Test and Measurement - Power transformers – are you covered?


It's easy to assume the substation on your site belongs to the power utility, but are you absolutely sure? If you get it wrong, says Damon Mount of Megger, and you're unlucky enough to suffer a transformer fault, you could find yourself landed with a bill for tens or even hundreds of thousands of pounds

In the substation, the power transformers are probably the most expensive items. And that's not the worst of it - the delivery time for a replacement transformer is typically months - or even years for the largest types. The direct and indirect costs associated with a transformer failure can, therefore, be enormous.

But there's surely no need for concern. All of the power transformers on your site are the responsibility of your energy supplier, aren't they? It may be a very good idea to check again. In a surprisingly large percentage of installations, the power transformers belong to the owner of the premises, and not to the power utility.

Of course, there's still no reason to worry, because transformer failures will certainly be covered by insurance, won't they? The answer is possibly not. Because of the huge costs involved, insurers are understandably cautious about making payouts relating to transformer faults and failures. If there is a claim, they will certainly ask for evidence to show the transformer has been regularly tested and maintained.

Since many companies are not even aware they are responsible for the power transformers on their sites, it's not too much of a surprise there are a lot of transformers that most certainly don't get the regular attention they need.

This is a special concern with the many transformers currently in use that have long exceeded their design lives. Although they may apparently still be working well, it is inevitable some of the materials used in their construction - in particular the insulating materials - will have started to deteriorate.

If an unmaintained transformer fails, whether it is old or new, it's perfectly possible that the insurers will contest the claim or refuse to pay. Let's take a look at what needs to be done to avoid this potentially devastating situation.

The first and most obvious step is for maintenance departments to check which of the transformers on their site are their responsibility. The next step is to implement a regular testing programme for these transformers.

But what form should the testing take? There are, of course, many types of conventional tests that can be applied to power transformers to check, for instance, the performance of the tap changers or the windings.

This means to build up a reasonably complete picture of the transformer's condition, a whole battery of tests is needed, which will take a considerable time to perform. During this time, the transformer will be out of service, which can be very inconvenient.

There are, however, two tests that between them can provide a wealth of information, not only about the presence of faults but also, in many cases, their type and location. These tests are sweep frequency response analysis (SFRA) and frequency domain spectroscopy (FDS).

Electrically, a transformer is made up of multiple capacitances, inductances and resistances. It is, in effect, a very complex circuit that produces a unique ‘fingerprint' when test signals are injected over a range of frequencies and the results plotted as a curve. In particular, the capacitances in the transformer are affected by the distance between conductors.
Movement of the windings, which can be caused by electrical overloads, mechanical shocks or simply by ageing will, therefore, alter the capacitances and change the shape of the frequency response curve.

The SFRA test technique for transformers is based on comparisons between measured curves, which allow variations to be detected. An SFRA test involves multiple sweeps and reveals whether the mechanical or electrical integrity of the transformer has been compromised.

SFRA tests are used to capture a ‘fingerprint' reference curve for each winding when the transformer is new or when it is known to be in good condition. These curves are subsequently used as the basis for comparisons during maintenance or when problems are suspected.

The best way to use SFRA testing is to take regular measurements on the same transformer over a period time, and to compare the results. However, it is also possible to use type-based comparisons between transformers with the same design. Finally, a construction-based comparison can be used in some circumstances, when comparing measurements between windings in the same transformer.

A single SFRA test can detect winding problems that would otherwise require multiple tests with various kinds of test equipment, as well as problems that cannot be detected at all by tests of other kinds.

As a general guide, magnetisation and other problems relating to the core alter the shape of the SFRA curve at the lowest frequencies, up to around 10 kHz. Medium frequencies, from 10 kHz to 100 kHz represent axial or radial movements in the windings, and high frequencies above 100 kHz correspond to problems involving the cables from the windings to bushings and tap changers. In modern SFRA test sets, built-in analytical tools simplify comparisons between curves.

While SFRA tests provide a lot of information about the condition of a transformer, they do not give an accurate indication of the presence of contaminants - in particular water - in the transformer insulation. Standard tests, such as the widely used Karl Fischer test, are, of course, available for accurately assessing the moisture content of transformer oil, but this is not the whole story.

In fact, it is usual for a much greater percentage of the moisture in a transformer to be held in solid insulation such as paper than is held in the oil. To further complicate matters, the moisture moves between the solid insulants and the oil in a way that is influenced by many factors including, in particular, temperature. 

Measuring the moisture content of the oil may not, therefore, provide dependable information about the moisture content of the transformer's solid insulation. This is a serious concern, as moisture in the insulation significantly accelerates the ageing process in transformers and, in addition, it can cause bubbles between windings that lead to sudden catastrophic failures.

To establish the moisture content in the transformer, the second of the tests mentioned earlier - frequency domain spectroscopy (FDS) - can be used. Initially, this may sound a lot like SFRA, as it involves measuring transformer characteristics at over a range of frequencies. This time, however, it's the dielectric properties of the insulation (capacitance, loss and power factor) that are measured over a range of frequencies, typically from one millihertz to one kilohertz.

These are, in essence, the same dielectric tests that are often carried out at power frequency, but testing at a single frequency provides far less information than is revealed by FDS testing. Unlike spot-frequency testing, FDS can, for example, reliably distinguish between a transformer that is dry but has bad oil, and one that is wet but has good oil. In the first case, the oil needs refurbishing or replacing; in the second the transformer only needs drying out.

FDS testing also has other benefits - it can be performed at any temperature, and the test can be completed quickly. Software can be used to calculate the water content in percentage terms, and modern FDS test sets typically provide accurate and detailed results in less than 20 minutes.

As we have seen, regular testing using the SFRA and FDS test techniques provides a reliable insight into the condition of power transformers, but how can this information best be used by the transformer owner?

A short-circuit fault on the transformer may cause unseen damage inside, and a damaged transformer put back into service could fail catastrophically. An SFRA test can be done before re-energising and compared to a reference trace taken while the transformer was in good working order. If the two traces match, nothing has changed and the transformer can be safely returned to service. Carrying out this test takes less than an hour, reducing outage time and saving money.

Ageing, mechanical damage and moisture content can be seen as a change in the frequency response of the transformer over time and may indicate that remedial action, such as drying out the transformer, is needed to guard against future failures. In other cases, it may show that the transformer is inevitably coming to the end of its useful life, but even then the information is invaluable.

In this situation, it may be possible, for example, to minimise the load on the transformer so it can continue in service until a replacement is obtained. And even in the worst case, there is at least a warning that failure is imminent, which can allow time for contingency plans to be made and put into place.

There is also another very valuable aspect of regular testing, which we touched on earlier. Insurance companies are more likely to honour a claim for failure of a power transformer that's been regularly tested and properly maintained so as to remedy any issues identified by the tests. Such a transformer is, of course, less likely to fail, but if it does there is at least the consolation that the insurers will foot the bill!

Even for those who are aware of their responsibilities in looking after power transformers, regular testing may appear as something of a burden. However, tests with modern instruments can be performed quickly and easily, and they yield dependable informative results. And, if the test regime eliminates just one unforeseen transformer failure that would otherwise have occurred, the effort involved in testing and the cost of the instruments used will have been repaid many times over.