Renewable and Sustainable Energy Reviews 76 (2017) 687–700

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

Evolution of transformer health index in the form of mathematical equation MARK

⁎
A. Azmi, J. Jasni , N. Azis, M.Z.A. Ab. Kadir
Centre for Electromagnetic and Lightning Protection Research (CELP), Department of Electrical and Electronic Engineering, Faculty of Engineering,
Universiti Putra Malaysia, 43400, Serdang, Selangor, Malaysia

A R T I C L E I N F O A BS T RAC T

Keywords: Energy is a basic necessity in every country. The worldwide demand for energy will rise due to the developments
Transformer asset management of power generation in industrial, service, and residential sectors. A healthy power system is therefore very
Transformer health index important to guarantee continuous electricity supply to the end users and this can be achieved through asset
Transformer condition monitoring management. A proper asset management will allow asset managers to conduct quality assessment of conditions
Renewable energy
and to develop future management strategies of the electrical assets such as transformers. The execution of
transformer asset management involves an investigation of the transformer's condition by employing
Transformer's Health Index (THI). Mathematical equation/algorithm or expert judgment has been investigated
by many previous studies as one of the technique to determine health index (HI). Some of the established
methods of HI determination such as scoring and ranking method, tier method, matrices and multi-feature
assessment model have led to the different interpretations of the final condition of a transformer. This paper
critically examines and explores the previous studies related to transformer health index by using mathematical
equation/algorithm or expert judgment. The concept of HI and its formulation are presented in this study.
Generally, there are three parts of HI formulation which are input, algorithm for HI and the output of HI. The
application of HI is discussed in terms of the performance of in-service transformer. The limitations of the
available methods are also discussed and future works to overcome the problems are suggested.

1. Introduction Therefore, transformer asset management has been widely adopted to

manage and prevent failures from suddenly occurs.
Energy is considered as the ‘back-bone’ of human activities and a Generally, asset management is considered as a series of processes
vital commodity for the survival of modern economies. It is an that begins with an examination that proceeds to analysis and
interesting subject to encourage electricity production [1]. There are prioritization of assets which are conducted across an organization.
many forms of energy resources such as conventional, renewable and There are many approaches in transformer asset management that lead
sustainable energy. Nowadays, more renewable and sustainable energy to final output decision of the transformer. One of the important
resources have been considered for electricity generation due to the approaches in transformer asset management is transformer condition
need to provide a safe and clean future electricity system. Moreover, assessment. In this approach, after the process of collecting and
renewable energy is abundant, sustainable, locally available and widely analyzing data, the data will then be converted to an index which
distributed [2]. Another reason for the increase in the adoption of indicates the health condition of the transformer. Risks and opportu-
renewable energy is because it helps to assure energy security, reduce nities can also be identified by using this index.
reliance on conventional energy resources as well as bridging the This paper will focus on transformer asset management, health
demand-supply gap [3,4]. index (HI), input of HI, previous mathematical equation/algorithm of
Transformer plays an important role in the energy transmission. It HI, output of HI, application of HI in-service, and future trends of asset
links the energy transmission from generation or extraction point to management with the integration of renewable energy resources.
the distribution or consumption point. In the event of transformer
failures, the utilities will experience major economic such as loss of 2. Transformer asset management
revenue and market backlash. As a result, the end users could face an
electrical shortage and which will also cause shutdown of hundreds Transformer asset management techniques have gradually evolved
industries, paralyzing production and exacerbating unemployment [2]. from time to time which involve intelligence framework techniques

⁎
Corresponding author.
E-mail address: [email protected] (J. Jasni).

http://dx.doi.org/10.1016/j.rser.2017.03.094
Received 16 June 2016; Received in revised form 21 January 2017; Accepted 20 March 2017
1364-0321/ © 2017 Elsevier Ltd. All rights reserved.
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

where the technique include fuzzy logic, digital signal processing developed using the algorithms to prevent transformer failure. Brittes
algorithms, pattern recognition algorithms, principal component ana- et al. introduced Technical & Economical Real Time Transformer
lysis (PCA) algorithms and back-propagation artificial neural network Assessment (T & ERTTA) approach for the technical and economic
(BP-ANN) algorithm. For example, an intelligence framework for power transformer assessments [13]. In this approach, real time data
condition monitoring and assessment of power transformers has been base is used to calculates real daily loss of technical transformer life.
developed by Ma et al. where various signal processing and pattern Dominelli et al. developed a computer-based program to determine the
recognition algorithms were used to provide dynamic information of a equipment health rating of power transformers [14]. This program
transformer's condition [5]. In a different study by Arshad et al., asset incorporates test, inspection data and equipment information to
management and decision-making model by using diagnostics and data diagnose and assess the condition of the transformers. In summary,
interpretation techniques has been developed based on fuzzy logic as observed from prior studies, there are many methods and strategies
approach [6]. The used of fuzzy logic model has successfully facilitates available to the utility industries for transformer asset management.
in addressing the criticalities of power transformers where premature The execution of transformer asset management involves historical
failures can be predicted and cost effective maintenance strategies can review, condition assessment, financial information, economic risk
be developed to further reduced any expensive risks. Abu-Elanien et al. management, and final decision whether to maintain, repair and
briefly discussed the classifications of transformer asset management upgrades, replace, monitor or conduct contingency control to the
which are condition monitoring (CM), condition assessment (CA), transformers [15,16]. This approach is important to identify transfor-
performing maintenance plans, aging, health and end-of-life assess- mers or groups of transformers that are at risk in providing the future
ment [7]. The estimation of transformer health condition were carried system resource needs [5].
out by Abu-Elanien et al. by using a feed forward artificial neural Fig. 1 depicts the basic process of asset management. The process is
network (FFANN) and it was found that the model is reliable in finding divided into two main sections historical review and financial informa-
the health condition of any working transformer [8]. Abu-Elanien et al. tion. In the historical review, data such as historical information,
later presented a new method for calculating a health index for loading history, diagnostic data, failure history, expected stresses in the
transformers by using fuzzy set theory [9]. The results which relies future at its particular location in the network are collected to assess
on the use of furan analysis, dissolved gas analysis, and other oil the conditions of a transformer. As shown in the figure, the second step
analysis showed 96.7% similarity results to those obtained by an in the process, a condition assessment, involves a careful inspection
experienced asset management and health assessment consulting and extensive testing of the transformer which include results of test
company [9]. Abiri-Jahromi et al. proposed a two-stage framework and measurement, observations, operation history, knowledge of the
for power transformer asset maintenance management [10,11]. The failure mechanisms and processes, previous experience with similar or
model optimizes maintenance outage schedule over a predefined comparable equipment and any other relevant knowledge and informa-
period of time by taking into account the actual and expected tion. The normal output of condition assessment can range from a
transformer assets’ condition where it incorporates joint midterm simple normal or abnormal assessment to a sophisticated ‘asset health
and short-term maintenance [10]. The numerical studies showed that index’ which refers to a ranking or scoring system on a single or
the two-stage framework are able to maximizes the transmission asset multiple scales to allow decisions on future maintenance or replace-
serviceability over a predefined period [11]. Trappey et al. developed an ment prioritized over a fleet of units [17]. Financial information
intelligence engineering asset management system by using principal approach is another important process to improve the efficiency of
component analysis (PCA) algorithms and back-propagation artificial grid assets management which includes life cycle cost model, invest-
neural network (BP-ANN) algorithm [12]. The algorithms are use to ment, operation, maintenance, failure and disposal cost [18]. In recent
predict transformer failure using data mining and fault prediction research, the life cycle cost model is now integrated with environmental
model. Further action such as maintenance and repair strategies is and preventive test cost [19].

Fig. 1. Power transformer asset management.

688
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

Fig. 2. Concept of health index.

Both financial information and condition assessment are combine infrared thermography, conductivity factor and polarization index.
to carry out economic risk management. Economic risk management DGA is normally considered as an important input for HI formula-
will be used to decide whether to maintain, repair and upgrades, tion because it is an established method to detect defects in transfor-
replace, monitor or conduct contingency control to the transformer. mers such as arcing, partial discharge, low energy sparking, severe
Condition assessments of the transformer are expected to be the main overloading and overheating. The presence of certain gases above the
decision driver in future planning and the budgeting of transformer recommendation limit by IEEE and IEC shows that fault occurs in a
asset management. transformer. One of the standards used to analyze dissolved gas is IEC
60599 [24] which provides a coded list of faults detectable by DGA.
3. Concept of health index and formulation Meanwhile, IEEE Standards C57.104 is the standard that introduces a
four level criterion to classify risks to a transformer for continued
HI is commonly known as a practical tool that combines the result operation at various combustible gas levels [25].
of operating observations, field inspections, and site and laboratory The second important parameter in the formulation of HI is furanic
testing. The combination of the results will be converted into an compound content. The measurement of furfural content in the oil can
objective and quantitative index which provides an overall health of be used for a bulk measurement of the degree of polymerization of the
the asset [20,21]. It plays an important role in determining the insulation paper. This parameter provides a good indication of paper
condition and assist in monitoring the health of a transformer. degradation. IEC 61198 shows the measurement of trace furanic
Undoubtedly, HI is an excellent indicator in reflecting the results of compounds [26]. Another parameter that is usually included in HI
optimal balance among capital investment, asset maintenance cost and calculation is oil quality test. This parameter provides a good indication
operating maintenance. HI concept consists of three components of the overall condition of the oil and internal components and can also
which are input, mathematical equation/algorithm for HI and output be used in preventing premature aging of a transformer. Most of the
as illustrated in Fig. 2. findings suggest that in general, the health of a transformer is highly
affected by the condition of its insulation oil [27].

3.1. Input of health index

3.2. Mathematical equation/algorithm of health index
The input for HI is usually obtained from the operating observa-
tions, field, site and laboratory testing. Some common diagnostic tests Previous researchers have developed several mathematical equa-
that are usually used for calculation of HI are dissolved gas analysis tions or algorithms for the formulation of HI. The mathematical
(DGA), oil quality (breakdown voltage (BDV), interfacial tension (IFT), equations or algorithms encompass scoring and ranking method, tier
acidity, water content (WC), color), furan analysis (FA), and degree of method, matrices/entropy weight health index (EWHI) method and
polymerization (DP). Some advanced diagnostic tests such as frequency multi-feature factors assessment model. Even though they are using
response analysis (FRA) and recovery voltage measurement (RVM) the same basic equation, some improvements have been made from
have also being used by several studies. Other parameters include tap time to time to make the equations more reliable and scientifically
changer and bushing, load history, maintenance data and age [15,20– proven. Table 4 shows the list of mathematical equations/algorithms
23]. The number of parameters used in the calculation of HI is also developed by previous researchers.
different among researchers. The list of diagnostic tests corresponds to
different transformer components and the numbers of parameters used 3.2.1. Scoring and ranking method
in previous researches are tabulated in Tables 1 and 2 respectively. In 2008, Naderian et al. has developed a condition-based HI by
The weighting factor represents the degree of importance or using scoring and ranking method [20]. The input of HI included test
contribution of any particular parameter that affects the condition of results of DGA, oil quality, furan, power factor, tap changer and
a transformer. The weighting factors for each parameter are presented bushing condition, physical observations, load history, maintenance
in Table 3 and the parameters that have the highest weighting factors work orders and age. 20 parameters have been used in this study and
are DGA, load history, power factor, global loss factor, thermal scans, the calculation involved condition ratings, weighting factor and as-

689
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

Table 1
List of diagnostic tests.

COMPONENT

INSULATING OIL BUSHING CORE TAP CHANGER WINDING TANK AND ASSOCIATED
DEVICES

Water content (WC) Capacitance Excitation currents Ratio Capacitance Visual inspection
Breakdown voltage (BDV) Dielectric loss Insulation Dynamic resistance Dielectric loss Infrared thermography
resistance
Acidity Partial discharge Motor currents Partial discharge Calibration
Interfacial tension (IFT) Infrared thermography Vibro-acoustic Insulation resistance
Dissolved gas analysis (DGA) Ratio/polarity/phase
Furan Winding resistance
Frequency response analysis
(FRA)
Recovery voltage measurement
(RVM)
Frequency domain spectroscopy
(FDS)
Degree of polymerization
Polarization and depolarization
current

Table 2 Table 3
Number of parameters. Weighting factor for parameter.

Author Year Number of parameter PARAMETER WEIGHTING

FACTOR
Hernanda et al. [28] 2014 3
En-wen and Bin [29] 2014 4 Dissolved gas analysis (DGA) [21,23,28,30,33,36,38] 10, 3
Martins [30] 2014 19 Load history [20,21,23,30,36] 10
Nelson et al. [31] 2014 4 Power factor [20,21,23] 10
Ashkezari et al. [32] 2013 3 Global loss factor [36] 10
Haema and Phadungthin [23] 2013 21 Thermo scan [23] 10
Taengko and Damrongkulkamjorn [33] 2013 14 Infrared thermography [30,36] 10, 8
Tanasescu et al. [34] 2013 17 Conductivity factor [36] 10
Malik et al. [35] 2012 6 Polarization index [36] 10
Gorgan et al. [36] 2010 27 Furanic compound content [20,21,23,28,30,36] 9, 6, 5
Jahromi et al. [21] 2009 24 Oil quality [20,21,23,28,30,32,36] 8, 6
Ghazali et al. [37] 2009 12 Overall transformer condition [20,21,36] 8, 6
Naderian et al. [20] 2008 20 Leakage reactance [21,30,36] 8
Winding resistance [21,30,33,36] 8, 6, 2
Bushing conditions [20,21,36] 7, 5
signed score. The weighting factor is assigned by the subjective Frequency response analysis (FRA) [30] 6
DGA of tap changer oil [20,21,23,30] 6
judgment from the expert. The scores for certain parameters are
Turns ratio [21,33,36] 5, 2
determined by using the recommendation limit from IEC and IEEE Tap changer contacts condition [30] 5
standard. In 2009, Jahromi et al. has improved the scoring and ranking Overall LTC condition [20,21,36] 5, 2
method by adding the number of parameters from 20 to 24 [21]. Age [30,33] 4
Paper insulation factor [32] 4
Naderian et al. has developed a mathematical equation to find the
Inspection and maintenance [33] 4
factor for each parameter [20]. The sum of ith score multiplied by ith Internal faults history [33] 4
weight for all components of a parameter and is then divided by the Dielectric breakdown test [33] 3
sum of ith weight for all components for a parameter in order to obtain Water content test [33] 3
the factor for a parameter. This factor computation is applied to DGA Surge arrester [23] 3
Cooling equipment condition [30] 3, 2
and oil quality test parameters. This factor determination is calculated
Tap changer oil quality [20,21,23,30,36] 3
to balance all components for each parameter. In the final equation of Location [33] 2
HI, the sum of jth weighting factor (Kj) is multiplied by jth health index Main tank corrosion [20,21,30,36] 2, 1
factor (HIFj) for all parameters and divided with the sum of the highest Insulation resistance test [33] 2
Core-to-ground connection [30,36] 2
HIF multiplied by Kj. 60% weighting is assigned to a transformer and
Oil leaks [20,21,30,36] 2, 1
40% is assigned to LTC [20,21]. Manufacturer, General condition, Conservator tank, Oil 1
The scoring and ranking method was further improved by Haema tank corrosion, Preservation system condition, Hot
and Phadungthin by dividing the parameters used in HI determination line oil filter, Foundation, Radiator and cooling
into three groups (electrical test, insulating oil test and visual inspec- system, Grounding, Transformer control cabinet,
Gaskets, Seals, Neutral Ground Reactor (NGR),
tion) [23,39]. The basic scoring and weighting method is used in the
Connectors, Protection equipment, regulating PT, oil
calculation of HI but the equation is multiplied by percentage (%) and level, OLTC compartment, OLTC control cabinet
the denominator is added with Smaxi for each parameter. The number [20,21,23,30,33,36]
of parameters used by Haema and Phadungthin is 21. In 2010, Gorgan
et al. expanded the list of parameters used from 24 to 27 diagnostic
factors by adding in loss factor at very low frequency, conductivity the used of 27 diagnostic factors allowed a more precise determination
factor and polarization index [36]. Gorgan et al. discovered that HI is of HI.
strongly influenced by those added factors. They also emphasized that In general, for the conventional HI formulation through scoring
and ranking method, there are several steps that need to be taken to

690
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

Table 4
List of algorithm.

AUTHOR YEAR PARAMETER EQUATION/ALGORITHM

Hernanda et al. [28] 2014 HI each parameter ∑in=1 Si × Wi

HI each parameter =
∑in=1 Wi

HI final ∑in=1 Si × Wi
HI final = ×100%
∑in=1 4Kj

En-wen and Bin [29] 2014 Main health index (HIm) HIm = HI0×e B ×(T 2− T1)
HIm=wanted health index
HI0=initial health index
B=aging coefficient
T1=year corresponding to HI0 (year put into operation)
T2=year studied (present year or some year in future)

Insulating paper health HI c, o = ∑i3=1 ωi ×Fc, o (i )

index (HIiso) ω=weight
Fc,o=parameters (CO, CO2, CO+CO2)
HIfur =3.344 × (Cfur )0.413
Cfur=furfural content
HIiso = ω1 HIc, o+ω2 HIfur
HIiso=insulating paper health index
ω1=weight factor HIc,o (0.3)
ω2=weight for HIfur (0.7)

Index based on DGA (HIC,H) FC , H (i )=ax (i )+b

HIC , H = ∑i5=1 ωi FC , H (i )
ω=weight of gases (H2, CH4, C2H6, C2H4, C2H2)
Index based on oil quality HIoil = ∑i4=1 ωi Foil (i )
factor (HIoil) ω=weight of acid, BDV, micro-water, dielectric loss
Comprehensive health index HIcom=f ∑i4=1 ωi HI (i )
(HIcom) f=fault correction factor
ωi=weight n z
∑ j =1 Kj Sj ∑ j = n +1 Kj Sj
Martins [30] 2014 Final HI HI = X % +Y % z
∑nj =1 4Kj ∑ j = n +1 4Kj

X=transformer
Y=tap changer LTC
Sj=score corresponding to parameter ‘j’
Kj=weighting factor corresponding to parameter ‘j’
j=number
n
of each diagnostic parameter used for calculation of the health index
∑ j =1 Kj HIFj
Nelson et al. [31] 2014 Final HI HI =
∑nj =1 4Kj

Kj=weightage of ith parameter

HIFj=health index factor of ith parameter

Ashkezari et al. [32] 2013 DGAF ∑i7=1 Si × Wi

DGAF =
∑i7=1 Wi
i=dissolved gas (H2, CH4, C2H6, C2H4, C2H2, CO and CO2)
Si=score value
Wi=weighting factor
OQF ∑i4=1 Si × Wi
OQF =
∑i4=1 Wi
i=oil tests (BDV, acidity, moisture, DDF)
Si=score value
Wi=weighting factor
HI ∑3j =1 Fi × Ki
HI1 =
∑3j =1 Ki

Fi=factor values for DGAF, OQF, PIF

Ki=values for DGAF (10), OQF (8), PIF (4)

P
Haema and Phadungthin [23] 2013 Power Factor %PF = ×100
VI
P=power
V=voltage
I=current

DGA of oil main tank i = CO

∑i = H 2 (Si × Wi )
2
(%) DGAF = i = CO
×100
∑i = H 2 (Smaxi × Wi
2

DGA of oil in OLTC ∑ii = n

=1 (Si × Wi )
compartment (%) DGATF = ×100
∑ii = n
=1 (Smaxi × Wi )

Quality of oil in main tank ∑ii = color

= BD (Si × Wi )
(%) OQF = ×100
∑ii = color
= BD (Smaxi × Wi )

(continued on next page)

691
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

Table 4 (continued)

AUTHOR YEAR PARAMETER EQUATION/ALGORITHM

Quality of oil in OLTC ∑ii = color

= BD (Si × Wi )
compartment (%) TOQF = ×100
∑ii = color
= BD (Smaxi × Wi )

Final HI ⎛ ∑jj =17 ∑jj =21 ⎞

=1 (Kj × HIFj ) =18 (Kj × HIFj ) ⎟
(%) HI = ⎜⎜0.6 j =17 +0.4 j =21
⎝ ∑j =1 (HIFmax × Kj ) ∑j =18 (4 × Kj ) ⎟ ⎠

Scattiggio and Pompili [44] 2013 HI HIoil =weighted rank test1 + weighted rank test 2 + …test n
HI electric + HI thermal + HI mechanical + HIoil
HI =
max
max=14
HI will be expressed in per unit (pu)
Taengko and Damrongkulkamjorn 2013 HI HI = (0.4 × history score)+(0.6 × condition score)
[33]
Tanasescu et al. [34] 2013 HI ∑in=1
−3 c . DI
i i ∑n ci . DIi
HI = A1. +A2. i =n n −3
∑in=1
−3 4 . c
i ∑i = n −3 4 . ci

ci=rating given to each state element

DIi=value of the diagnostic index
n=number of considered diagnostic factors
A1, A2=corresponding weight of n-3 factors (Tx, load tap changer
Malik et al. [35] 2012 HI Tier 1=Transformer condition assessment summary
(Total Score for Tier 1=(Ʃscore×weighting factor)
Tier 2=transformer condition summary
Net Transformer Condition Index=Tier 1-Tier 2

Gorgan et al. [36] 2010 Gas Analysis Factor ∑i7=1 ni ∙ pi

Fg =
∑i7=1 pi
Fg=gas analysis factor
ni=1…6 is the score given to gas i
pi=weight factor
i (30) − ir (30)
Conductivity Factor kc = a
ia (60) − ir (60)
kc=conductivity factor
ia,r (30,60) represent the values of absorption/resorption currents measured at 30 and 60 s from
the applied voltage start
i (60)
Polarization Index kp = a
ia (600)
kp=polarization index
ia (60) and ia (600) represent absorption current values measured at 60 s and 600 s from the
applied voltage
σ
U start
Ɛ"r +
Loss factor at f=1mHz ω ∙ Ɛ0
tgδ =
Ɛ′r
tgδ=loss factor at f=1 mHz
ε'r and ε"r=real and imaginary part of complex relative permittivity
σ=dc conductivity
ω=angler frequency
ε0=vacuum permittivity
Final HI ∑in=1
−3 c . DI
i i ∑n ci . DIi
HI = A1. +A2. i =n n −3
∑in=1
−3 4 . c
i ∑i = n −3 4 . ci

ci=rating given to each state element

DIi=value of the diagnostic index
n=number of considered diagnostic factors
A1, A2=corresponding weight of n-3 factors (Tx, load tap changer)

Jahromi et al. [21] 2009 DGAF ∑i7=1 Si × Wi

DGAF =
∑i7=1 Wi
Si=1–6
Wi=assigned weighting factor

Load factor ∑i4=0 (4 −i ) × Ni

LF =
∑i4=0 Ni
Final HI ∑21j =1 Kj HIFj ∑24
j =22 Kj HIFj
HI = 60% × + 40%
∑21
j =1 4Kj ∑24j =22 4Kj
Kj=weighting factor for parameter
HIFj=health index factor
60%=weighting factor assigned to transformer
40%=weighting factor assigned to LTC

determine the final HI for an asset. For the calculation of HI, there parameter factor will be converted to HIF by referring to a table of
are no fixed parameters have been used so far but the most common parameter factor and HIF value. The HIF will then be used in the
one are dissolved gas analysis (DGA), oil quality test and furan. The final calculation of HI along with constant weighting factor (Kj).
first step involves the determination of health index factor (HIF) for From previous research, for the final HI equation, the assigned
each parameter. HIF for certain parameters such as DGA and oil weighting factor for transformer and for load tap changing (LTC) are
quality are then normalized by using the scoring and weighting 60% and 40 respectively. This is based on the report by CIGRE
values which usually different between utilities. The final value of survey which showed that in large power transformers, 40% of the

692
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

Fig. 3. Scoring and ranking method [20,21,23,39].

failures are due to LTC [30,40]. The steps to obtain the final HI is
illustrated in Fig. 3 below.

3.2.2. Combination of scoring, ranking and tier method

In a different study, Zaidey et al. adopted the scoring and ranking
method into a three-tiered approach [37,41]. Tier 1 served as the
baseline audit for assessing the presence of faults, quality of the
Fig. 4. Combination of scoring and ranking method and tier method [37,41].
insulating oil, degradation level of insulation paper as well as physical,
thermal and operating performance of the transformers. The para-
judgment by experts. The final HI value is calculated by using Eq. (1):
meters involved in Tier 1 are DGA, oil quality analysis (BDV, water
content, acidity and power factor), FFA, inspection on physical condi- HI=(0.4 × history score) + (0.6 × condition score) (1)
tions and operating performance and thermography. Tier 2 will be
applied if the diagnosis from Tier 1 could not classify a transformer as
3.2.3. Matrices
normal or the value of transformer condition index (TCI) is below 55.
Another tool to calculate HI is by using matrices. Zhou et al. in their
Tier 2 involves diagnostic of transformer turns ratio measurement,
study combined entropy weight method and HI together, and use
winding resistance measurement, dielectric dissipation factor/tan delta
entropy weight health index (EWHI) to assess the status of power
measurement, excitation current measurement and insulation resis-
transformer [43]. The data of diagnostic tests are used to establish and
tance and polarization index (PI). Tier 3 will then be performed if Tier
calculate unitary judge-matrix. The entropy, entropy weight and EWHI
1 and Tier 2 diagnostics could not classify the condition of a
for each parameter were obtained for further comprehensive status
transformer as normal or the value of TCI is below 55. Tier 3 involved
evaluation of transformers. Fig. 5 summarizes the steps to obtain the
advanced diagnostic tests such as FRA and partial discharge (PD)
final value of HI by using the matrices method.
measurement. Each parameter has been assigned to certain weighting
factor and the scores are determined by using the recommendation
limit from IEEE and FIST [25,42]. The final THI will be compared to 3.2.4. Multi features factor assessment model
condition based ranking value and the recommended mitigating En-wen and Bin proposed a new assessment model that is based on
actions will be taken. Fig. 4 shows the step of this method. the multi-feature factors to overcome the disadvantages of the tradi-
The study on combination of scoring and ranking and tier method tional condition assessment [29]. The method comprises four sub
in the determination of HI was also carried out by Malik et al. [35]. factors which are main health index (HIm), insulating paper health
However, a two-tiered approach was implemented in the study instead index (HIiso), index based on oil quality factor and (HIoil). The final HI
of three-tiered approach. In this approach the diagnostic methods are is the fault correction factor multiplied with the sum of the weights of
classified into Tier-1 tests and Tier-2 tests. Tier-1 tests include the four indexes. The weight of each parameter was obtained by an
insulating oil analysis (DGA and furan), power factor and excitation analytic hierarchy process. Li En-wen used piecewise linear function to
current tests, operation & maintenance history and age while Tier-2 obtain the factors.
tests involved the data from the turn ratio test and sweep frequency
response analysis (SFRA). The transformer health index (THI) for 3.3. Output of health index
transformer condition assessment summary is determined in Tier 1.
Net TCI is obtained by subtracting Tier 2 from Tier 1 and the result will The output of the final HI will be subjected to a certain range where
be compared to the TCI range for further action. the preventive action will be taken accordingly. There are various
Another studies by Taengko and Damrongkulkamjorn used the ranges presented by researchers and there is no standard in determin-
same scoring and weighting method but the parameters are divided ing the range and the preventive action taken. Some used range 1–10
into two groups which are history group and condition group [33]. In where 1 shows a good condition of transformer while 10 refers to bad
history group, parameters involved are age, loading history, inspection condition of the transformer. Others used range of 100–0 where 100
and maintenance, internal faults history, location and manufacturer. indicate the good transformer while 0 indicates the bad transformer.
Dielectric breakdown test, water content test, DGA, insulation resis- Other examples of range are from 1 (excellent) to 4 (poor), 1 (excellent)
tance test, winding resistance test, turn ratio test, winding insulation to 5 (unsatisfactory), 0 (very good) to > 0.3 (poor), 0 (very poor) to 1
power factor test and oil insulation power factor test are in the (very good), and 0 (excellent) to 4 (further investigation). Some
condition group. The weighting factor is determined by subjective researchers have also added the failure rate or probability of failure

693
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

this level is data-driven with engineering support. In the predictive

level, HI is consistently used across all asset classes. Every data
collected has a defined use within the business and unnecessary data
collection has been stopped. The predictive level depends on data-
driven or fact-based decision making [50].
Utility companies which have adopted HI for their asset manage-
ment plan have proved that the application of HI is capable of reducing
the maintenance cost and balance the capital investment with opera-
tion and maintenance cost. Fig. 7 shows an example of the breakdown
maintenance cost from 2002 to 2006 [50].
Majority of utility companies adopt a basic form of health index
equation to determine transformer's health index and depend on expert
judgment in determining the weighting for each parameter. The
weighting for each parameter varies between utilities.
The available HI methods have some limitations where the values of
scoring and ranking are not fixed and vary according to experts in the
utilities. The HI will not be similar even if the same equation was used
due to inconsistency of the weighting factor and scoring method used
in the determination of HI. As for that reason, the improvement on the
determination of weighting factor could be determined mathematically
according to certain data.

5. Future trends of asset management with the integration of

renewable energy resources

Nowadays, the topic of renewable and sustainable energy resources

is one of the most active areas in electricity generation. To cater the
demand for energy, some countries have combined both conventional
and renewable energy resources in order to generate electricity. The
integration of renewable energy resources with the power system
network presents several challenges to the utilities especially in terms
of compatibility with the existing power system equipment. Currently,
there has been a number of studies carried out to examine the impact
of photovoltaic (PV) solar power generation on the power system
equipment [51–53].
Fig. 5. Matrices method [43].
Based on the previous studies conducted on the impact of PV solar
system, rapid changes in network demand or in supporting generation
to the description of the range. The outputs of health index by previous
(PV solar system) may potentially cause power quality issues such as
studies are presented in Table 5.
voltage flicker and harmonic distortion [54]. Harmonic in a power
system network may also cause heating on power cables and transfor-
4. Application of health index in service mers [55,56]. Given that transformers are among the most important
and expensive equipment, it is important to maintain transformers in-
In recent years, HI and condition-based methodologies have been service for as long as possible. Previous studies have found that the
widely adopted by utilities worldwide. This is because the global harmonic in a transformer could increase transformer losses and cause
engineering activities have been subject to much closer scrutiny and abnormal rise in temperature, which in turn will decrease the expected
there is great pressure to reduce cost while maintaining or improving lifetime [57–59]. Normally, temperature rises refer to the hot-spot
system performance. As a result, there is a greater need to provide temperature inside the transformer which is located in the transformer
technical and economic justification for engineering decisions and winding. Hot-spot temperature can cause advanced degradation of the
spending plans. oil and paper in transformers which, in severe case can lead to the
The application of HI is not limited to transformers only but can transformer failure. Therefore, the application of transformer asset
also be applied to all assets in utility companies. Since transformers are management specifically transformer health index is very important to
classified in Priority 1 (P1) class under prioritization of assets it detect the condition of the transformer itself.
represents the highest priority and is of high value (in terms of total
sustainment program expenditures) and high risk to the business. 6. Conclusion
There are basic and advanced techniques in determining HI. The
advanced techniques involved in neural network analysis and other Energy supply and demand are crucial aspects in governing a
advanced methodologies. Fig. 6 depicts the maturity level of the country. There are several forms of energy that can be converted into
adoption of HI. electricity production. The distribution of the electricity production
In the reactive level, HI is not used due to some limitations or no relies on the transformer. HI is an approach to determine the condition
consistent data available on most assets. In the controlled level, some of a transformer in transformer asset management. There are many
HI are in place but the consistency are questioned. There is regular/ methods that can be employed to determine HI such as scoring and
consistent data collection of asset. The decision-making process for ranking method, tier method, matrices method and multi-feature
both reactive and controlled level is engineer-driven and takes support- factor assessment model. The most popular method that has been
ing analysis into account. In the proactive level, HI is consistently used constantly used is the scoring and ranking method. Each of these
to compare similar assets. There is an abundance of data collection and methods has its own advantages and disadvantages. Recently, the
some analyses of data requirements. The decision-making process for matter of expert judgment in the weighting factor determination has

694
 Table 5
Output of health index.

AUTHOR YEAR OUTPUT

A. Azmi et al.

Waugh and Muir [45] 2015 Score Status Suggested Course of Action
≥7 and ≤10 Good Continue operation without restriction.
Repeat condition assessment as needed.
≥3 and < 7 Caution Continue operation but assess maintenance
practices. Repeat condition assessment as
needed.
≥0 and < 3 Poor Remove from operation and consult with
experts. Begin rehabilitation/replacement
process.

Amir and Muttalib [46] 2014 Health Index Condition Action Index Recommended action
10 Good 4 Normal monitoring/maintenance.
9 Good 4 Normal monitoring/maintenance.
8 Good 4 Normal monitoring/maintenance.
7 Fair 3 Investigate and monitor in 3–6
months interval. Trend data.
6 Fair 3 Investigate and monitor in 3–6
months interval. Trend data.
5 Poor 2 Repair and refurbish where possible.
4 Poor 2 Repair and refurbish where possible.
3 Bad 1 Remove from service.
2 Bad 1 Remove from service.
1 Bad 1 Remove from service.

Ashkezari et al. [38] 2014 Health Index Condition of Transformer Insulation

695
Levels
1 Excellent
2 Good
3 Fair
4 Poor

En-wen and Bin [29] 2014 Index value Health status Failure rate
0–3.5 Slightly aging Low
3.5–5.5 Obviously aging, but still belong to the normal Relatively low, but begin to increase
range
5.5–7 Aging beyond the normal range Significantly increase
7–10 Extremely poor state Fault may happen at any time

Hernanda et al. [28] 2014 Health Index Condition Description Approximate expected lifetime
85–100 Very good Some aging or minor deterioration of a More than 15 years
limited number of components.
70–85 Good Significant deterioration of some More than 10 years
components.
50–70 Fair Widespread significant deterioration or Up to 10 years
serious deterioration of specific
components.
30–50 Poor Widespread serious deterioration. Less than 3 years
0–30 Very poor Extensive serious deterioration. At end of life

Martins [30] 2014 HI (%) Condition Actions

85–100 Very good Normal maintenance
70–85 Good Normal maintenance
50–70 Fair Increase the number of diagnostic tests,
corrective maintenance or need of
(continued on next page)
Renewable and Sustainable Energy Reviews 76 (2017) 687–700
 Table 5 (continued)

AUTHOR YEAR OUTPUT

A. Azmi et al.

replacement, depending on the criticity.

30–50 Poor Start planning the replacement process or
repair, taking in account the risk.
0–30 Very poor Immediate risk assessment, replacement or
repair, depending on the case

Nelson et al. [31] 2014 Health Index Health status

(%)
100 < HI < 80 Excellent
80 < HI < 70 Good
70 < HI < 60 Alarm state
60 < HI < 50 Maintenance is required
HI < 50 Poor

Taengko and 2013 Mark Score Risk Level Remark

Damrongkulkamjorn [33]
A 4 None Normal
B 3 Low Quite normal
C 2 Moderate Quite normal
D 1 High Abnormal

Ashkezari et al. [32] 2013 Health index Insulation system condition

levels
1 Excellent
2 Good
3 Fair

696
4 Poor
5 Unsatisfactory

Scattiggio and Pompili [44] 2013 HI Condition Color

0–0.10 Very good Green
0.1–0.20 Good Pale blue
0.20–0.30 Fair Yellow
> 0.30 Poor Red

Haema and Phadungthin [23] 2013 %HI Condition Requirement Color Band
86–100 Very good Normal Maintenance Green
71–85 Good Normal Maintenance Blue
51–70 Fair Increase diagnostic testing Yellow
31–50 Poor Start planning process to replace or rebuild Orange
considering risk
0–30 Very poor Immediately assess risk Red

Tanasescu et al. [34] 2013 HI Condition Recommendations

0.85 < HI≤1.0 Very good Normal maintenance
0.7 < HI≤0.85 Good Normal maintenance
0.5 < HI≤0.7 Fair Increase diagnostic testing
0.3 < HI≤0.5 Poor Start planning process to repair or rebuild
HI < 0.3 Very poor Immediately assess risk; replace or rebuild
based on assessment

Ashkezari et al. [47] 2012 Health index Transformer health conditions

0 Excellent
1 Good
2 Fair
(continued on next page)
Renewable and Sustainable Energy Reviews 76 (2017) 687–700
 Table 5 (continued)

AUTHOR YEAR OUTPUT

A. Azmi et al.

3 Alarm
4 Further investigation or oil treatment

Haema and Phadungthin [39] 2012 %HI Condition Requirement Color band
86–100 Very good Normal maintenance Green
71–85 Good Normal maintenance Blue
51–70 Fair Increase diagnostic testing Yellow
31–50 Poor Start planning process to replace or rebuild Orange
considering risk
0–30 Very poor Immediately assess risk Red

Malik et al. [35] 2012 Tx condition Suggested course of action

range
8–10 (Good) Continue O & M without restriction; repeat this
condition assessment process as needed.
4–7 (Fair) Continue operation but re-evaluate O & M.
Consider using appropriate Tier 2 tests. Conduct
full time extension risk-economic assessment.
Repeat this condition assessment process as
needed.
0–3 (poor) Immediate evaluation including additional Tier 2
testing. Consultation with expert. Adjust O & M as
prudent. Begin replacement process.

Gorgan et al. [36] 2010 HI State Expected lifetime

0.85–1 Very good > 15 years

697
0.7–0.85 Good > 10 years
0.5–0.7 Satisfactory < 10years
0.3–0.5 Poor < 3 years
0–0.3 Very poor End-of-life

Jahromi et al. [21] 2009 Health index Condition Description Approximate expected lifetime
85–100 Very good Some aging or minor deterioration of a More than 15 years
limited number of components
70–85 Good Significant deterioration of some More than 10 years
components
50–70 Fair Widespread significant deterioration or Up to 10 years
serious deterioration of specific components
30–50 Poor Widespread serious deterioration Less than 3 years
0–30 Very poor Extensive serious deterioration At end-of-life

Ghazali et al. [37,41] 2009 THI Indication Recommended mitigating action

85≤THI≤100 Good Maintain normal frequency of current tier
test
55≤THI < 85 Fair Revise frequency of current tier tests to 6
months interval or as recommended by the
expert
10≤THI < 55 Poor Perform next tier tests (2nd priority)
THI < 10 Very poor Perform next tier tests (1st priority)

Naderian et al. [20] 2008 HI Condition Expected lifetime Requirements

85–100 Very good More than 15 years Normal maintenance
70–85 Good More than 10 years Normal maintenance
50–70 Fair From 3 to 10 years Increase diagnostic testing, possible
remedial work or replacement
(continued on next page)
Renewable and Sustainable Energy Reviews 76 (2017) 687–700
 Table 5 (continued)

AUTHOR YEAR OUTPUT

A. Azmi et al.

needed depending on criticality

30–50 Poor Less than 3 years Start planning process to replace or
rebuild considering risk and
consequence of failure
0–30 Very poor Near to the end of life Immediately assess risk; replace or
rebuild based on assessment

Anders et al. [48] 2006 Health index Condition Probability of failure (pof) Equivalent status on life curve Requirements
85–100 Very good Low First half of mean age Normal maintenance
70–85 Good Low but slightly increasing Second third of mean age Normal maintenance
50–70 Fair Rapidly increasing but lower than pof at Final third of mean age Increase diagnostic testing,
mean age possible remedial work or
replacement depending on
criticality
30–50 Poor Higher than pof at mean age and increasing First third after mean age Start planning process to replace
or rebuild considering risk and
consequences of failure
0–30 Very poor Very high, more than double the pof at mean Second third after mean age Immediately assess risk, replace or
age rebuild based on assessment

Hjartarson and Otal [49] 2006 Health index Condition Probability of failure (pof) Equivalent status on life curve Requirements
85–100 Very good Low First half of mean life expectancy Normal maintenance
70–85 Good Low but slightly increasing Second one-third of mean life Normal maintenance
expectancy
50–70 Fair Rapidly increasing but lower than pof at Final one-third of mean life Increase diagnostic testing,
mean age expectancy possible remedial work or

698
replacement depending on
criticality
30–50 Poor Higher than pof at mean age and increasing First one-third after the mean life Start planning process to replace
expectancy or rebuild considering risk and
consequences of failure
0–30 Very poor Very high, more than double the pof at mean Second one-third after the mean life Immediately assess risk, replace or
age expectancy rebuild based on assessment

Dominelli et al. [14] 2004 Name Rating Color Definition

Good 100–75 Green No noticeable deterioration or defects
Fair 74–50 Yellow Some deterioration or defects.
Function not affected
Poor 49–20 Orange Serious deterioration or defects in at
least some portions of the asset
Unsatisfied < 20 Red Extensive deterioration or defects. No
longer functions as designed.
Renewable and Sustainable Energy Reviews 76 (2017) 687–700
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

[11] Abiri-Jahromi A, Parvania M, Bouffard F, Fotuhi-Firuzabad M. A two-stage

framework for power transformer asset maintenance management – part II:
validation results. Power Syst IEEE Trans 2013;28(2):1404–14.
[12] Trappey AJC, Trappey CV, Ma L, Chang JCM. Intelligent engineering asset
management system for power transformer maintenance decision supports under
various operating conditions. Comput Ind Eng 2015;84:3–11.
[13] Brittes JLP, Nunes E, Jardini JA, Magrini LC, Kayano PSD., T & ERTTA, technical
& economical real time transformer assessment: an innovative approach on power
transformer life cycle management. In: Proceedings of the transmission &
distribution conference and exposition - Latin America (PES T & D-LA), 2014 IEEE
PES, p. 1–6. 2014.
[14] Dominelli N., Equipment Health rating of power transformers. In: Proceedings of
the conference record of the 2004 IEEE international symposium on electrical
insulation, no. September, p. 163–168. 2004.
[15] Boss P, Horst T, Lorin P, Pfammatter K, Fazlagic A, Perkins M. Life Assessment of
Power Transformers to Prepare a Rehabilitation Based on a Technical-Economical
Analysis. 2002.
[16] Suwnansri T., Asset Management of power transformer: optimization of operation
and maintenance costs. In: Proceedings of the international electrical engineering
Fig. 6. Maturity level of HI adoption [50]. congress (iEECON) international, p. 1–4. 2014.
[17] A2.34 WG. Guide for transformer maintenance. CIGRE, no. February, 2011.
[18] Wies RW, Johnson RA, Agrawal AN, Chubb TJ. Simulink model for economic
analysis and environmental impacts of a PV with diesel-battery system for remote
villages. IEEE Trans Power Syst 2005;20(2):692–700.
[19] Guo-Hua Q, Zheng R, Lei S, Bo Z, Jian-Gang X, Xiang-Ling Z., A new life cycle cost
model of power transformer and its comprehensive sensitivity analysis. In:
Proceedings of the power system technology (POWERCON) international confer-
ence on, p. 1342–1348. 2014.
[20] Naderian A, Cress S, Piercy R, Wang F, Service J., An approach to determine the
health index of power transformers. In: Proceedings of the international sympo-
soium on electrical insulation (ISEI 2008), p. 192–196. 2008.
[21] Jahromi AN, Piercy R, Cress S, Service JRR, Fan W. An approach to power
transformer asset management using health index. IEEE Electr Insul Mag
2009;I.25(2):20–34.
[22] Arshad M, Islam SM, Khaliq A, Power Transformer asset management. In:
Proceedings of the 2004 international conference on power system technology,
2004. PowerCon 2004, 2004, vol. 2, no. November, p. 1395–1398.
[23] Haema J, Phadungthin R., Development of condition evaluation for power
transformer maintenance. In: Proceedings of the 4th international conference on
Fig. 7. Breakdown maintenance cost [50]. power engineering, energy and electrical drives, no. May, p. 620–623. 2013.
[24] MS 60599:2013 – Mineral Oil-Impregnated Electrical Equipment in Service –
Guide to the Interpretation of Dissolved and Free Gases Analysis (IEC 60599:2007,
become an issue in obtaining the final HI. The subjective judgment of MOD). 2013.
weighting factor leads to different values of final HI. Therefore, the [25] IEEE Power & Energy Society. IEEE Std C57.104-2008: IEEE Guide for the
Interpretation of Gases Generated in Oil-Immersed Transformers. 2009.
work such as developing an algorithm for weighting factor could be [26] IEC 61198: Mineral Insulating Oils – Methods for the Determination of 2-Furfural
done to overcome the problem. and Related Compounds. 1994.
[27] Murugan R, Ramasamy R. Failure analysis of power transformer for effective
maintenance planning in electric utilities. Eng Fail Anal 2015;55:182–92.
Acknowledgements
[28] Satriyadi Hernanda IGN, Mulyana AC, Asfani DA, Negara IMY, Fahmi D.,
Application of health index method for transformer condition assessment. In:
The authors wish to thank Universiti Putra Malaysia for the Proceedings of the TENCON 2014–2014 IEEE region 10 conference, p. 1–6. 2014.
[29] En-wen L, Bin S., Transformer health status evaluation model based on multi-
financial support of this research under the PUTRA IPB Scheme (GP-
feature factors. In: Proceedings of the 2014 international conference on power
IPB/2014/9440804). The comments of the anonymous reviewers are system technology (POWERCON 2014), no. Powercon, p. 1417–1422.2014.
also appreciated. [30] Martins MA. Condition and risk assessment of power transformers: a general
approach to calculate a health index. Ciência Tecnol Dos Mater 2014;26(1):9–16.
[31] A AN;, Jaiswal GC, Ballal MS, Tutakne DR., Remote condition monitoring system
References for distribution transformer. In: Proceedings of the national power systems
conference (NPSC). 2014.
[1] Borunda M, Jaramillo OA, Reyes A, Ibargüengoytia PH. Bayesian networks in [32] Ashkezari AD, Ma H, Saha TK, Ekanayake C. Application of fuzzy support vector
renewable energy systems: a bibliographical survey. Renew Sustain Energy Rev machine for determining the health index of the insulation system of in-service
2016;62:32–45. power transformers. IEEE Trans Dielectr Electr Insul 2013;20(3):965–73.
[2] Rauf O, Wang S, Yuan P, Tan J. An overview of energy status and development in [33] Taengko K, Damrongkulkamjorn P., Risk assessment for power transformers in
Pakistan. Renew Sustain Energy Rev 2015;48:892–931. PEA substations using health index. In: Proceedings of the international conference
[3] Shaikh F, Ji Q, Fan Y. The diagnosis of an electricity crisis and alternative energy on electrical engineering/electronics, computer, telecommunications and
development in Pakistan. Renew Sustain Energy Rev 2015;52:1172–85. Information technology (ECTI-CON), p. 1–6. 2013.
[4] Karatayev M, Clarke ML. A review of current energy systems and green energy [34] Tanasescu G, Notingher PV, Dragomir O, Gorgan B, Voinescu L., Health index
potential in Kazakhstan. Renew Sustain Energy Rev 2016;55:491–504. calculation of electrical equipments using DiagConsole software. In: Proceedings of
[5] Ma H, Saha TK, Ekanayake C, Martin D. “smart transformer for smart grid- the 8th international symposium on advanced topics in electrical engineering, p. 1–
intelligent framework and techniques for power transformer asset management. 6. 2013.
IEEE Trans Smart Grid 2015;6(2):1026–34. [35] Malik H, Azeem A, Jarial RK., Application research based on modern-technology
[6] Arshad M, Islam Syed M, Khaliq A. Fuzzy logic approach in power transformers for transformer health index estimation. In: International multi-conference on
management and decision making. Dielectr Electr Insul IEEE Trans systems, signals and devices, SSD 2012 - summary proceedings. 2012.
2014;21(5):2343–54. [36] Gorgan B, Notingher PV, Badicu LV, Tanasescu G. Calculation of power transfor-
[7] Abu-Elanien AEB, Salama MMA. Asset management techniques for transformers. mers health indexes. Ann Univ Craiova Electr Eng Ser 2010(34):13–8.
Electr Power Syst Res 2010;80:456–64. [37] Yang Ghazali YZ, Talib MA, Rosli HAhmad,. TNB experience in condition
[8] Abu-Elanien AEB, Salama MMA, Ibrahim M. Determination of transformer health assessment and life management of distribution power transformers. In:
condition using artificial neural networks. In: Proceedings of the international Proceedings of the electricity distribution – part 1, 2009. CIRED 2009. 20th
symposium on innovations in intelligent systems and applications, p. 1–5. 2011. International conference and exhibition on, p. 1–4. 2009.
[9] Abu-Elanien AEB, Salama MMA, Ibrahim M. Calculation of a health index for oil- [38] Ashkezari AD, Ma H, Saha T, Cui Y. Investigation of feature selection techniques for
immersed transformers rated under 69 kV using fuzzy logic. IEEE Trans Power improving efficiency of power transformer condition assessment. IEEE Trans
Deliv 2012;27(4):2029–36. Dielectr Electr Insul 2014.
[10] Abiri-Jahromi A, Parvania M, Bouffard F, Fotuhi-Firuzabad M. A two-stage [39] Haema J, Phadungthin R. ,Condition assessment of the health index for power
framework for power transformer asset maintenance management – part I: models transformer. In: Proceedings of the power engineering and automation conference
and formulations. IEEE Trans Power Syst 2013;28(2):1395–403. (PEAM), 2012 IEEE, p. 1–4. 2012.

699
A. Azmi et al. Renewable and Sustainable Energy Reviews 76 (2017) 687–700

[40] WG A2.37 Report . Transformer reliability survey: interim report members. Electra [50] Deloitte and CEA. Asset Health Indices: A Utility Industry Necessity. 2014.
2012(261):46–9. [51] Eftekharnejad S, Vittal V, Heydt GT, Keel B, Loehr J. Impact of increased
[41] Yang Ghazali YZ, Talib MA, Rosli HAhmad., Condition assessment of power penetration of photovoltaic generation on power systems. Power Syst IEEE Trans
transformers in TNB distribution system and determination of transformer 2013;28(2):893–901.
condition index. In: Proceedings of the conference of the electric power supply [52] Enslin JHR. Network impacts of high penetration of photovoltaic solar power
industry (CEPSI). 2008. systems. IEEE PES Gen Meet 2010:1–5.
[42] Facilities Instructions, Standards and Techniques: Transformer Diagnostics (FIST [53] Liu H, Jin L, Le D, Chowdhury AA., Impact of high penetration of solar
3-31). 2003. photovoltaic generation on power system small signal stability. In: Proceedings of
[43] Zhou Y, Ma L, Yang J, Xia C., Entropy weight health index method of power the 2010 international conference on power system technology, p. 1–7. 2010.
transformer condition assessment. Proc. 2011 9th international conference reliab. [54] Ortega MJ, Hernández JC, García OG. Measurement and assessment of power
maintainab. saf., p. 426–431. 2011. quality characteristics for photovoltaic systems: harmonics, flicker, unbalance, and
[44] Scatiggio F, Pompili M,. Health index: the TERNA’s practical approach for slow voltage variations. Electr Power Syst Res 2013;96:23–35.
transformers fleet management. In: Proceedings of the electrical insulation [55] Hiranandani AK., Effects of harmonics on the current carrying capacity of insulated
conference (EIC), no. June, p. 178–182. 2013. power cables used in three phase electrical power distribution systems. In:
[45] Waugh NT, Muir DD., Improving the life cycle management of power transformers Proceedings of the 18th international conference on electricity distribution, no.
transforming data to life. In: Proceedings of the IEEE SoutheastCon 2015. 2015. June. 2005.
[46] Amir MDM, Muttalib ESA., Health index assessment of aged oil-filled ring main [56] Palmer JA, Degeneff RC, McKernan TM, Halleran TM. Determination of the effect
units. In: Proceedings of the power engineering and optimization conference of harmonics on pipe-type power cable AC/DC resistance ratios. IEEE Trans Magn
(PEOCO) IEEE 8th international, p. 347–351. 2014. 1993;29(2):1427–33.
[47] Ashkezari AD, Ma H, Ekanayake C, Saha TK. Multivariate analysis for correlations [57] Elmoudi A, Lehtonen M, Nordman H. Corrected winding eddy-current harmonic
among different transformer oil parameters to determine transformer health index. loss factor for transformers subject to nonsinusoidal load currents. 2005 IEEE Russ
IEEE Power Energy Soc Gen Meet 2012. Power Tech, Power 2005:1–6.
[48] Anders G, Otal S, Hjartarson T. Deriving Asset Probabilities of Failure: Effect of [58] Henderson RD, Rose PJ. Harmonics: the effects on power quality and transformers.
Condition and Maintenance Levels, p. 1–7. 2006. Ind Appl IEEE Trans 1994;30(3):528–32.
[49] Hjartarson T, Otal S., Predicting future asset condition based on current health [59] Elmoudi A;, Lehtonen M, Nordman H., Effect of harmonics on power transformers
index and maintenance level. Proceedings IEEE international conference transm. loss of life. In: Proceedings of the IEEE international symp. electr. insul. 2006.
distrib. constr. live line maintenance, ESMO, p. 1–8. 2006. 2006.

700