By Michael Belanger
Eventually your transformer will have a defect. What are the parameters to follow? When and how should you react? These are the questions that this series of articles will try to answer.
In this series of four articles, we will cover the diagnosis of the industrial transformer as a whole. Without doubt, a subject this vast cannot be covered in depth within the framework of these articles. The aim is to discuss the measurement techniques and analyzes which are available.
In the first part we will discuss the failure rate of the components of the electrical distribution system. We will put into perspective, with the help of the failure rate, the relative place of each of the components of the electrical distribution system. We will also examine which parts of the transformer are most susceptible to failure.
In the second part, we will concentrate on the norms which serve as the guide for evaluating the data collected. We will discuss the life span of the transformer in light of statistics collected within the industry in the last three years.
The third part will allow us to focus on essential tests that should be performed and on their frequency. Because it is not economically viable to carry out all possible tests on a transformer, we will try to rationalize the available tests and will introduce the notion of a test tracker.
Finally, in the fourth part we will discuss the interpretation of the data and the methods used to detect the reduction of the life span or any condition for a failure.
PART 1: Statistical justification
FOR preventATive
maintenance
Introduction
Among other things, preventative maintenance has an undeniable effect on the reliability of the system. It is neither efficient, nor economical to inspect the equipment right down to the last bolt. In this section we will try to put into perspective the effects of preventive maintenance as well as look at the components which are the most susceptible to failure. For this, we will use statistics which were collected on existing equipment and taken from a number of surveys, conducted by the Institute of Electrical and Electronic Engineers (IEEE)[1], DOBLE and ABB[7].
The first survey, conducted by IEEE [1], asked personnel with responsibility for maintenance of electrical equipment, for information on the reliability of their transformers. A total of 3787 transformers, with an accumulated 37692 years of operation, were part of the study. While we don't have similar information on the survey size from DOBLE or ABB, we believe that their statistical results also come from a large database.
Reliability of equipment and quality of
maintenance
The survey conducted by IEEE covered the quality of preventative maintenance and revealed that a company with a "poor" preventive maintenance program has a failure rate three times higher than a company having an "excellent" maintenance system.
According to IEEE, a poor maintenance program signifies that an irregular evaluation is executed on a component which is selected in an arbitrary fashion. An excellent maintenance program signifies that a regular evaluation is executed on all the equipment. The evaluation could take the form of a visual inspection or tests carried out at specific time intervals.
Table 2 was developed data from all the respondents to the IEEE survey. The failure rate of the transformers is similar to that of cables and switchgear. In terms of frequency of failure, the transformer is classed among the elements having a relatively high rate of failure, in the electrical power distribution network. The failure rate of an oil transformer is 0.0062 failure/unit year.
A failure rate of 0.0062 signifies that a transformer will have a failure within the next 160 years. Consequently, within a group of 10 transformers, 1 of these will have a problem within the next 16 years. So, the failure rate increases in proportion to the number of transformers. The reader should take into account the component location and how it will affect the downstream electrical distribution when comparing reliability data. More precisely, how will the plant be affected if a particular piece of equipment fails? Will it shut down a critical plant process?
When we take a failure rate and multiply it by the downtime per failure, we obtain the average downtime per year. The data has been compiled in Table 3. This table demonstrates the dependence of the electrical distribution vis-à-vis each of the major components. What is interesting in this table is that it allows for the balancing of the maintenance effort between each component. This table also shows that the replacement of a transformer has a downtime four times less than repairing a transformer. We are talking here of a major repair such as a rewind.
The failure rate of a transformer in the range of 300 kVA to 10 MVA is less than one half compared to other transformers of a capacity higher than 10 MVA.
Table 4 indicates that the transformers of an industrial class of 600 volts to 15 kV, have a better reliability rate. The category of transformer more than 10 MVA operating at a higher voltage than 15 kV, has a three times greater chance of failure than that of the class previously sited (item 1 of Table 4).
The next few tables show us the origin of the fault. Table 5 compiles the information of all transformer types [2][3], whereas Table 6 illustrates transformers with On Load Tap Changers [7] (OLTC).
First, let us look at the case where no distinction is made between tap changer types.
Table 5 indicates that a strong proportion of failures is related to insulation. The failure of insulation is often caused by the substantial reduction of the mechanical strength of the cellulose. with water being one of the principal agents of deterioration. In this condition, the insulation system does not have the necessary capacity to support the stress which is imposed upon it. This will eventually lead to an irreversible breakdown.
The second source of failure is related to damage created by external short-circuits (see Table 5). About 8 per cent of failures are related to a defect in the protection system. This could be prevented by carrying out a complete verification of the protection system every 5 to 10 years, depending on operating conditions and the immediate environment of the transformer protection network.
The third cause of breakdown is due to failure of the tap changer Ń both onload tap changers and offload tap changers. Our experience shows that offload tap changers need to be repaired too often. We believe that their current capacity or their magnetizing current withstand is too marginal and that the offload tap changer should be specified with an overload current capacity of 125 per cent.
Transformers equipped with OLTC have a different failure distribution. The failure ratings place the tap changer in the first position. Tap changer failures are dominated by faults of a mechanical nature (spring, bearing, shaft, drive mechanism), followed by default of an electrical nature such as choking of contact, burning of transition resistors and insulation problems.
Comparing Tables 5 and 6, we can understand the importance of considering the transformer's components when establishing inspection priorities.
Table 7 indicates that the majority of problems come from a manufacturing defect or from inadequate maintenance. This is where the justification for careful inspection and testing during start-up and later preventative maintenance comes in.
Table 8 illustrates that 20 per cent of transformer failure removals are done manually (the operator de-energized the transformer). This indicates that the use of a prevention program for tracking equipment problems allows for equipment withdrawal from the circuit where the condition is judged doubtful - avoiding unscheduled downtime.
We have noted that the periods most susceptible to failure are those that are related to equipment manufacturing, transport and installation. These problems manifest themselves within the first years after installation. The failures relative to aging start to appear around the twenty-fifth year, according to the surveys.
Based on a failure curve developed from the preceeding information, two periods of intense observation have to be implemented. These tracking periods follow premature failure and random failure, as well as when insulation begins aging.
The first tracking period follows problems related to premature failure experienced in the first year of utilization or following a major repair. The tracking starts when equipment is put into service, followed by consistent observation and a tight sampling campaign. This sampling rate is about two to three times the normal frequency since the probability of failure is higher during that period.
The second tracking period is related to random failures or insulation aging. A large percentage of breakdowns of transformers are related to a problem in the insulation system. These problems, generally detectable, are a consequence of a condition that evolved during a certain period. Since the principal element of failure is the insulation system, the latter should receive the biggest part of the maintenance effort. However, the other components should be inspected, but at less frequenct intervals. If the transformer is equipped with an OLTC, then greater maintenance effort should take place on the tap changer.
Besides considering these statistics, we must keep abreast of any changes in the transformer market which could affect the reliability of the components. Some things to consider here are new product developments, changes to manufacturing techniques, and the effects of power quality on the equipment. This occurrence reinforces the principle of putting in place a preventive maintenance system that adapts to the reality of the market. To do this, we will need to put in place an information system accessible to the industrial user, which will allow failure causes to be put into perspective.
This information system will be continually updated by the users. This update can be done from a database software specialized for transformers [5] or any other means judged efficient. From the database, new tendencies could be revealed allowing us to adjust our maintenance efforts or to modify future purchasing specifications.
We have two examples which justify putting into place an information system.
Examples
In the first example, the offload tap changer at a certain facility has become an increasing source of faults in transformers which are less than 5 MVA. After reviewing the information, it was the authorÕs opinion that the design of the contact point and its associated mechanism was not sturdy enough to withstand the repetitive surges from the magnetizing current. Moreover, the current capacity was sometimes marginal. This weakness became evident when a certain number of 15kV and 25kV electric boilers were installed during a provincial surplus energy program which was put into effect in the early 1980s.
The energy program lasted a few years. After one or two years in operation, a large portion of the transformers, specified in the context of this program, had develop hot spot problems in their offload tap changers. The application of the boiler transformers at 15 kV/ 25 kV were too demanding on the magnetization current and the short-circuit current. The frequent energization as well as the number of short-circuits of the boiler were damaging the contact points of the tap changers. A large number of these transformers had to be repaired.
The tap changers are still built and specified in the same way and we believe that these are not suited to offer a life span of fifty years. At purchase, specify a tap changer with an overload current capacity of 125 per cent. This overload capacity will diminish the damage to contacts caused by overheating. For a sealed transformer, it is suggested you use a terminal board located at the top inside the tank. A gasketed access panel allows you to make the proper connection change.
The second example involves the analysis of dissolved gases in electrical insulating oil, allowing for the detection of a latent fault within the tank of a transformer. These gases are produced by the degradation of the insulation materials under thermal and electrical stresses. The nature and the variation of the amount of each individual component in the dielectric fluid is used to determine the presence of abnormal (incipient) faults in the equipment. Some times these changes take years before they are recognized as being problematic, and thus reduce our ability to recognize trouble in a transformer.
A data base of dissolved gas testing results would allow for the recognition of these tendencies easily and quickly, because it allows certain trends to become obvious, which are not easily perceptible when the information is viewed in an isolated way. Once this information is available, we can act accordingly. Having an information system in place would be advantageous to the industrial user.
Conclusion
References
[1]IEEE Recommended Practice for the Design of
Reliable Industrial and Commercial Power System, IEEE Std
493-1990.
[2] Annual survey by Doble Engineering
Company.
[3] Kelly J.J., A Guide to Transformer Maintenance
Transformer Maintenance Institute, SD Myers, 1981.
[4] Report of
Transformer Reliability Survey - Industrial Plants and Commercial Buildings,
James W. Aquilino, IPSD 80-7.
[5] PERCEPTION 4.0, Software for
tracking and diagnosis of oil insulated equipment, SEIDEL,
www.seidel.qc.ca
[6] Diagnostic du transformateur. Recherche du
dysfonctionnement, Mˇthodologie de lÕentretien, Expertise. Michel Bˇlanger, mars
1999.
[7] Status and Trends in Transformer Monitoring. C. Bengtsson,
ABB Transformer, Ludvika, Sweden, IEEE transaction on Power Delivery, Vol. 11,
No. 3, July 1996.
[8] R. Sahu, Using Transformer Failure Data to Set
Spare Equipment Inventories, 1980.
Michael Belanger is with SEIDEL inc. He is a consultant in power system engineering and a transformer diagnosis software designer.
ET