Lord Consulting
Power Industry Asset Management Consultants
Lord Consulting
Power Industry Asset Management Consultants
Home » Technical Papers » On-Line Monitoring as a Strategic Tool to Enhance Supply System Reliability Monitors
Certain classes of electrical network fault can develop unseen over a long period and then rapidly deteriorate into a major outage, in some cases completely destroying the asset and causing substantial collateral damage. The disruptive impact on both the electrical network and the wider community can be considerable. The recent availability of on-line monitoring systems can assist electrical network operating companies to detect potential faults early enough to take appropriate action. This paper broadly overviews the field of on-line monitoring and suggests that not only will the use of on-line monitoring systems progressively become the industry norm for all critical network assets, but that in future network operators may be independently audited for their compliance in meeting new standards governing the on-going health of electrical networks.
Our supply networks are ageing. The growth boom of the late 1950's through to the mid 1970's means some of the assets in service are now well over 40 years old. While much research has been done on failure mechanisms of cables, transformers, tap changers, switchgear and high voltage bushings (which together make up the bulk of the long haul transmission and local distribution networks) the reality is that any of these items of equipment can and do fail without warning. Stated bluntly, we simply do not know how long these assets will last.
Continuous improvement is an integral component of the electricity supply industry. From the earliest days, power supply companies have wrestled with the optimum balance between high system reliability and low costs (both capital and operating). When assets were young, the problem of failure due to asset age was for a future era.
That era is now upon us. Surrounding communities trust their electricity suppliers to protect them from prolonged and disruptive outages. Electricity is so integral to modern life, major cities simply cease to function when the power supply is interrupted. Consequences of these assets failing include major supply disruption, considerable localised property damage, and occasionally loss of human life. Thankfully, these events have been reasonably rare to date, but they do happen and the combination of an ageing asset base coupled with the new commercial trend to push assets that much harder, means the frequency of failure could well increase.
With significant outages increasingly in the news, the bulk of installed assets approaching or exceeding their nominal design life, and deregulation pressing for ever-higher asset utilisations, it prompts us to ask some very professional questions of exact how well as electricity suppliers we really understand the underlying health of our core network distribution assets.
This change in focus will mean asset base integrity will no longer be solely the prerogative of the engineers and accountants. From now on, ensuring asset base integrity will become a corporate strategic issue with decisions taken in full knowledge of the tradeoffs involved and their likely consequences. We can no longer take for granted the assets that have served us so faithfully up to this point. We need to increasingly arm ourselves with new tools whose prime function is to nurture and protect the established base of electrical network assets.
This paper is presented as an introduction to, and overview of, the increasingly maturing field of on-line asset monitoring and assumes little previous knowledge of the subject. Readers who wish to follow up any particular issue raised are referred to the extended references that provided both background and specific information included in this paper.
Electrical network equipment failures appear to be on the increase. This should hardly be a surprise. Some of the equipment comprising the bulk of the world's electricity systems is now approaching (or in some cases has now well exceeded) its original nominal design lifetime. We are now contemplating the implication of this. Should we replace items as a matter of course that have reached their design life simply to ensure on-going supply reliability? Mechanical replacement of assets reaching a nominated service age is as impractical as it is unnecessary.
Numerous studies have been done around the world over the last 20 years on this subject. As just one example, IMIA1 conducted a study looking at 75 failures in large oil filled transformers (100MVA and above) between 1989 and 1995 [4]. A working group was convened to study this and concluded that failures in these transformers:
This example serves to illustrate why this issue is so important: the cost of disruption can often significantly exceed the value of the electrical assets themselves. We are increasingly witnessing power loss for lengthy periods on unprecedented scales. Recent widespread outages this year around the world, with numerous smaller (but still locally serious) problems involving substation fires and HV cable failures. While the causes vary widely, and on-line monitoring of relevant assets may or may not have assisted prevent any specific case, they do serve to give us a sobering wake-up call and remind us as network managers we need to be doing the best we can to ensure we minimise the likelihood of preventable incidents on these scales.
1 International Machinery Insurers Association, see Reference 4.
The main problem we have is that it is hard to see how healthy or otherwise our electrical distribution assets are. Outwardly, they may appear fine and even close inspection may not reveal a problem straight away.
A good analogy is that maintaining healthy network assets is like maintaining healthy people. In both, it is a life long activity. In both, it is a matter of degree. In both, damage can be cumulative over time, with little outward appearance of any inner trouble. Properly recognised and treated, a person with a potentially serious medical condition can live a long and productive life even though they may need a few restrictions placed on them as they get older. Likewise, an asset may have a known condition which if carefully managed can mean it will continue to operate normally well past its nominal design lifetime. However, ignore or mismanage this condition and premature failure could well be the result.
The trick is to know what to signs to look for and to know what appropriate action to take when they are found. In the same way that modern medicine now has non-invasive tools to look deep into the human body and check its health status, the electrical power industry needs similar tools that will help revolutionise the management of our distribution networks.
Whether you become the network equivalent of the local GP, the nurse, the hospital orderly, or the specialist surgeon, all of us will need to start thinking differently about the increasing population of old age pensioners out there we fondly call our "asset base".
In the good old days, as any competent electrical engineer over 70 years old will tell you, it was fatal to let accountants meddle in the affairs of power companies. All they did was ask awkward questions and talk about money all the time. Naturally, it was far more interesting to talk about power factors and zero sequence impedances.
Deregulation has changed all this, of course, and now the engineer welcomes the accountant's awkward questions with open arms. In fact, the accountant is the engineer's new secret weapon, since now they both have a common problem. How do we keep our network operating both reliably and economically when portions of it are potentially approaching the end of their working lives? Good question. Corporate management is now starting to pay close attention to this issue as well.
Clearly, a much closer and more targeted approach will be required looking for problems before they deteriorate to the point of tripping and/or threatening the asset itself. Better still, can we predict the "time to failure" of a fault quietly brewing in a cable or bushing, replacing the asset only at the latest possible moment.
While we have not reached this degree of sophistication just yet, there are some very clever monitoring systems now available to trend the "health trends" of network assets and detect potential problems early. Abnormal trends can alert us to taking the more targeted action as needed.
Deregulation has required us to focus on how to lower network operating costs without compromising system reliability at a point in the life of our networks where asset ageing is starting to get our full attention. Given the higher commercial risks deregulation brings, on-line monitoring is likely to become more than just a very useful life extension tool. It must start to form a core part of a corporate strategy to ensure the maximum is extracted from their network investments while minimising the risks the corporate entity faces in doing so.
On-line monitoring of critical network assets provides information previously unavailable. This in turn allows better asset management decisions and increased management confidence going forward. Rather than just hoping the network is in sound shape, this in now visible to everyone via hard data.
Table 1 gives a summary of the factors increasingly driving network operators towards on-line asset monitoring. Note that many of these are strategic corporate factors that are difficult to quantify in a purely financial business case. This is touched on above and discussed further in Section 9.
| Strategic Driving Factor |
Comment |
| Networks are ageing |
Some assets have already exceeded their design lifetimes without serious problems, while others have failed early |
| Allows higher asset utilisation |
With relevant data collected and made visible, network assets can be worked harder with increased confidence. |
| Lack of historical information |
For most assets, no real-time trend monitoring has been done and loading profile and other information has been lost forever. We are now in “information catch-up” mode. |
| JIT maintenance |
On-line monitoring is a major help in delaying maintenance as long as possible hence driving down routine costs. |
| Current sampling too slow |
Nyquist sampling theory says that to detect any asset fault, a sampling frequency at least twice the likely fault occurrence rate is required. Most present condition assessment takes place at much greater intervals than the ultimate common failure mechanisms. Catastrophic faults can appear very quickly (weeks, days, or even hours). |
| Replacement very difficult |
In certain cases, replacement of the asset would be extremely difficult and time consuming. Such critical assets are prime candidates for on-line monitoring. |
| NPV advantage |
Clear cut financial benefit of monitoring immediately |
| Limits direct costs |
On-line monitoring can help ensure any damage to the asset is minimised. Operators can be alerted and react quickly to reduce load or quarantine the asset as needed. |
| Limits indirect damage costs |
Cables faults, transformer fires, or explosions in switchgear or bushings, can all impact largely preventable damage on surrounding equipment or property. |
| Limits business interruption costs |
Risks and liabilities of sustained supply interruption are very real. On-line monitoring may well lower insurance premiums. A prolonged and widespread outage may invite possible legal challenge if on-line monitoring may have avoided it. |
| Aids health and safety |
Porcelain from exploding bushings can travel considerable distances at high velocity. Labour Unions are increasingly aware of the dangers workers can face in high voltage substations and are starting to insist on added safety precautions like on-line monitoring being used where it is reasonable to do so. |
| Reality today |
While on-line monitoring has been evolving for many years, it is fast maturing into a serious and reliable network tool. |
| Organisation Embarrassment |
Sometimes called the “CEO Embarrassment Factor” in a business case. Hard to quantify, but very real. |
| Brand damage |
It is not just the CEO reputation at stake, but also the organisation’s brand. This has a real “re-construction” cost as well. |
| Limited expertise |
On-line monitoring systems are effectively “expert systems” containing accumulated knowledge and experience. Given the world-wide shortage of experts who understand the complex failure mechanisms, these systems allow genuine management “leverage” of available knowledge. |
In this section, we look at some common failure modes for various asset classes and how on-line monitoring is assisting to detect the conditions early.
The following classes of assets are now regularly and economically monitored on-line:
For each of these asset classes, the common failure mechanisms are discussed below along with the telltale clues which monitoring systems need to watch for. (Section 8 contains a concise summary of this section in table form for future reference.)
Transformers have a natural limit to their life. In paper/oil insulated transformers, both paper and oil decompose over time. Excess heat within the transformer tank, particularly at unseen localised hot-spots, can dramatically reduce its useful working life. The cellulose in the paper breaks down over time and releases glucose, carbon monoxide, carbon dioxide, water, and acids. The glucose further breaks down into furans [9].
Oil also breaks down gradually over time, and this too is accelerated by heat from a fault. Depending on the type of fault it will produce a range of gases: hydrogen, methane, ethylene, ethane and acetylene in varying quantities. Figure 1 shows the relationship between these different types of transformer fault and the gases produced [12].
Water, heat and oxygen all accelerate the ageing process for transformers. Water can enter the tank from outside (either carried in as vapour in the air or enter directly via a leaky seal). However, water is also generated inside the tank by the breakdown of the papers themselves. Even in perfectly sealed transformer, the water dissolved in the oil will gradually increase.
Once water is dissolved, the oil acts as a water-transferring mechanism, carrying the water to every crevasse within the transformer. While the water remains dissolved in the oil, negative impact is generally minimal. However, the oil can become increasingly saturated with water, allowing water to come out of solution as the transformer cools below its "dew point". In these conditions, free water can appear anywhere within the transformer at short notice and cause immediate localised insulation failure [9].
Figure 1: Transformer fault classes and the various gases they produce.
It is worth noting that hydrogen (H2) is produced in all the above cases, therefore producing excellent early warning detection for all types of elevated temperature in-tank faults.
However, to see any of this we have to look for it. Traditionally the oil is checked using a DGA (dissolved gas analysis) test at intervals between six months and two years (for a good transformer) and perhaps as often as monthly if a critical transformer is giving indications of possible problems. However, in some cases the fault mechanisms described above can very quickly develop into a runaway condition which an intermittent checking regime, no matter how good, will simply not detect in a timely enough manner.
Figure 2 shows a thermal fault that advanced quickly in a 650 MVA generator transformer at Mirant Mid-Atlantic, LLC (USA). The graph shows a dramatic rise in H2 over a two-day period in September 2002. This was caught in time using a Morgan Schaffer Calisto on-line hydrogen monitor. Once the monitoring unit raised the alarm, an on-site DGA (dissolved gas analysis) test of the oil indicated a localised fault temperature in excess of 700C. The transformer was taken out of service and it was found that an internal high voltage lead to a bushing was burned and ready to fail. As can be seen, this is a very graphic demonstration of the effectiveness of on-line monitoring. As it happened, the monitoring unit alerted operations staff who promptly reduced the transformer loading to reduce the thermal heating at the location of the fault. This brought the situation under control and allowed an inspection outage to be scheduled for a more convenient time [13].
Figure 2: Sudden increase in hydrogen generating rate in a 650 MVA transformer
Many distribution transformers are accompanied by an on-load tap changers which account for a significant proportion of transformer related failures. These devices, fitted to the high voltage side, control secondary terminal voltage regulation as the loading on the transformer varies. Being mechanical devices in constant operation, tap changers are a significant point of weakness in the distribution system.
Tap changers can experience a number of problems, including:
Most of these conditions can cause a rise in tap changer oil temperature, sometimes quite quickly. If not attended to, this in turn will lead to damage to either or both of the mechanism and/or the insulation system. Measuring just the OLTC oil temperature will show up this rise, but can be misleading as changes in main tank due to load variations or changes in ambient temperature also affect the OLTC.
OLTC on-line monitoring equipment gets around this problem by measuring both the OLTC oil temperature and the main transformer tank oil temperature, together with the transformer load [10]. It then trends both the difference between the two and the deviation (the trend of this difference) to look for the various heat generating conditions. On-line monitoring units are employing sophisticated digital filtering techniques to avoid false alarms from short-term temperature changes. Best accuracy is obtained from temperature readings taken from thermowell probes. However for OLTC or transformers without thermowell pockets fitted, magnetic temperature probes can be attached to the outside of both tanks with good results. In additional, motor current can monitored to detect motor or mechanical drive problems.
Figure 3 shows a typical on-line monitor installation for the transformer core and tap changer. This equipment is monitoring the oil temperature of both the main transformer tank and the OLTC enclosure. An algorithm, having been preset with fixed data about the transformer during installation, calculates the winding temperatures and can both control local pumps and raise alarms for relaying via SCADA to a remote control room.
Some newer transformers use optical fibres installed with the winding at manufacture to allow direct measurement of the core temperature. However, surprisingly close correlation is achieved between this direct measure method using optical fibres and the indirect temperature algorithm approach. The algorithms used have been standardised (IEEE C57.91 and the IEC equivalent) which means consistent results can be obtained monitoring existing transformers without internal optical fibres.
Figure 3: On-line transformer and tap changer monitoring employing
temperature-based measurement and algorithms. These monitors are
gradually replacing older analogue gauges.
Figure 4: Bushing Currents
The main purpose of a bushing is to transfer load currents in and out of metal (grounded) enclosures at system voltages. High voltage switchyard bushings and CTs are particularly vulnerable to sudden and spectacular failure. Bushing failures also initiate 30% of transformer failures [15]. If the insulation system breaks down, bashing can explode without warning and throw out shards of porcelain at high velocity. However, bushings usually give warning signs that all is not well. Concern surrounding safety aspect of bushings is leading many network operators to look seriously at on-line monitoring.
The health of a bushing is best determined by measuring the current that flows through the bushing insulation at full system voltage. In a healthy bushing, this current is mostly capacitive but with a small resistive component causing heating losses in the dielectric of the insulation material. In a practical bushing, it is not possible to totally eliminate this resistive (heating) component. If the bushing insulation deteriorates, this resistive current component increases and alters the overall angle of the bushing insulation current with respect to the driving voltage.
HV bushings are constructed in different ways, but the most common type is made using alternate layers of oil impregnated paper and foil, protected inside a porcelain shell [30]. The foil layers control the voltage gradient within the bushing by acting as many individual "capacitors" in series. Microscopic "pores" in the paper, which reduce its insulating properties, mean many paper layers are required between each foil layer. If the paper between any two foil layers breaks down, the capacitance of the bushing will increase, since one of the series capacitors has been removed. This in turn increases the capacitive current through the bushing. Therefore, by measuring both the magnitude of the current flowing in the bushing insulation and the angle of this current with respect to the driving voltage, a good picture about the health of the bushing can be obtained.
The traditional way to check bushings is by measuring the "dissipation factor" of the insulation with an off-line test, usually at less than full system voltage. This involves removing the bushing from service and connecting it to an off-line HV "tan delta" (usually written as "tanD") test set. This measures the insulation leakage current magnitude and angle with respect to the purely capacitive bushing current (see Figure 4) against that of a known "test bushing". Off-line testing can be a rather disruptive exercise, and therefore is normally done reasonably infrequently (say once every two years).
However, it is also possible to check bushings dynamically on-line. To do this, a special signal coupler is placed into a bushing tap point located at the base of the bushing and connected to the other-most layer of bushing foil (see Figure 5). This allows a voltage signal to be generated proportional to the current flowing in the bushing insulation. Similar signals from other bushings in the same vicinity are also collected in real time and compared together using special software to watch for differences in the local group of bushings.
A minimum of three bushings is required for this technique, with the upper limit determined by the number of inputs on the monitoring system. The more that are included, the better the base-line picture against which can be built a picture of abnormal trends [19]. The tanD readings from all bushings in the local group are compared in a "round robin" fashion. This allows longer-term variations of insulation current being caused by changes inside the bushing to be separated from other variations caused by changes in local ambient conditions which affect all bushings in the monitored group equally.
It is interesting to note that that European failure statistics show 80% of all bushing failures on transformers occur between 12 and 20 years into their service life, in other words, around the mid-life of the transformer population [15]. This would suggest that bushings need much closer attention than we have been giving them to date, as the implicit assumption that they have a life span equal to the equipment onto which they are fitted may be quite incorrect.
A very similar approach can be taken with monitoring switchyard CT's on line. The main benefits of real time monitoring both bushings and CT dissipation factors on-line are the extended life of bushings and the safety increase for people working in the switchyards [19]. Network managers are starting to pay much closer attention to this area of their business.
Figure 5: A bushing tap coupler at the base of a high voltage bushing
The primary job of switchgear is to interrupt the very high currents which flow under fault conditions. Every other device in the network simply has to tolerate the flow of these high currents for short periods until a circuit breaker somewhere upstream of the fault opens.
Interrupting fault currents is the single most difficult task that must be performed in a network. Failure to adequately clear this current, typically 20 times the full load breaker's load current rating, can damage not only the breakers themselves but also other equipment within the network, such as cables, transformers and bushings [22]. The large physical and thermal stresses which the high faults currents generate can cause significant damage, particularly to older transformers where cellulose breakdown has already lead to a reduction in internal clamping pressure holding the winding structure rigidly in place [32].
Switchgear failure can be sudden and expensive. Being mechanical devices which must operate very quickly when required, switchgear experiences high operating forces and can be subject to physical stresses and wear. Like the protection relays which initiate the tripping signals, circuit breakers can sit for years doing nothing and then be called to operate in fractions of a second.
Failure mechanisms in breakers fall into a number of categories [22]:
The operating mechanism of a circuit breaker is vulnerable to degradation over time. It is tempting to think that a well-lubricated, well-sealed circuit breaker in good condition that hardly ever operates will require little if any maintenance. This is not so. Insulating oils gradually breakdown, and normal load currents (particularly at higher load factors) passing through the breaker will generate temperature rises causing loss of lubrication, oxidation of contacts and slow decay of the main insulation systems.
On-line monitors are available to handle both external yard circuit breakers and indoor metal clad circuit breakers, monitoring such parameters as:
As with other network assets, circuit breaker insulation can also be monitored for partial discharge. Oil can be monitored for dissolved gas and water (as is done with transformers), while gas insulated or vacuum breakers can (and certainly should) be pressure monitored to ensure the integrity of the main insulation container. The health of the tripping circuit, including the substation battery system which supplies the tripping energy to the circuit breaker, should also be monitored via trickle currents through the trip circuits and faults reported via SCADA. Failure to do this can leave an inadvertent but significant "hole" in the overall on-line circuit breaker monitoring system.
A number of digital protection relays also collect information to allow circuit breaker monitoring. In general, digital protection relays do not record mechanism timing sufficiently accurately, or monitor for re-strikes to reflect contact wear to standards such as IEEE/ANSI C37.04. Quite often, protection relays provide somewhat limited multistage alarm set points for the parameters monitored. It is likely in future that digital protection relays will incorporate many of the requirements of full on-line circuit breaker monitoring systems, but for the present separate monitoring units will probably remain necessary.
A well-designed circuit breaker monitoring programme will give network managers a good confidence level, but putting this in place is not a trivial exercise. With the increasing trend of deferring scheduled maintenance in favour of "just in time" (JIT) maintenance, it is important that network managers have confidence at all times that circuit breaker wear remains within normal limits. On-line monitoring can act as an alert trigger, predicting problems well ahead and allowing scheduled maintenance to be deferred as long as reasonable. Once again, information is the key.
2 Included is the switchgear enclosure itself, together with bus bar chamber and cable termination boxes.
Like other network assets, no one is quite sure how long high voltage power cables will last. Many of them have been in service around the world more than 50 years, with some still going strong after 90 years! In London, it was found that around 98% of all supply interruptions were the result of cable faults [23].
Cables experience high internal voltage gradients by design. With the HV cores so close together, the cable is totally reliant on the insulation system integrity (paper/oil, paper/gas, XLPE, etc) for continued operation. Even slight changes in the insulation properties can cause cables to fail quickly in service. It is remarkable they operate as reliably as they do.
Cables can suffer from mechanical damage during installation (particularly the outer sheath which if breached can allow water into the cable) and at any time while in service. Thermal stress from overloading of the cable, as with other electrical plant, can cause localised hot spots that can degrade the insulation material in the immediate area of the cable. This can weaken the cable system at that location causing eventual failure, even decades later.
Heat from an operating cable is normally conducted evenly away into the surrounding ground, avoiding hot spots developing. But over time, it is possible for ground material initially in good physical contact with the cable to slump away leaving inadequate thermal transfer. This situation can be very difficult to detect in existing cable systems. In new and particularly critical installations, an integral fibre optic core can be included in the cable and can be monitored on-line.
The bulk of cables in use today have insulation which is either paper/oil or polymeric (plastic). In the last 15 years, the swing has been largely in favour polymeric cables such as XLPE. Modern XLPE cables from reputable suppliers may be considered well made and reliable, but earlier XLPE cables (older than 10 years) sometimes suffered from tiny voids in the insulation material that could initiate partial discharge activity. These problems have now been largely overcome.
Partial discharge activity tends not occur to any appreciable extent in newer cables themselves, but rather in the joints and termination arrangements. Partial discharge monitoring is a very good technique for spotting problems early in joints and terminations, be they pitch or epoxy. These are common across all cable types and can arise from many causes, including insufficient jointing expertise, poor workmanship, inappropriate or loose ferrules, use of incorrect jointing kits for the cable type, changes in conductor type, and inadequate clamping of conductors in joint coffins.
On-line partial discharge monitors are now being used to detect this activity and watch the trends over time. Partial discharge in cables occurs when small voids in the insulation or terminations flashover internally as the voltage rises and falls during each half cycle. The high frequency currents resulting from these discharges flow to earth through the cable sheath and can be monitored via a special high frequency CT placed between the isolated cable sheath and the substation earth connection. An on-line monitoring system is available to watch for these RF (radio frequency) current signatures.
Figure 6: High frequency CT sensing of RF partial discharge currents in HV cables. (Note the insulated cable sheaths.)
The on-line monitors detect the tiny pulses of RF current and check them to ensure they correlate with the phase being monitored (this helps eliminate noise). They are then algorithmically summed over time to estimate the total discharge energy occurring, allowing an estimate of the cumulative damage being caused to the insulation. Recent advances have concentrated on using algorithms which improve the partial discharge pulse recognition against background noise, as well as paving the way for on-line cable fault location systems from partial discharge signatures.
Partial discharge can be present for years without causing problems, or it can rapidly degrade the insulation and cause a flashover. Therefore, like other on-line monitoring systems, it is the trend in the discharge currents that holds the most useful information. It is changes in trends that need to be watched for.
Polymeric cables can suffer from another problem known as "water treeing". Interest in this subject in cables was sparked a number of years ago by reports from the West Coast of the USA that "water trees" were being discovered in a high percentage of polyethylene extruded dielectric cables examined after 1 to 10 years in service [31]. It has since been reported as occurring in many countries.
"Water trees" begin to form when a cable is exposed to water and normal operating voltage over an extended period of time. Water can enter the cable through damage to the outer sheath (either during laying or once in service), or it could be water left over from the curing process (a problem in earlier XLPE cables). The base of the 'water tree' is located at the point where the growth starts, and the extremities tend to grow parallel to the lines of the localised electric field. Electrical forces acting on water molecules ("electrophoresis") at a microscopic point within the insulation, increase the separation between polymer units. These minute water droplets become oriented into a chain-line channel that is conducting. The result is a sharp electrode producing highly localised stresses [31].
Although partial discharge is a useful tool for examining cable joints and terminations on-line, it cannot always spot "water trees" in the cable itself until they are well advanced. The reason is that "water trees" do not create voids as such, hence do not create the tiny RF flashover current signatures that the partial discharge monitor looks for. Once the "water tree" has grown to the point where a flashover will occur, it is likely to be a full insulation breakdown between the core and the sheath with little if any partial discharge signature available beforehand. Even if it were available, the bulk of the damage would have already been done.
Cables are normally tested off-line at present for "water treeing", by measuring the tanD of the insulation similar to that for bushings (see Section 7.3 "Bushings and CTs"). Cables have a much higher capacitance than bushings, so nowadays the tests tend to be performed at very low frequencies (0.1Hz and lower) to keep the charging currents and test sets a manageable size. Cables with "water tree" damage tend to exhibit and increasing tanD figure with increasing applied voltage, this phenomena being quantifiable as a condition assessment tool (IEEE 400.2).
However, like most other assets, cables will benefit greatly from on-line monitoring. There is no reason in principle why tanD monitoring cannot be done for cables as it is for bushings. At this time, such an on-line monitoring system is not available.
|
Asset type |
Natural ageing processes |
Factors which accelerate ageing |
Natural outcome |
On-line monitoring signature to watch for |
Notes |
|
Transformers (tank & core) |
paper cellulose decomposes to CO, CO2, H2O, acids, & glucose (which further breaks down into furans) |
water ingress from outside, |
continuous degradation of paper insulation component |
water (H2O) |
1 |
|
oil decomposition into |
partial discharge (corona) |
localised cumulative insulation damage and eventual failure |
hydrogen (H2) |
|
|
|
thermal fault |
high localised stress on oil, paper and bus work leading to damaged components & functional failure |
hydrogen (H2), methane (CH4), ethane (C2H6), ethylene (C2H4) |
|
||
|
arcing fault |
rapid decomposition of oil and paper into gases, explosion likely if not stopped quickly |
hydrogen (H2), |
|
||
|
accumulated water dissolved in oil |
oil will saturate and become free when due point reached |
% water (H2O) dissolved in
oil |
|
||
|
electrical current flows |
excessive heat in core due to elevated load currents |
very rapid breakdown of insulation papers in core |
core temperature monitoring |
2 |
|
|
short circuit fault currents |
core distortion, paper damage and loss of clamping pressure |
frequency response analysis (FRA) |
3 |
||
|
On-Load Tap Changers |
breakdown of insulation system wear of mechanical components |
drive mechanism sticky or erratic in operation |
high wear and eventual drive mechanism failure |
drive motor rotor current |
|
|
drive mechanism totally jams (locked rotor) |
no secondary voltage control with likely damage to drive motor |
drive motor rotor current |
|
||
|
stops over correct tap contact but with insufficient contact pressure |
gradual heating of tap changer oil and cumulative damage to contact |
slower increase in OLTC temperature (wrt main tank temperature) when remaining on this tap setting for longer periods |
|
||
|
stops between two adjacent tap positions |
rapid heating of tap changer oil due to series resistor remaining in circuit continuously |
faster increase in OLTC temperature (wrt main tank temperature) with heating rate dependent on load current |
|
||
|
low oil level leading to arcing fault |
serious damage or explosion |
header tank oil level |
|
||
|
Bushings & CTs |
damage to porcelain |
rapid temperature swings or extreme
environmental conditions; |
crazing allowing dirt build-up and water ingress with reduction of insulation properties; |
changes in tan delta |
|
|
core paper deterioration |
elevated operating temps; |
localised partial discharge promoting further paper (and oil) deterioration (runaway condition) |
partial discharge RF currents |
|
|
|
oil deterioration |
similar mechanism as within transformers |
localised partial discharge promoting further oil (and paper) deterioration (runaway condition) |
partial discharge RF currents |
5 |
|
|
short circuits between foil layers |
design/manufacturing defects; |
higher capacitance leading to increased capacitive current |
increase in bushing C1 (and possibly C2) capacitance |
4 |
|
|
small voids within insulation system |
partial discharge (corona) |
localise insulation damage with formation of carbon and/or shorts between foil layers |
partial discharge RF currents |
|
|
|
large temperature variations |
differing coefficient of expansion of
bushing components; |
frequent thermal cycling results in
excessive seal wear in turn leading to oil
leak and/or water ingress; |
usually will show up as an increase in
tandelta and/or partial discharge RF
currents; |
6 |
|
|
Switchgear |
contact wear |
excessive interrupting duty: |
cumulative damage which can lead to catastrophic failure thru failure to clear fault arc in time |
accumulated I2T duty of main
contact; |
|
|
dielectric compromise |
partial discharge; |
result in longer arc clearance times |
partial discharge RF currents |
|
|
|
mechanism problems |
higher operations count; |
gradual (degradation) or sudden failure of breaker |
interval between trip command (or “a” contact opening) and “b” contact closing |
7 |
|
|
control circuit failure |
shorted or open trip coils; |
corrosion or gradual wear leading to insulation breakdown (short) or wire fatigue (open). Either way, trip circuit will malfunction. |
self-monitoring tripping relay; |
8 |
|
|
Cables |
outer sheath damage |
water ingress |
growth of “water trees” in solid insulation
(XLPE); |
increasing cable capacitance; |
9,10 |
|
thermal cycling with load |
inadequate joint clamping; |
high conductor stresses, conductor twisting and/or
joint damage; |
internal optical fibre; |
10 |
|
|
paper degradation |
water ingress; |
accelerated failure of primary insulation resulting in complete cable failure |
partial discharge RF currents; |
10 |
|
|
oil degradation |
similar mechanism as within transformers oil distribution problems in cable |
localised partial discharge promoting further oil (and paper) deterioration |
partial discharge RF currents |
||
|
Substation batteries |
thermal cycling |
mechanical fatigue of seals and internal linkages |
failure of sealing; |
open circuit (internal inks fail); |
|
|
loss of electrolyte |
improper float charging; |
pressure build-up opening battery chamber to
atmosphere; |
battery impedance rises with time as battery capacity falls |
||
|
plate growth |
overcharging |
plate buckling; |
battery impedance rises with sulphate build up on plates |
||
|
short circuits |
build up of sludge |
plates progressively shorted together |
reduction in battery impedance |
|
Note |
Comment |
|
1 |
In future, on line acid monitors may be developed |
|
2 |
Core temperature monitoring can be performed non-invasively by measuring tank oil temperature probe (in thermowell) and load current and applying a standard algorithm to calculate core temperature surprisingly accurately. This is a very cheap and effective retrofit option. |
|
3 |
Systems to perform on-line frequency response analysis are currently under development. |
|
4 |
A bushing capacitance will increase if any two adjacent equi-potential foil layers short together, since effectively there is now one fewer "capacitors" in series formed by the foil layers. This increases the C1 (conductor to tap point) capacitance and hence will allow a higher capacitive current to flow. If C1 deviates more than 5% from the manufacturers nameplate C1 value, the bushing should be monitored closely for trends and on-line tandelta monitoring should be seriously considered. If C1 deviates more than 10% the bushing should be replaced. |
|
5 |
DGA of oil in bushings can be (as is) performed but does not always give a reliable indication of fault development. Part of the problem is that oil does not flow in a bushing as it does in a transformer, hence getting a representative oil sample is very difficult. Gas tends to remain trapped in the tightly wound construction of the bushing near the developing fault site and will not necessarily migrate to the point where the oil sample is taken. |
|
6 |
In rare circumstances, very large temperature variation may exceed bushing design limits. Main problem with thermal cycling is oil O ring wear and crazing damage to outer porcelain shell. Possible bubble formation in the oil was a concern in the GE Type U (printed ink) bushing [30]. |
|
7 |
Monitoring the "a" and "b" contact timing can show up changes in operating characteristics of the trip coil, latch, and mechanical components with a current signature used to detect other faults. |
|
8 |
Healthy trip monitors work by continuously passing a small current through the tripping circuit and therefore confirm the circuit is not open circuited. This may or may not detect a short circuit fault in the control wiring, depending on the system and the nature of the short. This should also monitor the status of the tripping circuit fuse. |
|
9 |
Older XLPE (more than ten years old) may contain manufacturing defects that give rise to partial discharge. Newer XLPE cables from quality manufacturers exhibit very little partial discharge activity, but their accessories (joints and terminations) certainly do. Unfortunately, partial discharge is not currently a viable method for reliable detection of water treeing. Water trees give very little if any partial discharge signature. Partial discharge will occur briefly in the final phases, as the outer branches of the tree approach flashover distance. To detect water tree growth on-line will require on-line tandelta monitoring, which will spot the increases in the dielectric losses. The authors are unaware of a commercially available on-line tandelta monitoring system at this time, but there is no theoretical reason why such a system could not be successfully developed. |
|
10 |
Partial discharge is usually noted at joints and terminations. Continuing partial discharge can progressively destroy insulation through carbon tracking and the creation of corrosive by-products. |
|
Please note: |
The information in this table and the accompanying notes has been collated as the result of an extensive literature review. Notification of any suggested improvements this table (including correction of any errors or omissions) would be gratefully received. The intention is to continue to develop this table further and make it available on the Lord Consulting web site (www.lordconsulting.com) in due course. Please send any suggested changes to: graham@lordconsulting.com |
Generally, the first question that gets asked of those proposing on-line monitoring is: "what does all this cost ?"
A business case justification can, and should, be performed on a case-by-case basis for each deployment of on-line monitoring under consideration. The drivers that need to be considered are receiving increasing attention and can be broadly analysed under "direct" and "indirect" costs and benefits to the network business [27, 28].
However, there is a problem with constructing business cases for on-line monitoring not normally encountered. This lies with the variability encountered in the result obtained from multiplying together a very large number (the full cost to the network and its customers of the consequences of major failure) and a very small number (the likelihood of occurrence of such a failure in a nominated time period). Slight variations in either number could give virtually any outcome desired. Hence, the business case is vulnerable to manipulation.
Careful analysis with meaningful data can yield useful results. Provided the full costs of failure are factored in, these results serve to give a feel for the orders of magnitude of the costs and payback periods. Network managers should be cautioned not rely solely on detailed financial arguments to make decisions to deploy on-line monitoring. The business case analysis needs to be performed within a wider framework of the desire to improve the long-term network health on behalf of the community served. It is therefore a true governance decision, not solely a mechanical management decision.
Governance considerations by both Boards of Directors and senior management should be made in the light of all the factors listed in Table 1 (Section 6: Strategic Drivers Towards On-line Monitoring). The financial view should underpin a wider discussion that considers the core purpose of the organisation and what the wider community expects. In this regard, there is nothing wrong in principle with formulating the position of the organisation in consultation with community and business leaders. The incremental cost of on-line monitoring systems may well be regarded as trivial alongside the full cost of unexpected and prolonged outages.
Over 75% of the investment in the electricity sector is beyond the meter. In most cases, the electricity supply industry does not feel the economic cost of supply interruption. This is expected to progressively change and network companies would do well to get in early and stay ahead of the issue. Problems that previously may have caught the asset operator unawares will in future be detected early, permitting a full "health plan" for the asset concerned to be developed and implemented quickly.
The onus of proof will increasingly lie with network managers to prove they are following best practice in this area, and best practice will constantly improve. Networks should also anticipate their on-line monitoring programmes to be open to inspection from agencies acting in the best interests of the surrounding communities who stand to lose the most from the disruption of sudden asset failure.
The biggest challenge for electricity network managers the world over in the next 25 years will be to keep the present asset base operating reliably. Wholesale replacement on the basis of age alone is obviously out of the question. All assets are being expected to work more and more productively and they are all busy getting older while we scratch our collective heads to figure out how long they can ultimately live. It is unavoidable that the result will require a perpetually and careful balance of targeted nurturing, maintaining and replacing.
For some years now, the focus has been moving away from "determining asset life" (which is really unknowable) towards "continuous condition assessment". On-line monitoring takes us ahead one step further, making our assets monitor themselves and therefore that much smarter. In doing so, we enable them to tell us how they are feeling and their internal "expert systems" can advise us as to what to do should things start to go wrong. This is an elegant answer to our biggest problem of all: there are simply not enough asset life experts to go around.
On-line monitoring is becoming an integral part of the new landscape. It should be considered as setting a new benchmark for the electricity supply industry worldwide in the years ahead.
[1] "An international survey on failures of large power transformers in service" Final Report of Working Group 05 of Study Committee 12 (Transformers).
[2] "Australia / New Zealand transformer reliability survey 1996". Western Power Transmission Projects Branch.
[3] "Transformer Failures", Section 7, EEA/EA Technology Travel Award 2000, Ragu Balanathan, Summary Report.
[4] "Failure of large oil cooled transformers", IMIA 16-66 (96) E. Working Group of the 29th conference of the International Machinery Insurer's Association, September 1996.
[5] "Huntly power station transformer incident - why it was catastrophic", Harvey O'Sullivan, Tri-Sheras Ltd, EEA Conference, Auckland NZ, June 1999.
[6] "Transformer technology at a glance", Graeme Fincher, Fincher Consultants, NZ. Paper presented at the 3rd AVO NZ Third International Technical Conference, October 15-17, 2002, Methven, New Zealand .
[7] "Transformer life management", V. Sokolov, II Workshop on Power Transformers-Deregulation and Transformers Technical, Economic, and Strategical Issues, Salvador, Brazil, 29-31 August 2001.
[8] "Transformer risk assessment considerations", Sokolov, Bassetto, Mak, and Hanson, EuroTechCon 2002.
[9] "Life management techniques for power transformers". Prepared by CIGRE WG A2.18, 20 January 2003.
[10] "Transformer and LTC temperature monitoring ... field experience and updates", Gerald Lucak, Weschler Instruments, USA. Paper presented at 3rd AVO New Zealand/LORD Consulting Third International Technical Conference, Methven NZ, October 15-17, 2002.
[11] "Condition monitoring of high voltage electrical equipment (with an emphasis on transformers)", Ron Park, Park Consultants Ltd. Paper presented at 3rd AVO New Zealand/LORD Consulting International Technical Conference Methven NZ, October 15-17, 2002.
[12] "Preventative to predictive … the future of transformer oil testing ", William Morse, Morgan Schaffer Inc, Canada. Paper presented at 3rd AVO New Zealand/LORD Consulting Third International Technical Conference, Methven NZ, October 15-17, 2002.
[13] "Analysis of IED and instrument generated dissolved gas data prior to and after the repair of a critical transformer", Richard Clark, Trilok C. Garg, Mirant Mid-Atlantic, LLC, and Richard Bérubé, Morgan Schaffer Systems Inc.
[14] "The theory and application of power factor testing", Mr Rick Gaskey, Megger Limited, USA. Paper presented at 3rd AVO New Zealand/LORD Consulting Third International Technical Conference, Methven NZ, October 15-17, 2002.
[15] "Bushing failure rates and mechanisms", J Stead. Weidmann 2002 LV Conference Presentation.
[16] "500 kV bushings failures and bushing sampling program" Mike Lau, BC Hydro, 2001
[17] "On-line monitoring of high voltage bushings", Sokolov and Vanin, Proceedings. Of the 1995 International Conference of Doble Clients Sect 3-4.
[18] "Comparison of a new technique for power factor measurement at rated voltage with the standard off-line test on bushings and HVCTs", EPRI Project Opportunity-Substation Operation and Maintenance, February 2003
[19] "Managing high voltage current transformers and bushings using on-line insulation monitoring techniques", Terry Krieg, ElectraNet SA, and Jeff Benach, AVO International, TechCon Asia Pacific, 2002.
[20] "Innovations in hardware and software architecture applied to on-line monitoring of strategic HV substation assets", Terry Krieg, Power and Water, Darwin. Paper presented at the 3rd AVO New Zealand/LORD Consulting International Technical Conference Methven NZ, October 15-17, 2002.
[21] "Preparation of guidelines for collection and handling of reliability data". CIGRE Joint Task Force 23/12/13/21/22-26, Final Report September 12th, 2001.
[22] "Circuit breaker maintenance, sentinels on guard", Fred Tanguay, Article on The Electricity Forum website, 16 September 2003.
[23] "Experiences with continuous partial discharge monitoring", Michael Webb and Mark Lomax, MW Test Equipment Ltd. Paper given at AVO NZ 3rd International Technical Conference, October 15-17, 2002, Methven, New Zealand.
[24] "Some developments in diagnostics for high voltage cables", Dr Ross Mackinlay, High Voltage Solutions, UK. Paper presented at 3rd AVO New Zealand/LORD Consulting Third International Technical Conference, Methven NZ, October 15-17, 2002.
[25] "Partial discharge testing... experiences from the field", Peter Rhodes, Delta, Dunedin, New Zealand. Paper presented at 3rd AVO New Zealand/LORD Consulting Third International Technical Conference, Methven NZ, October 15-17, 2002.
[26] "Operations - extreme loading", Mark Janick, "Electrical World T&D" magazine, Nov/Dec 2001.
[27] "Risk assessment of power systems assets ...the insurance perspective", Paul Boman, Hartford Steam Boiler, USA. Paper presented at 3rd AVO New Zealand/LORD Consulting Third International Technical Conference, Methven NZ, October 15-17, 2002.
[28] "Successful economic justification of on-line monitoring systems for power system assets", Terry Krieg, Power and Water Corporation, Darwin, Australia.
[29] "On-line HV insulation condition monitoring…substation on-line system", RK Fricker, Power Vision Programme, CSIR, South Africa. Paper presented at the 1997 AVO International Technical Conference, Dallas, Texas, September. 14-17, 1997.
[30] "Bushing Presentation", John F Leech. Presented at the 1997 AVO International Technical Conference, Dallas, Texas, September. 14-17, 1997.
[31] "What are Water Trees?", from USA Wire & Cable, Inc. technical paper on their website. (http://www.usawire-cable.com/techpapers/water_trees.html).
[32] "The Effects on Winding Clamping Pressure Due To Changes In Moisture, Temperature and Insulation Age", by Tom Prevost (EHV-Weidmann, USA), David J. Woodcock (Weidmann Technical Services, Inc., USA) and Christoph Krause (H.Weidmann Ltd., Switzerland).
[33] "TanD Cable Testing", Power Transmission and Distribution magazine, Dec/Jan 2003.