Vibration Monitoring of industrial Motors

I am a maintenance electrician and want to show how important it is to monitor motor health by monitoring vibration of the motor itself while running under load. I'm wanting to take an arduino UNO and an MPU6050 accelerometer, place them in a box an attach it to the motor frame. Eventually, after I get the actual sensor operational, I want to use and ethernet shield and put it on our network, develop some type of graph to be able to monitor in real-time remotely from an office, control room, etc.

I cant seem to find any type of consistent information that someone has already accomplished this. Would anyone be able to provide some input on how I need to proceed? What type of code to start with? Do I have the correct hardware? What would be the first step in designing a page to monitor said sensor?

There have been commercial systems out there for over 15 years, takes some math to make sense of the measurements, was told to put vibration monitors for some 600 hp electric motors driving metal shredders - monitoring the bearings on the shredders - but could not find a system back then that was reliable enough that there would not be false shutdowns. Ended up putting temp sensors on the shredder bearings - ran fine for about 18 months then in the middle of the night one of the bearings started to fail and went over temp in about 3 minutes I think - system shut down and repairs were only about $5,000- and not the $50,000 had the bearings destructed and exploded sending shrapnel a long ways

the maint. dept did vibration monitoring of some equipment - helped but was not the cure all - probably better now -

good luck

[u]What range of g-forces do you need to measure?[/u]

The MPU-6050 is capable of measuring +-16g, so your "no alarm"- accelerations need to be below that. That may seem a lot, but on a small scale, like machine vibrations, I imagine that might be a common scenario. I could terribly wrong there though, I'm not a mechanical engineer :)

[u]What frequencies are you interested in?[/u]

The MPU-6050s accelerometer is capable of a sample rate up to 1 KHz, provided your code is fast enough. Accordingly to the sample theorem, that means you can measure frequencies up to 500Hz.

The first approach that came to MY mind however was a piezo + precision rectifier.

The vibration of an induction motor servicing a pump shouldn’t have any type of g forces applied if it was indicated properly. I don’t need alarms. Just an analog output that is represented in a form that is acceptable for data logging. I want to be able spot trends of motors are being lubricated properly, have been indicated properly, bearing wear, etc. maybe eventually be able to monitor voltage of each phase, current draw etc.

TheDudeAbidess: The vibration of an induction motor servicing a pump shouldn’t have any type of g forces applied if it was indicated properly.

Not sure what you mean by that. Vibration = change in movement speed and direction = acceleration/g forces.

As saildude already mentioned, there are many commercial solutions for that, some using piezos, some MEMS sensors like the MPU6050.

It depends on what frequencies, amplitudes and resolution you are looking for.

I think I was trying to imply that the vibration should be minimal if it was indicated properly.

If my amplitude needs to be between 2-16g’s because of the MPU6050’s capabilities then I guess I would be limited to around 2g’s for the initial design. The frequency would need to be variable right? So you could fine tune in to find where your most accurate readings would reside?

The frequency would need to be variable right?

The vibration depends on the motor, the bearings, the mount, the load, etc. and will almost certainly exhibit a very complex waveform composed of many frequencies.

I still don't understand. You can't "choose" to limit the acceleration amplitudes and frequencies you have to measure. They are just "there" and you have to choose a appropriate way to measure it.

Sparkfun has a project tutorial on this topic that might be fun/insightful

Piezo Vibration Sensor Hookup Guide

I second the suggestion from Saildude in reply #1.

Had a pump station with 2 - 415v 400hp vertical motors on level 5 of a pump well in a flood zone feeding a line shaft on each down to ground floor pumps.

So many bearing blocks on the way down and all monitored by temperature.

Prior to the sensors, one line shaft let go due to a failed bearing and near tore the internal structure apart as the motor kept running.

After the event and the temp sensors which shut down the motor, not alarms as it is an unattended station, everything has been fine for the last 20 years.

An occasional shut down now and then from a failed bearing.

Oh, and no Arduino, just the temperature sensor relay, a matter of safety. In these conditions one should never rely on electronic programs and such. As a maintenance lecky, I'm surprised you do not know this.