21, rue d’Artois, F-75008 PARIS http: //www.cigre.org A2-204 CIGRE 2008 TRANSFORMER REFURBISHMENT POLICY AT RTE CONDITIONED BY THE RESIDUAL LIFETIME ASSESSMENT. R. BLANC, G. BUFFIERE, J-P. TAISNE* RTE / CNER France P. LONG, E. MOUTIN Transformateurs Services Vénissieux France A. TANGUY, P. GUUINIC EDF / R&D France F. DEVAUX AREVA / T&D France SUMMARY On RTE's 225 kV and 400 kV networks in France, 22% of the 1250 power transformers in use are 40 years old or more. The feedback shows a satisfactory behaviour of this fleet with a rather small failure rate. This feedback shows also that a part of these transformers' serious damages is due to failures of the main equipments (on-load-tap changers and bushings). To extend the lifetime of these old transformers which are in operation, RTE has decided to put in place a new refurbishment policy which mainly concerns the equipment and dielectric oil. This report first describes all of the criteria retained to select the transformers to be refurbished, the evaluation of the residual lifetime of the active part and, finally, the economic criteria used in the decision-making process. The difficulties encountered when evaluating the residual lifetime based on the analysis of the furanic derivatives are illustrated through the studies carried out on a 100 MVA, 225 kV transformer which was in use at the Laval substation, so as to establish an accurate mapping of the ageing of cellulose insulation as well as to estimate its residual mechanical resistance to short circuits. Finally, this report mentions current studies and research in order to better characterise the ageing models of network transformers which are not, on average, heavily loaded. In this case, the goal is to extend the lifetime of the active part through targeted maintenance operations on the oil in particular. These models must be validated through measurements made on transformers and the Laval experiment results will be utilized in the process. KEYWORDS Power transformer - Behaviour - Refurbishment - Diagnosis - Lifetime - Furan - Ageing - Models. [email protected] Table 1: serious failures according to their origin Failures associated with the control panel. The aim of this study. as Failure origin Yearly failure rate failures (%) well as faults which occurred at the Active part 45 (38) 0. the 118 serious failures are shown according to the source of the failure and in line with the transformer's component parts.1. On January 1 2005. When the number of failures encountered is low. 13 % failures include the failures of the bushing Other 4 (3) 0. The breakdown of the fleet according to type and age is shown in figure 1 below. Therefore we define. the bushings.23 % connections are counted as on-load-tap Tap changer 45 (38) 0. which confirmed the good performance of the fleet. 'bushing' Bushing 24 (21) 0. Two-thirds of the RTE's fleet of power transformers is made up of High Voltage 225 kV winding transformers. the on-load-tap changers and its control panel.23 % changer failures. namely. this fleet was estimated at 1259 transformers with an average age of 27 years. for example. In table 1. 40% of the transformers were older than 30 years. to a small number of transformers with the same age range being in use.2004 and which were categorised according to the age of the transformer at the time of the failure and compared to the total transformer fleet with the same age over the period in question. the upper (λupper) and lower values (λlower) of the failure rate for which there is a 90% chance of the actual rate being situated between these two values. 2 . on the basis of the χ2 law. 90 and 63 kV winding transformers. the estimated rate is no longer relevant. When these upper and lower values tend to diverge. Likewise. The analysis differentiated between serious failures characterised by an outage time of more than eight days as well incidents with forced outage. was to provide the information required to draw up a refurbishment policy for the power transformers. this gives an average failure rate of 0. a behavioural study was carried out on RTE's transformer fleet. the cabinets. due. This study targeted each homogeneous group of transformers by specifying the behaviour of the active part and the main equipment parts. and the protection and control relays. the calculated failure rate is no longer representative on a statistical level. Type Three Phase Autotransformers Single phase Autotransformers Three phase Transformers Single Phase Transformers Total Number 208 6 960 85 1259 Figure 1: Distribution of RTE transformer fleet according to type and age Serious failures: The method used to determine the failure rate λi for age i consisted of counting the failures which occurred over the period 1990 .61 % occur between its end and the tank. the cooling system. Number of 'selector' and 'diverter switch' subsets. There were 118 serious failures over the period in question.6% per year.02 % itself as well as the internal flashover which Total 118 (100) 0. 23% of the fleet comprises High Voltage 400 kV winding transformers and the remainder is made up of High Voltage 150. Overview of the behaviour of the RTE transformer fleet Between 1990 and 2004 (inclusive). This drop is most likely due to the undertaking of work on the transformers when a switch to increase power was performed or when the condition of a piece of equipment no longer permitted its repair. In addition to the serious failures. This rate also heavily depends on the age of the transformers. it is above 1% for those which are over 30 years old. This can partly be put down to the fact that the number of transformers aged over 45 years is far smaller as well as poorer information gathering for transformers which are considered as old. While the forced outage anomaly rate remains under 4% between 7 and 25 years. these averages may conceal quite different forms of behaviour depending on the group of transformers considered and the equipment generation.2% in the 5-15 year range. which are counted in this forced outage anomaly rate. The initial result of this overview is that 60% of serious failures which generally lead to the transformer being sent for repair or to be scrapped are associated with the transformer's equipment and that the probability of these failures increases significantly after 30 years. incidents to do with cooling. Similar forms of behaviour are seen when focusing on failures associated with the various components and. the λupper and λlower failure rates grow too far apart to be representative of the estimated failure rate. Forced outage anomalies: The annual average forced outage anomaly rate measured over 1990 2004 is 4. in which the forced outage anomaly sliding rate over five years is shown. as shown in figure 3. There is a drop in the anomaly rate between 32 and 37 years.6%. Figure 2: Annual rate of serious failures according to the age of the transformers While the global failure rate remains below 0. 3 . The age of the transformers is also a determinant factor. as shown in figure 2 which specifies the rate of serious failures according to the age of the transformer. in particular. the control panels as well as the other pieces of equipment are also included. it then starts to grow and exceeds 10% after 45 years. Beyond 45 years.Nevertheless. as it was also the case for the serious failure rate. the on-load-tap changers whose failure rate increases significantly after 30 years. Furthermore.as well as a diagnosis of the active part to estimate its residual lifetime. Subsequently. safety and environmental conditions. so as to reduce the number of expert inspections in the first few years. as costs would be far higher. 1st line of improvements: Implementation of a refurbishment policy The average age of the RTE transformer fleet was 27 years in 2005 and it increases due to quite a low renewal rate and a relatively mature network in terms of development. we estimate around 15 expert inspections over the first few years will take place every year and 60% of these inspections could result in refurbishment. Carrying out the work while ensuring the replaced pieces of equipment are correctly suited to the transformer. its critical importance for the network as well as works scheduled on the cell's other equipment. which requires transportation to a workshop. criteria of a hierarchical nature were defined on the basis of the transformer's past history. The main criteria are the age of the transformer (35 years for 400 kV and 40 years for 225 kV) and the guarantee that the transformer is going to remain in its substation for more than ten years. Establishing a diagnosis of the equipment . Assessing the financial viability of performing all or part of the works. Nevertheless. Finally. 4 . a far larger number of expert inspections are envisaged in accordance with the initial results obtained.Figure 3: Forced outage anomaly rate according to the age of the transformers 2.to compile the works list and estimate the cost of the works . The principles adopted by RTE for the implementation of the refurbishment policy consist in: Selecting the transformers which may be refurbished. Likewise. The behavioural study reveals that the serious failure rate may be significantly reduced if the main equipment parts (on-load-tap changers and bushings) are repaired on a preventive basis before the failure of one of these parts leads to a serious failure of the active part. it is worth making sure that the transformer's residual lifetime justifies the cost of the refurbishment. if the active part of the transformer shows early signs of impending failure. the transformer requires repair rather than preventive maintenance. With these criteria and on the basis of the existing fleet. Renewing the other equipment parts at this stage would also serve to reduce forced outages beyond 40 years of age. refurbishment does not currently consider the active part. This made RTE to consider how to extend the lifetime of this infrastructure while maintaining acceptable availability. Then a measurement of the losses and of the tg δ is performed by studying in particular their variation in line with the test voltage. Methods to diagnose these contact elements on the basis of electrical and acoustic measurements are in the process of being assessed. oil retrofilling operations were suspended following several cases. which are yet to be clearly explained. However. measurement of the no-load current which may reveal any insulation degradation affecting the magnetic circuit sheets. Bushings: The diagnosis of the bushings starts with a visual examination. the manufacturer. etc. which represents a dielectric fault in the insulation. a significant and prolonged overload can give rise to concentration levels which are characteristic of a thermal effect over several months. oil processing is recommended according to dielectric rigidity measurements. . the age and gastight characteristics of the on-load-tap changer's diverter switch. Dissolved gas thresholds and ratios specified in international standards have been supplemented by RTE. These measurements.the forming of sludge due to the acidity of the oil which leads to higher thermal stress. which may work loose. let us move on to the next stage.3. this may reveal leaks whose seriousness will be assessed according to the nature of the porcelain envelop (single part or multipart). we prefer to analyse the dissolved gases rather than an electric measurement to detect any latent faults. Dielectric oil: The main failure modes which concern oil are given below: . oil temperature. In addition to this dissolved gas analysis. if they are comparable to the values from the transformer's factory tests. Another area of attention concerns the selector's contact elements which are not visible but which can generate flah over through thermal run-away in the event of carbonization. When the bushings are not fitted with capacitance taps. the following 1st level electrical measurements are also carried out: measurement of the insulation of the windings in relation to the mass. which led to partial discharges rise and hydrogen production. The diagnosis is based quite simply on the absence of alarms in previous months (Buchholz 1st stage. attention must be given to the previous analyses to ascertain whether they reveal a trend or if they are the result of a specific situation which affected the transformer.flashover in large volumes of oil due to the nature and content level of particles. water content in oil. Diagnosis of the condition of the active part and the main equipments 3. This meant lower thresholds than the maximum thresholds indicated in standard IEC 61 464 could be chosen. For example. On-load-tap changers: The preventive maintenance of the diverter switch foresees the replacement of the normal wear parts. The thresholds retained to replace a bushing or to categorise it as 'to be watched' were defined on the basis of measurement campaigns. A set of 'normal' gas concentration levels encountered in RTE's transformer fleet was established in accordance with the voltage level. the on-load-tap changers and the dielectric oil. These oil processing are preferable to a change of the oil load. .dielectric flashover due to a too high moisture level in the oil.2 Diagnosis of the main pieces of equipment The main pieces of equipment which can result in the transformer's serious failure are the bushings. after 30 years. The diverter switch is therefore sent to a specialised workshop for the mechanical section to be checked and repaired. if necessary.1 Diagnosis of the active part The first inspection consists in checking that the transformer's active part does not suffer from an incipient fault. To avoid these faults.) and on gas-in-oil analysis. acidity measurements as well as the number and nature of particles. If concentration levels exceed these normal thresholds. short circuit impedance measurement to characterise the geometric deformation of the windings. 3. Concentration levels are defined as 'normal' when 90% of the levels of the sub-group in question are not exceeded. In fact. attention should be given to the mechanical section of the tap changer's head. measurement of the turn ratio. 5 . 6 . due to the low displacement values and changes to the insulation distances.flash over due to static electrification phenomena on 'solid-liquid' interfaces. which is generally greater than 10. Residual lifetime assessment 4. .formation of copper sulphides which create local short circuits between conductors. Most users consider the detection of mechanical faults as the main purpose of FRA method as these faults are very hard to detect with other methods. 4. we do not look at the equipment's early failures and nor do we consider cases in which the operating constraints exceeded over time the specifications used for the original design. We usually consider the CO2/CO ratio as an indicator of the paper's thermal decomposition when this ratio.The following failure modes which also involve the paper and metallic catalysts which are present in the active part have not yet been found on RTE transformers and are not being specifically monitored under its refurbishment policy. The risk is associated with the loss of the paper's mechanical properties following the scission of cellulose chains in accordance with the chemical interactions between the paper and the oil which are subject to thermal stress. an indirect measurement is taken. the FRA technique can detect the following types: .Buckling of winding (significant displacement in the radial direction). It is important to have an estimation of the residual tightening of the windings as this significantly conditions the mechanical withstand of the transformer at the time of a short circuit. However. The ageing considered below therefore corresponds to a slow kinetic mechanism which mainly affects 'paper-oil' insulation″. Use of the FRA technique could be considered. . Among these faults. particularly in the frequency range which is under 1 MHz (range considered as interpretable for this type of test). we could also note that the concentration in [CO + CO2] also provided a good indication of the presence of furans when it exceeded the threshold of 300 μmol/l. is below 5. 3. The main criterion to assess the age of the paper consists of characterising their degree of polymerisation (DP). . as it is impossible to sample paper to carry out DP measurements from a transformer which is in use.Seriously damaged winding. The simplest measurement consists of analysing CO and CO2 concentration levels in the oil which are a consequence of the paper's decomposition.1 Paper ageing evaluation In this section. the electrical parameters (inductances and internal capacitances) will only vary very slightly. particularly at the windings' hot spots. The effect will therefore be difficult to interpret on FRA measurements. It is generally accepted that paper no longer has any mechanical resistance when the DP reaches 200. However. Nevertheless. On the basis of RTE's oil analysis database. Rupture of the cellulose insulation barrier caused by mechanical tension and compression stress at the time of a near fault can therefore turn into a serious failure through turn to turn short circuits. Other methods are therefore being assessed as this parameter is essential to ensure the transformer's lifetime. .Displacement of winding (significant displacement in the axial direction). particularly for CO2 in transformers which breathe freely. CO and CO2 may originate other than through ageing.3 Mechanical assessment of windings It is important to have information on the mechanical condition of the windings. particularly in the event of rising network short circuit power which results in mechanical stress levels which are close to the windings' maximum critical withstand levels. Work published by CIGRE [1] and carried out in a laboratory on sealed cells also revealed that the change in CO + CO2 dissolved gases was linear with the concentration in 2-FAL. It is therefore a global criterion which is mainly used to trigger a HPLC analysis following a DGA (dissolved gas analysis) carried out during regular maintenance at RTE. 4. Nevertheless. namely. which is not particularly compatible with oxidation. and the effect of the temperature on such partitioning. each of the three phenomena has a different activation energy (Ea). particularly due to the slowness of low temperature kinetics. RTE evaluates the degree of polymerisation (DP) according to the formula below in which the [2-FAL] concentration in the oil is expressed as mg/kg: DP = 500 – 333 x log [2-FAL] This formula. The formula currently recommended to describe chemical kinetics is the following [2]: 1 1 − = A× e DPt DP0 − Ea RT ×t A is the ageing coefficient (prefactor) Ea is the activation energy (kJ per mole) R is the constant of pure gases (8.01. acids with a low molecular weight would be more harmful [3]. namely.To estimate the degree of polymerisation of the paper.314 J/mole/°K) T is the temperature (°K) t is time (sec) In the specific case of RTE fleet transformers. which can only apply to [2-FAL] concentration levels of between 0. These tests revealed an exponential difference of the ageing coefficient A in line with the amount of water in the paper (mass ratio) as shown in figure 4 below. the adopting of a single formula was considered sufficient for the entire fleet. the ratio between the mass of cellulose insulation and that of the oil should be taken into consideration and thereby groups could be formed in accordance with the technology. The best known formula which is used in the IEC Loading guide is the doubling of the speed of ageing for every 6K increase with a reference period set at 98ºC for normal paper. 7 . Best practice presented by CIGRE WG indicates that prefactor AH relating to hydrolysis depends on moisture and acidity. laboratory tests were performed on sealed test tubes while varying the initial water content in the paper. RTE uses the only ageing indicator which is currently available and widely used. hydrolysis (130>T>70°C) and oxidation (70°C>T). The devices used to rapidly study chemical kinetics in a laboratory are often based on acceleration by increasing the temperature. Recent work by GT CIGRE TF D1. in a first step. hydrolysis and oxidation may be considered. winding voltages and the design to define a specific formula for them. corresponds to an average transformer defined on the basis of all of the RTE fleet's oil analyses. furan measurement through HPLC. The rupture of cellulose chain phenomenon has been widely studied. the partition of furans between the paper and the oil. If the oil were to have the same acidity level. Oxidation of the cellulose is not so well-known. to only consider the ageing coefficient associated with hydrolysis when characterising the lifetime of Kraft paper immersed in oil. RTE has decided. Certain recent publications reveal that the nature of the acids must be taken into consideration to estimate the precise value of prefactor AH. In fact.2 Assessment of the residual lifetime The paper-oil insulation ageing mechanisms are known to feature three chemical phenomena which interact among themselves and which have preferential activation areas in terms of temperature: pyrolisis (T>150°C). (Coefficient A was multiplied by 108 in the figure).10 recalls that the IEC formula is a simplification. To be more accurate. in view of the other factors of uncertainty.2 and 15 mg/kg. To quantify the effect of water content on this paper ageing factor. This research was supplemented with tank removal for a visual inspection of the active part. the cost of the refurbishment work is below 80% of the expected gain which corresponds to the investment made.3 Financial criterion in the refurbishment decision-making process On the basis of the diagnosis performed on the equipment (see § 3) and the financial analysis of the cost of the work needed to upgrade this equipment.1). Lessons from the LAVAL transformer case The case of the 100 MVA 225/93kV transformer in LAVAL. 8 . 5.5 ppm led to diagnosing a DP of less than 200 and therefore the premature ageing of cellulose insulation although this content level was not coherent with the CO2/CO ratio. RTE defined the following charts which let one estimate the residual lifetime of the active part by determining the average degree of polymerisation and the moisture contained in the papers. Figure 5 shows an example of chart giving the decrease of the polymerisation degree with age at 80°C temperature according to the initial water content in the paper. in so far as the [CO + CO2] concentration was high (386 μmol/l). mentioned at the CIGRE SC A2/D1 symposium in Bruges in 2007 [4]. clearly illustrates the issues that an operator must address when determining the degree of polymerisation on the basis of furans. The measurement of a concentration of 2-furfural of 8. Let us keep in mind that these charts are based only on the hydrolysis model for the time being. which was normal (7. Direct measurements of the degree of polymerisation (table 2) on the first and twelfth disks as well as on the pressboard insulating cylinder proved to be significantly higher than the values estimated on the basis of the generally accepted formulae. samples were taken from the conductors on the first 12 disks of the upper part of the MV winding (93kV). the decision to carry out the work depends on the following two criteria: the residual lifetime of the active part is above 15 years. which comes under the most thermal stress. Figure 4: Coefficient A according to moisture Figure 5: example of chart at 80°C 4. on the basis that the transformer is taken to the end of its live part's life without suffering a serious premature failure associated with one of its equipment parts. This led RTE to undertake research tests which initially showed that this material's mechanical withstand to short circui was good.On the basis of this law. the temperature and other factors? In the case of transformers with quite a low average load and operating temperatures which are generally below 70ºC. is the breakdown of the hemicellulose. concerns foreseeing the rise in the failure curve at the end of the transformers' lifetime and knowledge of the influencing factors. 2nd line of improvement: Delaying the ageing of the active parts As previously explained. We will then be able to anticipate this rise on the failure curve and. represented by the furans.Sample position on the disk Under yoke HV bushing side Disk 1 MV bushing side OLTC side Under yoke HV bushing side Disk 12 MV bushing side OLTC side pressboard insulating cylinder (without thermal stress) Sample position on the conductor Outer paper layer Inner paper layer Outer paper layer Inner paper layer Outer paper layer Inner paper layer Outer paper layer Inner paper layer Outer paper layer Inner paper layer Outer paper layer Inner paper layer Outer paper layer Inner paper layer Outer paper layer Inner paper layer Maximum value Minimum value Polymerisation degree 440 415 440 470 450 430 450 430 480 460 480 490 520 480 500 490 759 709 Table 2: Measurements of the degrees of polymerisation on the Laval transformer This difference highlights the fact that caution is needed when using furans as a indicator of ageing. foresee investments and even delay the rise by reinforcing or improving maintenance practices (periodic or conditional) on the oil in particular. this means that the aforementioned ageing mechanisms are activated at a low level. at low loads and therefore low temperatures. The numerous research projects under way at international level. as the substations are designed to be able to transmit the maximum powered demanded with N-1 transformers. This expert investigation opened up other fields of research on the following issues: How are furans distributed between the oil and the paper and how does this distribution evolve in line with the characteristics of the oil. on average. Up to now. Another important question concerning infrastructure management. It appears necessary to consolidate future laboratory-produced models with on-line monitoring and measurements on old 9 . particularly those presented at the last CIGRE symposium [6]. which is characteristic of the equipment's end-of-life. this difference reinforces the interest given to work in some relevant laboratories [5] in the search for new indicators to quantify the rupture of the cellulose chain. This particularity is explained by the fact that the power transmitted by the network experiences quite large seasonal and daily cyclical variations and. consequently. analysis of the serious failures affecting the live parts of RTE transformers has not revealed any underlying trends which correspond to a rise in the failure rate. the transformers are operated. also systematically associated with the rupture of the cellulose chains? Furthermore. RTE must therefore look at ageing associated with low power conditions. 6. RTE's refurbishment policy aims to limit the number of serious transformer failures through heavy preventive maintenance which is performed once in the transformer's lifetime. testify to the fact that this issue is currently at the centre of utility concerns. in light of the fleet's quite rapidly ageing nature. Long. [4] JP. operate at a low load. B. Finally. R. evolutions in the transformer oil preservation system (for example. Jalbert. [3] L.) o resorting to additives (after having validated stability over time). so as to anticipate its renewal. Tetreault. Morin. Moutin. Hansen. D. P. 'Ageing of Kraft paper by acid catalized hydrolysis' – IEEE Dielectric liquids.B. 10 .2007/s 10570-007-9124-1 – June 2007. Liland (SINTEF) 'verification of paper condition in aged transformers' . 'Furanic compounds analysis' . CIGRE colloquium A2/D1. contrary to generator step-up transformers. Pahlavanpour. and at a financial level.CIGRE WG 15. if the preponderance of the acid hydrolysis phenomenon on the RTE fleet is validated: o resorting to physical-chemical processes which aim to restore the oil original characteristics (regeneration on attapulgite clay. these research efforts must also allow ageing models to be defined for transformers which. A. S. M.1 July. [2] CIGRE TF D1. Lundgaard. 'Remnant life assessment of a thirty years substation transformer' . Taisne. by abandoning atmospheric breathing through a desiccator while having first validated the absence of risk of copper sulphides appearing in the operating temperature range). L. [5] J. Lessons drawn from the Laval experiment will therefore be used to supplement models which are more realistic while combining hydrolysis and oxidation of the cellulose and considering other accelerating factors such as acids [3]. 'Ageing of cellulose in mineral oil insulated transformers' . Following this work. resin based ionic treatment. E. W. etc.transformers which are in the process of being repaired or at the end of their life. evolutions in the maintenance of oil such as the regular dehumidification of the transformers or. 26 June .03 IWD 11 p1-21. Lillevik.E.01 – Bruges – October 2007. on average. the possible measurements that may be envisaged could concern: on the one hand. 7. Pablo. Gilbert. Tanguy.01. The development of ageing models will also provide an opportunity to define maintenance methods which will allow the lifetime of this transformer fleet to be extended.01.01 – Bruges . Dahlund. on the other hand. Liebert.4-β-glycosidic bonds of cellulose in an oil-impregnated insulating paper system” – Cellulose DOI 10. [6] L.October 2007. P. D? Linhjell. BIBLIOGRAPHY [1] A.CIGRE colloquium A2/D1. B.10. assessing the lifetime of its ageing transformer fleet is an important issue both at an industrial level. O. K. Such an assessment is characterised by expert knowledge on the ageing of these transformers. as it is most probably the case for a large number of Transmission System Operators in Europe.Brochure CIGRE N°323 (October 2007). Ingebrigtsen. pp 381 – 384. Devaux. F. ICDL 2005. Lessard-Deziel “Identification of a chemical indicator of the rupture of 1. Lundgaard. This objective means to continue research efforts undertaken in several laboratories so as to be better able to estimate the degree of polymerisation on the basis of furans or by fine-tuning methods based on other indicators of cellulose ageing in the oil. Conclusion For RTE. which rarely operate at full load.
Report "TRANSFORMER REFURBISHMENT POLICY AT RTE CONDITIONED BY THE RESIDUAL LIFETIME ASSESSMENT"