The Use of Health Index for Condition Assessment and Replacement Prioritizing

Presented By:
Claude Rajotte
Hydro-Quebec-TransEnergie Canada
Patrick Picher
IREQ, Hydro-Québec, Canada

TechCon 2017


Many utilities are facing a situation where assets are aging and the pressure to reduce operating costs is very large in a context where reliability is a priority. This paper shows how a large North American utility addresses this problem by using health indicators to prioritize maintenance and replacement strategy.

1. Introduction

Hydro-Quebec-TransEnergie operates the largest transmission network in North America and one of the largest in the world. Part of this transmission system is a fleet of almost 2300 power transformers and shunt reactors, including Generation Step-Up (GSU) units, transmission power transformers and transformers to feed the distribution network. The average age of this fleet is about 33 years and the age profile will create, as it has in several regions of the world, a need for intensive reinvestment in future years. To cope with this situation, Hydro-Quebec-TransEnergie developed an approach to use all information available to help to establishpriorities for reinvestment and maintenance activities.

2. Probability of failure or replacement

Aviation and Navy industries published several surveys concerning reliability patterns of assets as a function of age [1]. Six different reliability patterns are described in Figure 1 from these surveys and the associated proportions are indicated. These patterns are named, from top to bottom on Figure 1: bathtub, worst old, slow ageing, best new, linear and worst new. An interesting bit of information derived from these surveys is the fact that a minority of assets has a reliability pattern that shows a failure rate increasing as a function of age.

Figure 1 Reliability pattern from Aviation and Navy industries

The interpretation of such surveys should be done carefully. Aviation and military industries are particular and relatively different from other industries. For example, aircraft maintenance standards are very demanding and strictly respected which is not always the case in other industries. Moreover, despite the reliability profile shown in figure 1, aircraft users have to determine an “End of life point” as the “Run to failure” approach is obviously not an option for aircraft fleet management!

Poor condition and high consequences of failure are definitely the main drivers for “End of Life”. However, in the context of electrical utilities, several additional factors are affecting “End of Life”, in particular:

  • Inadequate apparatus rating (power, impedance, voltage level, ratio, etc.)
  • Inadequate apparatus availability
  • High maintenance and refurbishment costs and unavailability of replacement parts
  • Legislation, safety & environmental issues
  • Financial motivations – investment smoothing and reinvestment opportunity
  • Other: Obsolescence, orphans, low efficiency, etc.

Hydro-Quebec power transformer and shunt reactor historical inventory consists of close to 3000 units, including 800 scrapped units [2], replaced due to in-service failures or one of the reasons described above. By using the information from these scrapped units, in particular the age at the time of scrapping and a statistical model using AFT-Weibull analysis, it became possible to plot the failure/replacement rate as a function of age as shown in Figure 2.

Figure 2 FailureReplacement rate per year for transformers and shunt reactors

The curves clearly indicate the relationship between failure rate/replacement as a function of age even if there is no consensus about this relationship, in particular for transmission transformers [3]. The inventory also reveals that units in operation have an average age of about 33 years. Thus, it may also be said that a good way to maintain the fleet reliability at the current level would be to maintain the average age of the fleet as close as possible to the current average age.

Figure 3 shows the age profile of Hydro-Quebec’s power transformer and shunt reactor fleet. Considering that several apparatus have already reached or will reach their expected end of life in the near future, it was concluded that it would be impossible to maintain the average age of the fleet at its current level. Thus, it became evident that an approach considering apparatus Failure/replacement rate per year condition would be required to establish reinvestment priorities. Moreover, a strategy to smooth the level of reinvestment needed year after year also appeared necessary.

Figure 3 Hydro-Quebec-TransÉnergie transformer and shunt reactor age profile

3. Available information on condition

During the 2000’s, Hydro-Quebec-Trans Énergie made extensive efforts to optimize maintenance activities [4]. To do so, a compilation of corrective maintenance actions performed during a 5 year period was made and classified by failure mode. Figure 4 shows an example of corrective actions classification for a relatively homogeneous group of 228 transformers 120/25 kV. In such a diagram, called “Fishbone” or “Ishikawa”, failure mode and causes (components) are classified by order of importance. In the example shown, items are classified by costs and the number of events is also shown in parentheses.

Figure 4 Corrective maintenance actions classified by failure mode and main component

Once these diagrams are produced, a cost-effective maintenance strategy to address dominant failure modes can be developed. In figure 4, only costs of repair are indicated but in an RCM analysis, safety, environment and operation problems also have to be considered. Figure 5 shows a typical RCM decision process.

Figure 5 RCM Decision Process

The conclusion of this study revealed few differences between different transformer groups studied and, as a result, a uniform preventive maintenance program was established for all transformers. This program includes the following condition assessment actions:

  • DGA, water content and thermography yearly
  • Oil tests every 4 years
  • Inspection and electrical tests: every 6 to 8 years
    • Accessories inspections and functional tests
    • Capacitance and PF of bushings and windings
    • Core ground test
    • If OLTC present: Winding resistance and inspection of OLTC contacts and mechanism

4. Health Index Development

4.1 Purpose of using health index

It is very important to clearly define in advance the objectives of developing health indexes. For example, if the main purpose of development is to establish reinvestment priorities in the long term, it would not be a requirement to react for example very quickly to a sudden increase of dissolved gasses as other processes should already be in place to flag such events.

4.2 Review of the available data

A review of all relevant information should be performed. Data gathered by condition assessment actions represents the major source of information to develop health indexes. Nevertheless, there is also a lot of relevant information in the maintenance management system, in particular:

  • Detailed information about apparatus characteristics and its components for existing and scrapped units
  • Age at the time of apparatus scrapping (and in some cases, information about the reasons for scrapping)
  • Corrective maintenance actions tracking, as shown in Figure 4

In the process of selection of relevant information, some basic principles were defined. For example, it was decided to develop a health index indicator only if the information is available for at least 75% of the apparatus. It was also decided to consider the condition “Good” by default if the information is not available for a particular apparatus.

4.3 Selection of data

An exhaustive review of the information available was made to determine the parameter to be used. It appears that:

  • Several parameters are only available for a few transformers: ex: SFRA, DFR, etc.
  • Several parameters cannot be easily compiled: ex: winding resistance (OLTC)
  • Little information is available for some condition assessment actions: ex: visual inspection where the reporting is only made by exception

Table 2 Parameter retained by Hydro-Quebec-TransEnergie for health indexes

4.4 Data aggregation

Once the selection of data is completed, another important step of the development is to determine the method of aggregation to compile data in order to fulfill the main objective discussed in 4.1.

One interesting approach would be to reuse diagrams like Figure 4 to associate the condition assessment results to failure modes and probabilities as some authors have proposed [5]. Nevertheless, such an approach was not adopted because it appears to be very difficult to link bad test results with measurable failure probability as the current preventive maintenance program is using diagnostic techniques rather than prognosis techniques. It also seems important to keep age as a significant parameter, as the link between age and failure/replacement probability was demonstrated and also because age is seen as a good global health indicator that may compensate for the lack of precision and availability of all other parameters used.

It was thus decided to study the relationship between the selected health index parameters and age. To do so, it was necessary to find a way to aggregate all the health index parameters in a global health index. Weighted sum approach was selected to make the aggregation. In order to determine the weight of each index, it was decided to give more weight on indexes applied on irreversible conditions (ex: aged paper) and less weight for indexes for which condition can be improved by maintenance actions. Figure 3 describes the weighting factors applied for each indicator.

Figure 6 Health indexes as function of age

The red line of Figure 6 represents the average condition of apparatus as a function of age. All units above the red line are considered in worse condition compare to average and inversely, apparatus below the red line are in better condition compared to average. These observations lead to the creation of the apparent age concept.

Figure 7 shows the concept of apparent age. In the example of transformer “A”, the actual age is 23 but the condition is worse compared to the average. If a horizontal line is drawn as shown in Figure 7, transformer “A” apparent age is 44 years. Inversely, transformer “B” is 76 years old but its apparent age is younger at 62 years. For transformers far off the average, a procedure has been implemented to limit the correction (+15, -10 years) to avoid applying unrealistic apparent age.

Figure 7 Apparent age concept

5. Application of health index

5.1 Risk matrix Once apparent age has been determined for each apparatus, it becomes possible to reevaluate failure/replacement risks described in Figure 2 by using apparent age. Figure 8 shows the failure/replacement rate based on apparent age that is reported on the “X” (probability) axis of a 9by 9 risk matrix and where the “Y” axis represents the importance of the apparatus in the network and the consequence of the loss of this apparatus.

Figure 8 Risk matrix

All the data is available in a database and a list of priorities can be easily created by de development of a software application as shown in Figure 9. Such a list becomes a major input for power transformer and shunt reactor reinvestment strategy. Using such an application, it is also easy to identity transformers with specific problems (high DGA, aged paper, etc.) and visualization of its raw data is also easily possible.

Figure 9 Ranking of apparatus to establish reinvestment priorities

5.2 Corrective maintenance

Another interesting application of the apparent age concept is to use the difference between actual and apparent age to establish priorities to initiate corrective maintenance actions. Corrective actions like bushing replacement, on-load tap changer refurbishment, cooling system refurbishment, oil reclamation and gasket system replacement can be initiated. Cost analysis had also been integrated in the application in order to make the best decision between corrective maintenance and apparatus replacement.

5.3 Health index confidence level

Confidence level indexes were also developed to qualify each individual health index based specifically on the availability, recency and precision of the data. These individual health index confidence levels are then compiled to calculate a global health index confidence level. When the confidence level is considered insufficient, it may be possible to increase it by initiating condition assessment actions in order to obtain more recent or more accurate data to make a better decision.

6. Future work

Hydro-Quebec experience with the software application in the last five years has led to several improvements that were systematically implemented. The application is used in several departments (maintenance, expertise, management) and the main users are project planners.

Among the future improvements, it is planned to introduce in the software a way to manage special cases of apparatus that would require reinvestment but for which the health index indicators are not sensitive enough.

Moreover, as the data currently used are all off-line information, another improvement planned is the integration of the on-line continuous monitoring information. For example, more than 1000 transformers are equipped with DGA on-line monitoring sensors. The use of on-line monitoring information will have a direct effect on further improving health index levels of confidence.

Finally, authors of this paper are contributing to CIGRE working group WG A2.49 “Condition Assessment of Power Transformers” [6] that is planned to be completed in 2017 with a publication of a brochure. The recommendations of this WG will be considered to improve further, if possible, the existing process.

7. Conclusion

Hydro-Quebec-TransEnergie strategy for asset management of transformers and shunt reactors is to be pro-active by making preventive replacements in order to keep the average age as close as possible to the actual average age and to smooth the requested reinvestment.

In the mid-2000’s, an RCM program was established to optimize maintenance activities and to determine the best possible condition assessment actions. The combination of the data gathered from this condition assessment program combined with all relevant information contents in the maintenance management system was studied to determine the best way to develop health indexes.

The concept of apparent age, to combine age and condition information, was developed and used to feed a risk matrix to determine reinvestment priorities and corrective maintenance actions.

There are plenty of techniques that can be used in order to establish reinvestment priorities. It is nevertheless believed that with the use of the same data, all techniques would probably give similar results and ranking. Keeping in mind that this ranking is used as an input to the planning process, the main function of such an approach is to identify the most critical units.


Authors want to thank the following main collaborators from Hydro-Quebec for their contribution in the development of the approach described in this paper: Bernard Bérubé, Germain Bizier, Nicolas Di Gaetano, Bruno Girard, Stéphane Proulx and Carl Tardif.


[1] T. M. Allen. “U.S. Navy of Submarine Maintenance Data and the development of age and reliability profile”, Department of Defense, USA, 2001.

[2] P. Picher, J. Boudreau, A. Manga, C. Rajotte, C. Tardif, G. Bizier, N. d. Gaetano, D. Garon, B. Girard, J. Hamel and S. Proulx, “Use of Health Index and Reliability Data for Transformer Condition Assessment and Fleet Ranking“, Paper A2-101, Cigre Session, Paris, 2014.

[3] CIGRÉ WG A2.37, “Transformer Reliability Survey”, Brochure 642, Paris, 2015.

[4] C. Rajotte and A. Jolicoeur, “RCM Implementation in Hydro-Quebec Transmission System”, IEEE ESMO Conference, 2000.

[5] P. Lorin, L. Cheim, L. Pettersson, K. Gustafsson and E. TeNyenhuis, “Transformer Health Index and Probability of Failure based on Failure Mode Effects Analysis (FMEA) of a RCM Program”, CIGRE Paris Session, 2016.

[6] CIGRE SC A2 ToR for WG A2.49 “Condition Assessment of Power Transformer”, CIGRE SC A2 website http://a2.cigre.org.

Join our email list

We use cookies to give you the best online experience. By using this website you agree with our cookie policy.