How sensor performance supports condition monitoring solutions

“Advances in semiconductor technology and capabilities provide new opportunities for industrial applications (especially condition monitoring solutions) to detect, measure, interpret, and analyze data. The combination of a new generation of sensors based on MEMS technology and advanced algorithms for diagnostic and predictive applications expands the opportunities for measuring various machines and improving capabilities, helping to efficiently monitor equipment, extend uptime, enhance process quality, and increase output.

“

Advances in semiconductor technology and capabilities provide new opportunities for industrial applications (especially condition monitoring solutions) to detect, measure, interpret, and analyze data. The combination of a new generation of sensors based on MEMS technology and advanced algorithms for diagnostic and predictive applications expands the opportunities for measuring various machines and improving capabilities, helping to efficiently monitor equipment, extend uptime, enhance process quality, and increase output.

In order to realize these new capabilities and obtain the benefits of condition monitoring, new solutions must be accurate, reliable, and robust, so that real-time monitoring can extend beyond the basic detection of potential equipment failures and provide insightful and actionable information. The combination of the performance of next-generation technologies and system-level insights helps people gain a deeper understanding of the applications and requirements needed to solve these challenges.

Vibration is one of the key elements of machine diagnosis and has been reliably used to monitor the most critical equipment in various industrial applications. There is a large amount of literature to support the various diagnostic and predictive capabilities required to implement advanced vibration monitoring solutions. However, there is not much literature on the relationship between vibration sensor performance parameters (such as bandwidth and noise density) and the fault diagnosis capability of the final application. This article introduces the main types of machine failures in industrial automation applications, and determines the key performance parameters of vibration sensors related to specific failures.

The following focuses on several common fault types and their characteristics in order to gain an in-depth understanding of some of the key system requirements that must be considered when developing a condition monitoring solution. The failure types include, but are not limited to, imbalance, misalignment, gear failure, and rolling bearing defects.

unbalanced

What is imbalance and what causes it?

Unbalance refers to the uneven mass distribution, which will cause the load to deviate the center of mass from the center of rotation. System imbalance can be attributed to improper installation (such as coupling eccentricity), system design errors, component failures, and even accumulation of debris or other contaminants. For example, most of the built-in cooling fans of induction motors may become unbalanced due to uneven accumulation of dust and grease or damage to the fan blades.

Why is an unbalanced system a problem?

An unbalanced system will generate excessive vibrations, which will be mechanically coupled to other components in the system, such as bearings, couplings, and loads, which may cause accelerated deterioration of components in good operating conditions.

How to detect and diagnose imbalance

An increase in overall system vibration may indicate a potential failure caused by an unbalanced system, but the root cause of the increase in vibration needs to be diagnosed through frequency domain analysis. The unbalanced system generates a signal at the rotation rate of the system (usually called 1×), the amplitude of which is proportional to the square of the rotation rate, F = m×w². The 1× component usually always exists in the frequency domain. Therefore, an unbalanced system can be identified by measuring the amplitude of the 1x and harmonics. If the amplitude of 1× is higher than the baseline measurement and the harmonics are much smaller than 1×, there is likely to be an unbalanced system.Horizontal and vertical phase shift vibration components may also appear in unbalanced systems¹.

What system specifications must be considered when diagnosing an unbalanced system?

The noise must be low in order to reduce the influence of the sensor and support the detection of small signals generated by unbalanced systems. This is very important for sensors, signal conditioning and acquisition platforms.

In order to detect small imbalances, the acquisition system needs to have a high enough resolution to extract the signal (especially the baseline signal).

In addition, sufficient bandwidth is required to capture sufficient information (not just the rotation rate) to improve the accuracy and reliability of diagnosis. 1× harmonics may be affected by other system faults, such as misalignment or mechanical looseness, so analyzing the harmonics of the rotation rate (or 1× frequency) can help distinguish system noise from other potential faults¹. For slow rotating machines, the basic rotation rate may be much lower than 10 rpm, which means that the low-frequency response of the sensor is essential to capture the basic rotation rate. ADI’s MEMS sensor technology can detect signals as low as DC, and can measure slower rotating equipment, while also measuring wide bandwidths to obtain higher frequency content usually associated with bearing and gearbox defects.

Figure 1. An increase in the rotation rate or 1X frequency amplitude may indicate an unbalanced system.

Misaligned

What is misalignment and what causes it?

As the name implies, when the two rotating shafts are not aligned, system misalignment will occur. Figure 2 shows an ideal system where the alignment starts with the motor, then the shaft, coupling, and up to the load (in this case, the pump).

Figure 2. Ideal alignment system

Misalignment can occur in parallel and angular directions, or a combination of the two (see Figure 3). When the two shafts are misaligned in the horizontal or vertical direction, it is called parallel misalignment.When one axis is at an angle to the other axis, it is called angular misalignment².

Figure 3. Examples of different misalignments, including (a) angle, (b) parallel, or a combination of both.

Why is misalignment a problem?

Misalignment errors may force components to work under stresses or loads higher than originally designed, which can affect larger systems and may eventually lead to premature failure.

How to detect and diagnose misalignment

Misalignment errors usually appear as the second harmonic of the system’s rotation rate, called 2×. The 2x component does not necessarily exist in the frequency response, but when it exists, its amplitude relationship with 1x can be used to determine whether there is a misalignment.Increased alignment error can excite harmonics to 10×, depending on the type of misalignment, measurement location and direction information¹. Figure 4 highlights the features associated with potential misalignment failures.

Figure 4. The increasing 2× harmonics plus the increasing higher order harmonics indicate possible misalignment.

What system specifications must be considered when diagnosing a misaligned system?

In order to detect small misalignments, low noise and sufficiently high resolution are required. The machine type, system and process requirements, and rotation rate determine the allowable misalignment tolerance.

In addition, sufficient bandwidth is required to capture a sufficient frequency range to improve the accuracy and reliability of diagnosis. 1× harmonics may be affected by other system faults, such as misalignment, so analyzing harmonics of 1× frequency can help distinguish other system faults. This is especially suitable for higher speed machines. For example, in order to accurately and reliably detect imbalances, machines (machine tools, etc.) with speeds exceeding 10,000 rpm usually require high-quality information above 2 kHz.

The combination of system phase and directional vibration information can further improve the diagnosis of misalignment errors.Measure the vibration at different points on the machine and determine the difference between phase measurements or within the entire system, which helps to understand whether the misalignment is angular, parallel, or a combination of two types of misalignment¹.

Rolling element bearing defects

What are rolling element bearing defects and what causes these defects?

Rolling element bearing defects are usually the illusion of mechanically induced stress or lubrication problems. These problems produce small cracks or defects in the mechanical parts of the bearing, leading to increased vibration. Figure 5 provides some examples of rolling element bearings and shows several possible defects.

Figure 5. Examples of (top) rolling element bearings and (bottom) lubrication and discharge current defects

Why is rolling element bearing failure a problem?

Rolling element bearings are used in almost all types of rotating machinery, from large turbines to slow rotating motors, from relatively simple pumps and fans to high-speed CNC spindles. Bearing defects may be signs of lubrication contamination (Figure 5), improper installation, high-frequency discharge current (Figure 5), or increased system load. Failure can cause catastrophic system damage and have a significant impact on other system components.

How to detect and diagnose rolling element bearing faults?

There are a variety of techniques for diagnosing bearing faults, and due to the physical characteristics behind the bearing design, the defect frequency of each bearing can be calculated based on the bearing geometry, rotation speed and defect type, which helps diagnose the fault. The frequency of bearing defects is shown in Figure 6.

The analysis of vibration data of a specific machine or system often relies on the combination of time domain and frequency domain analysis. Time domain analysis can be used to detect the overall increase in the vibration level of the system. However, this analysis contains very little diagnostic information. Frequency domain analysis can improve diagnostic insights, but due to the influence of other system vibrations, determining the frequency of failures can be complicated.

For the early diagnosis of bearing defects, the harmonics of the defect frequency can be used to identify early or just-appearing failures, so that they can be monitored and maintained before catastrophic failures occur. In order to detect, diagnose, and understand the system impact of bearing failures, techniques such as envelope detection (as shown in Figure 7) are combined with spectrum analysis in the frequency domain, which can usually provide more insightful information.

What system specifications must be considered when diagnosing rolling element bearing failures?

Low noise and high enough resolution are essential for early detection of bearing defects. When the defect just appears, the amplitude of the defect feature is usually very low.Due to design tolerances, the inherent mechanical sliding of the bearing will propagate amplitude information to multiple bins in the bearing frequency response, thereby further reducing the vibration amplitude, so low noise is required to detect the signal earlier².

Bandwidth is essential for the early detection of bearing defects. During rotation, every time a defect is hit, a pulse containing high-frequency content is generated (see Figure 7). Monitoring the harmonics of the bearing defect frequency (not the rotation rate) can detect these early failures.Due to the relationship between bearing defect frequency and rotation rate, these early features can appear in the kilohertz range and extend beyond the 10 kHz to 20 kHz range².Even for low-speed equipment, the inherent nature of bearing defects requires a wider bandwidth in order to detect defects early and avoid the effects of system resonance and system noise (which will affect lower frequency bands)³.

Dynamic range is also important for bearing defect monitoring, because system loads and defects may affect the vibration experienced by the system. An increase in the load will result in an increase in the force acting on the bearing and the defect.Bearing defects can also produce shocks, excite structural resonance, and amplify the vibrations experienced by the system and sensors².As the speed of the machine rises and falls under stop/start conditions or during normal operation, the changing speed creates potential opportunities for system resonance excitation, leading to higher amplitude vibrations⁴. The saturation of the sensor may lead to loss of information, misdiagnosis, and even damage to the sensor element under certain technical conditions.

Figure 6. The frequency of bearing defects depends on the bearing type, geometry and rotation rate.

Figure 7. Techniques such as envelope detection can extract early bearing defect features from wide bandwidth vibration data.

Gear defect

What is a gear defect and what causes a gear defect?

Gear failures usually occur in the tooth pitches of the gear mechanism due to fatigue, spalling or pitting. It is manifested by cracks in the root of the tooth or metal being removed from the tooth surface.Causes include wear, overload, poor lubrication and backlash, and occasionally due to improper installation or manufacturing defects⁵.

Why is gear failure a problem?

Gears are the main components of power transmission in many industrial applications and are subject to considerable stress and load. The health of gears is critical to the normal operation of the entire mechanical system.There is a well-known example in the field of renewable energy. The biggest factor causing wind turbine downtime (and the corresponding loss of revenue) is the failure of the multi-stage gearbox in the main power system.⁵. Similar considerations apply to industrial applications.

How to detect and diagnose gear failures?

Due to the difficulty of installing the vibration sensor near the fault and the existence of considerable background noise caused by various mechanical excitations in the system, the detection of gear faults is very difficult.This is especially true in more complex gearbox systems, where there may be multiple rotation frequencies, gear ratios and meshing frequencies⁶. Therefore, multiple complementary methods may be used to detect gear failures, including acoustic emission analysis, current characteristic analysis, and oil residue analysis.

In terms of vibration analysis, the accelerometer is usually mounted on the gearbox housing, and the main vibration mode is axial vibration⁷. The frequency of the vibration characteristic of a healthy gear is the so-called gear mesh frequency, which is equal to the product of the shaft frequency and the number of gear teeth. There are usually some modulation sidebands related to manufacturing and assembly tolerances. These conditions of healthy gear are shown in Figure 8. When a partial failure such as a tooth crack occurs, the vibration signal in each rotation will include the mechanical response of the system to a relatively low energy level short-term impact.This is usually a low-amplitude broadband signal, which is generally considered aperiodic and non-static^7,8.

Figure 8. The spectrum of a healthy gear, the crankshaft speed is ~1000 rpm, the gear speed is ~290 rpm, and the number of gear teeth is 24.

Due to these characteristics, standard frequency domain technology alone cannot accurately identify gear faults. Since the impact energy is contained in the sideband modulation, which may also contain energy from other gear pairs and mechanical components, spectrum analysis may not be able to detect early gear failures.Time domain techniques (such as time synchronized averaging) or hybrid domain methods (such as wavelet analysis and envelope demodulation) are generally more appropriate⁹.

What system specifications must be considered when diagnosing gear failures?

Generally speaking, wide bandwidth is very important for gear fault detection, because the number of gear teeth is a multiplier in the frequency domain. Even for relatively low-speed systems, the required detection frequency range will quickly rise to several kHz. In addition, local failures further expand bandwidth requirements.

For many reasons, resolution and low noise are extremely critical. It is very difficult to install the vibration sensor near a specific fault area, which means that the mechanical system may attenuate the vibration signal to a higher degree, so it is very important to be able to detect low energy signals. In addition, since the signal is not a static periodic signal, it cannot rely on the standard FFT technology that extracts low-amplitude signals from high noise floor, and the noise floor of the sensor itself must be very low. This is especially true in a gearbox environment where multiple vibration characteristics of different components are mixed. In addition to these considerations, the importance of early detection is not only for asset protection reasons, but also for signal conditioning reasons. It has been shown that the vibration severity of the single-tooth fracture failure situation may be higher than that of the two or more tooth fracture failure situations, which means that it may be relatively easier to detect at an early stage.

Concluding remarks

Although common, imbalances, misalignments, rolling element bearing defects, and gear pitch faults are just a few of the many types of faults that can be detected and diagnosed by high-performance vibration sensors. The combination of higher sensor performance and appropriate system-level considerations will help achieve a new generation of condition monitoring solutions, allowing people to have a deeper understanding of the mechanical operation of various industrial equipment and applications. These solutions will change the way maintenance is performed and the way machines operate, ultimately reducing downtime, improving efficiency, and enabling the next generation of equipment to have new capabilities.

Table 1. Requirements for each sensor parameter

Fault type	bandwidth	Noise density	Dynamic Range	Resolution
Imbalance unbalanced	Low Low	Medium middle	High high	Medium middle
Misalignment Misaligned	Medium middle	Low/medium Low/medium	High high	Medium middle
Bearing Bearing	High/very high High/very high	Low Low	Medium middle	High high
Gears gear	Very high very high	Low Low	Low Low	High high

For Table 1, it is generally considered that the low bandwidth is less than 1 kHz, the medium bandwidth is between 1 kHz and 5 kHz, and the high bandwidth is greater than 5 kHz. The low noise density is greater than 1 mg/√Hz, the medium noise density is between 100 μg/√Hz to 1 mg/√Hz, and the high noise density is less than 100 μg/√Hz. The low dynamic range is less than 5 g, the medium dynamic range is between 5 g and 20 g, and the high dynamic range is greater than 20 g.

references

¹ Jason Mais. “Spectrum Analysis: The Main Features of Spectral Analysis”. SKF USA, Inc. 2002.

² Robert Bond Randall. Vibration-based condition monitoring: industrial, aerospace and automotive applications. John Wiley & Sons, Ltd. December 2010.

³ Scott Morris. “SKF Pulp and Paper Practice”. SKF Global Pulp and Paper Division, Issue 19, 2016.

⁴ Chris D. Powell, Erik Swanson, and Sorin Weissman. “A Practical Review of Critical Speeds and Modes of Rotating Machinery”. Sound and Vibration, May 2005.

2015 IEEE Workshop on Electrical Machines Design, Control and Diagnosis (WEMDCD), Torino, pp. 297-303, 2015.

⁵ Shahin Hedayati Kia, Humberto Henao and Gérard-André Capolino. “The trend of using Electronic feature analysis for gear fault detection in induction motor-based systems.” 2015 IEEE Symposium on Motor Design, Control and Diagnostics (WEMDCD), Turin, pages 297-303, 2015.

⁷ Giorgio Dalpiaz, Alessandro Rivola and Riccardo Rubini. “The effectiveness and sensitivity of vibration processing technology for gear partial failure detection.” Mechanical Systems and Signal Processing, Volume 14, Issue 3, 2000.

⁸ Wenyi Wang. “Use resonance demodulation technology to detect gear tooth cracks early.” Mechanical Systems and Signal Processing, Volume 15, Issue 5, 2001.

⁹ Kiran Vernekar, Hemantha Kumar, and KV Gangadharan. “Gear fault detection based on vibration analysis and continuous wavelet transform”. Procedia Materials Science, Volume 5, 2014.

Pete Sopcik

Pete Sopcik [[email protected]