Augustus 2007

-1- Copyright 2007 by Dynaflow Engineering B.V.
All rights reserved

DYNAMIC MODULE OF CAESAR II
Is it of any use??
lectures 2007
DYNAFLOW
30
th
August 2007
-2- Copyright 2007 by Dynaflow Engineering B.V. All rights reserved
Why is dynamics not used more
Coade: < 5% of Caesar II license holders also makes use of dynamic module
Dynamical effects are overlooked
Dynamical effects are underestimated
Pipe stress engineers prefer quasi static approach
Dynamic load * 2
Pipe stress engineer feels uncomfortable with dynamics
Function of the dynamical module
Facilitate accurate assessment of the dynamical effects
Effects of unsteady loads on stress and load levels in piping
systems.
When is it used??
During design (avoid fatigue, overloading, large displacements)
Control of Vibration problems (development of mitigation
measures)
Piping Incidents due to dynamic loads
Few examples of what might go wrong!!
Collateral Damage due Dynamic Effects from
Flange Failure
Support Damage (I)
Support Damage (II)
Support Damage (II)
Support Damage (III)
More Incidents due to fluid/gas transients
( small bore piping )
Examples of incidents as a result of fluid transients
Deluge FF System on Jetty when
Tested First Time (I)
Deluge FF System on Jetty when
Tested First Time (II)
Flange Failure
Dynamic loads are classified based on time history
Sustained dynamic loads
Periodic loads of moderate amplitude and long duration (minutes-
days)
- Forces due to pressure oscillations in pump/compressor
suction or discharge systems
- Imposed oscillating displacements at pump/compressor nozzle
connections.
- Oscillating forces originating from flow fluctuations/instabilities
Transient, intermittent dynamic loads
Relati vely large forces of relati vely short duration (seconds)
- Slug Loads
- Relief Loads
- Waterhammer, surge loads
- Earthquake
Effect of dynamic loads
Vibration with a periodicity equal to the excitation mechanism
Vibration amplitude depends on separation between excitation
frequency and natural frequency of the piping system
Failure mechanism: High cycle fatigue
Transient dynamic loads
All natural mode shapes and natural frequencies of the piping system
are affected.
Response may show large amplitude vibrations of short duration.
Failure mechanism:
Excessi ve support loads
Accumulation of strain, low or high cycle fatigue
Required data for dynamic analysis
Dynamic model of piping system
Good distribution of lumped masses, additional nodes
Sufficient number of DOF
- For lumped mass (FE) type calculations typically only the
first 33% of the modes shapes are accurate.
Accurate boundary conditions
Magnitude of the excitation forces
Mechanical
Fluid mechanical
- Slug load calculation (manual??)
- Acoustical simulation (pulsation study)
- Waterhammer (surge) load calculation, simulation or
manual
(Measurement results for bench marking the simulations)
Assessment of dynamic results
Material Fatigue data,
ASME B&PV section VIII div 2 appendix 5
API 579
AD Merkblatter
BS-5500
EN-13445
Vibration Assessment conform VDI 3842
Typical Vibration Level Limits conform VDI 3842, Vibrations in Piping Systems
1.0
10.0
100.0
1000.0
1 10 100 1000
Frequency [Hz.]
V
i
b
r
a
t
i
o
n

V
e
l
o
c
i
t
y

[
m
m
/
s
]

R
M
S
design marginal correction danger
Three Options for Dynamic Response Analysis
in Caesar II
1
2
3
0
Harmonic analysis
Application
Sustained vibrations (compressors & pumps)
Input
Periodic Loads can be applied at any node in the system
Per load case one excitation frequency
Many load cases are possible
Solution method
Since response frequency equals excitation frequency solution
procedure is quasi static (fast)
Output
Stress, displacement and load amplitude per frequency
Spectrum Analysis
Application
Transient vibrations: slug, waterhammer, relief
Input
Load time histories at many nodes in the system
Calculation method
Load time histories are translated into response spectra (time
phase between different loads is lost)
Natural frequencies and mode shapes are calculated
Response for each mode shape is determined
Mode shape responses are combined into a final system
response
Output
Maximum stress, largest modal contribution
Maximum loads, largest modal contribution
Maximum displacements
Response Spectrum Generation
Response Spectrum Generation
DLF Related to 4 harmonic cycles
DLF may grow > 2 if time history contains only a limited amount of consistent
periodicity
Time History Analysis
Application
Transient vibrations: slug, waterhammer, relief
Input
Load time histories at many nodes in the system
Calculation method
Load time histories are maintained (time phase between different loads is
conserved)
Natural frequencies and mode shapes are calculated
Response of each mode shape is determined
Mode shape responses are combined into a final system response
Output
Maximum stress, actual stress at user defined times (snapshots)
Maximum loads, actual load at user defined times
Maximum displacements, actual displacement at user defined times
By combination of several runs a time history of stress, loads
displacements can be composed
Missing Mass correction
Only limited number of mode shapes are extracted and participating in
the response
Only a fraction (preferably close to 100%) of the system mass is
participating
Only a fraction of the total excitation force is participating
The missing force fraction is calculated and applied statically after
multiplication by the largest DLF value above the frequency cut-off.
Important when large axial loads are applied (axial mode shapes have
relatively high natural frequencies.
Spectrum versus Time History
Spectrum
Frequency domain
Conservati ve results: only maximum response per mode shape is
calculated and combined in system response
Mode making maximum contribution is identified (advantage for
vibration control)
Only maxima are calculated, time phase is lost
Number of participating modes is finite (missing mass)
Time history
Time domain
More accurate results: time history response per mode is
conserved and time phasing between model maxima is
maintained during combination.
Load, stress time histories (enables fatigue assessment by means
of cumulati ve damage, counting of cycles)
Graphical response is possible
Modal info is not available
Calculation is memory intensi ve (limits simulation duration,
number of participating modes, time step resolution)
Number of participating modes is finite (missing mass)
Quasi-Static Approach
Dynamic Load amplitude (maximum) * DLF=2 as static load
Quasi static approach is simple and fast
Quasi static approach works when there is only one dominant mode
shape that is excited
Quasi static approach focuses on loads (dynamic response is not
considered), i.e. solution by change/elimination of modes is not possible.
Best Method for Transient (Impact) Loads Analysis
Quasi static
Simple but only if there is one dominant mode shape
Spectrum
Time phase between several impacts on one system is lost (e.g.
slug hitting consecuti ve elbows)
Conservati ve but output provides clues for problem solving.
Mode with largest contribution is identified.
Time history
Exact, timing relation between impacts is maintained (slug,
waterhammer)
Clues for diagnosis are less obvious
For transient loads a combination of spectrum and time history runs
provides the best opportunities
Protective Measures
Control of mechanical natural frequencies of the piping system
in relation to the excitation frequency
Support functions and support stiffness (in general high
stiffness)
Analysis accuracy is increased if support structure is included
in the model
Transient dynamic loads
Control of support and nozzle loads
Support flexibility is sometimes useful
Elimination of damaging mode shapes
Protecti ve measures for dynamics may be conflicting with statics
Example 1
Problem
High vibration level in compressor suction piping
Steps to solution
Vibration Measurements, identification of main contributions in
frequency domain
Verification of acoustical natural frequencies of piping system
(acoustical resonance)
Verification of mechanical natural frequencies (mechanical
resonance)
Identification of source of vibration problem
Modification proposal
Compressor Location
Steel Supporting (I)
Steel Supporting (II)
Compressor Layout
Step 1. Vibration Measurements and Compressor
Harmonics
0.00
20.00
40.00
60.00
80.00
100.00
120.00
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Frequency (Hz)
A
m
p
l
i
t
u
d
e

(
d
B
)
66 Hz
99 Hz
49 Hz
33 Hz
83 Hz
16 Hz
Intermediate Conclusion from Step 1
Vibrations are at compressor harmonics
Vibrations must be result of:
Acoustical resonance
or
Mechanical resonance
or
High pulsation forces without resonance (compressor bottle
sizing problem)
Step 2. Acoustical Natural Frequencies & Compressor
Harmonics (Search for acoustical resonance)
0
50
100
150
200
250
10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00
Frequency (Hz)
A
m
p
l
i
t
u
d
e
16 Hz
Purple vertical lines represent compressor harmonics
Intermediate Conclusion from Step 2
Maybe near to resonance condition at first compressor harmonic (16.5 Hz.).
No further acoustical resonance
Vibration peak at 16.5 Hz, most probably is due high shaking forces as a result
of near resonant condition.
The other vibration peaks must be the result of:
Mechanical resonance
or
High pulsation forces without resonance (compressor bottle sizing
problem)
Step 3. Vibration Measurements & Calculated Mech.
Natural Frequencies (Search for Mechanical Resonance)
0.00
10.00
20.00
30.00
40.00
50.00
60.00
70.00
80.00
90.00
100.00
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Frequency (Hz)
A
m
p
l
i
t
u
d
e

(
d
B
)
66 Hz.
33 Hz
83 Hz
Purple vertical lines represent pi pe system natural frequencies
Conclusion from Step 3 & Identification Cause of
Vibration Problem
Apparently there is mechanical resonance at 33 Hz and 66 Hz and near
mechanical resonance at 83 Hz.
No mechanical resonance condition at the first compressor harmonic (16.5 Hz.)
and at 49 Hz. and 99 Hz.
The high vibration levels 33 Hz, 66 Hz and 83 Hz are of mechanical nature.
The high vibration level at 16.5 Hz most probably is an acoustical resonance
problem.
The high vibration level at 49 Hz and 99 Hz. must be the result of:
High pulsation forces without resonance (compressor bottle sizing
problem)
Examination of Mechanical Behavior
66 Hz. Mode Shape
Large amplitude movement
in suction manifold
Step 4. Modifications
The high vibration levels 33 Hz, 66 Hz and 83 Hz are of mechanical nature
and need a mechanical solution
Better supporting
Improved support stiffness
The high vibration level at 16.5 Hz is due to acoustical resonance and
needs an acoustical solution, I.e. different bottles and/or orifice plates to
introduce more damping
The high vibration level at 49 Hz and 99 Hz. are the result of high pulsation
forces without resonance and must be resol ved by compressor bottle
(re)sizing.
As Built Supporting Structure of Compressor Manifold
Modified/Improved Supporting Structure of
Compressor Manifold
Modified Structure Implemented & Connected to
Attached Piping
Conclusion from Example 1
Compressor vibration problems are of a mixed nature
Part is mechanical
Part is acoustical
Each category requires a different approach and result in different
solutions
Not all vibration problems can be sol ved by mechanical measures.
Example 2
Problem
Failure in cooling pump discharge line (possibly vibration induced)
Steps to solution
No vibration measurements just visual observation
Identification of excitation mechanism thru fluid simulations (pump
trips & start-up and check val ve closures)
Time history mechanical simulation to verify stress levels
Identification of source of vibration problem
Modification proposal
Steady State Volume Flowrate [m
3
/s]
Model of the System
Pump trips at t=1 seconds,
Pump inertia: 8 kgm
2
Time History of Pumps
Check val ve closes in 0.5 second
Time History of Valve
Flow thru checkval ve
Time History of Flow through Valve
Maximum Transient Pressure during pump trip [Barg]
Maximum Transient Pressure
Unbalanced Load Time Histories (I)
Load i n Newton * 10
4
4
Unbalanced Load Time Histories (II)
4
Unbalanced Load Time Histories (III)
4
Unbalanced Load Time Histories (IV)
Unbalanced Load Time Histories available in CAESAR
Conclusion from Example 2:
Alternating stress amplitude of 233 MPa results in stress range of 466 MPa
may be responsible for LCF
Example 3
Problem
Excessi ve pressure in injection line.
Steps to solution
Assumed mechanism entrapped (undrained) fluid propelled by gas at
gas velocity during start-up hits val ve that is cracked open
Identification of source of problem thru simulation
Valve Damage
Model of the System
Pressure Time History at the Ball Valve in Case of
Entrapped Gas at the Valve
Val ve cracked open at t=0
Pressure in Barg
Maximum Transi ent Pressure in trapped gas
0
50
100
150
200
250
300
350
400
450
500
0 100 200 300 400 500 600 700 800 900 1000
Trapped gas volume [Liter at atmospheric pressure]
M
a
x
i
m
u
m

P
r
e
s
s
u
r
e

[
B
a
r
g
]

.
Slug 29 liter, 20 meter Slug 130 liter, 90 meter
Pressure Time History of Entrapped Gas
Conclusions for presentation
Dynamic analysis is an important and sometimes a critical element in
integrity analysis
Many failure modes are to be addressed
Excessi ve loads are to be handled
CAESAR offers several types of dynamic analyses to assist in the
demonstration of integrity:
1. Harmonic Analysis
2. Spectrum Analysis
3. Time History Analysis
Solutions to possible problems are often found by introduction of the right
supporting and/or supporting steel structure
END
Thank you for your attention

Augustus 2007

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Augustus 2007

Hochgeladen von

Copyright:

Verfügbare Formate

-1- Copyright 2007 by Dynaflow Engineering B.V.

All rights reserved

Das könnte Ihnen auch gefallen