hot tap

ELSEVlER 0308-0161(94)00051-8

hr. J. Pm. Ves. & Piping 68 (1996) 169-180 @) 1996 Published by Elsevier Science Limited

Printed in Northern Ireland. All rights reserved 0308-0161/96/$15.00

Strength of a hot tap reinforced Tee junction

Frank Nippard, Roy J. Pick Department of Mechanical Engineering, University of Waterloo, Canada N2L 3GI

&

David Horsley NOVA Gas Transmission Ltd

(Received 30 May 1995)

The connection of branches to pipelines in service often makes use of a reinforcing saddle assembled in a hot tapping operation. Although widely used, the stress levels in such junctions have not been established. This paper describes, in detail, the stress levels in a hot tapped saddle reinforced Tee junction between a 24 inch branch and 36 inch main pipe. Using the finite element method other geometries are also considered.

1 INTRODUCTION

On January 8, 1992, NOVA Gas Transmission Ltd. experienced a rupture’ on the NPS 36 Western Alberta Mainline at the James River Interchange (Fig. 1). The 24-inch branch of the junction was reinforced by a full encirclement saddle assembled on the main line in a hot tapping operation. The cause of the rupture was identified as a weld defect in the weld between the branch and main pipe at the four o’clock position (4:00) as shown in Fig. 2. During the investigation of the rupture a study to determine the stress levels in the junction was undertaken. This study considered the stress levels created in the weld due to the assembly operation, the operating pressure, and moments applied to the branch. In this investigation the saddle was modelled as being a frictionless sliding fit on the main pipe and welded to the branch at the stub of the saddle.

A finite-element model (Fig. 2) was used to investigate the stress and strain levels in the junction.’ Stresses produced by the hot tap assembly operation and the operating pressure were found to produce small amounts of plasticity in the weld between the branch and the main pipe at the location of the defect. Small

amounts of plasticity should be expected at stress concentrations since the design hoop stress in the main pipe can reach 72% of the yield stress at the operating pressure. Shakedown to elastic behaviour would be expected following hydrostatic testing and safe operation thereafter in the absence of defects.

The junction was in successful operation for a period of 11 years after installation. During the fall of 1991, additional piping was installed on the branch line adjacent to the junction being described (shown as new construction in Fig. 1). Failure occurred shortly thereafter, the first time the pipe reached the maximum allowable operating pressure. A piping analysis considering soil and foundation settlement was used to predict moments applied to the junction. Under the action of these moments the level of plasticity in the junction increased considerably. The predicted plastic strains suggest that the defect in the weld became critical during the growth of plasticity due to settlement. This led to a fracture of the junction and a running crack along the pipeline with a major loss of gas.

This paper provides details of the finite element investigation of the stress levels in the junction including the effect of dimensions, assembly, operating pressure, and moments

169

170 F. Nippard, R. J. Pick, D. Horsley

Fig. 1. Isometric of James River Interchange on NOVA’s NPS36 Western Alberta Mainline.

Fig. 3. Pipe/branch finite element mesh for in-plane moment analysis.

applied to the branch. In a companion paper’ the influence of the fit of the saddle on the pipe is considered and experimental results are described which indicate that the finite-element model is accurate for pressures below 6037 kPa (875 psi), which is the maximum operating pressure.

2 FINITE-ELEMENT MODEL

Figure 1 shows the overall geometry for the James River Interchange geometry. In creating a finite element model of the main pipe, the branch pipe and the saddle, there are two planes of symmetry for pressure loading and therefore only one quadrant of the junction is modelled (Fig. 2). For a moment applied to the branch pipe there is only one plane of symmetry and one half of the junction must be modelled; the plane of symmetry depending on the direction of the moment (Figs 3 and 4).

The area of interest in the analysis is the welded region between the branch pipe and main pipe under the saddle. Figure 5 shows the detailed finite element mesh of this region

Location of defect Main Pipe

Fig. 2. Quadrant of main pipe showing branch pipe and saddle.

between the 12:00 position (the 90” crotch) and the 3:00 position at the side of the branch. The failure in the junction at the James River Interchange initiated in a defect in the weld at approximately 4:00 (30” from the 90” crotch). In this region different materials are considered in the main pipe, the branch pipe, the saddle and the weld metal. Figure 6 shows a cross section through the weld and the linearized stress/strain behaviour measured for each material.

The fit between the saddle and the pipe is given by the geometry of each component, described by the node point coordinates of the elements used in the finite element model. To consider interaction between the inner surface of the saddle and the outer surface of the pipe, interface elements have been used between the coincident points on the pipe and saddle. The interface elements can model various conditions such as frictionless slipping, slipping with friction or a connection between the two components at the point. For the results reported in this paper a perfectly fitting saddle with frictionless sliding

Fig. 4. Pipe/branch finite element mesh for out-of-plane moment analysis.

Strength of a hot tap reinforced Tee junction 171

Fig. 5. Local view of the finite element mesh near the pipe/branch weld.

between the saddle and the main pipe has been assumed. Analyses with friction have shown reduced stresses. However, measurements of the fit between the saddle and pipe on similar

junctions show gaps between the pipe and saddle sufficient to suggest that sliding was probable. In this case frictionless sliding is a reasonable and conservative assumption.

1 PIPE MATERIAL

37%

2 BRANCH MATERIAL

3 WELD MATERIAL

30%

4 SADDLE MATERIAL

4 4 4 lililz ---I- 1

i- LOCATION OF DEFECT

Fig. 6. Material properties for various elements in the finite element mesh.

172 F. Nippnrd, R. J. Pick, D. Horsley

To obtain sufficient resolution of the stresses for fracture mechanics predictions in the weld, it was necessary to use a relatively fine mesh at the junction between the main pipe and branch pipe. Three solid elements through the thickness of the pipe and branch and three solid elements through the thickness of the saddle were used (Fig. 6). While the use of shell elements would be more efficient, resolution of stresses at the junction is difficult and inaccurate. Hybrid shell/solid element models are only slightly more efficient since the majority of elements are the solid elements used at the junction. In addition this modelling approach becomes complicated when contact between shells must be included.

The model was analysed using the commercial ABAQUS finite-element package,” Version 4.85. Unfortunately the large number of interface elements between the saddle and pipes make the analysis highly nonlinear, requiring a large number of iterations. This, with the large number of elements, led to long computing times and difficulty with closure. A successful and efficient solution strategy required considerable development.

2.1 Description of the results-location and orientation of the stresses

Isoparametric rectangular elements with 20 nodes were used in the finite element analyses. These elements provide the stress and strain history at each of 8 integration points within each element. At these points the stresses in the direction of the global coordinates, the principal stresses, the stresses in any specified direction, the Von Mises Equivalent Stress, and elastic and plastic strains are predicted. Thus each analysis provides considerable data. Since the failure at the James River Interchange was due to a weld defect at 4:00 only the elements in the region of the weld between 3:00 to 6:00 are reported in detail. Examination of the results at other locations have shown the 4:00 position to be an area of maximum stress.

3 ASSEMBLY PROCEDURE

The pipe junction is normally loaded by internal pressure. In the situation at the James River Interchange, the junction was also loaded by moments applied to the branch pipe due to settlement of the adjacent piping. The reinforced

Tee junction shown in Fig. 2 is assembled onto the pipe while it is at pressure in a Hot Tapping procedure. Thus the stresses created in the Tee junction are a function of the assembly procedure, the sequence in which the pressure is introduced and the applied moments. Initial analyses of a fully assembled junction loaded only by internal pressures produced stress levels approximately 15% below those predicted when the assembly procedure was considered.

In the finite-element analyses the assembly and loading history of the junction was simulated by five steps which allowed changes to the applied moments and internal pressure and changes to geometry through the addition or deletion of elements. Five steps were considered:

Step 1: The stresses in the main pipe were determined for an internal pressure of 3018 kPa which is one half of the operating pressure. To simulate the welding of the branch pipe to the main pipe the finite elements for the branch pipe are attached to the main pipe but considered stress free. This simulates the attachment of the branch to the pipeline at reduced pressure. Step 2: The pressure in the main pipe is increased to 6037 kPa. Because the branch pipe is attached, this step creates stress in the branch pipe due to the expansion of the main pipe. Step 3: In this step the saddle is added to the model as stress free material and then the branch is pressurized to 6037 kPa to match the pressure in the main pipe. Step 4: The hole is cut in the main pipe and the material removed from the model. The pressure of 6037 kPa is applied to the edges of the hole. This creates stress in the saddle due to ovalization of the branch pipe and contact between the saddle and the main pipe. Completion of this step provides the expected stresses in the junction as a result of the assembly and normal operating conditions.

At the James River Interchange an in-plane moment of 490 kN-m and an out-of-plane moment of 332 kN-m were indicated in a piping analysis as being applied to the branch pipe as a result of soil and foundation settlement (Figs. 3 and 4). The stresses created by these moments would add to the operating pressure stresses. As the next step (Step S), the in-plane moment of 490 kN-m was considered to be acting. This


required that the analysis (from Step 1) make use SP3 Wo)

of a simulation of one half of the Tee junction 1 0 2 60

(Fig. 3). 3 120 4 180

Subsequent consideration of an out-of-plane 5 240

moment would require that the analysis be 6 300 7 360

undertaken from Step 1 with a full model of the 0 420 9 400

junction, since the loading is no longer 10 540

symmetrical. This was not feasible without a 11 600

drastic reduction of the number of elements and interface elements, producing a coarser mesh and less accurate prediction of the stresses in the welds. To determine if this step was necessary, an elastic analysis with an out-of-plane moment of 332 kN-m acting (Fig. 3) was undertaken. The results were considered, in light of the stress and strain levels predicted in Step 5, to give an Fig. 8. Third principal stress at operating pressure of

approximate indication of the stress levels. 6037 kPa, after assembly of the junction. At a section through main pipe at 33” from 3:O0.

4 STRESS LEVELS DUE TO PRESSURE

For brevity the results on the surface of the main pipe and branch pipe at the weld between 3:00 and 6:00 are presented to indicate the stresses at the weld toe. In addition, the stresses through the thickness of the main/branch pipe at a section at 4:00 (30” from the 90” crotch) are presented as the stresses that initiated failure of the defect. The three principal stresses are considered: SP3 the maximum tensile principal stress which is generally aligned with the weld direction, SP2 and SPl which are generally perpendicular to the weld, with SPl being radial and SP2 being in the plane of the pipe. SP2 is the stress perpendicular

SP3

2 3 4 5 6 7 0 9 10

Wo) 0

60 120 180 240 300 360 420 460 540

to a long weld defect and is expected to be the critical stress. The principal stresses are only generally aligned with the weld. However transformation to the exact weld direction only altered the magnitude of the stresses slightly in the region from 3:00 to 4:O0.

Figures 7 and 8 show the third principal stress, SP3, resulting from the completion of the assembly operation with an internal pressure of 6037 kPa acting. Figures 9 and 10 show the second principal stress, SP2, perpendicular to the weld which is believed to have initiated the failure.

In considering the combined stresses indicated

SP2

2 50 3 100 4 150 5 200 6 250 7 300 0 350 9 400

10 450

ow 0

Fig. 7. Third principal stress at operating pressure of 6037 kPa, after assembly of the junction. On outside of main

pipe between 3:00 and 6:O0.

Fig. 9. Second principal stress at operating pressure of 6037 kPa, after assembly of the junction. On outside of main

pipe between 3:00 and 6:O0.


SP2

2 3 4 5 6 7 8 9 10 11

(MPa) -100

-50 0

50 100 150 200 250 300 350 400

Fig. 10. Second principal stress at operating pressure of 6037 kPa, after assembly of the juction. At a section through

main pipe at 33” from 390.

by a Von Mises stress, it is found that a small amount of plasticity occurs at the toe of the weld due to the operating pressure. This must be considered approximate as the amount of plasticity is small and as can be seen in Fig. 6, the weld toe is represented by the main pipe material with a yield strength of 383 MPa (Figs. 11 and 12).

5 STRESS LEVELS DUE TO APPLIED MOMENTS

Figures 13, 14 and 15 show the three principal

MISES (UPa)

2 160 3 189 4 218 6 275 8 332

10 390

Fig. 11. Von Mises stress at operating pressure of 6037 kPa, after assembly of the junction. On outside of main pipe

between 3:00 and 690.

MY (MPo) 96 2 134 3 172 4 210 5 240 6 286 7 324 8 362 9 400

10 438

Fig. 12. Von Mises stress at operating pressure of 6037 kPa, after assembly of the junction. At a section through main

pipe at 33” from 3:O0.

stresses (SPl, SP2, SP3) on the outside of the weld between the branch and main pipe after a moment of 490 kN-m is applied in-plane. Figures 16, 17 and 18 show the principal stresses through a section at 4:O0. Under the action of the moments the plastic zone increases significantly and the high strength weld material (yield strength = 460 MPa) yields. Figure 19 shows the equivalent plastic strain on the outside of the junction. The plastic zone is outlined by the outer contour in Figure 19.

Figures 20 to 22 show the elastic stress levels that were predicted with only the out-of-plane

SPI Wa) 1 -300 2 -260 3 -220 4 -180 5 -140 6 -100 7 -60 8 -20 9 20

10 60

Fig. 13. First principal stress at operating pressure of 6037 kPa, after assembly of the junction and application of a 490 kN-m in-plane moment. On outside of main pipe

between 3:00 and 6:O0.


SP2 Wa) 1 0 2 50 3 100 4 150 5 200 6 250 7 300 8 350 9 400

10 450 11 500

Fig. 14. Second principal stress at operating pressure of 6037 kPa, after assembly of the junction and application of a 490 kN-m in-plane moment. On outside main pipe between

3:00 and 6:O0.

moment of 332 kN-m applied. Adding these elastic stresses to the previous elastic-plastic stresses from the assembly, pressure, and in-plane moment is not valid. However as can be seen in Figs. 20 to 22 the stress levels resulting from the out-of-plane moment are relatively small compared to the stresses after Step 5 in the region from 3:00 to 4:O0. Therefore the out-of-plane moment will increase the size of the plastic zone somewhat and increase the plastic strain levels at the 4:00 location. Determination of the magnitude of the increase would require considerable additional computation and this was not felt necessary as the increase would be small

SP3 ww 1 0 2 60 3 120 4 180 5 240 6 300 7 360 8 420 9 480

10 540

Fig. 15. Third principal stress at operating pressure of Fig. 17. Second principal stress at operating pressure of 6037 kPa, after assembly of the junction and application of a 6037 kPa, after assembly of the junction and application of a 490 kN-m in-plane moment. On outside main pipe between 490 kN-m in-plane moment. At a section through main pipe

3:00 and 6:O0. at 33” from 3:O0.

SPl WW 1 -300 2 -240 3 -180 4 -120 5 -60 6 0 7 60 8 120

Fig. 16. First principal stress at operating pressure of 6037 kPa, after assembly of the junction and application of a 490 kN-m in-plane moment. At a section through main pipe

at 33” from 3:O0.

and would not change the conclusions of the study.

6 SUMMARY OF STRESS LEVELS

To summarize the behaviour of the junction at the location of the weld defect the stress levels at the weld toe at 4:00 have been calculated by averaging the stresses in the finite elements at this location. For the various steps the stresses perpendicular to the weld (termed SF?) and the equivalent plastic strains are indicated:

Step 1: The pressure was 437.5 psi and the

SP2 NW 1 -100 2 -40 3 20 4 80 5 140 6 200 7 260 8 320 9 380

10 440 11 500


SP3 0.w 1 0

2 60 3 120 4 180 5 200 6 300 7 360 8 420 9 460

10 540 11 600

Fig. 18. Third principal stress at operating pressure of 6037 kPa, after assembly of the junction and application of a 490 kN-m in-plane moment. At a section through main pipe

at 33” from 3:O0.

branch was attached to the main pipe as stress-free material.

Stress perpendicular to the weld = 33.6 MPa Equivalent plastic strain = 0%

Step 2: The pressure in the main pipe is increased to 6037 kPa.

Stress perpendicular to the weld = 70.0 MPa Equivalent plastic strain = 0%

Step 3: The saddle is added to the model as stress-free material. The branch is pressurized to 6037 kPa.

Stress perpendicular to the weld = 214.7 MPa Equivalent plastic strain = 0.0738% Step 4: The hole is cut in the main pipe.

EQUIVALENT PLASTIC STRAIN (/.i) 1 1260 2 2530 3 3600 4 5070 5 6340 6 7610 7 8080 8 10100

Fig. 19. Equivalent plastic strain indicating the extent of plastic deformation after the application of a 490 kN-m

Fig. 21. Second principal stress. Elastic material properties,

in-plane moment. no saddle/pipe gap, no internal pressure and a 332 kN-m

out-of-plane moment applied.

SP3 0-J) 1 0 2 70 3 140 4 210 5 200 6 350 7 420 a 490 9 560 10 630 11 700

Fig. 20. Third principal stress. Elastic material properties, no saddle/pipe gap, no internal pressure and a 332 kN-m


Stress perpendicular to the weld = 231.2 MPa Equivalent plastic strain = 0.125% Step 5: An in-plane moment of 490 kN-m is applied to the branch.

Stress perpendicular to the weld = 336.6 MPa Equivalent plastic strain = 0.458%

As the junction is assembled and pressurized some plasticity occurs at the toe of the main pipe/branch pipe weld at the 4:00 position. An equivalent plastic strain of 0.125% occurs in the parent metal at the weld toe while the adjacent weld metal remains below its yield strength. Significantly, the addition of the 490 kN-m

SP2 1 2 3 4 5 6 7 8 9

10 11

Wa) 0

30 60

19200 150 160 210 240 270


SPl w4 1 -70 2 -59 3 -48 4 -37 5 -26 6 -15 7 -4 a 7 9 ia

10 29 11 40

Fig. 22. First principal stress. Elastic material properties, no saddle/pipe gap, no internal pressure and a 332 kN-m


in-plane moment considerably increases the plasticity resulting in equivalent plastic strains of 0.458%. It is expected that the out-of-plane moment of 332 kN-m will increase this further although the effect will not be as great as with the in-plane moment.

Considering only the in-plane moment (Step 5) the total principal strains acting at the toe of the weld at the 154 position are: O-431%, 0.176% and -0.481% in the three principal directions with an indication of a significant gradient through the pipe wall. It is the combination of these strains minus the elastic portion that results in the equivalent plastic strain reported.

6.1 Failure scenario

Chiovelli, Dorling, Glover and Horsley’ determined that the lower bound CTOD of the NPS 36 WAML base material was 0.034mm at -5°C. Using Appendix K of CSA Z184-M 864 the strain required to initiate a failure from the defect found was approximately 0.11%. Previous work” indicates that this strain has a factor of safety of approximately 2. Therefore Chiovelli et al. stated that: ‘The strain required to propagate the observed pre-existing defect was concluded to be 0.22%. This implies that the local stress at the toe of the stub weld must have exceeded the yield stress of the material just prior to failure.’

Their findings appear to be supported by the finite-element results. At the operating pressure the equivalent plastic strain (0.125%) is below the critical strain of 0.22%. However as moments

are applied to the branch, strain levels increase well beyond this value suggesting that during settlement of the pipe the strains exceeded the critical value and fracture resulted.

6.2 Comments on the operating stress levels

The pipeline was operating at 72% of the nominal yield strength of the material and is expected to experience small amounts of plasticity at stress concentrations. Thus the predicted plasticity at the toe of the weld, after assembly and at the operating pressure is expected and is acceptable if the material is ductile and contains only small defects. However the analyses have shown that applied moments significantly increase the plasticity in the branch pipe/main pipe weld making it less tolerant of defects.

The most probable scenario of failure began with the creation of a weld defect during construction. This defect was below a critical size with respect to the normal assembly and operating stresses. Piping analysis has indicated that moments would be gradually applied to the branch pipe by soil settlement, ultimately reaching the values used in the analysis. It is believed that as the plasticity and strain levels increased the weld defect became critical and failure resulted.

The results reported above are for a 24/36 inch branch to main pipe diameter ratio and assuming a perfectly fitting saddle. Measurements on similar saddles have shown significant variation in the fit of the saddle on the main pipe. This has also been investigated and is reported in.2

7 INFLUENCE OF GEOMETRY

NOVA Gas Transmission Ltd. of Alberta, operates a system of approximately 18,400 kilometers within the province of Alberta, carrying 78% of all marketed Canadian gas each year. Within this system there are numerous hot tapped reinforced Tee junctions of various sizes. To study the effect of diameter ratio, a series of elastic analyses were undertaken considering a 36 inch diameter, 10.7 mm thick, main pipe, with various diameter branch pipes and saddles attached. For each analysis it was assumed that the saddle fitted perfectly, the material was elastic, and the junction was only loaded by a


pressure of 6454 kPa (creating a pipe hoop stress of the weld, at the 3:00 location (at the 90 of 72% of the yield strength). crotch, Fig. 2) are presented.

At the junction between the branch and the main pipe the weld size on the outside of the branch was set to 1.0 times the wall thickness of the thinner pipe and on the inside was set to one half this size. This gives a reasonable and consistent geometry in the crotch and approxim- ates the James River Interchange geometry. In general, the weld toe angle has a greater influence on the local elastic stress concentration than the size of the weld. This angle was set to 45 degrees for all analyses. The length of the branch stub on the saddle was fixed at the value used for the James River Interchange geometry. The local curvature of the inside surface of the saddle was varied so that the distance from the line of intersection of the branch and main pipe and the inside surface of the saddle was fixed. The diameter of the hole cut in the main pipe varied with branch size and was based on the outside diameter of the actual hot tap cutters used by NOVA.

The values of SP2 are summarized on Figs. 23 to 26 at a function of the branch diameter, saddle thickness and strength ratio. Note that a saddle thickness of 0.0 mm indicates the results for an unreinforced junction. All figures indicate that the presence of the reinforcing saddle substantially reduces the stress levels in the junction. Where stress levels exceed the yield strength of the material, plasticity is indicated with higher stress levels generally indicating larger plastic zones.

It was noted that in the case of thick saddles and thin branch pipes the maximum stress can occur at the toe of the weld of the saddle stub to the branch pipe.

7.2 Strength ratio = 1-O

A parametric study was undertaken with three major variables: saddle thickness, branch to main pipe diameter ratio and strength ratio. The saddle thickness was varied between O-0 mm (no saddle) to 22.0 mm (James River Interchange geometry). The strength ratio is the ratio of the nominal hoop stress in the main pipe to the nominal hoop stress in the branch pipe as defined by:

PDR 2t, D&s SR=-=---

PD, D,t, I -.. a3

Figures 23 and 24 show that the stress perpendicular to the weld (SP2) increases with branch diameter. For the limiting case of no saddle, the second principal stress (Fig. 23) ranges from 280MPa for a 12-inch branch to 825 MPa for a 28-inch branch. The third principal stress (parallel to the weld) remains relatively constant at approximately 1050 MPa indicating plasticity at the weld toe for any unreinforced junction within this size range. For the case of a 22aOmm saddle, the second principal stress ranges from 60MPa for a 12-inch branch to 320MPa for a 28-inch branch with the third principal stress between 360 MPa to 490MPa. Thus the stresses remain elastic for reinforced

where DB = branch diameter; tB = branch thickness; D, = main pipe diameter; tR = main pipe thickness.

In elastic analyses, this definition serves as a useful comparison of the basic stresses in the main and branch pipes. Since the main pipe hoop stress is normally set to a code allowable, this ratio indicates how much the branch pipe is over the code allowable (through increased thickness) compared to the main pipe. Strength ratios of 1-O and 2-O were considered.

7.1 Results from the parametric study

To summarize the results the second principal stress (SP2), perpendicular to the weld at the toe

9OO _ + O.Omm saddle 6.0mm saddle I 2.Omm saddle

-x- 18.0mm saddle

12 14 16 I8 20 22 24 26 28

Branch diameter (in.)

Fig. 23. Second principal stress for various branch diameters (elastic behaviour with a strength ratio = 1.0).


Q 28” branch -+- 24” branch

- 16” branch -x- 12” branch

O- I I I I I 0 5 10 I5 20 25

Saddle thickness (mm)

Fig. 24. Second principal stress for various saddle thicknesses (elastic behaviour with a strength ratio = 1.0).

junctions except for a small amount of plasticity for large diameter branch pipes.

7.3 Strength ratio = 2.0

Figures 25 and 26 show that the second principal stress ranges from 150MPa to 450 MPa (22 mm saddle and no saddle) for a 12-inch branch and 320 MPa to 800 MPa for a 2%inch branch.

Compared to the results for a strength ratio of 1.0, it is clear that for large diameter branches the strength ratio (i.e. branch thickness) has little effect on the second principal stress, but for small branches, thicker branch walls result in higher stresses in the main pipe. Since the second principal stress is primarily caused by local bending, a thicker walled branch will transfer more bending to the main pipe increasing the stress in the main pipe adjacent to the weld. This

e O.Omm saddle

-x- 18.0mm saddle

12 14 16 18 20 22 24 26 28

Branch diameter (in.)

Fig. 25. Second principal stress for various branch diameters (elastic behaviour with a strength ratio = 2.0).

VI 600-

+ 28” branch -+- 24’ branch -*- 16” branch -x- 12” branch

Ioil I I I I I 0 5 IO 15 20 25

Saddle thickness (mm)

Fig. 26. Second principal stress for various saddle thicknesses (elastic behaviour with a strength ratio = 2.0).

result disagrees with the often-used design philosophy based on ‘equivalent area reinforcement’, which promotes thicker branches as a means of increasing local reinforcement.

Within the various assumptions made of the detailed geometry the results indicate that the stresses become lower with smaller diameter ratios. It is expected that this conclusion would remain after considering the effect of assembly, fit and moments applied to the branch.

8 CONCLUSION

The finite-element method has been used to investigate the stress levels in the NPS 36 reinforced Tee junction that failed at the James River Interchange. The results of strain gauge and dial indicator measurements made on a similar junction removed from service and pressurized to burst showed that the finite element-model predicts the stress and strain levels in the pipe at pressures below 6037 kPa reasonably well.’ As this pressure corresponds to the maximum operation pressure, the finite- element results should accurately predict the stress and strain levels that existed in the junction at the time of failure.

For the analyses it was assumed that the saddle fit perfectly over the main pipe but was free to slide. To determine the operating stresses the assembly operation has to be modelled, as the hot tapping operation allowed the pressure to vary as the saddle was assembled over the junction and the hole in the main pipe was cut. It was found that after assembly, upon reaching the


MOP of 6037 kPa, a small amount of plasticity had occurred in the weld between the branch and saddle with an equivalent plastic strain level of 0.125% at the location at which fracture was believed to have initiated. Subsequently if an in-plane moment of 490 kN-m was applied to the junction the equivalent plastic strain level increased to 0.458%. Unfortunately it was not possible to consider a subsequent out-of-plane moment; however an elastic analysis indicated the plastic strain levels would increase further. The final strain level would substantially exceed the estimated critical strain of 0.22% for the defect. Thus it is believed that the failure resulted from the gradual build-up of moments resulting from the soil and footing settlement of some adjacent construction.

Small amounts of plasticity should be expected in the pipelines at areas of stress concentration as the general hoop stresses may reach 72% of the yield stress. At these locations shakedown is expected to occur and elastic behaviour result thereafter. This is undoubtedly the case in the James River Interchange even in the presence of the weld defect that was identified. However the addition of in-plane and out-of-moments increased the plastic strain levels considerably. It is believed that with the additional strains the defect became critical leading to the failure. Thus failure occurred some years after the initial assembly and pressurization as a result of moments applied to the branch by adjacent construction.

Measurements of similar junctions removed from service has indicated a significant gap or clearance between the main pipe and the saddle. This has been shown to increase the stress levels when the junction is loaded by pressure.* The actual clearances in the failed junction are not known; however, they undoubtedly contributed to somewhat higher stresses. Stress levels were found to be a function of the assembly procedure and consideration should be given to the manufacturing tolerances and assembly procedure of hot tapped reinforced Tee junctions to ensure that the fit is accurate. Considerations should be given to the use of grout if an accurate

fit cannot be achieved. The finite-element analyses indicated that the

stress levels are highest for the largest branch diameter to main pipe diameter and the largest diameter pipe. Thus the junction at the James River Interchange with a 36-inch main pipe and 24-inch branch pipe can be considered one of the most critical reinforced junctions in the NOVA pipeline system.

In junctions of this size the stress/strain levels are high under the normal operating pressure. In part the high stress levels in the junction are a result of the hot tapping assembly procedure and the necessary dimensional tolerances for the assembly operation. Unfortunately because of the high stress levels the junctions will have limited tolerance of defects and moments applied to the branch. At the James River Interchange a combination of a weld defect, high moments from the branch piping, and possible poor fit combined to raise the stress/strain levels beyond a critical level for fracture of the defect.

It must be concluded that care must be taken in the use of hot tapped reinforced junctions to avoid factors that will increase an already highly stressed fitting.

REFERENCES

1.

2.

3.

4.

5.

6.

Chiovelli, S., Dorling, D. V. Clover, A. G. & Horsley D. J., NPS 36 Western Alberta Mainline Rupture At James River Interchange, Proceedings 8th Symposium on Line Pipe Research, (1993) AGA, Houston, September. Pick, R. J. & Nippard, F., Summary Report, Stress and Strain Levels in the Pipeline Junction form the James River Interchange, (1994). Department of Mechanical Engineering Report, University of Waterloo. ABAQUS Users Manual, Version 4.8.5, (1994) Hibbitt, Karlsson and Sorenson, Pawtucket, Rhode Island, USA. CSA Z184-M86 Standard, Canadian Standards Associa- tion, Rexdale, Ontario, Canada. Glover, A. G., Coote, R. I. & Pick, R. J., Engineering Critical Assessment of Pipeline Girth Welds, Proceeding of Conference on Fitness for Purpose Validation of Welded Construction, London, (1981), The Welding Institute. Nippard, F., Strength of Saddle Reinforced Pipeline Junctions, MASc Thesis, (1995) Department of Mechani- cal Engineering, University of Waterloo.