Why condition monitoring might be the right maintenance strategy.
by Marko Preibisch & Christoph Hoeweler
July 19, 2018

The period for observing machine condition must be smaller than the amount of time it takes for faults to develop in machinery.

Although, figuring out the degradation rate may not be as simple, especially on complex machines depending on many different parameters.

Traditionally, the knowledge of the degradation rate on typical wear parts is experienced by plant operator’s maintenance personnel and may even deviate from manufacturer specifications due to installation and ambient conditions.

CBM conceptImage 2. The concept of CBM can be further improved thanks to the IIoT.

Additional Considerations

Regarding high value assets where unplanned shutdown time is costly or where breakage may negatively affect employees or the environment, permanent monitoring systems may be considered.

Below are further advantages of 24/7 monitoring:

  • statistical analyses on measurement signals to optimize process to ensure or sustain high product quality
  • provision to realize optimum spare part management of assets

How to Understand, Analyze & Interpret the Data

Keep in mind that monitoring 24/7 does result in a large volume of data and requires personnel trained in analyzing and interpreting the data. The key challenge resulting from this investigation is when to put gained knowledge to use. Interpretation of signals and the timing of when to schedule repair is difficult and requires experience. Traditional software is able to make only simple decisions like fault localization based on sensor position and peak to peak value evaluation.

Today with new software tools, we are able to do far more than that, including differentiation of faults based on known fault patterns, root cause analysis, residual lifetime prediction of components, driveline optimization and automatic reporting. Therefore, the usage of new software tools can significantly support the decision making process for machine owners.

Skills in instrument expertise, machine knowledge and digital signal processing are necessary in gaining physical understanding of the process.

To comprehend the physics, it may be a requirement to study mathematical models to gather the system behavior and distinguish between correct and faulty operation processes of machinery.

To increase autonomous support in terms of interpretation and timing of repairs, artificial intelligence (AI) may support data analysts and process experts in root cause analysis and decisions. However, AI based on neural networks only can interpolate and not extrapolate on training sets (fault condition), meaning that a fault condition must be known to the AI.

At this point, big data may come into play. Big data gathers machine faults on similar input variables on whole machine fleets generating those training sets. Over time, this establishes a robust assistant to monitor and diagnose machine failures.

Conclusion
Permanent and intermittent monitoring is essential in monitoring the right signals and receiving salient measurement signals for analyses and interpretation. As noted, in order to identify machine failures, it is important to understand the measured signal feature in condition monitoring for diagnosis.

There are many different monitoring strategies to consider, and condition monitoring is one of the most ideal for plant operators since it:

  • protects and optimizes equipment
  • extends operational life
  • avoids unplanned downtime
  • increases maintenance strategy effectiveness while reducing costs
  • increases employee safety.

Pages