T&D World Magazine
University of Arizona lightning data

Deriving Value from Lightning Data

Utilities have been exploiting cloud-to-ground lightning information for more than 30 years.

Real-time information about individual cloud-to-ground lightning flashes has been used by electric utilities in the U.S. since the mid-1980s. An impressive array of scientific and technical innovations has made it possible for operations center staff to receive information about a flash before the thunder reaches them.

This valuable information can be used to identify the closest line structures to a lightning stroke location, estimate the peak current of a stroke and provide the timing of return strokes with an accuracy of better than 1 μsec. Modern lightning locating systems (LLS) track the precise location of thunderstorms and provide information about storm arrival times at critical locations.

University of Arizona, lightning data
NLDN operational overview: 1) The sensor detects an event. 2) Data are transferred to the satellite. 3) The satellite forwards data to the hub. 4) A dedicated link sends data from the hub to a central analyzer. 5/6) Lightning data are forwarded to real-time users via a satellite link.

The NLDN

The U.S. National Lightning Detection Network (NLDN) has its technical roots in gated wideband magnetic direction-finding technology, first commercialized by Lightning Location and Protection Inc. in the mid-1970s for locating lightning-caused fires. The value of real-time information about lightning was immediately appreciated for other applications.

The basic elements of ground-based LLS have not changed since the early years. Several remote magnetic or electric field sensors are placed tens to hundreds of miles apart, covering the perimeter and interior of the region of interest. These sensors detect and measure lightning-generated electromagnetic fields, which also are the source of the crackling sound heard on AM radios during storms. The sensors then send this information to a central location that uses it to geo-locate the lightning and compute additional information about each discharge. Finally, the results are formatted and distributed to users with varying levels of detail.

One of the earliest LLS was the East Coast network operated out of the State University of New York at Albany, which installed its first sensors in 1982. The early success of this network interested the Electric Power Research Institute (EPRI), which then supported further development of a U.S. nationwide network. EPRI support began in 1983, and the network entered an expansion period that did not stop until the entire contiguous U.S. was covered in 1989.

During this six-year period, EPRI funded the development of the NLDN based on the electric utility needs for real-time lightning information for repair crew management and the long-term objective of producing an 11-year (solar cycle) lightning ground-flash-density data set for the continental U.S. The NLDN has been retaining information about all recorded cloud-to-ground lightning flashes throughout the U.S. for more than 30 years.

University of Arizona, lightning data
A modern NLDN sensor provides data to the central analyzer about the direction of the lightning from the sensor, field strength and timing.

Network Upgrades

The NLDN has undergone several improvements throughout its 30-year life, based on a sequence of scientific discovery and emerging technologies. In 1989, before the advent of practical sub-microsecond timing made possible by GPS, the NLDN’s sensors used
millisecond timing to correlate lightning data among the reporting sensors. Geo-location of events calculated the intersection of direction information provided by each sensor pointing to the lightning discharge.

Following the commercialization of the NLDN in the early 1990s, GPS timing became effective. By 1995, the 100+ sensors in the NDLN were upgraded to provide both directional data and arrival-time information with GPS accuracy. This change improved ground stroke location accuracy from a median error of 2 km to 4 km (1.24 miles to 2.5 miles) down to 0.5 km (0.3 miles). This same upgrade resulted in the ability to report 80% to 90% of all cloud-to-ground flashes and respond better to unusual multiple-stroke or large-amplitude lightning.

Another important upgrade occurred in 1998 along the NLDN’s border with Canada, as a result of the installation of Environment Canada’s Canadian Lightning Detection Network (CLDN). Performance of the NLDN was significantly improved along the border between the U.S. and Canada. The combined operation of the NLDN and CLDN is referred to as the North American Lightning Detection Network (NALDN), which provides continuous and consistent lightning information throughout a region of nearly 20 million sq km (7.72 million sq miles) with latitudes ranging from 25˚ to more than 60˚ north latitude.

A number of incremental improvements to the NLDN occurred between 2002 and 2012, which resulted in a better than 95% cloud-to-ground flash detection efficiency and typical location errors in the range of 200 m to 300 m (656 ft to 984 ft). Since 2012, typical location errors have been reduced to about 150 m (492 ft), making the data more useful for distribution-related applications than in the early years. The most-recent upgrade in 2013 resulted in the ability to also report about half of the lightning flashes that remain in the clouds, so-called “cloud flashes.”

University of Arizona, lightning data
An early objective of NLDN was to produce long-term ground flash density maps. Early maps showed only the data collected in the eastern U.S. Data are now available for the continental U.S. for more than two solar cycles.

General Data Uses

Electric utilities employ lightning data in a variety of ways. Data may be used within minutes of occurrence to help in decisions about reclosing, or over a period of years when used for long-term lightning incidence climatologies. The time and location of the lightning may be used to see where, in general, a storm is occurring and where it is going, or it may be used at the scale of milliseconds and a few tens of meters to help determine if a specific stroke in a flash caused a fault of interest. In the latter case, the peak current and polarity of the discharge also become important. The customer scope of use can be as broad as systemwide reliability indices or as focused as helping critical customers to understand a specific outage that disturbed their operations.

Storm Tracking Use

Early applications of NLDN and Canadian data in the control rooms of major electric utilities led to some initial disappointment. The additional warning time of approaching frontal storms was of little consequence. The system operators knew when lightning had moved into their service territory, on the basis of their crude detection network: the distribution and transmission system breakers would start to operate. Remedial actions such as rebalancing generation to prevent islanding would be taken after a few automatic operations.

However, the operators soon embraced the LLS technology, as it brought them a different and unexpected advantage: tracking the progress of storms allowed operations to reduce the duration of storm-mitigation measures, which normally reduced the transfer of power across critical interfaces to respect stability margins. At Ontario Hydro, storm limits on 230-kV and 500-kV lines were reduced from 24 hours to about 2 hours, placing the value of LLS use at CDN$16,000/hour, $350,000/thunder day and $10,000,000/year — and this evaluation was from 1986.

These numbers help to explain why a real-time LLS display remains an important feature in the Hydro One grid control center, commissioned in 2005. Better management of storm limits starting in 1986, combined with other system configuration factors, also postponed the need for construction of a relief line, costing $700 million, from the late 1980s to June 2012.

Use for Protection Design

Comparison of NLDN data with transmission line tripout times has resulted in serious consideration that low-amplitude flashes could be causing shielding failures or back flashovers on poorly grounded towers. This issue appeared in an early comparison of NLDN data with utility experience at the Tennessee Valley Authority (TVA) in 1990. Of the 18 diagnosed faults to its backbone 500-kV lines, TVA found 13 had estimated peak currents of less than 40 kA and only one had a peak current estimate exceeding its critical current of 100 kA.

Field tests showed structure footing resistance as high as 150 Ω. This may have played a role in TVA’s decision to measure the footing resistance of every 500-kV tower — at the time (1990-1994), a task involving more than 10,600 structures — and to improve the NLDN coverage in its service region.

The importance of NLDN data remains even when the lightning density is low. For example, outages on a pair of 230-kV unshielded transmission lines at Nalcor Energy in Newfoundland were evaluated against lightning flash times, locations and magnitudes.

This pair of lines used single-pole reclosing rather than overhead shield wires as its primary lightning protection. The success rate of the reclosing system was low, however, and lightning data from the new NALDN confirmed improvements were needed. These took the form of the first line surge arresters used at this voltage level, with large-diameter metal-oxide varistor blocks designed for the severe duty.

Line Reliability

An IEEE task force explored ways to use weather data to normalize annual variations in reliability indices. Electric utilities are continuously rated and ranked for SAIDI, SAIFI, CAIDI and MAIFI values on their distribution systems, often using the IEEE Standard 1366. The group provided strong evidence lightning was a dominant reliability issue. Lightning density by ZIP code area, averaged over four years, and monthly lightning flash counts from lightning detection networks were both used in this work, which was completed in 2011.

In the IEEE case study for Manhattan, Kansas, U.S., 79% of interruptions occurred during lightning, with only 8% in fair weather. With 100 to 125 hours of thunder per year in Kansas, this means only 1.3% of the total hours were providing the vast majority of problems. Severe weather increased the probability of faults but not necessarily the number of sustained interruptions.

Not surprisingly, areas with the highest lightning activity showed the highest level of agreement between the monthly flash density and monthly interruption rate. Expressed as a correlation coefficient, with 100% being perfect, the grades were 90% in Florida, 81% in the Carolinas and 54% in Michigan. This experience confirms any claimed improvements in line performance — whether from new equipment installation, system maintenance or reinstallation of stolen copper grounding — should be justified using lightning data before the work and checked against the data afterward.

Another high-value use of lightning data is the quantification of lightning incidence near individual transmission lines, making it possible to assess the tripout rate, normalized by the number of lightning challenges to or near the line. This provides precise and poignant information for explaining problem seasons to customers, setting appropriate budgets for line improvement efforts and accurately assessing the benefits in the face of significant year-to-year variability in lightning exposure. This use of lightning information does not require the precise timing data from advanced protective relays that now makes detailed fault detection and analysis practical.

University of Arizona, lightning data
NALDN coverage area and flash detection efficiency for cloud-to-ground lightning.

Fault Detection and Analysis

The detailed analysis of individual faults and seasonal fault behavior has emerged as one of the most valuable uses of lightning information in transmission-related application:

  • The ability to make near-real-time decisions to do a manual reclose, if the lockout was related to lightning, can have sizeable reliability implications.
  • Next-day assessment of lightning and sequence-of-event recorder data can be used to evaluate insulation coordination, follow up on unusual lightning damage cases and make decisions about line inspections.
  • Long-term accumulation of challenges and time-correlated faults can be used to identify poor-performing sections of lines and schedule specific remedial actions.

One clear example of this is a line in the Duke Energy service area. The line supplied approximately 25% of the retail customers in this area and was historically one of the worst performers on the system in terms of both outage duration and frequency. While confident lightning was the primary culprit, the utility required a quantitative analysis of time-correlated faults with lightning.

The analysis determined that almost all the outages occurred as the result of modest-current lightning on a very small section of the line. Targeted mitigation efforts, such as grounding improvements and arrester applications, dramatically improved performance. Given this line was 35 miles (56 km) long, a lot of time and money was saved by focusing improvement efforts on the problematic section.

Acknowledgement

The authors thank the following individuals for valuable discussions during the preparation of this article, helping to ensure it reflects the current operational value of lightning information to the electric utility industry: Mitch Cowan of Georgia Power, Eric Engdahl of American Electric Power, Theo Laughner of Tennessee Valley Authority and Mark Matthews of Duke Energy.


Kenneth L. Cummins ([email protected]) was the R&D manager and chief scientist for Vaisala’s thunderstorm business unit (formally Global Atmospherics) until 2005. He is currently a research professor in the Institute of Atmospheric Physics in the atmospheric sciences department at the University of Arizona. Cummins is a senior member of IEEE and has served on various IEEE and CIGRE working groups related to lightning. He has received NASA Silver Medals for his service on NASA’s Lightning Advisory Panel and his contributions to unmanned aerial vehicle-based observations of thunderstorm electric fields.

William A. Chisholm ([email protected]) is an IEEE Fellow and past chair of the IEEE Power & Energy Society Transmission & Distribution Committee, a former member of the Kinectrics technical staff, and now a utility consultant dealing with adverse weather effects including lightning, icing and low wind conditions.

 

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish