Laxmi Associate, India March 2017

Proposal for

TRANSFORMER FLEET
MANAGEMENT

Page 1 of 21
 Laxmi Associate, India March 2017

Forward

This proposal is prepared by M/s Laxmi Associates, Vadodara for Power Utility

Dr. Aradhana Ray

E-mail:rayaradhana@yahoo.com, Mobile No.09374076950
Laxmi Associates (Engineering consultancy & Service Provider)
Sun Enclave, Office No. 13-14-15, Near Suncity Paradise, Manjalpur, Vadodara-390 020
http://www.laxmiassociates.in/

COPYRIGHT © 2017 Laxmi Associates. India. ALL RIGHTS RESERVED

This document is protected by copyright vested in Laxmi Associates, India. No part of the document
may be reproduced or transmitted in any form by any means including, without limitation, electronic,
photocopying, recording or otherwise, without the prior written permission of Laxmi Associates. No
information embodied in the documents which is not already in the public domain shall be
communicated in any manner whatsoever to any third party without the prior written consent of
Sarawak.
Any breach of the above obligations may be restrained by legal proceedings seeking remedies including
injunctions, damages and costs.

Page 2 of 21
 Laxmi Associate, India March 2017

Table of Contents
1.0 Introduction

2.0 Transformer Health Index

3.0 Condition and Data Quality Indicators for Health Index estimation

4.0 Risk based Asset Management

5.0 Transformer failure and end of life definition

6.0 Transformer life time data statistical analysis

7.0 Case study on Statistical analysis of distribution transformer

failure

8.0 Case study on HI and Risk assessment of Power Transformer

Page 3 of 21
 Laxmi Associate, India March 2017

1.0 Introduction:

Transformers are considered the most crucial and expensive Asset within a
transmission & Distribution system. Most transmission systems currently
have large populations of aging transformers. With the growing demand for
electricity, the loading of transformers is increasing in the network.

The impact of a power transformer failure can be catastrophic. It would be

beneficial to know the risk status of the transformer population in order to
facilitate better asset management, optimisation of maintenance,
refurbishment and replacement strategies which will ensure maximum
asset utilisation and minimise system risk.

To meet these challenges, utilities are adopting asset management as

their framework for allocating capital and operation/maintenance budgets.
Simply stated, asset management consists of decision-making processes
that have the goal of deriving the most value from utility assets within the
available budget.

Among the asset management decisions that arise in dealing with

transformers are the following:
 Determining what level of maintenance to apply to a transformer
over its lifetime;
 Determining how frequently to inspect or test transformers and
what types of inspections or tests to perform (e.g., visual checks
for oil leaks, assessment of general condition [corrosion of
external metal work, tanks, pipe work, radiators, etc.], annual
infrared surveys, dissolved gas oil sampling, and others);
 Determining which transformers to repair and what repairs to
apply to individual transformer components, such as the radiator
or the pumps or the bushings;
 Determining whether and when to overhaul a transformer, thereby
rejuvenating it;
 Determining when to replace a transformer.

Of course, these issues are closely related, so that deciding, for instance,
to use a certain level of maintenance can affect the frequency of
inspections, tests, repairs, overhauls and replacements. Thus, balancing
Page 4 of 21
 Laxmi Associate, India March 2017

the many factors that influence the overall cost and reliability impacts of
transformer management strategies forms a complex, multi-faceted
decision problem.

This proposal describes a set of action plan for dealing with these
complexities. This framework is valuable for several reasons:
 It systematically and logically captures the interrelationships
among the factors that influence the cost effectiveness of
transformer management policies.
 It identifies the key information needed for making good
decisions.
 It provides an objective way to choose among decision
alternatives.
 It enables calculating the cost and performance consequences of
transformer management policies.

2.0 Transformer Health Index:

HI for each transformer is determined, based on the condition of the

Transformer. The HI is a quantification of condition measurements that are
taken on the transformer as mentioned below.

On-site Diagnostic Testing of the HV/EHV class of Transformers:

 Voltage Ratio and Vector group confirmation test,
 No-load excitation current measurement
 Capacitance and dissipation factor measurement of the inner insulation
 Capacitance and dissipation factor measurement of high-voltage bushings
 Insulation Resistance and polarization index
 Impedance measurements (short-circuit impedance, zero sequence
impedance, FRSL method)
 Analysis of winding damages with the FRA method
 OLTC operation and dynamic contact resistance measurement,
 Winding Resistance Measurement
 Moisture & Dielectric aging analysis of paper insulation by dielectric method
(PDC+ FDS) .
 Thermography , Noise , Vibration Measurement
 Routine Oil test, DGA, Furan
 DP test of paper insulation.
 Partial discharge measurements of power transformers

Page 5 of 21
 Laxmi Associate, India March 2017

HI is Systematic collection of available information to define asset condition

in terms of numbers as shown below:

To provide more clarity to health index number and its relevance a specific
condition is suggested to each number as follows:

HI Key word Condition Suggested action

10 Good All parameters are good Routine monitoring
9 Acceptable degradation No Maintenace
8 Average Moderate degradation Increase frequency
7 High degradation Trending
6 Poor High degradation closely watch Online monitoring & localize
fault
5 Minor criticality suspected Preventive measure
4 Bad Moderate criticality suspected Identify fault
3 Higher criticality found Quantify fault & Planned
shutdown for corrective action
2 Critical High criticality found Corrective measure
1 Very critical Replacement/ Refurbishment

Transformer condition indicator scoring is somewhat subjective, relying on

transformer condition. Weighting factors used in the HI assessment
methodology recognize that some condition indicators affect the
Transformer Health Index to a greater or lesser degree than other
indicators. These weighting factors were arrived at by consensus among
transformer design and maintenance personnel with extensive experience
and can be presented for fleet of transformer as shown below:

Page 6 of 21
 Laxmi Associate, India March 2017

Y0 HI Profile - Transformers
32
28
Number of Assets

24
20
16
12
8
4
0
0-1 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10
HI Category

Laxmi Associate has developed a proprietary health index algorithm for

transformer fleet based on various CIGRE and IEEE standards, which is
validated in various fleets in form of asset management project with Power
utilities in India and overseas and presented at several conferences.

over O&M ,
Rating Condition Plan all Dielectric Thermal Mehanical design REMARK
R-phse
bushing
16 MVA SUSPECT Test 5.975 3.8 5.1 7 7 monitoring

16 MVA OK 8.5 9 8 8 9

25MVA OK 8.625 9 8.5 9 8

25MVA OK 8.1 8.5 8.9 7 8

31.5MVA OK 8.45 8.6 8.2 9 8

Factory V-phase HV
53 MVA FAULT rebuild 1.875 1 1.5 3.6 1.4 winding Faulty

20MVA OK 8.525 8.8 7.8 9 8.5

OLTC Diverter
20MVA AGING Monitor 6.175 4.8 4.5 7 7.4 contact aging

25MVA OK 8.625 8 9 9.5 8

25MVA OK 8.45 8.4 8.2 9.2 8

16MVA OK 8.425 8.2 8.2 9.3 8

16MVA OK 8.575 8.9 8.3 9.1 8

Page 7 of 21
 Laxmi Associate, India March 2017

3.0 Condition and Data Quality Indicators for

Health Index estimation:
The initial Health index calculation is based on Tier 1 tests. The indicators are
expressed in numerical terms and are used to arrive at an overall Transformer
Health Index. Additional information regarding transformer condition may be
necessary to improve the accuracy and reliability of the Transformer Health
Index. Therefore, in addition to the Tier 1 condition indicators, this procedure
describes Tier 2 and Tier 3 tests and measurements that may be applied to
the Transformer Health Index, depending on the specific issue or problem
being addressed.

Tier 2 tests are performed routinely or when there is any indication from Tier
1 test results. Tier 3 tests are considered non-routine. However, if Tier 2 and
Tier 3 data are readily available, it may be used to supplement the Tier 1
assessment. Alternatively, Tier 2 and Tier 3 tests may be deliberately
performed to address Tier 1 findings. Results of the Tier 2 and Tier 3 analysis
may either increase or decrease the score of the Transformer Health Index
obtained from Tier 1. The Transformer Health Index score may indicate the
need for immediate corrective actions and/or follow-up testing. The
Transformer Health Index score is also suitable for use as an input to the risk-
and-economic analysis model.

However, an abnormal result of ANY Tier 1 test or measurement may be

adequate in itself to require immediate call for Tier 2 test to be conducted.
The following condition indicators are generally regarded for assessing
transformer condition:

Tier 1

Diagnostic Tests-No outage required

 Thermography , Noise , Vibration Measurement
 DP test of paper insulation.
 Partial discharge measurements of power
transformers
 Routine Analysis of insulating oil from Main
Tank
 DGA of insulating oil from Main Tank
 Furan content in insulating oil from main tank
 Routine analysis and DGA from oil of OLTC
Tank
 Operation and Maintenance Performance
including Age

Page 8 of 21
 Laxmi Associate, India March 2017

Tier 2 &3

Diagnostic Tests-Routine off line

 Voltage Ratio and Vector group confirmation test,
 No-load excitation current measurement
 Capacitance and dissipation factor measurement of the inner insulation
 Capacitance and dissipation factor measurement of high-voltage bushings
 Insulation Resistance and polarization index
 Impedance measurements (short-circuit impedance, zero sequence
impedance, FRSL method)
 Analysis of winding damages with the FRA method
 OLTC operation and dynamic contact resistance measurement,
 Winding Resistance Measurement
 Moisture & Dielectric aging analysis of paper insulation by dielectric method
(PDC+ FDS) .

Note: Test highlighted in YELLOW color is not applicable for distribution transformer.

4.0 Risk based Asset Management:

LA has developed an asset management tool for optimising asset

refurbishment/ replacement/retirement strategies.

The inputs to this model include Transformer Asset age, condition/HI,

probability of failure and consequences of failure. The output of the model
is the overall risk allowing analysis of cost/benefits of
replacement/refurbishment.

HI used to determine a probability of failure as a function of HI (P(HI)) as

indicated below.

Page 9 of 21
 Laxmi Associate, India March 2017

Statistical analysis on historical transformer failure data is therefore

required, in order to obtain a model to determine the probability of failure
of the transformers currently in service.

Bathtub curves, developed from Weibull distributions depict the probability

of failure of the plant as a function of age (P(age)).

If P(HI) > P(age), then P(age) is modified according to P(HI), i.e. the
bathtub curve becomes steeper, indicating an increased rate of aging, and
a new probability of failure is determined as a function of both age and HI
(P(age, HI)) & used for further analysis.

Page 10 of 21
 Laxmi Associate, India March 2017

5.0 Transformer Failure and End-of-life

Definition:
Transformer failure and the definition of transformer end-of-life are
illustrated in this section to indicate the failure of a general product.

A transformer is designed to withstand the expected growing load and

the system transient fault events. These transient fault events can be
catalogued as short-circuit
faults, lightning surge, switching surge and temporary over-voltages
coming out of the transformers. Initially a transformer does not fail as
its insulation withstand strength is higher than the normal operating or
the fault stress. As transformer ages, the insulation withstand strength
gradually reduces due to its normal degradation and the cumulative
effects from transient events; meanwhile transformer load increases
with age. Once the insulation withstand strength cannot sustain the
high operation stress, transformer fails. This process is illustrated in
Figure 1.

Figure 1 Transformer Failure Illustration

In Figure 1, the impulses in the actual stress curve represent the

sudden increased stresses from transient events. Such events occur
randomly during transformer operation and they may result in the
insulation withstand strength reduction. Each step change in the
insulation withstand curve shows a slightly reduction of insulation
strength. If the load increases, and/or a transient event occurs, then
the insulation withstand strength reduces. The crossover point of
insulation withstand curve and operation stress curve in Figure 2-2
indicates the expected operation lifetime of a transformer.

Page 11 of 21
 Laxmi Associate, India March 2017

If the load demand in Figure 1, is growing faster than expected, or the

transient events occur more frequently, or the fault stress exceeds the
insulation withstand strength, the transformer fails before the designed
age (Say 40 years). A transformer pre-mature failure before the
designed life, due to the significant effect from a transient event, is
shown in Figure 2.

Figure 2 Transformer Failure before Designed End-of-Life

The reality of transformer pre-mature and post-mature failure reveals

that the unique age of 40 years as transformer designed end-of-life
does not indicate transformer failure. In fact transformer failure would
depend on individual design, loading experience, maintenance scheme
and even the condition of the installation site.

Hence in order to precisely indicate transformer lifetime, transformer

failure needs to be well defined rather than being defined as the single
age of 40 years. Transformer failure is indicated by CIGRÉ as any
outage when the withstand strength is exceeded by operational
stress that requires the asset to be out of service, or in other
words, a transformer reaches its end-of-life when the unit does
not meet the operation requirements anymore.

Based on the above definition, transformer failure should be

distinguished from retirement. Hence ,End of life is based on remaining
life & not based on age
- Remaining life from loading profile, furan, moisture content or
statistical analysis
- Depends on acceptable failure probability (50 % end of life) as shown
below.

Page 12 of 21
 Laxmi Associate, India March 2017

Based on all above information a risk matrix can be derived for fleet of
transformer as shown in Figure below

Page 13 of 21
 Laxmi Associate, India March 2017

Risk Assessment

Probability Ranking
Severity
1 2 3 4 5
Ranking
Frequent Probable Occasional Remote Improbable

1 Catastrophic A A A A B

2 Critical A A B B C

3 Moderate A B B C C

4 Negligible A B C C C

Level of Risk Description of Risk

A High Risk

B Medium Risk

C Low Risk

Risk Based Maintenance Planning

Degree of Risk
Maintenance/Replacement Planning
(Composite Risk Index)

a. Immediately take out of service.

High risk b. Perform Offline Diagnostics & Invasive tests.
c. Load adjustment/ Relocating/
Replacement/Refurbishment may be required.
a. Condition Based Maintenance.
b. Both on-line condition monitoring & off-line
Medium risk
diagnostics will be required.
c. Monitoring frequency will be based on Health Index
and MTTF of the Asset.
a. Calendar Based Maintenance can be continued as
Low risk
per manufacturers’ guideline and regulatory
requirement.

Page 14 of 21
 Laxmi Associate, India March 2017

6.0 Transformer Lifetime Data Statistical

Analysis:

Since 1950’s and 1960’s biomedical studies suggested advanced analysis

methods; these methods were further developed by engineers in products’
design and manufacturing. Particularly in power systems, the lifetimes of
generator windings, engine fans, turbine wheels, cables, transformers and
insulation material strength are more and more concerned by statistics.

The common procedure according to which the product lifetimes are

analyzed statistically can be concluded in four steps as follows:

Step 1: collecting data;

Step 2: selecting the proper distribution model(s);
Step 3: fitting the data into the distribution model(s) and
determining the best-fitted parameter(s) by optimal approach;
Step 4: carrying out the goodness-of-fit test to check the
presumed distribution model(s).

The transformer population future failure trend in terms of hazard curve

against age can be further predicted according to the presumed distribution
and the obtained best-fitted parameters.

7.0 Case study on Statistical analysis of

distribution transformer failure:

Following table shows the failed transformers (t) during 2011-15 in

approx. 12000 nos. of distribution transformer fleet of a distribution
company.
Age (yr) Number of failed transformer
<1 35
2 23
3 24
4 19
5 22
6 21
7 21
8 12
9 17
Page 15 of 21
 Laxmi Associate, India March 2017

10 14
11 6
12 8
13 24
14 15
15 19
16 27
17 40
18 24
19 20
20 15
21 13
22 17
23 14
24 14
25 3
26 7
27 7
28 9
29 3
30 3
31 7
32 8
33 7
34 7
35 4
36 2
37 0
38 3
39 1
40 4

Age (Year)
Fig 3- Hazard function (left) and cumulative hazard function of distribution transformer

Page 16 of 21
 Laxmi Associate, India March 2017

The variation of hazard rate and cumulative hazard with age is shown in
Figure 3. The probability density function of above data is presented in
below Figure 4. Failures beyond 40 years (total 15 numbers) are not
considered for statistical analysis.

Age (year)

Fig 4- PDF (Probability Density Function) of distribution transformer failure

The derived distribution parameters (shape and scale) using two variables
of Weibull distribution function is estimated as 1.32 and 16.47 respectively.
The CDF (Cumulative Distribution Function) of Weibull distribution is shown
below:

Age (Year)
Fig 5- Variation of CDF (cumulative distribution function) with age

Age (year)

The estimated shape and scale parameter of two variable Weibull

distributions are 1.32 and 16.27 respectively. Shape parameter is an
indicative of rate of hazard. Higher shape parameter indicate higher hazard.
Whereas scale parameter (in year) estimated by Weibull is characteristic
Page 17 of 21
 Laxmi Associate, India March 2017

life of distribution transformer life. The observed mean life and median
lives are 15.1 and 15 years respectively. From above statistical analysis it
is evident that distribution transformer average service life span is around
15 year .

8.0 Case study on HI and Risk assessment of Power

Transformer:
Details of transformer, observed results of various diagnostic tests and
estimated health indices of twenty four generation transformers are
mentioned in table 1 and Figure 6 & 7. To correlate health index with
observed results these transformers are grouped in following three
categories:

Healthy:

Five transformers of S No 7, 8, 17, 18 and 21 are

having health indices (more than 7) 7.1, 8.6, 8.5, 7.5 and 9.5
respectively. Furan content of S No 7, 8 and 18 are not measured
resulting into lesser health index. DGA result of transformer of S No 7
indicates presence of electrical fault which was identified as defective
bushing. Bushing is replaced and transformer is now working
satisfactorily. These five transformers do not require any special
attention and normal monitoring shall be conducted.

Unhealthy:

Based on health index (less than 4.5), seven

transformers can be identified as unhealthy. Transformers of S no 2,
3, 13, 14, 15, 16 and 22 are having health index 4.2, 4.4, 3.7, 4.3,
4.3, 3.9 and 4.1 respectively. Among these seven transformers,
three are more than three decade old. S No 2 and 3 are of 38 years
and S No. 22 has completed 31 years of service. There health index
is lower due to normal aging. They are close to end of their expected
designed life and shall be planned for replacement. On the other
hand remaining four transformers namely S No 13, 14, 15 and 16 are

Page 18 of 21
 Laxmi Associate, India March 2017

24 years old but oil quality is poor and furan content is marginally on
higher side. These transformers need attention and after detailed
diagnostic tests appropriate corrective actions shall be planned.

Moderate:

Remaining twelve transformers are having health

index within 4.5 to 7. Specific attention is required to each
transformer. S No 5 and 6 indicates poor oil quality and moderately
high furan content. Whereas DGA result of S No 11 shows
development of internal fault. The other transformers are under
normal aging and routine diagnostics shall be conducted to monitor
rate.

Page 19 of 21
 Laxmi Associate, India March 2017

HI scores
10

4 HI scores

0
0 5 10 15 20 25 30 35

Figure 6: Heath Index distribution of GTs

Generation Txs - Age Profile

16
14
12
10
No of Tx

8
6
4
2
0
6-10 11-15 16-20 21-25 26-30 31-35 36-40
Age Bin

Figure 7: Age distribution of GT

Outcomes of fleet screening exercise may lead to review of condition

assessment methods/technologies/programs and also focusing of remedial
/corrective asset solutions on highly critical asset sub-groups or classes
within entire population.

Apart from providing more focused and prioritized asset strategy and
solutions, overall gaps and opportunities in fleet management will be made

Page 20 of 21
 Laxmi Associate, India March 2017

more transparent and understood for more conscious or integrated

development of holistic fleet management capabilities of critical assets.

• Standardizing the definitions of condition states, failure modes,

specifications
• Establishing useful long term data collection procedures by
standardized norms
• Calibrating test protocols and developing relationships between test
outcomes and asset condition .
• Conducting on-going failure cause analysis of selected equipment
recovered form the field
• Undertaking statistical analysis of the data and comparing with the
results of expert judgment.

Page 21 of 21