Using Predictive Analytics to Minimize Risk Associated with Aging Assets

By Mike Reed, InStep Software

It’s no secret that an aging asset infrastructure is of major concern to power generation companies. In fact, 74 percent of all coal-fired capacity in the United States was 30 years old or older by the end of 2012, according to the U.S. Energy Information Association. That infrastructure is further stressed by the growing populations and urbanization trends that demand increased generation capacity. Additionally, most utilities face pressure to keep electricity costs low while delivering reliable power, which can lead to challenging budget constraints. As a result, operators, engineers, and plant managers continually strive to make every plant’s operation and maintenance dollar stretch as far as possible.

While running assets for as long as possible can be cost effective and efficient, the practice can have quite the opposite outcome without proper preparations. Aging equipment can contribute to outages, failures, downtime, higher costs, decreased efficiency, and a number of other associated problems. Aging assets can also cause regulatory, environmental compliance, and safety issues. To ensure that these assets continue to provide value to the plant and organization, engineers and operators are tasked with determining the criticality of specific assets and assessing their risk. Such an assessment examines the probability of failure and the consequences of that failure. A number of factors are considered as well, including added costs for items like unplanned maintenance, impact on resources, environmental effects, safety issues, and public reputation, among others.

figure 1

In an ideal world, budgets would be unlimited and utilities could always replace equipment long before any signs of declining asset performance became apparent. While that is unfortunately not the case, there are a number of steps that can be taken to mitigate the risks and sometimes-costly consequences associated with aging assets. Equipment reliability processes and systems can prevent failures of both new and aging assets. That being said, ensuring equipment reliability is key when operating these aging assets.

Maintenance for Older Assets

Effective maintenance is critical to ensuring that assets, plants, and entire fleets continue to operate reliably for long periods of time. Because older assets have a higher propensity for failure, they usually require more frequent maintenance. Engineers and operators employ a combination of maintenance techniques depending on the criticality of each asset. Organizations that do not have a comprehensive maintenance strategy in place are putting the operation at risk.

On one extreme, if a potential asset failure has little to no immediate effect on the safe and reliable generation of electricity, an engineer may choose to run that asset to failure and then simply replace it. However, when used to manage assets that significantly impact generation, such a run-to-failure strategy can prove inefficient and possibly dangerous.

Typically, a preventative or condition-based approach will be used to ensure that an asset does not reach a point of failure. These techniques rely on calendars, intervals, and condition monitoring of the asset to determine when maintenance is performed. However, if a potential asset failure could result in significant damage, safety issues, or power outages, the risk is obviously much higher, and a more proactive maintenance approach is required.

One such proactive approach is Predictive Maintenance (PdM). PdM involves continuously monitoring the health of equipment and comparing its state to a model that defines normal operation. This is done to detect subtle early warning signs of potential failure in equipment. The practice of collecting, storing and trending data on an asset’s performance and health is ideal for determining if that older asset is heading toward the end of its life, or if it is likely to continue running as needed for months or years. This information allows for smarter maintenance decisions and avoids the sometimes arbitrary shortening or lengthening of maintenance intervals based on little performance evidence.

Predictive Analytics

PdM strategies are most beneficial with the implementation of proper online condition monitoring and analytics software. Typically, predictive analytics software analyzes information from an enterprise historian, ensuring that all historical and real-time data is included in the analysis and model building.

It is not possible to manually analyze the real-time data streaming from thousands of sensors which are simultaneously transmitting equipment health and performance-related information. There is simply no way to derive real-time insights under the crush of so much information. As a result, data can quickly become contaminated through human error, and equipment issues can arise before an engineer has time to even analyze the data. Predictive analytics software solutions are much more efficient when it comes to detecting potential failures or possibly disastrous problems with critical assets.

When used for asset performance management, predictive analytics software continuously monitors equipment through sensor data and uses various prediction engines and algorithms to provide advanced warnings for equipment problems and failures. One predictive analytic technique frequently used is Advanced Pattern-Recognition (APR). APR derives predictions from empirical models generated by “learning” from an asset’s unique operating history during all ambient and process conditions. The model effectively becomes the baseline to determine the normal operational profile for a piece of equipment or system. Depending on the software selected, the modeling process can require varying skill levels and time commitments.

The technology can compare an asset’s unique operational profile with real-time operating data to detect subtle changes in system behavior, which are often the early warning signs of impending equipment failure or performance problems. Engineers and operators are alerted well before the abnormal conditions reach standard alarm levels, creating more time for analysis and planning any corrective action. Thus, engineers and operators are able to better prioritize maintenance needs and reduce costs due to better planning efforts.

After an issue has been identified, predictive analytics solutions can provide root-cause analysis and fault diagnostics to help plant engineers understand why an issue occurred. That information can then be used to deter similar issues in the future. In addition, diagnostic technology reduces the likelihood that an engineer will attribute abnormal operating conditions to the wrong variable.

Organizations that implement predictive analytics software to monitor critical assets can feel comfortable knowing that they will receive early warning notifications of incipient issues. Rather than spending time searching for issues, or worse, waiting for an important piece of equipment to fail, they will be notified via an alert that an asset is not operating as expected. Depending on the software selected, alerts can be customizable and emailed immediately to the appropriate personnel. The insights from a predictive analytics solution will help engineers and plant operators better determine when an aging asset can continue running as is, should be serviced, or needs to be replaced.

figure 2


A predictive analytics solution used to monitor and improve the reliability of aging assets will pay for itself in a short time. Consider the probability that an older piece of equipment is likely to fail and think about the financial impact it would have on operations and resources.

For example, a hypothetical situation might look like this: an aging steam turbine has been operating normally until site personnel are alerted of a vibration step change via an email notification sent from the plant’s predictive analytics solution. The appropriate personnel are able to verify that a proximity probe and casing vibration have both changed. Further analysis indicates a likely loss of mass in the turbine blade path. Based on the unit’s history, site personnel immediately suspect shroud material has been lost. It is determined that the unit can continue to run at a reduced output, under increased observation, until a more convenient and strategic time is found to take the unit offline. Once the unit is brought offline, a borescope inspection verifies missing shroud material and several other segments that are close to liberating.

Had this issue not been identified through APR vibration modeling, it could have caused immediate unplanned downtime, loss of generation, possible catastrophic failure, and danger to personnel. The vibration step change was not significant enough to alert the operations staff of this impending condition via normal monitoring practices; it was the predictive analytics software and PdM protocols in place that brought about this positive outcome, resulting in a potential savings of millions of dollars in lost revenue and increased repair costs, in addition to the safety of the operating engineers.

The inclusion of a predictive maintenance plan in a comprehensive strategy creates benefits that are both immediate and long-term. Predictive maintenance allows plants to keep running aging assets even longer, with the ability to proactively manage and act on potential problems. Plants can effectively move from unexpected and immediate maintenance to planned, strategic maintenance.

When applying predictive maintenance strategies, power generation organizations are able to make smarter decisions about when and where maintenance should be performed. These decisions are based on the criticality of the asset, the asset’s performance history, and the goals of the plant managers. Predictive analytics solutions allow decision makers to extend maintenance windows by delaying maintenance that may not be immediately necessary. Rather than completing maintenance exactly as suggested by the original equipment manufacturer, the maintenance can be performed during a more convenient and cost-effective time. Additionally, maintenance duration can be reduced because plants receive early warning of an issue, thus allowing efficient use of resources. Both of these benefits reduce overall maintenance expenditures while continuing to esnure that older assets are performing well.

Predictive analytics software can also identify underperforming assets, which can be a problem with older equipment. This knowledge can help personnel understand what factors are contributing to that underperformance. In the same manner, predictive analytics technology can prevent equipment failures by providing early warning of subtle changes that may have otherwise gone unnoticed. The technology can sometimes identify problems months before they happen, allowing plants to be more proactive. Not only do organizations benefit by further extending the life of their equipment, thereby lengthening maintenance windows, increasing asset efficiency, and increasing availability; other savings are also realized when potential costs are averted, including downtime, replacement equipment, lost productivity, and additional man hours.

Moving Forward

The entire power industry is continuing to advance with new technologies, becoming smarter every day. The transition from traditional maintenance approaches to a comprehensive strategy involving predictive techniques is allowing organizations to safely run their equipment for as long as possible. As the power generation infrastructure in the United States continues to age, it’s more important than ever to understand how and why an asset is performing the way it is in order to avoid costly failures.

Plant engineers, operators, and managers can utilize predictive analytics solutions to help them work more effectively and efficiently by equipping them with the information needed to make proactive and well-informed decisions. Aging assets will continue to present challenges and raise warranted concerns for utilities.

This issue is not unique to power generation; rather it runs the gamut from generation to transmission and distribution, from coal- and gas-fired plants to nuclear facilities. However, the amount of data available to engineers and plant personnel also continues to grow, creating opportunities to further improve plant reliability and efficiency. Through predictive analytics solutions, this information is now being used to monitor the health and performance of equipment and can be used to circumvent the failure of older assets.


Mike Reed is the Manager of Analytical Services at InStep Software, a leading global provider of eDNA real-time performance management and PRiSM predictive asset analytics software.

More Power Engineering Issue Articles
Power Engineerng Issue Archives
View Power Generation Articles on

No posts to display