7/24/2019 Mitered Elbow Fea Report http://slidepdf.com/reader/full/mitered-elbow-fea-report 1/45 GTI PROJECT NUMBER 02222-121534 PE 4710 Mitered Elbow Finite Element Analysis Date Submitted: August 25, 2014Project Investors: PPI GTI Project Manager: Natalya Bates Project Manager 847-768-0953 [email protected]GTI Principal Investigators: Ernest Lever Senior Institute Engineer 847-768-3415 [email protected]Oren Lever Engineer 847-768-0668 [email protected]Gas Technology Institute 1700 S. Mount Prospect Rd. Des Plaines, Illinois 60018 www.gastechnology.orgFINALREPORT
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
This information was prepared by Gas Technology Institute (“GTI”) for Plastics Pipe Institute(PPI).
Neither GTI, the members of GTI, the Sponsor(s), nor any person acting on behalf of any ofthem:
a. Makes any warranty or representation, express or implied with respect to the accuracy,completeness, or usefulness of the information contained in this report, or that the use of anyinformation, apparatus, method, or process disclosed in this report may not infringe privately-owned rights. Inasmuch as this project is experimental in nature, the technical information,results, or conclusions cannot be predicted. Conclusions and analysis of results by GTIrepresent GTI's opinion based on inferences from measurements and empirical relationships,which inferences and assumptions are not infallible, and with respect to which competentspecialists may differ.
b. Assumes any liability with respect to the use of, or for any and all damages resulting from theuse of, any information, apparatus, method, or process disclosed in this report; any other use of,or reliance on, this report by any third party is at the third party's sole risk.
c. The results within this report relate only to the items tested
Figure 13. Comparison of Stresses in ID and OD Reinforced Elbows ........................................ 15
Figure 14. 3-Segment Bend, 1st, 2nd and 3rd Principal Stresses [psi] (Left To Right), ExternalView, Full Model .......................................................................................................................... 16
Figure 16. 3-Segment Bend, 1st, 2nd and 3rd Principal Stresses [psi] (Left To Right), InternalView, Full Model .......................................................................................................................... 18
Figure 18. 3-Segment Bend, Von Mises Stress [psi], Full Model ................................................ 20
Figure 19. 5-Segment Bend, 1st, 2nd and 3rd Principal Stresses [psi] (Left To Right), ExternalView, Full Model .......................................................................................................................... 21
Figure 20. 5-Segment Bend, 1st
, 2nd
and 3rd
Principal Stresses [psi] (Left To Right), InternalView, Full Model .......................................................................................................................... 22
Figure 21. Effect of Bend Radius on Elbow Stress – 1.5xOD Left, 2.5xOD Right ...................... 23
Figure 22. 3-Segment Bend, 1st Principal (Hoop) Stress [psi], Internal View, Joint with Bead ... 24
Figure 29. Relative Values of First Principal Stress at Various Geometric Locations ................. 29
Figure 30. Fit of FEA Results to Equation 5 to Determine Parameter “a” .............................. 32
Figure 31. 95% Prediction Surfaces for FEA Model Fit .............................................................. 33
Figure 32. Mitered Elbow Configuration for Calculations ............................................................ 36
Figure 33. Equations and Parameter Values Used in Calculations ............................................ 37
Figure 34. Comparison of ASME B31.3 Elbow Design Equations with Bend Radius set to 2.5 x
Pipe Outside Diameter and a=0.643 ........................................................................................... 38
Figure 35. Percent Difference Between ASME B31.3, 304.2.3 Equations 4a and 4b with BendRadius Set to 2.5 x Pipe Outside Diameter and a=0.643 (relative to eq. 4a) ............................. 38
Figure 36. Percent Difference Between ASME B31.3, 304.2.3 Equations 4a and 4b with BendRadius Set to 2.5 x Pipe Outside Diameter and a=0.2856 (relative to eq. 4a) ........................... 39
A series of seventy seven (77) nonlinear finite element analyses was conducted to evaluate the
performance of various configurations of fabricated PE 4710 elbows.
In order to properly understand how a particular design behaves it is necessary to conduct a large
number of analyses that properly capture the range of boundary conditions and material
properties. The approach taken in the study was to use a “Virtual” design of experiment. In this
approach each finite element analysis is viewed as a virtual experiment.
Preliminary analysis showed that the following five input parameters are sufficient to fully
understand the performance characteristics of a mitered elbow:
1. Dimension Ratio - A: DR
2. Miter Angle - B: Angle
3. GSR - C: Factor
4. Ambient Temperature - D: Temperature
5. Operating Pressure - E: Pressure
A face centered cubic response surface design was chosen. Fifty (50) design points would be
sufficient to fully resolve all second-order interactions of input parameters. An additional 27
design points were added to augment the design.
A high-quality response surface model was achieved with low standard error that is capable of
properly analyzing all of the input FEA results. Appendix 0 presents the initial model
information prior to actual analysis of the results that are presented in detail in Appendices 1
through 16. In Appendix 17 we present the stress strain curves on which the analyses were based
and define the stress ratio that is used in several of the analyses.
Temperature dependent material properties were used to assess the impact of ambient
temperature on elbow performance. The material model used reflects the stresses developed inan elbow upon pressurization and does not include creep of the material. This level of analysis is
sufficient as the Hydrostatic Design Stress (HDS) of the material accounts for long term creepeffects for calculated stresses equal to, or lower than the HDS. The stress/strain curves weredeveloped by detailed testing of physical test specimens of a typical PE 4710 material provided
by a leading resin manufacturer. To validate the suitability of the stress/strain curves used in this
model four different HDPE pipes that capture the range of materials typically used in the market
were obtained from North American pipe extruders. A single MDPE was also obtained forreference. Three replicate tensile tests were conducted on each material at 0°C, 22°C and 40°C.
The model used in the analysis accurately captures the average material behavior and will be
slightly conservative at strain rates less than 5% as the model elastic modulus is at the high end
of the measured modulus for the range of materials across all temperatures.
Four arcs shown in Figure 8. Probe Locations were chosen to fully describe the stress state of
each elbow configuration. Response surfaces were developed for the average stress among each
of these arcs, the maximum stress among each of these arcs, and the ratio of each of thesestresses to the temperature dependent yield stress of the material
The first principal stress P1 is a good indicator of the stress state of the fabricated elbow under
all loading conditions for all fitting configurations. The highest stresses occur at the intrados of
the miter joint; hence the first principle stress (hoop stress) at this location was used in all
calculations in the report.
The effect of fusion beads on the stress fields in the miter joint were evaluated and found to be
insignificant. Therefore, to avoid unnecessary modeling complexity, the full analysis did not
include fusion beads in the geometry of the elbow.
Internally and externally reinforced designs were compared and found to be equivalent. Only
externally reinforced elbows were modeled in the full analysis.
Analyses were run to verify that the elbow stresses for DR based designs are insensitive to pipe
diameter. The results differed by less than one half of a percent across all configurations and
diameters thus justifying the modeling of configurations for a single pipe size.
The impact of mesh size on the stress results was investigated. The mesh size was chosen tominimize analysis time without compromising the stress result. The stress results differed by less
than one tenth of a percent from FEA models with a high resolution mesh.
The analysis presented in this report justify the use of ASME B31.3-2012 Section 304.2.3
equation 4a (Equation 1 in this study) as the basis for PE 4710 mitered elbow design. This
A CAD model was created based on a draft ASTM specification that is being prepared by a task
group of the Plastics Pipe institute.
The model was constructed parametrically to allow it to accommodate all possible configurationsunder the draft standard and was driven by design tables which allowed for quick configurationadditions.
Butt-fusion beads were modeled in a single configuration to assess their impact on the local and
global stress fields. The results of this sub analysis are presented in the section Fusion Beads.The net effect of the beads on the stress fields is shown in Figure 25 through Figure 27. It can
be seen that the effect is minor and it was decided not to model them in all configurations under
the scope of this project due to their complex and inconsistent geometry, which presented majordifficulties in including them in a DoE.
All analyses assumed a bend radius of 2.5 times the OD. A bend radius of 1.5 OD was evaluatedto assess the impact on the intrados first principal stress. The results, summarized in the section
Effects of 1.5xOD Bend Radius on Elbow Stress in the body of the report, were within one tenth a
percent of each other (1358.4 psi for the 2.5 OD bend radius and 1359.7 for the 1.5 OD bendradius). The analyses can thus be considered conservative for bend radii greater than 1.5. Tighter
bend radii approaching 1 (sometimes used in large diameter mitered bends) need to be evaluated
The FEA used a 90°, constant ID miter bend for all simulations. A quarter model was used in
order to reduce computation time. The simulation assumed symmetry in all directions (see
Figure 4) meaning that, as pressure was applied, the elbow was free to expand (hoop) and to
move in the radial direction of the bend radius. The elbow’s end was not otherwise constrained.Defining symmetry in all directions constituted the simplest loading condition where nosecondary bending moments are introduced.
Figure 4. Symmetry Surfaces
Pressure was applied to all internal surfaces (Figure 5). Mesh element size was adjusted foreach geometric configuration such that the nominal pipe wall had one element across its
thickness and the joints have five, see Figure 6 and Figure 7, respectively. The elements were
second order elements that maintained a good balance of accuracy and computation time.
Figure 8 shows the locations where stresses were probed.
FEA used a nonlinear elastic material model based on HDPE tensile stress-strain curves at 0°C
(32°F), 20°C (68°F), and 40°C (104°F) temperatures, all pulled at 0.001/s engineering strain rate.The low strain rate was chosen to more closly simulate actual pressurization rates.
The nonlinear elastic model did not account for material relaxation or creep and, therefore, the
results of the FEA represent an estimation of the immediate stresses upon pressurization.
Poisson’s ratio was 0.41.
To validate the suitability of the stress/strain curves used in this model four different HDPE
pipes that capture the range of materials typically used in the market were obtained from NorthAmerican pipe extruders. A single MDPE was also obtained for reference. Three replicate tensile
tests were conducted on each material at 0°C, 22°C and 40°C. The results are presented in
Figure 10 to Figure 12 below. The model used in the analysis accurately captures the averagematerial behavior and will be slightly conservative at strain rates less than 5% as the model
elastic modulus is at the high end of the measured modulus for the range of materials across all
The results of the first set of tests (Table 1) showed that, given an elbow that is unrestricted in
the radial direction of its bend, pipe diameter has no effect on the stresses due to the absence ofsecondary bending moments. Stresses are dependent on pipe dimension ratio and applied
pressure, therefore, for example, a 34-inch SDR 11 pipe with a 3xOD miter bend will have thesame stress as a 14-inch SDR 11 pipe with a 3xOD miter bend.
Table 1. Average Hoop Stress (psi) at Miter Joint, SDR 17, Gore-DR 13.6, 125 psig, 68°F
Size 5-segment 4-segment 3-segment
14 IPS 962.4 1054.1 1230.9
34 IPS 968.0 1056.2 1232.0
Difference 0.58% 0.20% 0.09%
When the pipeline design is based on the DR of the pipe, stresses in an unrestrained miter elbow
have no dependence on pipe diameter, therefore the subsequent DoE matrix did not have
diameter as a variable in order to reduce the number of analyses.
Comparison of Internally and Externally Reinforced Elbow Stresses
The average first principal stress at the intrados was evaluated for 14-inch IPS DR 11 at two
extreme configurations; heavy wall five segment and thin wall three segment elbows for both theID reinforced and OD reinforced designs. There is less than 1% difference in the results,
therefore only OD reinforced elbows were considered in the FEA study.
Table 2. Comparison of stresses in ID and OD reinforced mitered elbows
Diameter
ControlConfiguration
Pressure
[psig]
Intrados P1
Avg.
[psi]
Extrados P1
Avg.
[psi]
OD reinforced 14IPS11-5-06-20C 250 736.04 447.93
OD reinforced 14IPS11-3-08-20C 250 1358.4 521.11
ID reinforced 14IPS11-5-06-20C 250 729.11 449.37
ID reinforced 14IPS11-3-08-20C 250 1349.5 492.66
ID/OD 14IPS11-5-06-20C 0.991 1.003
ID/OD 14IPS11-3-08-20C 0.993 0.945
Figure 13. Comparison of Stresses in ID and OD Reinforced Elbows
In order to determine the influence of the various geometry and boundary condition variables on
stresses in the elbows, DoE and RSM were employed. Using these techniques, an appropriateFEA test matrix was developed and the influences of the test variables could be resolved.
Two sets of analyses were carried out. The first set was used to find out the significance of pipediameter. The second test matrix was used to analyze the effects of pipe DR, joint angle, elbowDR factor, temperature, and pressure.
FEA Results
The following figures show typical stress plots from the DoE analyses. Figure 14 shows thequarter model of a 14-inch IPS, SDR 11, 3-segment miter elbow at 200 psig, at 32°F.
Figure 14. 3-Segment Bend, 1
st
, 2
nd
and 3
rd
Principal Stresses [psi] (Left To Right), External View,Full Model
The first and second principal stress plots indicate that higher stresses occur at the joint line, as
expected. The third principal stresses (radial force due to pressure) are not significant in tension(i.e., < 100 psi).
Figure 15 shows a closer view of the First principal stress (hoop stress in straight pipe) at the joint line. From this external perspective the intrados is in tension and the extrados is in
Figure 21. Effect of Bend Radius on Elbow Stress – 1.5xOD Left, 2.5xOD Right
The FEA study was based on a 2.5xOD bend radius. A bend radius of 1.5xOD was evaluated to
assess the impact on the intrados first principal stress. The results were within one tenth a percentof each other (1358.4 psi for the 2.5xOD bend radius and 1359.7 for the 1.5xOD bend radius).
Figure 21 gives a visual indication of the effect. Although the intrados stress does not change,
the region of higher stress around the intrados is beginning to merge at a bend radius of 1.5xOD.
There is good separation of the stress fields with the bend radius of 2.5xOD.
An attempt was made to analyze the miter bends with beads, however, because of their
complexity; they had to be modeled individually, per configuration. This precluded their usewithin the scope of this project and therefore only a single model with fusion beads was analyzed
for reference.
Figure 22. 3-Segment Bend, 1st Principal (Hoop) Stress [psi], Internal View, Joint with Bead
As can be seen in Figure 23 and Figure 24, the fusion beads reduce the hoop stress at the very
center of the joint, but cause a stress concentration where each bead meets the inner pipe wall.
Figure 25 shows a comparison of hoop stresses on a model with and without beads. Figure 26
and Figure 27show the extrados and intrados in detail, respectively.
Figure 25. Comparison of Hoop Stress [psi] on Model with and without Beads, Full Model
Figure 25 illustrates how the bead only affects stresses in its close vicinity. Figure 27 showshow ignoring the beads results in a higher average hoop stress at the intrados corner. The inner bead, however, may cause higher stresses at the edge of its contact with the pipe’s inner wall.
The sensitivity of the FEA result to mesh density was checked. The results are presented in
Figure 28 below. The ratio between the results for the average intrados first principal stress thatis used as the basis for this analysis was 0.99985 (high resolution mesh result / low resolution
mesh result). The mesh resolution used in the DoE overestimates the high quality mesh result byless than one tenth of a percent (15 thousandths of a percent).
Figure 29. Relative Values of First Principal Stress at Various Geometric Locations
The intrados average first principal stress is always the largest stress in the mitered elbow as can
be seen in Figure 29. This stress is used in all subsequent calculations and models requiring astress input. In a straight pipe segment the first principal stress corresponds to the hoop stress of
the pipe which is the reference stress used in all pipe design calculations. For consistency this is
the reference stress we have used in this report. The relative values of the first, second and third principal stresses can be seen in Figure 14 through Figure 24. It is also clear in reviewing these
plots that there is little difference between the stress patterns of the first principle stress and thevon Mises stress in all of the elbow configurations.
In a DR 11 pipe the third principal stress (radial) is the internal pressure in the pipe, the first principal stress (hoop) is 5 times the internal pressure and the second principal stress (axial) is
2.5 times the internal pressure. The von Mises stress is 1.04403 times the first principal stress
Equation 1 for calculating the maximum allowed pressure in a mitered elbow with miter cutangle θ less than or equal to 22.5°:
. Equation 1
Pm maximum allowable internal pressure
c sum of the mechanical allowancesD outside diameter of pipe
r 2 mean radius of pipe using nominal wall T
E quality factor P p internal design gage pressureS stress value for material
T pipe wall thickness (measured or minimum in accordance with the purchase
specification)t m minimum required thickness, including mechanical, corrosion, and erosion allowances
W weld joint strength reduction
Miter cut angle
Setting E and W = 1 and defining (T-c) as t we get a simplified equation:
.√ Equation 2
t pressure design thickness
Equation 2 can be manipulated to give:
.√ .
Equation 3
Recognizing that the LHS of Equation 3 is the nominal pipe wall thickness t0 we can rearrangethe equation to give: . Equation 4
Equation 4 is useful for calculating the required wall thickness for a mitered elbow. This
equation is derived from the work of Green and Emmerson [1]who solved the elasticity problemof an unrestrained discontinuous bend. The solution developed by Green and Emmerson has
been extensively validated for steel mitered elbows. Wood [2]used this equation to validate the
finite element solution he developed for steel mitered elbows. We will now use this equation tovalidate that the non-linear finite element study carried out by GTI is consistent with the ASME
design approach. To accomplish this validation we parameterize Equation 4 as follows:
Next we fit a model of form described in Equation 5 to the full set of FEA results and extract the
value of parameter “a” that provides the best fit.
Equation 6
Figure 30 and Figure 31 below show the results of the model fit and the 95% confidence
surfaces respectively. The adjusted R 2 of the model is 0.9657 and:
a = 0.2856 (0.2567, 0.3144)
This fitted value of the parameter “a” is valid for all temperatures in the range 32°F – 104°F and
all miter cut angles less than or equal to 22.5°. This is an extremely encouraging result as itshows that the non-linear temperature dependent stress-strain curves used in the analysis
correctly capture the elastic behavior of the elbows across the range of loading conditions, elbowgeometry and temperatures used in the study.
A very simple design equation for the required thicker wall of the elbow results:
. Equation 7
Equation 7 for HDPE mitered elbows reflects the fact that plastic materials are much more
compliant than steel and develop lower stress risers as a result (a=0.2856 as opposed to a=0.643for steel).
Equation 8 can be used directly to calculate the necessary increase in wall thickness as afunction of miter cut angle and pipe dimension ratio, DR=D/t. The results of this calculation for a
range of miter cut angles 2.813° - 22.5° and DR’s ranging from 7 – 32.5 are presented in
Table 3. The Geometric Strength Ratio is the inverse of the wall thickness multiplier and can beused to calculate the appropriate DR for the fabricated elbow.
. Equation 9
The results in Table 3 show that, the current industry practice of using a DR of one standard
dimension less than the pipe for the fabricated elbow, which results in an approximately 25%
increase in wall thickness, is a very reasonable approach and is nominally conservative up to a
miter cut of 22.5° and for DR’s less than or equal to 21.
ASME B31.3 section 304.2.3 requires the designer to address three design equations forcalculating the maximum allowable pressure for a mitered elbow; equations 4a, 4b and 4c. For
elbows with a miter cut less than or equal to 22.5° the lowest pressure calculated from equations
4a and 4b is specified. Equation 4c is similar in form to equation 4a and is intended for use withmiter cuts greater than 22.5°. Only equation 4a has been used in this work, hence it is appropriate
to briefly discuss the validity of this approach.
Figure 32. Mitered Elbow Configuration for Calculations
Figure 33. Equations and Parameter Values Used in Calculations
Figure 32 defines the geometry the calculations reference and Figure 33 shows the actualworksheet used. There is a point at which equations 4a and 4b are equal when:
√ Equation 10
Evaluating Equation 10 in the range of diameters and DR values relevant to HDPE miteredelbows shows that this occurs for very widely spaced elbows where equation 4b (that is derived
for a smooth torus), is valid.
Figure 34 below shows that in the range of miter cut angles and DR values used in the FEAstudy equation 4a always provides the lower operating pressure. All three equations coincide for
Figure 36. Percent Difference Between ASME B31.3, 304.2.3 Equations 4a and 4b with Bend
Radius Set to 2.5 x Pipe Outside Diameter and a=0.2856 (relative to eq. 4a)
Figure 35 and Figure 36 show the percent difference relative to equation 4a for a=0.643 as
defined in the standard and a=0.2856 as calculated in this study. The percent difference issmaller for HDPE elbows reflecting the greater compliance of the material.
The results presented above justify the use of ASME B31.3 equation 4a ( Equation 1 in thisstudy) as the basis for HDPE mitered elbow design and show that the analysis presented in this
Appendices 0 – 17 provide all the necessary detail to evaluate the 16 response surfacesgenerated in this study.
Future Work
The work done in this project has given valuable insight into the fundamental factors influencing
miter bend stresses, however, the models used were simple in terms of geometry (no beads) andmaterial model (nonlinear elastic). For further insight the following are possible in future work:
Modeling of fusion beads.
Evaluation of thermal expansion and contraction.
Evaluation of various end constraints.
Evaluation of loading due to internal flow.
Evaluation of bend radius influence.
Creep analysis of various configurations.
Stress relaxation analysis of various configurations.
Expansion of the temperature range analyzed to include high temperature hydrostatictesting regimes.
Bibliography
1. Green, A. and W. Emmerson, Stresses in a pipe with a discontinuous bend. Journal of the
Mechanics and Physics of Solids, 1961. 9(2): p. 91‐104.
2. Wood, J., A study of single mitre pipe bends. 1983, Paisley College of Technology.