By Caroline Winn and Patrick Lee, SDG&E, and Larry Kuhl, Microsoft
San Diego Gas & Electric recently piloted a “real-time condition-based maintenance” solution to bolster its asset management program. So, why is real-time data important for a condition-based maintenance (CBM) approach to asset management? Before we answer that question, let’s start with something a little more fundamental. Why is CBM, or asset management in general, important to a utility?
Click here to enlarge image
For San Diego Gas & Electric (SDG&E), the answer is simple: The need to maximize utilization and performance of existing assets and to reduce maintenance expenditures make CBM and asset management essential. SDG&E’s electric system consists of approximately 1,800 miles of transmission lines (mostly overhead) supported by 18,000 transmission structures. These transmission lines feed 127 greater than 5 MVA substations with 906 distribution circuits, 145,000 distribution transformers and more than 220,000 poles serving 1.3 million customers. These assets represent hundreds of millions of dollars of capital investment, and many are starting to reach the end of their expected life. Maximizing the return on these assets is paramount to SDG&E, its customers, and its stockholders.
SDG&E set out to develop and deploy an asset management program that would improve load forecasting and system planning, system design, system operations, and maintenance management of T&D assets. Determining the optimal time to add or replace a specific asset could result in deferral of capital expenditures. The knowledge of operating conditions and asset health would allow planners to optimize system configurations when planning for network reinforcements and replacements. Consolidated access to operations and equipment data would also provide the planning and operations departments with “one version of the truth” on equipment ratings. Furthermore, having easy access to both real-time and historic equipment condition information would facilitate better decisions to prioritize which assets need maintenance and when, thus reducing operations and maintenance expenditures. Finally, there was the potential for other reliability benefits such as avoiding catastrophic failures as well as improving power quality of the electric system.
The evolution of asset maintenance in the electric industry has gone from Stage 1, “interval-based,” to Stage 2, “condition-based,” and now to Stage 3, “real-time conditioned-based” programs.
Stage 1: Interval-based
For years, time-based interval maintenance practices have been the norm. Utilities typically relied on time intervals to determine when a specific maintenance procedure needed to be conducted. An example would be overhauling SF6 circuit breakers with new seals, contacts and nozzles every 10 years or 4,000 routine operations. This made the budgeting process simple. However, the result was a very large O&M budget. Plus, with interval-based maintenance, you miss what happens to the asset in between maintenance procedures.
Another type of interval-based maintenance practice is the counter-based method. Circuit breaker manufacturers have recommended certain procedures be based on the number of times a breaker has operated. However, this method doesn’t capture all the data necessary to optimally determine when the contacts really need to be refurbished or replaced. To better ascertain when an asset needs maintenance, you need to know more about the asset, like what the asset has been subjected to or is currently being subjected to. Utilities have found that knowing how much fault current the contacts have been exposed to is a good indicator for determining when the contacts need to be maintained. Continuing with the circuit breaker example, seeing that during the most recent breaker operation the actuator motor required more current than for past operations is another valuable piece of information. This knowledge provides advance notice to another type of potential failure and the need to trigger a maintenance work order.
Stage 2: Condition-based
The primary goals for any CBM program have typically been to reduce O&M spending, maximize asset utilization and improve overall asset reliability and performance. These goals have been realized by leveraging the routine collection of data from a variety of sources (i.e. the last dissolved gas analysis, oil dielectric strength, field inspection data like operation counter readings, temperature, pressure, current, voltage, insulation power factor, functional check results, and general asset data like ratings, age, type, and design age, etc.). Many of these data sources tend to be updated at intervals on the order of weeks, months and sometimes even annually. (See Figure 1, “Sample Implementation Overview.”)
Click here to enlarge image
Some of the items discovered with a strong CBM program include:
- Defective controls on older load tap changers,
- Defective counters,
- Low oil levels,
- Cylinder leaks,
- Controls out of calibration, and
- Insulation degradation.
A strong condition-based maintenance program can uncover defective controls on older load tap changers.Click here to enlarge image
The results of CBM programs have greatly exceeded the expectations of some utilities. Utilities have been able to avoid or defer million-dollar-plus capital expenditures and reduce O&M costs in the multi-million-dollar per year range.
Stage 3: Real-time Condition-based
However, to achieve the ultimate objective of avoiding potential catastrophic asset failures, one needs access to not only traditional data sources, but “real-time” data as well. The goals for a real-time conditioned-based maintenance (RtCBM) program are the same as for CBM. The difference lies in that with RtCBM there is potential for learning more about the asset, more timely identification of potential asset issues, and greater O&M cost savings.
While CBM provides access to the asset’s history, allows you to factor in a number of different parameters and is less expensive to deploy than RtCBM, the data sampling frequency does not allow for catching a variety of abnormalities. For example, when paint overspray gets into the cooling fans it could cause them to burn out a day or two later, which could result in an abnormal temperature rise in a substation transformer. Monitoring key transformer temperatures and comparing those temperatures to the temperature projected by the IEEE standard for a given load condition, ambient temperature and transformer characteristics can trigger an alert or alarm when critical thresholds are exceeded. Responding to the alert and taking the appropriate action can save the transformer. It always costs more to replace an asset when it fails unexpectedly than when the replacement is planned. The catch could also preclude a potentially catastrophic outage event and unnecessary stress on other assets.
Consequently, real-time data adds a new dimension to a utility’s analysis capabilities. One example of a real-time calculation would be to use a combination of substation transformer top oil and hot spot temperature monitoring. Using ambient temperature, the current loading, and the specific name plate data, expected temperatures can be calculated using standard IEEE equations that can then be adjusted to take into account other parameters like the results of the most recent dissolved gas in oil analyses, on-line hydrogen in oil monitoring, or on-line bushing power factor to continually monitor the transformer’s real-time condition. This real-time monitoring can trigger alerts to the appropriate personnel when specific thresholds are reached. With this capability, the utility can take action prior to experiencing a potentially catastrophic situation.
So, by adding real-time data you know exactly what each monitored asset is being subjected to. Consequently, you can:
- Quickly determine whether or not you need to send someone out to observe a situation in the field,
- Pinpoint when a particular maintenance procedure needs to be done, and,
- Assess and better predict whether an asset has or will have spare capacity.
Ultimately, you can operate an asset much closer to its design limits, without exceeding those limits.
To address all the primary drivers for its new asset management program, SDG&E looked at various alternatives and ultimately chose to utilize OSIsoft’s PI System to pilot a real-time CBM solution. A number of software factors and core capabilities drove this decision. One key element was the need to have access to both historical and real-time data so the true condition of an asset could be determined. The solution needed to leverage other data sources, such as weekly general inspection info obtained with manual loggers and from insulation power factor (Doble), dissolved gas analysis, outage history, transformer turns ratio, Delta X, substation automation, distribution automation, digital fault recorders, relays, PQ monitors, etc. In addition to data access, the solution needed to interface with a variety of industry maintenance management systems (MMS’s) in order to autosync with asset hierarchies maintained in the MMS, trigger work orders and monitor progress in work orders.
Flexibility was another important requirement as changes to the range of assets, level of asset monitoring and resulting need to add or modify condition asset algorithms was expected. Scalability was another key factor since the number and types of assets were expected to increase over time, resulting in higher data volumes. The level of asset monitoring is also expected to expand. Intuitive web-based visualization (see Figure 2) of integrated information with drill-down navigation abilities were high on the list of capabilities. Using technology that was familiar, easy to deploy, and easy to use was also important to minimize training and support costs. Leveraging the automated assessments, the solution needed to be able to notify the appropriate personnel of required actions via pager, e-mail or phone. Further, the solution had to be able to escalate the notification in case the primary contact could not be reached. Last, but not least, the solution needed extensive reporting capabilities and superior software support.
A sample PI System substation transformer display leveraging Microsoft’s SharePoint collaborative web-based environment for integrating several data sources.Click here to enlarge image
SDG&E is currently embarking on a multi-year real-time CBM deployment. With real-time data and analytics, SDG&E can:
- Spot situations that require immediate attention,
- Perform maintenance only when and where it is needed,
- Operate assets closer to their operating limit,.
- Enhance ability to determine the optimal time to replace a specific asset,
- Reduce manpower requirements to manage data-intensive CBM programs, and most importantly,
- Avoid catastrophic failures.
So what does the future of RtCBM and asset management look like? Advanced metering infrastructure deployments will play a key role. The use of AMR/AMI data will enable more predictive capabilities. Imagine what a utility could do with hourly or more frequent interval AMI data. You would have detailed distribution circuit load profiles so you would know exactly what the substation assets will be seeing at any given point in time. There are also examples of asset signature technology being used to predict asset failures.
The bottom line is that future asset management practices will be even more predictive in nature.
Caroline A. Winn is the director of transmission and distribution asset management at San Diego Gas & Electric Company. Her present responsibilities include development of the asset strategy, investment planning, and technology innovation and development. In this capacity, she provides staff support for T&D construction line activities including distribution planning, electric reliability, compliance management, and information technology management.
Patrick Lee has worked at San Diego Gas & Electric since 1991, holding various positions in electric T&D with varying responsibilities. He is currently director of electric regional operations overseeing electric distribution construction and maintenance, field operations, skills and compliance training, business processes support. Prior to joining SDG&E, Patrick worked for the Sacramento Municipal Utility District and the Electric Department at the City of Roseville in California.
Larry Kuhl is Microsoft’s U.S. utility solutions executive and has more than 28 years experience in identifying, engineering, developing, selling, and implementing solutions for the utility industry. This experience combines 20 years of working for Niagara Mohawk and Carolina Power & Light and eight-plus years of delivering solutions while in leadership positions at Coherent Networks, Osmose Utilities Services and OSIsoft.