Conditions Have Consequences

Regardless of how condition score is tallied, the qualities most often sought in a condition measure are that it is simple and easy to use, and that it relates to:

Ability to do its job: to serve demand, support a certain weight under certain wind loadings, break a certain amount of currently within a certain time, etc.

Durability: the expectation that it can continue in its current role, or not, for a foreseeable time: remaining service life, expected failure rate, etc.

Actions required by the owner/operator: run at full rating, or reduce loading by 20%, or withdraw from service immediately. Inspect every three years or inspect annually. Service this unit immediately.

Conditions have consequences, either with regard to operating capabilities and restrictions, or actions needed to be taken (or not taken) for the equipment (Figure 11.1).

Figure 11.1 Relationship of state, conditions, and consequences and actions as it relates to the management of electric T&D assets.

Condition, State, and Priority

State is a technical description that is unambiguous in context. In theory, state refers to a set of data that comprehensively defines a system so that its response and behavior can be completely understood and predicted. From a practical standpoint, state usually involves several measures that are needed to understand and define the status of a unit unambiguously for modeling, analysis, and prediction.¹ Often these state factors are component related: the “state” of a power transformer involves its core, bushings, oil, case, LTC mechanism, and cooling systems, etc. Failure or poor status of any one could lead to condition assessment of poor – needs attention now.

Condition assessment is ambiguous as to the details. State is not. A condition of “poor – needs attention now” might mean that oil readings are outside of permissible range, or that DGA readings indicate the core is deteriorated, or that the case is rusted and leaking, or that bushing diagnostics indicate a developing problem. Condition ratings are ambiguous in this situation, but the point is that these details are not needed to address the three managerial aspects listed above. By contrast, the state assessment or model will provide all those details: the core is like this, the oil like this, the case like this, and similarly for the bushings, LTC, cooling, and all other equipment associated with the unit. Even for a commodity unit such as a pole, state might give more details behind the “condition.” (This pole is located in an acidic soil area. This pole is in a high risk location.)

Condition is unambiguous as to actions needed; state might be too complicated to immediately determine that. But the value a simple condition score has is that it is unambiguous as to management action required – attention is needed now. Condition is the most succinct and unambiguous scoring possible that achieved that one quality, and that is all that should be expected of it. The first step in the attention the unit receives will almost certainly be for someone to gather up the state estimates and the data itself, and examine them in detail, but the condition score has done its job at that point. An important distinction is that modeling and prediction should always be based on state, not condition, as illustrated in Figure 11.2. Condition, by its nature, is not as useful for modeling and prediction as state. State, by its very nature, is not as useful a summary of management issues as condition.

¹ Model, as used here, is not necessarily a computer program or even a numerical representation. It could be only the concept used to predict behavior: “My boss is a stickler for detail and abrupt in the morning, particularly on days when she has meeting with her boss. She is easier going in the afternoon, particularly when she is tired. She can be distracted from an unpleasant topic by bringing up the Duke women’s basketball team. . . .”

Figure 11.2 Methods that make predictions and analysis based only on condition tend to do much poorer as planning and management tools than models that use state representation. See text for details.

Priority is Used for Optimizing or Trying to Optimize of Results

Priority, or prioritization grade is a rating of importance with respect to spending and budget. Usually priority is based on both condition (often including details depending on state) and a unit’s importance (reliability, grid systems security or phooey customer service needs). Units that score highly in terms both condition needs and high importance are given priority over those that do not. Figure 11.3 shows a prioritization matrix as used by one utility. Prioritization will be discussed further in Section 11.10.

Figure 11.3 Prioritization is based on condition (or state) and importance (potential impact on reliability, system interconnected security, and operating results). This matrix summarizes one utility’s representation of the priority it gives to equipment based on both evaluations, and is used by it to communicate how it manages its assets. Equipment in poor condition is withdrawn from service regardless of criticality. Critical equipment that is of concern is given high priority of attention. A detailed numerical prioritization that also takes into account the cost of actions and the cost of consequences (good and bad) is actually used to make all final decisions on what to do and where to focus resources.

A Data Mining Perspective

A modern term for analysis and use of very large data sets and their conversion to useful knowledge is data mining. It can and is described in various references in slightly different ways, but regardless is always a process that takes data all the way to useable knowledge, usually in stages. Figure 11.5 shows the authors depiction of it based on several common web and technical references (Wikipedia among them). Four techniques are used to develop knowledge from data, usually in stages. These are similar overall to the analysis-to-knowledge path described above. .

Clustering is finds patterns or groups of assets that are similar to one another, etc.: “This set of transformers has operating and histories that are very similar and distinctly different than average.” “Failure rate of an OH service transformer is 20% higher in the three months after a lightning outage on the line within a quarter mile of its site.”

Classification seeks to identify groups and patterns and put them in context with regard to the framework needed: “We have two groups here: wet transformers have both high rust rates and water-in-oil contamination rates; dry transformers have neither. There are very few transformers not in one group or the other.”

Regression attempts to find a relationship between variables: “Annual failure rate is equal to this function of time in service, annual peak load, climate area code and manufacturer code.” Readers should note that this term does not mean just the classic multiple linear regression method of polynomial curve fitting of data (it is among the simpler and more direct of dozens of methods used): Regression includes all methods that seek to identify a functional relation among variables, and includes grade logic methods: “When all factors are GOOD, the problem rate for the next year will be LOW, MOST of the time.”

Rule Development uses patterns, classifications, and relationships found by clustering, classification, and regression to build applicable rules and guidelines. “Wet transformers over 15 years in service should be inspected for rust annually and the oil tested for contamination bi-annually.”

Data mining is not just a term but a professional practice area with established theory and rigorous methodology that, if followed correctly, generally guarantees good results, or at least freedom from the effects of many common mistakes and oversights that are often made in data analysis. An example of the methodology is relational normalization, which does not refer to numerical normalization (i.e., adjust all values to a scale of 0-100), but to methods that adjust the structure of the database and its rules to avoid duplication, inconsistent updates, conflict among different parts of the mode, and error, and logical completeness which assures that the resulting model covers the range of expected application fully, without gaps.

Transformer Aging and Deterioration

Transformers “age” in the sense that the strengths of their components deteriorate as a function of chronological time, time in service and loading, and due to severe abnormal events like through-faults and switching surges. The amount of load served has a great deal to do with the rate of deterioration. Chapter 7 discusses in detail aging and loss of life with service time.

Many aspects of a transformer deteriorate with time: 1) the insulation level of its windings; 2) its oil and its bushings; 3) the mechanical strength of its core stack and internal bracing and electrical connections; 4) the desired chemical and physical properties of its materials (anti-oxidants in the oil, corrosion resistance of paints, flexibility of insulation, etc.). However, deterioration of the insulation (of both winding and the oil itself) is the major area of concern, and the predominant area evaluated, to determine a transformer’s “condition.”

There are several reasons for the focus on internal insulation strength. First, a core winding insulation failure will not only prevent the transformer from doing its job, but probably lead to a large fault causing very severe damage if not catastrophic failure. Bushing failures, corrosion, and other types of deterioration can lead to failure of the device, but rarely cause such high levels of damage.

⁵ See the Power Distribution Planning Reference Book, Chapter 7.

Secondly, many other defects, when found, can be repaired or replaced rather quickly and inexpensively. Bushings can be replaced, tap changers repaired, and most non-core/winding related equipment serviced while the unit is left in place but de-energized. However, the winding insulation is both the most difficult and the most expensive item to repair in a transformer. The unit must be withdrawn from service for a long period of time and returned to a factory or refurbishment center for what is essentially a complete rebuild.

The insulating oil in a transformer, if significantly degraded, can be reconditioned or replaced. This will improve the oil and have a positive effect on the condition of the winding cellulose insulation. But in some cases the winding insulation is sufficiently deteriorated so that replacing the oil alone will not necessarily make a satisfactory improvement in the transformer as a whole.

For all these reasons, the main focus of power transformer condition assessment is on insulation strength, the various tests focusing on direct (measures of dielectric strength) and indirect (measures of contaminates, dissolved gas analysis) means of determining the winding and oil insulation strength. However, other portions and aspects of the unit should be checked regularly and maintained as needed.

Acceptance Tests

New units are usually tested for insulation resistance, dielectric strength, turns ratio, amount of losses, and run through AC hi-pot and power factor tests. They may be subjected to load cycle (thermal tests). Acceptance tests are not a factor in condition assessment of older equipment and will not be discussed here.

Routine Transformer Inspection

Routine physical inspection of transformers includes : Examining the exterior of the unit for signs of leakage or corrosion of the case; examining radiator joints and grounding straps, etc., for cracked or dirty bushings, for damage (vandalism, weather), deterioration (paint chalking), and loosening of brackets or seals for attached sub-components (radiator fans, pressure sensors). It should also include testing the operation of ancillary devices such as oil pumps, fans, pressure relief valves, and the tap changer mechanism and its control system. This level of inspection should be done annually, if not on a more frequent basis.

Overall, inspection tends to focus on aspects of the transformer other than insulation strength and winding condition. Inspection may include viewing the unit through an infrared scanner to enhance detection of hot-spots caused by loose terminations of incoming lines, leakage current, or other similar developing flaws. Similarly, it can include enhanced “audible noise” analysis using a spectrum analyzer and pattern recognizer programmed to identify acoustic signatures indicative of trouble.

Routine inspection should include a thorough review of all operating records (temperature, pressure, load records) since the last routine inspection, with particular attention paid to abnormal events such as nearby lightning strikes, through faults, and storm damage.

Depending on the results of these inspections, follow up tests or diagnostics aimed at determining more precisely the condition and any necessary maintenance may be required.

Routine or Periodic Tests

Along with routine inspection, certain tests are done on a periodic basis. These tests have two uses:

1. “Good to go” evaluation. The tests determine if any measured values deviate from their permissible or recommended ranges. This indicates a need for further testing or for maintenance, and caution in using the device at high loadings. Values above a certain event lead to recommendations that the unit be withdrawn from service.

2. Condition monitoring. Periodic values are compared to identify long term trends – a particular value (e.g., dissolved methane in the oil) could still be below the maximum permitted – but significantly higher than its value in the last test, indicating a possible problem just beginning to develop.

Many of the most common periodic tests and diagnostics for a transformer focus on measuring some aspect of insulation strength, or look for the products or signs of insulation and/or oil deterioration. Table 11.2 lists the periodic inspections and tests that are done on power transformers. The frequency of tests shown are the authors’ impressions of typical industry practice (not necessarily recommended practice) with regard to units that are not thought to be high risk. Units that are suspected of having problems or a high potential for failure should be checked much more often – in extreme cases on a daily basis if not continuously (with real time monitoring).

Two guides useful sources of information on transformer tests and their use are Fifty Years – A Guide to Transformer Maintenance by Myers, Kelly and Parrish (1981), and Service Handbook for Power Transformers, by ABB’s Transformer Remanufacturing and Engineering Services Division (2006).

Four Categories of Risk/Recommended Power Transformer Evaluation

IEEE Standard C57-1004-1991classifies oil-filled power transformers into four categories or conditions of risk based on the results of dissolved combustible gases tests in their oil. Within each of the four categories, the standard then recommends specific inspection and test intervals that vary depending on the rate of change of the gas content as measured from interval to interval. It also contains recommendations for use of units and if and when they should be tested again and if they should be removed from service, based on the gas test results.

Table 11.2 Inspections, Tests, and Diagnostics Performed on Power Transformers

Table 11.3 IEEE Standard C57.104-1991 Power Transformer Risk Conditions

Table 11.4 Summary of Risk Categories and Recommended Actions Based on TCGA Tracking, from IEEE Standard C57.104-1991

Table 11.5 Limits to Amounts of Individual Gases in Each Category

The total dissolved combustible gases total does not include CO₂, which is non-combustible.

The full standard gives more detailed breakdowns of recommended testing frequency versus TDCG totals and increase rates than shown here. Table 11.3 summarizes the meaning of the four categories, which are defined by the totals shown in column two of Table 11.4. The table also shows how rate of increase data is interpreted in each of the four categories. Table 11.5 gives the concentrations of various gases (which must be determined by DGA) that are considered tolerable within each category.

Thermal Load Tests

The electrical losses created inside a transformer produce heat, which leads to elevated temperatures unless the heat is moved away from the core with some means of cooling. If the unit has internal problems that create higher than expected amounts of heat, the unit might operate at temperatures noticeably higher than it should for a given load. Sources of excess heat include short circuits, severe sludging problems, damaged radiators or damaged pumps. This could shorten lifetime significantly and bring about premature failure.

Units can be “laboratory tested” to determine their temperature versus load characteristic by running test currents of known amounts through them for known periods of time, while measuring their internal temperature rise. However, such tests are expensive (the unit has to be de-energized, taken to the lab, and left there for a considerable period).

Nearly the same accuracy in determining thermal behavior of a transformer can be carried out in the field. This is done by monitoring load, ambient temperature, and temperature inside the unit on a periodic basis (every 15 minutes) and then calculating the thermal time constant and the maximum rise expected at full load from the resulting data (Figure 11.6). The actual computation is straightforward, although it is generally best to gather much more data than theoretically sufficient – several weeks of readings. Use of an analytical method based on signal processing, rather than time-series analysis methods, is also recommended.

This computed “thermal response” signature of the transformer will reveal whether the unit is performing up to its design specifications or if thermal capability is degraded for some reason. Typically, such field tests will not indicate the cause of the problem, only that a problem exists. For example, Figure 11.6 gives no indication whether the deviation from expected performance is due to a degradation in cooling capability (due perhaps to sludge buildup or blocked cooling ducts), or to creation of greater amounts of heat inside the unit due to a defect such as an internal fault.

The causes of thermal problems have to be traced through other inspections and tests, some of which will be prescribed based on the initial, basic test results (see the aforementioned books). These begin with an inspection and test of the entire cooling system – pumps, fans, radiators. If these are fine, then tests for sludging, such as acidity and interfacial tension tests of the transformer oil, are recommended.

Figure 11.6 A thermal load “test” of a power transformer. Top, electrical load and temperature rise above ambient are monitored over a 24-hour or longer period. Bottom, de-convolution of the two time-series produces a calculated impulse response function (temperature rise and time for an arbitrarily short pulse of loading) for the unit. This can then be used to compute the step response shown, which is the rate of rise and final temperature that would result from having the unit serve a load equal to its rating on a continuous basis. This example reveals an as-tested signature that both have a higher rise time and asymptote (high eventual temperature), indicating some type of internal problem.

Circuit Breakers

Circuit breakers are electromechanical devices. They are tested in both mechanical and electrical performance and for signs of deterioration.

Switchgear

Switchgear other than breakers is generally tested and inspected on the same periodic basis as breakers. Switches and other mechanical devices should be routinely (semi-annually or annually) operated. Unlike the situation with breakers, in many cases visual inspection of contacts and mechanisms is possible with much of this equipment.

• Routine electrical tests include only the insulation resistance tests. They are relatively easy and quick to apply in the field and put no particularly high stress on the equipment. If they indicate problems other tests might be called for. These include hi-potential tests and power factor tests.

• Radio interference is a very good way to identify problems due to dirty or cracked bushings, post insulators, and corrosion-caused weaknesses in connections. As mentioned earlier, a small AM radio can detect the interference given off by some types of electrical problems.

A detailed discussion of breaker and switchgear inspection and testing procedures, including comprehensive lists of recommended procedures on a step-by-step basis and a method for a detailed RIV (radio interference voltage) test of substations is included in Electrical Power Equipment Maintenance and Testing, Chapters 7 and 8 (Gill, 1998).

Underground cables

Underground cable is available in a wide variety of types, including three-phase, single phase, either carrying or not carrying a concentric neutral, and with various types and degrees of “armor” or protective sheathes. Cables are made from a diverse range of materials, including copper, aluminum, and alloys for the conductor. Insulation and construction may be composed of any large set of materials with somewhat differing properties, and different construction methods, including paper, cotton derivatives (varnished cambric), natural or synthetic rubbers, and other materials. Cables also vary in the type or material and construction of their outer sheath.

With all of these types of materials and construction, there is no one type of inspection or testing program that is applicable to all types of cable. Each manufacturer provides information or recommendations. Table 11.6 lists general categories of inspection and testing and the typical periods of their application.

Fault Location

When an underground cable fails due to a fault or open circuit, various test procedures can be used to try to locate the fault. The traditional method is a manual sectionalization to isolate the fault, section by section, which is tedious and time consuming, and often abuses equipment.⁶ Other methods include the radar method and arc-reflection method (two forms of time reflectometry) and a resonance method in which electrical signals or pulses are tracked down the cable, pulse return timing or resonant frequency indicating distance to the fault. Several methods inject a signal or pulse that excites some type of electromagnetic signal from the fault location, which can then be found using a portable location (this only works if it is possible to walk above the cable route).

Fault location is not part of “condition assessment” of cable operating in normal fashion, and will not be discussed further here. A very comprehensive discussion is included in Malik et al.(1998).

Damage, Not Deterioration, Is Often the Major Concern

A majority of power distribution in most utility systems consists of overhead primary (MV) and overhead service, or secondary (LV) lines and equipment. The basic materials for overhead lines include poles, conductor, insulators, and their associated hardware, all of which are incredibly robust equipment. Their expected lifetimes in normal service are 50 years or greater. While deterioration is a concern, the very slow rate of change under most conditions and the very great times to failure for this equipment generally mean that routine monitoring to track the rate of deterioration is not an effective or justifiable O&M procedure.

⁶ An expedient for finding faults in URD systems at many utilities involves “opening and closing” URD primary sections by pulling elbow connectors with a hot-stick. This is not recommended and most elbow connectors are not designed or rated for such duty. The stress put on the elbows is unknown and cumulatively could lead to early failure.

This is not to say that condition assessment is not an issue in overhead distribution. Far from that, assessment and tracking of condition should be a high priority in many distribution systems throughout the United States, particularly in those where large portions of the system exceed 50 years of age, an age where deterioration of wood, metal, and bell insulators may have reached a point of concern. Condition assessment of overhead equipment in these areas is a critical part of good reliability management.

But one definite fact that colors all approaches to inspection and maintenance of overhead lines is that by its very nature, overhead equipment is quite exposed both to the elements and to external sources of damage such as errant automobile traffic, nearby vegetation and fires, etc. A majority of overhead outages occur due to damage, not deterioration, whether that damage is from severe weather, contact with trees, automobiles and construction accidents, forest and building fires, and vandalism. For these reasons, identification of damage, not deterioration, is a large part of inspection and testing for overhead lines.

Inspection of Overhead Lines

Generally, elements of the primary and service voltage systems are not tested with electrical tests like insulation resistance, power factor, hi-potential etc., as are substation equipment and UG cable systems. This is not to say that they could not be, and that such tests have not been done on a special or experimental basis. Such tests aim to identify deterioration of insulation, and as mentioned above, deterioration is typically not a big concern on overhead lines; damage from external causes is. But while such tests could perhaps identify some types of damage, visual inspection of overhead lines can reveal a great deal more. Visual inspection, done to a high standard and organized with good record keeping will identify damage and in fact can detect signs of deterioration in many elements of the system. Typically, the best procedures utilize written classification criteria (“Less than ten square inches of visible rust per side” “Scaling rust flaking off”). Such routine inspection might be augmented using infrared imaging to identify hot equipment and fittings, and/or the use of developing acoustic/radio frequency detection devices that find traces of leakage currents, etc.

Therefore, inspection is the chief means of determining overhead system condition. The only electrical test routinely done on overhead distribution involves checking the grounding of poles and neutrals in grounded and multi-grounded Y-connected distribution systems.

Poles and Pole Assemblies

Poles are made of natural wood (treated, straight tree trunks), laminates of wood, composite wood (wood epoxy mixes), concrete, fiberglass, or steel. Concrete, fiberglass, and steel poles have very long lifetimes, if their materials are properly manufactured (cured, etc.).

The vast majority of power distribution poles in use are wood. Wood poles have an expected lifetime of from 35 to 85 years, depending on the type of wood, chemical preservative and treatment method, installation, soil and climate conditions, and periodic inspection and treatment given them. Cross-arms have a similar lifetime. Deterioration takes two forms: 1) a general weakening of mechanical strength of the entire pole, due to aging, cracking, etc., and 2) rotting of the pole near ground level due to trapped moisture. They are also subject to damage from automobile accidents, trees falling, fire and all manner of unusual situations (a property owner nearby may use the pole as one corner of a garage or barn being built, etc.). Deterioration, like damage, tends to be very pole-specific. One pole can be near failure due to deterioration; those nearby might have a considerable lifetime left.

Visual inspection is notoriously unreliable in determining the state of deterioration of a wood, although it is very good at identifying damage to poles and their assemblies. Testing for mechanical strength and rotting takes several forms. The simplest, but least exact, is the hammer test. The pole is struck with a large hammer near ground level. The resulting “thump” has a very different sound depending on whether the pole is sound or rotted. This test is very quick (many poles can be checked per person-hour and inexpensive (the equipment, a standard small sledge, costs very little). However it is not particularly quantitative. It tends to identify poles with severe stages of deterioration, but does not measure interim levels of deterioration. In addition, dependability of results has a good deal to do with the skill of the person doing the test.

A type of sonogram approach uses a device that sonically scans a cross-section of the pole (usually near ground level) to determine its density and detect any voids within. The sonogram device is reasonably portable (about the size of a small rug vacuum cleaner) and is taken from pole to pole. This test is more expensive than the hammer test, and requires more operator skill, and a good deal more time for each pole than the hammer test, but it is much more accurate in determining condition on a pole. Coring to obtain a sample of interior wood for analysis is also often used.

A type of indention-mechanical/electric resistance test can also be used. A portable device about the size of a small suitcase is attached to the pole near ground level, using a web belt to tighten it to the pole. The device then pushes two thin electrical probes into the pole, measuring the mechanical resistance to their insertion, one measure of the condition of the pole. The electrical resistance through the pole is then measured with the probes, using an ammeter. The combined mechanical-electrical values can be correlated with the pole’s mechanical strength (different tables are needed for each type of wood, e.g., fir).

Deteriorated poles can be repaired by injection of strengthening and/or drying materials into them near the base or by reinforcement with sheathes or stub-poles. The most effective long-term approach however, is replacement.

Inspection of Cross-arms, Insulators, Conductors and Hardware

Overhead line equipment includes cross-arms, mostly wooden, which can deteriorate much like poles can, insulators and attachment hardware, and conductor and associated hardware like splices and vibration dampers.

• Careful visual inspection of a distribution line, often called “driving the line,” can identify damage of cross-arms, hardware, conductor, and deterioration if it has reached advanced stages (near failure). One factor in favor of such inspection is that it is inexpensive, requiring no special equipment. However, it fails to find some incipient problems (in fact, a majority may go undetected), and it is not easy to carry out on sub-transmission or distribution lines that do not parallel roads or highways.

• Radio interference tests. As mentioned at the beginning of this chapter, a simple and inexpensive enhancement to visual inspection of electrical equipment exposed to the elements is radio interference monitoring. Routine policy at several utilities is to leave a radio on the AM band but not tuned to any station while driving line routes for visual inspection of equipment. Cracked insulators, corroded brackets, and broken conductor strands create low levels of radio interference, so a sudden increase in radio “static” and noise is a sign of a nearby problem. Generally, inexpensive analog radios work best for this purpose: expensive stereo systems and digital radios have filters and signal enhancement circuits that eliminate all or most static and interference.

• Infrared scanning. Two improvements in overhead line inspection are infrared scanning and inspection by helicopter, often combined into a much more effective and rapid means of inspection lines. Infrared scanners are essentially infrared video camera-recorders that display on a CRT an image of equipment as seen in the thermal infrared band. Intensity of items in the image is proportional to temperature, so brackets, splices, or conductor sections that are overheating for whatever reason show up clearly.

• Inspection by helicopter provides several advantages. First, a number of important deterioration modes are visible from the air better than from the ground. Cross-arms crack and rot from above where standing water tends to collect after rains. They may look fine from below but can have signs of serious deterioration if viewed from the air. Loose attachments of conductor to insulators are easier to see when the insulator does not block view of the tie point, as it does if viewed from the ground.

Second, in all but very heavily congested urban areas, a helicopter can “fly” line routes while staying close enough to visually inspect in detail. Often, it is possible to approach closely enough that service-level equipment and lines can also be checked.

Third, the helicopter can cover much more line per hour than inspection by foot or automobile. Helicopter surveys often use a video recorder in both visual and infrared bands to record the inspection for later review in detail as needed.

Replacement and Repair of Basic Line Equipment

Damage to or deficiencies in basic overhead line equipment found by inspection can be repaired on an as-identified basis by replacement of the failed or failing components. Little of this equipment is “repairable” in the sense that it can be taken down, reworked or refurbished, and put back up. Instead, replacement with new is the only alternative. The exception is conductor. Birdcaged conductor, broken strands, and similar damage can be repaired by replacement (splicing) of only the damaged part of the conductor.

One recommended exception is annealing. Experience has shown that failures or severe deterioration due to annealing at one part of an overhead line will generally indicate severe deterioration in other parts nearby. Ideally, if a conductor falls down or other problems from annealing are suspected, the entire segment should be replaced or carefully inspected at close range (from a bucket truck) for signs of annealing.

Switches and Cutouts

A portion of manually operating switches in many electric distribution systems are suspect simply because they have not been operated in years. Over time, switches will become inoperable for a number of reasons. Even a minor amount of corrosion on the contacts of closed switches, or in the operating mechanism of open or closed switches, will freeze the switch so it cannot be operated. Water working its way into mechanisms can freeze, warping rotating surfaces and jamming devices. Closing a switch into a current higher than its rating can “weld” its contacts closed, so it will not open next time an attempt is made to operate it. Overloading can reduce the temper in the switch springs, causing it to malfunction. In addition, connections sometimes just weaken or loosen.

Cutouts and fuse connections also suffer from the same types of problems as line switches. However, they are almost never test-operated. There is some likelihood that each will go bad, but given their number, the difficulty in reaching them, and that fact that any quick test would assuredly cause a customer interruption,⁷ few utilities test anything but a handful of cutouts, and only at special sites, each year.

⁷ To test without an outage a jumper has to be placed in parallel with the cutout prior to the test, then removed afterward. While certainly within the capabilities of any well-trained line crews, this increases the time, and hence the cost, of such tests.

Infrared scanning of both switches and cutouts can reveal a large portion of mechanical and electrical problems. Loose and corroded connections or a partially broken conductor near them shows up as a hot spot. The corrosion that causes frozen contacts often creates higher impedance in the switch contacts, creating a hot spot at the top of the blades.

Fuses

Fuses can be damaged by overheat, nearby lightning strikes, or long periods of high loads. However such problems are very rare, and fuses have very long lifetimes in service, assuming no fault occurs that requires them to blow.

One of the problems with fusing experienced to varying degrees by some utilities is a mismatch between designed and actual protection size of fuses on overhead lines. Mismatches occur because of mistakes made by line crews, expediency (a trouble-man has no 30 amp fuses so installs a 40 amp), and “coppered out” fuses.

The last problem arises due to either laziness on the part of line crews (rare) or to good intentions, overwork, and poor coordination of line repair records (more common than is appreciated). After long outages, particularly in very cold or very hot weather, “cold load pickup” can be a nuisance when restoring service. Starting current and loss of diversity due to the lengthy outage may mean that the “cold loads” are up to three times higher than normal. As a result, on primary feeders with tight fuse coordination, the initial load seen on a branch or lateral that is closed into service after a long outage may exceed the specified fuse rating. The fuse links keep blowing when the line crews try to close the cutouts to put the customers back into service.

The proper way to “fix” this problem is to jumper around the fuse for a period of thirty minutes to one hour, until load diversity returns. However, one expedient solution frequently used by line crews working against a tight deadline (common during storms) is to replace the fuse link with something that will not blow. This might simply be a larger fuse link than specified, but even a larger fuse size (i.e., a 40 amp instead of a 30 amp) may not provide enough capacity for the initial cold load surge. However, a length of #6 solid copper, or a large strand of aluminum from AA conductor, will fit in a fuse barrel, and “cures the problem.” Usually, the intent of the line crews is to return after one or two hours and change the fuse link back to the proper fuse. But due to difficult schedules and forgetfulness, a noticeable portion of such coppered-out fuses probably remain in service. Accumulated over many years, it can be a noticeable problem – engineering records and visual inspection show that fusing is installed on the system, but protection doesn’t work properly.

It is impossible without pulling the fuse element to verify the size of fuse actually installed in a fuse barrel. The “solution” to this inspection problem is a detailed assessment of trouble reports from the field. Statistical analysis of outages in which protection coordination is suspected of not functioning properly can reveal the degree of the problem – the “advanced TLM” discussed in Chapter 15 is somewhat successful at this. However, nothing short of site by site inspection, will identify all problems.

Regulators, Capacitor Banks, Reclosers, and Sectionalizers

These types of power system equipment are considerably more complicated than basic line equipment, and should be inspected, tested, and maintained in a manner similar to the transformers, tap changers, breakers and station capacitor banks it resembles. Routine inspection of this equipment, including visual and infrared inspection from the ground or helicopter, is typically included with inspection of the entire line.

Unlike the case with the basic overhead line equipment, electrical tests are typically applied to this equipment. Line regulators are tested in the same manner as small medium-voltage transformers/tap changers. Reclosers and sectionalizers fall into a category similar to breakers and relay sets. All of this equipment is routinely tested on something like a 2-5 year cycle for proper operation, and usually with instrumentation to assure proper turns ratios or operating speed and synchronization, as the case may be. Insulation resistance is also measured as a routine course. Reclosers and sectionalizers will have specific inspection, test, and maintenance procedures recommended by their manufacturers.

Shunt and series capacitor banks have unique characteristics (high capacitance) which require special test procedures, generally involving DC potential testing and induced potential tests. Capacitor switches in some systems are particularly failure prone and are tested annually, generally just prior to peak period.

Theory of Condition and Correlation with Test Results

The theory of equipment aging and deterioration is well understood by the power industry. Three major factors “age” electrical equipment: voltage-frequency stress, thermal stress, and water damage.

1. Voltage aging. Given enough time, equipment that is simply energized (e.g., a cable kept in service but carrying no load) will fail due to the accumulated stress of voltage – somewhere, somehow, its insulation will break down. The level of stress, and hence the rate of deterioration in insulation strength, is a function of both voltage level and frequency. Insulation lifetime is exponentially related to voltage level by something between a fifth and ninth order relationship depending on type of material and design. This means doubling voltage will cut lifetime by anywhere between a factor of 2⁵ (32) and 2²⁵ (33 ½ million) an incredibly wide range, indicating that insulation type and design are key factors in equipment quality.

The voltage-lifetime relationships for typical electrical insulation is in the range of fifth to tenth order. For example, XLPE insulation has a relationship that is ninth order. Doubling voltage means lifetime is cut by a factor of 2⁹ (512), meaning the unit loses the equivalent of about 3 weeks of lifetime for every hour it is energized to twice its nominal voltage. Its lifetime will be cut by 35% if it is operated at 1.05 PU rather than 1.0 voltage.⁸

2. Frequency also plays an important role in aging. Time to failure is related to frequency, roughly on a linear basis (doubling frequency will roughly halve lifetime, halving it will roughly double lifetime. Frequency is regulated so tightly – within .02% even in “sloppy” systems – that the equipment installed in any power system will see very nearly the same operating frequency all its lifetime. Basically one can assume that the equipment sees nominal system frequency during all hours of operation.

However, voltage spikes due to lightning strikes, switching surges or other unusual transients have a very high frequency component due to the rapid rise and fall times of their transient pulse. By their very nature these events are higher than normal voltage, and that higher voltage causes added stress in and of itself. But even though brief, these disturbances can cause considerable stress on insulation due to their high frequency content. For example: assuming that lifetime and voltage are inversely related by a ninth order relationship (XLP cable), one can compute that a spike to 1.5 PU voltage, lasting one half millisecond (.0005 sec), would cause roughly 50% more stress than normal due to its higher voltage, multiplied by a factor of 16.6⁹ (almost 150 billion) due to the higher frequency component of the spike. This means that this event lasting half of a millisecond ages the cable by the equivalent of about twoyears. In fact the impact may be somewhat different than this computation predicts, due to a host of secondary effects that can only be computed with an EMTP analysis.⁹ But the stress will still be great. Brief voltage events put undue stress on insulation.

Secondly, many diagnostic procedures such as hi-potential and partial discharge tests, and some cable fault location methods, use pulses (again, a high frequency content) or variable frequencies higher than normal. They can produce noticeably high stress on insulation.

⁸ This does not include the effects of any current increase due to the raised voltage. Generally, raising the voltage on a cable or unit of power system equipment will also raise the current passing through it (since it is a constant or very close to constant impedance). That increases the losses and hence the rate of thermal aging. The effects discussed here are due only to voltage – increasing the applied voltage on an unloaded section of XLPE cable by 5%, for example, is most likely to cut time until voltage breakdown occurs due to voltage stress alone, by a factor of about 35%. If that also leads to an increase in load, the additional thermal loading could contribute further acceleration of the cable aging.

⁹ This very simple computation neglects a host of electromagnetic transient (EMT) related factors that alter the induced stress, including changes in: 1) the impedance of the cable or insulators as a function of frequency, 2) the non-uniform distribution of stress that occurs as a function of high frequency, and 3) the fact that the exponential order of the relationship is itself a function of frequency.

3. High temperature has two effects on electrical insulation. First, some types of insulation (XLPE) soften at high temperatures to the point that plastic deformation occurs in the cable. The concentric conductor falls or “flows” through the insulation; mechanical stress on the cable may result in deformation. Minimum insulation distances can decrease as a result and failure can occur.

Secondly, and more important for the vast majority of equipment which does not suffer plastic deformation, higher temperatures accelerate all types of deterioration. The different modes of deterioration and the different deterioration rates of different materials are all affected in varying degrees, but qualitatively all forms of power system equipment lifetime loss are accelerated by higher temperature. A rough rule of thumb is that lifetime of cables is halved for every 9°C increase in temperature. Transformer insulation behaves in a roughly similar manner.

4. Combined stress of all three factors is what all-electrical equipment in service sees, the simultaneous effects of voltage, frequency and temperature, over time. Basically, the relative stress levels from these three factors are multiplicative, meaning that each exacerbates any problems due to the others:

Thus, if voltage and frequency at the moment are causing high levels of stress, and the insulation is at high temperature – say 9°C higher than normal (which can be expected to roughly halve lifetime), then the accelerated rates of lifetime loss due to the voltage and frequency are doubled.

Condition Assessment from Inspection and Test Data Go/No-Go versus Condition Measurement

Table 11.7 reprises data from Chapter 6’s Table 6.4 on the types of test and diagnostic methods available for power system equipment. Some of the tests are basically “go/no-go” tests of equipment, such as the high potential test and a transformer turns ratio test. A unit will either pass, or fail this test. Units that pass are put in service. Units that fail should be withdrawn from service.

Other tests, such as the total combustible gas test (TCGA) provide a numerical measure that is useful in assessing condition, and by tracking over time, provides clues as to the gradual deterioration of the unit and its continued capability to provide dependable service. These diagnostic tests lead to classification and fuel decision making on future inspection, testing, and/or replacement of units. For example, TCGA is used to determine classification of transformers by risk category and to recommend inspection and testing intervals (IEEE Standard C57.104-1991).

Some tests can be viewed as both go/no-go and condition assessment. Insulation resistance can be used both as a certification test and diagnostic. Below a certain value of insulation resistance, the unit is not put into service, and the value can be tracked over time so that changes in the unit can be detected. Like it, most of the test procedures discussed in Chapter 6 provide quantitative results – numerical values that can vary over some range. For example, insulation resistance tests measure resistance in megaohms, and can return a value of from 1100 to 1500 ohms when tested on the windings of high voltage power transformer. A power factor test for cross-linked cable can give a value of from just above 0.0 to over 2.0%. Transformer turns ratio tests for a 138/12.47 kV transformer should give 11.06, but can give values below or above that ratio depending on internal faults in the high or low side windings.

Generally, go/no go tests are dependable indicators of the viability of power equipment for service during the next year. If a unit “passes” the appropriate tests, it will usually provide good service in the next year. Leave a unit that has failed such a test in service and it will very likely fail, often catastrophically but nearly always to a point of at least destroying any residual future value in the next year.

Periodic and Frequency Testing is Required for Efficacy

The data shown here and the results they give in application are typical. Despite careful measurement, estimation of remaining lifetime – long-term condition assessment – is inexact. This test (and others) bears out several lessons:

1. Multiple tests help, but mostly because it gives more opportunities to “flunk” a unit on the basis of a single test. Note that unit looked fine from the standpoint of insulation resistance. But it scored so poorly on the TCGA test, that it received almost immediate attention.

2. Explanations based on other data help make decisions. Unit 4 was an old unit, something known to the test personnel from their property and maintenance records. That age data provided an explanation for why unit 4 had a lower resistance reading – age means deterioration means lower resistance. Therefore unit 4’s reading was basically expected. Had unit 4 been three years old, the interpretation of that test data would have been far different – there is a problem.

3. The best of tests are imperfect predictors. As mentioned above, roughly 4 out of 12 units (2, 3, 5 and 7) in this set ultimately had service lifetimes different than might have been expected based on their test results. These strike the authors as typical results – long-term condition analysis is dependable only about 50% of the time.

Thus, the overall conclusion one reaches about using test data for condition assessment is that:

1. Multiple tests have value because they test for different things or test the same thing (insulation strength) in different ways, and therefore increase the number of ways that flaws or weaknesses can be detected.

2. Good records are a critical part of inspection and testing and can be used in conjunction with test data to improve evaluation.

3. Periodic testing must be done, because long-term predictive capability of the tests is far from perfect.

4. Models of equipment behavior and function are essential, as they permit the test results to be interpreted both for consistency and importance.¹²

This mirrors what was said in Chapters 6 and 7 when comparing modern equipment diagnostic technology to medical diagnostic capabilities. In both fields, the tests are fairly good at identifying a reason (if there is one) for immediate concern due to an impending failure, but poor at predicting how condition will deteriorate, or not, over the long term.

Basically, both the results shown here and experience gained elsewhere all indicate the same thing: that testing and diagnostics, while giving some indication of long-term condition prognosis and remaining lifetime, are basically accurate only at identifying flaws and weaknesses that currently or soon will limit capability. Thus, the only viable way for a utility to maintain a good picture of its asset base’s condition, is to perform frequent “check ups” on that equipment.

People Often Say Condition Tracking When They Mean State Tracking

Using the distinctions made in the summary, tracking of condition can be useful but not nearly as useful as tracking of state (Figure 11.10). The trigger that brings a unit experiencing change to management’s attention will invariably be a change in condition (“Bayfront Substation’s 345/138 auto transformer has gone from B to C in the last week.”). But details related to its state and perhaps even the raw data itself may be needed immediately in order to understand the significance of the change and any urgency for action. (“Gassing has increased fourfold since the through-fault on the 19th of April . . . Traces of acetylene have been detected”). Thus, when most people or organizations talk about “condition tracking” they actually mean the periodic archiving of inspection, operating, and test data along with the state estimates and the overall condition determination. Condition is the factor that is tracked in terms of alarm criteria and used as the trigger for an alarm, notification, or recommendation for action – i.e., state might change but if the unit’s condition remains the same then any recommended actions or consequences of its condition remain the same.

Understand the Difference Between Condition Assessment and Failure Prediction

As mentioned in Chapter 6, exact knowledge of when a unit would fail would be worth a great deal to a utility. But even the best condition-assessment methods can not predict exactly the expected remaining lifetime, other than in extreme cases where the unit’s failure is imminent. In many such cases monitoring and testing can usually determine the unit is about to fail.

In all but those “near to failure” cases, the best that good condition assessment can do is to rate the relative likelihood that a unit will fail. One hundred transformers judged to be in “poor condition” will see far more failures in the next twelve months than one hundred that have been assessed to be in “very good” condition. Just how poor and just how good the units have been assessed will give some indication of failure rate. The information provided is a step in the right direction and worth a good deal toward improving service quality and allowing optimal management of equipment lifetimes. But again, this condition assessment is not a prediction of expected lifetime. Instead, it can and should be used to set priorities and general operating policy.

Good Record Keeping

Good records on the service and maintenance history of major equipment should be a part of any comprehensive condition assessment and improvement program. Records of the annual peak load and duration of peak served, and kWh served through the unit should be maintained. This equipment operating record library should also contain information on the number, type, and time of all through-faults and other types of abnormal events that occurred to the unit. Similarly, records of all inspections and maintenance should be maintained along with any comments on findings during inspections and repair (“Lower northwest corner of case dented”).

Record keeping has a good deal of value. Recall Chapter 6’s discussion of the value that exact knowledge of future time-of-failure would be worth. Accurate prediction of the exact failure of time is probably not feasible and is certainly not within the capabilities of existing technologies. Nevertheless, an old transformer is worth considerably more to a utility, in terms of being able to estimate its likelihood of failure, if records are maintained so its operating history is available to help determine its condition. Therefore, good record keeping has real value.

The most common failing among utilities in terms of record-keeping is a lack of historical data on through-faults, which many power systems engineers (including the authors) believe is a leading cause of transformer failure.

Inspection, Test, and Diagnosis/Tracking Program

The methods covered in this chapter are worthwhile only if applied in well-managed programs aimed at improving and maintaining service quality at high levels while keeping costs as low as possible. The recommended method of determining priority and of setting the intervals for inspection, testing, and diagnostic evaluation is to use the appropriate approach as described in that chapter.