Formed in 1903 from the consolidation of more than 400 gas, electric and transportation companies in the state of New Jersey, Public Service Electric & Gas Co. (PSE&G; Newark, New Jersey, U.S.) is one of the oldest and largest combined electric and gas companies in the United States, currently serving 2 million electric and 1.6 million gas customers. PSE&G has always worked hard to provide uninterrupted electric power to customers in its 2600-sq-mile (6734-sq-km) service area for more than a century.

As with any other public utility, PSE&G found it difficult to achieve 100% reliability, because the mean time between failures (MTBF) of transmission and distribution equipment is difficult to predict. Most utilities follow manufacturer recommendations for preventive maintenance using a calendar-based approach to repairing or replacing equipment. PSE&G had taken this same approach for years, but felt the need to develop and implement a more intelligent plan that would allow the company to change maintenance practices from reactive to proactive. It implemented the SAP PM (plant maintenance) system in 2000. While this application improved maintenance, it did not provide the proactive, predictive approach to T&D maintenance for which the company was looking.

Despite this setback, PSE&G did not give up. Upper management knew it wanted to create a more formal method that would allow the utility to make better decisions based on the criticality of the asset, failure modes of the asset, maintenance history and expected replacement timing. The goal was to spend capital dollars more efficiently than it had in the past.

In 2001, PSE&G's Electric Delivery Asset Management Department pursued another solution: launching a project to create a computerized maintenance management system (CMMS) that focused on the electrical T&D arm of the delivery organization and would use a proactive condition-based maintenance approach to help predict equipment failures before they occurred. The primary assets of concern included transformers, phase-angle regulators, high-voltage regulators, load tap changers (LTCs) and breakers.

Installed in 2002, the Real-time Performance Management (RtPM) platform and software application modules from OSIsoft (San Leandro, California, U.S.) make up the foundation on which PSE&G's engineering and IT staffs created this CMMS solution. As a result, PSE&G began to predict failures using a combination of conditioned assessment algorithm results, experience from internal subject matter experts (SMEs) and OSIsoft's client tools (ProcessBook and DataLink).

Calling for a full return on investment by 2005, the system was launched in early 2002 and began delivering cost savings in its first year. In 2003, PSE&G reduced T&D maintenance costs by more than US$300,000. It also is extending the life of aging equipment through more efficient monitoring of performance.

Taking a New Approach

The main objectives of this new system were to:

• Change maintenance practices from reactive to proactive.

• Re-orient corrective maintenance procedures.

• Redirect capital expenditures more efficiently, based on planned replacement of equipment rather than replacement following a failure.

• Reduce equipment failures.

• Better prioritized maintenance orders based on SAP PM applications.

• Capture and institutionalize the expert system knowledge of PSE&G's more experienced personnel to elevate the level of all technical people.

To make the most effective use of maintenance resources for electrical inside-plant assets, the staff developed new equipment condition-monitoring applications that would automatically trigger maintenance tasks and provide management indices that would allow the company to measure program effectiveness.

The CMMS integrates the existing work management, supervisory control and data acquisition (SCADA) and laboratory information management systems for inside-plant assets. Process data is retained in a time-series database for ease of analysis, and transactional data is periodically extracted from the SAP PM module and placed in an SQL server relational database for reporting and analysis purposes.

PSE&G chose OSIsoft's platform as the foundation for its CMMS system, because it allowed the company to store existing control system data and additional operational data in one central place and also because of the fact that it provided an interface to the PM module of SAP enterprise systems. Several OSIsoft modules were used to manage specific tasks for the CMMS, including:

• The PI System for archiving operational data.

• Module Database for sorting real-time data, aliases, parameters and process specs into groups for use in programs and displays.

• ProcessBook, a graphics package for creating dynamic, interactive displays of data.

• RLINK for interfacing to the SAP PM module for generating work orders and notifications.

• Manual Logger (PIML) coupled with mobile data terminals (MDTs) for collecting weekly inspection data.

Building the Platform

Using the OSIsoft tools, PSE&G was able to build a condition-based monitoring platform that looked at factors such as counter operations and gas tests from comparative equipment, and ranked the equipment according to the sum of the factors. This platform has allowed the company to identify comparative asset performance. Conditional-effects algorithms identify the performance parameters of equipment by the class of equipment. SMEs look at the condition assessment results, do further research to verify results and then manually create condition-based PM orders in SAP. The field performs the work and reports on the findings. Based on these findings, modifications and adjustments are made to the algorithms to produce results that are more accurate to the operations of the equipment.

Take the utility's LTCs as an example. The algorithms run monthly and take into account the most recent lab tests that determine how much water is in the oil as well as any hot metal gases, methane and ethane. The algorithm also factors in the number of operations based on inspection data. These raw values are put through case statements and scored accordingly; the subscores are then totaled to a final equipment score, which identifies the condition of the equipment (10 being the worst performer and one being the best). Differences in characteristics are also factored in among equipment from specific manufacturers, since it's been found that some LTCs can withstand high hot metal gas levels, while others have problems with corrosion.

From this analysis, the utility gets a detailed breakdown on equipment costs and man-hours to service. This information was not readily accessible before because it would have taken the staff several months to gather and analyze the information without the use of the RtPM platform. In fact, PSE&G has developed conditional algorithms for use with 982 transformers, 400 LTCs and 1500 breakers.

Equipment in Action

PSE&G began test usage of the CMMS in 2002 and gradually fine-tuned the operations. Condition-monitoring goals for 2003 targeted 16 LTCs for maintenance as a result of wet oil, low dielectric strength and excessive levels of hot metal gases. Of the 16 targeted LTCs, five were found to have impacted contacts and interior components. Overheating and/or failure of internal components of any LTC is a major maintenance item because the cost of switching, teardown and repair is expensive. The estimated savings is more than $150,000 in maintenance and capital costs for LTCs alone.

The overriding goal of condition assessment was to implement preventive maintenance rather than repair equipment following a breakdown. PSE&G generates preventive and corrective maintenance orders using SAP PM. The company uses PI Manual Logger and mobile data terminals (MDTs) to enter weekly inspection data into PI.

There are close to 800 performance equations just to look at LTC operations, cap breaker operations, gas breaker run hours, ATB compressor run hours and gas breaker temperatures. The algorithm looks to see if there has been excessive operations or temperatures over a three-week period, and it uses RLINK to automatically generate a notification in SAP. The best part is PSE&G knows the information is much more accurate and reliable now because there are more people analyzing the data.

In the past, operators recorded weekly inspection data on paper; these papers were filed within the division. Identifying which LTC or breakers had excessive operations over time was a time-consuming task.

With the new system, the PI servers do the work. The PI Performance Equations (PEs) have identified controls that were out of calibration, leaky blast valves, incorrect CMV settings and defective controls on some of the older LTCs that needed to be replaced. The SAP corrective maintenance orders were automatically generated through RLINK via PI PE triggers.

Custom Reports and Data Links

In addition to the preventive maintenance approach, the new CMMS allowed PSE&G to create custom DataLink reports for the field planners to identify heavily loaded areas. The team also worked with the company's system planners and IT staff in Newark to develop performance equations to calculate substation loads. The system planners need to plan for system growth because when electric loads get to a certain demand point, they have to expand capacity or adjust the load.

When the PI System was originally set up, encrypted tag names were assigned to 85,000 points. When users were trained on the CMMS, it was difficult to grasp the tag-naming convention. Therefore, a display was created that would allow users to easily locate and trend PI data. This was easily accomplished via ProcessBook displays, a little VBA code and the PI Module Database. A hierarchy structure was set up in the Module Database that mimicked the SAP IPE structure, then the PI operational data was linked to SAP equipment or functional locations via tag attributes and assigned real names — tag aliases — to all of the points. This display allows the user to navigate the SAP IPE structure in the Module Database and trend the associated operational points.

Knowing the tag-naming convention isn't required to find a point, users just navigate to the desired station, scroll down to the device and get the information. It's as simple as that. DataLink was also used to create custom reports for users, in particular, maintenance supervisors. Now, there are reports that monitor transformer nitrogen pressure readings, LTC and breaker operations, and breaker compressor run hours. Reports are easily accessed. There's one click to refresh, and the latest data is there immediately. Readings that fall below or above a predefined limit are highlighted in yellow; maintenance crews investigate these readings and report findings back to the maintenance supervisors.

Currently, there are about 40 to 50 users accessing PI data, either via ProcessBook displays, DataLink reports or Web Intelligent reports. Sometime next year, the utility plans to implement RtPortal, a highly collaborative environment PSE&G believes will help double the user base.

Bottom-Line Benefits

One important benefit PSE&G realized from the new system is that the SMEs are aging and soon will be retiring. Its idea was to capture as much information as possible to systematize or institutionalize it. Another benefit is the ability to do a better job of expending capital budgets. Previously, PSE&G used SAP order costs and field knowledge to determine which equipment needed replacing. Today, it combines SAP cost data, field knowledge, lab results, load profiles, location and age to determine equipment replacements.

Looking back, the condition-based maintenance system has definitely paid off. Its measure of success was to prevent one major transformer failure every two years. Although it's too early to claim success, the utility's 2003 maintenance and capital cost savings are an extremely good sign.

Instead of operating in a “run-to-failure” mode, PSE&G has succeeded in capturing system knowledge in intelligent systems that allow it to identify problematic equipment before a failure occurs — and there's no real way to calculate the return on investment of a predictive maintenance approach because you can't measure failures that never happened.

Angela Rothweiler is a principal engineer for PSE&G Delivery in Newark, New Jersey. angela.rothweiler@pseg.com