Many engineers wonder when and how to choose among proximity, velocity, and acceleration sensors when they want to monitor and analyze machinery or structure vibration. The traditional canned answer is “If you are expecting high frequency vibration then use an accelerometer, if you are expecting lower frequency vibrations then you are good with proximity, and velocity is an all-rounder”. We believe this answer was correct at one point of time, but it is not 100% correct any more. Read more below to understand where we are coming from!
To make it simple, if your machine has journal bearings, then go for proximity probes. This is true in at least 95% of the cases. The machine rotor is the piece that initiates most of the vibrations, and proximity probes look directly at this important part, so they are the best to monitor it! The closer you are to the source of vibration, the better.
Modern proximity systems are good up to 5000 Hz (sometimes up to 8000 Hz also), which is more than enough in most cases, so off goes the claim that proximity is good for low frequency vibrations only. Yet, if you are monitoring a high-speed gearbox you may want to use an accelerometer.
Standard Proximity measurements are looking for the movements of the machine shaft relative to the probe mounting place, which is mostly the bearing case. We see the shaft moving/vibrating all the time, but we should remember that the bearing is moving too, and that the reading of the proximity system is a relative vibration reading. This is not a problem since anyway in 99% of the cases what we care for is nothing but the relative movement of the shaft compared to the bearing.
Remember, above recommendations are applicable only if your machine has journal bearings. Proximity does not do very well in monitoring machines with roller element bearings. The later have high frequency and low amplitude vibrations generally.
If your machine has anti-friction bearings (rolling element bearings) the expected vibrations are usually very small in displacement amplitude to be discovered clearly with a proximity probe. So you’d rather focus on the force in vibration rather than the displacement, which means you shall use an accelerometer. Some vibration systems and instruments manufacturers tried using proximity on roller element bearings before, and the results were not so impressive.
Don’t worry about the low frequency vibrations, modern accelerometers from good companies can measure at as low as 0.5 Hz, and as high as 30,000 Hz.
Accelerometers are by far the most used sensor for any application that does not allow for proximity probes.
So what about velocity sensors?
Vibration measurement in velocity units is the oldest casing vibration measurement method. It utilized sensors that had internal moving parts (permanent magnet, coil, known mass, and a spring suspension system). These systems are good but had many limitations that put constrains on their use.
If you are measuring velocity now, then you are most probably using piezovelocity sensors, which are nothing but accelerometers with added inbuilt signal integration circuits. The accelerometer like design has many advantages over the moving coil design, but the inbuilt integration circuit introduces a limitation in detecting high frequency vibrations. The integration process amplified low frequency signal components which may include noise, and attenuates high frequency components which may include gear mesh or bearing frequencies. The added electronics also mean that piezovelocity sensors are more expensive, heavier, and bigger in size than accelerometers.
So why should you bother with these sensors?! The answer is that you don’t really have to! use accelerometers if you can.
Unfortunately, many standard vibration limits were written long time back in units of velocity (when velocity was the common option). If you are using these standards in your plant then you might find yourself cornered into working with velocity units. If this is the case (and you can’t escape it) then always use a piezovelocity sensor, and avoid using an accelerometer and then integrating the signal in the system to get velocity units. The later method may amplify low frequency noise collected through the field wiring.
Some special OEMs still use moving coil velocity sensors; those have special applications and are to be used for these applications only. For example, the Metrix high temperature velocity sensor used in the protection of the GE frame 5 gas turbines.
Our rule of thumb is that you should always analyze signals in the unit they were collected in. As much as possible avoid integrating acceleration and then analyzing the resultant velocity signal; this will be confusing, and the signal quality may be degraded due to amplified noises and attenuated high frequency components. If it’s a case of scaling clarity for the low frequency region, then try getting used to looking at acceleration in Log scales. The same applies to integrating velocity into displacement.
If you collect data using an accelerometer then try to analyze in acceleration units only, and if you collect using a piezovelocity sensor then analyze in velocity units only.
To conclude, use proximity whenever you can, and use acceleration in all other cases. Use velocity only when someone forces you to do it!