Alarms in a diagnostic maintenance program are critical for the efficiency of the process. They show where the user’s attention should be focused.
This is because most machines in a plant do not show any problems. Eventually if the alarms are set correctly the user will deal with assets that have a real problem.
The simplest alarm method is the comparison of the current value with the base value. However, changes in the operation of the equipment (speed, load) or even environmental conditions changes, can affect the size that the sensors we use measure creating the risk of false alarms. These reduce confidence in the diagnostic maintenance system implemented by enhancing the risk of hiding real damage.
The international maintenance community is increasingly recommending the use of statistic methods for monitoring equipment to combat the phenomenon. The statistical analysis explains how much something can deviate from the ‘normal’ situation. This method is based on the notion of standard deviation or σ. The distribution of data is visually described by the following figure:
Measurements up to 1 x σ from the average, constitute the 68% of the measurements while measurements up to 2 x σ constitute the 95% of the measurements. As can be understood, data even further from the center are certainly indications that the machine’s operating condition is outside ‘normal’.
As part of the continuous improvement of our DataRunner software, we have added an assistant guide for the definition of statistical alarms.
The user having collected data with any of the supported methods, transitions to the measurement point. There they will come across the new option: Statistical Alarms
Then they will be asked to choose the measurement on which they will be based on. For oscillation measurements this might be RMS, Peak, Peak-to-Peak, Crest factor acceleration/velocity/offset. Regarding spectral analysis statistical alarms are also applied to arrange region value.
Based on the measurements selected, the guide suggests values for the alarms. At this point the used who is acquainted with the machine’s behavior may want to make changed and eventually finalize the alarms.
For a predictive maintenance program, the time spend on alarms is one of the most cost-effective throughout the program. We ensure that these costs are even lower by optimizing the outcome.
If you are interested in improving the reliability of your equipment click here.