MANAGING DISTRIBUTION SYSTEM RELIABILITY– INNOVATIONS TO ENABLE THE NEXT RELIABILITY BREAKTHROUGH
Advances in IoT (Internet of Things) technologies and Artificial Intelligence (AI) are enabling breakthrough applications for distribution system management. Proactive asset management and outage prevention are the next frontier of applications that are being adopted to improve SAIDI, SAIFI, and MAIFI metrics simultaneously while delivering significant efficiencies in optimizing O&M costs and reducing risk of wildfires. As often stated, the best unplanned outage or asset failure is the one that never occurs! But deploying innovative solutions in the utility sector takes more than just the technology. It requires a proactive attitude and a culture of innovation to break down the traditional silos and champion change management across the enterprise. In this article, we delve into the nuances introduced in this paradigm shift and provide a recipe for success using experiences gained from early adopters.
Is it possible to predict outages and affect the change management process?
Outages in distribution systems have many causes and each distribution circuit has unique electrical and load characteristics. Therefore, it’s worth looking into the different causes of distribution system outages. Some are very random and unpredictable, such as incidents caused by a motorist hitting a power pole or an animal causing a short on the power line. But two categories of faults which are among the leading causes of outages are often instigated by incipient faults prior to a full-blown failure: faults caused by equipment failure and vegetation encroachment.
These two types of faults can be predicted reliably using intelligent devices and advanced analytics applied to the grid data they produce. These devices can be configured to capture circuit disturbances, some of which are good indicators of a fault in the making. By segregating noise (operational signatures) from the signal (failure signatures) using domain-specific analytics and machine learning, it’s quite possible to detect early signs of failures and track them over time. These predictors, a.k.a. precursor anomalies, often manifest themselves as small deviations in the current and/or voltage which are too small in magnitude or short in duration to be detected by conventional protection equipment or SCADA systems.
In reality, a vast majority of disturbances represent normal operating conditions such as a large motor starting, a capacitor bank switching in, or a sympathetic response to adjacent feeder events. These operational anomalies need to be classified and filtered out through advanced analytics so we can further identify whether the remaining anomalies match signatures that are indicative of an impending failure.
Intelligent line devices have a higher sampling frequency (130 samples per cycle or higher) and sensitivity than conventional protection devices. They can be cost-effectively placed in multiple locations along the feeder for more locational precision. Often a set of devices are installed along a feeder to segment the feeder into 3-5 mile segments. This becomes important for location precision when a proactive action needs to be taken on predictions, which is discussed later.
Is the enterprise ready to act proactively and preempt outages?
This may be the most important change management question for utilities to consider. Predictive analytics provides breakthrough results for reliability improvement if proper processes are in place. Taking actions to investigate, locate, and mitigate impending outages creates the need for new processes or modifications of existing processes. It takes utility innovation, regulatory support, a proactive attitude, and a culture of innovation to break down the traditional silos and champion change management across the enterprise.
To be successful in achieving breakthrough results, it’s important to assess readiness in four key areas:
Planning, Operations, and Maintenance Readiness
Let’s take the case of an equipment failure prediction. When an alert is issued in reaction to emerging precursor anomalies indicating a high probability of an impending equipment failure, this information is sent to a responsible unit e.g. a circuit planner to investigate. By comparing data from multiple devices on the feeder, we can identify the segment of the feeder and the phase where these anomalies are occurring. This is important because if we can give the field inspectors a smaller area to investigate, the probability of finding the failing equipment increases significantly.
The target feeder segment inspector is looking for possible culprits. It could be that arcing is occurring due to a pin insulator that is broken, a cracked fuse cutout, or a conductor lead for a lightning arrestor has detached. These are problems that an experienced inspector can visually identify during a targeted field inspection.
Once the issue is identified, a quick repair is needed from the maintenance team in order to preempt the outage. After all, the outage clock is ticking. Best practice is to inspect and repair within 1-2 days for the best chance of improving reliability metrics.
If the planning, operations, and maintenance teams are able to inspect a feeder segment and repair problems quickly, then they could support this type of process for preempting outages without significant change management.
Vegetation Management Readiness
The process for preventing vegetation outages is similar to equipment failure in some ways but different in others. There are three steps – PREDICT, INSPECT, REPAIR, but the REPAIR in this case is often tree trimming.
The action from the predictive model indicating an impending outage due to vegetation encroachment would ultimately go to the utility’s vegetation management contact and this person would perform a routine inspection of the identified feeder segment. Once the vegetation encroachment is identified, the next step is to have a vegetation management crew trim, ideally within a day or two of the alert for best reliability results.
It's important to assess how tree trimming is managed. Normally there is a cyclic trimming program that may repeat every 7-10 years. The trimming plan is optimized for efficiency as it is resource and equipment intensive. The cyclic trimming program is not a feasible option to address active vegetation encroachment causing incipient faults as this type of outage mitigation is reactive and requires a faster response to preempt the outage. A proactive data driven trimming capability or team must be part of the overall vegetation management program in the new proactive paradigm.
If the utility has data driven trimming capabilities and can respond in 1-2 days, then they are ready to support preempting outages with predictive analytics. To facilitate this practice, the optimal approach is a phased change management. Start by building trust in the outcomes, receiving alerts, validating the alerts through routine inspections, and then confirming the subsequent outage. When trust is built with the stakeholders, then it becomes easier to justify the creation of a nimble spot trimming program to realize the associated reliability gains and capital efficiencies.
Information Technology Readiness
Proactive outage and asset management requires more data and analytics than traditional IT programs. It’s essential that the IT organization is ready to support such a progressive initiative. One big factor is the use of scalable deployments. Most utility operational technology (OT) systems run in on-premises data centers, but for the advanced solution described above there are advantages to using a scalable software deployment model that allows for life cycle management of machine learning models, security by design, and elastic computing.
Leadership Readiness
Utility analytics relies on new technologies, and therefore return on investment is less predictable over a short time frame compared to traditional solutions due to the probabilistic nature of predictions. To ensure successful outcomes, the leadership needs to be comfortable with more uncertainty in the payback timing of these deployments and the change management aspects. It’s a key enabler to go from pilot to production. One way to overcome the initial stigma is to track both “realized” and “unrealized” benefits.
Realized reliability improvements are when a predictive alert is sent to the utility, the inspector finds the failing equipment or vegetation encroachment, the equipment is repaired or trees are trimmed, and the outage is subsequently avoided.
The realized value can be calculated in various ways. One method is to estimate the CMI that would have been observed had nothing been done. This calculation utilizes feeder level historical CAIDI and estimates the closest upstream protective device that would have operated had the outage occurred. For example, avoiding an estimated 3-hour outage for 50 customers would yield a potential CMI benefit of 9000 CMI. If CMI is valued at $2 per minute saved, $18,000 of value is recognized as “Realized”.
In some cases, there will be unrealized benefits, too. There are several scenarios where benefits are unrealized after a predictive alert is raised to a utility:
-
An inspector is not able to locate the failing equipment or vegetation encroachment through a visual inspection of the identified feeder segment
-
An outage occurs before the inspection or repair is completed
-
Problem is found during inspection but there is no process/effort to repair
Utilities should identify which unrealized scenarios could be addressed with change management and the associated reliability value of potentially preventable outages. For example, let’s say vegetation contact is identified during inspection, but a crew was not sent to do the necessary trimming and an outage occurred with an estimated CMI value of $12,000 [outage affected 60 customers and lasted for an hour]. This is the unrealized value that could be realized with changes in vegetation management processes.
For the leadership, it’s important to track and quantify both realized and unrealized value. Realized value is used to measure the ROI of the program and unrealized value is used to drive change management and program effectiveness to shift the unrealized value into realized value over time.
Summary
For innovative electric utilities looking to materially enhance system reliability and resiliency through proactive outage and asset management, intelligent line sensing and predictive analytics offer a cost-effective path to realize that vision. Field results from existing deployments show solid ROI when change management is addressed and predictive alerts are acted upon. This requires an assessment of the needs and plans to enable this journey into the future. It is vital that the planning, operations, maintenance, IT, and executive management are ready, able, and willing to embrace the necessary change management elements needed to make this type of ground-breaking project successful. In short, the rhetoric of innovation must be matched with genuine action and implementation.