Enhancing the Potential of Condition Monitoring Using Vibration Information

When fully utilized, condition monitoring can transform how power plants operate and maintain their assets. It can lead to increased efficiency, reduced maintenance costs, improved safety, and better overall equipment performance. In essence, this article argues that to exploit the full potential of condition monitoring, it’s necessary to move beyond the simple collection of data. Today, it’s about using that data to acquire actionable insights, optimize maintenance strategies, and make data-driven decisions that improve power generation business outcomes.

Rotational machines are the workhorses of power generation: They transform various energy sources into the rotational motion needed to drive electricity production. Therefore, it’s essential that these machines operate efficiently and reliably for power plants to function effectively.

The energy source used depends on the type of power plant (coal, nuclear, solar, hydro, and others). Rotating components like bearings, shafts, and turbine blades are constantly subjected to friction and stress during operation. This can lead to gradual wear over time, reducing efficiency and increasing the risk of failure. That’s why monitoring the status of these assets and implementing the right maintenance strategies to prevent breakdowns and failures is fundamental—and it’s not a new issue.

As noted by Oil & Gas iQ, the world’s largest community of oil and gas professionals, production capacity in power generation is reduced by up to 5% by unplanned shutdowns every year. In addition, the repair costs for broken machines have increased by 50%.

From Reactive and Run-to-Failure Maintenance to Data Condition-Based Maintenance

Rotating machinery maintenance strategies have evolved over time and are getting closer and closer to meeting power plant goals. Condition monitoring is no longer considered to be a luxury or a sophisticated and difficult-to-access tool: Instead of waiting for breakdowns to occur, it can now be used as part of a proactive maintenance strategy that identifies subtle patterns and trends by analyzing large volumes of sensor data.

To understand how power plants are now able to schedule maintenance exactly when it’s needed to avoid costly downtime and prolong asset life, it’s essential to look back at the different maintenance approaches used in the past and how they’ve evolved.

Run-to-Failure. Run-to-failure maintenance is a conservative approach in which machines operate until they break down; this approach provides the longest time between shutdowns. However, a breakdown can be disastrous and cause serious damage, resulting in increased costs for production losses compared to the cost of the machines.

Preventive. Time-based maintenance is performed at regular intervals, usually shorter than the pre-planned “time between failures” interval. This method allows for better planning and reduces catastrophic failures. Time-based preventive maintenance is indicated when the time to failure can be accurately predicted, because some assets may tend to wear or fatigue at a reasonably predictable rate. However, in other components, like rolling element bearings, there’s a large statistical dispersion around the mean, where the mean time to failure is two to three times the minimum.

Condition-Based. Data-driven condition-based maintenance has been recognized in recent years as the most efficient maintenance strategy. This strategy allows customers to anticipate potential failures and perform maintenance at the optimal time. It requires reliable condition monitoring techniques that note the current condition and make reasonable predictions of the remaining operating life. This strategy has been used successfully for the past 35 to 45 years with great success in industries like energy and petrochemical. In these industries, machinery tends to operate at a near-constant speed and with a stable load, so the technical problems associated with condition monitoring have been greatly reduced.

Most Common Condition Monitoring Techniques

Condition monitoring involves continuous inspection and prediction of the current and future status of machines while they’re in operation, which requires measuring and analyzing the condition-related characteristics of an asset and understanding their influence. Several diagnostic methods have been applied to these maintenance strategies over the years. Most of them include oil analysis, vibration signal analysis, particle analysis, corrosion monitoring, acoustic signal analysis, and wear debris analysis.

Among them, vibration and lubricant analysis are two techniques commonly used to determine the internal condition of a machine. Vibration analysis involves analyzing the machine’s vibrational signature, which changes as failure approaches; lubricant analysis focuses on wear particles and chemical contaminants in circulating-oil lubrication systems. For example, many bearings contain metals with the same chemical composition, but only the defective one will show increased vibration.

Therefore, vibration analysis is especially advantageous for rotational machinery compared to other methods. It reacts immediately to changes and can be used for both continuous and intermittent data acquisition. Compared to oil analysis, vibration analysis is more likely to indicate the actual defective component.

Vibration-Based Diagnostic Monitoring of Rotational Machines

In vibration analysis, powerful signal processing techniques can be applied to extract very weak fault indications from noise and other masking signals. Signal processing is the core of vibration-based diagnostic monitoring of rotating machines; it takes the raw electrical signals from vibration sensors and transforms them into meaningful information that’s used for fault diagnosis.

Vibration signals and process parameters are continuously digitalized and then processed. From the vibration signals, important characteristic values like amplitudes and phases of harmonic vibration components are calculated. These values—along with other process parameters—are compiled into a time-stamped record called the condition vector. The diagnostic monitoring of the rotational assets is realized by continuously comparing the current characteristics to reference values derived earlier (Figure 1).

1. Diagram of vibration monitoring and signal processing on a turbogenerator. Siemens Energy

Many rotating machine condition indicators can be derived from measuring vibration. For large machines like turbines, generators, pumps, and fans, the most meaningful indicators are the amplitude and phase angle of specific vibration frequencies. These well-known frequencies include the first and second harmonic frequency, which represent the movement of the machine’s shaft and bearings at turning speed and at twice that frequency. To obtain the movement’s correct phase angle information for a position on the shaft or a specific bearing, it’s essential to measure vibration in two perpendicular directions and to synchronize the measurements with the rotor position.

Calculating meaningful indicators from huge amounts of digitalized vibration signals also helps to compress the data so it can be stored for future analysis. For comprehensive machine diagnosis, it’s important to keep track of the slightest changes in the machine’s motion along with corresponding process data that might reveal the cause of the observed change.

Vibration Analysis Tools: The Diagnostician’s Strongest Partner

Using a powerful storage concept, condition monitoring software needs to offer the right set of tools to the analyst that enable comprehensive machine diagnosis and informed operation and maintenance decisions. The basic online data-processing functions include the following:

■ Continuous measurement of vibration in appropriate time and value resolution.

■ Derivation of characteristic indicators like amplitudes and phase angles at meaningful frequencies (for example, first and second harmonic frequencies, etc.) from perpendicular vibration measurements.

■ Compilation of these parameters along with related process signals into coherent data records for both online monitoring and for storage and later analysis.

■ Evaluation of the current indicators by comparing them to reference limits.

■ Message communication to operators and maintenance engineers whenever the machine appears to deviate slightly from its normal condition.

Statistical functions of condition monitoring software also account for a variety of different operating conditions that in turn are indicated to the system and serve as leading variables. This feature provides very tight and accurate reference boundaries for each indicator to identify the slightest deviations from normal conditions. This allows operators to detect developing machine problems at the earliest possible point in time.

Who Benefits from Vibration Monitoring?

Once the issue has been reported, the severity of the problem can be actively monitored, and the operator can accommodate the problem: for example, by avoiding critical operating modes. It also enables maintenance engineers to analyze the condition of the machine and generate valuable information for decision-making and activities like overhaul preparation. To achieve this, it’s essential to have a comprehensive set of analysis tools provided by condition monitoring software.

Siemens Energy’s condition monitoring solutions have been applied in more than 1,000 projects around the world to meet a variety of customer needs. Typically, power plants implement a new condition monitoring system when replacing old rotating equipment or updating their existing diagnostic systems.

One of these successful projects was executed at an energy supplier based in Germany. The company decided to completely modernize its machinery diagnostic systems in 2021 for nine turbogenerator sets that are distributed over four sites and connected in a network.

The existing diagnostic systems offered a high level of safety and reliability thanks to a comprehensive redundancy concept. However, the utility wanted to modernize its diagnostic system and take advantage of advanced functions; they also wanted to transfer the stored data to the new system.

All protection and vibration analysis systems were replaced and synchronized, and the independent systems in each unit were centralized in one diagnostic software installation (a multi-machine system). The entire integrated vibration diagnosis system now provides in-depth knowledge about the operating behavior of the machines and has been offering valuable information for decision-making in maintenance planning. The option to network machines and plants, as well as the convenient integration into the plant control system, facilitates the use and exchange of machine status information.

The Future of Vibration Monitoring

Machine learning algorithms introduced in the visualization functionalities have improved the quality of data obtained from condition monitoring software; these are state-of-the-art techniques that have already been discussed in maintenance strategies. Archived historical data along with real-time status data allows events to be connected to reveal emerging patterns. The rapid identification of factors that are combined to create a view can be automatically linked to actions—from email notifications to operational management and triggering additional actions to gathering more information and developing a more granular view.

Normal operating behavior, when observed over an extended period, provides a baseline for analysis enabled by artificial intelligence. These tools permit vast amounts of data collected by the monitoring unit in asset sensors to be rapidly turned into actionable insights. This is what enables earlier intervention and more effective predictive maintenance regimes.

The condition of critical operating assets is even more important today than in the past: The nature of the market means that unplanned downtime is unacceptable, and there are no excuses for failing to meet availability contracts. Power generation businesses need to be completely in control of their assets, including having minute-by-minute insights into their operating condition. This is the only way to ensure a competitive performance in energy markets worldwide. A reliable equipment condition monitoring and diagnostic solution is therefore a crucial aspect of any company’s strategy.

Siemens Energy’s vibration-based condition monitoring solution consists of hardware components and software tools for data analysis. Its research and development (R&D) team is aware of emerging technology trends and the constant need to offer digital products that meet customer requirements.

—Torsten Wetzel and Daniela Quiroga Hernández are members of Siemens Energy’s vibration-based condition monitoring portfolio product management team.