How can possible failures in rolling cages be identified?

How can possible failures in rolling cages be identified?

How can possible failures in rolling cages be identified? To answer this question, it is necessary to analyze the context of the natural wear of bearings.

In industries in different productive sectors, rotating machines use bearings to support the load and maintain the clearances between stationary and rotating elements.. in this universe of machinery more than 90% have bearings. Unfortunately, rolling element bearings are prone to a multitude of premature failures.

It is observed in technical literature that only 10% of bearings reach the limit of their design service life (L10), that is, the expected service life of 90% of the bearings of a certain type under similar operating conditions. In general, initial failures are attributed to lubrication, misalignment, overload, design and application errors, and even pre-existing problems that were not detected during manufacture.

A comprehensive maintenance program based on condition analysis, which incorporates predictive maintenance, must be prepared to detect the onset of wear and deterioration of the bearings throughout their life cycle. A mature maintenance program provides not only indications of wear on the bearings, but also an assessment of the severity and recommendations for corrective actions for consequent maintenance.

The most common rolling cage failures are generally related to its main elements such as inner race, outer race, cage and rolling element. These bearing elements are illustrated in Figure 01. Once the bearing is mounted in its housing, a new dynamic system is obtained, and due to the high rigidity, it is expected that this system will always have high-frequency resonance components (> 2 kHz). In the bearing resonance, it is then possible to show the frequency of the bearing defect: internal raceway (BPFI) and external raceway (BPFO) passing frequency, rolling element turning frequency (BSF) and cage frequency (FTF).

 elements of a bearing on a transparent background

Figure 01: Bearing elements

One of the failure modes in a bearing is the breakage and/or wear of the cage. Although less common, these failure modes have a high severity, because when they occur they cause an immediate stop of the bearing as it compromises the performance of its function.

The way to capture this failure mode by vibration analysis requires some important observations and precautions. To begin, it is necessary to understand how the phenomenon of bearing cage failure occurs. The failure occurs for reasons such as:

  • Wear of the cage where the rolling elements are mounted;
  • Some of the cage elements becoming detached;
  • Friction of the cage on the other parts of the bearing, or even by direct rupture of the cage.

Shortly after some of these failure modes occur, the bearing collapse does not take long to happen. With that, the time to capture the fault becomes short.

The failure mode of the cage, with this particularity, becomes more challenging for the vibration monitoring system, as well as for any experienced vibration analyst. This is because the failure can manifest itself in the natural frequency of the bearing (if it has impacts). Another way to capture this failure is to do the demodulation using the Envelope technique using a filter in the high-frequency region, taking advantage of the natural frequency of the bearing to amplify the periodic frequencies.

Due to the low frequency of the cage, that is, approximately between 1/4 and 2/5 of its rotation, special care must be taken when parameterizing the data collection, where it is always considered that the duration of the collection is at least a period of two to four turns of the cage on its axis.

Another important aspect happens during the measurement, when it must always be ensured that the vibration transmitted by the structure is well received at the sensor monitoring site, at the same time that the sensor is rigidly and properly fixed at the monitoring point. For this reason, the vibration sensor should be as close as possible to the load area of ​​the bearing under investigation. The measurement direction of the sensor should be approximately perpendicular to the load zone, as shown in Figure 02.

Transmissão de vibração em um rolamento

Figure 02: Vibration transmission transmitted by the structure from its origin (cage defect) to the measurement position

It is important to note that it is not advisable to monitor the defect of the cage in the acceleration/speed spectrum at low frequency because the defect has no energy/power in amplitude to exceed the values ​​of conventional frequencies, such as the rotation of the axis and the vibrations of the machine itself, which manifest themselves in low frequency.

Considering the points presented above about the monitoring and detection of the defect in the rolling cage, Dynamox technology shows the ability to demonstrate them with reliability. The Dynalogger HF (High Frequency) sensor allows monitoring at high frequency up to 6.4 kHz, the region close to where the bearing’s response occurs at its natural frequency, so if there is any impact of the cage on the other parts of the bearing it will be visible in the spectrum.

Another important point to consider is the possibility of using the envelope in more than 5 frequency filters. It is also important that using the number of lines and maximum frequency, we have the desired period to parameterize the number of cycles necessary for a reliable collection. An example of a bearing envelope spectrum is shown in Figure 03. In addition, this technology makes it possible to make collections from 1 to 60 minutes, which favors and increases the probability of identifying the fault, since we will have faster information due to the frequency of samples with trends, helping significantly in identifying the failure.

Figure 04: Example of using kurtosis and crest factor acceleration with and without bearing defect

In addition, with a frequency bank (BPFO, BPFI, BSF and FTF) of almost 70 thousand bearings of the main brands and models, the DynaPredic Platform (part of DynaPredict Solution) is ready to add automatic and manual markings (See Figure 04) in the spectrum to facilitate data interpretation and expedite information collection.

Therefore, the obstacles to assertive diagnoses of this type of detection are common to all vibration platforms, thus, the differential is to meet the listed criteria and be able to count on the diversity of solutions presented by Dynamox.