 3.11 .

Initial temperatures for top oil (θoil,i), bottom oil (θbo,i) and hot spot temperatures for
both high voltage and low voltage windings (θhs,i)are taken from measurement
records of Test 2 which is varying load test. Initial temperatures of hottest spots and
top oil were not same with ambient temperature. These temperatures were slightly
higher because of the, effect of previous temperature rise test.

All thermal models, are solved for same load cycle which is applied during Test 2
which is given in Table 3.2.

Exponential equations for IEC and IEEE method which are given and explained in
Section 2.1 and Section 2.2 are solved for each interval and results are plotted in
Figure 3.25 , Figure 3.26 and Figure 3.27.

Differential equations in D. Susa’s method which are given and explainedin Section
2.3, are solved with fourth order Runge-Kutta method. Obtained results are plotted
on Figure 3.25 , Figure 3.26 and Figure 3.27.

3.3 Comparison Of Measured Values With Results Of Thermal Models

Results obtained by thermal models in Section 3.2 are compared in this part. Both
top-oil temperatures and hot-spot temperatures for high voltage and low voltage
windings are determined with thermal models. These calculated values are plotted

49
and compared with measured values.Identificaiton of methods on graphs are as
following:

“Measurement” refers to results measured during test 2 ( varying load test ),

“IEEE Method” refers to the model which is given in Section 2.1, “IEC
Method” refers to the model , which is given in Section 2.2“TM”
refers to the model, which is given in Section 2.3

In Figure 3.25, top oil temperature values which are measured and calculated with
thermal models are given. It is observed that IEEE method give slightly higher
results, IEC method give slightly lower results and D.Susa thermal model give more
accurate results compared to measured values.

140,0

120,0

100,0
Top oil temperature, ̊C

80,0

60,0

40,0

20,0

0,0
0 50 100 150 200 250 300 350
Time, minutes
Measurement IEEE Method
IEC Method TM

Figure 3.25 : Top oil temperature for varying load test

In Figure 3.26, hot-spot temperatures for low voltage winding’s are plotted. In
Figure 3.26, both measured and calculated with thermal models are given. It is
observed that IEEE method give significantly higher results, IEC method give
slightly lower results and D. Susa thermal model gives more accurate results

50
compared to measured values. On the other hand D. Susa method, does not give
accurate results in first minutes of load step change which shows winding time
constant is not defined correctly.

200,0
LV hottest spot temperature, ̊C

150,0

100,0

50,0

0,0
0 50 100 150 200 250 300 350
Time, minutes

Measurement IEEE Method

IEC Method TM
Ambient Temperature

Figure 3.26 : LV hot-spot temperature for varying load test

200,0
HV hottest spot temperature, ̊C

160,0

120,0

80,0

40,0

0,0
0 50 100 150 200 250 300 350
Time, minutes
Measurement IEEE Method
IEC Method TM
Ambient Temperature

Figure 3.27 : HV hot-spot temperature for varying load test

51
In Figure 3.27, hot spot temperature values for high voltage winding which are
measured and calculated with thermal models are given. It is observed that
differences between measured values and calculated values are very similar to
differences for low voltage winding which is shown in Figure 3.26. It is noticed that
IEEE method give significantly higher results and IEC method gives slightly lower
results. D. Susa method gives accurate results especially for the hottest temperature.

52
4. CONCLUSIONS

Distribution transformers are one of the most important equipmentsin electrical

distribution systems. Operation of transformer with higher temperatures, which is
why they are designed for, causes increase in aging rate. An aging transformer, could
fail before economical life time, this will obviously cause additional costs. The
common reason of failures in this case, is insulation material deformations on layer
insulations and also turn to turn insulations.

In the industry two basic thermal models are widely used which are proposed by IEC
and IEEE. These two models uses exponentials formulas and neglects the affect of
viscosity change with oil temperature change. Thermal model which is developed by
Dejan Susa, takes into account the oil viscosity changes and loss variations with
temperature changes. In addition to oil viscosity and loss variations, it also takes into
account the change on time constants.

In this study, average winding to average oil gradients which are calculated
according to different methods were compared with each other. It is observed that ,
IEC method gives very similar results with well established calculations methods.
On the other hand, IEEE method, gives significantly higher values compared to well-
established calculation methods.

For top oil temperature, three different models’ results are compared with measured
values. In all models, proper rated oil time temperature rise values are used which
were obtained during temperature rise tests with well-established methods. Also
affects of ambient temperature changes are adapted on all thermal models. Therefore
it is noticed that, all three methods supplied similar results to measured values.
Thermal model which is developed by D. Susa, gave the best result for all load
cycles except slight difference on cooling down condition. It is observed that values
calculated according to IEEE method , are considerably higher from measured values
and for IEC method , it is just the opposite and calculated values are considerably
lower than measured values.

53
For hot-spot temperature, three different models’ results are compared with the
measured values for both high voltage and low voltage windings. As in top oil
model, in hot-spot model also, proper rated winding hot-spot temperature rise values
are used. It is observed that the values obtained from IEEE method are slightly
higher than the measured values and the values obtained from IEC method are
slightly lower than measured values. According to results obtained from thermal
model which developed by D. Susa, highest hottest-spot temperature is calculated
very accurately compared to measured data. On the other hand, a significant
difference is noticed during first minutes of step load change. This has shown that
winding time constant for this transformer is not defined properly.

According to temperature rise tests, it is observed that, highest winding temperature

was occurred in B phase which is located in the mid of active part.

It is observed that, results are very dependent on accurate input data for defined rated
load parameters like rated temperature rise values and rated top-oil time constants.

Considering the difficulty of defining, winding time constant with temperature rise
tests, further research and development is needed to improved defining the winding
time constant for distribution transformers.

Distribution transformers which are feeding non-sinusoidal loads are continuously

increasing in the industry. Hot-spot temperature rises are highly affected with the
affect of harmonic contents. Considering IEEE C57-110-98 standard is conservative
and does not consider real load cycle, further research and developments are needed
for ONAN cooled transformers which are without external cooling that areloaded
with harmonic contents. This has been left as a further research activity.

54
REFERENCES

[1] Del Vecchio, R. M., Poulin, B., Feghali, P.T., Shah, D.M. and Ahuja, R.
(2002).Transformer Design Principles: With Applications to Core
Form Transformers.CRC Press 2001.
[2] IEEE (2006). IEEE Std C57.12.90-2006 : IEEE Standard Test Code for Liquid-
Immersed Distribution,Power and Regulating Transformers, IEEE.
[3] IEC (2011).IEC 60076-1 ed3.0Power transformers – Part 1: General, IEC.
[4]Kömürgöz, G. (2002). Kendi kendine soğuyan kuru tip güç transformatörlerinde
sargı ısınma hesabına katkılar, PhD Thesis, ITU.
[5] IEC (2007). IEC 60085-1 ed4.0Electrical insulation – Thermal evaluation and
designation, IEC.
[6] IEEE (1995). IEEE Std C57.91-1995 IEEE Guide for Loading Mineral-Oil-
Immersed Transformers, IEEE.
[7] IEEE (2002). IEEE Std C57.91-1995/Cor 1-2002 IEEE Guide for Loading
Mineral-Oil-Immersed Transformers Corrigendum 1, IEEE.
[8]Betta, G., Pietrosanto, A., and Scaglione, A. R. (2000).An enhanced fiber optic
temperature sensor system for power transformer monitoring,
Instrumentation and Measurement Technology Conference, IMTC
2000, Proceedings of the 17th IEEE, Vol.1, 1-4 May 2000.
[9]Feng, J.Q., Sun, P., Tang, W.H., Buse, D.P., Wu, Q.H., Richardson, Z. and
Fitch, J. (2002). Implementation of a power transformer temperature
monitoring system, International Conference on Power System
Technology, ProocedingsPowerCon 2002., Volume 3, 13-17 October
2002, p. 466-473.
[10]Lachman, M.F., Griffin, P.J., Walter, W. and Wilson, A. (2003). Real-time
dynamic loading and thermal diagnostic of power transformers, IEEE
Transactions on Power Delivery, Volume 18, Issue 1, January 2003, p.
142-148.
[11] Swift, G.W., Zocholl, E.S., Bajpai, M., Burger, J.F., Castro, C.H., Chano,
S.R., Cobelo, F.S.P., Fennell, E.C., Gilbert, J.G., Grier, S.E., Haas,
R.W., Hartmann, W.G., Hedding, R.A., Kerrigan, P., Mazumdar,
S., Miller, D.H., Mysore, P.G., Nagpal, M., Rebbapragada, R.V.,
Thaden, M.V.,Uchiyama, J.T., Usman, S.M., Wardlow, J.D. and
Yalla M. (2001) Adaptive transformer thermal overload protection,
IEEE Transactions on Power Delivery, Volume 16, Issue 4, October
2001, p. 516-520.
[12]Lesieutre, B.C., Hagman, W.H., Kirtley and J.L.Jr. (1997).An improved
transformer top oil temperature model for us in an on-line monitoring

55
 and diagnostic system, IEEE Transactions on Power Delivery,
Volume 12, Issue 1, January 1997, p. 249-256.
[13]Bergman, L. T., Incropera, F.P., Lavine, A.S. and Dewitt, D.P. (2011).
Fundamentals of Heat and Mass Transfer, Seventh Edition, John
Wiley & Sons, Inc.
[14]Aubin, J.R., Bergeron, R. and Morin, R. (1990). Distribution Transformer
Overloading Capability Under Cold-Load Pickup Conditions, IEEE
Transactions on Power Delivery, Volume 5, Issue 4, October 1990
1997, p. 1883-1891.
[15]Blume, L. F., Boyajian, A., Camilli, G., Lennox, T.C., Minecci, S. and
Montsinger V.M. (1951). Transformer Engineering, 2nd edition,
John Wiley & Sons, Inc.
[16] Grubb, R.L., Hudis, M. and Traut, A.R. (1981). A Transformer Thermal Duct
Study of Various Insulating Fluids, IEEE Transactions on Power
Apparatus and Systems, Volume PAS-100, No. 2, February 1981, p.
466-473.
[17]Karsai, K., Kerenyi, D. and Kiss L. (1987). Large Power Transformers,
Elsevier Science Publishers.
[18]Nordman, H., Hironniemi, E. and Pesonen, A.J. (1990). Determination of hot-
spot temperature rise at rated load and at overload, CIGRE Paper 12-
103.
[19]Nordman, H., Rafsback, N., and Susa D. (2003). Temperature responses to
step changes in the load current of power transformers, IEEE
Transactions on Power Delivery, Volume 18, Issue 4, October 2003,
p. 1110-1117.
[20] Tang, W.H., Wu, Q.H., and Richardson Z.J. (2002). Equivalent heat circuit
based power transformer thermal model, Electric Power Applications,
IEE Proceedings, Volume 149, Issue 2, March 2002, p. 87-92.
[21]Tang, W.H., Wu, Q.H., and Richardson Z.J. (2004). A simplified transformer
thermal model based on thermal-electric analogy, IEEE Transactions
on Power Delivery, Volume 19, Issue 3, July 2004, p. 1112-1119.
[22]Pierce, L.W. (1992). An investigation of the thermal performance of an oil filled
transformer winding, IEEE Transactions on Power Delivery, Volume
7, Issue 3, July 1992, p. 1347-1358.
[23]Radakovic, Z., Feser, K., (2003). A new method for the calculation of the hot-
spot temperature in power transformers with ONAN cooling, IEEE
Transactions on Power Delivery, Volume 18, Issue 4, October 2003,
p. 1284-1292.
[24]Susa, D., Lehtonen, M. andNordman, H. (2005). Dynamic Thermal Modelling
of Distribution Transformers, IEEE Transactions on Power Delivery,
Volume 20, Issue 3, July 2005, p. 1919-1929.
[25] Swift, G., Molinski, T.S., and Lehn, W. (2001). A fundamental approach to
transformer thermal modeling-I, Theory and equivalent circuit, IEEE
Transactions on Power Delivery, Volume 16, Issue 2, April 2001, p.
171-175.

56
[26] Swift, G., Molinski, T.S., Bray, R. and Menzies, R. (2001). A fundamental
approach to transformer thermal modeling-II, Field Verification, IEEE
Transactions on Power Delivery, Volume 16, Issue 2, April 2001, p.
176-180.
[27]Susa, D. (2005). Dynamic Thermal Modelling of Power Transformers , Doctoral
Dissertation, Helsinki University of Technology.
[28] IEC (2005).IEC 60076-7 ed1.0Power transformers – Part 7: Loading guide for
oil-immersed power transformers, IEC.
[29]Susa, D., Lehtonen, M. andNordman, H. (2005). Dynamic Thermal Modelling
of Power Transformers, IEEE Transactions on Power Delivery,
Volume 20, Issue 1, January 2005, p. 197-204.
[30]Rivera, L.J., Tylavsky, D.J. (2007). Dynamic Thermal Modelling of Power
Transformers, Power Systems Engineering Research Center,
November 2007 .
[31] Susa, D. and Nordman, H. (2012). IEC 60076-7 loading guide thermal model
constants estimation, European Transactions on Electrical Power 2012
, DOI: 10.1002/etep.1631.
[32] IEC (2011). IEC 60076-2 ed3.0Power transformers – Part 2: Temperature rise
for liquid-immersed transformers, IEC.
[33]Radakovic, Z., Kalic, D., (1997). Results of a novel algorithm for the
calculation of the characteristic temperatures in power oil
transformers, Electrical Engineering 80, p. 205-214.

57
58
APPENDICES

APPENDIX A:Test Equipment DataAnd Photos

59
APPENDIX A

In this appendix part, data of measurement devices are given. During this study,
mainly following devices are used; thermocouples, temperature monitoring and
recording devices, winding resistance measurement device. In this section datasheets
for temperature monitoring and recording devices are given in Figure A.1, Figure
A.3, and Figure A.4. In addition to this, in Figure A.2, Figure A.5, Figure A.6 ,
Figure A.7, Figure A.8 and Figure A.9, photos of the test devices which are used in
this study are shown.

Figure A.1 : Temperature monitoring and recording device

60
Table A.1 : Technical datasheet for temperature monitoring and recording devices

61
Table A.2 : Technical datasheet of temperature monitoring device

62
Figure A.2 : Temperature monitoring and recording devices in varying load test.

63
 Figure A.3 : Thermal camera

Table A.3 : CPC 100 winding resistance measurement test device datasheet

64
Figure A.4 : CPC 100 winding resistance measurement test device view.

65
Figure A.5 : Test room while recording temperature reading during test 1

66
Figure A.6 : Tested transformer view from test laboratory during test

67
68
CURRICULUM VITAE

Name Surname: Oluş Sönmez

Place and Date of Birth:Kadıköy, 1987

Address: MehtapSokakBurçak Apt. Daire:4 No:18 Kadıköy / Ġstanbul

E-Mail: [email protected]

B.Sc.: Yıldız Technical University / Electrical Engineering

69