## GQZ Technology Knowledge: Experimental Studies on Relationship between Rolling Bearing Life and State Characteristic Parameter

Hot tags: cylindrical roller bearing(9)

**Bearing life test platform and relationship between bearing life characteristics and operation time.**

** **

** **

**Abstract: Using the self-made bearing experiment bench, the tracking experiments in load with five conditions were conducted, such as bearing in good condition, bearing with damage in outer ring and in roller respectively. So a large number of vibration signals were obtained which were analyzed in order to obtain the relationship among amplitude domain, frequency domain and wavelet packets frequency division energy which reflected the development law of bearing life. Also the development law during bearing wear was mastered better. **

** **

**Key words: bearing; vibration character parameter; life; experiment study **

** **

**0 Introduction**

Through the self-made bearing test-bed, the on load operation tracking test under five working conditions, including intact bearing, damaged bearing outer ring and damaged roller, was completed, and a large amount of vibration information was measured. After analysis and processing, the variation relations of parameters such as amplitude range, frequency domain and wavelet packet sub-band energy value reflecting the evolution law of bearing life with operation time were obtained, and the development and change law of bearing wear and damage was better mastered, It provides a basis for the life prediction of the power end of the drilling pump.

**1 Bearing life test platform**

**1.1 Research object and test method**

(1) Determination of research object

Generally speaking, bearing life refers to the fatigue life of bearings. In order to study the service life of the bearing of the drilling pump, on the one hand, the service life experimental data provided by the bearing factory can be obtained through calculation and from the bearing factory. On the other hand, the working life of the bearing of the field drilling pump can be statistically analyzed as a reference frame for life prediction. A bearing test-bed is established in the laboratory to track and monitor the damage state of artificially manufactured bearings with different initial damage states under different loads and different operating speeds, so as to master the damage development law of bearings and the relationship between them and state characteristic parameters, so as to provide a reference basis for judging the damage law of large bearings.

(2) Working principle of test bench

The motor is controlled by the frequency converter to drive the bearing test bench to operate at the required speed. Three accelerometers are installed on the tested bearing pedestal to measure the bearing vibration signal, one of which is horizontally installed and connected with the charge amplifier, and two are vertically installed. There is an adjustable loading device under the bearing pedestal, and the device is equipped with a tension pressure sensor, the spindle of the test-bed is also equipped with a photoelectric sensor (depressurized by a ring rheostat) to measure the time scale signals. These signals are collected at the wiring terminal and sent to the data acquisition card for collection, and then sent to the computer for recording, analysis and processing, so as to obtain the vibration, force, fault and other state information of the bearing. After analysis and processing, the bearing fault diagnosis conclusion can be obtained.

**1.2 Test principle**

Combined with the actual situation of bearing life test, track the changes and trends of vibration parameters in various domains of the bearing at different damage stages, and find the most sensitive parameters to the development of bearing damage, which can be used as an important index for the life prediction of the bearing at the power end of the drilling pump. Therefore, the constant stress accelerated life test method is selected.

Test steps:

(1) For the selection of acceleration stress, the method of increasing the bearing load is selected here. The selected bearing is ZWZ22206, which is a double row cylindrical roller bearing with inner ring without flange. Now the loading load is 24kN, and the speed of the test bench is 581r/min;

(2) Determine the accelerated stress level. In the constant stress accelerated life test, the lowest stress level should be appropriately higher than the normal working stress level, while the highest stress level should try to choose the stress level that can achieve the maximum acceleration effect without changing its failure mechanism. Considering that this test is mainly to find the parameters of each domain that can directly reflect the bearing damage, the accelerated stress level is 24kN (pure radial load, no axial load, at this time, the calculated rated life is 63.77h).

(3) Selection and grouping of test samples

Take two intact bearings, one bearing with initial 1-point and 2-point faults in the outer ring, one bearing with initial 1-point and 2-point faults in the roller, and one bearing with initial 1-point and 2-point faults in the inner ring as test samples for experimental research. Install these bearings on the test bench and run until they are damaged under a certain speed and load. Study and analyze the changes of their vibration state information and the law of bearing damage.

In terms of the time from the operation to damage of the test bearing, the bearing has been operated for less than 20h from the beginning of operation to complete damage, whether it is a good bearing or a bearing with local failure, which is lower than its rated life. On the one hand, the determination of the rated life of the bearing has its probability factor, and is related to the lubrication conditions and temperature of the bearing. During the test, in order to accelerate its damage, it was not well lubricated and cooled. As for the three kinds of bearings that had faults before, because they had faults before the test run and lacked the conditions to compare with the rated life, it was normal for them to be damaged in a short time.

**2 Relationship between bearing life characteristics and operation time**

The vibration signals measured by vibration accelerometers are analyzed in amplitude domain, frequency domain and wavelet packet respectively, and the relevant indicators of different operation times are classified and compared, from which the parameter indicators most related to the bearing life are found, and the relationship between the operation time and the life is calculated as the basis for the prediction of the remaining working life of the bearing.

**2.1 Frequency domain parameter index**

The power spectrum of signal reflects the distribution of signal energy with frequency. When the energy of each frequency component in the signal changes, the energy ratio of the power spectrum component also changes. On the other hand, when the frequency component of the signal increases, the distribution of energy in the power spectrum will be dispersed. When the frequency component of the signal is small, the energy distribution in the power spectrum will be concentrated. It can be seen that by describing the change of the position of the main frequency band in the power spectrum and the dispersion of the energy distribution on the spectrum, the change of the frequency domain characteristics of the signal can be better described.

Frequency domain parameter indexes barycentric frequencyFC, mean square frequency MSF and root mean square frequency RMSF all describe the change of the position of the main frequency band of the power spectrum, while frequency variance VF and frequency standard deviation RVF describe the dispersion of spectral energy. The frequency domain parameter index can be used to roughly judge the fault of rolling bearing. When there is no fault in the bearing, the frequency component is mainly in low frequency and FC is small; In case of local damage fault, due to the resonance caused by impact, the main frequency region shifts to the right and FC increases.

**2.2 Amplitude range index**

The amplitude range analysis method is to extract the statistics reflecting the characteristics of the signal through the calculation and analysis of the time-domain signal, and then identify the state and fault on this basis. These statistics mainly include kurtosis and margin indicators, which can reflect the impact intensity.

For the research, the more valuable parameters are effective value, waveform index, peak index, pulse index, margin index and kurtosis index. The effective value indicates the magnitude of the signal vibration, and the last five dimensionless parameters can describe the magnitude of the impact component in the signal. Most of these indexes have been applied in the research of condition monitoring and fault diagnosis of the bearing at the power end of the drilling pump.

**2.3 Relationship between amplitude range, frequency domain parameter characteristics and operation time**

The relationship between the typical amplitude range and frequency domain indexes of a perfect bearing and the running time of the bearing from the beginning of operation to the failure of operation can be made by graphics. Figure 1 shows the relationship between the standard deviation of the bearing frequency and the running time when the bearing is in good condition, figure 2 shows the relationship between the bearing deflection coefficient and the running time when the bearing is in good condition, and the relationship between other parameter indexes and the running time can also be similarly calculated.

Fig. 1 Relationship between standard deviation of bearing frequency in good condition and operation time

Fig. 2 Relationship between deflection coefficient of bearing in good condition and operation time

By carefully comparing the changes of graphic curves of various parameters, their differences can be identified. From the above indicators, at this time, each parameter - operation time curve has an inflection point after running for a period of time, and then the curve has a strong upward (or downward) trend, indicating that the bearing has failed. When each parameter falls sharply, it indicates that the bearing has been damaged. Among these parameters, K factor, waveform index, domain coefficient, pulse index, skewness and three frequency domain parameters have obvious changes, so the changes of these parameters can be used as important reference indexes in fault diagnosis and bearing life prediction.

Through the experimental study of the relationship between the parameters of each domain of the bearing in different initial states and the running time, the following laws can be found: the parameters of the intact bearing are basically stable and change little before the fault occurs; The K factor value of the roller with a little fault at the beginning is very high; The parameters in the amplitude domain of the roller with two initial faults fluctuated at the beginning of operation, presenting an unstable state; When there is an initial fault in the outer ring, the parameters in the amplitude domain fluctuate in an upward trend, and the initial values of the three frequency domain parameters are significantly lower than the initial values of the intact bearing; When there are two initial faults in the outer ring, the skewness coefficient and K-factor value are significantly higher than the initial value of the intact bearing, and the other parameters are significantly lower than the initial value of the intact bearing, and the skewness coefficient operation time/h frequency standard deviation l/s in addition to the skewness coefficient and K-factor value rising, the other parameters will drop sharply; From the initial failure to damage of the inner ring, except for the skewness coefficient and the convex K-factor curve, all other parameters are concave. The above conclusions can provide reference for fault diagnosis and life prediction of bearings.

**2.4 Relationship between wavelet packet frequency division energy value and operation time**

On the bearing test-bed, the same kind of bearings with different initial states are tested for on load operation life. At the same time, the vibration signals are collected, and then the wavelet packet analysis is carried out. A series of wavelet packet frequency division band energy diagrams of bearings with different initial states are obtained. See Figure 3 for the change of wavelet packet frequency division band energy spectrum from an initially intact bearing to damage. At the beginning, except for the high energy value in the first frequency band, the energy value in other frequency bands is very low. As previously analyzed, the bearing starts to damage when it runs for 6h, so the energy value in the second, third, fourth and fifth frequency bands begins to increase continuously. When the bearing is completely damaged and cannot work (running for 9h), the ratio of the energy value in the second frequency band to the energy value in the first frequency band has exceeded 50%, The ratio of the energy value in the third frequency band to the energy value in the first frequency band has also been close to 1/3, which is enough to explain that the distribution of the energy value in the wavelet packet frequency division band is closely related to the state of the bearing.

(a) Wavelet packet frequency division energy spectrum of intact bearing

(b) Wavelet packet frequency division energy spectrum of damaged bearing

Fig. 3 wavelet packet frequency division energy value when an initially intact bearing runs to damage

Similarly, the frequency division band energy diagram of rolling bearings with different initial faults is investigated. It is found that the wavelet packet frequency division band energy value of the faultless bearing is concentrated in the low frequency domain (400Hz in the first frequency band), and the energy value in the rest of the frequency division band is very small, while the situation of the faulty bearing is different. For example, when there are two faults in the outer ring, the frequency division band energy value of the rolling bearing in the second and third frequency bands is already very high, and the situation of the rest of the faults is also roughly similar. Therefore, the distribution of wavelet packet frequency division band energy value can be used as an important basis for rolling bearing fault diagnosis.

From the test results, we can know that the characteristic parameters such as amplitude range K factor, waveform index, domain coefficient, pulse index, skewness, frequency domain center of gravity frequency, root mean square frequency, frequency standard deviation and wavelet packet sub-band energy value are closely related to the damage and life of the bearing. From their changes, we can understand the deterioration of the bearing and take them as important indicators for life prediction.

**3 Conclusion**

Through the operation test of several bearings in different initial states on the test bench, a comprehensive experimental research and theoretical analysis are carried out on the relationship between the operating state characteristic parameters of the bearings and the operating time, so as to obtain the general relationship between the amplitude range, frequency domain parameters, wavelet packet sub-band energy and the operating time, as well as the variation law of bearing deterioration:

(1) Variation of parameters in amplitude range and frequency domain

The parameters of the intact bearing are basically stable before the fault occurs, with little change; The parameters of the fault bearing in the amplitude range and frequency domain have large or small changes, which are closely related to the damage and life of the bearing. From these changes, we can understand the deterioration of the bearing, and take them as an important indicator of life prediction.

(2) Variation of energy value in wavelet packet frequency division

The energy value of wavelet packet frequency division band of the fault free bearing is concentrated around the first-order natural vibration frequency of the bearing (400Hz in the first frequency band in this example), and the energy value in the rest of the frequency division band is very small. Because the fault can excite the high-frequency vibration of the system, the energy value of the frequency division band (especially in the high-frequency domain) of the fault bearing in its remaining frequency band is very high. Therefore, the distribution of wavelet packet frequency division band energy value can be used as an important basis for rolling bearing fault diagnosis.

More about GQZ Cylindrical Roller Bearing：

GQZ precious-quality cylindrical roller bearing with reasonable price for customers all over the world and first-class service and unrivaled products.

Cylindrical roller bearings (CRBs) have a simple structure with their cylindrical rollers in linear contact with the raceways. They offer high load capacity under primarily radial loads. Low friction between the rollers and ring ribs makes these bearings suited for high speed rotation.