Budgeting Decisions for Underground Primary Cable Replacement are often based upon a reactive approach, replace it as it fails. If preventive replacement approaches are attempted, they are typically based upon looking back at the cable fault history in conjunction with estimating the amount of at-risk footage and making an “educated guess” as to how much and what vintages of cable to replace.

Even if accurate fault and length data are available, without the ability to accurately forecast future faults and quantify their impact on reliability, budgeting decisions are still open ended in that future fault behavior and reliability impact are assumed.

For the past four years, the Maintenance Engineering department at Salt River Project (SRP; Phoenix, Arizona, U.S.) has been working on a process and modeling approach to better understand the impact of primary cable replacement on reliability and long-range budgeting.

The preventive underground primary cable-replacement process and modeling approach undertaken by SRP, known as the Long-Range Fault and Reliability (LRFR) model, consists of three main elements: good cable vintage data, the ability to accurately forecast cable faults and a reliability criterion that reflects faulting impact of feeder and underground residential distribution (URD) cable.

Each of these elements are critical to establishing a viable process and model that can provide an understanding of cable-replacement impact on long-range fault behavior and reliability.

CABLE VINTAGE DATA

The cable vintage data consist of annual fault counts and total in-ground length for specific cable installation/vintage dates. Both the annual fault count and length data are used to establish the cable vintage fault parameters. The cable vintage fault parameters in turn are used in the LRFR model, as well as the vintage length totals. Therefore, it is imperative that attention be paid to obtaining good vintage fault and length data. For the past several years, SRP has used a four-step approach to date underground cable faults:

Step 1

At the end of the calendar year, with all cable faults posted to the faults database, a copy of the current year's undated cable faults is extracted and copied to the Cable Vintage Fault database, which was established in 1996.

Step 2

Faulted segments from the current year's undated cable faults are then compared to previously dated faulted segments to check for additional segment faults and to date the current fault should a match occur.

Step 3

After updating the previously faulted segments, the new undated segments are researched to determine the date of installation. Installation dates are primarily determined from the underground operating number Pad Book database. The Pad Book database contains the individual identification numbers of field equipment and their data of installation. Dating is done either directly or indirectly by reviewing the connected devices or all devices and surrounding cable in the area. If no dates are found or if a date conflict exists, installation work orders are researched.

Step 4

If step 3 does not yield a satisfactory result, the Pad Book information is coupled with a review of the area map and the segment is dated by the oldest transformer in the area. An alternative to dating using the area map is by the service start date in the Transformer Load Management database.

SRP has found this process to yield an accurate result for dating underground primary cable.

Determination of cable footage based upon installation dates has been considerably more challenging and has taken major efforts from both Maintenance Engineering and GIS Services. The reconciliation of the footage installation dates and amounts is broken out into four categories:

1. Accepted by Job Numbers

   Installation dates associated with job numbers within GIS are available as a result of new construction, cable replacement or system improvement. However, the majority of this information is only available for cable segments that have been installed after 1998 with the introduction of the current GIS.

2. Accepted by End Device Dates

   RELIABILITY CRITERION

   An installation date is accepted if the installation dates of the devices connected to both ends of the cable segment are approximately the same age.

3. Unverified

   A date has been found associated with the segment but there is no information at present to verify its authenticity.

4. Unknown

   A segment has a null terminator or missing/corrupted date.

LONG-RANGE FAULT AND RELIABILITY MODEL

As of June 2007, approximately 80,000-plus segments remained either unverified or unknown regarding installation dates. SRP is currently in the process of validating the dates and footages on these segments, which represent about 30% of the entire system.

The criterion used to evaluate the fault impact on primary underground cable reliability is a weighted fault based upon customer outage minutes per segment fault.

In researching random 500-kcmil, 4/0 and underground segment faults, SRP found the average customer outage minutes per segment fault on the primary underground cable was approximately 17,500 customer minutes for 500 kcmil and 4/0 and 2500 customer minutes for URD, a 7-to-1 reliability impact. However, rather than using a single weight factor of seven for the 500 kcmil and 4/0, the LRFR model randomly selects the weight factor from a uniform distribution ranging from three to eight for each 30-year forecast. The URD weight factor is always fixed at 1.

The annual reliability for each individual cable type is determined by dividing the forecasted faults by the remaining 100 cable miles and multiplying this value by the weight factor. The overall primary underground cable reliability is determined by summing the weighted faults per 100 cable mile for each primary cable type and comparing this value to its historic average, which serves as the reliability target.

The LRFR model provides a 30-year fault and reliability forecast for individual underground cable types based on a defined cable-length replacement program. Faults in the model are determined from the power law reliability model as previously mentioned. However, to address the 30-year forecast period effect on underground-cable reliability, the assumption was made to increase the fault rate, β, beginning at a cable age of 25 years and at 5-year increments thereafter. The increment in which β is increased is randomly selected from a uniform distribution. λ remains fixed through the aging process; therefore, the cumulative faults/100 cable mile increase in a stepwise manner for each vintage of cable once 25 years of age is obtained.

The replacement length for each primary cable type is allocated as a percentage of the total replacement footage. This percentage remains fixed throughout the 30-year forecast period. If either the 500 kcmil or 4/0 vintage length is exhausted prior to the end of the 30-year forecast period, their replacement length balance is transferred to the URD cable.

The selection of where to assign the replacement lengths is based on ranking each of the cable vintages in terms of faults per 100 cable mile and assigning 100% of the cable replacement to the worst-performing cable, the cable with the highest faults per 100 cable mile. If the amount of replacement length is greater than the worst-performing vintage length, the balance is cascaded to the 2^nd, 3^rd … 10^th worst-performing vintages until exhausted. The faults and reliability parameters are then calculated from the remaining length for each cable vintage in each of the 30 forecast years.

Thirty simulations or trials for each of the 30 forecast years are performed, and the average and standard deviation of the faults and reliability parameter are calculated. The forecast is updated annually based on the currently available fault and length data and annual fault analysis thereof.