Take a fresh look at your lifestyle.

Enabling condition-based monitoring solutions with Sensor performance

Advancements in semiconductor technologies and capabilities are enabling new possibilities to sense, measure, interpret, and analyze data in industrial applications and, particularly, condition-based monitoring solutions. Next-generation sensors according to MEMS technology, coupled with advanced algorithms for diagnostic and prognostic applications, expand opportunities to measure a variety of machines and enhance the capability to effectively monitor equipment, improve uptime, enhance process quality, and increase throughput.

To enable these new capabilities and capture the advantages of condition-based monitoring, new solutions must be accurate, reliable, and robust to ensure that real-time monitoring can expand beyond basic detection of potential equipment faults to provide insightful and actionable information. Performance of next-generation technologies combined with system-level insights enable a deeper understanding into the application and requirements essential to solve these challenges.

Vibration, one of the key aspects of machine diagnostics, has been reliably accustomed to monitor the most critical equipment across an array of industrial applications. A significant amount of literature exists to aid the various diagnostic and predictive capabilities necessary to enable advanced vibration monitoring solutions. Less well covered may be the relationship between vibration sensor performance parameters, such as bandwidth and noise density, and end application fault diagnostic capabilities. This short article addresses the main machine fault types in industrial automation applications and identifies the important thing vibration sensor performance parameters which are highly relevant to the specific faults.

Several common fault types as well as their characteristics are highlighted below to provide insights into a few of the key system requirements that must be considered when developing a condition-based monitoring solution. These include-but aren't limited to-imbalance, misalignment, gear faults, and rolling bearing defects.

Imbalance

What is imbalance and what causes it?

Imbalance is an unequal distribution of mass that causes the load to shift the middle of mass away from the center of rotation. System imbalances could be attributed to improper installations such as coupling eccentricity, system design errors, component faults, as well as accumulation of debris or other contaminates. As an example, the cooling fans included in most induction motors can become unbalanced because of an uneven accumulation of dust and grease, or because of broken fan blades.

Why is definitely an unbalanced system a concern?

Unbalanced systems create excess vibrations that mechanically couple with other components inside the system such as bearings, couplings, and loads-potentially accelerating the deterioration of components which are in good operating condition.

How to detect and diagnose imbalance

Increases in overall system vibration can point to a potential fault created by an unbalanced system, but proper diagnosis of the main cause of the increased vibration is conducted through analysis in the frequency domain. Unbalanced systems produce a signal at the rotational rate of the system-typically referred to as 1×-with a magnitude that is proportional towards the square from the rotational rate, F = m × w2. The 1× component is usually always present in the frequency domain, so identification of an unbalanced product is made by measuring the magnitude from the 1× and the harmonics. If the magnitude of the 1× is higher than the baseline measurement and the harmonics are much under the 1×, then an unbalanced product is likely. Both horizontally and vertically phase-shifted vibration components will also be likely within an unbalanced system.

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

Low noise is required to lessen the sensor influence and let detection of small signals developed by an unbalanced system. This is important for the sensor, signal conditioning, and acquisition platform.

Sufficient resolution of the acquisition system to extract the signal (particularly the baseline signal) is required to detect these small imbalances.

Bandwidth is essential to capture sufficient information beyond just the rotational rates to improve the accuracy and confidence of a diagnosis. The 1× harmonic can be influenced by other system faults, such as misalignment or mechanical looseness, so analysis of the harmonics from the rotation rate, or 1× frequency, might help differentiate from system noise and other potential faults.1 For slower rotating machines, fundamental rotation rates could be well below 10 rpm, meaning the low-frequency response from the sensor is crucial for capturing the fundamental rotation rates. Analog Devices' MEMS sensor technology enables detection of signals down to dc and provides the opportunity to measure slower rotation equipment, while also enabling measurement of wide bandwidths for you can hear content typically related to bearing and gearbox defects.

Misalignment

What is misalignment and what causes it?

System misalignments, as the name suggests, occur when two rotating shafts are not aligned. Figure 2 shows an ideal system where alignment is achieved beginning with the motor, then your shaft, the coupling, and all sorts of way to the burden (which, in this instance, is really a pump).

Misalignments can occur in the parallel direction as well as in the angular direction and may also be a mix of both (see Figure 3). Parallel misalignment occurs when the two shafts are displaced in the horizontal or vertical directions. Angular misalignment is the place among the shafts reaches an angle highly relevant to another.2

Why is misalignment a concern?

Misalignment errors could affect the higher system by forcing components to function under higher stresses, or loads, than what the components were originally designed to handle and may ultimately cause premature failures.

How to detect and diagnose misalignments

Misalignment errors typically show up because the second harmonic from the rotational rate from the system, known as 2×. The 2× component is not always present in the regularity response, but when it is, the connection of the magnitude to the 1× can be used to see whether a misalignment is present. Increased misalignments can excite harmonics out to 10× with respect to the kind of misalignment, the place at which it's measured, and also the directional information. Figure 4 highlights the signatures related to potential misalignment failures.

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

Low noise and sufficient resolution have to detect small misalignments. Machine types, system and process requirements, and rotational rates dictate the allowable misalignment tolerances.

Bandwidth is necessary to capture sufficient frequency range and enhance the accuracy and confidence of the diagnosis. The 1× harmonic can be influenced by other system faults, for example misalignment, so analysis of the harmonics from the 1× frequency might help differentiate using their company system faults. This is also true for higher rotational speed machines. As an example, machines operating above 10,000 rpm, for example machine tools, will typically require quality information beyond 2 kHz to be able to accurately detect imbalance with high confidence.

Multidirectional information also improves the accuracy of the diagnosis and offers understanding of the type of misalignment error and the direction from the misalignment.

The phase of the system, combined with directional vibration information, further increases the diagnostics of the misalignment error. Measuring the vibration at different points around the machine and determining the main difference in the phase measurements or across the system provides insights into whether the misalignment is either an angular, parallel, or combination of the two misalignment types.1

Rolling Element Bearing Defects

What are rolling element bearing defects and what causes them?

Rolling element bearing defects are usually artifacts of mechanically induced stresses or lubrication issues that create small cracks or defects within the mechanical components of the bearing, resulting in increased vibration. Figure 5 provides some examples of rolling element bearings and depicts a couple of the defects that can occur.

 

Why are rolling element bearing failures an issue?

Rolling element bearings are located in almost all kinds of rotating machinery, which range from large turbines to slower rotating motors completely from easy pumps and fans to high speed CNC spindles. Bearing defects could be a manifestation of contaminated lubrication (Figure 5), improper installations, high frequency discharge currents (Figure 5), or increased loading from the system. Failures may cause catastrophic system damage and have significant impacts on other system components.

How are rolling element bearing faults detected and diagnosed?

There are a number of techniques accustomed to diagnose bearing faults and because of the physics behind bearing design, each bearing's defect frequencies can be computed in line with the bearing geometries, the speed of rotation, and also the defect type, which helps with diagnosing faults. Bearing defect frequencies are indexed by Figure 6.

Analysis from the vibration data from the particular machine or system often relies on a mixture of both some time and frequency domain analysis. Time domain analysis is wonderful for detecting trends in the overall increase of system vibration levels. However, very little diagnostic information is contained in this analysis. Frequency domain analysis improves diagnostic insights, but identifying the fault frequencies can be complex due to influences from other system vibrations.

For early diagnosis of bearing defects, harmonics from the defect frequencies are utilized to find out the initial phase, or incipient, faults to enable them to be monitored and maintained before a catastrophic failure. In order to detect, diagnose, and comprehend the system implications of a bearing fault, techniques such as envelope detection, shown in Figure 7, coupled with spectral analysis within the frequency domain typically provide more insightful information.

What system specifications should be considered when diagnosing a rolling element bearing fault?

Low noise and sufficient resolution are important to the detection of initial phase bearing defects. Typically, these defect signatures are lower in amplitude throughout the onset of a defect. Mechanical slip, inherent to bearings due to design tolerances, further cuts down on the magnitude of the vibrations by spreading amplitude information across multiple bins in the frequency response of the bearing, thus requiring low noise to detect the signals earlier.2

Bandwidth is crucial for early detection of bearing defects. An impulse containing high frequency content is created every time the defect is struck throughout a revolution (see Figure 7). Harmonics from the bearing defect frequencies, not the rotational rate, are monitored for these early stage faults. Due to the relationship of the bearing defect frequencies to rotation rates, these early signatures can occur within the several kilohertz range and extend well beyond the 10 kHz to 20 kHz range.2 For lower speed equipment, the inherent nature of bearing defects requires wider bandwidths for early detection to avoid influences from system resonances and system noise that influence the lower frequency bands.

Dynamic range can also be essential for bearing defect monitoring as system loads and defects can impact the vibrations felt by the machine. Increased loads result in increased forces acting on the bearing and the defect. Bearing defects also create impulses that excite structural resonances, amplifying the vibrations felt by the machine and the sensor.2 As machines ramp up and down in speeds during stop/start conditions or normal operation, the changing speeds create potential opportunities for system resonances being excited, resulting in higher amplitude vibrations. Saturation of the sensor can result in missing information, misdiagnosis, and-in the case of certain technologies-damage to the sensor elements.

Gear Defects

What are gear defects and just what causes them?

Gear faults typically occur in one's teeth of a gear mechanism because of fatigue, spalling, or pitting. These can be manifested as cracks in the gear root or removal of metal in the tooth surface. They may be brought on by wear, excessive loads, poor lubrication, backlash, and occasionally improper installation or manufacturing defects.

Why are gear faults a concern?

Gears are the main elements of power transmission in many industrial applications and therefore are subjected to significant stresses and loading. Their own health is critical towards the proper operation from the entire mechanical system. A well-known illustration of this within the renewables field is the fact that the greatest contributor to wind turbine downtime (and consequent revenue erosion) is the failure from the multistage gearbox in the primary powertrain. Similar considerations apply in industrial applications.

How are gear faults detected and diagnosed?

Gear faults are challenging detect due to the difficulty in installation of vibration sensors close to the fault and the presence of significant background noise due to multiple mechanical excitations inside the system. This is especially true in more complex gearbox systems, in which there might be multiple rotational frequencies, gear ratios, and meshing frequencies. Consequently, multiple and complementary approaches could be taken in the detection of gear faults, including acoustic emissions analysis, current signature analysis, and oil debris analysis.

In relation to vibration analysis, the gearbox casing is the typical mounting location for an accelerometer, using the dominant vibration mode finding yourself in the axial direction.7Healthy gears produce a vibration signature at a frequency referred to as gear mesh frequency. This really is equal to the merchandise from the shaft frequency and the quantity of gear teeth. There typically also exist some modulation sidebands associated with manufacturing and assembly tolerances. This is illustrated for any healthy gear in Figure 8. When a localized fault such as a tooth crack occurs, the vibration signal in every revolution includes the mechanical response of the system to a short duration impact in a relatively low energy level. This is typically a low amplitude, broadband signal that's generally considered to be non-periodic and non-stationary.

As a result of these specific characteristics, standard frequency domain techniques by themselves are not thought to be ideal for accurate identification of gear faults. Spectral analysis may be not able to detect initial phase gear failures because the impact energy is contained in sideband modulation, which can also contain energy using their company gear pairs and mechanical components. Time domain techniques such as time-synchronous averaging or mixed-domain approaches such as wavelet analysis and envelope demodulation are generally more appropriate.

What system specifications must be considered when diagnosing a gear fault?

Wide bandwidth is generally very critical in gear fault detection, since the quantity of gear teeth provides a multiplier within the frequency domain. For relatively low speed systems, the necessary detection frequency range is easily pushed in to the multiple kHz region. Moreover, localized faults further extend the bandwidth requirement.

Resolution and low noise are incredibly critical for several reasons. The problem of mounting vibration sensors close to specific fault zones implies that there is potentially higher attenuation of the vibration signal through the mechanical system, which makes it vital to have the ability to detect low energy signals. Furthermore, because the signals are not static periodic signals, standard FFT strategies to extract low amplitude signals from a high noise floor can't be depended on-the noise floor from the sensor itself must be low. This is particularly true in a gearbox environment by which there's a mixing of multiple vibration signatures from different elements of the gearbox. Put into these considerations is the importance of early detection not only for asset protection reasons, but for signal conditioning reasons. It has been shown that vibration severity could be higher in the case of a one-tooth breakage fault, as opposed to a fault with two-or-more-tooth breakage, implying that detection may be relatively easier in the early stages.

Summary

While common, imbalance, misalignments, rolling element bearing defects, and equipment tooth faults are only a some of the many fault types that may be detected and diagnosed with high end vibration sensors. Higher sensor performance, combined with appropriate system-level considerations, enable next-generation condition-based monitoring solutions that delivers deeper amounts of understanding of the mechanical operation of a number of industrial equipment and applications. These solutions will transform how maintenance is conducted and how machines operate, ultimately reducing downtimes, improving efficiencies, and delivering new capabilities to next-generation equipment.

Table 1. Requirements on Each Sensor Parameter
Fault Type Bandwidth Noise Density Dynamic Range Resolution
Imbalance Low Medium High Medium
Misalignment Medium Low/medium High Medium
Bearing High/very high Low Medium High
Gears Very high Low Low High

 

For Table 1, a low bandwidth is recognized as <1 kHz, a medium bandwidth is between 1 kHz to 5 kHz, and a high bandwidth is recognized as >5 kHz. A minimal noise density is recognized as >1 mg/√Hz, a medium noise density is between 100 μg/√Hz to at least one mg/√Hz, along with a high noise density is considered <100 μg/√Hz. A minimal dynamic range is recognized as <5 g, a medium dynamic range is between 5 g to 20 g, and a high dynamic range is recognized as >20 g.