Sunteți pe pagina 1din 8

INFLUENCE OF ELASTIC FOUNDATION STRUCTURES ON THE ROTOR

DYNAMICS OF DRIVE TRAINS


Dr. Horst Kuemmlee

Dr. Gnter Siegl

Dr. Peter Woywode

Siemens AG
13624 Berlin, P.O. Box 13 03 00
Germany

Siemens AG
13624 Berlin, P.O. Box 13 03 00
Germany

Siemens AG
47015 Duisburg, P.O. Box 10 15 07
Germany

Abstract - For large motor-compressor drive trains the


influence of the foundation and the supporting structure
is essential for the rotor dynamics and vibration behavior of the complete system and must be taken into
account during the design process. This paper is a
contribution to the understanding of how rotating machines that were originally designed for operation on
rigid concrete foundations can be operated safely on
flexible foundation structures. It will be shown, that
despite the physically necessary reversal of some of
the originally demanded criteria, a safe plant operation
can be achieved. Examples of large motor-compressor
systems show the principal difference between a rigid
and a soft installation. The interaction of all system
components is modeled in detail and compared to
common Standards, like DIN 4024. The paper will give
a guideline for performing such a complete system
evaluation and discuss the acceptance criteria in a
case study of a typical large motor-compressor application.

I. INTRODUCTION
For large electric motor compressor trains the individual machine cannot be considered as an independent unit. The interactions between the components and
the surroundings should be taken into account for vibration purposes. The electric motor must also be able
to run uncoupled and coupled to a loading machine in
the manufacturers test field, as well as coupled to a
complete drive unit with the compressor and skid in the
manufacturers test field and finally on site at the end
destination. Fig. 1 shows a motor-compressor unit for a
natural gas application.

These units are normally the heart of the complete oil


& gas plant. Therefore very high requirements must be
demanded for this type of drive train to get a maximum
of availability and reliability.

II. DESIGN BASIS


In general the vibration design of rotor trains is based
on international Standards like IEC or API. Normally
additional specific customer requirements have to be
considered for the mechanical design of rotors for electrical machines, compressors and other driven machines. The demands on vibration levels have been
tightened over the years. For electrical motors IEC
60034, API 541 and API 546 are the most common
regulations, for compressors API 617 is normally used.
The requirements regarding the location of bending
critical speeds are defined as well as their verification
proof in these standards. The standards for the oil &
gas industry are recognized worldwide. In general
these standards have to be complied with for contracts
coming from this economic sector. Experience shows
that these requirements can be achieved regarding the
rotor dynamic design. Sometimes problems can arise if
the mounting of the machine is different between customers site and manufacturers test field regarding
dynamic behavior. This paper discusses some examples of different mounting conditions for the rotor dynamic behavior of drive trains. It will be shown that
Eigenvalues were shifted in the originally defined speed
range including separation margins and how, by appropriate system evaluation, possible problems can be
avoided.
A. Rotor
The finite element model of the rotor is always simplified to rotating beam elements with all rotational
masses, inertias and other restraints like magnetic pull
of the electric motor included. For high-speed applications the gyroscopic effects of shaft and components
must be taken into account as well as the flexibility of
stiff-coupled components.
B. Bearing system

Fig. 1: Motor-compressor unit for natural gas application

For the rotor dynamic design of a machine shaft it is


normally assumed that the machine is mounted on a
very stiff concrete foundation. This means that the
dynamic characteristic of the foundation is decoupled
from the rotor train, and no interaction between the
natural frequencies of rotor and foundation exists. The
usual standard bearing conditions are the speeddependent oil film stiffness and damping combined with
bearing pedestal stiffness and modal masses of the
bearing shell and pedestal, as well as the modal mass
of the base frame (Fig. 2).

The next refinement step for shaft trains mounted on


sub-critical foundations is to add a simplified frame
structure and modal masses for the foundation. For stiff
coupled machines it is necessary to model both rotors
for motor and compressor in coupled condition (Fig. 4)

Fig. 4: Motor and compressor rotor and simplified frame


structure on sub-critical foundation

Fig. 2: model of sleeve bearing and bearing support


C. Foundation
For foundation design the stiffness of the machines
normally will not be considered. The mass and mass
moment of inertia, e.g. of the motor, will be modeled
using single mass points. The connection of the mass
points can be realized using rigid beams to the foot
point of the machines.
In an ideal case the stiffness of turbo machinery
foundations should be higher than the stiffness of the
machinery oil film and bearing housing, respectively
casing stiffness. For large turbo machinery trains this
cannot be realized in every case due to arrangement
and cost/space restrictions.
In accordance with DIN 4024 there are two kinds of
foundations. First there are so called rigid foundations,
meaning the first Eigenvalues are higher than the operational speed. In most cases so called flexible foundations can only be realized, that means the first or
more Eigenvalues are within the operational speed
range. In that case especially the stiffness of the foundation may influence the rotor dynamic calculation.
To avoid foundation caused vibrations the first Eigenmodes of the foundation should be at the minimum
20 % lower or 25% higher than the operational speed.
The higher mode should have a separation margin of
more than 10 %.
The second design criteria also used for foundation
design are effective vibration amplitudes at the machinery bearing housing and/or casing in accordance
with DIN ISO 10816. For motor design vibration, velocity in accordance with DIN ISO 10816-3, should not
exceed 4.5 mm/s for rigid foundations and 7.1 mm/s for
flexible foundations. The values are for Zone B/C which
means during operation. For the response analysis an
imbalance quality of Q=6.3 and a structural damping of
2 % in accordance with DIN 4024 should be assumed.
D. Evaluation steps for system analysis
In the first design stage, motor and compressor rotors are handled as individual units connected by linear
spring elements, which represent the surroundings like
base frame and foundation structure (fig. 3).

For super-critical foundations or machines running


near to a significant foundation Eigenvalue a complete
system analysis including motor frame and foundation
structure is required (Fig. 5). Soft mounted parts - like
motor hubs or auxiliaries or parts where detailed
structural information is not available like the gear
box in the example figure are modeled as rigid mass
and inertia connected by spring elements.

Fig. 5: Complete system analysis including foundation


structure
III. EXAMPLES FOR SYSTEM VERIFICATION
The system units discussed normally have life times
of some decades and a high grade of investment.
Therefore in many cases only parts of the plant were
renewed after reaching their life time or due to performance improvements. The following chapter deals
with a revamp of an old compressor station and the
second part with planning for a new train.
A. Revamp of a motor-compressor set for an air
separation plant
Fig. 6 shows an example of a revamp of a stiff coupled motor-compressor train for an air separation plant
mounted on a concrete table foundation. An old synchronous motor is substituted by a new type, the com-

Fig. 3: Motor rotor on stiff foundation

pressor remains. The foundation must be modified to fit


the new motor frame.

structural deformation of the housing is negligible


compared to the piles.
6000

7
7

[1/MIN]

4140.0

FREQUENZ

4000
3600.0

3060.0
6

6
5

5
4

2000

1
4

4
3

2
1

2
1

3
1
1

1
1

Fig. 6: Revamp of an air separation compressor train

1000

2000

3000

4000

5000

[1/MIN]

ROTATIONAL SPEED
100

[%]
1

DAEMPFUNG

80

60
1

40
4

4 1

20

1000

69
7
5
23
8

9
5
7
2
3
1
8

2
1
5
9
7
3
8

4
1
6

2000

3000

64
1

4000

5000

[1/MIN]

ROTATIONAL SPEED

(
4140

(
3060

p
4140

(
3060

3060

Fig. 9: Shaft vibrations caused by in-phase imbalance at


motor active part
4140

In the first design stage the rotor train is modeled on


a rigid foundation represented by spring elements below the bearings (Fig. 7). Fig. 8 shows the critical
speed chart with the Eigenfrequencies and the related
modal damping of the system as a function of the rotational speed. The operational speed range including the
separation margin is free of natural bending frequencies, according to API standard 546.
Imbalance response calculations according to API 546
show no inadmissible resonance amplifications in the
operational speed range. Fig. 9 shows the in-phase
imbalance at the motor active part and Fig. 10 shows the
out-of-phase imbalance reaction. The resonance
amplification of the system is different for the individual
imbalance settings; also the interaction between motor
and compressor is affected.
The next step in the refinement of the vibration model
is to take into account the modal mass of the machine
housing and the foundation table (Fig. 11). In this case
the foundation stiffness is also modeled as linear
springs. The machine housing is modeled as a 3
dimensional rigid body, the mass and moments of inertia
were concentrated in the center of gravity. The rotor
bearing nodes were coupled by rigid links. Some
compressor manufacturers use only vertical steel piles
for height adjustment of the motor and compressor. This
leads to weak supporting stiffness values which can
affect the rotor-dynamic behavior of the system
significantly. This model assumes that the dynamic

3060

Fig. 7: Standard model for the rotor-compressor shaft


string

Fig. 8: Critical speed chart of drive train system on a


very stiff foundation

4140

The existing compressor is rigid coupled to the new


motor rotor and this cannot be changed in the new
design due to a natural frequency location on the compressor side, because of this stiff coupling condition
between motor and compressor the interaction of both
rotors cannot be neglected for the evaluation of dynamic behavior. The complete shaft string has to be
treated as a statically undefined support system.

Fig. 10: Shaft vibrations caused by out-of-phase imbalance at motor active part

4140

3060

4140

3060

Fig. 11: Model extended by housing and foundation plate


masses
Due to these additional degrees of freedom additional
Eigenfrequencies can be created and these may fall into
the operational speed range. This can cause the risk that
the originally planned separation margin of the API
cannot be met any more. In the present application no
problems occurred (compare Fig. 12 to 14).
6000

7
7

[1/MIN]

4140.0

FREQUENZ

4000
3600.0

3060.0
6

6
5
5
4

2000

9
4

2
1

2
1

2
9

0
0

1000

2000

3000

4000

5000

[1/MIN]

ROTATIONAL SPEED
100

DAEMPFUNG

[%]
80

60

40
4

20

1000

6
7
5
23
8

5
7
2
3
1
8

2
1
5
7
3
8

4
1
6

2000

3000

4000

5000

[1/MIN]

ROTATIONAL SPEED

Fig. 12: Critical speed chart of a drive train system


including housing stiffness on a very stiff
foundation
p

The next step in refinement of the vibration model is to


take into account that the motor and compressor are
mounted on a table shaped concrete or steel foundation
(Fig. 15). The foundation table of this frame itself seems
to be very stiff but it must be taken into account that this
big mass rests on soft and high lateral supports. In
connection with the large additional masses of the
machines and the additional aggregates/auxiliaries this
always produces a complete system with structural
Eigenvalues and related mode shapes that have to be
considered during planning of the installation. In the
process alternating forces between the rotor and the
remaining structure can lead to unforeseen resonance
behavior.
The critical speed chart of this system shows a large
number of Eigenvalues which are also located within the
operational speed range (Fig. 16). The Eigenvalues of
the system generated by the interaction with the
foundation must be taken into consideration because of
possible interference by the rotor modes. The example
shows that normally the separation margins for
calculated foundation modes defined in several
standards cannot be met. The severity of each individual
foundation mode on the rotor-dynamic behavior of the
drive train must be examined and checked for
interference with the rotor. In simplified rotor-dynamic
models only the damping of the oil film is included.
Therefore pure foundation modes lead to zero damping
and high local amplifications. To build a sufficiently
accurate model of the complete system to verify
amplification factors, material damping for foundation
and frame structures must be taken into account.

(
4140

(
3060

p
4140

(
3060

Fig. 14: Shaft vibrations caused by out-of-phase imbalance at motor side

Fig. 13: Shaft vibrations caused by in-phase imbalance


at motor side

Fig. 15: Concrete foundation with motor and compressor


as in previous model

5000

[RPM]

4000

FREQUENCY

3600.0

S
R
Q

P
O

P
O

P
O

M
L

K
J

K
J

K
J

I
H

I
H

I
H

G
F

3000

G
F
E

D
C

D
C

D
C

9
A
8

1000

G
F

2000

T
S
R
Q

S
R
Q

A
8

A
8

7
6

7
6

5
4

5
4

2
1

2
1

5
3
4
2
1

0
0

1000

2000

3000

4000

5000

[RPM]

ROTATIONAL SPEED

Fig. 18: Shaft vibration at E-motor bearings for out-ofphase imbalance

60
4

DAMPING

[%]

40
4

20
9
4
9

0
0

1000

4
6
T
B
5
S
A
C
8
3
E
L
D
J
K
G
Q
O
1
2
7
H
N
R
M
P
IF

5
6
S
B
T
C
A
3
8
E
D
J
1
F
G
K
L
Q
2
7
IH
M
N
O
P
R

2000

5T
3B
A
S
6
C
8
E
L
K
Q
D
J
1
G
O
2
7
H
N
R
M
P
IF

3000

ROTATIONAL SPEED

6
5
3

4000

5000

[RPM]

Fig.16: Critical speed chart of the machine set up on a


concrete table foundation
Despite the big number of natural modes near the
operational speed indicated in the critical speed chart the
motor shows no significant resonance amplifications
(Fig. 17 and 18).
The vibration level at the foundation is verified by
vibration velocity at six selected points (Fig. 19). Fig. 20
and 21 show the diagrams for in-phase and out-of-phase
imbalance at the motor rotor. The results show that
resonance peaks at the foundation (Fig. 22 and 23) can
be excited without creating high shaft vibrations.
Dependent on the imbalance distribution, local
foundation resonances can occur.

Fig. 19: Overview of the complete model with numbered sensor points.

Fig. 17: Shaft vibration at E-motor bearings for in-phase


imbalance
Fig. 20: Foundation velocities at sensor points 1 to 6 for
in-phase imbalance at motor rotor

B. Design of a large motor-compressor set for a process


plant
The following example shows a drive arrangement
with a gas turbine elastic coupled on the left side, to a
modeled feed-through E-motor/generator and two
compressors coupled to the right end and mounted on a
table shaped steel foundation. The calculation model
(Fig. 24) shows the highest resolved step of rotordynamics calculation. The motor with all rotating and
non-rotating parts is modeled in detail. The rotor is
modeled similarly to the previously shown example. The
coupling of the rotor bearing with the lower structure is
again conveyed by the rotation speed dependent
stiffness- and the damping matrix of the oil film.

Fig. 21: Foundation velocities at sensor points 1 to 6 for


out-of-phase imbalance at motor rotor.

Fig. 24: Starter-helper-motor on a steel table foundation

Fig. 22: Foundation mode at 2395 rpm with D=3.0 %

Fig. 25: Sketch of a similar motor-compressor train on a


concrete foundation

Fig. 23: Foundation mode at 3880 rpm with D=0.4 %

Looking at the critical speed chart one can surmise


that, due to the high number of Eigenfrequencies in the
originally defined separation margin region, high
resonance levels can be expected (Fig. 26).
For assessment of the Eigenvalues in a first step
modal mass analysis can be used. Local modes of the
foundation normally can be neglected.
Usually there is no interaction between the first
foundation modes and the critical rotor speeds. But there
might be an influence on rotor modes due to stiffness
influence. Lowering of the foundation stiffness may lower
the critical speeds. This can be assessed by amplitude
and phase response analysis. Different kinds of

excitation scenarios due to rotor imbalance have to be


considered. For a better assessment of the situation it
can be surmised that, due to the many Eigenfrequencies
in the originally defined cut-off area, high vibration levels
can be expected. For the existing situation the modal
damping of the structure has to be considered.
5000

T
U
S
R
Q
P
O
M
N
L
J
K
IH
F
G
E
D
C
B
987654
A
321
00
0Z
Y
X
W
V
U
T
S
R
Q
P
O
N
M
LK
J
I
H
G
E
F
D
C
B
A
9876
54321
00
Y
Z
W
X
V
U
T
R
S
Q
P
O
N
M
LK
JI
H
G
F
E
D
C
B
A
987
6
543
210
0
0Z
Y
X
W
V
U
S
R
T
Q
P
O
N
M
LK

[RPM]
4140.0
4000

FREQUENCY

3600.0

3060.0

3000

2000

1000

W
V

T
U
S
R
Q
P
O
N
M
L
J
K
H
I
F
E
G
D
C
B
98764A
35
0210
0Y
Z
X
W
V
U
T
R
S
Q
P
N
M
LJO
K
I
H
G
E
F
D
C
B
A
8697
54321
00
Y
Z
X
W
V
U
T
R
S
Q
P
O
N
M
LJK
I
H
G
F
E
D
C
B
987A
6
543
210
0
0Z
Y
X
W
V
U
S
R
Q
P
T
O
N
M
LK

T
S
U
R
Q
P
O
N
M
L
JK
H
I
F
E
G
D
C
B
98764A
35
0210
0Y
Z
X
W
V
U
T
R
Q
S
P
O
N
LM
JK
I
H
G
E
F
D
C
B
9658A
420731
0Y
Z
X
W
V
U
T
R
S
Q
P
O
N
M
LJK
I
H
G
F
E
D
C
B
987A
6
543
210
0
0Z
Y
X
W
V
U
S
R
Q
P
O
T
N
M
LK

T
S
U
R
Q
P
O
N
M
L
JK
H
I
F
E
D
G
C
B
98764A
5
01032
0Y
Z
X
W
V
U
T
R
Q
S
P
O
N
LM
JK
I
H
G
E
F
D
B
C
9658A
420317
0Y
Z
X
W
V
U
T
R
S
Q
P
O
N
M
LJK
I
H
G
F
E
D
C
B
987A
6
543
210
0
0Z
Y
X
W
V
U
S
R
Q
P
O
T
N
M
LK

JI

JI

JI

JI

H
G
F

H
G
F

H
G
F

H
G
F

E
D
C
B

E
D
C
B

A
9

A
9

E
D
C
BY
A
9

8
7
6
5

8
7
6
5

4 X
3
2
1

V.

E
D
Y
C
B
A
9

Z
8
7
6
5

8
7
6
5

4
3

4
3

4
3

2
1

2
1

2
1

0
0

1000

2000

3000

ROTATIONAL SPEED

structures, a large Eigenvalue will be found within the


separation margin region. In this case the shaft string
with its existing Eigenmodes is in strong interaction with
the surrounding structural parts, so that the vibration
level could be changed significantly compared to an
assumed stiff foundation assembly.
As shown by practical experience, assemblies like the
one discussed in the above example do not cause
technical vibration problems. In the described case study
the shaft and bearing vibrations, as well as the vibration
speeds at selected foundation points, do not exceed the
given limit values.
But the results show also that the separation margin
criteria alone combined with a simplified rotor-dynamics
model alone may not be sufficient to evaluate the train
behavior under exciting forces and moments, like
imbalances. To avoid vibration problems on site under
any circumstances the dynamics of the complete system
must be studied in more detail.
Evaluation and testing of single components, for
example the motor, is not sufficient to judge the later
behavior of the string on site.

4000

5000

[RPM]

Fig. 22: Critical speed chart of the motor on a steel


foundation
The motor manufacturer always pays close attention
to the standards given to him by the customer for the
rotor design as defined under the condition it will be
operated on a very stiff foundation. If there is a deviation
from this type of set up, it could lead to difficulties in
keeping to the cut-off regions. But the results and
calculations show also that as a rule, the additional
Eigenfrequencies blended into the cut-off region, can be
relatively clearly identified. If the critical speed charts are
compared to the calculation this can be clearly
recognized.
IV. SUMMARY
Looking at the calculation results for system models
with simplified stiff foundation, the API standard in
respect to the separation margin criteria is fulfilled. The
stiffness values of the spring elements which represent
the foundation were chosen by experience and can be
verified in motor manufacturers test field during the
acceptance test. Even with the modal motor frame
masses included the results show little difference to a
very stiff mounted shaft string with spring-damper
bearings. Therefore in most of the applications the
simplified rotor/bearing calculation is sufficient.
For machine sets mounted on table foundations the
critical speed chart shows a number of Eigenvalues
within the API separation margins. This is generally the
case if the support structure is flexible compared to the
shaft train behavior. Looking at the steel foundation

LESSONS LEARNED

This report is a contribution to the understanding of


how rotating machines that were originally designed for
operation on rigid concrete foundations can be operated
safely on flexible foundation structures.
The rotor dynamic behavior of shaft trains, consisting
of an electric motor with a flexible coupled compressor
usually have fulfilled relevant standards like API 546 for
electrical synchronous machines as well as the
compressor standard API 617 or similar alternative
customer requirements. Essential requirements are that
the natural frequency of the critical bending speeds of
the rotor may not be in the speed range and have to be
kept at a minimum distance from the upper and lower
limits of the operational speed range.
These boundary conditions are taken into account
during the rotor design and can be proven by the
installation of the machine on a very stiff test field floor
during the factory acceptance test.
For similar conditions at the final installation on site,
regarding the foundation as in the test field, the machine
will show an equal rotor dynamic behavior. However, a
more flexible foundation construction at the customer site
will show structural resonances in and close to the speed
range. In general the API distance conditions for the
natural frequency of the bending modes can not be
fulfilled for the whole system consisting of shaft train,
motor- and compressor enclosures and the foundation
structure. But on the critical speed chart for the whole
system the Eigenvalue characteristics of the simplified
model can be recognized. The study of the systems
mode shapes shows additional combination modes of
rotor and foundation structure. These, as well as the
structural mode shapes are only non-critical for the
operation if they are not excited by residual imbalances
and therefore dont lead to excessive vibration
amplitudes on selected relevant structure points. This
can be proven by different imbalance response
calculations. If the results show only a very small modal
amplification and limited modal coupling between the
pure rotor lateral vibrations and the structure
movements, then the shaft train can be operated on a
stiff as well as on a specially designed softer foundation.
In this paper it was shown, that despite the physically
necessary reversal of some of the originally required

criteria, a safe system operation is possible. This is also


in harmony with practical experience that operations on
softer foundations made of concrete or steel are, in most
cases, without problems if special design criteria are
adhered to.
VI. REFERENCES
[1]
[2]
[3]
[4]
[5]

API 546 - Brushless Synchronous Machines - 500


KVA and Larger
API 617 Centrifugal Compressors for Petroleum,
Chemical and Gas Service Industry
API 684 Tutorial on API Standard Paragraphs
covering Rotordynamics and Balancing
DIN 4024 Machine foundations
DIN ISO 10816 Mechanical vibration,
Evaluation of machine vibration by measurements
on non-rotating parts
VII. CURRICULAE VITA

Dr.-Ing. Horst Kuemmlee is senior engineer and head


of the R&D Department for large rotating electrical
machines in Siemens Dynamowerk, Berlin. He is responsible for basic mechanical and electrical design, rotor
dynamics, simulation of electro-mechanical systems,
measurement procedures and systems and standardization. He has worked at Dynamowerk since 1985 as a
research engineer, head of the special machines order
processing and design department. Dr. Kmmlee graduated from the Technical University Berlin with a Dipl.Ing. degree in Mechanical Engineering (1980) and received the degree of Dr.-Ing. for his work on hyperelastic
coupling elements and dampers (1985). He is a member
of the API 541 and 546 technical subcommittees, vice
chair of ISO TC108/SC2/WG1 and secretary of the PCIC
international subcommittee.
Dr.-Ing. Guenther Siegl is a senior engineer and key
career expert in the R&D Department for large rotating
electrical machines in Siemens Dynamowerk, Berlin and
is responsible for the basic development of experimental
rotor dynamics and the calculation and simulation of
vibration and acoustic problems of electro-mechanical
systems. He has worked at Dynamowerk since 1992 as
a research engineer. Dr. Siegl graduated from the
Technical University Berlin with a Dipl.-Ing. degree in
Physical Engineering science (1976) and received the
degree of Dr.-Ing. for his work on the stability of rotors
which include laminated cores (1982). Following
university he worked one and a half years as a
development engineer in the AEG large drives factory in
Berlin. From 1983 to the end of 1991 Dr. Siegl was
project manager at IABG Ottobrunn for the field of
system dynamics. He was responsible for program
development in the sector of non-linear multi body
systems and vehicle and rolling contact dynamics.
Dr. Peter Woywode is the head of the Department of
Technology and Projects of Siemens Oil&Gas in
Duisburg. Included within this function is the technical
responsibility for large gas turbine driven compressor
trains. He has worked with Mannesmann Demag and
Siemens since 1992 as a research engineer for CAE and
as the head of the departments of mechanical order
engineering and R&D plant design. Dr. Woywode
graduated from the Technical University of Dresden with

a Dipl.-Ing. degree in Technical Mechanics (1984) and


received the degree of Dr.-Ing. for his work on analysis
of valve casings from the Technical University of
Magdeburg (1989). He is member of the working group
for calculation and design of the IIW.

S-ar putea să vă placă și