Cause and origin of transformer failures
An Analysis of International Transformer Failures, Part 1
IntroductionMajor losses involving large oil-cooled transformers continue to occur on a frequent basis. A working group of the International Association of Engineering Insurers (IMIA) (www.imia.com) was established in 1995 to examine this topic and presented a report at the 1996 Conference. The magnitude of the losses has increased significantly since the last study. Increased equipment utilization, deferred capital expenditures and reduced maintenance expenses are all part of today's strategies for transformer owners. To make matters worse, world power consumption is increasing, and the load on each aging transformer continues to grow.
Scope of the StudyA request was sent to all IMIA national delegations seeking information on losses of transformers rated at 25 MegaVoltAmperes (MVA) and above for the period 1997 through 2001. Information was requested concerning year of loss, size in MVA, age at failure, application (such as utilities, industrials), cause of failure, property damage portion, and business interruption portion. Data was obtained on 94 cases. An estimate of the total population of power transformers would have been useful, but it is impractical to obtain this information. Some of the contributors were not able to identify the age of the transformers, and in some cases, the size of the transformer. The analysis is annotated wherever data is missing. All amounts of losses were converted to U.S. dollars, using the following exchange rates: 0.9278 euros; 8.542 Swedish kronas; and 6.0858 French francs.
Five-Year TrendDuring this period, the number of transformer claims reached a peak (25) in 1998. But, the dollars paid out, reached a maximum in 2000 due to several claims in the multi-million dollar range, plus one large business interruption loss. The largest transformer loss also occurred in 2000, at a power plant, with a business interruption portion the equivalent of more than U.S. $86 million. Three of the top four property damage claims were in industrial plants.
Table 1 displays the annual transformer claims. Not all of the data contributed had size information. Therefore, we could only analyze 78 claims for cost per size. The average cost (for property damage only) was approximately $9,000 per MVA (or $9 per kVA). Table 1A displays the annual transformer claims and cost per MVA.
Type of ApplicationDuring this period, the largest number of transformer claims (38) occurred in the Utility Substation sector, but the highest paid category was Generator Step Up transformers, with a total of more than $200millon. If the extraordinary business interruption loss is ignored, the Generator Step Up transformer category is still significantly higher than any other category. This is to be expected due to the very large size of these transformers. Table 2 displays the annual claims by application.
Cause of FailureFor the transformer failures reported, the leading cause was "insulation failure." This category includes inadequate or defective installation, insulation deterioration, and short circuits, but not exterior surges such as lightning and line faults. Table 3 lists the number of failures and percentage of total costs paid for each cause of failure. A description of each cause category is found below.
The risk of a transformer failure is actually two-dimensional: the frequency of failure and the severity of failure (see "Transformer Asset Management,” The Locomotive, Spring 2003, Vol. 77, No. 2).
Cause of Transformer FailureInsulation Failures – Insulation failures were the leading cause of failure in this study. This category excludes those failures where there was evidence of a lightning or a line surge. There are actually four factors that are responsible for insulation deterioration: pyrolosis (heat), oxidation, acidity and moisture. But moisture is reported separately. The average age of the transformers that failed due to insulation was 18 years.
Design /Manufacturing Errors – This category includes conditions such as: loose or unsupported leads; loose blocking; poor brazing; inadequate core insulation; inferior short circuit strength; and foreign objects left in the tank. In this study, this is the second leading cause of transformer failures.
Oil Contamination – This category pertains to those cases where oil contamination can be established as the cause of the failure. This includes sludging and carbon tracking.
Overloading – This category pertains to those cases where actual overloading could be established as the cause of the failure. It includes only those transformers that experienced a sustained load that exceeded the nameplate capacity.
Fire /Explosion – This category pertains to those cases where a fire or explosion outside the transformer can be established as the cause of the failure. This does not include internal failures that resulted in a fire or explosion.
Line Surge – This category includes switching surges, voltage spikes, line faults/flashovers, and other T&D abnormalities. This significant portion of transformer failures suggests that more attention should be given to surge protection, or the adequacy of coil clamping and short circuit strength.
Maintenance /Operation – Inadequate or improper maintenance and operation was a major cause of transformer failures, when you include overloading, loose connections and moisture. This category includes disconnected or improperly set controls, loss of coolant, accumulation of dirt and oil, and corrosion. Inadequate maintenance is to blame for transformer owners not discovering incipient troubles when there was ample time to correct them.
Flood – The flood category includes failures caused by inundation of the transformer due to man-made or naturally caused floods. It also includes mudslides.
Loose Connections – This category includes workmanship and maintenance in making electrical connections. One problem is the improper mating of dissimilar metals, although this has decreased somewhat in recent years. Another problem is improper torquing of bolted connections. Loose connections could be included in the maintenance category, but we customarily report it separately.
Lightning – Lightning surges are considerably fewer than in previous studies we have conducted. Unless there is confirmation of a lightning strike, a surge type failure is categorized as a Line Surge.
Moisture – The moisture category includes failures caused by leaky pipes, leaking roofs, water entering the tanks through leaking bushings or fittings, and confirmed presence of moisture in the insulating oil. Moisture could be included in the inadequate maintenance or the insulation failure category above, but we customarily report it separately.
An Analysis of International Transformer Failures, Part 2
IntroductionIn our previous discussion of a five-year international study of loss trends for transformer failures, we concluded that Insulation Failure was the leading cause. Combined with Design/Material/Workmanship and Unknown causes, these three categories accounted for 65 percent of the total number of transformer breakdowns in our study of losses as reported by members of the International Association of Engineering Insurers (IMIA) (imia.com) and 85 percent of the total amount paid out for these claims.
Other causes of loss were spread among: Oil Contamination; Overloading; Fire/Explosion; Line/Surge; Improper Maintenance/Operation; Flood; Loose Connection; Lightning and Moisture (see An Analysis of International Transformer Failures, Part 1, The Locomotive, Winter 2004, Vol. 78, No. 1).
Transformer AgingWe did not categorize "age" as a cause of failure. Aging of the insulation system reduces both the mechanical and dielectric-withstand strength of the transformer. As the transformer ages, it is subjected to faults that result in high radial and compressive forces. As the load increases, with system growth, the operating stresses increase. In an aging transformer failure, typically the conductor insulation is weakened to the point where it can no longer sustain mechanical stresses of a fault. Turn-to-turn insulation then suffers a dielectric failure, or a fault causes a loosening of winding clamping pressure, which reduces the transformer's ability to withstand future short circuit forces.
Table 1 displays the distribution of transformer failures by age. The average age at failure among those transformers in our study was 18 years.
Age at Failure
Number of Failures
Cost of Failure
|0 to 5 years|
6 to 10…
|11 to 15…|
|16 to 20…|
|21 to 25…|
|Over 25 Years|
|Age Unknown *|
The most influential parametric mortality model in published actuarial literature is that proposed in 1825 by Benjamin Gompertz, who recognized that an exponential pattern in age captured the behavior of human mortality. He proposed the failure function:
In this article, the author will discuss an updated model for transformer failure predictions in the coming decade and shares the observations of industry executives about the condition of the transformer marketplace.
A New Failure ModelHSB’s first publication on transformer failure predictions used the Gompertz model. In 1860, W.M. Makeham modified the Gompertz equation because it failed to capture the behavior of mortality due to accidental death, by adding a constant term in order to correct for this deficiency. The constant can be thought of as representing the risk of failure by causes that are independent of age (or random events such as lightning, or vandalism).
The instantaneous failure rate for transformers in a given year is the probability of failure per unit time for the population of transformers that has survived up until time “t.” To include the frequency of random events (lightning, collisions, vandalism) separate from the aging component, the constant “A” is set at 0.005 (which represents one-half percent of 1 percent). Figure 2 is the corresponding exponential curve for a 50 percent failure rate at the age of 50.
Number of failures (in GVA) at year "t," = [Failure rate] x [population that is still surviving]
Future Transformer FailuresUsing the population profile from Figure 1, the predicted failures can be plotted for all U.S. utility transformers built between 1964 and 1992. The prediction is simply intended to illustrate the magnitude of the problem facing the utility industry and the insurance industry. Figure 3 is the failure distribution. The X-axis is the year of predicted failures. The Y-axis is the population of the failures (expressed in GVA). It should be noted that the graph is a failure rate of those that survived, until time ("t"). In this graph, a vertical line depicts each vintage. By 1975, each year has a cluster of six different vintages (1964, ‘66, ‘68, ’70, ’72 and ’74); and after 1992, each cluster is 15 vintages.
Although we have not yet seen an alarming increase in end of life failures, such a rise must be expected eventually. The most difficult task for the utility engineer is to predict the future reliability of the transformer fleet, and to replace each one the day before it fails. Meeting the growing demand of the grid and at the same time maintaining system reliability with this aging fleet will require significant changes in the way the utility operates and cares for its transformers.
Action PlanOne conservative strategy suggests that the industry start a massive capital replacement program that duplicates the construction profile of the 1960s and 1970s. But this would needlessly replace many transformers and cost the utility industry billions of U.S. dollars.
The ideal strategy is a life assessment, or life cycle management program, that sets loading priorities and provides direction to identify: a.) transformer defects that can be corrected; b.) transformers that can be modified or refurbished; c.) transformers that should be relocated; d.) transformers that should be retired. The insurance industry should be aware that both the Institute of Electrical and Electronics Engineers, Inc. (IEEE), and the International Council on Large Electrical Systems (CIGRE) are developing guidelines for aging transformers. [5, 6]
Electric Utilities and the Transformer IndustryThe deregulation of wholesale electricity supply around the world has led to a number of changes and new challenges for the electric utility industry and its suppliers. In the last few years, many electric utilities have merged to form larger international utilities, and others have sold off their generating assets. All of this is being done in an attempt to enhance revenue streams, reduce the incremental cost per MW or react to spot market opportunities.
Years ago, utilities knew the needs of their native markets and built an infrastructure to keep pace with those needs, with associated construction costs being passed back to the ratepayers. Starting in the 1980s, utilities in the United States had to contend with regulatory mandates to utilize independent power producers to satisfy supply and meet demand. They were not able to plan projects for their native load projections. In this environment, it was possible that the utility's capital projects may not be afforded a favorable rate structure from the local utility commission in an openly competitive market. Therefore, many utilities understandably halted most of their capital spending, due to this regulatory uncertainty.
This significantly limited the activity taking place in terms of expanding the industry's infrastructure, including their transmission and distribution assets. In the 1990s, capital spending on new and replacement transformers was at its lowest level in decades. Many of the major manufacturers exited the power transformer business. Many of the remaining manufacturers have undertaken cost-cutting measures to survive.
"The Boom is Over..."Then in 1999-2000, the transformer market experienced a brief upswing in activity primarily due to a rush to build gas-turbine generating plants. The demand for generator step-up transformers in the United States almost doubled during these peak months. At that time, there were predictions that 750 Gigawatts of new generating capacity would be installed worldwide between 2000 and 2010.
But, the rush to build power plants in the United States has subsided; many of the energy companies are now drowning in debt. Many developers and investors had to sell their interests in existing plants in order to finance the completion of new plants. In 2001, projects worth 91 GW of generating capacity in the United States were cancelled (out of 500 GW). And in the first quarter of 2002, orders for 57 GW of capacity were cancelled.
Again, capital spending in the utility industry sharply declined. According to Dennis Boman, director of marketing, ABB Inc. Power Transformers, "the decline has far exceeded anyone's prediction to levels that post-dated the increase. Within a short six-month period, the power transformer market dropped by over 50 percent,” he said. Added Joe Durante, vice president, VA Tech Elin Transformers, "...the boom of the late 1990s and early 2000 is over, and most likely won't be seen for another 30 years. Replacement opportunities will continue to remain flat and customer spending will continue only when necessary."
Based on Hartford Steam Boiler claim experience, new transformer prices are significantly lower than they were a few years ago. It is truly a buyer's market. New power transformers are being sold at a price less than the cost of a rewind, and the manufacturers are now providing three-year and five-year warranties.
Peter Fuchs, vice president sales and marketing, Geschaftsgebiet Transformers (Siemens), predicts “a stagnant market, on average, for the United States , Europe and the Far East .” However, “in other parts of the world, economic growth and business development are proceeding at high levels, including a resurgence in Asia ,” he continued. “The need for power in this area already exists, and as international funding becomes available, we expect to see increased activity in this region.”
Transmission Growth Opportunities?Today, many of the transformer manufacturing plants and repair facilities have very little activity. Is this "slump" in the market due solely to government regulation —or deregulation? The three major manufacturers point to a number of different problems. According to Bowman of ABB, “We have seen a shift in focus to ‘first cost’ buying with little regard for any long term impact on buy decisions." Many buyers are choosing the lowest bidder, with little regard to quality, reliability or factory service. Fuchs of Siemens observes that “in addition to the price-driven decision, there is very little technical evaluation, and ‘price-dumping’ continues to go unpunished." Durante of VA Tech Elin confirms that the major obstacle is "ongoing deregulation uncertainty which is hindering capital investment.”
Durante believes that the next growth opportunity in the North American utility market is the transmission segment. This includes inter-tie transformers, phase-shifter transformers and autotransformers. "However, this market is heavily influenced by government regulations and decisions,” he added.
The U.S. Federal Energy Regulatory Commission (FERC) has mandated that all generators have equal access to transmission systems and required integrated utilities to turn over their transmission systems to independent entities. Some utilities have decided to sell their transmission assets and purchase transmission service. Other utilities are joining together and rolling their transmission assets into limited liability companies. But many utilities first want to understand exactly how transmission will be regulated. In other words, utility investors want to know whether the U.S. federal government or state governments will regulate the transmission assets. Until this is clear, overall capital spending will be deferred.
SummaryElectricity is much more than just another commodity. It is the life-blood of the economy and our quality of life. Failure to meet the expectations of society for universally available low-cost power is simply not an option. As the world moves into the digital age, our dependency on power quality will grow accordingly. The infrastructure of our power delivery system and the strategies and policies of our insureds must keep pace with escalating demand. Unfortunately, with regulators driving toward retail competition, the utility business priority is competitiveness and related cost-cutting — and not reliability.
- William H. Bartley, HSB, Analysis of Transformer Failures, Proceedings of the Sixty-Seventh Annual International Doble Client Conference, Boston MA , 2000.
- William H. Bartley, HSB, Failure History of Transformers-Theoretical Projections for Random Failures, Proceedings of the TJH2B TechCon, Mesa AZ , 2001.
- William H. Bartley, HSB, Transformers Failures, presented as Keynote Address at the annual ABB Technical Conference, Alamo , TN , 2003.
- Tim Higgins, Mathematical Models Of Mortality, presented at the Workshop on Mortality Modeling and Forecasting, Australian National University , February 2003.
- IEEE C57.140, Draft 9 March, 2003 , IEEE Guide for the Evaluation and Reconditioning of Liquid Immersed Power Transformers, Rowland James & William Bartley Co- Chair.
- CIGRE 12-20 Guide on Economics of Transformer Management (draft 23.7.02).
An Analysis of Transformer Failures, Part 1 – 1988 through 1997
Transformers Losses Are SignificantAs an object class, transformers have consistently been ranked in the top five objects for claims paid by Hartford Steam Boiler during the past several decades. Over the years, HSB has investigated thousands of transformer losses. Some were covered claims and some were not. Using the data collected, HSB has conducted a number of studies on transformer losses and published them in The Locomotive.
In previous analyses, we examined Initial Parts to Fail, Age at Failure, and Cause of Failure. To improve our loss experience for transformers, the most recent study was conducted last year, and covers a 10-year period from 1988 through 1997. In 1999, we deleted the Initial Parts category and added two new categories: the Size and Frequency of Failures, and Where the Failures Were Occurring (by industry or occupancy).
10-Year Trend - Losses by Transformer TypeOver the past 10 years, HSB has paid hundreds of transformer claims that represent many millions of dollars. Chart No. 1, below, shows a breakdown of claims, according to the transformer type. The chart shows power transformers, askarel-filled (PCB), dry type, arc furnace, induction furnace, and rectifier transformers. Except for 1988, the power transformer dominates our loss history.
Occupancies Where Failures OccurredHSB insures literally hundreds of different types of occupancies: shopping malls, bakeries, apparel manufacturers, electric utilities and steel mills. In order to analyze the transformer risk, we divided this long list of occupancies into 10 categories we call exposure groups. An exposure group is a set of occupancies with similar equipment, operations and loss profiles. Table 1, below, lists the 10 exposure groups, and identifies some specific examples within those groups.
Frequency and SeverityIn his book, Risk-Based Management: A Reliability-Centered Approach1, HSB’s Dr. Rick Jones defines "risk" as the product of probability and consequence, or in insurance engineering terms, the frequency and severity of the losses. The severity can be defined as the average annual gross loss, and the frequency (or probability) can be defined as the average number of losses, divided by the population. Since we don’t have a true transformer population, we needed to make a substitute. Because we are ranking the relative risk of exposure groups, we used the number of locations insured in each group for our "population." Thus, for any given exposure group:
Frequency equals the Number of Losses divided by the Number of Locations.
(For example, if we have had an average of 10 losses per year, in a given exposure group and we insure 1,000 locations in that group, the probability of a failure is .01 each year, at any location in that group.) Therefore, we can rank our transformer risk by occupancy, using the product of frequency and severity. (Risk = Frequency x Severity).
The graph below, Chart No. 2, is a Frequency-Severity "scatter plot" for transformer risks in our 10 exposure groups, based on the last 10-year period. With each group plotted, frequency on the X axis and severity (or average gross loss) on the Y axis, the X-Y plot becomes a risk coordinate system. The diagonal lines are called equivalent lines of risk (for example, a probability of 0.1 for $1,000 and a probability of 0.01 for $10,000 can be considered an equal risk.) Coordinates in the upper right quadrant are the highest risk.
Transformer AgeTransformer design engineers tell us that a transformer can be expected to last 30 to 40 years under "ideal conditions." But, that is clearly not the case. In the 1975 study, it was found that the average age at the time of failure was 9.4 years. In our 1985 study, the average age was 11.4 years. In this study, the average age at failure was 14.9 years. One would expect to see a "bathtub curve" with infant mortality in the early years, and aging equipment at the far right. Instead, our claim statistics show that transformers do not have an indeterminate life. Chart No. 3, below, shows the age statistics for this study. These statistics should justify the time and expense to periodically check the condition of the transformer.
There are actually two problems here: our nation’s transformer fleet is aging, and to compound this, the load on each transformer, (or its utilization), continues to grow. While installation of new transformers is declining, power consumption continues to grow at a rate of about 2 percent per year. Capital deferment has led to the increased overall utilization of transformers in the United States. Due to the steady growth in power consumption over the last 20 years, it is obvious that the utilization factor for transformers has increased significantly.
Summary[In Part 2 of this article: The author discusses the causes of transformer failure, including lightning surges, line surges, poor workmanship, deterioration of insulation, overloading and other factors. He recommends a good maintenance program and concludes with several recommendations to help achieve maximum service life.]
[Footnotes: 1 R.B. Jones, Risk-Based Management: A Reliability-Centered Approach, Gulf Publishing Co., Houston, TX, 1995]
An Analysis of Transformer Failures, Part 2: Causes, Prevention, and Maximum Service Life
IntroductionOver the years, Hartford Steam Boiler has investigated thousands of transformer failures, compiling an extensive database of loss information. In part one of this article, the author used data from HSB’s latest 10-year study of transformer claims to examine the types of breakdowns, frequency, severity, and the issue of transformer age. In part two, he discusses the causes of transformer failures, recommends a maintenance program and concludes with ways to help achieve maximum service life.
Cause of FailureHartford Steam Boiler has collected information about transformer failures for decades. Analysis has shown that while aging trends and utilization may change (see part 1), the basic causal factors of these failures remain the same. In the article "Factors Affecting the Life of Insulation of Electrical Apparatus," published in the July 1949 issue of The Locomotive, HSB’s J.B. Swering, chief engineer of the Electrical Division, wrote:
"There are a number of factors which affect the life expectancy of insulation and these should receive the careful consideration of persons responsible for the operation of electrical equipment. These factors include:
- High Operating Temperature
- Lightning or Line Surges
- Care of Control Equipment
- Lack of Cleanliness
- Care of Idle or Spare Equipment
- Improper Lubrication
- Careless or Negligent Operation
LightningLightning surges are considerably less than previous studies, because of our changes in categorizing the cause. Today, unless we have confirmation of a lightning strike, a surge type failure is categorized as "Line Surge." This is one of the departures from the previous studies.
Line SurgesAccording to our database, the Line Surge (or Line Disturbance) is the number one cause for all types of transformers failures. This category includes switching surges, voltage spikes, line faults/flashovers, and other transmission and distribution (T&D) abnormalities. This significant portion of transformer losses indicates that more attention should be given to providing surge protection, or testing the adequacy of existing surge protection.
Poor Workmanship/ManufacturerIn the 1998 HSB study, only a few percent of the total claims were attributed to Poor Workmanship or Manufacturer’s Defects. Among the conditions found were such things as loose or unsupported leads, loose blocking, poor brazing, inadequate core insulation, inferior short circuit strength, and foreign objects left in the tank.
Deterioration of InsulationInsulation Deterioration was the second leading cause of failure over the past 10 years. The average age of the transformers that failed due to insulation deterioration was 17.8 years — a far cry from the expected life of 35 to 40 years! In 1983, the average age at failure was 20 years.
OverloadingThis category pertains to those cases where actual Overloading could be established as the cause of the failure. It includes only those transformers that experienced a sustained load that exceeded the nameplate capacity.
Often, the overloading occurs when the plant or the utility slowly increases the load in small increments over time. The capacity of the transformer is eventually exceeded, resulting in excessive temperatures that prematurely ages the insulation. As the transformer’s paper insulation ages, the strength of the paper is reduced. Then, forces from an outside fault may cause a deterioration of the insulation, leading to failure.
MoistureThe Moisture category includes failures caused by floods, leaky pipes, leaking roofs, water entering the tanks through leaking bushings or fittings, and confirmed presence of moisture in the insulating oil.
Inadequate MaintenanceInadequate Maintenance was the fourth leading cause of transformer failures. This category includes disconnected or improperly set controls, loss of coolant, accumulation of dirt and oil, and corrosion. Inadequate maintenance has to bear the blame for not discovering incipient troubles when there was ample time to correct it.
Sabotage and Malicious MischiefThis category is usually assigned when willful damage was evident. Surprisingly, there were no reports of transformer damage in the last 10 years due to this cause.
Loose ConnectionsLoose Connections could be included in the Maintenance category, but there was a sufficient number of reports to list it separately. This is another departure from previous studies. This category includes workmanship and maintenance in making electrical connections. One problem is the improper mating of dissimilar metals, although this has decreased somewhat in recent years. Another problem is improper torquing of bolted connections.
All OthersThis category encompasses all that could not be attributed to the above categories, including "Cause Undetermined."
SummaryA review indicates that a planned program of maintenance, inspection and testing would significantly reduce the number of transformer failures, and the unexpected interruption of power. From a cost standpoint, not only has the cost of repair increased dramatically, so has the cost of downtime. Rewinding or rebuilding a large power transformer can take six to 12 months. A good maintenance program should include the following recommendations to help achieve maximum service life.
Installation and Operation
- Keep the electrical loading within the design range of the transformer. In liquid-cooled transformers, carefully monitor the top oil temperature.
- Install transformers in locations that are compatible with their design and construction. If placed outdoors, make sure the unit is rated for outdoor operation.
- Protect transformers from surges and other external hazards.
Test the Oil"The dielectric strength of transformer oil decreases rapidly with the absorption of moisture. One part water in 10,000 parts oil has been known to decrease the dielectric strength 50 percent. Oil samples from each tank, except of course small distribution transformers, should be given a break-down test at least once each year … so that moisture may be promptly detected and removed by filtering." (From The Locomotive, April 1925).
- Gas-in-oil analysis should be performed annually to measure the dissolved gases in the oil that are created by developing faults in the transformer. The specific gas and the amount of gas can identify the type of fault. The fluid screen test should be performed annually to determine the oil’s ability to perform as an insulant. These tests include dielectric breakdown, acidity, interfacial tension, etc.
- Keep the porcelain bushings and insulators clean.
- On liquid-cooled units, check the radiators for leaks, rust, accumulation of dirt, and any mechanical damage that would restrict the oil flow.
- Keep electrical connections tight.
- Inspect tap changes on a regular basis. Check the contacts for tightness, burning, pitting, freedom of movement, and alignment.
- The transformer windings, bushings, and arresters should have a Power Factor test on a three-year basis.
- Check the ground connection on the surge arrester annually. The connection should be tight, and the lead should be as short as possible. The earth resistance should be checked during the dry season, and should not exceed 5 ohms.
- Consider on-line transformer monitor system for the most critical transformers. There are a number of on-line systems currently on the market. The system vendors assemble a variety of probes and sensors, connect them to a data acquisition unit [DAU] and provide for remote telecommunications through a modem. The systems also incorporate an "expert system" to diagnose the problem and distinguish between events that are harmless and events that are dangerous.