Elevating operations practices to be best in class

Elevating operations practices to be best in class

Jim S. Tracey and E.K. Mayfield

Introduction
'The Flight to Technology'
Distribution Operations has recently become a primary area of emphasis for the application of new technologies. While over the last decade there have been technological advances in areas such as field equipment and telecommunications equipment, the basic processes of distribution operations relating to outage management, network management, and power quality are still predominately supported by manual or semi-automated processes. In some cases, there have been some significant levels of automation where home-grown utility applications have been implemented to support the operational needs, but they often are standalone solutions that do not keep up with the growing and ever-changing demands of the utility they support.

This is especially true where there is a need to interface with other processes outside of distribution operations such as asset management or accounting.

Over the last several years, there has been a tremendous increase in the demand for technology that can automate and improve operational processes. This "flight to technology" and the emphasis on automation has been slowly evolving as a result of changes over the last 10 years in the utility environment. The past and current emphasis on reducing costs has reduced the number and experience of operations personnel while, at the same time, utility infrastructures have grown in size and age. Due to these factors, utilities are looking to technology to help deal with the loss of resources and experience. However, the introduction of technology in distribution operations has proven to be much more challenging than in other business areas of the utility such as customer service, call centers and even engineering. The high demand for reliability, performance, and usability has resulted in many project failures where the technology did not live up to expectations. Why is this?

Usability is probably the one key reason (performance and reliability are outright demanded and expected). In many cases, the functional goals of proposed technology solutions are not linked to the benefits that were originally identified in the business case. Nor are they specific enough so that they can be proven later on. For example, some typical benefits identified in business cases for an outage management system are:

Reduce service unavailability (SAIDI)
Improve customer care
Reduce operational costs
Improve operational efficiency

Sound familiar? Who can argue with them? The big problem with the general nature of these goals is that unless there is a complete and detailed understanding of the real deficiencies of the current processes, the application of technology fails to address the core problems. Although end-users may be given new tools that change their entire workflow, if the underlying process problems were never identified, the tools can not help them-or the utility-to achieve the original goals.

An example - Identifying the improvement areas
One of the primary benefits of an outage management system mentioned above is reducing service unavailability (known as the System Average Interruption Duration Index - SAIDI). SADI = Total Customer Minutes Interrupted / Total Customer Served SAIDI is a standard in the United States for measuring the ability of electric utilities to minimize the effects of service interruptions on their customers. It is typically used as a summary of the utility's entire restoration process over a period of time and is calculated as follows:

When you look at the components of SAIDI you will see the following: SAIDI = Sum(# Customers Affected X Length of Interruption / Total Customers Served While the SAIDI index itself is a standard, the basic definitions of each of the factors in the numerator of the equation are not. The numerator represents many different types of interruption and restoration processes. The processes reflected by the numerator are numerous and unique to the types of interruptions that occur, as well as the characteristics and types of information available to the utility at the time. Figure 1 reflects some of the inconsistencies typically found in the numerator and denominator of SAIDI.

Figure 1 - Defining Standards for SAIDI

The first challenge is to define the components of the index so that reporting is consistent within and across electric utilities. A utility's ability to define the components listed above is often based on its level of automation and the results can be advantageous or disadvantageous. In some cases, the utility that has better reporting capability will look "worse" than the utility that has little or no automation. For example, a utility that does not have SCADA will use the first customer call as an outage start time instead of using the breaker operation time. Likewise, feeder outages appear longer in duration for a utility that has SCADA and that uses breaker operation time in automated outage management systems (OMS).

Targeting improvement areas
'If you can't measure it, you don't understand it'
It is not always necessary to wait for the application of technology to address many of these issues and to start measuring and defining the potential improvement areas. Once the criteria for calculating service unavailability has been determined, the steps to identifying improvement areas are as follows:

Stratify the types of outages by their contribution to service unavailability (see Figure 2)
Target areas for improvement that can be addressed immediately and easily (pick the low hanging fruit)
Define the components of the current processes and measure them. Identify improvement areas.
Change the processes and measure again (this step may still not involve technology yet)
Apply technology where appropriate and measure again.

Steps 1 & 2 - Stratify the Interruptions & Target Improvement Areas
Figure 2 shows a typical distribution of the types of outages utilities experience over a period of time (several months or a year).

Figure 2 - Typical Stratification of Outages

Figure 2 categorizes outages by the protective device (feeder breaker, lateral fuse, etc.) that opened, and organizes them by how much they contributed to service unavailability. The outages are measured against total customer minutes interrupted (CMI) on the left and the percent of the total CMI on the right. The line with the dots reflects the cumulative percentage for each category of interruption. Not surprisingly, the infrequent feeder and recloser outages are typically the biggest contributors to service unavailability because of the number of customers they affect. Lateral, transformer, and other outages, while more frequent, have a lesser impact on overall unavailability. This paper focuses on feeder and recloser outages because they offer the biggest opportunities for improvement in most utilities. Reducing the duration of an outage on a feeder with 1000 customers by one minute will have a more dramatic impact on service unavailability than reducing a lateral or transformer outage by one minute (assuming non-feeder interruptions remain somewhat level).

Step 3 - Define the Process Components and Measure Them
To improve the duration of any of the categories by one minute, the components of the restoration process for each category must be defined and measured as well. There may be 10 different processes used by a utility to restore feeder outages depending on the type of feeder (overhead versus underground), the level of existing automation (SCADA versus no SCADA), and the information available at the time of the outage (such as known fault location versus patrolling required). As in any industrial process, the goal in improving any of the restoration processes is to reduce variation. So each process should be documented and have a measurement of its own. For feeder restoration, the goal should be to utilize the exact same steps for each process identified. The key is to identify the components of the restoration process to whatever level of detail is possible so it can be measured and improved upon. Figure 3 shows a breakdown of the steps involved in a typical restoration procedure.

Figure 3 - Components of the Restoration Process

Step 4 - Change the Process and Measure Again
For many utilities, the first time they can measure their performance in each of the process steps is when technology (such as an outage management system) is applied. This explains why it is possible for reliability indicators to look worse after going from a completely manual process to an automated one. Of course it is really due to the fact that the reporting is more accurate.

The steps outlined in Figure 3 apply to most types of outages and are typically time-stamped so that they can be measured by category. In the case of feeder outages, we can group the steps in the following categories and determine some of the countermeasures that can be applied to improve them individually. If we can reduce each category by several minutes, the result is a dramatic reduction in CMI for the entire outage, as well as for the entire category of outages for a specified time period. For example, some of the process improvement changes that should be considered for the steps in Figure 3 are listed below. In some categories, there are many more considerations that could be listed, but it was just not possible to do so in this paper. These considerations may not be new to everyone but they certainly are not always considered. This is especially true when there is not enough operations experience involved in the planning.

Outage Generation Time (Steps 0-1):

Outage generation time is the period of time from the actual start of the outage to the time the utility determines the probable device that has opened. For SCADA controlled feeders, there should be little delay in generating the outage. There are unique situations (such as storms and other conditions where temporary faults are suspected on overhead feeders) where consideration should be given to try to reclose feeder breakers within a specified time period. In many cases, this can avoid a SAIDI-defined interruption entirely and put the incident into the momentary bucket (MAIFI). During storms, when distribution feeders are affected by these conditions, this process can have a 50% or higher success rate and have a dramatic impact in reducing CMI. To take advantage of this procedure, appropriate feeders would have to be marked as candidates for reclosing when temporary faults are suspected and there would have to be a means to determine that no information is available to indicate conditions such as downed conductors.

For non-SCADA feeders, the delay in generating a feeder outage can be significant, even with automation. Even manual prediction procedures can be improved if it can be determined that customer calls are coming from a minimum of three or four different protective devices downstream on the same feeder. In some cases, remote-monitoring devices can help as well, if the information is available to the dispatch center.

Queue Time (Steps 1 - 4):
Queue time is probably the biggest contributor to CMI especially during storm events when resources are already being utilized. Some key areas for improvement are:

Use of single-person investigators. This has become popular as it can double the number of resources available to quickly restore the majority of outages (those that do not require major repairs). Doubling up investigators to perform feeder restoration switching can also have a dramatic impact in reducing CMI.

Use of non-trouble crews during storms. Identifying crews that can be quickly utilized during storm events can help reduce queue time, especially if the outages that require major work can be identified (such as pole hits or wire downs). This will not only help the queue time for the outages the multiple-person crews need to handle anyway, but will also keep the investigators (single-person crews) moving and not getting tied up investigating outages they can not repair.

The ability to quickly add investigators and dispatchers when experiencing high volumes of outages. Queue times often become extended when resources (such as contractors) that can help during severe conditions are not identified in advance.

Other improvements to be considered are prioritization of outages based on customer type (industrial vs. residential), trouble reported type (such as an accident), and number of customers affected.

Travel Time Steps (4 - 5):

For feeder interruptions, travel time can be reduced in several ways:

Rapid identification of the closest switchable device to the current location of a crew that can restore a large number of customers.
Accurate switch device locations available on demand.
Again, using multiple crews for feeder restoration, when possible, will cut CMI dramatically.
7 x 24 crew availability.
SCADA-controlled switches on remote devices or the worst performing feeders of a utility.

Restoration Time (Steps 5 - 6):
This category has many types of processes that could be utilized. For feeder interruptions, the processes must be clearly identified so that when the type of outage is recognized, the specific process can be utilized. The types of processes to be utilized are dependent on the following factors:

Fault known: Immediately isolate and restore rest of feeder.
Fault unknown: Use pre-determined switch points on the feeder to achieve partial restoration in conjunction with patrolling. Using multiple crews will cut the time significantly.
Full restoration: Customer-reported or investigator information (such as a tree in conductors reported) indicates it may be quicker to remove the reason for the fault and get all customers back with one step.

Referral Queue, Travel and Restoration Time (Steps 6 - 11):
When multiple person crews are necessary for feeder restoration, the ability to identify available resources is particularly important, especially in the off-hours. Many of the same considerations identified in steps 0 - 6 can be utilized for the referral process.

Assuming that most of the feeder restoration efforts occur in steps 0-6, improvements in steps 6- 11 typically will not have the same level of impact on CMI. However, steps 6-11 tend to be longer in duration and must be addressed to avoid extremely long interruptions (such as underground cable failures).

Step 5 - Apply the Technology and Measure Again
Bring on the technology! As mentioned previously, the automation is sometimes applied earlier, but it is still important to know what the measurements were before so the improvement opportunities can be measured after to determine if the expected impact was achieved. While the measurements taken before may not be completely accurate, this is better than nothing. If technology is to be selected and implemented after the improvement areas are identifed, these areas of improvement must be the basis of the technology decision. Some of the biggest reasons for failure when technology selections are made are listed below:

Turnover in team members that did the original analysis.
Little or no business unit representation. Operations does not "own" the decision.
The procurement and implementation of the technology is considered to be the end and not the means.
The accuracy of the network and customer data (the data by which SAIDI will be calculated) is not considered in the decision. This is one of the biggest success factors in OMS implementations.

It is extremely important to focus the criteria for the technology selections. The automation should help sustain the benefits already achieved and add additional reductions that only automation and technology can provide. Examples of where technology provides breakthrough benefits are:

Outage Generation Time (Steps 0-1):

An outage management system can drastically reduce the variation that may exist in determining the probable device involved in the interruption. Providing configurable rulebased algorithms will provide accurate and consistent outage prediction with little manual intervention.
Outage management applications can quickly identify the feeders to be considered for reclosing and can quickly determine if information is available that indicates dangerous conditions (such as a pole hit). The affected circuit can also be displayed in a full graphical display.

Queue Time (Steps 1 - 4):

Automated functions that support the management of crews can drastically reduce decision time as to which crew is available and is closest to the areas of concern.
OMS can also automate the entire allocation and re-allocation of work that occurs when outages are generated and later when resources are shifted. (Shift change is a common issue for dispatch centers.)
Prioritization and re-prioritization of outages can be automated (based on critical customers, dangerous conditions reported, etc.)

Travel Time Steps (4 - 5):

Technology can provide better location information for field crews. Spatial information is available in technologies such as Geographical Information Systems (GIS). GPS can provide accurate routing of crews in unfamiliar geographical areas.

Restoration Time (Steps 5 - 6):

Technology can assist in switching procedures by providing full spatially-referenced network management tools. These tools can reduce the time required to decide which restoration process to use and can simulate results to ensure the desired outcome.

Referral Queue, Travel and Restoration Time (Steps 6 - 11):

When necessary to involve other crews or locations, technology can provide the functionality necessary to manage the referral work so that the same tools and the resulting benefits can be used in the referral process. This includes deciding the types of crews necessary and the transfer of information relating to the efforts performed by previous crews.

The examples above are just a small sample of how technology can support the targeted improvements for each utility. However, the implementation of the technology is not the main success factor. Success is achieved when the targeted improvements are realized and verified through measurement. This is challenging because the success requires the operations personnel to take a direct role in planning, implementation, and training.

If planning is done up front, the improvements can become evident on the first day of implementation. Even modest gains in improving the individual components of the restoration process will be evident for each outage type. When looking at SAIDI over a period of time, the effect is much more dramatic.

While this paper focuses predominately on feeder interruptions, non-feeder interruptions such as lateral and transformer outages can also have dramatically improved restoration times using some of the same techniques. They require different supporting processes to be developed and measured. Queue time, for example, can be dramatically improved by applying technologies such as mobile data terminals.

Summary
The challenges in elebrvating operations practices and implementing technology in distribution operations are greater than in any other area of a utility, but the potential for breakthrough gains are significant. The demand for improvements is increasing and the new competitive actions taken by utilities, such as mergers and acquisitions, are making the need for improvements and technology greater than ever. There are also more infrastructures to manage spread across large geographical areas. Utilities must take advantage of the right applications of technology in conjunction with improving core operational processes. Their success will depend on how well the problems are defined and measured, as well as their ability to select and implement the technologies that best support the areas targeted for improvement.