A Distribution Automation Pilot Project is in Place at National Grid that will establish the value relationship between distribution automation (DA) and other options at the utility's disposal for maintaining and improving reliability performance. The DA pilot is also informing National Grid's (Waltham, Massachusetts, U.S.) Smart Grid pilot, which is much more ambitious than DA alone.

These are some of the issues being addressed in the DA pilot and being vetted through firsthand experiences:

• To which circuits should DA be applied?

• Which technology should be used?

• How should the value be measured?

FIRST STEPS

DA includes automation of both distribution (15 kV and below) and subtransmission (23 kV to 69 V) systems. While DA includes supervisory control and data acquisition (SCADA) capability, it is much more than that. The control system sectionalizes (using feeder breakers, reclosers and switches) to reduce the number of customers interrupted and aid crews in trouble location, communicates back to dispatchers where a fault current has occurred or not occurred (using data from devices and fault indicators) and restores as much of the system as possible through alternative sources, but it does all this faster than human intervention (switching manually in the field or using SCADA) can accomplish.

Additionally, DA allows the equipment to “think” by knowing if there is capacity available in real time from an alternate source. This allows as many customers as possible to be restored very quickly in real-time conditions.

In 2003 and 2004, National Grid began investigating DA and potential benefits for the system and customers. At that time, the utility determined that addressing root causes through normal mitigation measures was most cost-effective.

Figure 1 shows a combined view of the cost benefit of these various programs from that time. Each point on the curve is a feeder on National Grid's system and is color-coded by the mitigation program applied. Thus, a feeder would have several points on the curve, one for each program examined. The figure illustrates that the value of the first DA project was well up on the cost-benefit curve.

SELECTION CRITERIA

The expectation was that in a couple of years, the utility would have captured the most cost-effective benefits and moved up the cost-benefit curve to a point where DA value would be relatively comparable to other mitigations. National Grid expected that the DA capability would continue to improve and hoped the cost also would improve or at least remain stable.

Improvement in DA technology did indeed happen and costs remained stable. National Grid captured the low-hanging fruit by addressing root causes. The utility improved its reliability performance, digging out of the regulatory penalties it used to accrue. As the utility approached the cost-benefit ratios comparable to its expectations for DA, National Grid acted to further solidify its reliability-performance gains. Thus, in 2007, the utility started its DA pilot program on a fast track. National Grid moved from interest in DA to DA equipment installed in the field in about a year and a half.

National Grid decided to use the system average interruption duration index (SAIDI) as its measure for reliability performance, because SAIDI incorporates both customers interrupted and interruption duration in the statistic. National Grid wanted rapid change, so it wasn't interested in the risk associated with unproven technology. Serial number one of anything was not acceptable for a pilot that, in addition to growing the utility's corporate understanding, had to deliver reliability improvement on day one. National Grid also wanted visibility for its dispatchers and needed the operations staff to quickly accept any new equipment placed in service.

To identify which distribution feeders should be automated, the following criteria were established:

• The feeder had to have one or more existing manual tie points. National Grid wanted quick wins and was not prepared for the time it would take to create new ties. Of course, as DA matures on the system, the utility will eliminate this criterion.

• Existing ties to alternate feeders had to have reserve capacity in excess of 50 A during peak times. This would mean there would be even more capability during off-peak periods.

• The feeder had to have experienced interruptions on the primary within the last four years. This would ensure the utility was addressing its reliability issues with the pilot and that the pilot would be challenged to perform.

PILOT CIRCUITS

Using these criteria, National Grid identified 290 feeders as candidates for DA. This list was prioritized using dollars per change in customer minutes interrupted (CMI is proportional to SAIDI when the number of customers served is stable).

Estimating the cost was easy. For material cost, National Grid obtained price estimates from vendors and then added in labor for installation. To estimate the reliability value the utility would obtain through DA implementation, an ideal feeder was assumed for the first pass. Assuming an even distribution of customers and an even distribution of interruption causes along a feeder would result in a theoretical maximum value of a 25% reduction in CMI. National Grid realized reality would impact this figure, but for the time being, it applied this factor to the historical four-year average CMI incurred by each feeder. As the utility gains experience, it will adjust this factor for future evaluations.

National Grid focused the subtransmission review in its New York territory. Circuits with more than 10 miles (16 km) between circuits were considered because distances less than that would be very short exposure at this level in the system. Issues on short line exposures are usually best handled using root-cause mitigation methods. National Grid also added to the list any repeat offenders from an annual review. The list of potential feeder candidates reached 17 circuits.

National Grid then prioritized these circuits based on expected reliability improvement using a similar process as that used for the distribution feeder selection. As a final filter, the utility eliminated circuits where major remediation was in progress or budgeted. This work would improve the reliability performance of these circuits; thus, there would be no benefit to adding DA to them at that time.

Six distribution feeders and two subtransmission lines were selected for the DA pilot. An additional four subtransmission lines were selected for a reduced implementation that would allow the utility to test the communications in very difficult terrain while not overburdening the pilot workforce.

TECHNOLOGY USED FOR THE PILOT

All of the noted circuits would employ distributed intelligence for the control scheme. Another four distribution circuits were selected for a centralized intelligence control scheme. As the pilot progressed, the centralized scheme encountered issues that slowed its progress, which were not related to the technology. National Grid decided to place those issues on hold and continue with the rest of the pilot. These feeders and alternative technologies will be revisited as part of the utility's Smart Grid pilot.

The circuits selected for the DA pilot are as follows:

• Duguid 26551, 26552 and 26553 (all are 15 kV)

• Bridgeport 16852, 16853 and 16854 (all are 15 kV)

• HOW TO MEASURE SUCCESS

  Boonville-Lowville 22 line (23 kV)

• Lighthouse Hill-Mallory 22 line (35 kV)

• Chestertown-Schroon Lake 3 line (35 kV)

• Battenkill-Cement Mountain 5 line (35 kV)

• Cement Mountain-Cambridge 2 line (35 kV)

• Cambridge-Hoosick 3 line (35 kV)

NEXT STEPS

The distribution circuits are highlighted in Fig. 2. In addition to the selection criteria noted previously, these feeders were in convenient locations so the pilot team could access them frequently.

This pilot uses distributed intelligence through S&C (Chicago, Illinois, U.S.) IntelliTeam II controls. ScadaMate switches are used as well as the S&C Universal Interface Module to automate Cooper (Waukesha, Wisconsin, U.S.) reclosers with Form 6 controls.

The peer-to-peer communication is achieved with Cellnet's (Norwalk, Connecticut, U.S.) UtiliNet spread-spectrum 900-MHz radios. To keep the communications consistent, National Grid also uses this same radio technology for point-to-point backhaul communication to the nearest points for uplink to the utility's existing energy management system (EMS).

These uplink points are in substations already enabled with EMS, which is a SCADA system principally for transmission and substation control. By uplinking to EMS, National Grid was able to bring SCADA capability of the DA devices to its control centers. In addition, all data (some 86 points per device) was brought back to a central database warehouse.

Table 1 lists the hardware used in this pilot. National Grid used far more repeater radios than anticipated because the severe terrain (elevation changes and heavy long-needle pine forestation) blocked the 900-MHz signals. Despite this, the project flow went relatively smoothly because the utility had anticipated this issue. Therefore, it had included a radio field survey in addition to the typical GIS analysis. Table 2 provides information about the comparison of the original pilot cost estimate with the actual cost. The actual total pilot cost is just under the original estimate.

The principal subtransmission circuits have been operating since January 2009 and the distribution circuit devices are about to be activated. For the subtransmission, National Grid is experiencing better-than-expected reliability improvement and the costs are meeting expectations.

One might ask how the utility bettered its starting assumptions, which assumed theoretical best values. Two factors provided this delta. First, the utility had assumed a perfectly even distribution of load and root causes. Obviously, reality is not so evenhanded. The utility's final filter was to pass the theoretical results by engineers who were familiar with the circuits and allow them to adjust DA-controlled device locations based on their knowledge of root-cause hot spots. These engineers also adjusted based on actual customer locations rather than the assumed even distribution.

These two factors resulted in an improved performance above the model expectations. As GIS data becomes more targeted, National Grid will be able to add this sophistication directly into the modeling process.

The utility has estimated significant reliability improvement. The reduction of customers interrupted is estimated at 1460 for the pilot distribution feeders and 16,000 for the subtransmission circuits. (Customers interrupted is proportional to the system average interruption frequency index [SAIFI]). The CMI reduction is estimated to be 164,000 for the pilot distribution circuits and 346,000 for the subtransmission circuits. (CMI is proportional to SAIDI.)

National Grid has achieved about one-third of its subtransmission annual expected improvements in the first three months of operation. With costs on budget, the resulting benefit-cost ratios are better than the anticipated US$10/ÄCMI to $12/ÄCMI. The pilot feeders are depicted on the cost-benefit curve in Fig. 3. While certainly not inexpensive, the benefit for the incremental cost is within the ballpark of other mitigation measures.

Moving to a production mode will require careful discernment to ensure the lowest-cost mitigation is used. Sometimes this will be attacking the root cause as National Grid has historically done; other times it will be a DA implementation or a combination of the new and older approaches. DA is proving to be not a panacea but a very useful tool to add to the tool bag of options.

National Grid will complete the activation of the remaining units at the distribution level and then monitor the performance of the pilot systems. When sufficient operating experience has been documented, a summary report will be created.

The utility is beginning a pilot for the Smart Grid, which will have a self-healing portion that is informed by the DA pilot. Integration of many facets within the Smart Grid will involve changes to the approach used in the DA pilot in order to maximize value for such things as a common communication system. These changes will not alter the underlying value relationships determined through the DA pilot.

With a thorough understanding of the cost and benefit for DA, National Grid will be in a position to choose the most-successful reliability options with minimal-cost impact for its customers. The utility is gaining firsthand knowledge through the DA pilot and also improving its reliability performance. Results to date are better than expected.

---

Vincent J. Forte Jr. (Vincent.Forte@us.ngrid.com) is a principal engineer in Smart Grid for National Grid. A licensed professional engineer in New York, Forte earned an associate's degree in engineering science from Hudson Valley Community College (Troy, New York) in 1977, and BSEE and MSEE degrees from Rensselaer Polytechnic Institute (Troy) in 1978 and 1979, respectively. He has co-authored papers and articles on customer valuation of interruptions, RF signal transmission over power distribution systems, and asset management methods for improving reliability through targeted mitigation. He is a member of IEEE, the National Society of Professional Engineers and the Eta Kappa Nu Electrical and Computer Engineering Honor Society.

Table 1. Hardware used in this pilot by circuit.

Circuit	Voltage	Number of reclosers	Number of ScadaMate switches	Number of ScadaMate tie points	Number of radios	Number of teams
Booneville-Lowville 22	23 kV	0	2	1	9	2
Lighthouse Hill-Mallory 22	35 kV	0	5	1	21	5
Chestertown-Schroon Lake 3	35 kV	1	0	0	8	0
Battenkill-Cement Mountain 5	35 kV	0	1	0	4	0
Cement Mountain-Cambridge 2	35 kV	0	1	0	4	0
Cambridge-Hoosick 3	35 kV	0	1	0	12	0
Duguid 26551	15 kV	3	1	2	***	4
Duguid 26552	15 kV	2	1	2	***	3
Duguid 26553	15 kV	0	1	1*	***	1
Duguid-Dewitt for back haul	—	0	0	0	***	0
Bridgeport 16852	15 kV	2	1	2	****	3
Bridgeport 16853	15 kV	2	1	3*	****	3
Bridgeport 16854	15 kV	1	1	2**	****	2
Total		11	16	9	153	23
* Counted in another feeder; ** One counted on another feeder; *** 48 radios for Duguid; **** 47 radios for Bridgeport.

Table 2. Pilot cost information (original vs. actual).

	Per unit cost without adders	Original unit cost estimate without adders	Percent of actual spend compared to original estimate
Dist-Central	$68,125.59	$86,787.00	78%
Subt-Central	$69,668.84	$61,762.00	113%
Pilot base total	$68,897.21	$77,218.00	89%