Machine Health Monitoring (MHM)

30 March 2017

Machine Health Monitoring provides tremendous leverage by reducing downtime and preserving assets. Vibration monitoring enables preventive maintenance on almost any type of machine. It could be done by recurrent control or real time analysis. Condition monitoring involves fixing sensors to mechanical or electrical parts of the machines in order to track failures and malfunctions.

Nowadays, accelerometers are the most common and flexible sensors for this task. Just about any machine or motor could be analysed through this process. The table below shows a non-exhaustive list of applications:

Components Monitored Possible Faults Frequency Range
Support structure Imbalance, external vibrations, oil whirl DC 0 Hz – 100 Hz
Shaft, gear, slide-bearing Imbalance, misalignment, hunting mesh, looseness DC 0 Hz – 1’000 Hz
Gear, bearing Mesh faults, gear pitting, shocks 100 – 10’000 Hz
Bearing, motor Acoustic vibrations 10’000 – 100’000 Hz

Due to the large number of applications and contents of MHM, this document focuses exclusively on low frequency vibrations, from DC 0 Hz to 1’000 Hz.  The following applications could be considered:

·         Hunting tooth detection

·         Fan monitoring in cooling towers

·         Slow speed agitators in industries

·         Geophysical measurement equipment

The market is mostly targeted by industry. As an example, more than 30’000 plants in the US apply monitoring for critical, expensive components, each plant owning from 3 to 5 vibration transmitters.

Explanation of the System

MHM will be set up according to the type of faults to analyse, by measuring different frequency ranges. In general, spectrum analysis is used to identify the different vibration frequencies. An example of a configuration is shown in figure 1.

The signal output is transformed by a DSP unit, generating a frequency spectrum for each sampling cycle. The spectrum is then compared to the component’s reference spectrum, given by the supplier of the component. This operation allows determining whether the frequency of a vibration is caused by the component monitored or by an external fault. If the amplitude of a vibration exceeds a threshold, an alarm is triggered and impending repairs are scheduled.

Environment Analysis

The purpose of MHM is to prevent mechanical failure by detecting growing vibrations. A fault within a system may trigger a devastating chain reaction: for example if a tooth on a pinion gear is broken, it will inevitably damage the contact gears. The broken gears could cause imbalance in their respective shafts, potentially harming the bearings as well.

MHM requires reliable sensors, resistant to possibly high temperature and repetitive shocks. The system must be able to operate in a wet and corrosive environment (e.g. contact with steam in a cooling tower). Furthermore, the sensors must last as long as the component being analysed.

There are various ways to set up a sensing system to the desired component. Handheld measurements are the simplest option, but they require frequent maintenance reviews. Sensors mounted on magnets measure rotating components; on the other hand, they influence the inertia and could cause vibrations. Proximity measurements are also a possibility, despite being difficult to install. Lastly, stud mounts satisfy the stipulation of permanent monitoring and bring little if any disturbance to the system.

MHM Conditions

Monitoring at low frequencies implies two major requirements. First of all, the noise produced by the sensor must be as low as possible: the lesser the amount of electronic noise generated, the earlier a fault is detected. As an example, some CNC machinery demands sensors with a noise level between 75 and 90 dB, operating at a low frequency (DC 0 Hz – 1’000 Hz). Secondly, maximizing voltage output is key to set a sufficient alarm threshold. A good range would be from ±2g to ±10g.

The frequency range of the sensor depends on both the operating frequency and the nature of the potential fault. Fans located in a cooling tower turns at a very low frequency, typically from 100 to 1000 RPM (~ 2 Hz – 20 Hz). Motors generally turn from 500 up to 3’000 RPM (~ 10 Hz – 50 Hz). The table below describes the frequency of the most common defects in terms of the system’s rotation speed.

Technical Solutions

A wide variety of sensors could be used for monitoring of machinery. Although many physical properties could be measured to obtain a vibration analysis, the accelerometers remain in the top position. The table below focuses only on acceleration, velocity and displacement sensors.

Accelerometers offer the best solution in term of installation costs and measurement flexibility. Piezoceramic accelerometers include PE and PR sensors. Both sensors have a broad bandwidth, but cannot measure very low frequency vibrations. MEMS capacitive accelerometers are more accurate than the piezoceramic sensors due to a compensation for thermal drift.



Eddy Current Probe Electrodynamic Sensor Piezoceramic Accelerometer MEMS capacitive Accelerometer
Frequency Range (Hz) @ ±3 dB (max.) 0 – 10’000 10 – 2’000 5 – 10’000 0 – 7’000
Acceleration Measurement NO NO YES YES
Velocity Measurement NO YES YES YES
Displacement Measurement YES YES YES YES
Installation Costs High Low Low Low
Electronic Noise Low High High Low
Durability High Low Low High

Colibrys: a Trustworthy Company

All sensors stated above are able to withstand harsh environments: to enable certification, many tests are done on each sensor after assembly in order to respect the performance required. The sensor is sealed by a ceramic housing for excellent hermeticity. A self-test function is available on a pin to check if a problem within the sensor has appeared (e.g. electrostatic disturbance, saturation).

The vibration sensors are for this application, particularly the VS1000 series. A recommended range of measurement would be between ±2g and ±10g. A noise level of 7 µg/√Hz – equivalent to class C at 90 dB – makes the VS1002 sensor suitable for very accurate measurements. The VS1000 sensors also have a non-linearity of 0.1% FS (typically), a linear frequency response from DC 0 Hz to 2500 Hz at ±5% (typically), and up to 7000 Hz at ±3 dB (typically).

Does it help you ?

1 0

Share content

Leave a Reply

Your email address will not be published. Required fields are marked *

Rediffusion Webinar

03.11.2015 14:00
The video is not available on this post. Please visit LINK to access...
29.10.2015 5 PM CET

Agenda Webinar

No webinar available for the moment.