You are on page 1of 23

Introduction to Rotor Dynamics

Introduction to Rotor Dynamics:


A Physical Interpretation of the Principles and Application of Rotor Dynamics

by

Krish Ramesh, Ph.D.,


Senior Product Technology Engineer,
Dresser-Rand, Houston, TX – 77043.

ABSTRACT

The intent of this short course is to:


• Explain the basic concepts of rotor dynamics
• Describe the application of rotor dynamics to rotating equipment
• Present an overview of the newer technologies (hardware) available to solve rotor
dynamic problems

The objective of the short course is to provide a physical understanding to rotor


dynamics, specifically the vibration characteristics of rotating machinery. This course
covers the principles of lateral vibration of turbomachinery. Because most of these
machines operate in critical services in the oil and gas industries, they are designed to
operate reliably when used properly. The dynamic characteristics of the turbomachinery
need to be completely understood before a machine is placed in service. A basic
knowledge of the underlying principles of rotor dynamics will help promote a better
understanding of the behavior of rotating machinery.

Contents:
• Review of basic vibration principles
• Terminology used in rotor dynamics
• Discussion of journal bearings
• Introduction to rotor dynamics
• Analytical methods: Critical speeds, unbalance response and stability
• Interpretation of results
• Overview of hardware used in solving stability problems in turbomachinery

Short Course at GMRC 2005 Page 1 of 23


Introduction to Rotor Dynamics

Table of Contents

Abstract 1
1.0 Introduction 3
2.0 Basic Vibration Principles and Definitions 3
3.0 Discussion of Journal bearings 6
3.1 Motion of the shaft in the bearing 7
3.2 Bearing stiffness and damping coefficients 7
3.3 A closer look at bearing instability (Oil Whirl) 8
4.0 Entering the World of Rotor Dynamics 10
4.1 Rotor supported on rigid supports 10
4.2 Rotor supported on flexible supports 11
5.0 Rotor Dynamic Analyses 12
5.1 Undamped critical speed analysis 12
5.2 Unbalance response analysis 15
5.3 Damped eigenvalue analysis 17
5.4 Stability analysis 17
6.0 Technologies to Improve the Stability of Rotor-bearing Systems 20
7.0 References 22

Short Course at GMRC 2005 Page 2 of 23


Introduction to Rotor Dynamics

1.0 Introduction

In very simple terms, the turbomachinery consists of a rotor (with impellers/bladed disks,
etc) supported on bearings and rotating in the bearing clearance space. Basically there are
three forms of vibrations associated with the motion of the rotor: torsional, axial and
lateral. Torsional vibration is the dynamics of the shaft in the angular/rotational direction.
Normally, this is little influenced by the bearings that support the rotor. Axial vibration is
the dynamics of the rotor in the axial direction and is generally not a major problem.
Lateral vibration, the primary concern, is the vibration of the rotor in lateral directions.
The bearings play a huge part in determining the lateral vibrations of the rotor. In this
short course, we will study the basic concepts of the lateral rotor dynamics of
turbomachinery. [1,2,3]

2.0 Basic Vibration Principles and Definitions


The rotor dynamic terminology that is commonly used is as follows:
• Rotor: rotating element consisting of the shaft, impellers/bladed disks, shrunk-on
components like sleeves, balance piston, etc
• Bearings: journal bearings that support the rotor in the lateral direction. Thrust
bearings support the axial forces generated during the operation of the
turbomachinery.
• Damper: a device usually in series with a journal bearing used to provide additional
damping to the rotor-bearing system
• Stiffness: a property of a spring defined as force per unit displacement (units: lb/in.).
The effect of stiffness is to cause a sinusoidal motion as shown in Figure 1A.
• Damping: a property (typically of dampers and bearings) defined as force per unit
velocity (units: lb-s/in.). The effect of damping is to cause an exponential decrease in
motion as shown in Figure 1B.
Effect of Stiffness Effect of Damping
Amplitude
Amplitude

Time
Time

Fig. 1A. Effect of stiffness (spring) Fig. 1B. Effect of damping (dashpot)

• Natural frequency: the frequency of vibration of a system (e.g.. rotor-bearing system)


under free conditions (i.e., without external forces). This is a function of the system.
Each system has its own natural frequencies. Consider a very simple system – a mass

Short Course at GMRC 2005 Page 3 of 23


Introduction to Rotor Dynamics

supported by a spring, as shown in Figure 2. The natural frequency of this system is


given by:

m = mass (lb-s2/in) ωn =
k
(Equation 1)
m
k = stiffness (lb/in.)

Fig.2. A simple mass-elastic system

• Damped Natural frequency: In real systems, the damping is almost always present,
though in many cases it may be relatively small. The frequency of vibration of a
system (e.g., a rotor-bearing system) with damping is called the “damped natural
frequency”. Consider a simple system – a mass supported by a spring and damper, as
shown in Figure 3A. The variation of amplitude with time, for the mass-spring-
damper system is shown in Figure 3B. The sinusoidal part of the curve is a result of
stiffness and the exponentially decreasing part of the curve is the result of damping.
The combined effect is an exponentially decreasing sine wave. The assumption made
here is that the damping ratio ζ, is less than 1. The damped natural frequency (again,
in simple terms) of this system is given by:

ω d = ω n (1 - ς 2 ) (Equation 2)
c
where, ς = , and c c = 2 k * m
cc

Damped Vibration of Mass-Spring-Damper System

m=mass (lb-s2/in)
Amplitude

k=stiffness (lb/in.) c=damping (lb-s/in.)

Time

Fig.3A. A simple mass-spring-damper system Fig. 3B. Damped vibration

• Critical speed: When the operating(running) speed of a machine coincides with the
damped natural frequency, it is termed “critical speed” by definition as given by API
617. [4]

Short Course at GMRC 2005 Page 4 of 23


Introduction to Rotor Dynamics

In most cases, this simple equation (Equation 1) provides significant insight into the
physics of the problem. Let us examine this equation more closely. Consider the case
of small machinery. These machines have relatively small mass, but have large values
of stiffness resulting from the combination of shaft stiffness and bearing stiffness.
Hence the damped frequency will be relatively high. This design permits the smaller
machines (assuming the speed of operating is low) to operate in a range below their
damped natural frequencies. This kind of operation is known as subcritical operation
and is highly desirable if it can be attained.

On the other hand consider the case of large rotating machinery like centrifugal
compressors, gas turbines and steam turbines. The mass of the rotor is usually large
and by design there is a limit of the shaft diameter that can be used. Using Equation 2,
it can be seen that as the mass increases, the damped natural frequencies of the larger
machines are much lower. Hence during the operation, these machines typically have
to pass through one critical speed before they reach their actual operating speeds. This
is known as supercritical operation. The main problem is that the machine has to go
through the critical speed during start-up and shutdown. The challenge in designing
such machines would be to properly locate the bearings and ensure that the system
has enough damping to pass through the critical speed.

• Response: The classical dynamic behavior of the rotor-bearing system is plotted in


the form of response plots. The amplitude of the rotor at a particular axial location
(usually the probes, located near the journal bearings) is plotted as a function of the
running speed of the rotor. The term “forced response” or “unbalance response” is
also used to refer to these plots. The external force acting on the rotor is in the form
of unbalance. Unbalance is caused by the mass distribution of the rotor, i.e., a result
of the manufacturing process, when geometric center and mass center do not
coincide. This leaves behind what is known as “residual unbalance”. In the real
world, there is always a finite amount of residual unbalance in the rotor system.

• Stability: The stability of a system is defined as the “reaction” to any external


perturbation. In other words, consider a rotor that is rotating in journal bearings.
Imagine “tapping” the rotor with a “hammer” (perturbation). If the rotor comes back
to its “original” position in a finite amount of time, the rotor is said to be stable. If the
rotor amplitude increases with time (eventually distressing the machine), the rotor is
said to be unstable. Later in this paper we shall discuss in detail the causes for rotor
instability and the various methods and technologies that are available to solve this
problem.

Short Course at GMRC 2005 Page 5 of 23


Introduction to Rotor Dynamics

3.0 Discussion of Journal bearings


The fluid (oil) film bearings are the most common means of supporting turbomachinery.
Of course, rolling element bearings are also used in many gas turbine applications. But
we will discuss the effect of journal bearings on the rotor dynamic behavior later in this
paper. The term journal applies to the portion of the shaft that is supported by a bearing.
An understanding of the role played by the bearings on the dynamic behavior of the rotor
is an essential requirement for engineers who deal with the design and operation of
bearings for turbomachinery.

At zero speed (non-rotating) the shaft is at rest at the bottom of the clearance space. As
the rotor picks up speed, it tends to “climb” on to the inner surface of the bearing. The
convergent wedge formed between the rotating shaft and the inner surface of the bearing
housing acts as a “pump”, pumping the oil beneath the shaft. This lifts the shaft and at
speed, the shaft occupies an equilibrium position. These phases are shown in Figure 4.

Figure 4. Motion of the shaft in the bearing from rest to speed

The rotating shaft is supported by a thin film of oil. The thin oil
film that is “squeezed” between the shaft and the housing,
generates a pressure that supports the rotor weight. Figure 5
shows the pressure profile of a simple sleeve bearing. The
distance between the geometric center of the bearing and the
center of the rotating shaft is known as eccentricity.

Figure 5. Pressure profile in a simple


sleeve bearing

Short Course at GMRC 2005 Page 6 of 23


Introduction to Rotor Dynamics

3.1 Motion of the Shaft in the Bearing:


Figure 6 shows the schematic of a simple type of journal bearing (also known as sleeve
bearing). The shaft rotates in the clearance space (exaggerated) of the bearing as shown.
The clearance is usually about 1.5 mils/inch of the journal diameter. Typically for a 4-in.
journal diameter, the bearing would have a clearance of 6 mils. When the shaft rotates in
the bearing clearance space, it has two kinds of motions.
• The first motion is the rotation or the spin of the shaft. This is the same as the running
speed of the shaft.
• The second type of motion is the precession which is the rotation of the center of the
shaft with respect to the geometric center of the bearing. (An analogy would be the
motion of the earth which rotates about its axis and also revolves round the sun) This
precession (in rotor dynamic terminology) is more commonly known as whirl. This
whirl motion is further classified as forward whirl (shown in Figure 6) and backward
whirl. Forward whirl is the motion in which the center of the shaft moves in the same
direction as the rotation of the shaft. The backward whirl is the motion in which the
center of the shaft moves in the opposite direction as the rotation of the shaft. In
general, the whirl orbit is elliptical.

Whirl = Vibration
Amplitude Spin or Rotation

+++
Whirl direction
(motion of the marker)

Marker on the shaft


++ Bearing inner surface

Figure 6. Example of forward whirl in the bearing clearance space

3.2 Bearing stiffness and damping coefficients:


When the shaft is not rotating it is resting at the bottom of the clearance space in the
bearing. With the clearance space filled with oil, as the shaft starts to rotate, it acts as a
“pump”, “pushing” the oil underneath itself! This generates the lift of the shaft. At any
constant rotating speed, the center of the shaft is located away from the geometric center
of the bearing as shown in Figure 5. This is known as the eccentricity of the journal. The
oil “wedge” supports the shaft. The properties associated with the oil film are stiffness
and damping. These are inherent properties of the oil and are a function of oil type,
viscosity, temperature, etc. For analytical purposes the stiffness and damping are oriented
towards the horizontal and vertical axes – hence, the bearing is said to have a horizontal
stiffness and vertical stiffness (same for the damping). The horizontal stiffness is
indicated by Kxx and the vertical stiffness by Kyy. Similarly the damping is indicated as
Cxx and Cyy. To complicate (real life!) things, because the oil film is continuous around
the shaft, there exist components of the stiffness and damping in the x-y direction also!

Short Course at GMRC 2005 Page 7 of 23


Introduction to Rotor Dynamics

The role played by this cross-coupled stiffness (Kxy) is very important in the
understanding of the stability of rotor-bearing systems.

Bearing clearance
space with oil
Kxx, Cxx

Kyy, Cyy
Figure 7. A simple journal bearing geometry

3.3 A closer look at bearing instability (Oil Whirl):


The definition of Kxx is the “force of the rotor in x direction caused by a displacement in
the x direction”. Similarly, Kxy is the “force of the rotor in x direction caused by a
displacement in the y direction”. Now, Kxy and Kyx are complementary. Because of the
“link” in the x and y forces and displacements, these are called the cross-coupling
coefficients.

Let’s take the example of the plain sleeve bearing in which the Kxy and Kyx values are
relatively high. Recalling the definitions of Kxy and Kyx, a displacement in x-direction
causes a force in y-direction which causes the rotor to move in the y-direction. This
motion in y-direction causes a force in the x-direction, which results in the movement of
the rotor in x-direction! This “feed-forward” mechanism eventually grows to a significant
amount and finally results in instability. In a plain sleeve-type journal bearing, this
happens when the whirl speed coincides with the natural frequency of the rotor. The oil
film looses its capacity to support the load. This could result in a catastrophic failure of
the bearing. The design of the journal bearing has since evolved to in order to improve
the stability of the bearing. Table 1 gives a list of most common types of journal bearings
in increasing order of stability.

Relative stability of different types of Journal bearings:

Stability Design

1. (Least stable) Plain Cylindrical


2. (More stable) Lemon, Multi-lobe
3. (More stable) Offset halves, Pocket, Pressure Dam
4. (Most stable) Tiltpad-design

Table 1. Table showing Journal bearing hierarchy with respect to stability

Short Course at GMRC 2005 Page 8 of 23


Introduction to Rotor Dynamics

The different types of fixed pad (sleeve-type) journal bearings that are used for
supporting rotors are as shown in Figure 8.

Figure 8. Different journal bearing geometries

The latest type of journal bearings are the tiltpad journal bearings where the rotor is
supported by four or five small radial pads that are pivoted inside the bearing housing as
shown in Figure 9.

Figure 9. Tiltpad Journal bearing

The tiltpad bearing design allows each of the pads to rotate about its pivot and attain an
equilibrium position with respect to the rotating shaft. Also, the oil film exists only along
the pad in the circumferential direction. This has practically eliminated the cross-coupling
stiffness in the journal bearing. Therefore the tiltpad journal bearings are the most stable
bearings.

Short Course at GMRC 2005 Page 9 of 23


Introduction to Rotor Dynamics

4.0 Entering the World of Rotor Dynamics


As defined earlier, turbomachinery consists of a rotor (with impellers, bladed disks, etc)
supported on bearings and rotating in the bearing clearance space. To understand the
basic principles of the dynamic behavior of the rotor, let us look at a simple rotor-bearing
system and then extend these principles to the more complicated real-world
turbomachinery.

4.1 Rotor Supported on Rigid Supports


Figure 10 shows a classical Jeffcott rotor – a rotor with an external concentrated mass at
the center, a massless shaft, and supported by bearings at each end.

Figure 10. Jeffcott Rotor – concentrated mass at the midspan and supported at the ends by rigid bearings.

Let us assume that the mass is concentrated at the midspan. The bearings are assumed to
be rigid supports. Thus the rotor can be assumed to be simply supported. Using the theory
of beams, the stiffness of the simply supported beam can be written as,

k =
48 E I 48 E π d 4
=
( )
l3 64 l 3

where, l = “length” refers to the bearing span (axial distance between the bearing
centerlines) and d = diameter of the shaft.

Using the equation 1 for natural frequency, we obtain,


k 48 E (π d 4 )
ωn = =
m 64 l 3 m

and thus proportional to d2/ l1.5. Then for rotors with small (slender) shafts with large
external masses the critical frequency is directly proportional to the square of the
diameter of the shaft and inversely proportional to the 1.5 power of the bearing span. This
important relationship can be used to physically understand the effects of the design
changes to these machines.

For distributed system, the derivation of stiffness becomes a little more complicated. If
we assume no (or negligible) external mass on the shaft of diameter d and length l, and
the shaft mass m, the equation for natural frequency can be written as,
k eq π 4 E (π d 4 ) 4  π 4E  d 2  d
ωn = = * =   = f  
m 64 l 3 ρπ d 2 l  16 ρ  l 4  l2 
 

Short Course at GMRC 2005 Page 10 of 23


Introduction to Rotor Dynamics

In other words, for rotors with small external masses compared to the shaft, the natural
frequency is directly proportional to the diameter and inversely proportional to the second
power of the length. This important relationship also can be used to physically understand
the effects of the design changes to the machine.

In both cases as the diameter of the shaft increases the natural frequency increases
and as the bearing span increases the natural frequency decreases.

4.2 Rotor Supported on Flexible Supports


In the real world, the support is not infinitely rigid. There is a finite amount of stiffness
that a journal bearing provides at the supports. Figure 11 shows the rotor supported on
actual bearings. The bearing stiffness and damping are given by Kb and Cb respectively.

Kb Cb Kb Cb

Figure 11. Jeffcott Rotor – concentrated mass at the midspan and supported at the ends by actual bearings.

Adding the bearing stiffness in series with the shaft stiffness, reduces the effective
stiffness as in the equation,

1 1 1
= +
K eff 2K b K s

Hence, using the equation for natural frequency (ωn = sqrt(Keff/m)), we can see that
adding the bearing stiffness, reduces the natural frequency of the system!

Short Course at GMRC 2005 Page 11 of 23


Introduction to Rotor Dynamics

5.0 Rotor Dynamic Analyses

A rotor dynamic analyses consists of:


• Undamped critical speed analysis
• Unbalance response analysis
• Damped eigenvalue analysis
• Stability Analysis

5.1 Undamped Critical speed analysis


In simple terms “undamped critical speed analysis” means “calculating the critical speeds
of the machines without the effects of damping”! The inclusion of damping adds a huge
degree of complexity in formulating and solving the problem. Historically, the undamped
analysis was the first and in many cases the only analysis done. Most of the calculations
in rotor dynamics are based on matrix manipulations. Because of the lack of high-speed
computers, the matrix calculations had to be done, to a large extent, manually in a time-
consuming manner.

Let’s look at the physics of the analysis and what the output looks like. As defined above,
the output of the analysis is the undamped critical speeds of the rotor-bearing system. The
methodology is to vary the support stiffness from a very low value (flexible supports) to a
very high value (rigid supports) in discrete steps. At each step (or, value of the support
stiffness), the undamped natural frequency of the rotor is determined. The result is plotted
on a chart that is classically known as the “undamped critical speed map” in the
turbomachinery industry. A typical undamped critical speed map is shown in Figure 12.

Actual bearing stiffness

f1-4

f1-3

f1-2

f1-1
1 2 3 4

Flexible supports/Rigid rotor


Rigid supports/Flexible rotor

Figure 12. A typlical undamped critical speed map.

Short Course at GMRC 2005 Page 12 of 23


Introduction to Rotor Dynamics

Now, let’s look into how this plot can be “physically” generated. Let’s start the
“experiment” with the rotor supported on very soft supports. Also the rotor is assumed to
be appropriately instrumented. An accelerometer is placed on the rotor, say at midspan
and connected to a FFT analyzer. This instrument can measure the frequency at which the
rotor vibrates. Let’s assume the support stiffness is 1.0 x 103 lb/in. as shown by “1” in
Figure 12. This represents a very soft spring (“flexible support”). Imagine “striking” the
rotor with a hammer and allowing the rotor to vibrate. The frequency at which it will
vibrate is its natural frequency.

Now, the rotor being a continuous system (as compared to a point or concentrated mass),
it will have many natural frequencies. Usually we would be interested in the first four
natural frequencies. With the help of the instrumentation we measure the first four
frequencies of the rotor. Also, if we measure the movement of the shaft at these
frequencies, we could get the mode shape that corresponds to each of these frequencies.
These four frequencies are shown in figure 12 as f1-1, f1-2, f1-3 and f1-4. The corresponding
mode shapes are shown in figure 13.

Mode shape at f1-1 Mode shape at f1-2 Mode shape at f1-3 Mode shape at f1-4

Figure 13. The Mode shapes of the rotor for the first four frequencies at low support stiffness

Repeat the same experiment of striking with a hammer and measuring the first four
frequencies, now at a increased value of the support stiffness (say, 1.0 x 105 lb/in. as
shown by “2” in Figure 12. This gives four more points on the plot. Repeating the above
process for increasing value of support stiffness gives the four frequencies at each of the
support stiffness. Hence the four curves can be generated. Now, we can see that as the
support stiffness increases, the natural frequencies increase. Let’s look at how the mode
shape for the first critical speed varies with increasing support stiffness, as shown in
Figure 14.A. Figure 14.B shows the mode shape for the second critical speed.

Figure 14.A. The variation of mode shape of the rotor for the first frequency

Short Course at GMRC 2005 Page 13 of 23


Introduction to Rotor Dynamics

Figure 14.B. The variation of mode shape of the rotor for the second frequency

With low support stiffness, the rotor does not bend much. The rotor is referred to as Rigid
Rotor. As the support stiffness increases, the rotor does get to bend. With very high
values of support stiffness (1.0 x 108 lb/in.), the rotor is bent to an extent that the bearings
essentially “lock” the rotor with negligible displacement at the bearing locations. This
condition is referred to a Flexible Rotor, or Rigid Supports. As you can see from Figure
12, the natural frequency of the rotor-bearing system is a function of the support stiffness.
The common terms used in rotor dynamic world are “first rigid-bearing critical speed” or
“first undamped critical speed on rigid supports” and “second rigid-bearing critical
speed” or “second undamped critical speed on rigid supports”. These refer to the first and
second natural frequencies at the condition when the support stiffness is extremely high –
in other words, rigid supports.

Most turbomachinery operates above the first critical speed. Hence, the criteria widely
used in the industry are based on the first rigid critical speed. The ratio, which is very
popularly used in the turbomachinery industry is:
MCOS Max Continuous Speed of the machine
=
Nc 1r First Rigid bearing critical speed

This ratio gives a rough idea of how low the first critical speed of the machine is, with
respect to the maximum continuous speed. API 617 [4] has defined the rules for rotor
dynamic acceptability for centrifugal compressors based on a plot of this ratio vs. the
average gas density of the application. See [5] for an extensive database of centrifugal
compressor applications on this plot.

The bearing has a finite amount of stiffness and damping values. The bearing coefficients
are typically calculated by a “bearing program” that takes in the bearing geometry and
churns out the coefficients. As we have seen in the bearing discussion, the position of the
shaft in the bearing clearance space is a function of speed. Thus the oil film thickness is a
function of speed. Thus the bearing coefficients (K and C) are a function of speed!

Short Course at GMRC 2005 Page 14 of 23


Introduction to Rotor Dynamics

If we plot the bearing stiffness on the undamped critical speed plot, we can gain a wealth
of information about the possible damped critical speeds of the rotor! As shown in Figure
12, the bearing coefficients are plotted on the critical speed map. The intersection of the
bearing stiffness line with the rotor frequency line indicates a potential damped critical
speed of the rotor. The details of this “potential damped critical speed” will be discussed
in the section on unbalance response calculations.

5.2 Unbalance response analysis


Due to nature of manufacturing of the rotor and its components, unbalance always exists,
although in small quantities. The amount of unbalance can be reduced to a tolerable level
by different unbalance techniques. Nevertheless, the residual unbalance that is always
present in the rotor system acts as a forcing function on the rotor and tends to “pull” the
rotor away from the undeformed rotor centerline during operation. Figure 15 shows the
centrifugal force generated by an unbalance. Typically, the unbalance is indicated in “oz-
in”. Assume an unbalance of mass “m” located at a radius “r” and the rotor spinning at
“ω” rad/s. The unbalance force acting on the rotor, F = m*ω2*r. This displacement of the
rotor is the “response of the rotor to the unbalance”.

F=m ω2r
F = m*ω2*r m
m
ω

Figure 15. Unbalance force on a rotor

The response plot is a plot of the displacement of the rotor (at a particular location) as a
function of running speed. The main purpose of an unbalance response calculation is to
determine the actual critical speed and the corresponding amplitude as the rotor increases
from zero speed to its running speed. In the analytical world, to simulate a forced
response, a known amount of unbalance is located on the rotor at specific locations. We
know from previous discussion (Figure 14A) that the first critical speed will have a “half-
sine wave” as its mode shape. Therefore to “force” the rotor to bend at the midspan, the
appropriate location of the “theoretical” unbalance would be at the midspan! The
response of the rotor, typically at the probe locations, is noted down as a function of
speed. Figure 16 shows a typical response plot of a rotor for a midspan unbalance. As the
rotor increases in speed, let us look at one of the probe locations of the machine. This is
the probe located next to the bearing housing (in most cases) at the intake end of the
machine (typical in a compressor). As the speed increases, the amplitude of vibration
increases. The amplitude is maximum at the critical speed of the machine and then
decreases. Typically, machines are designed such that the critical speed is well below
their operating speed. API 617 [4] has rules on how far the peak of the critical speed can
be from the operating speed range.

Short Course at GMRC 2005 Page 15 of 23


Introduction to Rotor Dynamics

Figure 16. Forced response of the rotor for a midspan unbalance

There are a few parameters that are checked using the unbalance response plot.

• Amplification factor: When the rotor response goes through the critical speed, the
response follows a peak. Amplification factor, in simple terms, is the “sharpness of
the peak”. The popular half-power method is used to calculate the amplification
factor. This is described in Figure 17.

x1

0.707 * x1

The half-power method is used to


obtain the “sharpness” of the peak.
Draw a horizontal line at (x1/sqrt(2))
(or in other words, at 0.707*x1) to
get the intercepts N1 and N2.
N1 Np N2
0 A.F. = Np / (N2-N1)
0 5 10

Figure 17. Calculation of Amplification Factor at the Critical Speed

Short Course at GMRC 2005 Page 16 of 23


Introduction to Rotor Dynamics

• Separation margin: This is the distance of the peak of the critical speed with respect
to the nearest operating speed. API 617 [4] has defined the separation margin as given
below. Let us assume that the critical speed peak is at Nc1 rpm. If NMIN (Minimum
operating speed) rpm is the minimum operating speed of the machine and assuming
NMIN > Nc1, the separation margin is given by:
(NMIN - N c1 )
Separation margin (%) = 100 *
NMIN

API 617 [4] has rules on the minimum separation margin based on the Amplification
Factor.

• Amplitude at critical speed: The amplitude at critical speed is indicative of the


severity of the vibration as the rotor coasts through the critical speed. The limit of the
amplitude is determined by the turbomachinery manufacturer, the end user and the
process conditions under which the machines are operated. API 617 [4] has rules on
the limit of amplitude.

5.3 Damped Eigenvalue analysis


In this type of analysis, the natural frequencies of the rotor, supported on its bearings, are
determined. For the analysis, no external forces are assumed. It is similar to allowing the
rotor to rotate, then “hitting” it with a hammer and recording the frequencies of vibration.
It is quite similar to the undamped critical speed analysis, but now the rotor is supported
on actual bearings. Usually the first eight frequencies are noted down as a function of
running speed. The outputs of this analysis are the damped natural frequencies and the
corresponding logarithmic decrement.

5.4 Stability analysis


This is a measure of how the rotor system responds to external excitation. It is measured
by how fast the vibration “decays” with respect to time. There are three definitions of
stability in systems as shown in Figure 18.
1

0.8

0.6

0.4

0.2
Stable system: Exponentially decreasing Sine wave,
0

-0.2
Logdec > 0
-0.4

-0.6

-0.8

1.5

0.5

0
Neutrally Stable system: Sine wave, Logdec = 0
-0.5

-1

-1.5

Short Course at GMRC 2005 Page 17 of 23


Introduction to Rotor Dynamics

2
Unstable system: Exponentially increasing Sine wave,
0

-2 Logdec < 0
-4

-6

-8

-10

Figure 18. Stability of systems

The quantity that is used to measure the stability is called logarithmic decrement, or in
short, log dec. Log dec is the natural logarithm of the ratio of the amplitude of one peak
over the amplitude of the next peak as shown in Figure 19.

0.8

0.6 Log dec = ln(x1/x2)


0.4 x1
Amplitude

x2
0.2 Stable : if log dec > 0
0 Unstable: if log dec <0
-0.2

-0.4

-0.6

-0.8
Tim e

Figure 19. Definition of logarithmic decrement

In rotor-bearing systems, an unstable behavior is characterized by an increase in vibration


at a particular frequency, over time. Eventually the amplitude of vibration could be large
enough to damage the bearings, seals, etc.

The gases, while passing through narrow passages in the aero-path in the compressor,
generate forces on the rotor. Imagine small molecules of the gases passing through the
narrow paths, twisting and turning. These molecules have velocity in all the three
directions. The tangential velocity is the one that generates the “exciting” forces on the
shaft. This is also known as the Swirl velocity. The forces generated by the swirl
component can be modeled as cross-coupling stiffness (“Kxy terms”). (Recall that the
cross-coupling forces are “bad” elements and that such forces reduce the stability of the
system.)

These aero-induced forces are almost always present in the system. To quantify these
forces, an empirical relation is used to estimate them, as given by API [4] and Memmott
[6]. It was derived from the Wachel [7] and Alford [8] numbers. It gives the cross-
coupling forces created by the aero forces (function of stage horse-power, gas density,
speed, etc.). Let’s call this the Wnumber. These aero-dynamic influences can be
considered as “stresses” that are already present in the system. The purpose of the

Short Course at GMRC 2005 Page 18 of 23


Introduction to Rotor Dynamics

stability analysis is to determine how much “cross-coupling” forces can the rotor-bearing
system withstand before the system becomes unstable. For analytical purposes, the rotor
is simulated with an “artificial cross-coupling” at midspan. As this value of cross-
coupling is increased, the frequency (the first forward mode) and the corresponding log
dec are noted down. As the cross-coupling increases, the stability will drop (indicated by
a decrease in log dec). This is done until the log dec becomes zero. This is the threshold
of instability. The amount of cross-coupling gives the maximum aero-forces the system
can withstand before it goes unstable. Let’s call this value “Kxy-threshold”. From the above
discussion we know that the system already has a cross-coupled force of Wnumber.
Now the limit that the system can withstand is Kxy-threshold.

Thus the “Stability Margin” = Kxy-threshold / Wnumber.

API [4] and each manufacturer of rotating machines have acceptable values for the log
dec and Stability Margin based on their experience. The results of the stability analysis
are as shown in Figure 20.

“Inherent” cross-coupling
in the system

Threshold cross-coupling that the


system can withstand = x2

Margin = x2 / “Inherent” Kxy

Figure 20. Results of Stability analysis

Short Course at GMRC 2005 Page 19 of 23


Introduction to Rotor Dynamics

6.0 Technologies to improve the stability of rotor-bearing systems


Lack of stability is one of the cause for poor operation of turbomachinery. It is critical
that the components selected (bearings, seals, dampers, etc) are able to satisfactorily
operate under the given conditions without failure. [9-17]

The two major sources of potential instability in a turbomachinery are the aero-flow path
through the stages (always present) and the passage of the gas through the clearance
space between the laby and the shaft. The cross-coupling generated when the gas passes
through the clearance space under the laby can be reduced by new technological
advances like the swirl brakes and damper seals.

• Damper bearings: In the turbomachinery world, a common way to increase the


stability of rotor-bearing systems would be the addition of damping to the system by
adding a damper. Functionally, the damper that goes behind the journal bearings, is a
ring (also known as the cage) that houses the bearings and is in turn housed and free
to move in the bearing housing. The schematic representation of the damper bearing
configuration is shown in Figure 21. There is an oil film that is trapped between the
outside diameter of the cage and housing. The axial leakage of the oil is prevented by
O-rings. This oil film is historically known as “squeeze film” and hence the
terminology “squeeze film damper”. The oil is fed through the housing to the space
between the cage and housing and then onto the journal. This damper acts as a
“cushion” for the vibrations and hence “dampens” the vibration amplitude. The
stability of the rotor is also increased by this additional damping. The newer design of
this damper assembly has alternate load-carrying components instead of the O-rings.

Figure 21. Damper bearing configuration

• Damper seals: The new technology of damper seals, used to replace toothed
labyrinths at the balance piston or division wall of a centrifugal compressor, has

Short Course at GMRC 2005 Page 20 of 23


Introduction to Rotor Dynamics

proven to be a good means to add additional damping to the rotor system to improve
stability. In principle, the damper seal is like a silencer for a gun. It has radial holes
which “trap” the gases and produce small “pockets of damping”. These technologies
have been put to good use in the recent years and have reduced the number of
stability problems in the turbomachinery industry.

• Swirl brakes: The cause for aero-induced exciting forces, as noted above, is the swirl
or the tangential velocity component of the gas. If we are able to redirect the gas to
reduce the swirl velocity, we could potentially reduce the excitation forces. The swirl
brake is a stationary ring with blade-like axial “teeth” along its circumference. The
teeth redirect the gas flow and reduce the swirl velocity. This device improves the
stability of the rotor-bearing system. It is used at the impeller eye seals and at the
balance piston or division wall seal of a centrifugal compressor.

• Shunt holes: Similar to swirl brakes, taking gas off the diffuser of the last stage
impeller and injecting it before the balance piston or division wall will also reduce the
swirl velocity. They are typically used at the division wall of a back-to-back
compressor.

Acknowledgement
The information contained in this presentation consists of factual data and technical
interpretations and opinions which, while believed to be accurate, is offered solely for
information purposes. No representation or warranty is made concerning the accuracy of
such data, interpretations and opinions.

Short Course at GMRC 2005 Page 21 of 23


Introduction to Rotor Dynamics

7.0 References:

1. Kirk, R.G., 1980, “Stability and Damped Critical Speeds: How to Calculate and Interpret the
Results,” CAGI Technical DIGEST, Vol. 12, No. 2, pp. 375-383

2. Ruddy, A.V., and Summers-Smith, D., 1980, “An Introduction to the influence of the bearings
on the dynamics of rotating machinery”, Tribology International, October.

3. White, D.C., 1972, “The Dynamics of Rigid Rotor Supported on Squeeze Film Bearings,”
I.Mech.E., Conference on Vibration in Rotating Systems, pp. 213-229.

4. API Standard 617 7th Edition, July 2002, “Axial and Centrifugal Compressors and Expander-
compressors for Petroleum, Chemical and Gas Industry Services”

5. Memmott, E. A., 2002, “Lateral Rotordynamic Stability Criteria for Centrifugal Compressors,”
CMVA, Proceedings of the 20th Machinery Dynamics Seminar, Quebec City, pp. 6.23-6.32,
October 21-23

6. Memmott, E. A., 2000, “Empirical Estimation of a Load Related Cross-Coupled Stiffness and
The Lateral Stability Of Centrifugal Compressors,” CMVA, Proceedings of the 18th Machinery
Dynamics Seminar, Halifax, pp. 9-20, April 26-28.

7. Wachel, J. C. and von Nimitz, W. W., 1981, "Ensuring the Reliability of Offshore Gas
Compressor Systems," Journal of Petroleum Technology, pp. 2252-2260, Nov.

8. Alford, J. S., 1965, "Protecting Turbomachinery from Self-Excited Whirl," ASME Journal of
Engineering for Power, Vol. 38, pp. 333-344

9. Memmott, E. A., 1992, "Stability of Centrifugal Compressors by Applications of Tilt Pad


Seals, Damper Bearings, and Shunt Holes," IMechE, 5th International Conference on Vibrations
in Rotating Machinery, Bath, pp. 99-106, September 7-10.

10. Marshall, D. F., Hustak, J. F., and Memmott, E. A., 1993, "Elimination of Subsynchronous
Vibration Problems in a Centrifugal Compressor by the Application of Damper Bearings, Tilting
Pad Seals, and Shunt Holes," NJIT-ASME-HI-STLE, Rotating Machinery Conference and
Exposition, Somerset, New Jersey, November 10-12.

11. Memmott, E. A., 1994, "Stability of a High Pressure Centrifugal Compressor Through
Application of Shunt Holes and a Honeycomb Labyrinth," CMVA, 13th Machinery Dynamics
Seminar, Toronto, Canada, pp. 211-233, September 12-13.

12. Kuzdzal, M. J. and Hustak, J. F., 1996, "Squeeze Film Damper Bearing Experimental vs.
Analytical Results for Various Damper Configurations," Proceedings of the Twenty Fifth
Turbomachinery Symposium, Turbomachinery Laboratory, Department of Mechanical
Engineering, Texas A&M University, College Station, Texas, September 17-19.

13. Ramesh, K and Kirk, R. G., 1999, “Nonlinear Response of Rotors Supported on Squeeze Film
Dampers,” Proceedings of DETC99 – 17th Biennial ASME Vibrations Conference, ASME Paper
DETC99/VIB-8294, Las Vegas, Nevada, September 12-15

Short Course at GMRC 2005 Page 22 of 23


Introduction to Rotor Dynamics

14. Moore, J. J. and Hill, D. L., 2000, “Design of Swirl Brakes for High Pressure Centrifugal
Compressors Using CFD Techniques,” 8th International Symposium on Transport Phenomena
and Rotating Machinery, ISROMAC-8, Vol. II, Honolulu, Hawaii, pp. 1124-1132, March 26-30

15. Ramesh, K, 2002, "A State-of-the-art Rotor Dynamic Analysis Program," Proceedings of the
9th International Symposium on Transport Phenomena and Dynamics of Rotating Machinery,
Honolulu, Hawaii, February 10-14

16. Moore, J.J, Walker, S.T., Kuzdzal, M.J., 2002, “Rotordynamic Stability Measurement During
Full-Load, Full-Pressure Testing of a 6000 PSE Re-Injection Centrifugal ”, 32nd Turbo
Machinery Symposium, Houston, TX.

17. Moore, J. J. and Soulas, T. A., 2003, “Damper Seal Comparison in a High Pressure Re-
Injection Centrifugal Compressor During Full-Load, Full-Pressure Testing Using Direct
Rotordynamic Stability Measurement,” ASME, Proceedings of the 19th Biennial Conference on
Mechanical Vibration and Noise, International 2003 DETC, Chicago, Illinois, DETC2003/VIB-
48458, September 2-6

Short Course at GMRC 2005 Page 23 of 23

You might also like