Bearing failure undetected on vibration readings

[JeanMicheling]

Hello,

I was hoping maybe someone could have an idea on a failure I just had. A vibration analysis has been done a week ago on a spindle and everythink was looking ok. I just had a call this afternoon reporting a spindle failure. Can someone has an idea on what could have caused this unexpected failure? Could it be a floating issue or a lubrication issue? All the spindles are grease lubricated and they run at 3600 rpm. I assume that no external factors influenced the operation like operators.

 

[svanels]

Hm.. a spindle failure, did the spindle break in two halves? Did the spindle seize? Is the "bearing" failure as stated in the header a conseqence of another failure?
Is it a spindle on a pump or on a high precision lathe?

 

[briand2]

Assuming it was a rotating elemenent (ball or roller) bearing:

Although as they start to fail the vibration increases, I have seen instances where once they've already "failed", i.e. they are now sliding instead of rolling, the vibration actually reduces!
When the bearing tested OK, it might have been well on the road to failure already!

 

[JeanMicheling]

The spindle seized up and it has been caused by a bearing failure.

I would really be surprise that the bearings have slided a week long with the kind of application we have. The spindle is motorized with a 150 hp motor and carry high impacts and high vibrations with a lot of dust and they are filled with grease.

Can the preload be the problem?

 

[TPL]

Well...there's vibration analysis and there's vibration analysis. Its not neccessarily the 'catch-all' that most people think.

What parameters were measured the week before the failure? If you make measurements only in terms of velocity you are likely to miss the high frequencies that indicate the earliest onset of bearing failure?
Do you trend the measured parameters or simply compare against some sort of standard chart of acceptability? High and steady is OK - low and changing is not.

It is quite difficult to catch cage failures using vibration analysis, and depending on the type of bearing an excessive axial loading might overstress the cage.

Cages made of soft material such as brass or plastic can fail with no warning at all: once the rolling elements come in to contact then failure can occurs in minutes

 

[JeanMicheling]

TPL, I disussed with the guy in charge of the vibration readings,and I found out he didn't have bearing defect frequencies. What he did is that he made a 0-500 Hz envelop analysis with velocity and he trended the outside of the envelop. So I think you're right, he hasn't been able to detect the earliest failure stage. He compared his results with a tendencie he built up with other data from the same kind of machine.

 

[Dnicejr]

I am not sure I understand the 0-500 envelope analysis. But at 3600rpm and a 500hz spectrum you are only catching 8.333 orders of run speed, which isn't going to catch your bearing until it's to late. At a minimum he needs to be taking a 4000hz spectrum to get you out closer to 70 orders of run speed.