Bearing slow down calculation

[Kristian311]

Homework Statement

So, I have a 2 lists of values, first one has bearing rotations in hz, the other one has the times that the measurement was taken. I have a list of good bearing values.
I have to:
1. Make a way to calculate if the bearing is good based on it's own values compared to the values of the good one.
Also the problem is that the good one starts at 50hz and ends at 0 ant the bearings that are going to be tested are going to start at random different values from 40 to 60 hz.

The rotors are the same at all bearings

What approach should I make?

Relevant Equations

No Idea

I thought that I can get a function that is representing the values and compare it to the function of the sample bearing, is this right? or is there and easier approach?

 

[berkeman]

Welcome to PF.

I would start by plotting the values to see if there is an exponential decay or some other decay function. That would help me to figure out what kind of function to fit to the data, and that would help me to start to see how to differentiate between good and bad bearing sets.

Can you try that and upload some screenshots of your plots? Use the "Attach files" link below the Edit window to attach your plots to your post.

EDIT/ADD -- Do you know how to plot the data using Excel?

 

[kuruman]

OK, so the bad bearings start at different values. Do they all end at 0 Hz? If so, isn't it true that a bad bearing starting at, say 40 Hz will take less time to reach 0 Hz than the good bearing over the same 40 to 0 Hz interval?

I would suspect that bad bearings have more friction than good ones. If I were doing this analysis, I would plot the data as @berkeman suggested and use a spreadsheet to fit a polynomial of order 2 or 3. Then I would look at the coefficient of the linear term which is the first derivative of the angular speed, i.e. the acceleration. It should be negative. The larger the magnitude of that coefficient, the worse the bearing. This method does not care at what speed the bearing starts because that's the constant term in the polynomial fit.

 

[jbriggs444]

It might help to have a model for the "badness" in a bearing along with the model for "goodness".

A bearing that catches and comes to a sudden stop would be bad. Maybe there is grit, a broken ball or some foreign object.

A bearing that decellerates rapidly through some ranges of rotation rate would be bad. Maybe there is a "howl" or "chatter" [loose tolerances means that the shaft is not rotating smoothly, but is spiralling or chattering within its raceway].

A bearing that slows too rapidly would be bad. Maybe it was over-tightened or lacks lubrication.

Even a bearing that slows too slowly could be bad. Maybe it was under-tightened.

 

[erobz]

What do you mean you have a list of good bearing values? If its from a manufacturer, then just know that they controlled/measured the loading during the testing. If you are not doing the same (assuming you are in industry - where loads are "best guesses") I wouldn't expect a 1:1 comparison can be made.