Sie sind auf Seite 1von 42

Design, Analysis and Testing of

Saddle Supported Pipes and Vessels


Paulin Research Group
11211 Richmond Ave. Suite 109
Houston, Texas 77082
www.paulin.com
281-920-9775
November 18, 2003
Audience
Vessel Engineers Saddles with or
without wearplates
Piping Engineers Saddles & Pipe Shoes
with or without wearplates
Significant Features of the
Geometry
Wearplate attached to shell/pipe only on outer edge by fillet or stitch weld
Support members welded only to wearplate not directly to shell
Saddle external load transfer to shell is different for Support and Nozzle.
Trends are dominated by the ratio t
wearplate
/t
shell
Integral and NonIntegral model types are used most often
Integral vs Non-Integral Wearplates
Finite Element Analysis can evaluate geometries and
loads when no other solutions are possible
CONSISTENT DESIGNS ARE PRODUCED.
External Loads (sloshing, wind, seismic, piping induced thermal)
Fatigue Analysis
Multiple Saddles
Ovalization effect due to liquid head
Influence of Pressure on Circumferential Bending
Geometries beyond Zicks Scope (b +10t, etc)
Recognized Limitations of Standard Zick Approach
Adjusted beam theory and ring equations to match strain gage results.
Weight Only Weight Only No External Loads on Saddle Supports No External Loads on Saddle Supports
Integral Wear Pad Model
Minimum Extent of Wear Pad (b + 10t, R/10, 12 Limit)
Shell is fixed at Saddle Horn (Circumferential Bending)
Rigid Support not Connected to Shell (Ring Compression)
Pressure Stress Effects Not Included
Limited Load Cases
Comparing FEA to Zick
Compare secondary stress intensity (Pl+Pb+Q) to Zick
circumferential stress in shell for weight loads only.
Do NOT compare peak stress intensities (Pl+Pb+Q+F) values to
Zick stresses.
Zick theory and equations are adjusted to match strain gage
results. Strain gage stresses are location and gage size dependant
and reflect an average of the stress over the effective gage length.
In general, FEA secondary stresses will be more conservative than
Zick because the stress is calculated at all points in the geometry
and is not limited by the location or size of a strain gage.
Weight (Zick-Type) Stress Trends at Saddles
Highest stress usually at saddle horn (Compressive on Outside)
Vessel length, diameter, and thickness will influence shift of stresses
the vessel MUST be allowed to ovalize if there are no stiffening rings.
Stresses due to external loads DO NOT necessarily follow these
trends.
Tensile stress on outer surface
near meridian
Compressive stress on outer
surface at saddle horn
Stresses concentrate on inside
edge of saddle
Ovalization
When L*D/t > 15 vessel weight and fluid caused ovalization can
contribute significantly to stresses at saddle horn for vessels without
stiffening rings.
NozzlePRO analyzes filled, partially filled or empty vessels.
Ovalization does not affect stresses due to external loads as strongly.
Saddle Design Guidance
Flexible wear plate reduces stresses in shell BUT may increase the stress
in the wearplate. Zick will not show this shift. Run the FEA calculation.
Optimize the design by balancing the stress in shell and the wear plate.
Wearplates in the parameter range 0.50 < t/T < 1.0 are usually best for
weight. For t/T < 1.0, high weight stress is between the saddle horn and
edge of pad. For t/T = 1.0, high weight stress is near top edge of pad. For
t/T > 1.0, high weight stress is above top edge of pad.
For external load stresses, the wearplate thickness does not as strongly
effect the shell stresses. This is true for 0.50 < t/T < 1.40.
Wearplate thickness is typically selected equal to shell thickness.
FEA shows that optimum designs can be between 0.5 < t/T < 1.25
All recommendations use saddle angles between 120 and 150 degrees and
a 12 degree extension of the wear plate past the outer web plates.
Wearplates are NOT always required. Using FEA analysis guidelines
below NozzlePRO users can quickly determine the need for wearplates.
Influence of Wear Plate Thickness
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
t/T
S
t
r
e
s
s

R
e
d
u
c
t
i
o
n
Integral Pad
Non-Integral Pad
Zick (Saddle Horn)
Zick (Above Pad)
61%
52%
24%
Influence of Wearplate Thickness on
Weight Stresses in the Shell
Non-Integral Model shows lowest shell stresses due to weight and external
loads and compares best to Zick and PRG strain gage data.
Weight Stress Distributions
t / T = 0.375 t / T = 0.50
t / T = 1.0
t / T = 1.50
Notes for Analyzing External Loads
Axial loads tend to concentrate stresses at corners of saddle
Both positive and negative axial load directions need to be evaluated
because of the interaction with weight stresses.
Integral models tend to give the best stresses in the wearplate and
shell under the wearplate.
Non-integral models tend to give the best stresses in the shell
adjacent to the wearplate.
Stresses in the shell are not affected so much by the wearplate
thickness for external loadings.
These notes apply for wearplates that are equal to b+10t where b is
the width of the widget and t is the thickness of the shell.
0
10000
20000
30000
40000
50000
60000
70000
80000
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
t/T
S
t
r
e
s
s

(
p
s
i
)
Vessel Wall (Integral Pad)
Wear Plate (Integral Pad)
Vessel Wall (Non-Integral)
Wear Plate (Non-Integral)
Stress Trends for External Loads on Saddles
Stress in Shell is not strongly effected by the wearplate thickness for external loads.
Below t/T of 1.0 the high stress due to external loads moves into the wearplate.
Non-Integral models are the most conservative.
Piping Support Axial Loads are
typically due to primary and
fatigue loading:
Thermal Expansion of Pipe (Fatigue?)
Seismic/Wind (Primary & Fatigue?)
Water Hammer (Primary?)
Pulsations (Fatigue?)
Pitch & Roll (Primary & Fatigue?)
Horizontal Vessel Support Axial
Loads are typically due to
primary and fatigue loading:
Seismic (Primary & Fatigue?)
Pitch & Roll (Primary & Fatigue?)
Acoustic (Surge Protection and Noise
Attenuation Devices) (Fatigue?)
Primary Load Failures
Produce Collapse or Excessive Distortion
Seismic/Wind loads could produce collapse or excessive
distortion
Thermal loads tend not to produce collapse or excessive
distortion because they are strain limited or secondary.
Pitch & Roll Loads could produce collapse or excessive
distortion of saddle supported vessels.
Waterhammer loads could produce collapse or
excessive distortion of the saddle supported vessels or
pipes.
Liquid loads due to hydrotest can produce excessive
distortion.
Fatigue Load Failures
Seismic loads do have a cyclic component.
Seismic # of cycles can be taken to be around 100.
Seismic loads could contribute or add to fatigue failures from other
sources
Thermal cycling is a major cause of fatigue failures in pipe systems.
Pressure cycling can cause fatigue failures, but tend to be covered
by added awareness and design precautions and not by a Div 1 or
B31 type analysis.
Pitch & Roll Loads have a cyclic quality and may initiate fatigue
cracks that jeopardize the pressure boundary or cause support
failure due to crack growth and loss of load carrying area.
Acoustic Problems typically have several fatigue causing
mechanisms related to cyclic load of supports and restrained vessel
dilation.
Saddle Supported Geometries should be
designed to satisfy two loading criteria
Primary Load Failures Collapse or Excessive
Distortion
Secondary/Fatigue Failures Crack Propogation
A) Thru the pressure boundary most
dangerous crack.
B) Thru the load carrying members.
ASME Section VIII Div 2 Primary
Stress Criteria
Local primary membrane stress must be
less than 1.5 Sm. (Pl < 1.5Sm )
Local primary bending stress must be less
than 3 Sm (exception 4-136.7)
Must not be elastically unstable. (Most
PVP geometries are not.)
ASME Section VIII Div 2
Secondary/Peak Stress Criteria
(plus Markl and B31 fatigue criteria)
Secondary Stress must be less than 3Sm (Pl+Pb+Q < 3Sm)
Alternating Peak Stress (Fatigue) must be less than Sa
(Pl+Pb+Q+F < Sa).
Markl Test alternating Peak Stress limit for carbon steels
= 245,000 N
-0.2
psi
B31 Alternating Peak Stress Limit
= f(1.25(Sc+Sh)) = 6N
-0.2
(Sc+Sh)
Recommended FEA Method (NozzlePRO/FEPipe/Ansys/Abacus etc.)
Programs mentioned are the top three used in Europe as listed in the Current
Industrial Practices for Pressure Equipment Design Against Fatigue.
Only 10 values are needed to describe the saddle geometry.
A variety of options are available for saddle
geometry, model and boundary condition type.
Use checkbox
to toggle
between
integral and
nonintegral
repads.
NozzlePRO Non-Integral Saddle and
WearPlate model
Wear Plate Model Weld
Size Assumption
ASME VIII Code Stress Results
Tabular Reports
Load
Entered
produced a
300%
overstress.
Test at PRG Lab in Houston of Same Vessel
Axial Displacements
(Strain Gage #1)
0
5000
10000
15000
20000
25000
30000
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Axial Displacement (in.)
L
o
a
d

(
l
b
)
Displacement (Test #1)
Displacements (Test #2)
Displacements (Test #3)
Using
NozzlePRO
and ASME
rules a 28,000
lb. load
showed to be
291%
overstressed
100% of
allowable
would be
here.
Plastic
Deformation
Safety Factor
Primary Load Evaluation with NozzlePRO gives the desired Safety Factor
Accurate Fatigue predictions are difficult
to make. It has been said:
The Good Lord watches over:
1) Children
2) Fools
3) People that try to predict fatigue failures
of PVP geometries with FEA programs.
Of critical importance in cyclically loaded systems is weld profile
and quality. (The analysis doesnt mean anything if youve got a poor
weld. The welds shown below by Cen-Tex Marine were exceptional.)
NozzlePRO fatigue evaluation per ASME Section VIII
Div. 2 Appendix 4 and 5.
490,000 N
-0.2
= Stress-to-Failure
490,000 N
-0.2
= 137,614
N to Fail = (137,614/490,000)
-5
= 572 cycles
Load that
produced
plastic
deformation
in the saddle
would
overstress
the vessel if
applied 7000
times.
Anticipated
number of
fatigue cycles
to failure.
NozzlePRO Automatically
Calculates Primary and
Secondary/Fatigue Stresses
Primary Stress Evaluation
(Pl < 1.5Sm) using NozzlePRO gives
desired Safety Factor.
Fatigue Stress Evaluation
(Pl+Pb+Q+F < Sa) using NozzlePRO gives
desired safety factor.
Dont Let FEA Be the ONLY Design
Criteria Be sure to evaluate:
Cycles If none then can be less conservative
Service Cold Water, Low Pressure then can be less conservative
(generally)
Load Designation Pilot Plants tend to be more dangerous and loads are
less certain.
Corrosion/SCC Where metal damage can result design approaches
should be more conservative.
Dynamic Loads Waterhammer/Pulsation Take more conservative
approaches.
High Temperature In the creep range be more conservative with the
How fabricated? (Does the lowest bidder provide the best product?)
Conclusions
Two stress catagories must be evaluated:
Primary (Pl<1.5Sm)
Fatigue (Pl+Pb+Q<3Sm and Pl+Pb+Q+F < Sa)
Saddles on wearplates can fail thru the pressure containment section or
thru the support section.
Integral Models are Less Conservative but tend to more accurately predict
the stress. Non-Integral Models are More Conservative but may simulate
fit-up and actual fabrication better.
Fatigue Integrity is VERY weld profile and weld quality dependant.
FEA can be used to produce good design guidance.
Watch www.paulin.com for latest on test results.
SPLASH is a 2d CFD Program that generates
sloshing and earthquake loads due to fluids in
Vessels and Pipes.
Pick the Vessel Size,
the Grid Size and the
depth of fluid
1) Select Type of Geometry
2) Enter pitch and roll sloshing period and
maximum g load or El Centro Earthquake
multiplier.
3) SPLASH Calculates
shear loads and fluid
motion.
Maximum Base
Shear to use in
NozzlePRO
Saddle Caculation
Natural Frequency of
Fluid in Vessel
SPLASH Calculates base
shears on horizontal vessels
and overturning moments for
vertical vessels.
The natural period of the fluid
can be also be determined to
see if baffles should be
added.
Movie files are created and
stored with text and graphic
report files that can be
electronically transmitted to
the client.

Das könnte Ihnen auch gefallen