Keeping your fleet in top condi­tion

How long will our trans­formers last? This ques­tion is faced by more and more grid oper­a­tors around the world whose fleets are becoming increas­ingly dated. Previ­ously, the deci­sion to replace a trans­former was often exclu­sively based on its age, regard­less of its actual condi­tion. With a new method, this will all change.

Jan Patrick Linossier and Rolf Funk have racked their brains over the so-called „bathtub curve“many times already. Both are respon­sible for strategic asset manage­ment at the „Rheinische NETZGe­sellschaft“ (RNG), Rhineland‘s elec­tricity grid oper­ator, and are there­fore tasked with assessing the condi­tion and risk factors of the equip­ment in use. This includes around 100 power trans­formers which the distri­b­u­tion system oper­ator manages from Cologne for an elec­tricity grid span­ning roughly 20,000 kilo­me­ters. „We have to know the exact condi­tion of our trans­formers. Whether they‘re still service­able or whether we need to invest in new ones is a ques­tion we are constantly faced with. The problem is, nobody in the industry knows exactly how long a trans­former will ulti­mately last,“ explains Linossier, who heads up strategic asset manage­ment at RNG.


The bathtub curve is a model for describing the suscep­ti­bility of equip­ment to failure over its entire service life. Just after commis­sioning, the failure rate is very high. Once the initial prob­lems have been fixed, the failure rate drops rapidly and hovers at a constantly low rate. For a long period of time, this doesn‘t change. But as the system gets older, the prob­lems start to stack up again after a certain point. If you repre­sented this trend as a graph, with time along the X-axis and the failure rate along the Y-axis, the curve would look some­thing like the cross-section of a bathtub.

Jan Patrick Linossier, Head of Asset Manage­ment at RNG (right), and Alexei Babizki, Port­folio Manager at MR, are happy with the results of the pilot study. (© Dirk Moll)

But what Linossier and Funk have been racking their brains over is the following: At what point of the bathtub curve are their trans­formers currently at? When does the failure rate start to climb again? „That is the point at which we should be arranging a replace­ment,“ says Mr. Funk, who special­izes in strategic asset manage­ment for power trans­formers. Since the average age of trans­formers at RNG is 45.7 years, this ques­tion is becoming increas­ingly pressing. Most of these trans­formers were commis­sioned back in the 1950s and 1960s. The deci­sion to replace these trans­formers not only has tech­nical rami­fi­ca­tions, but also has finan­cial consid­er­a­tions as well, since a new power trans­former of this scale would cost almost half a million euros. „Previ­ously we would simply replace the trans­formers with new ones after a certain amount of time, but the cost pres­sures today are much higher. For this reason, we want to get the most out of the equip­ment for as long as possible,“ says Mr. Linossier.

„There are trans­formers used in indus­trial fields which have to replaced after 15 years of usage, and then there are those which only operate at 30 percent load and have been in oper­a­tion for over 50 years already. So the transformer‘s age alone doesn‘t neces­sarily say anything about its condi­tion.“
Alexei Babizki

And RNG isn‘t the only oper­ator with a fleet of aging trans­formers: In the two decades following the Second World War, the demand for more and more elec­tricity increased rapidly. New power plants were built, cables were laid, and numerous primary substa­tions were erected. Huge swaths of the German elec­tricity grid today are still based on the plans drawn up during this time. And this was not limited to Germany, but was rather a phenom­enon of all modern indus­trial nations. That said, the network load fore­casts at the time were much different.

„With the increasing use of renew­able ener­gies, grid struc­tures need to be completely redesigned and it‘s there­fore very impor­tant that we know the exact condi­tion of our trans­formers,“ stresses Mr. Linossier. On the one hand, because the utiliza­tion of the trans­formers changes over time and, on the other, because upgrading the network will make some instal­la­tions redun­dant in the future. „If we replace the trans­formers too early, that could turn out later to be a bad invest­ment,“ adds Mr. Funk. At present there is no method for assessing the exact condi­tion of the trans­former fleet.


However, all this is set to change thanks to a chance encounter in a tram in Lyon. Alexei Babizki, Port­folio Manager at MR and Markus Zdrallek, a professor at the Univer­sity of Wuppertal, got into a chance conver­sa­tion on the way to the airport, both having just attended the same congress. Zdrallek holds the senior profes­sor­ship for elec­trical energy supply tech­nology and has been working on the condi­tion assess­ment of equip­ment in elec­tricity grids for many years. „We want to offer grid oper­a­tors a scien­tific basis for intel­li­gent network renewal strate­gies,“ says Mr. Zdrallek, „but power trans­formers have not yet been exam­ined in my research.“


The condi­tion of the trans­formers is exam­ined in up to three stages, which can reduce the required down­time of the trans­formers:

1 In the first stage, the oper­ator provides the data which is already avail­able. This includes, for example, the age of the trans­formers, measure­ments that have already been carried out, and main­te­nance history infor­ma­tion.

2 The second stage involves an on-site appraisal which is carried out while the trans­formers are still in oper­a­tion. In the visual inspec­tion, an expert uses a check­list to examine the trans­former based on its external char­ac­ter­is­tics: Are there any signs of corro­sion or leaks? Is there visible damage to the bush­ings or fans? How old is the motor-drive unit? For the measure­ments, ther­mo­graphic tests are performed and oil samples are taken.

3 If, after this stage, there are any indi­ca­tions of more signif­i­cant damage, the third exam­i­na­tion stage will follow. In this stage,exten­sive measure­ments are performed, including a dynamic resis­tance measure­ment and partial discharge measure­ment. However, to carry out these tests, the trans­former must be removed from the grid.

The data collected using this method is eval­u­ated. The analysis is carried out using the algo­rithm devel­oped at the Univer­sity of Wuppertal in asso­ci­a­tion with MR. Then the data is trans­ferred into two indexes, with one revealing infor­ma­tion about life-time wear and the other about the failure rate. For the assess­ment, the results of the two indexes are trans­ferred into one graphic. The condi­tion of the trans­former fleet can thus be visu­al­ized simply; the more trans­formers located in the quad­rants near the origin, the better the condi­tion of the fleet and vice versa.

This imme­di­ately aroused the interest of Mr. Babizki, since he was looking for a system for deter­mining the condi­tion of power trans­formers. He also knows that this helps provide an impor­tant service for MR‘s customers: „It‘s impos­sible to give a general answer to the ques­tion of how long a trans­former will last. There are trans­formers used in indus­trial fields which have to replaced after 15 years of usage, and then there are those which only operate at 30 percent load and have been in oper­a­tion for over 50 years already. So the transformer‘s age alone doesn‘t neces­sarily say anything about its condi­tion,“ explains Mr. Babizki. He and Mr. Zdrallek quickly agreed that this would be the ideal topic for collab­o­ra­tion.

„It is of strategic impor­tance for us that we know the exact condi­tion of our trans­formers.“Jan Patrick Linossier

„We didn‘t have the well-grounded under­standing of trans­formers we needed, but we did have exten­sive exper­tise in the devel­op­ment of condi­tion assess­ments,“ recalled Mr. Zdrallek. The researcher hopes that the project will shed more light on how trans­formers dete­ri­o­rate over time, since he is also inter­ested in the bathtub curve. „The curve is often described in the theo­ret­ical liter­a­ture and has even been proven for some elements of elec­tronics. But I was skep­tical of whether the curve even exists for elec­trical grid compo­nents, and more specif­i­cally for trans­formers, because nobody has yet managed to prove this,“ adds Mr. Zdrallek. And thus began the collab­o­ra­tion between the two part­ners. In a series of meet­ings and work­shops, they worked together to develop an assess­ment system.


What was still lacking, however, were trans­formers on which they could test their assess­ment system. Mr. Zdrallek thought imme­di­ately of RNG, with which he had already completed numerous projects. Linossier and Funk didn‘t have to think about it for long: „The approach seemed very promising to us, so we imme­di­ately agreed to collab­o­rate in a pilot study,“ says Mr. Linossier. What made this approach special is that the system (see box below) regards the trans­former from two perspec­tives. On the one hand, there is the long-term view which focuses on longevity and is thus impor­tant for invest­ment deci­sions. And on the other hand, there is the short-term view which is targeted at the risk of failure, and influ­ences deci­sion-making relating to main­te­nance.

Markus Zdrallek, professor at the Univer­sity of Wuppertal, hopes that the system will open up new avenues of research for the effects of aging on power trans­formers. (© Dirk Moll)

All the main compo­nents of the trans­formers are exam­ined care­fully, from the on-load tap-changer and motor-drive unit to the cooling fans. A total of 200 para­me­ters ulti­mately influ­ence the outcome of the assess­ment, including main­te­nance history data and results from on-site measure­ments. The collected data is then analyzed using a special algo­rithm. „There are many indi­ca­tors we can use to ascer­tain the condi­tion of the trans­former. If, for instance, the ther­mo­graphic measure­ment shows that a partic­ular area of the trans­former is hot, it is possible to localize the fault. Then detailed measure­ment methods are required,“ explains Mr. Zdrallek. Another crucial element for the assess­ment is the oil analysis, which can be carried out by MR in the labo­ra­tory of its subsidiary Messko. As Mr. Babizki makes clear, „it‘s a bit like doing someone‘s blood-work at the lab, where an increased white blood cell count might indi­cate an infec­tion, for example. Simi­larly, for trans­former oil, a higher acetyl value is an indi­ca­tion of elec­trical arcing, which may be a sign that the trans­former is in poor condi­tion.“


RNG provided nine trans­formers for the pilot study. „Our choice of trans­formers reflected a repre­sen­ta­tive sample of our fleet,“ explains Mr. Linossier. After exten­sive tests and subse­quent analysis, the results were in: All trans­formers were found to be in good condi­tion. For some units, the risk of failure could even be reduced with simple measures, thus extending their service life. „That‘s the advan­tage of the method we have devel­oped. Not only do we offer our customers a condi­tion assess­ment, but also recom­men­da­tions for correc­tive action,“ empha­sizes Mr. Babizki.

The part­ners also take special cases into consid­er­a­tion. For example, in one trans­former a high acetyl value suggested there was a problem. As Mr. Linossier stresses, „MR pointed out that this value can vary depending on model.“ The limit-value being exceeded in the trans­former oil was there­fore trivial, which made renewal unnec­es­sary. „We designed the algo­rithm to take into account the type of tap changer used,“ adds Mr. Babizki.

Mr. Zdrallek is also happy with the new method: „It is exciting to me from a scien­tific point of view. If many network oper­a­tors start working with it, then we might be able to use the mass of collected data one day to find out whether the bathtub curve is actu­ally real.“ In any case, RNG has decided to use this method to examine the majority of its fleet of trans­formers. „We now have an instru­ment for objec­tively eval­u­ating the condi­tion of our fleet. This is an enor­mously im portant devel­op­ment for our strategic deci­sion-making,“ under­scores Mr. Linossier.

Learn more about this collab­o­ra­tion in an inter­view with Jan Patrick Linossier and Markus Zdrallek.


Would you also like to apply the new method to assess your fleet of trans­formers?

Alexei Babizki is avail­able for your ques­tions: