Over the last 14 years, Georgia Transmission Corp. (GTC) has been working to reduce its momentary average interruption frequency index (MAIFI). In 1997, its overall MAIFI was about two momentary outages per consumer per year, with about 10 on the worst-performing line. Today, the company-wide MAIFI is less than 1.0 and has decreased from 10 to one on that troublesome line.
Since its incorporation, one of GTC's goals has been to improve the traditional IEEE metrics, including MAIFI. GTC engineers began by determining what was causing the momentary outages and verified lightning was the major problem.
Lightning Central
It is hardly surprising that lightning is the major cause of GTC's MAIFI since, after Florida, Georgia has some of the highest lightning stroke densities in the United States. Some parts of GTC's service area are near the Gulf Coast, while other parts are close to the Atlantic Ocean. Weather fronts come through year-round with multiple lightning occurrences happening every month. The most vicious storms are in the summer with literally thousands of strikes and heavy downpours.
GTC collects outage data and, to better understand the data, participates in a benchmarking program with about 30 other utilities. Together, the utilities account for about half the U.S. electric grid. Unfortunately, because of lightning, GTC ranked near the bottom of the group for momentary outage performance. GTC has made steady progress in matching the MAIFI of its peer utilities in the Southeast. Noteworthy is the fact that GTC's sustained outage frequency is on par with the national average. But its momentary outages are worse and thus have received much more attention.
Lightning can cause momentary outages in three distinct ways:
- First is the classic induced overvoltage from a nearby high current stroke. The voltage builds up and flashes from the insulators to ground. This outage mode is not a frequent occurrence on transmission lines due to the line insulation levels.
- Second is a shielding failure. This occurs when lightning strikes the phase conductor directly instead of the overhead grounding (shield) wire. The shielding angle of the overhead ground wire dictates the likelihood of this failure mode.
- Third is a back flashover, which happens when lightning strikes the shield wire or tower and, as the stroke current passes through the footing resistance, a sufficient rise of ground potential overstresses the phase insulators, causing a flashover. Improving pole grounding reduces the occurrence of this flashover, but it does not eliminate it.
To discover just how problematic lightning was, in 2001, GTC started subscribing to the Vaisala lightning data service. Vaisala sends information to GTC on all the lightning strikes in the state of Georgia. This includes the GPS location and magnitude of the strike, the polarity, the date and the time (down to the millisecond).
If a momentary outage occurs, the utility takes the time stamp from the digital fault recorder (also in milliseconds) at the fault's inception and compares it to the lightning data. If the fault inception is within 10 msec of the lightning flash and the coordinates of the stroke are within 1 km (0.62 mile) of the transmission line, then it is almost certain (95% sure) that lightning caused the momentary outage. In any given year, anywhere from 50% to 75% of momentary outages at GTC are due to lightning.
Technology Leads to Change
In 1997, it was widely believed a 115-kV line (based on its basic insulation level and other characteristics) should be able to withstand a lightning strike of 30,000 A or less and not flashover. If the lightning strike was more than 30,000 A, engineers expected the strike would invariably lead to an unpreventable backflash and momentary outage. Workers tried regrounding towers to prevent problems caused by lightning strikes of less than 30,000 A, but they were not making much headway. Shortly thereafter, GTC began using the Electric Power Research Institute's TFlash lightning performance software to model typical line configurations.
To get the best results, the model requires a lot of information, including type of terrain, tree cover, structure size and height, wire data, insulator information and so forth. The software then models the transmission line and lightning characteristics (stroke density and stroke current distribution) and predicts the number of momentary outages that will occur per year.
Once the line configuration is entered, engineers can develop a feel for the change in lightning performance caused by varying line parameters. Typical modifications include adjustments to shielding, grounding and insulation, and application of lightning arresters. This led to an interesting discovery.
When engineers modeled the lines with an arrester on each structure and on each phase, TFlash predicted 100% elimination of momentary outages due to lightning. When they modeled the lines with two lightning arresters on each structure — one on the top phase and one on the bottom phase — TFlash predicted one momentary outage every 16 years. So GTC chose the top and bottom configuration as its standard, because this configuration would reduce cost and one momentary outage every 16 years would be more than tolerable.
From Simulations to Reality
For the last 10 years, GTC has been installing lightning arresters on its transmission lines, focusing on load-serving lines first (46 kV, 69 kV and 115 kV). Typically, the results have been excellent.
For its worst-performing line, the one with 10 momentary outages per year, GTC installed lightning arresters on all three phases at every structure. Momentary outages due to lightning went from 10 outages per year to zero, as predicted by TFlash, the first two years. For the last two years, the line has experienced one momentary outage per year — both during summer storms — caused by lightning. In both cases, the stroke magnitude was greater than 100 kA. This is a stroke current magnitude expected for only about 1% of lightning. The Vaisala data indicates that GTC still has the same quantity and magnitude of lightning within a 1-km buffer around that line. So, to go from 10 momentary outages to about one per year is a remarkable improvement.
Now, with that transmission line taken care of, GTC has now reached a point where its worst lines do not have more than three or four momentary outages per year. The utility is still installing lightning arresters and otherwise improving lines where a majority of momentary outages are caused by lightning.
In 2009, GTC had its best year ever for momentary outages: 0.951 momentary outages per consumer. In 2010, the lightning flash count for Georgia was up 20% over 2009. However, MAIFI was slightly worse (approximately 7%) at 1.018 momentary outages per consumer.
Lightning Arresters Versus Moisture
With so much riding on the lightning arresters, GTC keeps a close watch on them to ensure they are not an enemy of the index. To assist with this, the failed arresters are sent to ArresterWorks for analysis. GTC believes that to make informed decisions about arresters, it must know why they fail when they do.
Analysis by ArresterWorks has indicated that moisture ingress is the leading cause of arrester failures on the GTC system. To reduce problems associated with moisture ingress, GTC uses Protecta*Lite arresters from Ohio Brass and has found them to be a reliable solution with a very low failure rate. So far, the few Ohio Brass failures have been due to installation problems, which also have been overcome by more robust connections.
Keeping Up With Customer Expectations
GTC's drive to improve MAIFI goes beyond just providing reliable service to its members. Reliability is necessary to ensure the utility's future.
In the 1970s, a momentary outage simply meant the lights blinked. Customers did not complain. Linemen would search for the cause of the problem, but, in general, everyone was happy because the protection scheme worked properly: The breaker did not lock out.
It is a very different world today. The digital age brought computers and, now, just about everyone has a dozen high-tech devices in their homes and offices. Momentary outages are a big problem.
Also, the state of Georgia has been experiencing an economic boom. High-tech companies are building new facilities. Invariably, these manufacturers are using computer-controlled equipment. While it may be easy to put an uninterruptible power supply on a single computer, it is much harder to protect an entire plant. For a manufacturing plant, a momentary outage can mean anywhere from two hours to four hours of downtime. This is simply unacceptable.
With momentary outages negatively affecting all of its members, GTC needed to get its momentary outages in line with its peer utilities. GTC has accomplished that goal. This is vital because, in Georgia, there is competition for new electrical loads. If a company moves into the area, it will typically look at three or four sites. Invariably someone will ask, “If we build at site X, in this industrial park, how reliable will our electric supply be?” To prosper, GTC must provide reliability for the digital age. Improving MAIFI is just one part of an ongoing effort.
Doug Maddox ([email protected]) is manager of system reliability for Georgia Transmission Corp. (GTC) in Tucker, Georgia. He holds a bachelor's degree in industrial engineering from the Georgia Institute of Technology and has a background in distribution, lighting and transmission work. At GTC, he has worked in customer service, maintenance, planning and reliability.
Jonathan Woodworth ([email protected]) started ArresterWorks, a surge protection consulting firm dedicated to assisting others in the mitigation of surges through the application of arresters, in 2008. For the previous 30 years, he was with Cooper Power Systems, where he was involved in the design, manufacture, testing and marketing of surge arresters. He and his partner Deborah Limburg also created and maintain ArresterWorks.com, which educates stakeholders about the surge protection of power systems.
Companies mentioned:
ArresterWorks www.arresterworks.com
Electric Power Research Institute www.epri.com
Georgia Transmission Corp. www.gatrans.com
Ohio Brass www.hubbell.com/Power/OhioBrass.aspx
Vaisala www.vaisala.com
Understanding Arrester Failures
Lightning is a constant and real cause of momentary and sustained outages for any utility located in the southeastern United States. Surge arresters are used to mitigate the effect of lightning strikes, and in the process, sometimes they are involved in the outage themselves. If the arrester fails during one of these events, more often than not they are believed to have failed because of lightning, but this is not always the case.
An arrester may fail during a surge from a lightning stroke, but it may be that the arrester was already nearing its end of life, and the lightning surge was the final act that brought about its end. In fact, years of failed-arrester evaluation reveal that arresters seldom fail because of the energy of the lightning stroke. Oftentimes during an examination of a failed arrester, it is found that the cause of the failure is moisture that has found its way into the arrester over the years. However, metal-oxide-varistor (MOV)-type surge arresters manufactured today are extremely robust. They are designed to withstand very severe surge events, offering many years of circuit protection.
Fortunately for those involved in forensic analysis of MOV-type surge arresters, severe lightning strokes leave a subtle change in the semiconducting material used in the arrester. This footprint does not disappear in time, and even if an arrester has been horrendously destroyed during a fault event, the minor change can be identified if analyzed correctly. When an arrester experiences an end-of-life event during a lightning storm, there are several reasons other than high current that might lead to failure: moisture ingress, short-term overvoltage stress, switching surge stress and even contamination-assisted flashover.
Consider a lightning stroke to the overhead ground wire just a span or two outside a substation, where station class arresters are protecting the transformer. If the ground impedance of the towers is too high, it is quite likely one of the phase insulators will back flash, causing a direct fault on that phase. During the fault, the voltage on the faulted phase will be lower than normal, and the power frequency voltage on the unfaulted phases can be substantially higher than normal. This elevated voltage can result in higher stress or failure of the arresters on these phases.
So even though the arrester failed during the lightning event, technically it was a power-frequency overvoltage that caused the failure. If the arrester had been previously compromised by moisture, it might not withstand the temporary overvoltage and fail even with a minor overvoltage.
Knowing the root cause of an arrester failure can be very important when trying to reduce the momentary average interruption frequency index (MAIFI) level. Passively calling an arrester failure a natural event caused by lightning may lead to more future outages, especially if the cause is actually an underlying arrester problem. By taking a proactive approach and clarifying the root cause of the failure, it may become apparent that additional arresters on the system should be examined or removed.
If a more systemic problem is identified as a result of the review, a plan can then be put in place to replace arresters that have a high failure risk, thus avoiding future outages. Knowing when to remove arresters and when not to is a tough but important decision. A properly conducted arrester forensic analysis can help in making this decision.
