Back to Basics - Rubbing or Not?

Published on: 

Metrix SETPOINT Contition Monitoring Group; Minden, NV


The first item to be realized is that radial or axial rubbing is not a machinery malfunction.  A rub is secondary indicator that occurs when there is contact between rotating and non-rotating components.  Some of the primary causes that can lead to a rub are:

  • high vibration
  • tight or incorrect clearance
  • thermal growth
  • rotor bowing, etc.  
  • distorted / twisted turbine casing or bearing housing

A rub can be radial, axial or a combination of the two.  When the actual rub/stator contact occurs over a small fraction of the vibration cycle, it is called a partial rub. When it occurs over a majority or all of the vibration cycle, maintaining continuous contact, it is called a full annular rub.  A partial rub is the most common manifestation of a rub.


Axial rubs can result from a mis-match in the thermal growth rates between the rotor and casing.  During a cold startup, the turbine rotor expands faster than the more massive casing.  A related problem can occur when a slide key or trunion on the machine casing hangs up and prevents free movement of the casing during startup or shutdown. The constrained machine casing can deform, displace internal parts, and cause an internal rub.

The mechanism of partial rub usually involves a temporary, sliding rotor contact with the stationary part. During some part of its vibration cycle, the rotating shaft approaches the stationary part. While the stationary part has zero velocity, the shaft surface velocity is non-zero. The shaft contacts the stationary part and maintains contact for some period of time determined by the dynamics of the situation.  Due to the tangential friction force, a rub will always add a certain amount of localized heating to the rotor which will bow the rotor.

At some point during a partial rub, the rotor breaks contact with the stationary part and moves away to complete the vibration cycle. During the next vibration cycle, the rotor repeats the process.  A partial radial rub involves repetition of this process.  The period of time during which the rotor maintains contact with the stationary part is called the dwell time.  Local friction forces can be quite high and generate sufficient local heating to cause rotor bowing, severe wear, local melting or welding of the contacting surfaces, or plastic deformation of the shaft.

Remember that vibration displacement (mils  or µm peak to peak) is the ratio of the input force to the Dynamic Stiffness of the rotor system; i.e. displacement is equal to force / stiffness.  Rubs produce nonlinear changes in both the force and the Dynamic Stiffness, so the rotor dynamics of rub can become quite complicated. 

When a rub occurs, contact forces suddenly appear and disappear.  As the rotor contacts the stationary part, the stator pushes on the rotor while the rotor pushes in an equal and opposite manner on the stator.  This contact force can be separated into radial and tangential (frictional) components.  When contact occurs, the radial force acts in the direction of the rotor center to strongly accelerate the rotor away from the contact point.  The radial force changes during the dwell time of the contact period and is proportional to the instantaneous radial acceleration, a, of the rotor (F = Ma). 

During the period of contact, the tangential friction force appears, which is proportional to the magnitude of the radial force and the coefficient of friction at the sliding interface.  The tangential friction force acts opposite to the surface velocity of the shaft.  It produces a torque on the rotor and, at the same time, tries to accelerate the rotor centerline in the reverse precession direction.  For this reason, a rub will usually produce reverse components in the full spectrum.

This situation is similar to a wheel spinning on ice that suddenly encounters dry pavement.  The sudden contact between the tire and the road produces a friction force that pushes the car in a direction opposite to the surface velocity of the tire at the contact point.  The amount of force depends on the weight of the car and the coefficient of friction between the road and tire.  A side effect of the tangential friction force is that it acts as an agent to transfer the energy of rotation to lateral vibration.  Thus, the magnitude of unfiltered vibration is likely to change with rub onset, as is the amplitude and phase of filtered vibration. Direct (unfiltered) orbits can reveal changes in rotor trajectory due to rub.

For partial radial rubs, a shallow (low angle of incidence) approach to the contact zone will produce a gentle, wiping contact, which does not greatly change the forces in the system.  This kind of rub usually produces only a modified 1X response.

However, if the approach to the contact zone is steep, the rub contact can also be sudden and relatively violent.  In this case, the radial and tangential frictional forces due to rub are impulsive by nature.  They appear suddenly, build to high levels, only to disappear suddenly.  The effect is similar to hitting the rotor shaft with a hammer; an impulse force produces an impulse response that excites many of the free lateral vibration modes in the rotor.  When the rotor rebounds from the contact, it will ring in free vibration at one or more natural frequencies.

Rub produces changes in both the forces and the Dynamic Stiffness of the rotor system.  Because a rub involves rotor interaction with a hard constraint, a rub also introduces nonlinearities in the rotor system.  The result of these effects is a complex rotor dynamic response that produces a wide variety of symptoms.  Like most malfunctions, diagnosis of rub involves correlation of different types of data.  It is important to look at steady state and transient data, including direct orbit and timebase plots; full spectrum, including full spectrum cascade plots; 1X Bode and polar plots; and average shaft centerline data plots.

Rubs are further classified as either being “lubricated” or “dry”.  

  • A “lubricated rub” will typically exhibit very slow wear and can persist for an extended period of time.  
  • If the point of contact between the rotor and stator for a “dry rub” involves dis-similar stator and rotor materials, (bronze stator labyrinth seals / steel rotor), typically the rub will “wear in” and “clear” itself over time. 
  • However, if the point of contact between the rotor and stator for a “dry rub” involves similar stator and rotor materials (steel or stainless steel labyrinth stator seals / steel rotor) typically the vibration levels will increase over time due localized heating and bowing of the rotor without the rub clearing.

There are several symptoms that indicate that a rub is occurring:

  • Changes in 1X vibration behavior at constant speed.Rub begins after reaching constant speed and bows the rotor which results in the 1X vibration changing with time.
  • High 1X vibration amplitude while trying to pass through a critical speed.Rub begins before reaching the critical speed and is already bowing the rotor.  The bowed rotor changes the effective unbalance resulting in high vibration amplitudes when trying to pass through the critical speed.
  • Abnormal (elliptical or highly elliptical) orbit shape
  • Significant reverse precession components
  • Harmonic spectral components; i.e. super or sub-synchronous vibration componentsA rub mechanism imparts energy across the vibration spectra.  Under the right set of circumstances, if operating speed is sufficiently above a critical speed (resonance frequency), then a sub-synchronous frequency may be present at a sub-synchronous frequency equal to the critical speed frequency / operating speed.
  • Temporary thermal bow
  • Abnormal changes in shaft centerline position
  • Wear, damage, loss of efficiency
  • Noise
  • Leaks

The most common effect of a rub observed in the vibration response data is changes in the 1X vector due to thermal bowing of the rotor.  The frictional forces that act during radial rub produce local heating of the surface.  If, at a steady operating speed, a rub occurs repeatedly in the same place on the rotor, the frictional heating of the surface and associated thermal expansion in that area will cause the rotor to bow in the direction of the rub contact.  This bow effectively changes the unbalance magnitude and direction, which changes 1X rotor response.  Under special circumstances, the local heating due to a light rub can produce a continuously changing 1X response vector.

Examination of slow roll data is also critical to the process of evaluating if a rub has occurred.  Rotor bowing and changing 1X slow roll amplitudes can be very conclusive evidence of rotor rubbing conditions.  

During some types of light rubbing, the 1X vibration vector can change over time at steady state speed conditions.  This changing 1X vector over time is sometime referred to in the literature as “The Morton Effect”.  Although much has been written over the past 15 – 20 years relating to the “Morton Effect”, it is not a newly discovered phenomenon. 

The rub location occurs at the high spot location on the rotor (1X vibration vector).  As the rotor rubs at the high spot, it heats the local surface metal and causes it to expand.  This local metal expansion causes the rotor to bow.  The bow changes the net balance condition of the rotor.

Since the phase angle between the heavy spot (angular location of the unbalance) and the high spot (1X vibration vector) is constant for similar operating conditions, if the net unbalance of the rotor changes to a new location due to thermal bowing of the rotor, the high spot 1X vibration vector must change as well in order to maintain the fixed angular relationship between heavy spot and high spot.

When the rotor continues to rub, it rubs in the location of the new high spot.  The rub bows the rotor in a new direction, thus changing the net unbalance of the rotor again as well as the location of the 1X vibration vector or high spot.

This constantly changing bow location results in a rotating 1X vibration vector.  If the 1X vector revolves around the polar center, the net balance change to the rotor is predominately due to the rub/bow.  If the 1X vector revolves in a circle in one quadrant of the polar plot, the initial unbalance of the rotor dominates the location of the vibration vector and is only slightly modified by the rubbing and bowing of the high spot.

This type of rub can last for several months.  If sufficient clearance “wears in”, then the 1X amplitude and phase change will gradually diminish and disappear as the seal clearances open enough to prevent rubbing.



  • Bently, D. E., Hatch, C. T., “Fundamentals of Rotating Machinery Diagnostics”, First Edition, Bently Pressurized Bearing Press, 2002