Hydropower, Storage

Using a Predictive Asset Management System to Monitor Equipment Condition

Issue 4 and Volume 16.

By Neil Gregory

Meridian Energy uses a predictive asset management system to assess the health of generating equipment in its hydro plants. The system analyzes equipment condition data and then turns this data into easily understood information. Results are used to make decisions related to equipment maintenance, refurbishment, and replacement.

Meridian Energy, the largest electricity generator in New Zealand, is using a new predictive asset management (PAM) system. The purpose of this system is to provide an early warning of changes in the condition of equipment at the company’s nine hydroelectric stations. These stations provide 30 percent of the country’s electricity needs. The PAM system integrates with Meridian’s existing computerized maintenance management system (CMMS) and plant historian databases. The system — using engineering expertise, international standards and limits, factory test data, mathematical models, rules, and historical data — produces an indicative “percentage health classification” for each piece of equipment.

Background on plant condition monitoring

Plant condition monitoring has been an integral part of the maintenance strategy at Meridian for many years. In the early 1990s, plant condition monitoring typically involved the routine collection of condition data — such as temperature, pressure, and perhaps a physical description of the plant — using a simple record sheet. The results may have been reviewed by an engineer or plant manager before being “filed” for future reference.

In the mid-1990s, plant condition monitoring received a technology boost through the introduction of Maximo, a powerful CMMS. This CMMS made data storage, but not necessarily data retrieval, easier than ever before. Many thousands of measurements could now be stored in a single location.

By the late 1990s, online real-time data was available via plant historian databases. Meridian Energy uses OSIsoft’s PI System plant historian, which gathers, archives, and processes operational data from automation and control systems. Data integrity also improved with the introduction of hand-held data collection devices (such as the Palm Pilot), enabling data to be entered just one time before being uploaded to the CMMS.1

The new millennium heralded an increase in the range of condition monitoring technologies available and the potential to integrate some of the many systems used.

Despite the improvements in data collection and analysis over the past ten years, Meridian, like other hydro utilities, continues to face some interesting challenges in asset management:

  • Aging plants, rapidly approaching the point where they require refurbishment, replacement, or at least increasing maintenance work;
  • An overwhelming quantity of data collected on plant and equipment condition, combined with a lack of resources to both analyze that data and turn it into useful information; and
  • A worldwide skill shortage as experienced staff members, from managers to technicians, upon retiring, leave the organization searching for new opportunities, or simply move to exciting endeavours within the organization.

The above issues create a “vicious circle.” The aging plant demands more attention at the same time experienced staff members are leaving, taking with them knowledge built up over many years. Inspections and test results continue to be recorded, but the expertise and skills to analyze that data is not available. Thus, analysis is not performed on time (if at all), the aging plant continues to deteriorate undetected, and the downward spiral continues.

How can an organization break out of this vicious circle? There are a limited number of options, each with advantages and disadvantages:

  • Recruit more staff to analyze increasing amounts of data collected. The problem with this option is that it is difficult to replace experienced engineers and almost impossible to immediately replace the knowledge they may have acquired over decades. Also, organizations tend to work toward reducing their operating costs, not increasing them by hiring additional staff.
  • Improve existing systems and processes. This is typically the option chosen and fits well with a continuous improvement program. However, this option does little to reduce the intellectual property being lost.
  • Develop or introduce more specialist applications. There are now data storage and analysis applications for almost every conceivable item in a plant. The problem with this option is that it is expensive and requires integrating what are typically disparate systems.

At Meridian, we chose to introduce a “smart” decision support system that could be integrated with our existing maintenance systems.

Choosing a PAM system

In 2005, after an extensive review of the available systems, Meridian Energy selected Matrikon Pty to develop and implement a PAM system for its nine hydroelectric facilities, which have a total generating capacity of about 2,400 mw. ARC Advisory Group, a leading research and advisory firm in manufacturing and supply chain solutions, defines a PAM system as “a combination of hardware, software and services used to assess the health of plant assets by monitoring asset condition periodically or in real time to identify problems before they affect production or lead to a catastrophic failure.”

In addition to meeting the criteria in the ARC definition, Meridian had some other specific requirements for its PAM system:

  • Capture asset management knowledge and intellectual property so that it can be made available to others in the company;
  • Automate monitoring and analysis to provide deep insight into plant performance and condition;
  • Leverage existing systems, such as the PI plant historian and Maximo CMMS, to ensure the information contained in these systems is analyzed and turned into action; and
  • Convert masses of data into useful information that can be acted upon to, for example, perform a simple retest or recommend increasing the frequency of testing.

By integrating different applications into the PAM system, the benefits and functionality of each application can be used to provide a complete health monitoring solution.


Figure 1
The high-level station view screen on Meridian Energy’s predictive asset management system gives an overall assessment of the health of the facility. Clicking on a station takes users to more detailed information on that particular plant.
Click here to enlarge image

The benefits of this integration include:

  • Establishment of the current condition and rate of deterioration of the assets’ “health,” providing early warning on any changes;
  • Recommendations regarding appropriate actions to take to extend the service life of the asset;
  • Assessment and comparison of assets across a wide range of distributed hydro stations;
  • Reduction in maintenance and labor costs;
  • Prediction of impending faults;
  • Capture and retention of engineering knowledge before employees leave the organization;
  • Timely, standardized, and objective analysis of results for decision-making purposes; and
  • Immediate access to information from various locations through a single web-based user interface.

Meridian Energy’s PAM system was installed and commissioned in January 2006. The first application was to monitor a fleet of 34 transformers. Turbines were added to the PAM system in February 2007, and models for the governors and generator monitoring were added in July 2007. Future planned development includes models to assess the health of circuit breakers and batteries at the plants.

How the PAM system works

Meridian’s PAM system was developed primarily to support plant condition analysis based on both on-line and off-line data.

On-line analysis is run at a pre-scheduled interval. The frequency of this interval varies depending on the asset being monitored.


Figure 2
The transformer overview screen provides a DTF (days to test failure) value. This pre-set value indicates the point at which personnel will need to take some action to extend the life of the transformer.
Click here to enlarge image

Results of off-line tests are input into the CMMS as they are completed. Again, the frequency of this is linked to maintenance plans for the specific asset. All data collected through these analyses is then routed to the PAM system, where complex calculations — such as dissolved gas analysis for transformers and oil analysis for turbine guide bearings — are performed. Essentially, the PAM system assesses the plant condition and recommends an appropriate action.

Once the PAM system performs an analysis, it sends an e-mail notification to specific users. This notification alerts users to test results and provides a link to the corresponding PAM page. This link allows a more detailed analysis of the results using a ‘drill down’ process to view the raw data input into the system.

From a knowledge transfer perspective, the value of the PAM system lies in its ability to turn complex data into easily understood information that can be used by:

  • Asset coordinators/plant managers, enabling them to view updates and alerts on changes in condition, as well as flag possible work to be undertaken from an operational perspective;
  • Maintenance personnel, from a maintenance and plant condition perspective;
  • Tactical engineering teams, in order to investigate specific alerts and trend information in support of improved reliability;
  • Strategic engineering staff, to assist with planning and updates to the asset management plan, i.e. maintenance may be deferred or brought forward as a result of improved knowledge of the asset condition;
  • Generation controllers, in order to see alerts as they arise and understand the general health of the assets across all sites;
  • Administrators, to deal with behavior of the mathematical models, such as scheduling, enabling and disabling analyses, running historical analyses, and changes to functionality as the system evolves; and
  • Management, to view the overall “picture” of the asset health by site.

Information in the PAM system can be displayed at several different levels, using the drill down function mentioned earlier. Figure 1 shows a sample high-level station view, created by assessing the health of critical assets at each station. From this computer screen, users can click on a particular station (such as Aviemore) to view more detailed information. The transformer overview screen shown in Figure 2 provides information about health of the five transformers at this station.

One important piece of information supplied on this screen is the days to test failure (DTF) value. For each transformer, this value indicates the number of days left before personnel must take some sort of action to extend the life of the equipment. DTF is not days until the actual complete failure of the asset.

The DTF values are established by Meridian Energy personnel and are moving targets. First, for each test performed on the transformers, Meridian Energy personnel determine the values that would indicate a problem(s) with the equipment. Then, each time personnel perform a particular test, they record the results in the PAM system. The PAM system charts the results over time and determines, given past performance, the number of days until the test result could be expected to reach this pre-set value.

For example, for transformer AVI02 in Figure 2, 918 DTF indicates the days until the value for one of the tests performed to monitor the health of this transformer reaches this pre-set number. At this time, personnel would intervene with the appropriate action. Depending on the test, this action might be a simple retest, or cleaning of the transformer oil to remove impurities.

To determine the particular test result establishing this DTF value, the user can click on that transformer icon (AV102). Figure 3 shows the more detailed information revealed about the particular transformer. On this screen, each test condition analyzed for this transformer is assigned a key performance indicator, represented by:

  • Green (acceptable);
  • Yellow (questionable);
  • Orange (warning); and
  • Red (alarm).


Figure 3
Details of a particular transformer’s condition provided by Meridian Energy’s predictive asset management system feature key performance indicators for each test condition. Indicator levels range from green (acceptable) to red (alarm).
Click here to enlarge image

The PAM system also features e-mail alerts. An alert is sent when an off-line analysis result initiated by the CMMS becomes available or when an on-line analysis produces an exception or indicates an abnormal event. This e-mail contains the name of the test, its key performance indicator (from acceptable to alarm), the test status (pass, fail, retest, or reschedule), and a message description containing the diagnosis and recommended actions. The e-mail also features a section that contains additional information pertaining to the test in question, as well as a link to the corresponding display in the PAM system.

Benefits and lessons learned

Overall, a PAM system:

  • Combines data collection and analysis and integrity checks from multiple sources;
  • Forces pro-active decision-making through decision support e-mails and early detection of impending faults using predictive analysis;
  • Aids in maintenance scheduling based on current and historical analysis results;
  • Provides access to asset information through one easy-to-use corporate interface;
  • Allows viewing of summarized data at the corporate level and the ability to drill down to the condition monitoring points at the analysis level; and
  • Captures the intellectual property of an organization’s most experienced and skilled staff.

There is an adage, “Information is power.” With the introduction of the PAM system, this can be changed to “Shared information is power;” the power to understand the condition of critical assets and to make well-informed decisions that can extend the life of those assets.

The most important aspect of implementing a PAM system at a hydroelectric facility is to accurately define the scope and specifications of the system. With such leading-edge technology, you cannot simply look at what someone else has done. It is necessary to carefully evaluate current maintenance practices and instrumentation available across all sites and aim for a consistent approach. In addition, any existing systems must interface with the new system in order to get the best benefits from your existing systems.

Development of models to assess the health of the various assets (turbines, governors, circuit breakers, etc.) is particularly difficult because it requires combining smart technology with advice from subject matter experts, international standards, and best practices in maintenance.

Finally, not everyone will agree with what you are trying to achieve. Listen to all concerns and address them as they arise. Be transparent. Communicate issues and concerns that have been raised and the answers to them.

Mr. Gregory may be reached at Meridian Energy, 322 Manchester Street, P.O. Box 2454, Canterbury, South Island 8003 New Zealand; (64) 3-3579728; E-mail: [email protected] meridianenergy.co.nz.

Figures 1, 2, and 3 were simulated to demonstrate full predictive asset management system functionality and do not represent the current state of Meridian assets.

Note

1Gregory, Neil, “‘PDA’ over Paper: Improving Data Collection,” HRW, Volume 12, No. 6, December 2004, pages 37-39.

Meridian Energy’s predictive asset management system won the 2007 Engineers Australia Excellence Awards (Newcastle Division). In giving this award, judges cited the high degree of collaboration between Meridian Energy and Matrikon Inc.

Neil Gregory, technology and process strategist for Meridian Energy, is responsible for the evaluation process and subsequent operational use of the predictive asset management system.