Documentos de Académico
Documentos de Profesional
Documentos de Cultura
Component Modeling
• Nozzle/Vessel models in AutoPIPE
• Nozzle/Vessel Stresses Using WINNOZL
• Expansion Joints
• Jacketed Piping
• Valves
• Reducers
• Flanged elbows, miters and bends
• Tees
• Frame
Nonlinear Analysis
• Analysis assumptions (linear and non-linear)
• Support non-linearity’s
• Load sequencing
• Non-linear occasional loads
• Result interpretation
Dynamics
• Analysis assumptions
• Analysis algorithms
• Frequency and Mode Shapes
• Response spectrum analysis
• Spectrum Enveloping
• Static Correction
• Harmonic analysis
• Force spectrum analysis
• Seismic Anchor Movement Analysis
• Time history analysis
• Dynamic Load Factor
Fluid Transients
• Water hammer analysis
• Steam relief valve analysis
• Slug flow analysis
Miscellaneous
• Buried pipe analysis
• Submerged piping and wave loads
• Special Modeling Cases
• Open discussion
Bentley Confidential 1
Vessel and Nozzle Modeling Considerations
• Nozzles of equipment like pump and compressors are modeled as
Anchor
• Other than finite element model, all methods are approximate and only
valid for a specified range of nozzles. Finite element method will take
longer time and is impractical for everyday design.
• Nozzle option only models the local effects of nozzle and vessel. The
vessel must be modeled separately.
Bentley Confidential 2
Pressure Thrust on Vessel/Nozzle Junctions.
Pressure Thrust
A
B
Anchor or Axial
Support
Figure 1
Where:
Bentley Confidential 3
Pressure Thrust (P*A)
Figure 2
The force F is in equilibrium with the two spring forces F1 and F2:
F = F1 + F2 (1)
K1 = F1 / δ1
K2 = F2 / δ2
So:
F = δ1 * K1 + δ2 * K2
So:
F = δ * ( K1 + K2 )
δ = F / ( K1 + K2 )
If the piping system on the other side of the applied load (Spring 2) is stiff, for
example due to an anchor, then pressure thrust will be absorbed by the
anchor. Thus, the nozzle will experience very little direct axial stress. This
can be seen from equation 2. Note that a greater K2 results in a lower thrust
force F1. Therefore, in this case including all of the pressure thrust into
analysis will be conservative. However if the pipe shown by spring 2 is
flexible (maybe an expansion loop or small diameter pipe with bends) then
the nozzle will see more of the force due to the pressure thrust. Therefore it
is appropriate to analyze the local vessel/nozzle stresses due to most of the
pressure thrust load.
Bentley Confidential 4
Pressure Thrust Guidelines
WinNOZL WRC368 within its geometric limits provides a good design check
of pressure stress levels which includes the full thrust load otherwise use
FEA analysis to obtain more accurate combined stresses.
If the full pressure thrust is acting on the vessel/nozzle junction e.g. nozzle
with a blind flange then FEA would generally be the most accurate analysis
tool to evaluate.
Note: FEA programs have limitations due to the accuracy of the type of
elements used e.g. many programs use thin shell elements which do not
capture transverse shear effects of thick shell elements.
Bentley Confidential 5
Modeling Vessel
MODEL 1 (Sample Model: Vessel1.dat )
• Simple to model
• Hard to predict movements due to other loads like wind and earthquake
Bentley Confidential 6
MODEL 2 (Sample Model: Vessel2.dat )
• Requires more steps for the modeling but more accurate stiffness
model.
Bentley Confidential 7
Nozzle Stresses Using WinNOZL
2 and PD5500.
Bentley Confidential 8
Expansion Joints
• Used to absorb thermal expansion to reduce movement of pipe at
equipment
• Tie rods can be modeled using tie link. This is a simplified model
and does not capture bending resistance due to locking of rods
Bentley Confidential 9
Jacketed Piping
• Carrier pipe and jacket modeled as two separate segments with
different pipe identifiers e.g. Jacket6 and carrier8
• If both carrier and jacket are liquid filled then adjust jacket SG.
Bentley Confidential 10
MODEL 1 – Beam connected model
(Sample model: jacket1.dat)
• Flanged ends are modeled as rigid beams between a point on the
carrier segment and the jacket segment
STEPS
1. Open Jacket_1A
3. Select range C02 to C12, and also branch B16 to B04 so highlighted
red.
Bentley Confidential 11
5. Select / Clear
6. Click cursor on point C02 (so point name = RED) and Edit/Paste ,
uncheck the “connect to select points” then click Ok (This creates
jacket segments D and E)
Bentley Confidential 12
14. Click on point E07 (previously a reducer on carrier pipe) and
Modify/convert point to run
15. Delete the additional flanges at C02, D00 and E10, Select / Flanges,
Press Delete key
The Jacket is now connected to the carrier at C02, B04 and C12 using
rigid beams.
Bentley Confidential 13
Now add spacer supports between Jacket and Carrier
20. Click on E01, connected point = C03, Insert / Support > Guide, gaps =
0 and friction = 0.1
21. Repeat for E03 to C05, E05 to C06, E08 to C09
Bentley Confidential 14
The model should now look like Jacket_1b below
• The point name on the carrier segment is same as the point name on
the jacket segment (Similar to modeling a branch connection)
Bentley Confidential 15
STEPS
1. Open Jacket_2A
3. Note: Previously we had clicked on point C12 and Insert /Run before
C12 , length = 0.05’ (this is a small run C11 close to the valve since we
cannot insert a segment at start of a Valve)
4. Select range C02 to C11, and also branch B16 to B04 so highlighted
red. Hint: Select point C11 first, shift and click on C02
6. Select / Clear
7. Click cursor on point C02 (so point name = RED) and Edit/Paste ,
“connect to select points” = checked then click Ok (This creates jacket
segments D and E, connected at C02)
Bentley Confidential 16
9. Select / Clear
10. Select Segment E and Modify/Pipe properties over Range, and select
pipe identifier = Jacket8
11. Select / Clear
12. Select Segment D and Modify/Pipe properties over Range, and select
pipe identifier = Jacket6
Bentley Confidential 17
13. Click on point E07 (previously a reducer on carrier pipe) and
Modify/convert point to run
14. Delete the additional flanges at C02 and D00 Select / Flanges, Press
Delete key
Bentley Confidential 18
17. Similarly click on point E10 (or F5 goto point) and modify/ Point and
rename to C11
The Jacket is now connected to the carrier at C02, B04 and C11 using
tee segment connections
Now change all the Welding tees at C02, B04 and C11 to Tee type =
Other with SIF = 1.0.
18. This can easily be done in the Tee Input Grid using multiple Select of
“Type “ cell and use CTRL key to change to OTHER and then press
CTRL + Enter
Bentley Confidential 19
The model should now look like Jacket_2b below
Bentley Confidential 20
Valves
• Construction of valves makes it stiffer than pipe. The stiffness of valve
cannot be estimated without a detailed finite element analysis of the
valve.
• AutoPIPE models a valve as 100 times stiffer than the pipe at starting
point of the valve. This is achieved by increasing the modulus of
elasticity of the pipe.
ANGLE VALVES
• Angle valves and Relief valve can be modeled using valve component
and specifying offsets of the far end from the valve point
• The valve is defined as a tilted valve. The exact form of the valve is not
important as long as end points are defined at correct location
VALVE OPERATORS
• Heavy operator far from valve center of gravity can induce significant
force into the piping system in seismic event.
• Exact model of the operator requires a rigid beam with weight at the
free end. Modeling is complex and may not be necessary
Bentley Confidential 21
Reducers
• Used at locations where pipe size changes
• Cone angle of the reducer (for SIF) is calculated based on the full
length of the reducer.
• This model captures the exact axial behavior. The bending behavior is
approximate
Bentley Confidential 22
Bends and Elbows
• Elbows tend to become more flexible with increase in plane bending
due to ovaling – Von Karman effect
• Internal pressure also stiffens the elbow. The change is not significant
for most operating pressure ranges.
• ASME codes require different flexibility factor for elbows flanged at one
and both ends
• Miters are modeled as bends with modified flexibility factors and SIF's
Bentley Confidential 23
Tees
• Tee component is modeled as single point connecting, three pipes
• End points of a tee components are not modeled and the change in
thickness/diameter is ignored
• Some codes (B31.3) do not provide any guidelines for use of SIF's at
tees for sustained loading and occasional loading
• SIF's for tees are empirical based on fatigue test a series of simple full
size tees. SIF's for other types of tees (reduced) are derived from this
study.
• Piping codes do not specify SIF's for connections like laterals, Y's and
crosses. User must specify SIF's for code compliance. Bonney forge
does provide a published technical paper on calculating SIF’s for
lateralets.
Bentley Confidential 24
FRAMES
• To model racks and pipe supports
Options
• Add
• Delete
• Modify
Features
• Beta angle
• Rigid lengths
• End releases
Bentley Confidential 25
Analysis Assumptions
• Finite element Analysis (stiffness method)
• Elastic response - small deformation theory (1st order only). e.g. One
rule of thumb: Check that the maximum slope angle in radians of the deformed pipe
= approx. sin(slope angle) then the solution should be ok.
Bentley Confidential 26
Nonlinear Analysis
• Supports with Gaps
Bentley Confidential 27
Load Sequencing & Interpretation
• The end state of the piping system is always the same i.e. Operating
case GR + T1 + P1 results are the same if use load sequence Gr -> T1
-> P1 or Gr -> P1 -> T1
Bentley Confidential 28
Introduction to Dynamic Analysis
Dynamic loading tends to increase the response of the
structure beyond the response obtained if same load is applied
statically.
• Linear supports
• Gaps are ignored and supports are assumed linear
• Friction is ignored (frictionless)
• No yielding of soil
• Pipe material remains elastic
Bentley Confidential 29
Static correction can be used to capture effect of mass not
captured by eigenvalue analysis
• Missing Mass Correction
• Zero Period Acceleration (ZPA)
Bentley Confidential 30
Use ZPA method to capture
correct support load
Static Dynamic
Load varies with time or frequency e.g.
earthquake, fluid transient like water hammer,
Steady State, slow applied loads not varying
relief valve discharging (jet effect), mechanical
with time e.g. deadweight, temperature, wind
or fluid induced vibration, slugging flow, blast
(generally with gust factor)
loads, vortex shedding due to wave or wind
loading etc
Piping system is in equilibrium i.e. sum of
Piping system is not in equilibrium
forces and moments = 0
Piping system remains at rest due to balanced Piping system has unbalanced forces and
forces system moves due mass x acceleration
Support and Anchor reactions may be higher or
Support and Anchor reactions are equal to the
lower than to the distribution of applied
distribution of applied static loads
dynamic loads
A static load is simply a dynamic load with a long duration so the piping
system can fully respond to it.
Bentley Confidential 31
Natural Frequency and Mode Shapes
(Sample model: apham1.dat)
Natural frequency and mode shapes are property of the
structure and depend on the mass and elasticity.
They describe the tendency of the structure to vibrate
when subjected to dynamic loading.
Number of frequencies and mode shapes with which a
structure can vibrate depends on the number of mass
degrees of freedom in the structure.
In a lumped mass modal used by AutoPIPE, each node
has three mass degrees of freedom. Nodes with eccentric
weight can have up to six mass degrees of freedom
Mode shapes describe relative displacement of the
structure (mass normalized)
Bentley Confidential 32
Modal Period = 1 / natural frequency
Fundamental frequency or free vibration is based on the
1st natural frequency.
Bentley Confidential 33
Response Spectrum Analysis
(Sample model: response1.dat)
Bentley Confidential 34
Maximum response for each oscillator
can occur at different times during the
earthquake.
Bentley Confidential 35
The amplitude with which a certain mode in the structure will
be excited by an earthquake is be determined by the response
of a single degree of freedom oscillator at that frequency
(response spectrum).
Bentley Confidential 36
RESPONSE SPECTRUM PLOT
Maximum Response for each single DOF oscillator
for different damping ratios
Bentley Confidential 37
Nuclear Regulatory guide 1.60 (published in Dec 1973) describes the
requirements for generating a design seismic response spectra for nuclear
power plants.
Since response spectrum is based on SDOF system, the response can be
approximately represented by a sine function.
Displacement= A sin(wt+phase)
Velocity = Aw.cos(wt+phase)
Acceleration = -Aw2.sin(wt+phase)
Due to this fact, the displacement, acceleration and velocity spectra can be
plotted on a single log-log plot also called tripartite plot. The plot provided in
AutoPIPE notes corresponds to NRC displacement spectra included in the
AutoPIPE directory. All these spectra are normalized to a maximum ground
acceleration of 1g.
Although the response appears to be a smooth line, this is not actually the
case. As you discretize the SDOF frequency (more points on x-axis), the
curve will have an erratic shape with several peaks and ‘valleys’.
Unfortunately using such spectrum at a valley location will be un-
conservative. This is due to the fact that the structural frequency will change
during an earthquake. This change is caused by material yield or support
failure. A design response spectrum like NRC ones provided in AutoPIPE
are the result of averaging multiple earthquake spectra and enveloping the
resulting spectrum. NRC has a procedure to generate a design spectrum
from a single earthquake record. The procedure involves ignoring the valley
points.
Bentley Confidential 38
Response spectrum is more suited for design than time history analysis for
the reason mentioned above. The only disadvantage of response spectrum is
that it is limited to a linear system. For very important structure, nonlinear
time history analysis is sometimes performed. For such cases, 15 or 20 such
time history cases are performed and the time histories themselves are
generated from a design response spectrum. These are called response
spectrum compatible earthquakes. Each response spectrum can correspond
to infinite number of time histories as the phase information is lost in a
response spectrum.
The UBC and more recently the IBC has a procedure to generate a design
response spectrum based earthquake geographic zone, soil conditions and
other factors. The procedure does not involve any actual earthquake time
records.
Bentley Confidential 39
NRC Guide 1.60 RESPONSE SPECTRUM PLOT
Maximum accelerations, velocities and displacements
60
33Hz
Bentley Confidential 40
Typical Response Spectrum Plots
North-OBE Response
X-Dir
Period (s)
0.0100 0.1000 1.0000 10.0000
1.0000
Accel (g)
0.1000
South-OBE Response
Y-Dir
Period (s)
0.0100 0.1000 1.0000 10.0000
1.0000 Accel (g)
0.1000
Bentley Confidential 41
East-OBE Response Z-Dir
Period (s)
0.0100 0.1000 1.0000 10.0000
1.0000
Accel (g)
0.1000
Bentley Confidential 42
Spectrum Enveloping
¾ Different points in the system may be subjected to different response
spectra. For example, supports attached to different floors of a building
will have different excitation.
¾ The logical approach is to use the worst spectrum for all points in the
model and apply a single envelope spectrum to all supported points.
Bentley Confidential 43
Static Correction
To capture total dynamic response of a structure all possible
modes should be captured.
For a large system this is impractical due to large number of
modes.
High frequency modes that do not contribute significantly to
the final response can be approximated using static
correction methods.
Static correction is approximate and may not predict local
response
Zero Period Acceleration (ZPA)
The structure is subjected to the peak ground acceleration.
Entire mass of the structure is considered.
Static response is obtained for the structure.
Larger of the static and dynamic response is used.
Available all Dynamic Analyses
Missing Mass Correction
The amount of mass captured by all extracted modes is
subtracted from the total mass.
The structure is subjected to acceleration equal to the cut-off
frequency. Only the uncaptured mass is considered.
Static response is obtained for the structure.
The static response is combined with the dynamic response
using the specified combination method.
Available for harmonic, response & force spectrum only.
Bentley Confidential 44
Harmonic Analysis
(Sample model : hrmexp.dat)
Bentley Confidential 45
Mechanical Vibration
Bentley Confidential 46
Force Spectrum Analysis
Used to analyze response of a system due to short duration,
impulsive loads
• Water hammer
• Relief valve blow down, etc.
Bentley Confidential 47
Seismic Anchor Movement Analysis
Predicts stresses on piping system due to differential
movement of buildings or floors during a seismic event.
PROCEDURE
1. Piping system is solved for SAM in global X direction
for Phase 1
2. Piping system is solved for SAM in global X direction
for each remaining phase
3. Maximum response in X direction is calculated using
absolute sum.
4. Step 1, 2 and 3 are repeated for global Y and Z
directions
Bentley Confidential 48
Time History
(Sample model : apham1.dat)
where,
m= Mass matrix of the structure
u (t ) = Structural response (displacement) matrix
(function of time)
u&(t ) = Structural response (velocity) matrix (function of
time)
u&&(t ) = Structural response (acceleration) matrix
(function of time)
c = Structural damping matrix
k = Structural stiffness matrix
r (t ) = is a vector of arbitrary time-varying loads or of
effective loads which result from ground motion.
Bentley Confidential 49
Assumptions and Limitations:
¾ A single damping ratio is assumed for all modes.
¾ The same time step will be used for all modes.
¾ Time history files do not support displacement or velocity
input only accelerations or force.
¾ All supports are considered linear for this analysis i.e.
nonlinear behavior of supports is not considered. This is
consistent with assumption made for other dynamic
analyses.
¾ Acceleration inputs will be applied to all free degrees of
freedom with mass. The user cannot input accelerations
for specific points. The user will not be able to prescribe
displacement, acceleration, or velocity of supported
points. However support displacements can be modeled
by entering forces equal to the displacement times
support stiffness.
¾ Plotting of time history response at individual points is not
available.
Bentley Confidential 50
Dynamic Load Factor (DLF)
• The maximum dynamic response to an impact load is 2 x the static
load response. The ratio of dynamic to static response is termed the
dynamic load factor or DLF.
• Common impact loads on piping systems include :
• Relief Valve Discharge
• Water or Steam Hammer
• Slug Flow
Time
10
Bentley Confidential 51
B31.1-2004 Relief Valve DLF curve having only single DOF on a rigidly
supported pipeline.
• The low frequency (flexible) response is generally ignored, but the drop
from a DLF of 2 to a conservative 1.1 is set by the opening time of the
valve.
• The valve manufacturer can provide valve opening time then a more
realistic force response spectrum can be generated as shown below.
Bentley Confidential 52
Determine the DLF Factor.
Define time history load TEST.TIH starting at time 0 and load = 89000N
Notice at 20.4 to 510 Hz, the force is about 176000N giving a DLF of about 2.
Bentley Confidential 53
At very high frequency, >2000 Hz, it is 105480 with DLF= 105480 / 89000 = 1.1
The rise time of 0.0004 sec appears unrealistic. If changed to 0.01 sec the force spectrum
becomes:
Time (s)
0 0.01 0.02 0.03 0.04 0.05
100000
90000
80000
70000
60000
Force (N)
50000
40000
30000
20000
10000
Bentley Confidential 54
Test Response
Frequency (Hz)
-100 400 900 1400 1900
160000
100000
Force (N)
80000
60000
40000
Slope determined by overall Duration i.e.
Longer duration = steeper initial curve 20000
Bentley Confidential 55
Question: What are the Participation Factors and the Captured Modal Mass in
the frequency report?
The participation factors are calculated from the product of the mode shape, the mass
matrix and a vector of ones. For mode i the participation factor is calculated as:
The mode shape is mass normalized. The above equation is used three times for X, Y
and Z directions.
The mass participation report illustrates how sensitive each of the piping system’s
modes are to the dynamic loading. High modal participation factors indicate that
the mode is easily excited by the applied dynamic forces. If subsequent
displacement reports indicate high dynamic responses then the modes having high
participation factors must be dampened or eliminated. Once a particular mode is
targeted as being a problem, it may be viewed in the mode shape report, or graphically
via the animated mode shape plots.
--------------------------------------------------------------------------------
APHAM1A_SI_1
09/08/2005 SAMPLE MODEL OF WATER HAMMER BENTLEY
10:58 AM AutoPIPE+8.60 RESULT PAGE 3
--------------------------------------------------------------------------------
F R E Q U E N C I E S
Bentley Confidential 56
Note: Participation factors and captured modal mass apply
only to Response Spectrum and Earthquake Time History
load cases. Captured modal mass for other load
types can be found in the analysis summary sub-report.
A common rule of thumb is to capture at least 75% of the modal mass but the
missing mass correction will capture the remaining modal mass for an accurate
dynamic analysis.
Question: How do we enter response spectra for a given floor, with different
Frequency-Acceleration data for each direction X,Y,Z (The program requests
for scaler multiples for other directions)?
Answer: It is not possible to perform a dynamically independent response spectra
analysis i.e. different response spectra to different parts of the model scheduled for
Q2/2006.
a) This data can be input using AutoPIPE time history dynamic analysis which allows
different loadings at different parts of the piping model at different times.
b) Use static earthquake and graphically select ranges within your model and apply a point
static earthquake scale factor to account for different accelerations at different floors.
Question: What is the maximum DLF expected on an in-line rigid axial support
under time history analysis?
Answer: If it is really rigid, and the pipe is rigid axially, then it is assumed no dynamic
amplification, i.e. DLF=1.0
Question: How can I Estimate the cutoff frequency for my dynamic water
hammer analysis?
Answer: The maximum frequency cutoff can be estimated from SQRT (E/p )/L
where: E = Pipe material modulus of elasticity, p = Pipe material density, L = Length
of a single pipe element in the primary run that is to have accurate stresses computed
due to the passing of the water hammer originated acoustic stress wave. Calculation of
the maximum cutoff frequency for the 45-75 elbow-elbow pair for the 20 ft pipe
lengths is given as follows:
Bentley Confidential 57
= 134.2 Hz.
Then calculate pipe span based on this fundamental frequency (1st mode) for both fixed
and simply supported ends. A typical pipe behaves somewhere between these two
boundary conditions. Ref Roark Table 36 case 1b, 2b
where :
W = weight /unit length (kg/mm)
I = moment of inertia (mm4)
f1 = frequency (Hz)
E = Youngs Modulus (Mpa)
kn = 3.55 (Fixed Ends), 1.57 (simply supported)
Since the dynamic load factor (DLF) drops with increasing frequency, the best solution is
to increase the 1st modal frequency. This is done by increasing the stiffness of the system
e.g. adding supports or increasing support stiffness or reducing its mass e.g. thinner pipe
or move valves closer to support locations.
http://selectservices.bentley.com/en-
US/Support/Support+Tools/TechNotes+and+FAQs/Bentley+AutoPIPE/Index.htm
in particular technote
8274 - Compressor Vibration Method
Bentley Confidential 58
Fluid Transient
(Sample model : apham1.dat)
¾ Used to determine the force-time histories along a single pipeline due to fluid
transient events such as valve closures, pump shutdowns, flow demand changes,
pump startups, air venting from lines, failure of flow or pressure regulators, or pipe
rupture.
¾ For a piping system with branches, the user must select a single main pipeline in
which to generate the force-time history. The effect of branches on the surge
pressure will be ignored.
¾ The shock wave travels back upstream from the valve but the forces are acting
downstream. Consider free body diagram below - the +ve pressure rise at the
Valve results in higher pressure acting at valve end compared to 1st bend end
hence the net force is acting towards the valve (in the direction of the flow). Thus
the shock wave (and pressure rise) propagates back up the pipe resulting in leg
forces acting in the same direction as the flow.
Bentley Confidential 59
Typical pressure wave from a quick acting Valve Closure
Typically two shock waves are generated. A +ve wave on upstream side and
a –ve pressure wave on the downstream side
Pressure (psi)
1000
800
600
P_up
400
P_dow n
200
Time (sec)
0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
-200
Actual rise
Bentley Confidential 60
A Force vs. Time history is generated at each elbow node (Near or Far) and also at
the origin point of the shock wave. The profile looks like:
Force vs Time
Force
where,
L or S, S is preferred
Bentley Confidential 61
Calculated force in
each pipe leg
Bentley Confidential 62
• The reduction in the magnitude of the hammer loads is best achieved by a slow
valve closure or gradual pump shutdown. The maximum closure time can be
estimated from tmax = 2. L / a
where,
If the pump or valve closure time is less than tmax, then a hammer analysis should
be performed.
• When performing a modal analysis on the piping system, impulse loading such as
water hammer may have high excitation frequencies even as high as 200-300Hz.
For large piping systems it may be impractical and time consuming to extract all
natural frequency modes hence static correction methods ZPA and missing mass
are available in AutoPIPE. Generally the missing mass method is more accurate.
For small piping systems the extraction of high frequency modes is relatively fast
and will more accurately predict local dynamic responses than the static correction
methods.
Assumptions
• The maximum surge pressure is based on instantaneous valve closure. Longer
closure times which can significantly reduce surge forces, will not change the
maximum surge pressure.
• The maximum fluid transient loading will pass through the piping system once only
and no reflections will be considered.
• When a pump is shutdown, there are two shock waves generated. A positive
pressure wave on the suction end and a negative pressure wave on the discharge
end are generated. AutoPIPE will not check for pump cavitation i.e. the discharge
pressure < the liquid vapor pressure. The sudden pressure drop on the discharge
end may also cause a backflow, which will create its own water hammer effect
when it slams against the idle pump. AutoPIPE will ignore this effect also.
The magnitude of the pressure wave is estimated from the Joukowski formula:
∆P = ρa∆v
The items in the equation include the fluid density, speed of sound and change in fluid
velocity.
Bentley Confidential 63
Question: I am performing a steam hammer analysis but when should I use
ZPA correction method under Time History analysis?
Answer: We recommend to perform two analyses, one with ZPA and one without. For
flexible legs (legs with flexible or no axial supports) use no ZPA correction. If the system
has pipe legs with rigid axial supports, use ZPA correction to determine realistic loads on
these axial supports. Note: ZPA can be very conservative for flexible legs.
Define the flow rate with correct sign. Flow rate is positive for negative pressure rise.
Note: When a pump is shutdown, there are two shock waves generated. A positive
pressure wave on the suction end and a negative pressure wave on the discharge end are
generated. The maximum possible negative pressure wave is equal in magnitude to the
pump discharge pressure(Ps) less the liquid vapor pressure (Pv). The pressure wave
amplitude is calculated in AutoPIPE using the Joukowski formula. Be sure to check your
vapor pressure is not below Static - maximum pressure drop (i.e. Ps-Pv) otherwise
cavitation will occur
• This pressure wave = dP should be less than Ps-Pv to avoid cavitation. This
condition should be avoided since the AutoPIPE results will be invalid. Similarly the
pressure rise will be positive upstream of a closed valve and negative downstream
of an open valve.
• Define time history duration as 1st period (1/first modal frequency, hz) + transient
duration (as shown in the THL file i.e. TOTAL WAVE TRAVEL TIME (L/a))
• When click ok to the fluid transient check the red highlighted sections of piping are
correct.
• Run the modal analysis with cut-off frequency at least 100 to 150hz.
Recommended to perform modal followed by time history analysis at both cut-off
frequencies to confirm the solution has converge i.e. time history results are similar.
• Run time history with and without ZPA correction. See FAQ Q31.
Note: Recommend to set under tools/model options/Edit "Mass points per span" = A to
allow the program to automatically perform mass discretization on your model for
improved accuracy for the dynamic analysis.
Bentley Confidential 64
Support solution
Flexible is better. The restraint should only be stiff enough to sufficiently attenuate the low
frequency gross deformation.
Question : The AutoPIPE help states “When the rise time several times larger than
the 2L/a time, the calculated pressure rise in AutoPIPE might be conservative. For
this special case, the use of a fluid simulation software is recommended if P2 case
is critical.” What does this mean?
Answer: Check for maximum surge pressure (static (362)+ rise (228) =590 psi). This
should be added as a second pressure case (P2). Use Tools/Model Options/General and
set number of operating cases to 2. Use Select/All Points and follow by Modify/Pressure &
Temperature and set design pressure for P2 to 590 psi. When the rise time several times
larger than the 2L/a time, the calculated pressure rise in AutoPIPE might be conservative.
For this special case, the use of a fluid simulation software is recommended if P2 case is
critical and causes an overstress condition in the pipework
If P2=590 psi governs the design, that is critical, the use of fluid simulation software is
recommended since AutoPIPE value would be too conservative.
Bentley Confidential 65
Steam Relief
(Sample model : ap50SR1.dat)
• Used to calculate the thrust loads at the valve exit piping due to discharge of the
relief valve and automatically generate the time history files to perform a dynamic
time history analysis.
• Only the thrust loads at the valve exit piping will be calculated in accordance with
Appendix II of ASME B31.1 code. Since these thrust loads are dynamic in nature,
they can be analyzed more accurately using the AutoPIPE time-history solution
than using ASME 31.1 Appendix II section 3.5.1.
• The steam relief event is dependent on the type of vent piping either discharging to
atmosphere or a closed manifold. Using AutoPIPE, the configuration of relief and
vent piping is modeled as either:
• Open discharge to atmosphere- with non-integral relief and vent piping
• Open discharge to atmosphere- with integral relief and vent piping
• Closed discharge to manifold - with integral relief and vent piping
• A Force vs. Time history is generated at either the pipe/vent interface point or the
vent exit point depending on the type of vent. The profile looks like:
F1
where,
• AutoPIPE automatically calculates the value of f ∑(L/D) from the relief valve exit to
the pipe/vent interface point or vent exit depending on the type of vent.
• For non-integral open vent systems, AutoPIPE checks the blow-back condition. If
steam blowback does occur, AutoPIPE can re-size the vent pipe so that the
blowback condition is satisfied.
Bentley Confidential 66
Enter Valve exit point name,
each name = separate report
Point 1
Point 3
Point 2
Bentley Confidential 67
Thrust Loads on a Safety Valve
Safety valve thrust loads for an open/ non-integral discharge system (see Figure (a)) but the calculations are
similar for the other integral vent and relief piping configurations except the thrust time history is calculated only
at the vent exit (point 3) and no steam blowback condition occurs. For the integral systems (open or closed) the
vent thrust is calculated at point 3.
1. For a open, non-integral system, the thrust forces F2 and F3 acting on the vent pipe are not applied in
the program. The user may create a resultant static load equal to (F3 - F2) and apply it to the vent exit.
Or create a force-time history file with this same applied force as a profile shown in Figure G-6.
2. For a open, integral discharge system, the thrust force F1 is acting at the vent exit in the vent
discharge direction.
3. For a closed discharge, the vent discharge direction boxes are closed and the thrust force F1 is acting
at the vent exit in the direction of the pipe.
Bentley Confidential 68
Typical Safety Valve Installations
Integral
Non-Integral
Non-Integral
Integral
Non-Integral
1. Open discharge to atmosphere- with non-integral relief and vent piping (see Figure (a)).
2. Open discharge to atmosphere- with integral relief and vent piping (see Figure (b)).
3. Closed discharge to manifold - with integral relief and vent piping (see Figure (d)).
Cases in Figures c and e. can also be modeled with the appropriate inlet and exit points as defined in Case 1. Non-
Integral system is modeled as a separate segment for the vent pipe and positioned above the relief exit piping.
Case 2. is normally modeled by building the vent pipe on the same segment as the relief exit piping. Then using the
steam relief utility, the user will enter the same point for the vent inlet as the pipe/vent interface point. AutoPIPE will
now assume the system is a integral discharge system.
Bentley Confidential 69
Steam Relief Non-integral Example Design Conditions
Piping Code = ASME B31.1
Saturated Design temperature = 201.4 ºC
Steam Line Design Pressure = 1.5 N/mm2
Saturated steam line = 8” std weight
Fluid = Saturated Steam
Vent Pipe (Enthalpy = 2790.5 KJ/kg)
(Specific heat ratio =1.1)
Steam
Drum
Vent Pipe
Segment C is
disconnected
Discharge from Seg D
Vessel OD = 3500mm Pipe
thickness = 50mm
insul =100mm
Relief Valve
Bentley Confidential 70
Slug Flow
(Sample model : slug.dat)
•
• Distance or time phase between successive slugs
• Slug forces are generated when changes in flow direction occurs as at elbows.
The change in momentum causes these forces. The amplitude of the slug force
for 90-deg elbow is:
Bentley Confidential 71
• Typical forcing functions on 90-deg and 45-deg elbows are shown below
Rθ
• It takes the slug head seconds to exit the bend and to attain the maximum
V
slug force ρAV 2 , where Rθ is the length of the bend. The peak slug force is
L − Rθ Rθ
sustained for seconds. It then drops in seconds.
V V
Bentley Confidential 72
file and one THL file need to be entered for each slug. The TIH file profile is
shown above. The THL file defines points of application of the TIH file and the
direction of load application. It also specifies the time the load will act at each
point. A force will be applied at each elbow Near and Far points as shown in the
sketch.
• The TIH file depends on slug length, but the THL does not. THL vary with slug
speed.
• Most of the response to slug flow is primarily caused by low frequency modes
that have large modal displacements in the direction of slug loading.
Bentley Confidential 73
Sample Slug Problem:
ASME B31.4
Pressure: 1.034 N/mm2
Temperature: 93.3 deg.C
Slug Length = 3.05 m
Fluid density = 800.92 kg/m3
Fluid velocity = 15.24 m/sec
Fluid diameter = 304.8mm (12 inches i.e.12”STD), Fluid area = 0.072966 m2
45-deg bends:
Rθ
Tr= = 1.5*1.57/50/2 = 0.024 seconds
V
L − Rθ
Td = = 0.176 sec
V
Tr+Td = 0.2 sec
2Tr+Td = 0.224 sec
90-deg bends:
Rθ
Tr= = 1.5*1.57/50 = 0.047 seconds
V
L − Rθ
Td = = 0.153 sec
V
Tr+Td = 0.2 sec
2Tr+Td = 0.247 sec
Bentley Confidential 74
Bentley Confidential 75
45-deg slug definition
Bentley Confidential 76
90-deg slug definition
Bentley Confidential 77
Submerged Piping and Wave Loads
(Sample model : RISER_SEABED_SOIL_1.dat and RISER_SEABED_VSTOPS_1.dat)
• Submerged piping is subjected to upward buoyant force reducing its
effective weight
WAVE
DYNAMICS
• Natural frequencies of submerged piping system are lower than
frequencies of unsubmerged piping
Bentley Confidential 78
Modeling Riser & Sea bed piping
A) Model : Riser_Seabed_soil_2
This model uses soil properties to simulate the pipeline laying on seabed as
a semi-embedded pipe which means soil stiffnesses with all reasonable
calculated values for horizontal pipe (including submerged effect on soil)
except vertical up resistance K1, P1 and K2 = 0.
SOIL PROPERTIES
Bentley Confidential 79
Soil shown in blue
using view/show /soil
e.g. typical hydrodynamic data defined for the seabed piping including
non-zero lift coefficient.
Typical Hydrodynamic data defined for the riser, zero lift coefficient
Bentley Confidential 80
To model weight of concrete
mattresses:
Either
a) Add additional weight using
insulation thickness and density
b) Uniform distributed load
A. Insert soil over this area with equivalent stiffness to match rigidity of
mattresses
B. Insert V-stops with high friction e.g. much greater than 1.0
This model uses V-stops with high friction factor to simulate the pipeline
laying on seabed
Bentley Confidential 82
Key Modeling Points:
Yes if you define the 2 loadings below this is normally sufficient to provide reasonable
static results for riser modeling.
Note: AutoPIPE does not automatically include the effect of buoyancy with the wave
loading. Buoyancy parameters must be specified using the Analyze/Buoyancy command.
The loads due to buoyancy are automatically placed in the gravity load case (GR) by
AutoPIPE. Therefore, a combination should be defined which includes GR and the load
case which holds the wave loading to see the total effect on a submerged piping system.
a) You do not need to model so many points from bend e.g B16 since the section of piping
is buried or has soil properties, autopipe automatically calculates soil springs based on the
soil spacing specified.
b) Is your water elevation = 0 correct? This means the water surface is at a global y-coord
= 0 i.e 28.23ft below anchor A00.
e) Your wave direction is required to be defined e.g 1,0,0 is in the global X direction
f) You need to input all 5 data pairs for the current loading otherwise the program may get
confused with 2 different current velocities at depth = 0.
Note: It is important to understand the non-linear load sequencing. The default load
sequence which has the wave/current loading following the gravity cold condition but you
may wish to change the sequence to consider Gr -> T1 -> U1, U2 i.e. wave cases U1, U2
etc are applied in the hot operating condition. See Q17 & Q18 in attached AutoPIPE
FAQ's
Answer: It is important to add many points along the riser section of pipe e.g. at
approximately every 8 to 10 feet to provide adequate mass discretization so the program
can capture the distributed wave loading accurately across the riser pipe. Riser pipes are
typically sloped at 10 to 15 deg and guide supports on the riser will be normal to the pipe
axis and the reaction loads normal to the pipe can be seen in the a support forces report
Bentley Confidential 83
which shows Local and global displacement's and reactions. Note: Platform wave
displacements should be applied at the platform anchor and riser guides.
Answer. When the pipeline does not experience the wave or current effects then under
xtra/hydro data set Cm=0, Cd =0 and CL = 0 across the range of pipe selected e.g the
pipe is in a J-tube, seabed pipe is buried or when concrete mattresses are applied to the
seabed piping. These Hydrodynamic coefficients will over-ride the ones defined under
Load/Wave.
Answer. Cm under buoyancy is only used to compute added mass effects during a modal
analysis.
Answer: Either a) calculate "equivalent" soil properties for the concrete mattresses then
insert these soil properties over this range b) Model Vstops over this section of seabed
piping and use high value of friction e.g 1.5 to 2.0 plus additional distributed weight loading
from the concrete mass.
Answer: Marine growth thickness usually varies with depth therefore it is recommended to
add a distributed load down the riser which can be triangular profile to simulate the varying
thickness vs depth.
Note: There is no marine growth above mean water level, i.e., marine growth is assumed
zero above water level for drag and inertia wave calculations.
Answer: Referring to the On-line Help a "value of Cm (coefficient of inertia) for cylindrical
bodies in a incompressible, frictionless fluid is 2.0".
Bentley Confidential 84
Also refer to DNV 1981 A.3.2 and fig A.7 which shows added mass coefficient as a
function of M/D where M is distance from a fixed boundary. If no influence from a fixed
boundary then use Cm = 1.0 otherwise Cm = 2.29 to 1.0. Most of offshore users use
default value = 2.0
Question 62: How do I define the coefficient of lift for wave loading?
Answer: The lift coefficient (CL) is typically applied only to the seabed piping and is only
defined under Insert/xtra data/hydrodynamic data. By default CL = 0.
Answer: Sorry for the confusion Cm and Ci are inter-changeable in Autopipe i.e mass
(inertia) coefficient in Buoyancy, Wave Load Hydrodynamic Data dialogs.
So Autopipe (Cm)is inertia coefficient but DNV 81 (Cm) is the added mass coefficient.
Where Autopipe Mass coefficient (Cm) is the inertia coefficient(i.e. 1 + added mass coeff)
[where added mass coeff = Range 2.29 to 1.0 (no fixed boundary) as per DNV'81 Figure
A.7), hence Autopipe Cm(Ci) = 3.29 to 2.0]
We will be updating the program and help in v7.0 to clarify the definition of
these coefficients.
The only Cm (inertia) used in the modal analysis is the Cm value in the buoyancy loading
dialog.
Answer: In accordance with DNV 2000, AutoPIPE currently can output the following Local
Forces and Moments results:
1. GR = N + PeAe
2. P1 = internal pressure forces in pipe wall not including PiAi (capped pressure term).
Bentley Confidential 85
3. GR + P1- PiAi = S = N + PeAe - PiAi
Since the sign conventions for S (Effective axial force) , N (True axial force in the pipe
wall) , PeAe is consistent with respect to tension or compression i.e signs are
automatically calculated by the program and included in the GR and P1 load cases.
Answer: When the 2nd soil stiffness (K2) is set to zero, the pipe displacement can be
large or can cause instability as the soil yields. Most soils especially sand have a parabolic
force-displacement shape and a larger K2 value is justified but it is conservative to
assume a small K2 value. It is advisable with a model so sensitive to changing the K2
value, to examine the soil displacements and forces for more detailed evaluation of
yielding e.g. as seen in the soil forces report the horizontal soil force is about 176 kg/m at
A78 compared to the P1 value is only 145.8 kg/m hence the soil does not have enough
lateral resistance to support this pipe.
Bentley Confidential 86
Buried Pipe Analysis
(Sample model : pipesoil.dat)
Bentley Confidential 87
• Soil Stiffness are defined in the local coordinate system which correctly
captures soil interaction for sloped pipe.
Vertical
Horizontal
Axial or Longitudinal
Bentley Confidential 88
Bentley Confidential 89
Each supported point has stiffness in longitudinal, transverse and vertical
directions
Bentley Confidential 90
• Unified Soil Classification System can be used to determine soil certain
properties.
Bentley Confidential 91
• Several theories exist to determine stiffness of soil from physical soil
data
• Soil springs only provides bi-linear elastic behaviour i.e. soil ‘friction’ is
not path dependant unlike AutoPIPE supports i.e. unloads along same
path, no residual displacements upon unloading or hysteresis.
Bentley Confidential 92
Close soil spacing around
bends Ground
Level
Bentley Confidential 93
Some Key Design Points:
A) Ensure AutoPIPE soil spacing guidelines in the Reference online Help are
followed:
Bentley Confidential 94
Question : My buried piping system is showing large Displacement's in the
gravity case?
ANSWER. Soil supports are specified for most models with a final stiffness K2 = 0,
although AutoPIPE can still solve this problem the results may be invalid if the nonlinear
system gives a large deformation. P1 becomes the ultimate soil restraint force for any
displacement that is greater than the corresponding yield displacement.
The soil should still have some stiffness after yield to restrain the pipe, this is
accomplished by specifying a non-zero K2 value (e.g. K2= 0.01) or to include more
restraints in the system or reduce the soil span in the soil identifier. Note that may also be
possible that the K1 values for the soil are not providing enough stiffness to the system.
The value of 0.01 for k2 is arbitrary. It is meant to denote elastic-plastic case. The use of
zero could cause instability or unreasonably large displacements.
Conversely, if k2 is not zero, large displacements could cause less accurate calculations
for the soil restraint force (since there is no cutoff point for the upper end of the k2 region).
Answer: Assuming the bend has supports along it then model the bend as series of
straight pipe run sections with offsets calculated as a segmented miter bend. Alternatively
enter the bend with radius = 50x12 = 600 inches (for ENGLISH units, by overriding the
word Long or Short in the radius field. Radius units is displayed in the lower right of the
main AutoPIPE window) and insert soil over the large bend with large value of downward
soil stiffness, vertical up and lateral stiffness = 0, longitudinal stiffness = 0 (or some
nominal value to include some frictional resistance).
Answer: If the piping system is above ground then at the v-stop and guide supports we
recommend to apply settlement displacement in a user case e.g. U1. This will allow the
soil settlement to be observed as a unique loading. Then edit the sustained code case
combination SUS + Gr to include U1 i.e. the settlement is added to the sustained stresses.
If the piping system is buried then the imposed displacement at defined node points (with
soil) would give the soil settlement profile. In other words, the base of each soil spring is
subject to a displacement interpolated from adjacent nodes with imposed displacements.
Hence select the range of points and Insert/xtra data / imposed support displacement =
settlement value in the U1 case.
Bentley Confidential 95
1. Using model A, add the settlement in the T1 case and run the hanger selection to size
the spring.
2. Copy model A to Model B.
3. Using model B, change the settlement from case T1 to case U1 and create user non-
code combinations using U1 and add U1 into the sustained stress case GR+MaxP as
above.
4. Using a non-code combination e.g. GR+T1+U1 the combined movement and load on
the supports including any spring hangers can be evaluated.
Answer: When the 2nd soil stiffness (K2) is set to zero, the pipe displacement
can be large or can cause instability as the soil yields. A perfectly plastic soil K2 = 0 is not
typical unless it is called plastic clay. Most soils especially sand have a parabolic force-
displacement shape and a larger K2 value is justified but it is conservative to assume a
small K2 value. It is advisable with a model so sensitive to changing the K2 value, to
examine the soil displacements and forces for more detailed evaluation of yielding e.g. as
seen in the soil forces report the horizontal soil force is about 176 kg/m at A78 compared
to the P1 value is only 145.8 kg/m hence the soil does not have enough lateral resistance
to support this pipe.
Answer: The primary reason for the large displacements is the ultimate soil strength (P1).
As seen in the soil forces report the vertical soil force is about 290-A295 lb/ft at A23 to
A19 compared to the P1 is only 193.7 lb/ft hence the
soil does not have enough resistance to carry this 28" pipe.
Answer: The Buried pipe option can be used to model semi-embedded or non-embedded
piping. The difficulty is calculating a transverse horizontal(lateral) and longitudinal soil
stiffness.
Bentley Confidential 96
d) The longitudinal K1 and P1 would go to zero with Z=0 but there may be an alternative
equation in some textbook which would calculate a non-zero
longitudinal K1 and P1 although suspect it would be a low value since only assumed line
contact of soil with the pipe.
You may wish to enter some non-zero K1 and P1 values to evaluate the longitudinal
frictional stiffness effect back on the pipe system.
It sounds like you want to model the seabed piping as semi embedded i.e.
all soil stiffnesses with all reasonable calculated values for horizontal pipe
(including submerged effect on soil) except vertical up resistance K1, P1 and K2 = 0
Refer to Workbook / Pipe soil transition model for calculating values of soil stiffness.
Answer: One method to simulate subsidence of known amount e.g. 25mm. Set the yield
P1 = 25 as shown below and K1 (initial) value = 0.001 and calculated soil stiffness = K2
(final).
SOIL PROPERTIES
If the maximum settlement is unknown, then model the soft waterlogged soil over the
known span and known firmer soil at ends. Apply the over-burden sand weight as UDL
load or vertical wind load (to simulate the pressure of the 4m sand) on the pipe and
observe the deformation and check the resulting maximum slope.
All beam theory piping analysis programs such as AutoPIPE, Caesar, pipeplus are based
on small deformation theory and large displacements can give an ill-conditioned solution
and erroneous results.
As a check that AutoPIPE can handle these large deformations use the following rule of
thumb. Check that the maximum slope angle in radians of the deformed pipe = approx.
sin(slope angle) then the solution should be ok.
Bentley Confidential 97
Question : How do I model soil friction for my buried pipe?
Answer: There is no concept of friction for buried pipe. What you enter are stiffness and
not friction. You may be able to simulate friction, but that is left to the user ingenuity. The
axial soil force of T1=19900 N is correct when the pipe displaces more than 1 mm
(y1=P1/K1). As you see that value is the P1 value which the user can set manually.
Question : Can you please provide some references for soil properties.
Answer: Guidelines for Design of Buried Steel Pipe. July 2001 published by ASCE
references the AutoPIPE program.
http://www.americanlifelinesalliance.org/pdf/buried_pipe.pdf
Answer: Technically you need a support where your imposed displacement is. The
support should have a component in the direction of imposed displacement. The imposed
displacement is actually applied to the base of the spring away from the pipe and that is
the reason for the difference. When your support is rigid, the pipe displacement would be
the same as the imposed displacement.
In case of soil supported piping, the imposed displacement at the nodes would be the soil
settlement profile. In other words, the base of each soil spring is subject to a displacement
interpolated from adjacent nodes with imposed displacements. You might need to apply
imposed displacements at all soil supported node, especially the near and far bend points,
if any.
Answer: To resolve a total force (lb) at a straight run point - add a dummy concentrated
force = 0 at the pipe point. Perform a static analysis and look at the local forces and
moments report as seen below and will see a - and + point, the
Bentley Confidential 98
difference between the two is the load taken by the A10 soil spring. E.g. Z in T1 case =
4860 - -22519 = 27379 lb
L O C A L F O R C E S & M O M E N T S
Question : Why are the soil forces and deformations not reported on the same
row as the point name?
Answer: Soil forces and displacements are reported at middle of the element, not at the
node. See report below:
D I S P L A C E M E N T S
S O I L F O R C E S & D E F O R M A T I O N S
Note: Soil forces are always in lb per linear foot of pipe, Hence the actual force (LB) at a
soil spring in a longitudinal (axial) or Horizontal (lateral) or vertical direction can be
calculated by the average length between the upstream and downstream points
of the current point.
Bentley Confidential 99
Question : Why is my buried model unstable?
Answer: It appears the horizontal soil yield displacement is very small 0.0012 to 0.004in
which would appear to cause local yielding in so many areas for T1 and P1 cases that was
unstable.
I changed K2 value = 0.1 for all soil identifiers but the model still did not converge
in pressure case P1 so I investigated the LOG file and discovered a soil error = 532lb. I
searched the log file for 5.32 and found this error at B43 in the P1 vertical case. I removed
the soil from B42 to B44 and the P1 case also converged. You could try to tweek the soil
properties of 24H78 to also achieve convergence.
Note: The log file must be in English units (set input/output units = English).
Depending on the requirements of the design specification, the live-load effect may be
based on AASHTO HS-20 truck loads, Cooper E-80 railroad loads or a 180 kip airplane
gear assembly load.
The modified Iowa deflection formula may be used to calculate the pipe ovality under earth
and live loads.
In steel pipelines, buckling typically occurs when the ovality reaches about 20%. Other
construction and code requirements typically limit the amount of permissible cross section
ovality for new steel pipelines to much smaller values (e.g., 3% in API RP-1102).
• Legal document
Note: The option "Include Pressure Thrust in GR, HY Case" in the Load
Combination - Load Cases tab is generally an over-conservative option which
applies pressure thrust in the radial load of the gravity case GR and
hydrotest case HY.
WRC368 within its geometric limits provides a good design check of pressure
stress levels which includes the full thrust load otherwise use FE analysis to
obtain more accurate combined stresses.
It is good practice to design these types of piping systems with all natural
frequencies 10Hz or above to avoid large displacements due to seismic,
wind, impact loads or flow induced vibration.
A configuration below can be used to to add vertical restraint (e.g. rod hanger)
on spring supported system. It will add some torsional affects but may
provide enough vertical thermal movement and vertical restraint to avoid low
frequency modes. Other support types like guides and dampers are common
also.
• One rule of thumb is to keep mode shapes above 6Hz for reciprocating
pump or compressor systems which create pulsation flow (acoustic
vibration) of liquid and gases.
• Low horizontal or axial modes “may” be ignored if the piping system is
supported on sliding frictional supports like pipe shoes since dynamic
analysis does capture non-linear effects like friction and gaps. To
simulate these effects, a stiffness or rigid restraint e.g. guide could be
provided in these directions.
• Include all weight components including valve operators, dummy
support legs with baseplates or any other inline or offset weight loading.
• Consider both mechanical and flow induced vibration effects.
• Driving frequency of the applied harmonic load should be equal to
source pressure pulsation frequency or motor speed.
• The analyzed harmonic load displacement can be verified against the
actual measured displacement to capture a realistic harmonic loading.
• Add vertical and lateral supports to attenuate the harmonic loading but
ensure static load stresses e.g. sustained and thermal are satisfied.
0 stiffness in X dir
• Step 2: Apply a unit "concentrated force" (fA) at the anchored point for each
released DOF. For example, if only the "X" DOF was set to zero, apply the force in
the global X direction (fAX). If all three DOF's were zeroed, apply forces in each
global direction (fAX, fAY, and fAZ). The applied forces should be placed into separate,
isolated load cases (such as fAX -> U1, fAY -> U2, and fAZ -> U3; where no other load
effects are considered in Ul - U3).
NOTE: Any applied force value can be used for the unit force since its magnitude
scales the corresponding displacement calculated by AutoPIPE. However, care
should be taken when considering the magnitude of the resulting displacement
value because of the significant figures limitations in AutoPIPE reports.
NOTE: The applied force magnitude (fA ) can influence a nonlinear analysis result
set (lift-off and gap closure), and AutoPIPE is only capable of linear supports in a
dynamic analysis.
• Step 4: Calculate the stiffnesses (k) of the anchor for each released DOF as
follows:
fA
K= × 10 3
∆A
this will result in a corresponding kX, kY, and kZ value. The value "10E3" is
somewhat arbitrary, but it is a reasonable assumption based on the stiffness of a
"rigid" anchor DOF. Then change the anchor stiffnesses, which were set to zero in
Step 1, to the values calculated in this step.
• Step 5: Calculate the (piping system) mass of the anchor point (manc). This is
done by summing the weights of any components defined at the anchor point (i.e.
flange, or weight) and the weight of the pipe (and contents) based on the half
length to the next point (piping, soil, or mass). If the length of pipe between the
anchored point and the next adjacent piping point is buried or if "automatic mass
discretization" has been employed, the MODEL "list" report must be consulted to
determine the number of "transparent" points added by AutoPIPE in order to
calculate the correct half length. Once this total weight has been summed, divide by
"g" (using appropriate units) to obtain the mass.
• Step 6: Calculate the cut-off frequency (fc) to be specified for the modal analysis
from the following equation:
• Step 7: Perform a modal analysis and specify the cut-off frequency calculated in
Step 6. Make sure that the last mode calculated by AutoPIPE reaches the cut-off
frequency. If it does not, rerun the modal analysis with a greater number of modes
specified (along with the calculated fc). Iterate until the last mode reaches fc.
NOTE: This requirement can result in a large number of modes being captured.
Missing mass and ZPA corrections do not impact the requirement of capturing the
appropriate number of modes.
• Step 8: The harmonic force (FH) loads are calculated from the following equation:
where "k" corresponds to the appropriate global direction, (kx, kv, and kz). Thus,
there may be FHX, FHY, and FHZ (these are the values entered in the HARMONIC
• Step 9: Now, move on to the next (vibration source) anchor and repeat Steps 1 - 8
in order to calculate the harmonic forces (FH) to be applied at this location.
Note: Step 9 can be ignored for a single vibration source and 0 phase angle used.
Where multiple vibration sources exist in a piping system, each set of harmonic
forces can be applied as an individual load case (one-to-one correspondence of
source and load case), or they may be grouped into a single load case. If the latter
scenario is desired, the phase angle relationship can be evaluated.
• Step 10: Once all harmonic forces are modeled and defined as desired, perform a
final Modal analysis where the cut-off frequency is the maximum calculated for all
vibration sources. Then, perform the harmonic analysis. The desired results can be
obtained from the "Displacement" and "Restraint" reports, created using the Result
/ Output report, for the appropriate load cases (Hi - H3) or combinations of load
cases.
If all relevant modes have been captured, the resultant anchor displacements
should be within 2% of the measured amplitude ( ). Piping forces and
displacements for the remainder of the system will also be correct.
However, if fewer modes are extracted, the anchor displacements reported
by AutoPIPE will be near zero and the corresponding forces will be incorrect.
Piping forces and displacements for the remainder of the system will be close
to the correct values.
Note in Step 4, a sensitivity study should be made on the model with a
stiffness factor between 1000 and 1.0E8. The cut off frequency is not as
important as capturing the modal frequencies near the harmonic frequency.
Pipe span guidelines from industry practice to avoid resonance from gas
compressor accoustic vibration.
1. Typically for low speed motors e.g. 300rpm use multiplier 2.4 up to 4.0
times the 1st harmonic frequency, i.e. 5hz x 2.4 = 12hz
2. High speed compressors e.g. 1800rpm use multiplier 1.4, i.e. 30hz x
1.4 = 42hz