Please can someone explain, simply, why one of a pair of BN displacement probes may consistently give a higher value than the 90deg twin? It seems to be a common occurrence. No machine fault is suspected.

Ta,

Paul

Usually, the reason is because the vibration differs in the two directions.

With fluid film bearings (particularly fixed pad bearings), one expects the vibration to differ in orthogonal directions.

The probes are not considered redundant in this (x-y) configuration. They give distinct information.

Can't add anything to what Bill has said technically. However, sometimes probes may be mounted rolled a 45 off a 90 or true Horz - Vert. Either way they are X-Y probes; just a slight different orientation.

If you are concerned do a calibration check. Sometimes the electronics in the driver will drift. How often do you check the calibration?

Paul, you didn't stipulate how much higher the one probe is compared to the other.

One thing I'd like to add to the coments by Bill Foiles is differences are likely to be the norm, not the exception.

By the way experience has shown that the next likely question is why do two probes at 90° physical mounting angle often have 1X filtered phase lag angles that are not 90° different. Very often I get questioned on calibration for this reason. You can prove this mathematically but suffice it to say that the only time the phase lag angles of two orthogonal probes will differ by exactly 90° is when the major axis of the filtered orbit is directly in line with a probe mounting angle.

John

I really don't do any work with prox probes but wouldn't the phase shift be affected by resonance, misalignment or other forces besides unbalance?

quote:

Originally posted by Jim Crowe:
I really don't do any work with prox probes but wouldn't the phase shift be affected by resonance, misalignment or other forces besides unbalance?

Most of us are familiar with Hooke's law for springs, F = k * x. This is applicable to vibration as well, except all of the parameters are vectors and are frequency dependent.

So, the vibration response as measured by the prox probes is given by x = F / k.

Changes in the external forces, F (misalignment, balance, rubbing) or changes in stiffness, k (resonance, looseness) will have an effect on the vibration response.

Picking up on what Steve said,

Remember that prox probes are measuring the Relative motion between shaft and bearing housing. Both the shaft and bearing housing have different Absolute horizontal and vertical vector motions. With the exception of unbalance, the dynamic forces acting on the shaft may not be uniform over 360-degrees. Bearing housings typically have dynamic stiffness that varies by direction.
It takes two probes to reasonably define shaft motion within the bearing clearance as well as locating its static position on the oil film. The preferred arrangement is two probes 90-degrees apart to resolve two vibration or position vectors. Common installation practice is to locate probes +/- 45-degrees (90-degrees total) from vertical, because many bearing housings have a split-joint in the horizontal position, and they may have hold-down bolts or other obstructions in the vertical direction.
I am not sure I fully understand Paul's statement:
"why one of a pair of BN displacement probes may consistently give a higher value than the 90deg twin?"

I would not expect major differences in measurements from a set of 90-degree probes no matter what the orientation.

Walt

Fixed arc bearings not only have asymmetric stiffness, they also couple the horizontal and vertical directions depending upon load and direction of load (and speed, oil properties). This yields an elliptical orbit at an angle to the load in the bearing even with equal structrual support stiffnesses in the horizontal and vertical directions.

The orientation of the orbit depends on the rotation direction, too. With a vertical load due to gravity this results in an orbit with greater vibration toward one of the probes mounted at 45 degrees (if this is where the probes are). With a machine going the other way (ccw vs cw), the probe with the expected higher vibration will be switched. This typical relationship should be watched by the vibration analyst.

Adding to this may be varying support stiffnesses (including resonance).

quote:

Originally posted by Walt Strong:

The preferred arrangement is two probes 90-degrees apart to resolve two vibration or position vectors. Common installation practice is to locate probes +/- 45-degrees (90-degrees total) from vertical,

I also don't deal with prox probes in real life and that's why the question below popped up in my head just now.

It appears to me that depending on selected X-Y orientation (such as +/- 45-degrees or 0/90 degrees) the 1x filtered orbit SHAPE will also vary ( dependant on effective sforces acting in given direction) and therefore subsequent diagnosis will vary as well. Does it make any sense?

David

This message has been edited. Last edited by: David_G,

Rememeber what Walt Strong said about the fact that proximity probes are making a measurement of shaft "relative". Now, just to simplify things, if we make the assumption that the casing isn't moving then a pair of probes at 0° and 90° right will show the same orbit shape as a pair of probes at 45° left and 45° right of top dead center. Any pair of orthogonal probes will observe the same orbit. The orbit is the path of the shaft centerline, it doesn't vary because we have different mounting angles. The key is that the means of observing the orbit corrects for orientation. In the case of using an o'scope with probes at 45° left and 45° right then to observe the true orbit you need to tilt the o'scope 45°. Today, with most decent software this is handled so that the displayed orbits are corrected for orientation. This of course is subject to the correct transducer mounting angles being input into the software.

Shaft relative probes observe the shaft relative to their mountings. It shouldn't be necessary to assume the casing isn't moving to get the same relative view from different probe orientations. One does needs that to what ever the probes are mounted their distance to each other doesn't change dynamically, or a fair assumption that does this is that their mounting translates as a rigid body -- you don't want to mount the probes to an object that flexes in addition to translating as a rigid body, e.g. long cantilevered probe mounts ... bad idea.

Thanks for all your input. The question arose from a client's gas generator where the inlet X shows 16 microns (I'm a heathen & don't work in mils!) & the Y probe 9. The exit bearing shows X is 11 & Y = 5 microns.

Bill's original remark re vibration in that plane has to be right to a degree. However further light has been shed by Walt Strong wrt vectors & dynamic stiffness.

Thanks all. My senior engineer's back from holiday next week!

Cheers,
Paul

It is not unusual to get differing readings on the probes. It basically depends on the stiffness in the direction that the probes point.
We had some old Clark air compressors with a bull gear in the middle and pinions on each side. I do not remember the exact values but I do remember that there was significant difference due to the pinion not being able to move toward the bull gear. I think that the probe direction was not engineered very well.

I cannot resist commenting what was said above on mounting probes at 45 degrees was/is a way to circumvent practical diffculties in mounting them 12/3 o´clock as mentioned above. This is unimportant as long as you are actually looking for orbits and are only concerned about the bearing oil film behaviour. But if you are looking for the actual rotor response in vector form (which can be more important for early crack detection and balancing aspects) and also keen to monitor for instance how the rotor behaves in steam or gas seals, it is much more informative to place the relative sensors 12/3 o´clock plus move them axially to cover areas of interest, such as next to seals. We have many years of parallel measurements in bearings and on seal areas and with absolute sensors outside bearings structures. We normally find direct information that is useful away from the oil film and ambigous info inside the bearings.

In cases we have to smile a bit since we see other users of sister machines fighting to control them at high expenses (such as several test runs) using just in-bearing-mounted relative probes, as from seal oriented sensors we see in one run good and repeatable data.

Paul, your -very- low varying levels below 20 microns should not cause any kind of concern. Only x1 RPM components have been addressed in above comments. Your levels could be composed from a range of other frequencies as well. I would rather be suspicious if the readings you have would be similar!

quote:

But if you are looking for the actual rotor response in vector form (which can be more important for early crack detection and balancing aspects) and also keen to monitor for instance how the rotor behaves in steam or gas seals, it is much more informative to place the relative sensors 12/3 o´clock plus move them axially to cover areas of interest, such as next to seals.

Common sense and guides like API call for placing the sensors away from nodal locations (pick the favorable side of the bearing), but insist on measuring close to the bearing. Some, like D. Bently, have suggested placing extra probes for identification purposes, and sometimes machines have these but not often.

I don’t understand why mounting the comment about “actual rotor response in vector form.” It can help to rotate the sensors virtually when viewing modal response, but computers can handle this, and besides one doesn’t know ahead of time where to rotate them (Although, a good rotordynamics/structural model could help predict this.). Our industry has used 45’s with great success for many years.

I have identified a number of cracked shafts using probes at 45’s.

One big U.S. (really global) OEM of steam turbines used to mount their shaft riders (shaft absolute with limitations) at 60 degrees from the horizontal for optimal positioning. When the machines got retrofits for shaft probes, the shaft relative measurements would go in at this location sometimes. Some times one could end up with probes, when using two in a plane) mounted only 60 degrees apart instead of 90 degrees, definitely not a best practice. The other big U.S. manufacturer (no longer a U.S. manufacturer, but global) with the probes typically used a vertical mount.

If one is using the probes to monitor a horizontal machine (between pad tilt pads would be an exception, generally have symmetric stiffness and little cross coupling – other tilt pads, too), then one of the 45 degree probes, under normal conditions (including proper alignment) would observe nearly the highest vibration. In this way 45 degree mounting can be a more conservative approach, but the real reason is practical in terms of mounting a bearing with a horizontal split line. Physically, one observes the same information from a pair (hopefully orthogonally mounted) of probes.

What I often dealt with in the past was shaft relative probes at 45’s and a redundant pair of casing transducers mounted vertically. I cannot argue in favor of this design, but I did have to work with it. We also used a number of probes at 55 and 35 degrees, not a problem and probes at the bottom 45’s. Having two orthogonal probes is an advantage no matter the orientation.