Shaking things up08 January 2020

Some analysis tools promise to automate the interpretation of vibration signals to extend the job of collecting data to unskilled trades, instead of just trained experts. OE explores how they work

Vibration monitoring can be a powerful diagnostic tool, provided an expert is available to analyse the signals. However, often the greatest value is in preventing catastrophic failure, with the accompanying potential for secondary damage, safety incidents and excessive downtime. For this, a simple traffic light system may be sufficient, giving a warning when anything changes beyond a threshold value. New systems from SKF and Schaeffler enable easy, non-expert set-up and can potentially do a lot more.

Any change in the vibration signature of a machine is a potential indicator that a fault has developed. With proper diagnostics, this can enable replacement parts to be ordered and preventative maintenance to be scheduled, minimising downtime. When the vibration signature of a machine is fully understood, it can be used to diagnose specific issues without stopping the machine. Issues with balance, bearings, gears, electrical faults and lubrication can all be identified.

Traditionally, vibration analysis has been a job for an expert, although software has always handled the complex mathematics of processing the signals. Accelerometers attached to rotating machinery are used to measure the vibration that is caused by many different sources, such as misalignments, play and wear. To provide greater sensitivity to lower frequencies, the acceleration signal is often integrated to give the velocity and possibly also the displacement. Software then performs spectrum analysis to identify the underlying frequencies for each source of vibration. It may also separate out the high-amplitude, low-frequency vibration so that high-frequency shock pulses can be identified.

Experts have been required to position sensors and interpret what the underlying frequencies indicate about the condition of machinery. Accelerometers measure motion along a single axis, perpendicular to the surface they are attached to. Some faults produce primarily radial motion, while others produce axial motion. For example, unbalance and many bearing faults produce mostly radial vibrations, while angular misalignment and some bearing faults produce predominantly axial motion. The expert vibration analyst must, therefore, consider the positioning of sensors when interpreting signals. If a number of bearings of the same type are located along a shaft, then the amplitude of vibrations will normally be highest when the sensor is placed directly over the bearing with the fault.

The underlying frequencies, determined using spectral analysis, are also key to understanding the machine. Spectral analysis using Fast Fourier Transforms (FFT) to convert a vibration signal, expressed as acceleration or velocity against time, into a spectral plot giving the amplitude against frequency. This plot will show spikes indicating the frequency of the primary sources of vibration. It is often useful to express the frequencies as multiples of the shaft running speed, denoted 1X.

Faults like unbalance and misalignment have frequencies at shaft running speed, or a few multiples of it. When the 1X spike is relatively low, and all others are much lower, this is generally a good sign. Higher multiples may indicate damaged gears, pumps or fans. Looseness causes harmonics which is seen as spikes at many different integer multiples of X and possibly an increase in the noise floor. Bearing faults produce much higher frequencies that are a function of the number of balls and the rate of precession around the shaft and the housing; this is almost never an exact multiple of the shaft running speed.

Electrical faults on motors may cause vibrations at frequencies that are multiples of the AC supply frequency. Because multiple types of fault may produce vibration at the same frequency, it is sometimes necessary to look at the shape of the waveform in the time-waveform plot to differentiate between them. For example, misalignment produces a sinusoidal waveform, while a damaged gear tooth or roller will produce periodic impacts that appear more like spikes. Differences in the phase of vibration at different positions along a shaft may provide further information.


Correctly diagnosing faults using vibration analysis requires an understanding of the machine. For example, relevant details might include running speeds, numbers of teeth in gears, belt and pulley sizes and bearing details. If this information is made available to a software system, it is possible for it to check the vibration signatures for each component so that faults can be automatically diagnosed. Even if this information is not available, if the baseline condition of a machine is recorded, then any change in the vibration signature can be used to indicate an issue with the condition of the machine.

The Enlight ProCollect from SKF (pictured) is a new portable vibration monitoring solution, that is said to be easy for non-specialists to use, allowing personnel to incorporate vibration monitoring tasks into their activities.

The solution incorporates an updated SKF QuickCollect hand-held sensor, together with SKF ProCollect – a new mobile app – to simplify the collection, interpretation and communication of both operational and machine condition data. As an example, SKF says that pre-programmed inspection routines can be downloaded to a ProCollect device, which will then guide the operator through the steps necessary to collect data. That data is then transferred automatically to SKF’s Enlight Centre platform, where it can be analysed.

Pre-programmed alarms can also be used to help operators and maintenance staff to diagnose and fix problems, and maintenance teams can use the platform’s tools to spot trends, diagnose faults and conduct root-cause analyses. The visualisation capabilities of the Enlight Centre platform, meanwhile, can generate dashboards that provide an overview of plant performance.

Schaeffler has also produced a number of systems that can apply such principles to varying degrees. The most basic of these is SmartCheck. At its simplest, this can take a ‘black box’ approach in which baseline measurements of the characteristic vibration for a machine are used to ‘learn’ the appropriate alarm thresholds.

This approach doesn’t require any specialist knowledge of vibration analysis. An operator can attach the sensors, instruct the software to record the baseline condition and then just start monitoring. If there are any significant changes in machine vibrations, then a traffic light system can be used to indicate that further attention is required. The only thing the operator needs to be careful of, in this case, is that the machine is in a fault-free condition before starting the baseline measurement. When used in this type of black box configuration, SmartCheck can only tell that something is wrong. It can’t tell you what component is causing the fault or whether a replacement part might be needed.

The next step in getting more out of the SmartCheck system is to provide information on the components within a machine. Through a simple web interface, the user can fill out forms to describe the components. Standard templates are available for components such as bearings, motors, fans, pulleys and gears. The user would simply enter the bearing number for a Schaeffler bearing, and the system is then able to obtain the kinematic (or defect) frequencies from an internal database.

For other bearing manufacturers, the defect frequencies must be entered manually and can be saved in the database. For most other components, only some basic information is required, such as the number of blades on a fan or teeth on a gear wheel. The shaft running speed will also need to be provided, either by entering a fixed speed or by connecting a speed sensor to the device.

Once SmartCheck has all of this information on the configuration of the machine, it is able to more intelligently set a low pass filter and lines of resolution to monitor each component individually. When a change in the frequencies associated with a particular component occurs, it can indicate where there appears to be a fault.

If a fault, which cannot be easily diagnosed, is detected, the system can still provide all of the detailed information that a vibration analyst would expect. Because the initial setup has ensured that data collection is relevant to the components being monitored, this data will be immediately useable.

Jody Muelaner

This material is protected by MA Business copyright
See Terms and Conditions.
One-off usage is permitted but bulk copying is not.
For multiple copies contact the sales team.