Modal Screeninig

Dave Diehl

CAESAR II
Dynamic analysis screening
a qualitative approach

Modal Evalution as a
Dynamic Screening Tool
Introduction Analytical design and evaluation
versus real world systems
Concept Using systems lowest natural
frequency to measure likelihood of failure
Rule Setting your lowest allowed natural
frequency
Application Proper screening in CAESAR II

CAU Express 2012

Modal Screeninig
Dave Diehl

Most new piping design

specifications do not require
dynamic analysis
Seismic analysis is the exception but, even then,
static evaluation may be adequate
But operating systems may suffer from dynamic
response
Typical sources of these dynamic loads include:

Harmonic
Mechanical vibration (connecting equipment)
Flow-induced vibration
Pulsation

Impulse
Hammer loads
Relief valve discharge

Applied load vs. induced load

Due to system timing and load timing, the
maximum response to a changing load may not
equal the response to same magnitude static
load.
Call the static load the applied load.
An induced load is the load magnitude that
would produce the equivalent maximum
dynamic response to the changing load.
The ratio of induced load to the applied load is
the dynamic load factor or DLF.

CAU Express 2012

Modal Screeninig
Dave Diehl

Applied load vs. induced load

0
-0.2

Applied load

-0.4

load

-0.6

time

-0.8
-1
-1.2

Static response

-1.4
-1.6

Maximum
dynamic response

-1.8
-2
0

20

40

60

80

100

120

140

Since deflection is proportional to load, here,

DLF=1.9

Comparing response to
impulse & harmonic loads

CAU Express 2012

The magnitude of impulse loads may be large

but their dynamic amplification (DLF) is limited
to no more than two.

Modal Screeninig
Dave Diehl

Comparing response to
impulse & harmonic loads
The magnitude of impulse loads may be large
but their dynamic amplification (DLF) is limited
to no more than two.
Even small cyclic loads, however, may produce
large dynamic response.

Comparing response to
impulse & harmonic loads
The magnitude of impulse loads may be large
but their dynamic amplification (DLF) is limited
to no more than two.
Even small cyclic loads, however, may produce
large dynamic response.
Eliminating the source of cyclic loads (and
impulse loads for that matter) is useful. But
Attention to this f/n ratio will limit response
to cyclic loads. This is our focus here.

CAU Express 2012

Modal Screeninig
Dave Diehl

Reduce risk of piping failure by

controlling natural frequencies
In many cases it is our dumb luck that our
systems respond to any one of the many
harmonic sources
We can evaluate risk of response by monitoring
system natural frequencies
We can better reduce risk of failure by
increasing system natural frequencies
CAESAR II modal analysis of new designs can
be used to improve system reliability

Qualifying new designs using

modal analysis
Modal analysis calculates the natural frequencies
and mode shapes
The lowest natural frequency (LNF) indicates the
likelihood of piping vibration.
Typical acceptance criteria: keep LNF > 3 to 5 Hz
Based on experience
This method and acceptance criteria is not
reflected in traditional piping codes and standards

CAU Express 2012

Modal Screeninig
Dave Diehl

But how accurate can we be?

Quoting a presentation on API 618 Fifth Edition*

Inaccuracy of Mechanical Natural Frequency

Calculations Field experience shows that the
accuracy of predicting mechanical natural frequencies
in piping systems is suspect even under the best of
circumstances. Error margins of 20 percent are
obtainable only where accurate boundary conditions
are known and extensive, detailed modeling of both
the piping system and the supporting structure is
performed.

* Presentation at the 2004 Gas Machinery Conference entitled: From Understanding the Pulsation &
Vibration Control concepts in the New API 618 Fifth Edition K. Atkins (Engineering Dynamics Inc),
A. Pyle (Shell Global Solutions) & J. Tison (Engineering Dynamics Inc.)

But how accurate can we be?

Quoting

Other items which influence the accuracy of these

models are:
Uncertainty of stiffness (six degrees of freedom) of
clamps/hold downs and supporting structure
Difficulty in accurately predicting coefficients of friction
Nonlinear effects (e.g., gaps closing due to thermal growth)
Uncertainty of as-built piping layout and dimensions,
weights, etc.
Difficulty and complexity of modeling rack support structure
Uncertainties in soil stiffness effects on concrete piers
Settling of supports resulting in loss of piping contact

CAU Express 2012

Modal Screeninig
Dave Diehl

Modal Analysis Acceptance

Criteria
API 618 compressor piping is not our sole focus
here
The acceptance criteria is set based on
managing risk.
DNV-RP-D101

A typical system supported in accordance with a good

pipe support standard should result in a lowest natural
frequency not less than 4 to 5 Hz

DNV-RP-D101
Det Norske Veritas Recommended Practice
Structural Analysis of Piping Systems
2.2.7.1 Modal Analysis

CAU Express 2012

Finding the piping systems natural frequencies are

essential in order to determine the size of Dynamic
Load Factors (DLF) and to determine the correct
pipe-support spacing in order to avoid detrimental
vibrations caused by internal flow, pressure
transients, and vortex shedding oscillations from wind
or sea currents passing over the piping.

Modal Screeninig
Dave Diehl

DNV-RP-D101

2.2.7.1 Modal Analysis

Modal analysis of a static model is usually not time

consuming and should therefore be carried out to
determine the lowest natural frequency of the system.
A typical system supported in accordance with a good
pipe support standard should result in a lowest
natural frequency not less than 4 to 5 Hz.

DNV-RP-D101

3.12 Fatigue Calculations

3.12.1 General
A modal analysis of all piping systems should be
performed and it is desirable and a common practice
to keep the piping systems natural frequency above
4Hz to mitigate circumstances where fatigue can be
induced by low frequencies of vibration. For situations
where large expansion loops are required to absorb
large movements, natural frequencies above 4Hz
may be difficult to achieve.

CAU Express 2012

Modal Screeninig
Dave Diehl

EI Guidelines*

Technical Module T9 Specialist Predictive

Techniques

See T9.2: Structural Finite Element Analysis

Technical Module T10 Main Line corrective

Actions
See T10.3.2.1&2: De-tuning pipework
Excitation frequency should not be within 20% of
systems natural frequencies

LOF (Likelihood of Failure) score based on

fundamental frequency

* Energy Institutes Guidelines for the Avoidance of Vibration Induced Fatigue Failure in Process Pipework

Piping Support Arrangement

T2 Quantitative Main Line LOF Assessment
T2.2.3.3 Determining Support Arrangement
Support Arrangement designation for Typical
Fundamental Natural Frequencies (Table T2-1)

Stiff 14-16 Hz
Medium stiff 7 Hz
Medium 4 Hz
Flexible 1 Hz

CAU Express 2012

Flexible support arrangements have higher LOF

Modal Screeninig
Dave Diehl

API 618 5th Edition

Reciprocating Compressors for Petroleum,
Chemical, and Gas Industry Services provides a
Separation Margin Guideline
Again, working off the f/n ratio

f will be based on running speed of the compressor

n is the fundamental frequency or the Minimum
Mechanical Natural Frequency (Minimum MNF) of the
piping system

API 618 5th Edition

The minimum mechanical natural frequency

should be 20% above the 2nd harmonic of the
running speed

Illustration and data provided by Beta

Machinery Analysis Application Note 1b:
Implications of the New API 618 (5th Edition)
for Packagers, OEMs and End Users.

CAU Express 2012

10

Modal Screeninig
Dave Diehl

API 618 5th Edition

Vibration criteria for these compressor systems
is very specific and severe, and are included in
the piping design
We will not use this acceptance criterion of 20%
separation from 2nd harmonic for general piping
systems

Modal analysis acceptance

criteria
Piping systems which produce a high cycle
fatigue failure typically have a calculated lowest
natural frequency (LNF) of less than 1Hz
Again, published guidelines suggest:
LNF > 4 Hz
LNF acceptance criteria at higher frequencies
will impact other aspects of design, e.g. system
response to thermal strain.
Higher LNF criteria will incur higher costs.

22

CAU Express 2012

11

Modal Screeninig
Dave Diehl

CAESAR II modal analysis

Can be applied to all existing CAESAR II models
Quick and easy analysis
A good indicator of robust design

An example
6 inch standard pipe
Water filled
2 inches of insulation
No guides

24

CAU Express 2012

12

Modal Screeninig
Dave Diehl

An example

Modal analysis shows LNF = 3.28 Hz

25

An example one guide added

Modal analysis shows LNF = 4.91 Hz

26

CAU Express 2012

13

Modal Screeninig
Dave Diehl

An example two guides

Modal analysis shows LNF = 8.40 Hz

27

Example: Heater discharge piping

Existing system with a

history of vibration
problems

No guides or stops

LNF 1.2 Hz

Guides and stops added

and LNF increased to
3.7 Hz

28

CAU Express 2012

14

Modal Screeninig
Dave Diehl

Achieving a more realistic

response

Mass Distribution
In the earlier example extra nodes were added so that
maximum spacing between nodes was less than 1
foot per nominal inch (using the Break command)
Natural frequencies calculation used the Consistent
Mass distribution

Achieving a more realistic

response

Mass Distribution

The numbers may be

small here, but significant.

Compare that (fine, consistent) first mode of 3.282 Hz

with:
A coarse model, lumped mass value of 3.674 Hz
A coarse model, consistent mass value of 3.560 Hz

coarse

fine

CAU Express 2012

15

Modal Screeninig
Dave Diehl

Achieving a more realistic

response

Mass distribution review

Coarse distribution misses a mode (shape) and
Locks up more mass at restraints

Coarse&Lumped Coarse&Consistent
Mode Freq(Hz) Mode Freq(Hz)
1
1
3.674
3.56
2
2
5.517
6.017
3
3
9.344
9.25
4
4
11.795
11.425
5
6
7
8
9
10
11

13.926
16.585
17.294
19.299
21.381
23.932
34.174

5
6
7
8
9
10
11

12.923
14.62
17.196
18.515
23.174
26.63
36.036

Fine&Lumped
Fine&Consistent
Mode Freq(Hz) Mode Freq(Hz)
1
1
3.287
3.282
2
2
5.864
5.861
3
3
9.501
9.471
4
4
10.923
10.886
5
5
11.758
11.709
6
6
13.86
13.857
7
7
16.704
16.721
8
8
20.147
20.262
9
9
22
22.02
10
10
24.297
24.433
11
11
28.526
28.776
12
12
34.719
35.098

Mass Distribution Review

CAU Express 2012

16

Modal Screeninig
Dave Diehl

Achieving a more realistic

response
Support stiffness affects the frequencies
Is it rigid?

Achieving a more realistic

response

CAU Express 2012

Support stiffness affects the frequencies

17

Modal Screeninig
Dave Diehl

Achieving a more realistic

response
Friction may play a role
Is this loop free to wag?

Achieving a more realistic

response

Friction may play a role

A quick look at mode shapes helps here

CAU Express 2012

18

Modal Screeninig
Dave Diehl

Achieving a more realistic

response

Friction may play a role

FRIC=100
FRIC=1000
FRIC=0
Guided
Mode# Freq(Hz) Mode# Freq(Hz) Mode# Freq(Hz) Mode# Freq(Hz)
1
1.071
2
2.202
3
5.905
4
11.576
1
11.495
1
11.531
1
11.519
5
13.846
2
11.992
2
12.122
2
12.112
3
14.053
3
13.911
6
14.468
4
14.65
3
14.566
7
15.667
5
15.617
4
16.151
4
15.6
5
18.012
5
17.912
8
19.572
6
19.227
6
20.362
6
20.005
9
21.906
7
21.595
7
22.914
7
21.444
8
22.325
10
23.781
9
24.133
8
24.172
8
24.207
11
25.683
10
25.751
9
25.94
9
25.975
12
26.609
11
26.277
10
26.431
10
26.454
13
34.306
12
34.347
11
34.375
11
34.379

LNF Rule Applied Statistics

Source: Nigel Marsh of Peak Engineering Ltd.

CAU Express 2012

19

Modal Screeninig
Dave Diehl

Conclusions
Simply designing piping systems for thermal and
deadweight loads ignores dynamic response
common in operating systems.
Small cyclic loads can cause significant
response when a systems natural frequency
coincides with a harmonic forcing frequency.
Monitoring the systems fundamental frequency
is a good screening tool to evaluate risk of these
environmental loads

Conclusions
Keeping the systems first mode of vibration
the fundamental frequency above ~5 Hertz
is a useful guide in new system design
Its easy to apply and can improve system
reliability
Higher values may be necessary in specific
applications e.g. compressor piping (API 618)
and offshore systems
These requirements must compete with thermal
strain limits on equipment

CAU Express 2012

20

Modal Screeninig
Dave Diehl

Thank You for Your Attention

CAU Express 2012

Any Questions?

21