Surface Defects Vibration Measurements of Automotive Tapered Roller Bearings

The paper investigates vibration testing methodologies for automotive tapered roller bearings. The European Union regulations have indirectly led to improvements in the NVH performance of the bearings that affect both the industrial and automotive markets. The goal of this study was to examine the influence of localized small defects on the vibrations’ magnitudes within tapered roller bearings. A Rockwell indenter was used to create three different surface defects. These three levels, representing different indentation pressures, mimic defects that could occur during manufacturing or bearing installation. Indents were placed on inner raceways, the outer raceways, and the roller bodies. After the assembly of the components, different vibration testing methodologies were applied to quantify those types of defects. This paper presents a new envelope analysis approach to monitoring for defects. The paper presents the current Fast Fourier Transform (FFT) testing approach, whereby the machine sensitivity is representative of a typical quality control bearing thresholds. This paper also presents a new, more robust method and analyzed for its suitability in the detection of localized defects.


Introduction
Rolling element bearings play a critical role in the functionality of machinery.Almost all rotating applications contain bearings, and rolling element bearings are thus found in almost all industries.Like all componentry where friction is present, noise is generated.This noise can be exacerbated by geometric imperfections, such as those that can be generated due to manufacturing issues, part assembly, or in application bearing operation.
A new fault detection methodology lends itself to improved noise and vibrating monitoring of the imperfections.To reduce the number of bearings with imperfections, it's necessary to understand the type of vibration signals produced during the bearing's lifetime.Thus, engineers can explore new approaches that involve optimizing the bearing's geometric design, material selection, lubrication, and manufacturing process to minimize unwanted noise and vibrations.
Nowadays, requirements and standards regarding noise and vibration are more strictly applied in industrial settings.To address those requirements in the manufacturing process, vibration measurement and data interpretation were introduced.Condition monitoring serves as a detector of early stages of bearing damage, enabling forecasted repairs to be scheduled by the maintenance team.This approach helps to reduce downtime and optimize maintenance practices.Also, for quality control, at the end of each assembly line, vibration measurement equipment is installed to detect manufacturing imperfections before the bearing is sent to the customer.1303 (2024) 012022 IOP Publishing doi:10.1088/1757-899X/1303/1/012022 2 High volume production, in any size range, necessitated the development of new detection methods.The envelope detection method is a qualitative technique used by many industries, including the aerospace [1].The technique of envelope analysis is detailed by S. Kim [2], with the aim of analyzing roller bearing elements in mechanical transmissions.
Konstantin-Hansen and Herlufsen [3] describe envelope analysis as the Fast Fourier Transform of a modulated signal.This methodology can be used for investigation of machinery faults possessing a modulating effect with frequency characteristics.It can be used on gearboxes, motors, turbines, etc., and is also a great tool to find local defects such as cracks and spalls on bearing functional surfaces.
Relative motion of the rollers over a defect generates impacts that excite bearing periodic frequencies.Each bearing component, when in relative motion, generates a different periodic defect frequency.Those frequencies are calculated by using geometric characteristics and the rotating speed in the application.The ball pass frequency outer race (BPFO), ball pass frequency inner race (BPFI), ball spin frequency (BSF), and fundamental train frequency (FTF) can be calculated using modules available online or from the bearing producers.
Desheng Li and Young Sup Kang [4] published a simulation model for tapered roller bearing vibration introduced by a geometrical imperfection on the outer ring of the assembly.The model was correlated with measurements.Doguer and Strackeljan [5] presents a focus on bearing fault simulation to improve the understanding of fault-induced vibration and early-state fault detection.The authors propose a multibody system model of a rolling element bearing and introduce fault elements to simulate different fault geometries.
Patel, Pandey, and Tandon [6] developed a model to analyze the vibrations of deep groove ball bearings.The model takes into consideration multiple defects on the surfaces of the inner and outer races.Results from the model are compared to experimental measurements obtained from both undamaged and defective deep groove ball bearings.The Runge-Kutta method is used to solve the motion equations.This method calculates the vibration levels both in time and frequency domains for the shaft, balls, and housing.R.M. Stoica [7] created a simple mechanical transmission integrating an electric motor, a rigid coupling, a driven shaft, two housing bearings, and one monoaxial accelerometer that was used to collect data in three directions in the proximity of the bearing.
Adamczak, Domagalski, Wrzochal, Piotrowicz, and Wnuk [8] describe a new industrial test rig designed specifically to measure the frictional moment of rolling bearings during the production stage.The frictional moment is a critical parameter that influences the quality and suitability of bearings for various applications including the automotive industry, machine industry, and household appliances.By measuring the frictional moment, information can be obtained about the bearing's structure, the quality of elements, its cleanliness, and the specific properties of the lubricant used.Reducing friction in rolling bearings is crucial to increase bearing efficiency, reduce energy consumption, and protect the environment.
The envelope analysis was determined in [9] using the kurtogram of bearing fault detection.With the help of a kurtogram, the frequency band of the raw vibration signal with the highest impulsivity and its corresponding central frequency and bandwidth can be determined.
Liu J, Wang L, and Shi Z [10] focused on studying local defects in cylindrical roller bearings (CRB), specifically on the contact surfaces.The authors investigate the acceleration peak that occurs when the rolling element passes over the defect.
H. Arslan and N. Aktürk [11] developed a shaft bearing model to study the vibrations of an angular contact ball bearing with and without defects.The assembly of the shaft and bearing is modeled to be a mass-spring system that exhibits nonlinear characteristics under dynamic conditions.The equations of motion for the shaft and rolling elements are derived in both the radial and axial directions.
In [12], an experimental investigation focuses on analyzing the vibration signature generated by rolling elements as they enter and exit a defect in the outer raceway of a bearing.The study aims to measure and analyze the vibration responses of the bearing housing and the displacement between the raceways.

Bearing component defect determination
To observe the actual irregularities that influence the dimension allowance and to make it possible to reject the faulty bearings, The Timken Company often uses a testing machine at the end of the line inspection.The measurements are performed with a Vertical Sound Test machine.
The current methodology quantifies the level of the vibration by measuring the dynamic force level of each tested bearing while operating at a specific preload and speed.The level of the measured dynamic force is transformed using a Fast Fourier Transform (FFT).
This paper proposes a new testing methodology to identify potential defects.It is based on the envelope analysis method.Envelope analysis is an established technique that identifies the faults of the bearings' rolling elements.It uses a handpass-filtered signal with a rectifier and smoothing circuit.From the frequency spectrum of the signal, the filter extracts the resonance excited by the bearing fault.For improved results, in this study a new transducer was added in addition to the force transducer and mounted in the radial direction: the accelerometer.

Bearing periodic defect frequencies determination
In rotating machinery systems using bearings, each component rotates at a different speed.Consequently, each component generates noise and vibration at a different frequency.The bearing vibration behaviour depends on shaft speed and the bearing geometry.
With this information, is possible to calculate the bearing periodic frequencies.For this study, tapered roller bearing (TRB) assemblies were used.The following equations detail the formulas used to determine the bearing defect frequencies of the TRBs.

Testing preparation
Regardless of the methodology, the study was performed on a tapered roller bearing part number produced by The Timken Company.The bearing's boundary dimensions and characteristics are presented in figure 1.The Rockwell hardness test method, defined by ASTM E-18, is the most used hardness testing method for metals.It is preferred over other methods due to its ease of use and high accuracy.This hardness test measures the permanent depth of indentation caused by the force/load applied by the indenter.The process involves applying a preliminary test force (preload or minor load) to the sample using a diamond cone indenter.This preload breaks through the surface to spoil the surface finish.
Three levels of localized defect were intentionally created for each part.The diamond indenter was operated at three pressure levels on the inner ring raceway, the outer ring raceway, and the roller body.The values of force applied to the bearing components were 60 kgf, 100 kgf and 150 kgf, and were defined by defect magnitude level: A, B, and C. The localized defects were quantified using a profile gauge from the metrology department.
The process used for indenting the bearing components is presented in figure 2 -the inner ring raceway (left), the outer ring raceway (centre), and the roller body (right).The principle of vibration identification with raised material typically involves analyzing the oscillations occurring on the surface of an object or structure that has raised features or protrusions.
The profile was measured with a Jenoptik HOMMEL T8000 tester.It is designed to measure and analyze the surface roughness parameters of various types of surfaces for quality control, process optimization, and inspection tasks across various industries.Figure 3 presents how the profile was measured for the outer ring imperfection.Using this tester, it was possible to determine the height of the raised material.Compared to the normal profile of the raceway, after the artificial defects were created, the resulting values of the heights were obtained (figure 4): the smallest one of 7.3 microns for an indentation obtained with 60 kgf (left side), the next-highest one of 12.5 microns for 100 kgf (center), and the highest one of 15.

Vibration measurement equipment
The noise, vibration, and harshness (NVH) laboratory equipment available at The Timken Company-Romania facility consists of an acquisition system from Bruel and Kjaer (B&K) and measurement equipment from PCB Piezotronics.The data acquisition system used to acquire the data was the B&K Type 3050 with six input channels.The measuring equipment used included the PCB force transducer 208C03, which measures the bearing axial axis with a nominal sensitivity of 8.91 mV/lb, and the PCB accelerometer, Type 352C34, which measures the bearing radial axis with a nominal sensitivity of 99.8 mV/g.All the acquired data was processed with NVH dedicated software from Bruel and Kjaer -B&K Connect.Figures 5 and 6 present the measurement equipment used, the B&K Connect system, and the Vertical Sound Test machine.

Experimental results
The tests were performed on nine assemblies with the same characteristics, on bearings produced by The Timken Company.Each tested bearing assembly had a defect produced by different applied magnitudes on every bearing component.The recorded data were obtained from two measurement points (Figure 6).The first point is represented by a force transducer and the second one by an accelerometer.The force transducer measures the vibration in the axial direction of the bearing.The accelerometer measures the vibration in the radial direction of the bearing.
The measurements were made at a turning speed of 1800 rpm.The peak amplitude of the spindle is identified for frequencies of 30 Hz and 60 Hz.
The results are presented in the following figures for all the components of the tested bearings.The top row of the figures shows the radial signal from the accelerometer and the bottom row shows the axial signal from the force transducer.The recordings began with the defect with the highest magnitude, defect level C, and concluded with the smallest one, defect level A.

The inner ring vibration measurement
Figure 7 presents the envelope frequency spectrum for the inner ring defect with level C magnitude.The envelope spectrum results indicate that the radially placed accelerometer seems to be more sensitive to the raceway imperfections.The envelope spectrum indicates that the force transducer placed in the axial direction identified the defects, but the amplitude of the defect harmonics is smaller.Additionally, other frequencies with similar amplitudes are present there.
The plotted results identify the first, second, and third BPFI at frequencies of 309, 618, and 927 Hz.The amplitude recorded by the accelerometer for 309 Hz was 798 mm/s 2 .The second defect-specific frequency (618 Hz) had an amplitude of 435 mm/s 2 and the third (927 Hz) is at 254 mm/s 2 .When the results from the force transducer are analyzed, the following amplitudes can be viewed in the graph: 2.78 N, 1.83 N, and 2.92 N.
Also, as mentioned before, the spindle-specific frequencies of 30 Hz and 60 Hz can be observed in the measurements made by the force transducer and accelerometer.Figure 8 presents the acquired data from the force transducer and accelerometer measurement points for an inner ring defect with level B amplitude.The acceleration data also identifies the BPFI with reduced amplitude, and the first, second, and third defect harmonics are present.When the force transducer data is analyzed, the amplitude decreases substantially compared to the C magnitude.The amplitude recorded by the accelerometer for 309 Hz is 312 mm/s 2 , with 182 mm/s 2 for the second defect-specific frequency at 618 Hz and 93 mm/s 2 for the third at 927 Hz.When the results from the force transducer are analyzed, the following amplitudes can be viewed in the graph: 0.54 N, 0.86 N, and 2.92 N. In figure 9 are presented the results for the inner ring with a level A magnitude defect.From the envelope spectrum, the conclusion is that the defect magnitude is not big enough to be detected by vibration measurement with the accelerometer.

The outer ring vibration measurement
When the data for the outer ring is analyzed, the amplitudes of the vibrations from the same defects are higher compared to the inner ring.BPFO defect harmonics can be identified in figure 10.The identified specific harmonics of the defect are the first, second, third, and fourth.The amplitude recorded by the accelerometer for 231 Hz is 1158 mm/s 2 , 694 mm/s 2 for the second-defect specific frequency (462 Hz), 502 mm/s 2 for the third (693 Hz), and 235 mm/s 2 for the fourth (924 Hz).When the results from the force transducer are analyzed, the following amplitudes can be viewed in the graph: 1.25 N, 1.78 N, 0.45 N, and 0.43 N.  The results for the outer ring defect with magnitude A are presented in figure 12. From the envelope spectrum, we can conclude that the defect amplitude is not of sufficient magnitude to be detected by the vibration measurement.

The roller vibration measurement
The next phase of this study was the identification of the roller dent defect.In figure 13 are presented the two envelope spectra from the studied measurement points.The roller defect was created on the roller body; thus, it will be translated to roller defect harmonics or BSF multiples of two.They are identified as the second, fourth, sixth, and eighth roller defect harmonics.The radial acceleration measurement more precisely identifies this defect.
The values recorded by the accelerometer are 818 mm/s 2 at 198 Hz, 444.9 mm/s 2 at 396 Hz, 346.2 mm/s 2 at 594 Hz, and 302.25 mm/s 2 at 792 Hz.The force transducer recorded amplitudes for the same frequencies with the following values: 2.57 N -198 Hz, 1.31 N -396 Hz, 1.1 N -594 Hz, and 1.24 N -792 Hz.

Conclusions
To ensure high quality, every single bearing must be inspected before sending it to the customer.
This study focused on the influence of localized surface variations on tapered roller bearing components.It achieves its main objective of finding a more suitable testing methodology to identify tiny imperfections on tapered roller bearing components.
Three defect magnitudes were created using a Rockwell indenter to simulate defects that can occur during the manufacturing process or bearing installation.The envelope analysis proved to be suitable for identifying dents on the ring raceways or the roller body having 12.5 microns or more of raised material.Compared to the fast Fourier transform method, the envelope method is more accurate in those situations.
When the recorded data from the axial force transducer and the radial accelerometer is combined, it is fast and easy to observe possible defects on the components.
One added value of this study is that most of the available research was performed on ball bearings in general, and this paper is specifically about tapered roller bearing localized defects.

F
Finner race or BPFI -Ball Pass Frequency Inner

Figure 2 .
Figure 2. Identing process for inner and outer ring and roller body: (from left to right: the inner ring raceway, the outer ring raceway, and the roller body).

Figure 6 .
Figure 6.Force transducer and accelerometer position on the Vertical Sound Test machine.

Figure 7 .
Figure 7. Inner ring defect level C envelope analysis.

Figure 8 .
Figure 8. Inner ring defect level B envelope analysis.

Figure 9 .
Figure 9. Inner ring defect type A envelope analysis.

Figure 10 .
Figure 10.The outer ring defect level C envelope analysis.For the outer ring with a magnitude B defect (figure11), the force transducer measurements are not relevant.The defect amplitudes are too small and can't be clearly identified on the envelope spectrum.When discussing recorded measurement values, the accelerometer measured 264 mm/s 2 at 231 Hz, 213 mm/s2 at 462 Hz, 146 mm/s2 at 693 Hz, and 87.8 mm/s 2 at 924 Hz.

Figure 11 .
Figure 11.The outer ring defect level B envelope analysis.

10 Figure 12 .
Figure 12.Outer ring defect level A envelope analysis.

Figure 14 .
Figure 14.Roller defect level B envelope analysisAs for the magnitude A defect, on one roller body of the assembly its amplitude is too small to be identified either by the accelerometer or the force transducer.

Figure 15 .
Figure 15.Roller defect level A envelope analysis