Lightning, or the related cause of equipment failure after a lightning storm, has been found to be the cause of many distribution-system interruptions. Such is the case for National Grid (serving portions of New York, Massachusetts, New Hampshire and Rhode Island), which has developed new methods to view its reliability performance in combination with geographic information system (GIS) and lightning-performance data to mitigate future lightning interruptions where it will achieve the most reliability improvement. National Grid is also mining its existing data for new information to find areas where it can improve standards or construction practices before major lightning-caused impacts occur.

METRICS

The system average interruption duration index (SAIDI) is developed by dividing the cumulative customer interruption durations by the total customers served. This commonly used metric for the “average” customer reliability can be impacted by system design, construction practices or maintenance. One additional metric, helpful for the analysis of lightning-caused events, is the number of lightning-caused interruptions on a feeder by year. This metric varies due to the system capability to resist lightning damage and also to the lightning challenge experienced in the area served by the feeder. With geographic information of each lightning strike, the variability resulting from nature can be minimized, allowing system-condition trends to be more visible.

National Grid has historically collected interruption data for its system. In recent years, this process and the resultant data has improved with the aid of automation, intelligent electronic devices (IED) and increased attentiveness to the reporting process. In addition, system attributes such as location of customers, physical plant type and location, protective devices, and so forth have always been tracked. But, through GIS, access to and use of this data is now more efficient, allowing the company to conduct more analysis with it.

The National Lightning Detection Network's (Vaisala; Tucson, Arizona, U.S.) data and GIS now allow the mapping of specific lightning strokes into multiyear densities. By overlaying this data in polygons of small areas, such as operating areas or ZIP codes, engineers can easily determine areas where lightning is more likely to strike with a finer granularity than was previously possible. In addition to the location, this lightning data has information regarding the current and polarity of each stroke. So this, too, can be factored into the analysis.

MODELING

National Grid began by downloading interruption database extracts for lightning-coded interruptions only. To avoid single-year peculiarities from skewing results, five years of data are used to arrive at typical annual values. Four rankings are developed in the model. The number of customers served is ranked from high to low. This data helps to model the future avoided interruption value. The more customers served, the more likely greater customer minutes interrupted (CMI) will be incurred for a future interruption.

CMI per interruption event is ranked from high to low. This ranking provides a historic severity of interruption events. CMI is proportional to SAIDI when the number of customers served is large and fairly constant.

Another ranking is interruption events per mile, which is ranked from high to low. This provides an indication of the density of interruption events.

The fourth ranking is cost to mitigate in dollars per change in CMI avoided through mitigation. This is ranked from low to high and measures expected mitigation effectiveness.

The expected CMI improvements by feeder were developed assuming a less-than-perfect result from remediation efforts. Based on consensus discussions with operations engineers and from previous studies, National Grid assumed a 70% success factor. The estimates of labor hours required to mitigate were obtained from field-survey estimates, sampling of construction orders and discussions with field operations personnel.

Lightning-mitigation measures typically include such things as proper grounding, bonding, arrester application and insulation coordination. The most significant means for dissipating the transient voltage wave induced, due to a nearby lightning strike, is the use of arresters. Grounding and bonding help reduce the possibility of equipment damage.

The four rankings noted are combined into a single ranking using equal weighting for each. This allows cost/benefit curves (in dollars per change in CMI improvement), similar to Fig. 1, to be generated. Each point on the curve represents a specific feeder on the system. The resources in terms of dollars, man-hours and the expected improvement in CMI for each feeder allow for the selection of those feeders with the least cost and most improvement. Thus, an efficient overall prioritization is possible, as well as an associated budget and work force requirement by feeder or area. This business case can be compared against other reliability improvement cases (such as vegetation management and animals) to achieve the best program mix for reliability improvement.

NORMALIZATION

Mitigation should result in performance changes that can be tracked using the analysis method. Observing feeder performance year by year prior to and after mitigation should show a step change in performance. This is often not obvious due to the variable year-to-year nature of lightning in an area. The randomness of lightning can be accounted for, to a degree, and the benefit obtained for a certain cost can be calculated by normalizing the results by the amount of lightning for that period.

To normalize the variability of lightning, a per-unit exposure factor is created by dividing the number of lightning strikes in a single year by the maximum annual strikes for the period studied. This factor is then used to adjust the annual CMI values.

The five most recent years of lightning density by ZIP code for the area studied was used to normalize the reliability results obtained from the lightning-mitigation program and to compare this program against all feeders normalized in the same way. Figure 2 shows the lightning counts in the area where mitigation was performed. Figure 3 shows the reliability improvement with and without normalization.

Without normalization, the rate of worsening performance slows several years after mitigation is complete, but no significant improvement is visible. The significant increase in lightning activity in the area served by these feeders masks any benefit obtained from the mitigation. The lower curve in Fig. 3 shows the benefit if lightning had remained more stable. Here a significant benefit is visible. Figure 4 summarizes the CMI for the mitigated feeders and plots them against a backdrop of CMI for all feeders that were not mitigated in the area those feeders serve, but were normalized for comparison.

A marked improvement in reliability performance has been achieved through the lightning-mitigation program. The improvement is also contrasted by the system trend for unmitigated feeders, which is deteriorating. National Grid will continue the program since it was able to achieve and document significant improvement in reliability performance.

ANTICIPATION

The previous analysis does a good job of targeting the feeders that are performing poorly and tracking results to ensure mitigation is effective. While a valid approach, it misses the mark for determining and highlighting the adequacy of standards, construction practices and deteriorated arresters or grounding in areas not hit hard by lightning during the analysis period. The challenge is to identify low lightning-density areas that are performing acceptably when compared against all areas, but that are performing poorly when compared to other low-lightning areas. With this understanding in hand, an anticipatory program can be initiated to address lightning mitigation before a high number of interruptions are actually experienced.

Taking the interruption data and normalizing it by the location-specific lightning-density data and then displaying this new metric geographically, it is easy to see where on the system lightning-caused interruptions are being experienced with higher or lower frequency than expected given their exposure to lightning. Figure 5 shows the performance of the study area with respect to lightning, normalized by lightning-strike exposures. This can be very informative regarding the system's lightning-hardened status relative to other areas. Knowing how hardened a portion of the system is can help predict the future likelihood of inadequate performance.

Some areas impacted often by lightning (for example, location A in Fig. 5) perform better than other areas impacted infrequently by lightning (for example, location B). Either the construction standard is not being adhered to, the standard is inadequate to provide acceptable performance or equipment such as arresters have deteriorated. A quick field review of a small number of feeders will uncover which scenario is the case.

If the former is true, re-education of the labor force, spot inspections of completed jobs, and end-of-job walk-arounds should improve future performance. Remedial mitigation will bring the feeders identified to an acceptable performance level.

If the standard needs improvement, the lightning-data history will provide valuable data about lightning density, the distance of strikes to the lines, current magnitudes and polarity for use in reviewing the standards. Also, this information will help to decide whether the changes from the existing standard needs to be for the overall system or just for a particular portion of it. If deteriorated equipment is the root cause, a program for system hardening can be identified from the information at hand.

HIT THE TARGET

Modeling for prioritization and for business-case development allows for targeted mitigation, which optimizes dollars spent and results obtained. Using GIS tools in conjunction with interruption data and lightning data has expanded National Grid's ability to visualize and highlight areas where standards or their implementation are not achieving the desired results, or where equipment has deteriorated. Inspecting a couple feeders will identify which is the case. Thus, appropriate corrections can be made before denser lightning patterns are experienced, and larger more significant interruptions can be avoided.

Taken together, the ranking model and the GIS analysis form a good basis for deciding where to harden a system to improve its performance with respect to lightning. The tools allow backward views to ensure desired results are being achieved and forward views to target improvement before performance becomes an issue.

---

Vincent J. Forte, Jr. is a principal engineer in asset strategy and investment planning for National Grid. He is a licensed professional engineer in New York State. Forte has held several engineering and management positions in the electric utility industry. Vincent.Forte@us.ngrid.com

Cheryl A. Warren is vice president of asset strategy and investment planning for National Grid. She has held several engineering and management positions in the electric power industry in both the investor-owned utility and the consulting arenas. Cheryl. Warren@us.ngrid.com