# Transformer Design and Optimization: A Literature Survey

**ABSTRACT** With the fast-paced changing technologies in the power industry, new references addressing new technologies are coming to the market. Based on this fact, there is an urgent need to keep track of international experiences and activities taking place in the field of modern transformer design. The complexity of transformer design demands reliable and rigorous solution methods. A survey of current research reveals the continued interest in application of advanced techniques for transformer design optimization. This paper conducts a literature survey and reveals general backgrounds of research and developments in the field of transformer design and optimization for the past 35 years, based on more than 420 published articles, 50 transformer books, and 65 standards.

**3**Bookmarks

**·**

**457**Views

- [Show abstract] [Hide abstract]

**ABSTRACT:**This paper compares the application of two deter-ministic and three nondeterministic optimization algorithms to global transformer design optimization (TDO). Two deterministic optimization algorithms (mixed-integer nonlinear programming and heuristic algorithm) are compared to three nondeterministic approaches (harmony search, differential evolution, and genetic algorithm). All these algorithms are integrated in design optimiza-tion software applied and verified in the manufacturing industry. The comparison yields significant conclusions on the efficiency of the algorithms and the selection of the most suitable one for the TDO problem.IEEE Transactions on Industrial Electronics 01/2014; 50(1). · 5.17 Impact Factor - SourceAvailable from: Eleftherios I. Amoiralis[Show abstract] [Hide abstract]

**ABSTRACT:**Increase of temperature rise and hot spot values affects dramatically transformer aging and life expectancy. The present article investigates several improved designs of ONAN transformer cooling system by means of advanced numerical heat transfer-fluid flow model. Novel tank designs are examined in conjunction with other crucial parameters, as the number and location of the winding cooling ducts, so as to define the best geometry that ensures maximum efficiency of the transformer cooling system, with the aim of expanding transformer lifetime.IEEE Transactions on Dielectrics and Electrical Insulation 06/2012; 1900. · 1.36 Impact Factor - [Show abstract] [Hide abstract]

**ABSTRACT:**This paper discusses issues related to the transformer design for switch-mode power supplies based on the use of geometric form factors (GFFs) of magnetic cores. It is shown that GFFs allow a straightforward and reliable transformer design complying with loss and size constraints. It is also proved that inherent assumptions underlying the application of previous design methods based on some geometric constants can be penalizing if the effects of specific combinations of GFFs of magnetic cores are not properly considered. Examples regarding the design of transformers for forward converters are presented and discussed.IEEE Transactions on Industrial Electronics 06/2013; 60(6):2158-2166. · 5.17 Impact Factor

Page 1

IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 20091999

Transformer Design and Optimization:

A Literature Survey

Eleftherios I. Amoiralis, Member, IEEE, Marina A. Tsili, Member, IEEE, and Antonios G. Kladas, Member, IEEE

Abstract—With the fast-paced changing technologies in the

power industry, new references addressing new technologies are

coming to the market. Based on this fact, there is an urgent need

to keep track of international experiences and activities taking

place in the field of modern transformer design. The complexity

of transformer design demands reliable and rigorous solution

methods. A survey of current research reveals the continued

interest in application of advanced techniques for transformer

design optimization. This paper conducts a literature survey and

reveals general backgrounds of research and developments in the

field of transformer design and optimization for the past 35 years,

based on more than 420 published articles, 50 transformer books,

and 65 standards.

Index Terms—Analytical methods, artificial intelligence, equiv-

alent circuits, experimental methods, hybrid methods, numerical

techniques, standards, survey, transformer books, transformer

design, transformer design optimization, transformer modeling,

transformers.

I. INTRODUCTION

I

books, and reports about new models have been published in

the technical literature due mostly to the improvement of the

computer power availability, new innovative optimization algo-

rithms,andthegreateruncertaintylevelintroducedbythepower

sector deregulation.

Transformerdesignisacomplextaskinwhichengineershave

to ensure that compatibility with the imposed specifications is

met, while keeping manufacturing costs low. Moreover, the de-

sign methodology mayvarysignificantly according to the trans-

former type (distribution, power or instrument transformer) and

its operating frequency (ranging between 50/60 Hz and a few

megahertz), while many alterations according to the core con-

structionalcharacteristics,thecoolingmethod,orthetypeofthe

magnetic material may be encountered [1], [2]. This paper pro-

vides anoverviewofresearch,development,and theapplication

ofvariouscomputationalmethodsfortransformerdesign,based

N the last years, research in the area of transformer de-

sign experienced an expansion. Many papers, standards,

Manuscript received November 14, 2008; revised May 10, 2009 Current ver-

sionpublishedSeptember 23,2009. Thispaper ispartofthe 03ED045 Research

Project that is co-financed by E.U.-European Social Fund (75%) and the Greek

Ministry of Development-GSRT (25%). Paper no. TPWRD-00839-2008.

E. I. Amoiralis is with the Department of Production Engineering & Man-

agement, Technical University of Crete, Chania GR-73100, Greece (e-mail:

eamir@tee.gr).

M. A. Tsili and A. G. Kladas are with the Faculty of Electrical & Com-

puter Engineering,NationalTechnicalUniversityofAthens,Athens GR-15780,

Greece (e-mail: mtsili@central.ntua.gr; kladasel@central.ntua.gr).

Color versions of one or more of the figures in this paper are available online

at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TPWRD.2009.2028763

Fig. 1. Sources of the literature survey.

on an extensive number of published papers [3]. The referred

publications are mainly extracted from IEEE TRANSACTIONS,

IEEE Magazines, IEEE Proceedings, IEE/IET (Institution of

Electrical Engineers, currently Institution of Engineering and

Technology) Proceedings as well as a few, yet very important

conferences in an effort to cover the majority of published pa-

persinthetransformerdesignfield(Figs.1and2).However,due

to the amplitude of this field, such a survey, no matter how com-

prehensive, cannot be exhaustive. The review is divided into six

major sections: 1) Research efforts focusing on the prediction

and/or optimization of specific transformer characteristics; 2)

techniques adopted for global transformer design optimization,

taking into account all of the relevant performance parameters;

(3) transformer post-design performance and modeling; 4) stan-

dards governing the transformer qualification; 5) recent trends

on transformer technology; and 6) transformer books giving to

the reader a convenient starting point concerning important as-

pects of transformer engineering. The references are grouped

andpresentedaccordingtotheirmethodologicalapproach,how-

ever, no comparative analysis or specific details of the method-

ologies are provided in order to keep the survey as compact

and comprehensive as possible. The research focuses mainly on

power and distribution transformers and other types of trans-

formers operating at low frequencies, and is not expanded to

transformers for high frequency applications, apart from sev-

eralcasesofdesignmodelsandmethodologiesapplicabletothis

range of frequencies.

The present bibliographical survey will be particularly useful

for: 1) transformer designers and researchers engaged in trans-

formerdesign,optimization,andquality-enhancementactivities

0885-8977/$26.00 © 2009 IEEE

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 2

2000IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 2009

Fig. 2. Percentage participation of different IEEE Transaction Journals in the

overall amount of IEEE Transaction Journals of the survey.

in today’s competitive environment; 2) utility engineers who

would like to enrich their educational background about the

system interaction aspects of transformers in a power system;

and 3) undergraduate and postgraduate students who wish to in-

tegrate traditional transformer theory with modern computing

practices.

The paper is organized as follows: Section II describes the

various transformer types that are present in the relevant bib-

liography as well as the main considerations during the trans-

former design process. Section III includes the survey overview

of research dedicated to transformer characteristics, while Sec-

tionIVprovidesanoverviewoftheresearchconductedontrans-

former design optimization. Section V lists publications dedi-

cated to modeling transformers for power system studies. Sec-

tion VIprovidesa comprehensiveoverviewoftransformer stan-

dards that are issued by international organizations. Section VII

addresses new considerations in transformer design, under the

light of recent developments in the electric power industry as

well as the technologies involved in transformer construction,

providing an up-to-date review of modern trends in transformer

design. Section VIII presents a brief description of transformer

books. Finally, Section IX concludes this paper.

II. TRANSFORMER DESIGN

A transformer has been defined by ANSI/IEEE [4] as a static

electric device consisting of a winding, or two or more coupled

windings, with or without a magnetic core, for introducing

mutual coupling between electric circuits. Transformers are

extensively used in electric power systems to transfer power

by electromagnetic induction between circuits at the same

frequency, usually with changed values of voltage and cur-

rent. Transformers are one of the primary components for the

transmission and distribution of electrical energy. Their design

results mainly from the range of application, the construction,

the rated power and the voltage level.

A. Transformer Types

Different kinds of transformer types may be encountered, ac-

cording to their use, their cooling method or the construction

of their magnetic circuit. A major classification is realized ac-

cording to the power and voltage ratings: transformers with a

rated power up to 2.5 MVA and a voltage up to 36 kV are re-

ferred to as distribution transformers, while all transformers of

higher ratings are classified as power transformers.

In particular, a power transformer has been defined by

ANSI/IEEE [4] as a transformer that transfers electric energy

in any part of the circuit between the generator and the dis-

tribution primary circuits. Power transformers may be further

classified according to their scope of application, as described

in the following. Transformers that are directly connected to

the generator of a power station are called generator trans-

formers. Their power range goes up to far above 1000 MVA

and their voltage range extends to approximately 1500 kV. The

connection between the different high-voltage system levels is

made via network transformers (network interconnecting trans-

formers), which are mainly autotransformers (i.e., transformers

where the primary and secondary winding of each phase have

a common section). Their power range exceeds 1000 MVA and

their voltage range exceeds 1500 kV.

On the other hand-side, distribution transformers are used in

the distribution networks in order to transmit energy from the

medium voltage network to the low voltage network of the con-

sumers.Inparticular,adistributiontransformerhasbeendefined

by ANSI/IEEE [4] as a transformer for transferring electrical

energy from a primary distribution circuit to a secondary dis-

tribution circuit or consumer’s service circuit. In addition, there

arevariousspecialpurposetransformerssuchasconvertertrans-

formers, test transformers, instrument transformers or telecom-

munications transformers, which can be both in the range of

powertransformersandintherangeofdistributiontransformers

as far as rated power and rated voltage are concerned.

As far as the cooling method is concerned, transformers may

be designed either as liquid-immersed or dry type transformers.

In liquid-immersed transformers, the insulating medium is min-

eral oil or synthetic insulating liquid while in dry type trans-

formers,thecoolingis implementedwithnaturalaircirculation.

Theidentificationofliquid-immersedtransformersaccordingto

the cooling method is expressed by a four-letter code (Fig. 3)

(IEEE C57.12.00 and IEC 60076-2 standards). The first letter

expressesthe internal cooling medium in contact with the wind-

ings (Letter

designates the mineral oil or synthetic insulating

liquid with fire point below 300 C, letter

sulating liquid with fire point above 300 C, and letter

nates the insulating liquid with no measurable fire point). The

second letter identifies the circulation mechanism for internal

cooling medium (Letter

designates the natural convection

flow through cooling equipment and in windings, letter

ignates the forced circulation through cooling equipment (i.e.,

coolant pumps) and natural convection flow in windings (also

called nondirected flow), and letter

culation through cooling equipment, directed from the cooling

equipment into at least the main windings). The third letter ex-

presses the external cooling medium (Letter

air and letter

designates the water). The fourth letter iden-

tifies the circulation mechanism for external cooling medium

(Letter

designates the natural convection and letter

nates the forced circulation [fans (air cooling) or pumps (water

cooling)]). For example, if the internal cooling medium is min-

eral oil, which is circulated with natural flow, and the external

designates the in-

desig-

des-

designates the forced cir-

designates the

desig-

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 3

AMOIRALIS et al.: TRANSFORMER DESIGN AND OPTIMIZATION: A LITERATURE SURVEY2001

Fig. 3. Transformer cooling designations.

cooling medium is air, which is circulated with natural convec-

tion, then this cooling method is coded as ONAN (Oil Natural

Air Natural). In power transformers, various cooling methods

are used including oil circulation by pumps, or forced air cir-

culation by fans, or both of the above. As a result, a number of

different cooling methods exist: oil natural air forced (ONAF),

oil forced air natural (OFAN), oil forced air forced (OFAF), oil

forced water forced (OFWF). Combinations such as ONAN/

ONAF, ONAN/OFAN, or ONAN/OFAF are also applicable [5].

Transformer magnetic circuit is constructed in either a

shell or a core structure. They are distinguished from each

other by the manner in which the primary and secondary coils

are placed around the laminated steel core. The shell-type

transformer is one where the windings are completely sur-

rounded by transformer steel in the plane of the coil. In core

type, the windings surround the laminated iron core. There

are two different technologies for stacking the sheets of the

magnetic material of the core, providing a further distinction

to a) stack-core transformers, where the layers of the sheets

of the magnetic material are placed one over the other and

the vertical and the horizontal layers are over lapped, and b)

wound-core transformers, where the magnetic circuit is of shell

type and the sheets are wound. Multi-winding transformers, as

well as poly-phase transformers, can be made in either shell

or core type designs, with a magnetic circuit that consists of

five (in shell-type transformers) or three legs (in core-type

transformers), respectively.

B. Transformer Survey Structure

Transformer design must take into account numerous per-

formance parameters and technical constraints. The research in

the relevant literature may deal with each one of these param-

eters separately, or concern the overall transformer optimiza-

tion. Fig. 4 presents the main categories of the literature survey,

which define the structure of the survey overview presented in

the next Sections.

III. RESEARCH DEDICATED TO SPECIFIC

TRANSFORMER CHARACTERISTICS

The numerous computational methods and engineering

models proposed for transformer analysis and the accurate pre-

diction of their characteristics can be roughly categorized into

six main groups: 1) numerical techniques (NT) that consist of

some of the most widely used tools for transformer simulation.

Among the proposed techniques of this group, the finite-ele-

ment method (FEM) is the most prevalent one; 2) improved

versions of the transformer equivalent circuit (EC), in order

to include semiempirical descriptions of the core and winding

characteristics that affect the accuracy of calculations. The

Fig. 4. Transformer survey structure.

use of the EC is still common in the manufacturing industry,

due to its simplicity and its ability to provide reliable results,

especially in cases of standardized geometries; 3) analytical

methods (AM), employing analytical formulas for the represen-

tation of the transformer electromagnetic field as well as other

operational characteristics (such as the current distribution),

providing alternative modeling with less computational com-

plexity compared to numerical methods; 4) stochastic methods,

including artificial-intelligence (AI) techniques, such as genetic

algorithms (GAs), which have seen increased usage in the

transformer design area over the last few years; 5) experimental

methods (EM), combining data provided by measurements

with analytical or other methods, in order to provide efficient

models for the accurate representation of certain transformer

characteristics; 6) hybrid methods (HM) (i.e., combinations of

one or more of the methods listed before).

Table I illustrates an overview of the references presented in

Sections III-A–III-I, sorted by the subject and methodological

approach (based on the six categories defined before). Further

details on each reference or group of references are included in

the following paragraphs, providing the necessary background

for their classification. Since several papers were not entirely

dedicated to a single transformer characteristic or methodolog-

ical approach, their classification was based on the main axes of

the proposed methodology or the basic aspects of the addressed

research problems. The observation of Table I not only facili-

tates researchers in the field to categorize previous works but

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 4

2002IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 2009

TABLE I

CLASSIFICATION OF REFERENCES PRESENTED IN SECTION III

BY SUBJECT AND METHODOLOGICAL APPROACH

also reveals fieldsthat havenot yetbeen covered,proposing fur-

ther research areas.

A. No-Load Losses

No-load losses are the continuous losses of a transformer, re-

gardless of load, namely, they exist whenever the unit is en-

ergized [6]. No-load losses are also called iron or core losses

because they are mainly a function of the core materials. The

two main components of no-load losses are eddy currents and

hysteresis. Hysteresis describes the memory of a magnetic ma-

terial. More force is necessary to demagnetize magnetic mate-

rial than it takes to magnetize it; the magnetic domains in the

material resist realignment. Eddy current losses are small cir-

culating currents in the core material. The steel core is a con-

ductor that carries an alternating magnetic field, which induces

circulating currents in the core. These currents through the re-

sistive conductor generate heat and losses. Cores are typically

made from cold-rolled, grain-oriented silicon steel laminations.

A third component of core loss is also present, that cannot be

directly attributed to eddy current or hysteresis phenomena, and

is often called stray, extra or anomalous loss.

FEM has been extensively employed in the no-load losses

prediction problem. The application of two-dimensional (2-D)

FEM in the calculation of transformer core losses is realized in

[7] and [8], where tools for the calculation of stray and eddy

losses are developed, while in [9]–[11], hysteresis models are

developed. Detailed modeling of the core magnetic properties

based on measurements is presented in [12]–[16], while in [17],

eddycurrentlossintransformersoperatedbypulsewidth-modu-

lated inverters is presented. Magnetic anisotropy models for use

in 2-D numerical analysis are developed in [18]–[20]. Igarashi

et al. propose a 2-D FEM method which reduces the number

of unknowns in the finite-element analysis of steel laminations

in [21]. A numerical method for the calculation of the power

losses of nonlinear laminated ferromagnetic cores is developed

in [22], performing a two-step analysis (the first step considers

the anisotropic conductivity of the material, while the second

one introduces its inhomogeneous permeability) and employs

FEM models as a tool for the correction and refinement of re-

sults provided by the first step. Anisotropic magnetic material

properties in conjunction with 2-D FEM are used for the anal-

ysis of transformer magnetic material properties frequency de-

pendence [23]. 2-D FEM modeling combined with 2-D and 3-D

calculations is carried out in [24] in order to derive a dedicated

model of dynamic hysteresis and extra losses in transformer

soft magnetic materials. Excessive core losses and temperature

rises due to the half-turn effect (the phenomenon where the

winding leads are taken out from the different sides of the core

leadingtoanadditionalhalf-turninoneofthecorewindowsina

single-phase transformer) are studied through 2-D FEM in [25].

Virtualairgapscreatedinatransformercorebyachangeincore

permeabilityinthevicinityofauxiliarywindingscarryingdirect

current (dc) are also investigated by 2-D FEM in [26]. In order

to achieve more detailed analysis and computation of the trans-

former magnetic field, the 3-D analysis becomes necessary, as

proposed in [27]–[31]. Moreover, Koppikar et al. [32] describe

details of statistical analysis used in conjunction with 2-D FEM

for quantifying the effect of various factors affecting flitch plate

loss along with the results of 3-D FEM simulations carried on

slotted and laminated flitch plates. To reduce the computational

complexity of 3-D FEM, a practical modeling method of core

lamination modeling is investigated in [33]. Nakata et al. inves-

tigate the influence of transformer core step lap joints on its loss

in [34]. Furthermore, a rigorous analytical study using the finite

difference method for magnetic-field calculation is performed

by several authors in the literature in order to understand the

role of joints in determining the performance of cores of power

and distribution transformers [35] and to accurately calculate

the 3-D spatial distribution, components, and total core losses

in power transformer stacked cores [36], [37]. A dynamic core

lossmodeltoestimatecore lossinsoftferromagneticandpower

ferrite materials with arbitrary flux waveforms, for application

in 2-D and 3-D transient finite-element analysis is proposed in

[38].

In addition, various studies [39]–[46] explore the local flux

distribution in transformer cores as a function of joint design

and its relevance for power loss and noise. It is worth noting

that although transformer joint air gaps have been well studied

usingFEM,theyareseldomtakenintoaccountincircuitmodels

[47].

The application of AI in loss evaluation is addressed in [48]

and [49], where the no-load losses as a function of core de-

sign parameters are predicted by means of artificial neural net-

works (ANNs). Georgilakis et al. [50], [51] also used ANNs

to reduce the iron losses of assembled transformers while opti-

mizingtheproductionprocessofindividualcoresusingTaguchi

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 5

AMOIRALIS et al.: TRANSFORMER DESIGN AND OPTIMIZATION: A LITERATURE SURVEY 2003

methods. In addition, a combination of three AI methods is pre-

sented in [52] and [53], namely, ANNs, decision trees (DTs),

and GAs,in ordertominimize ironlosses during manufacturing

of wound-core distribution transformers. In [54], the differen-

tial evolution method is applied to determine the magnetically

nonlinearcharacteristicsoftransformers.Hybridmethods,com-

bining AI techniques with numerical methods, have been em-

ployed for the calculation and minimization of core losses, as in

[55], where a 3-D permeability tensor FEM is combined with

simulated annealing in order to define appropriate design vari-

ables of wound cores constructed by a combination of standard

and high magnetization grade steel.

Accurate calculation of losses is also provided by the devel-

opment of improved equivalent circuits, as proposed in [56],

where a circuitthat isable to predictlosses underdifferent oper-

ating conditions is presented. Moreover, a final EC of sufficient

accuracy, combining hysteresis and eddy current losses contri-

butions, is proposed in [57]. In addition, Adly [58] presented

an analytical investigation of the various transformer losses

resulting from semirotating flux excitation. The main feature

of this analysis is that core magnetic properties are accurately

represented and simulated using recently developed vector

Preisach-type models of hysteresis. Reference [59] introduces

an accurate Preisach-type model of the hysteretic inductor to

represent a single-phase transformer for the investigation of the

ferroresonance phenomenon, while [60] introduces an accurate

transformer core model, using the Preisach theory, to represent

the core magnetization characteristic in order to simulate fer-

roresonance in voltage transformers. An improved magnetic

anisotropy model, by use of tensor reluctivity, to accurately

express the phase difference between the magnetic flux density

and the magnetic-field intensity is developed in [61]. Guerra

and Mota present a nonlinear electric circuit to describe the

behavior of magnetic cores in low-frequency applications in

[62].

Experimentalmethods,combiningdataprovidedbymeasure-

ments with analytical or other methods, in order to provide ef-

ficient models for the accurate representation of hysteresis and

power losses in the laminations of power transformers, are pro-

posed in [63] and [64], based on data supplied from the steel

manufacturer. Moses [65] reported results in which localized

flux density and losses have been measured experimentally in

model cores and compared with data obtained from an FEM

analysis of the same core geometry. In addition, Girgis et al.

[66] carried out an analytical study in an attempt to determine

the magnitude of the effects of a number of core production at-

tributes. Albach et al. [67] present a practical method for pre-

dicting the core losses in magnetic components for an arbi-

trary shape of the magnetizing current. Furthermore, Dolinar

[68] determined a magnetically nonlinear iron core model of

a three-phase three-limb transformer and compared it with the

classicalsaturatedironcoremodel[69].In[70],anexperimental

investigation of the factors that influence the harmonic content

of magnetizing current (namely the flux density, the degree of

saturation and the core stacking technique) is performed.

Many researchers have studied rotational iron losses over the

years and a number of techniques for making measurements

have been described in the literature. Stranges and Findlay [71]

described an apparatus capable of determining iron losses due

to rotational flux. In addition, Findlay et al. [72] and Davies and

Moses [73] have carried out experiments on various samples in

order to test the hypothesis that different stacking patterns of

grain oriented silicon steel laminations, cut at small angles to

the rolling direction, can reduce the iron core losses in power

transformers. Fiorillo et al. have experimentally investigated

the magnetic variables of grain oriented alloys, in order to de-

rive models of magnetization curve, hysteresis loops, and losses

in any direction [74]. In addition, Marketos and Meydan [75]

introduced a novel method of fabricating consolidated stacks

of electrical steel into single-phase transformer cores, which

can considerably speed up the time required to build the cores

by attempting to further reduce the amount of flux that deviates

from the rolling direction at the corners of transformer cores.

Experimental study of harmonic flux effects in transformer

ferromagnetic materials is realized in [76], while a method

for predicting the core losses under the sinusoidal conditions

with an adequate accuracy from the test results obtained with

the real nonsinusoidal voltage waveform is proposed in [77].

Anisotropic behavior of transformer core loss material is ex-

perimentally studied and interpreted in [78]. Finally, the silicon

steel complex permeability at high frequencies is experimen-

tally determined in [79].

B. Load Losses

Load losses result from load currents flowing through the

transformer [6]. Load losses are also called copper or wire or

winding losses. The two components of the load losses are the

Joule losses (deriving from the product

stands for the winding current and symbol R represents the

winding resistance) and the stray losses.

on the measured dc resistance, the bulk of which is due to the

winding conductors, and the current at a given load. The stray

losses are a term given to the accumulation of the additional

losses experienced by the transformer, which includes winding

eddy losses and losses due to theeffects of leakage flux entering

internal metallic structures. Auxiliary losses refer to the power

required to run auxiliary cooling equipment, such as fans and

pumps, and are not typically included in the total losses.

A review of about 50 papers were conducted by Kulkarni and

Khaparde [80], which have dealt with one or more components

of stray loss from the point of view of estimation and reduction.

In this case, stray losses include eddy and circulating current

loss in windings, losses in flitch plate, core edge loss, loss due

to high current field, and frame and tank losses.

Moreover, a short methodological survey is carried out by

Krawczyk and Turowski [81], showing the need of eddy cur-

rent analysis in electric devices. Robert provides a theoretical

discussion about the layer copper factor used in winding loss

calculation in [82], focusing on a review of the relevant tech-

nical literature.

The FEM analysis is quite commonly used for the eddy loss

calculations[83],[84].PernandYeh[85]areengagedinthefor-

mulation of a finite-element method based on vector magnetic

potential formulation to simulate the electromagnetic field and

current distribution in the windings of power transformers with

non-negligible circulating current. In addition, the foil-winding

eddy loss is estimated by Ram [86], and his model is afterwards

exploited for the study of the variation of transformer sheet

winding eddy current loss with frequency [87]. Furthermore,

the eddy current field due to both windings and heavy current

, where symbol I

losses are based

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 6

2004IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 2009

leads in large power transformers is analyzed in [88] by using

an improved FEM method based on scalar magnetic potential

formulation. This formulation is also employed in [89] for the

calculation of three-dimensional distributions of eddy-current

densities and loss densities in metals near heavy current leads

in a large transformer. On the other hand, vector magnetic po-

tential is used to determine the magnetic field distribution in

transformer windings considering inhomogeneous current dis-

tribution in [90] and time-periodic nonlinear magnetic fields

with eddy currents in [91]. A numerical analysis of losses gen-

erated in the tank-wall surrounding the high-current bushings

of pad-mounted transformers using a 3-D FEM is conducted in

[92]. A special typeof finite element is developed in [93] for the

calculation of eddy current losses in transformer tank shields.

3-D eddy current calculation is also conducted by means of in-

tegral equation method in [94], which, as opposed to 3-D FEM,

provides less computational complexity. Analytical representa-

tions of the electromagnetic field are also used for the predic-

tion of eddy current losses in transformer tank covers in [95],

an investigation that is expanded in [96] by 1) an analytical for-

mulation;2)athree-dimensionalfinite-elementmethod;3)from

measuredvaluesoftheinitialtemperaturerise;and4)frommea-

sured values of the steady-state temperature rise. Eddy losses

due to high current leads in transformers are calculated by an-

alytical methods as well as 2-D and 3-D FEM in [97]. In [98],

an accurate 3-D formulation to study the boundary eddy current

field arising from the heavy winding and terminal lead currents

in a compact power transformer is developed. The 3-D reluc-

tancenetworkmethodis alsoproposedfortransformer loadloss

prediction in [99] as a fast low cost tool for obtaining power and

hottest-spot output data. The same method is employed for the

assessment of the effectiveness of a laminated flux collector in

controlling thepowerlosses ofthetransformer in[100].Finally,

FEM has been employed for the investigation of proximity ef-

fects on conductor losses in [101].

Experimental study of load losses is also encountered in

the relevant literature. When a three-phase three-limb core

type transformer is subjected to the load-loss measurement

test, it is found that the losses and currents of the phases are

asymmetrical (and unequal). Reference [102] attempts to give

a comprehensive explanation for the asymmetry in the loss

values of the three phases during the load-loss test. Moreover,

eddy current losses are experimentally determined in order

to analyze the derating of single-phase transformers under

nonlinear loads in [103], [104].

C. Leakage Field and Short-Circuit Impedance

The calculation of transformer leakage flux is a prerequisite

to the calculation of reactance, short-circuit impedance, short-

circuit forces, and eddy current losses. A review of the most

common winding arrangements in dual voltage transformers

(i.e., in transformers with primary or/and secondary windings

that can be reconnected in order to produce different operating

voltages)andtheirimpactonshort-circuitstrengthiscarriedout

in [105]. A further analysis on transformers with tap changers,

resulting to variable volts per turn, is carried out in [106], fo-

cusing not only to the leakage field but on losses, noise and

weight as well.

The finite-element method has been extensively employed

in transformer leakage field evaluation. The first research at-

tempts, presented over three decades ago [107], [108], focused

on 2-D modeling, due to the restricted performance abilities

provided by the early development of personal computers. 2-D

FEM remains an efficient tool for leakage field and short-circuit

impedance evaluation and more recent developments in the

prediction of these parameters are presented in the literature

[109]–[111]. Although the 2-D modeling is convenient and

useful in some design problems, it can be found insufficient for

detailed analysis and computation of the transformer magnetic

field, and therefore the 3-D solution, initially proposed by

Demerdash et al. at the beginning of 1980 [112], becomes

necessary [113]–[117]. In this context, formulas for the ex-

ploitation of 3-D FEM model results in winding flux linkageare

proposed in [118]. The boundary-element method is another

numerical technique extensively used for electromagnetic prob-

lems [119]. The main attraction of this method is the simplicity

of the data required to solve these problems, along with the

high accuracy obtained with boundary elements. Moreover, the

combination of boundary and finite elements is another widely

used numerical field analysis technique presenting significant

advantages in transformer leakage field modeling [120]–[123].

To overcome the main numerical techniques drawback,

namely the complexity of the required mesh size, especially

in 3-D configurations, alternative leakage field evaluation

models have been proposed, with the use of a 3-D reluctance

network method [124], [125], falling into the category of EC

representation. An alternative method of transformer leakage

field calculation is based on simplified analytical formulas,

as in [126], where the calculation of self and mutual imped-

ances between sections of transformer windings is performed

or [127], where analytical calculations are carried out using

Maxwell’s differential equations and introducing the vector

potential, for the components of the magnetic induction in

two-dimensional field space. Analytical methods are often

employed by transformer manufacturers in order to simplify the

time and complexity of the calculations required in automated

design process. Tomczuk and Zakrewski and Tomczuk propose

the integral equation method for the calculation of magnetic

leakage fields in [128] and [129], respectively.

Stochastic methods are also employed for solving problems

of this category, as by Thilagar and Rao [130], who suggested

an exact EC model for the estimation of all impedance param-

eters of three-winding transformers, with the use of GAs. The

suggested method also estimates geometrically a complex pa-

rameter, that is, mutual leakage between secondary and tertiary

windings.

In terms of leakage reactance experimental investigation, the

influence of the test circuit (involving low voltage and single-

phase excitation) on the respective measurements is presented

in [131]. The effects of aluminum magnetic shielding and mild

steel magnetic shunts on the leakage flux in the steel tank of a

single-phase transformer are experimentally examined in [132].

D. Inrush Current

Transformer inrush currents are high-magnitude, harmonic-

rich currents generated when transformer cores are driven into

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 7

AMOIRALIS et al.: TRANSFORMER DESIGN AND OPTIMIZATION: A LITERATURE SURVEY2005

saturation during energization. These currents have undesirable

effects, including potential damage or loss-of-life to the trans-

former and reduced power quality on the system [133]. More-

over, a significant impact on transformer winding mechanical

stress may be observed [134]. It is also shown that under special

conditions,thecurrentsobservedaftertransformerswitchingon

do not contain enoughrestraining information (e.g., second har-

monic), resulting in protective relay misoperation and posing

a great problem for protective relaying of power transformers

[135]. Inrush current prediction is therefore another important

issue during transformer design and various approaches to deal

with it are present in the technical literature.

Numerical techniques are present in the above context of

inrush current prediction, as in [134] and [136], where 2-D

and 3-D FEM is applied to three-legged power transformers

for the evaluation of forces on the windings due to inrush

current and their comparison to the respective short-circuit

forces. However, the majority of the methods used for inrush

current simulations are based on the derivation of appropriate

equivalent circuits, taking into account the core geometry [137]

and structural characteristics [138], the core material nonlinear

characteristics [139], [140], using real-time measurements

[141], [142], adopting proper parameters for the core magnetic

hysteresis Jiles–Atherton model [143] or by proper simulation

of the voltage sags caused by inrush currents [144]. Other

models take into account the effect of transformer energization

to other parallel connected transformers [145]. Frequency

domain solution techniques have also been proposed for the

simulation of inrush current variation, in order to overcome

numerical problems due to the transient nature of the phenom-

enon [146]. Artificial neural networks have also been employed

for the computation of inrush current and the resultant forces

on the transformer windings [147].

Practical methods of elimination of transformer inrush

current are also proposed in the relevant bibliography as in

[148], where proper control strategies of circuit breakers that

control transformer switching are presented, or in [149]–[151],

by means of sequential phase energization (i.e., by energizing

each phase of the transformer in sequence). Other methods may

be applied during the design stage, by modifying the winding

configuration [152], [153].

Inrushcurrentphenomenamaybeexploitedtoestimatetrans-

former performance as in [154] where they are used as a crite-

rion to assess core saturation characteristics.

E. Dynamic Behavior Under Short Circuits and Seismic Stress

The short-circuit current in a transformer creates enormous

forces on the turns of the windings. The short-circuit currents in

a large transformer are typically 8 to 10 times larger than rated

and in a small transformer 20 to 25 times larger than rated. The

forces on the windings due to the short-circuit current vary as

the square of the current, so whereas the forces at rated current

may be only a few newtons, under short-circuit conditions these

forces can be tens of thousands of newtons. These mechanical

and thermal stresses on the windings must be taken into consid-

eration during the design of the transformer. Transformer be-

havior under short circuits or internal faults is one of the major

concerns during their design, since the ability to overcome the

resulting stresses and currents without significant consequences

on their operation is a requirement often present in the inter-

national technical standards. The accurate representation of this

behaviorismainlyrealizedthroughnumericaltechniquesdueto

their prevalence in the transformer detailed magnetic field cal-

culation: In [155], 2-D and 3-D FEM are compared in terms of

accurate prediction of exerted forces on shell-type transformers

under short circuit. In [156], a method was developed to apply

2-D FEM to calculate the parameters for an EC of the trans-

former with an internal short-circuit fault, while in [157] the

method is expanded, representing in detail winding insulation

deterioration. Winding internal faults are also modeled by 3-D

FEM in [158] and by coupling the finite-element transformer’s

physical model with external electric circuit equations in [159].

3-D FEM based on scalar and vector magnetic potential formu-

lation is used for transient eddy current and short-circuit forces

estimation, in [160] and [161], respectively.

Equivalent circuit approaches are also employed, although

their application is usually coupled to some detailed calcula-

tion,oftenbasedonnumericalmethods[162],inorder toensure

better representation of the phenomena occurring during short

circuits.

Analytical models based on the theory of symmetrical com-

ponents are also used for the simulation of transformer faults

[163],whilethestabilityofcontinuouslytransposedcablewind-

ings under axial short-circuit forces is investigated through an-

alytical formulas in [164] and the vibration analysis of faulted

transformers helical windings is performed in [165]. Analytical

methods are proposed in [166] for the calculation of the power

flow during transformer internal faults.

Seismic stress is another transient phenomenon with partic-

ular interest to transformer engineers, as it can cause severe

damage including anchorage failure ripping the transformer

case and oil leakage, as well as fracture of porcelain bushings.

Therefore, seismic qualification of transformers, focusing on

high voltage bushings, is another design consideration that is

explored in [167], [168] by means of 3-D FEM analysis, and in

[169] by a specially developed experimental setup.

F. Noise

Transformers located near a residential area should have

sound level as low as possible. Techniques for power trans-

former noise control have been proposed over four decades ago

[170]–[173] mainly through experimental study and statistical

analysis of measurements used to determine the principal

factors affecting transformer performance. The sound intensity

method for power transformer noise measurements is described

and results of its application are demonstrated in [174], while

methods based on this technique are also developed in [175]

and [176]. In [177] and [178], appropriate conditions for accu-

rate outdoors and indoors measurement of transformer noise

are derived and the inherent inaccuracies in the measurements

are determined, resulting to the proposition of modifications

to the existing IEC and IEEE industry standards of measuring

transformer noise.

Sincethecoremagneticpropertiesandstructurearethemajor

factors influencing transformer noise, a lot of research focusses

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 8

2006IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 2009

on the analysis and improvement of the core attributes with re-

gard to transformer noise: A simple technique of measuring the

dynamic magnetostriction is used to illustrate the effects of im-

provingthestresssensitivityofsteelandreducingcorevibration

by a suitable bonding technique which can also reduce trans-

former noise in [179]. 3-D FEM structural dynamic analysis

is used for the examination of the influence of core lamination

upontransformernoisein[180],whilevibro-acousticmodelling

is also proposed for further analysis. Finally, the relevance of

the core magnetic properties for the generation of audible noise

in transformer cores is analytically and experimentally investi-

gated in [181].

Methods of noise level reduction are also proposed in the

bibliography, through the addition of equipment in transformer

substations [182], [183]. More recent approaches perform noise

level optimization by means of a reverse calculation method

and Linear Programming using an empirical formula for esti-

mating noise levels at the boundary points around the substa-

tion premises [184]. The design and the manufacture of a trans-

former with low sound level require indepth analysis of noise

sources. Core, windings and cooling equipment are three main

factors of noise, with the first factor the paramount one. To de-

termine a method for the optimum design of the noise-reduc-

tion transformer, noise attenuation of a simple structured proto-

typetransformerthatutilizesC-coresisquantitativelydiscussed

based on the EC analysis [185], [186]. Similar work is pre-

sentedin[187].Arecentlydevelopedcalculationschemeforthe

computer modelling of the load-controlled noise of liquid-in-

sulated three-phase power transformers is presented in [188].

This modelling scheme allows the precise and efficient compu-

tation of the coupled electromagnetic, mechanical and acoustic

fields. The equations are solved by using the FEM as well as the

boundary-element method (BEM).

G. Insulation

The insulation of a transformer is linked to its ability to with-

stand surge phenomena and overvoltages likely to occur during

its operation. For this purpose, the related work may deal with

the analysis of such phenomena, so as to design an adequate

transformer insulation system. Other factors that affect trans-

former insulation life are vibration or mechanical stress, repeti-

tive expansion and contraction, exposure to moisture and other

contaminants, and electrical and mechanical stress due to over-

voltage and short-circuit currents.

Numerical methods are more scarcely applied for the simula-

tion of the aforementioned phenomena, and are mainly used for

the calculation of the transformer electric field [189]–[191]. On

the other hand, the majority of research is based on EC repre-

sentation for transformer analysis under overvoltages, with re-

spect to their geometrical characteristics [192], their dynamic

behavior [193], their frequency-response characteristics [194]

or the characteristics of the network where they are connected

[195]. Other attempts to model the insulation structure of trans-

formers and the quantitative analysis of its dielectric response

are also encountered [196], [197]. Hybrid methods, combining

finite-element simulations for the derivation of EC parameters

have also been proposed [198].

Finally, the exploitation of measurements and the experience

from the impact of the operational characteristics on the insula-

tion life assessment is often one of the main methods to design

aninsulationsystemandpredictitsabilitytowithstandanytran-

sient phenomena during the transformer life [199].

Insulation condition assessment is a widely covered topic,

and various published works deal with this subject. A theoret-

ical discussion on the aging of oil-impregnated paper in power

transformersiscarriedoutin[200],while[201]–[203]and[204]

present experimental methods to determine cellulose and oil

aging. Transformer oil breakdown is experimentally and the-

oretically analyzed in [205]. Effects of moisture and aging on

the oilpaper insulation of transformers are investigated by re-

turn voltage measurements (i.e., the voltage that is built up be-

tweentheelectrodesonadielectricaftertheapplicationofdirect

voltage for a long period of time) in [206]. An analytical model

establishing the time to failure of the insulation of transformers

given their operating history is developed in [207], based on

hourly load and ambient temperature measurements that extract

the operating profile of the equipment and IEEE life consump-

tion models to assess the consumed life of insulation. Partial

discharge measurements are used to determine dielectric char-

acteristics of transformer oils in [208], [209] and [210]. Fre-

quency response of oil-impregnated pressboard and paper are

used for estimating moisture in transformer insulation in [211]

and [212]. Experimental investigation of bushing insulation is

analyzed in [213]. Methods to overcome the insulating mate-

rials degradation with time in service are also proposed, as in

[214], where the use of synthetic minerals for the absorption of

moisture in paper insulation is discussed.

H. Cooling

Transformer cooling is one of the most important parameters

governingatransformer’slifeexpectancy.Thetotaltemperature

is the sum of the ambient and the temperature rise. The temper-

ature rise in a transformer is intrinsic to that of a transformer at

a fixed load. The design of the cooling system is based on the

hottest-spottemperaturevalue,anddifferentmethodsforitspre-

dictionareproposedintheliterature,alongwiththeoveralltem-

perature distribution prediction, according to the transformer

coolingmethod.Furthermore,theimprovedknowledgeoftrans-

former thermal characteristics can allow transformer engineers

to achieve enhanced designs and manufacturers are particularly

interested in its accurate prediction.

The finite difference method is proposed by Pierce [215] for

hottest-spot temperature prediction in dry-type transformers.

2-D FEM thermal calculation is proposed in [216] for the

calculation of core hottest-spot temperature in power and dis-

tribution transformers. Moreover, it is employed to model the

effect of harmonic currents in the winding temperature [217]

and to perform heat transfer analysis and obtain the steady

state and the transient temperature distribution of

cooled-insulated power transformers [218]. Most recent trends

in thermal modeling employ coupled electromagnetic-thermal

finite-element models [219], [220]. In [221] a 3-D FEM model

using a magnetic scalar potential formulation is combined with

a mixed analytical and numerical form of the electrical circuit

equation to take into account the skin and proximity effects in

gas

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 9

AMOIRALIS et al.: TRANSFORMER DESIGN AND OPTIMIZATION: A LITERATURE SURVEY2007

the windings, resulting to current densities that are used as in-

puts to a steady state thermal FEM.In [222] a method that relies

on the combination of analytical calculations, 2-D thermal and

3-D electromagnetic FEM is presented for transformer thermal

modeling. Rosas et al. propose the finite volume method as a

means of predicting the improvement of the cooling process

of liquid-immersed electrical transformers using heat pipes

[223]. For the proper representation of the cooling medium

circulation, which affects the transformer thermal performance,

thermal FEM is often coupled to computational fluid dynamics

(CFD), in 2-D [224], [225] or 3-D models [226], [227]. Cou-

pled fluid flow, heat transfer and electromagnetic numerical

analysis is also encountered in the relevant litterature [228],

further enhancing the accuracy in the prediction of transformer

temperature.

Transformer thermal performance is usually predicted

through analytical formulas that use approximations and con-

stants derived from experimental results. Significant research

effort is focusing on the improvement of these formulas and the

derivation of more efficient calculation methods. In [229] the

authors present analytical methods for estimating the tempera-

ture and its distribution at different points of the transformer,

while in [230] they extend this work, taking into account the

thermal inhomogeniety of the windings. This method is later

employed to perform an elapsed life assessment study by ac-

quiring insulation-aging data under accelerated thermal stresses

in [231]. Calculation methodologies of top oil temperatures and

hottest-spot temperatures are evaluated in [232] by comparison

to respective measured values in power transformers. Two

different thermal analytical models to predict temperature rises

in an OFAF transformer at OFAN cooling mode in service

are presented and experimentally investigated in [233]. In

[234], new thermal loading guides are proposed, based on test

results in factories and measured data in the field that are used

to determine parameters of thermal differential equations. A

mathematical model of the winding hottest-spot response to

step changes in the load current of power transformers is pre-

sented in [235]. Ryder presents an analytical model to predict

winding temperature gradient in power transformers, based

on electrical analogy in [236]. The analytical assessment of

the impact of ambient temperature rise, as a result of climate

change, to distribution transformer loss of life is performed

in [237]. In [238] and [239], four top-oil thermal models that

require only parameters available from heat-run data and need

only measurements utilities routinely monitorare presented and

compared. Finally, a short review of developments in analytical

thermal models is presented in [240].

Differentkindsofequivalentthermalcircuitsareencountered

in the literature, dependent on the geometry of the core and

windings and the type of cooling. Appropriate equivalent cir-

cuits are developed in [241] for the core rise temperature calcu-

lation and power transformer thermal distribution [242]–[244].

In [245] and [246], the parameters of the proposed liquid-im-

mersed transformer thermal model are derived with the use of

GAs. Dynamic thermal models are also proposed in the litera-

ture [247]–[249], taking proper account of oil viscosity changes

and loss variation with temperature, while in [250] a reliability

analysisofvariousdynamicthermalmodelsiscarriedout.Since

transformer cooling is dependent on the hydrodynamic proper-

tiesofoil,hydraulicmodelsformassflowdistributionhavebeen

proposed in the literature, so as to provide detailed representa-

tion of the oil flow and pressure in ONAN transformers, as a

function of the number and configuration of the cooling ducts

[251], [252]. Hydraulic models are combined with heat convec-

tion models in [253] and [254].

The variation of transformer loading directly affects its

thermal performance and must be correlated to the transformer

time constants in order to derive safe conclusions for the re-

sulting thermal loading. Therefore, several works have focused

on developing proper load models, suitable for adoption in

transformer thermal studies. In [255], a probabilistic model is

presented by using load profiles, where variance and covariance

are included. A methodology for specifying the winter and

summer peak-load limits for substation transformers that carry

a temperature-sensitive load, taking into account the random

nature of load and ambient temperature as well as their cor-

relation is presented in [256]. Residential loading profiles are

extracted in [257], through statistical processing of measured

data, proposing a methodology for sizing the transformers to

serve these kinds of loads. In [258] a risk-based probabilistic

method is presented to assess transformer loading capability,

taking into account the probabilistic nature of time-varying

loads and ambient temperature. Finally, a method for the eval-

uation of cyclic loading of power transformers is presented in

[259].

Particle swarm method, neural networks, and neurofuzzy

networks are also encountered in the relevant research field

[260]–[263]. Monte Carlo methods are also used for sensitivity

analysis of transformer hottest-spot and equivalent aging in

[264].

Experimental investigation of thermal distribution of cast-

resin and liquid-filled transformers is carried out in [265] and

[266],respectively.Experimental data are used for theimprove-

ment of analytical equations that predict thermal distribution of

liquid-filled transformers in [267] and [268]. Transformer oil

characteristics before and after modifications of the forced-oil

cooling system are experimentally studied in [269].

I. DC Bias

DC current can flow in alternating-current (ac) power lines if

a dc potential difference exists between the various grounding

points. Such a difference can be caused by a geomagnetic storm

[geomagnetically induced current (GIC)] or the injection of dc

current by one of the ground electrodes of a dc link [270]. DC

flowing through the earthed neutrals of transformer windings

causes a dc component in the magnetising current. Owing to

nonlinearity, the waveform of this current is strongly distorted.

The prediction and impact of this phenomenon has been studied

with finite-element method [271]–[275] and equivalent mag-

netic circuits [276]–[280]. The experimental study of the phe-

nomenon has also been performed, by field tests in a power and

distribution transformer in [281] and [282], respectively, and

by reactive power measurements in various distribution trans-

former ratings in [283].

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 10

2008IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 2009

TABLE II

CLASSIFICATION OF REFERENCES PRESENTED IN SECTION IV

BY PROBLEM TYPE AND PROBLEM NAME

IV. TRANSFORMER DESIGN OPTIMIZATION

Transformer manufacturers use cost optimization techniques

during the design phase to minimize material costs and sat-

isfy the utility’s loss evaluation requirement. The difficulty in

achieving the optimum balance between the transformer cost

and performance is a complicated task, and the techniques that

areemployedforitssolutionmustbeabletodealwiththedesign

considerationsofSectionIII,soastoprovideadesignoptimum,

while remaining cost-effective and flexible. The research asso-

ciated with design optimization is therefore more restricted in-

volving different mathematical optimization methods.

Whatever the chosen optimization method is, there is also the

question of how much detail to include in the problem descrip-

tion. Despite the fact that the goal is to find the lowest cost, one

might wish that the solution should provide sufficient informa-

tion so that an actual design could be produced with little ad-

ditional work. However, it would be unrealistic to expect that

the optimum cost design for a transformer would automatically

satisfy all of the mechanical, thermal, and electrical constraints

that require sophisticated design algorithms to evaluate. Based

onthese,ourmaingoalistopresentthetransformerdesignopti-

mization techniquesthat deal with theminimization of theman-

ufacturing as well as operating cost. These techniques are sum-

marized in Table II according to the category of the addressed

optimization problem and they are discussed in the forthcoming

subsections.

A. Manufacturing Cost Minimization

In optimum design of transformers, the main target is to min-

imize the manufacturing cost. Therefore, the objective function

is a cost function with many terms, including material costs,

labor costs, and overhead costs. These component costs, as well

astheconstraintfunctions,mustbeexpressedintermsofabasic

set of design variables.

In order to compete successfully in a global economy,

transformer manufacturers need design software capable of

producing manufacturable and optimal designs in a very short

time. Traditionally, the transformer design problem has been

surrounded by much transformer designer art. The first trans-

former design was made on computer in 1955 [284]. Later

on, more research in transformer design using computers

was pioneered by [285]–[290]. Several design procedures for

low-frequency and high-frequency transformers have appeared

in the literature after the 70’s. Judd and Kressler [291] pre-

sented a technique for designing transformers with given size

and type of structure to have maximum volt-ampere (VA)

output while at the same time insuring the satisfaction of a

number of design constraints. The resulting design technique

eliminates overdesign problems in that the smallest physical

size structure will result consistent with the design objectives.

An improved solution of the described problem was presented

by Hurley et al. [292]. Poloujadoff et al. [293] show the varia-

tion in the price of the transformer depending on the primary

turns, which is an approximately hyperbolic function. Also cost

curves of the transformer against the magnetic flux density and

against the current density are presented. Jeweel [294] does

a functional proposal with students in electrical engineering,

in which the student designs, builds and tests a 10-VA trans-

former. Grady et al. [295] deal with the teaching of design of

dry type transformers, based on a computer program, where the

user optimizes its design based on trial and error. Furthermore,

Rubaai[296] describesa computerprogramyielding an optimal

design of a distribution transformer based on user input data

(classified in given, independent and dependent). The author

includes design transformer formulas used by the program.

Andersen [297] presented an optimizing routine,Monica, based

on Monte Carlo simulation. Basically, his routine uses random

numbers to generate feasible designs from which the lowest

cost design is chosen. Hernandez and Arjona [298] develop

an object-oriented knowledge-based distribution transformer

design system, in conjunction with FEM, which is used as a

tool for design performance validation.

Deterministic methods provide robust solutions to the

transformer design optimization problem. In this context,

the deterministic method of geometric programming has

been proposed in [299] in order to deal with the design opti-

mization problem of both low frequency and high frequency

transformers. Furthermore, the complex optimum overall trans-

former design problem, which is formulated as a mixed-integer

nonlinear programming problem, by introducing an integrated

design optimization methodology based on evolutionary al-

gorithms and numerical electromagnetic and thermal field

computations, is addressed in [3], [300]. However, the overall

manufacturing cost minimization is scarcely addressed in the

technical literature, and the main approaches deal with the

cost minimization of specific components such as the magnetic

material [301], the no-load loss minimization [302], [303]

or the load loss minimization [304]. Techniques that include

mathematical models employing analytical formulas, based on

design constants and approximations for the calculation of the

transformer parameters are often the base of the design process

adopted by transformer manufacturers [305].

Apart from deterministic methods, Artificial Intelligence

techniques have been extensively used in order to cope with

the complex problem of transformer design optimization, such

as GAs that have been used for transformer cost minimization

[306], performance optimization of cast-resin distribution

transformers with stack-core technology [307] or toroidal core

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 11

AMOIRALIS et al.: TRANSFORMER DESIGN AND OPTIMIZATION: A LITERATURE SURVEY 2009

transformers [308]. Neural network techniques are also em-

ployed as a means of design optimization as in [3], [309] and

[310], where they are used for winding material selection and

prediction of transformer losses and reactance, respectively.

Another aspect of transformer design optimization consists

in providing design solutions in order to maintain certain as-

pects of transformer performance within the limits imposed by

the technical specifications. In this context, the maintenance of

short-circuit impedance and losses within the acceptable toler-

ance is often addressed, as in [121], [311], [312], and [313],

[314], respectively.

In a nutshell, it is clear that the transformer design optimiza-

tion problem remains an active research area.

B. Operating Cost Minimization

Apartfromthetransformermanufacturingcost,anothercrite-

rionusedfortransformerevaluationandoptimizationisthetotal

owning cost (TOC) taking into account the cost of purchase as

wellasthecostofenergylossesthroughoutthetransformerlife-

time [315].

TheTOCtechniqueis themostwidelyusedtransformereval-

uation method for determining the cost-effectiveness of energy-

efficient transformers, providing a balance between cost of pur-

chaseandcostofenergylosses.TheTOCevaluationmethodhas

been developed as a handy tool to reflect the unique financial

environment faced by each electric utility when purchasing dis-

tribution transformers. According to this method, the variability

of the cost of electric energy, capacity and financing costs is ex-

pressed through two evaluation factors, called

corresponding to the unit cost of no-load and load losses, re-

spectively. It is important to note that the method that defines

these two factors varies according to the role of the transformer

purchaser in the energy market (two major categories can be

considered: electric utilities and industrial users) and the depth

of the analysis (depending on the accuracy of the representation

ofthetransformerloadingcharacteristics).Itisimportanttorec-

ognizethattheperspectiveoftheelectricutilityisdifferentfrom

the perspective of the industrial and commercial users of trans-

formers. The transformer loss evaluation procedurefor the elec-

tric utility involves understanding and assessing the total cost

of generation, transmission, and distribution transformer losses,

whilethetransformerlossevaluationprocedureforanindustrial

and commercial user requires an understanding and assessment

of the electric rates they pay to the electric utility.

An important part of the transformer cost optimization re-

searchisdevotedtotheTOCminimization,asfollows.Distribu-

tion transformer TOC optimization is analyzed in [316]–[320].

Since the load losses are directly linked to the type of the con-

sidered load and the specific details of the network at the trans-

formerinstallationpoint,anumberofversatilefactorsshouldbe

incorporatedintheTOCanalysis.Suchananalysisisperformed

in depth in [321], [322].

Furthermore, energy losses of transformers throughout their

life cycle increase significantly their operational costs, resulting

in TOC values much higher than their purchase price. For the

above reason, the decision for what transformer to purchase

shouldnot be basedonly onitspurchase price.In general, trans-

formerswiththelowestpurchasepricearealsotheoneswiththe

andfactors,

highestTOC.Therefore,inordertochoosethemosteconomical

transformer in the long term, the TOC value during the lifespan

ofthetransformershouldbetakenintoaccount[323].Moreover,

the external environmental costs [3], [324] should be taken into

consideration as well (i.e., the costs that are associated with var-

ious types of emissions resulting from the combustion of fossil

fuels) so as to compensate for transformer losses.

Recently, the impact of transformer environmental external-

ities and the contribution of losses to the greenhouse gas emis-

sionsgeneratedbytheglobalpowergenerationmixhasbeenad-

dressed [325]–[327]. Furthermore, ways to promote the policy

to encourage the use of efficient transformers in the Spanish

marketareproposedbyFrauetal.[328],whereincentivestopri-

vateusersandelectricutilitiesareintroduced,changingSpanish

lossesregulation,andallowingutilitiestoparticipateinthe

emissions market. Moreover, an overview of options available

to distribution transformer specifiers, taking advantage of the

efficiency and environmental benefits, has been provided [329].

However, a methodology to quantify the impact of environ-

mental externalities on transformer TOC has not yet been de-

veloped.

V. POST-DESIGN TRANSFORMER PERFORMANCE

The main incentive of the research presented in the previous

sections was to develop models for transformer simulation

and adopt methodologies that were able to optimize their

performance according to their constructional characteristics,

providing several criteria for transformer design optimization.

The works presented in this section do not focus on the pre-

diction and evaluation of transformer characteristics during

the design stage, but the derivation of proper models for given

performance characteristics, to be included in power system

studies or other engineering studies, where transformers are

involved. The relevant research can be divided into two major

categories: models suitable for harmonic load flow studies and

models for electromagnetic transient studies. Table III lists

the relevant publications by category, further classifying them

according to their methodological approach. The acronyms NT,

AM, EM, and HM of Table III correspond to the categories

defined in Section III (and included in Table I). It must be

noted that the majority of the articles concern the derivation of

transformer equivalent circuits; therefore, they are all included

in the category EC (equivalent circuits) defined in Section III,

a classification that is not explicitly denoted in Table III.

A. Harmonic Modeling

The research interest on harmonic load flow studies is con-

tinuously growing, due to the increase of nonlinear devices in

power systems. Since transformers are key elements in these

systems, their modeling is an important subject in harmonic

load-flow studies and a number of different approaches have

been proposed in the literature. Stensland et al. develop a trans-

former model where the iron and copper losses under low fre-

quency voltage harmonics may be determined either analyti-

callyorbyFEM,suitableforpowersystemstudies[330].Single

and three-phase equivalent circuits taking into account the non-

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 12

2010IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 2009

TABLE III

CLASSIFICATION OF REFERENCES PRESENTED IN SECTION V

BY SUBJECT AND METHODOLOGICAL APPROACH

linearityofthecorearepresentedin[331].Mohammedetal.im-

plement a transformer transient FEM coupled to external elec-

tric circuits and the wavelet packet transfer algorithm for the

analysis of harmonic behavior of the transformer currents and

the dc load current [332]. In [333], a complete analytical model

isdevelopedtocalculatethetimedomainwaveformandthehar-

monic components of the transformer excitation current. Ma-

soumet al. developa nonlineartransformer modeling technique

for steady-state operation under unbalanced, asymmetric, and

nonsinusoidal operation, capable of computing derating factors

in [334].

B. Transient and Dynamic Modeling

One of the weakest components of modern transient simula-

tionsoftwareisthetransformermodel.Manyopportunitiesexist

to improve the simulation of its complicated behaviors, which

include magnetic saturation of the core, frequency-dependency,

capacitivecoupling,andtopologicalcorrectnessofcoreandcoil

structure [335]. Martinez and Mork present a review of the ex-

isting models in [335], providing an overview of their main de-

velopments, while Martinez et al. provide guidelines for the es-

timation of transformer model parameters for low- and mid-fre-

quency transient simulations in [336].

Inthistypeofmodeling,classicalmethodstodeterminetrans-

formermagneticcircuitandwindingssuchastheonesdescribed

in Section III are combined with frequency and time domain

modeling techniques, as presented in the following.

Woivre et al. propose a model for shell-type transformer

overvoltage study, where the main R, L, and C parameters

are calculated with analytical and numerical methods, while

transient overvoltages of the transformer are calculated from

frequency response by Fourier transform [337]. Reduction

techniques of linear and nonlinear lumped parameters models

are proposed by Gutierez et al. [338] and Degeneff et al.

[339], respectively. De Leon and Semlyen introduce a simple

dynamic hysteresis loss model in [340] and the calculation of

transformer EC elementary parameters (namely leakage induc-

tance and capacitance) on a turn-to-turn basis in [341] which

are used to derive a winding model suitable for transformer

transients in [342]. Moreover, they investigate time domain

modeling of eddy currents for electromagnetic transients study

in [343] and propose techniques for time step reduction in

electromagnetic transient simulation in [344], while in [345], a

complete three-phase transformer model is proposed based on

their previous works in [341]–[343]. This model is improved in

[346] to include the capacitive effects between turns or sections

of a winding. An alternate model for low-frequency electro-

magnetic transients is proposed in [347] and its more simplified

versionis presentedin[348]. Papadiasetal. present three-phase

transformer models for the study of switching fast electromag-

netic transients in [349]. Distributed equivalent magnetic and

electric circuits are introduced in transformer transient analysis

in [350]. Tokic et al. develop numerical methods to solve the

system of differential equations in state space, describing the

transformertransientbehaviorin[351],whileTokicandUglesic

develop an original method of modeling nonlinear elements,

for the elimination of overshooting effects and suppression of

numerical oscillations in transformer transient calculations in

[352]. Frequency-response analysis is used in [353] to study the

transient recovery voltage associated with power transformer

terminal faults. Abeywickrama et al. present a 3-D model of

electromagnetic (EM)-field distribution in a power transformer

at high frequencies for use in frequency-response analysis in

[354] and its results are exploited in [355] for high-frequency

modeling of power transformers. In [356], the principles of

modal analysis are presented, while in [357], modal analysis

is used to consider frequency-dependent effects of internal

capacitance, inductance, and resistance of windings in order to

analyze the transient characteristics of a transformer. In [358],

a model is presented for a multiwinding multiphase transformer

developed by the nodal inverse inductance matrix, which can

be used for transient and steady-state analysis in complicated

winding arrangements and network configurations, while in

[359] and [360] a multiterminal transformer model is developed

for balanced and unbalanced load, respectively. A three-phase

transformer dynamic model, providing a good compromise

between accuracy and excessive complexity arising in dynamic

simulations is presented in [361], while in [362] frequency-de-

pendent time-varying resistance of the transformer winding

is considered during modeling the response to lightning im-

pulse wave. A model reference approach for classification

of faults that can occur during impulse tests on power trans-

formers is proposed in [363]. Stuehm et al. and Mork develop

five-legged wound-core transformer models in [364], [365],

while Mork et al. propose a hybrid transformer model based

on four typically available sources of information: factory

test reports, design data, basic ratings and direct laboratory

measurements [366], [367]. In [368], Mork et al. detail the

parameter estimation methods developed for the five-legged

core of the aforementioned hybrid model. Very fast transient

voltage analysis is performed in [369]–[373]. Mombello and

Moller present a model with accurate representation of winding

losses, developed for the determination of maximal stresses

during resonance phenomena within transformers [374], while

Mombello performs a deep analysis of the behavior of trans-

former winding impedances for high frequencies by analyzing

the propertiesof inductance matrices in [375]. In [376], a model

that reproduces not only the impedance characteristics seen

from each terminal of a core-type distribution transformer but

also the surge-transfer characteristics between the primary and

secondary sides in a wide range of frequencies is presented.

Thecouplingofnumericalmethodswithothertransientmod-

eling techniques is proposed by other researchers in the field.

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 13

AMOIRALIS et al.: TRANSFORMER DESIGN AND OPTIMIZATION: A LITERATURE SURVEY 2011

Fig. 5. Percentage participation of IEEE, CENELEC, and IEC standards in the

overall amount of standards survey.

An EC based on winding resonances computed by FEM is pre-

sented in [377] for the simulation of winding electromagnetic

oscillations. Mohammed et al. use a circuit coupled FEM anal-

ysisundersinusoidalwaveformstoderiveinductancesasafunc-

tion of the amplitude of ac flux as well as its phase angle during

acompleteaccycleforthree-phasetransformermodeling[378],

a work which is later expanded to single-phase transformers

[379].

Experimental methods to define models for electromagnetic

transientstudiesarealsoemployed, asin[380],whichdescribes

the measurement setup for the extraction of a frequency depen-

dent model of a two winding transformer. A three-phase trans-

former model including saturation and based on experimental

parameters is developed in [381], for the examination of sym-

metrical and unsymmetrical voltage sag effects on three-legged

transformers in [382] and [383], respectively.

VI. TRANSFORMER STANDARDS

A number of transformer relevant standards are listed in this

section. These standards are issued by the three institutions:

ANSI/IEEE, CENELEC, and IEC (Fig. 5).

IEEE stands for The Institute of Electrical and Electronics

Engineers, Inc. IEEE standards are national standards prepared

and issued in the United States of America by the IEEE So-

cieties. The use of these standards is wholly voluntary, and it

becomes mandatory only when specified in a contractual rela-

tionship or when required by a duly constituted legal authority.

The IEEE clearly indicates that the existence of an IEEE Stan-

dard does not imply that there are no other ways to deal with

matters related to the scope of the IEEE standard.

CENELEC stands from Comité Européen de Normalisa-

tion Electrotechnique (in English: European Committee for

Electrotechnical Standardization). It was created in 1973. CEN-

ELEC standards (EN standards) are international standards

prepared by working groups and approved by weighted voting

among countries being members of the European Union (EU)

and other countries included in the European Economic Area

Agreement. Furthermore, CENELEC has issued a number of

Harmonization Documents (HD), and their formal status is

practically the same as that of the EN standards. CENELEC

has decided to phase out the Harmonization Documents and

replace them by EN standards.

IEC stands for International Electrotechnical Commission,

whichwasofficiallyfoundedin1906inLondon.TheuseofIEC

standardsisavoluntarymatter.Technicalcommitteenumber14

(TC14) deals with power transformers. The IEC standards do

not prescribe how to design and produce transformers. There-

fore, it would be meaningless or it would at least be impre-

cise use of language to say that a transformer shall be or is de-

signedandproducedaccordingtoIECstandards.TheIECtrans-

former standards establish a series of performance, safety, ap-

plication, selection and other requirements to be satisfied by the

equipment, including performance tests for their certification.

In this context, they define certain tests the transformers shall

be subjected to before delivery from the factory and state the

acceptance criteria. The purpose of the tests is that transformers

that have passed these tests shall have good prospects of a long

life and high service reliability, when adequately protected and

maintained.

It is important to note that between the standards that are is-

sued bythese threeinstitutions, there are some basic differences

which are not in the scope of this work to emphasize. How-

ever, IEC and IEEE have expressed the intention gradually to

decrease or remove these basic differences between their stan-

dards. It is envisaged that a closer co-operation between these

organizations will make the future standardization work more

cost efficient.

A. ANSI/IEEE Standards

IEEE Std. C57.12.00-2006—IEEE Standard for Standard

General Requirements for Liquid-Immersed Distribution,

Power, and Regulating Transformers;

IEEE Std. C57.12.01-2005—IEEE Standard General Re-

quirements for Dry-Type Distribution and Power Trans-

formers,IncludingThosewithSolid-Castand/orResinEn-

capsulated Windings, Revision of C57.12.01-1998;

IEEE Std. C57.12.10-1997—American National Standard

for Transformers—230 kV and Below 833/958 through

8333/10 417 kVA, Single-Phase, and 750/862 through 60

000/80 000/100 000 kVA, Three-Phase Without Load Tap

Changing; and 3750/4687 through 60 000/80 000/100 000

kVA with Load Tap Changing-Safety Requirement;

IEEESTDC57.12.20-2005—IEEEstandardforoverhead-

type distribution transformers, 500 kVA and smaller: high

voltage, 34 500 V and below; low voltage, 7970/13 800y

V and below;

IEEE Std. C57.12.21-1992—American National Standard

Requirements for Pad-Mounted, Compartmental-Type

Self-Cooled, Single-Phase

with High Voltage Bushings; High-Voltage, 34500

GRYD/19920 Volts and Below; Low-Voltage, 240/120

Volts; 167 kVA and Smaller;

IEEE Std. C57.12.22-1993—American National Standard

for Transformers—Pad-Mounted, Compartmental-Type,

Self-Cooled Three-Phase Distribution Transformers With

High-Voltage Bushings, 2500 kVA and Smaller: High

Voltage, 34 500 Grd Y/19 920 Volts and Below; Low

Voltage, 480 Volts and Below, Reaffirmed 1998;

IEEE Std. C57.12.23-2002—IEEE Standard for Under-

ground Type, Self-Cooled, Single-Phase, Distribution

Distribution Transformers

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 14

2012IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 4, OCTOBER 2009

Transformers with Separable Insulated High-Voltage Con-

nectors; High Voltage 25 000 V and Below; Low Voltage

600 V and Below; 167 kVA and Smaller Revision of

C57.12.23-1992;

IEEE Std. C57.12.24-2000 Withdrawn—American Na-

tional Standard for Transformers Underground-Type

Three-Phase Distribution Transformers, 2500 kVA and

Smaller; High Voltage, 34 500 GrdY/19 920 Volts and

Below; Low Voltage, 480 Volts and Below Requirements;

IEEE Std. C57.12.25-1990—American National Standard

for Transformers—Pad-mounted, Compartmental-type,

Self-cooled,Single-phase

with Separable Insulated High-voltage Connectors; High

Voltage, 34 500 Grdy/19 920 Volts and Below; Low

Voltage, 240/120 Volts; 167 kVA and Smaller Require-

ments;

IEEEStd. C57.12.26-1992—IEEE

Pad-mounted, Compartmental-Type,

Three-Phase Distribution Transformers for Use with

Separable Insulated High-Voltage Connectors (34 500 Grd

Y/19 920 V and Below, 2500 kVA and Smaller);

IEEE Std. C57.12.28-2005—IEEE Std. C57.12.28—2005

IEEE Standard for Pad-Mounted Equipment—Enclosure

Integrity;

IEEEStd. C57.12.29-2005—IEEE

Pad-MountedEquipment—EnclosureIntegrityforCoastal

Environments;

IEEE Std. C57.12.31-2002—IEEE standard for Pole

Mounted Equipment—Enclosure Integrity;

IEEE Std. C57.12.32-2002—IEEE Standard for Sub-

mersible Equipment—Enclosure Integrity, Reaffirmed

2008;

IEEEStd.C57.12.34-2004—IEEEStandardRequirements

for Pad-Mounted, Compartmental-Type, Self-Cooled,

three-phase distribution transformers (2500 kVA and

smaller)—High-voltage: 34 500 GrdY/19 920 volts and

below; low-voltage: 480 volts and below;

IEEE Std. C57.12.35-2007—IEEE Standard for Bar

Coding for Distribution Transformers and Step-Voltage

Regulators, Revision of C57.12.35-1996;

IEEE Std. C57.12.36-2007—IEEE Standard Require-

ments for Liquid-Immersed

Transformers;

IEEE Std. C57.12.37-2006—IEEE Standard for the Elec-

tronic Reporting of Distribution Transformer Test Data,

Revision of 1388-2000;

IEEEStd.C57.12.40-2006—IEEEStandardRequirements

for Secondary Network Transformers, Subway and Vault

Types (Liquid Immersed);

IEEEStd.C57.12.44-2005—IEEEStandardRequirements

forSecondaryNetworkProtectors,RevisionofC57.12.44-

2000;

IEEE Std. C57.12.50-1981—American National Standard

Requirements for Ventilated Dry-Type Distribution Trans-

formers, 1 to 500 kVA, Single-Phase, and 15 to 500 kVA,

Three-Phase, withHigh-Voltage601to 34500Volts,Low-

Voltage 120 to 600 Volts;

IEEE Std. C57.12.51-1981—American National Stan-

dard Requirements for Ventilated Dry-Type Power

DistributionTransformers

Standard

Self-Cooled,

for

Standard for

Distribution Substation

Transformers, 501 kVA and Larger, Three-Phase, with

High-Voltage 601 to 34 500 Volts, Low-Voltage 208Y/120

to 4160 Volts;

IEEE Std. C57.12.52-1981—American National Standard

Requirements for Sealed Dry-Type Power Transformers,

501 kVA and Larger, Three-Phase, with High-Voltage 601

to 34 500 Volts, Low-Voltage 208Y/120 to 4160 Volts;

IEEE Std. C57.12.55-1987—American National Standard

for Transformers—Used in Unit Installations, Including

Unit Substations-Conformance Standard;

IEEE Std. C57.12.56-1986—IEEE standard Test Proce-

dureforThermalEvaluationofInsulationSystemsforVen-

tilated Dry-Type Power and Distribution transformers;

IEEE Std. C57.12.58-1991—IEEE Guide for conducting

a transient voltage analysis of a dry-type transformer coil,

Reaffirmed 1996, 2002, 2008;

IEEEStd.C57.12.59-2001—IEEEGuidefordr-typetrans-

former through-fault current duration, Reaffirmed 2006;

IEEE Std. C57.12.60-1998—IEEE Guide for test pro-

cedures for thermal evaluation of insulation systems for

solid-cast and resin-encapsulated power and distribution

transformers;

IEEE Std. C57.12.70-2000—IEEE Standard Terminal

Markings and Connections for Distribution and Power

Transformers;

IEEE Std. C57.12.80-2002—IEEE Standard Terminology

for power and distribution transformers, Revision of

C57.12.80-1978;

IEEE Std. C57.12.90-2006—IEEE Standard Test Code

for Liquid-Immersed Distribution, Power, and Regulating

Transformers;

IEEE Std. C57.12.91-2001—IEEE Standard Test Code for

Dry-Type Distribution and Power Transformers;

B. CENELEC Standards

EN 60076-1:1997/A12:2002—Part 1: General;

EN 60076-2:1997—Part 2: Temperature rise;

EN 60076-3:2001—Part 3: Insulation levels, dielectric

tests and external clearances in air;

EN 60076-4:2002—Part 4: Guide to the lightning impulse

and switching impulse testing—Power transformers and

reactors;

EN 60076-5:2006—Part 5: Ability to withstand short cir-

cuit;

EN 60076-6:2008—Part 6: Reactors;

EN 60076-10:2001—Part 10: Determination of sound

levels;

EN 60076-11:2004—Part 11: Dry-type transformers;

EN 60076-13:2006—Part 13: Self-protected liquid-filled

transformers;

EN 50216-1:2002—Part 1: General;

EN 50216-2:2002—Part 2: Gas and oil actuated relay for

liquid immersed transformers and reactors with conser-

vator;

EN 50216-3:2002—Part 3: Protective relay for hermet-

ically sealed liquid-immersed transformers and reactors

without gaseous cushion;

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

Page 15

AMOIRALIS et al.: TRANSFORMER DESIGN AND OPTIMIZATION: A LITERATURE SURVEY2013

EN 50216-4:2002—Part 4: Basic accessories (earthing

terminal, drain and filling devices, thermometer pocket,

wheel assembly);

EN 50216-5:2002—Part 5: Liquid level, pressure and

flow indicators, pressure relief devices and dehydrating

breathers;

EN 50216-6:2002—Part 6: Cooling equipment—remov-

able radiators for oil-immersed transformers;

EN 50216-7:2002—Part 7: Electric pumps for transformer

oil;

HD 428.1 S1:1992/A1:1995 Three-phase oil-immersed

distribution transformers 50Hz, from 50to 2500 kVAwith

highest voltage for equipment not exceeding 36 kV—Part

1:Generalrequirementsandrequirementsfortransformers

with highest voltage for equipment not exceeding 24 kV;

HD 428.3 S1:1994 Three-phase oil-immersed distribution

transformers 50 Hz, from 50 to 2500 kVA, with highest

voltage for equipment not exceeding 36 kV—Part 3: Sup-

plementary requirements for transformers with highest

voltage for equipment equal to 36 kV;

HD 428.1 S1:1992 Three-phase oil-immersed distribution

transformers 50 Hz, from 50 to 2500 kVA with highest

voltage for equipment not exceeding 36 kV—Part 1: Gen-

eral requirements and requirements for transformers with

highest voltage for equipment not exceeding 24 kV.

C. IEC Standards

IEC 60076-1 (2000-04) Power transformers—Part 1: Gen-

eral;

IEC 60076-2 (1993-04)Power transformers—Part2: Tem-

perature rise;

IEC 60076-3 (2000-03) Power transformers—Part 3: Insu-

lation levels, dielectric tests and external clearances in air;

IEC 60076-4 (2002-06) Power transformers—Part 4:

Guide to the lightning impulse and switching impulse

testing—power transformers and reactors;

IEC 60076-5 (2006-02) Power transformers—Part 5:

Ability to withstand short circuit;

IEC 60076-6 (in process) Powertransformers—Part 6: Re-

actors;

IEC 60076-7 (2005-12) Power transformers—Part 7:

Loading guide for oil-immersed power transformers;

IEC 60076-8 (1997-11) Power transformers—Part 8: Ap-

plication guide;

IEC 60076-10 (2005-07) Power transformers—Part 10:

Determination of sound levels;

IEC 60076-10-1 (2005-10) Power transformers—Part

10-1: Determination of sound levels—Application guide;

IEC 60076-11 (2004-05) Power transformers—Part 11:

Dry-type transformers;

IEC 60076-12 (in process) Power transformers—Part 12:

Loading guide for dry-type power transformers;

IEC 60076-13 (2006-05) Power transformers—Part 13:

Self-protected liquid-filled transformers;

IEC/TS60076-14(2004-11)Powertransformers—Part14:

Design and application of liquid-immersed power trans-

formers using high-temperature insulation materials;

IEC 60076-15 (in process) Power transformers—Part 15:

Gas-filled power transformers;

TABLE IV

CLASSIFICATION OF REFERENCES PRESENTED IN SECTION VII

BY PROBLEM TYPE AND PROBLEM NAME

VII. RECENT TRENDS IN TRANSFORMER TECHNOLOGY

Inthelastdecade,rapidchangesand developmentshavebeen

madeinthefieldoftransformerdesign.Thephenomenalgrowth

of power systems has put tremendous responsibilities on the

transformer industry to supply reliable and cost-effective trans-

formers. This section identifies the recent trends in research

and development in materials, insulations systems, accessories,

and diagnostic techniques, by quoting keyworks that address

them, giving pointers to readers desirous of pursuing research

in transformers. The references of this Section are summarized

in Table IV and they are discussed in further detail in the forth-

coming subsections.

A. Magnetic Circuit

There has been a steady development of core steel material

in the last century. The trend of reduction in transformer losses

in the last few decades is related to a considerable increase in

energy costs. One of the ways to reduce the core losses is to

use better and thinner grades of core steels, but their price is

higher. However, continuous efforts are directed at developing

improved electrical steels with lower iron losses for energy-ef-

ficient transformers. It is well known that low magnetic losses

of amorphous material are attributable to the material’s amor-

phous condition and small thickness of the ribbon [384], [385].

The core losses can be limited by insulating coatings [386], and

various types of coatings have been developed for application

to both fully processed and semiprocessed electrical steels. In

addition, Matsuura et al. [387] and Ezure et al. [388] presented

long-term property reliability for iron-based amorphous alloy

for use in liquid-immersed transformer cores.

B. Windings

The advent of high-temperature superconducting (HTS)

materials has renewed interest in research and development

of superconducting transformers. The principal advantages of

HTS transformers are: much lower winding material content

and losses, higher overload capacity and possibility of coreless

design. Some considerations from design point of view are

discussed in [389], [390], while in [391] new perspectives of

HTS transformer design are introduced. The development of

technology based on liquid nitrogen at temperatures up to 79 K

Authorized licensed use limited to: Marina Tsili. Downloaded on October 1, 2009 at 05:58 from IEEE Xplore. Restrictions apply.

#### View other sources

#### Hide other sources

- Available from Eleftherios I. Amoiralis · May 27, 2014
- Available from ntua.gr