You are on page 1of 8

Proceedings of PVP2007

2007 ASME Pressure Vessels and Piping Division Conference


July 22-26, 2007, San Antonio, Texas

Proceedings of ASME PVP-07:


PVP2007-26138
Pressure Vessel and Piping Conference
July 22-26 2007

PVP200726138

AN APPROXIMATE METHOD FOR DETERMINING THE LIMIT PRESSURE OF


PIPING BRANCH JUNCTIONS WITH d / D 0.5
K S Lee D N Moreton
Materials Performance Centre Department of Engineering
University of Manchester The University of Liverpool
Sackville Street Brownlow Hill
PO Box 88 Liverpool, L69 3GH
Manchester, M60 1QD United Kingdom
United Kingdom

D G Moffat
Department of Engineering
The University of Liverpool
Brownlow Hill
Liverpool, L69 3GH
United Kingdom

ABSTRACT defect free branch junction by assuming it to be a plain cylinder


containing a fully penetrating axial crack, for which there are
A series of finite element computations have been standard solutions available such as cited in the R6 defect
conducted to establish the limit pressures for a range of piping assessment procedure [1] and Miller [2]. The current R6 code
branch junctions with d/D 0.5. The results have been [1] does not have a limit pressure solution for branch junctions.
compared with those obtained using existing assessment
procedures. In general, the latter have been found to be In all that follows, the term limit load or limit pressure is used
conservative. An approximate method for calculating the limit as recommended by Gerdeen [3]. That is, it is the maximum
pressure of a branch junction is proposed. This method calculated load, or pressure, that the component or structure can
supposes that the branch junction may be treated as a plain sustain based upon the assumption that the material is elastic
cylinder containing an axial through-wall crack. The crack (or rigid) / perfectly-plastic and that deformations remain small.
length is dependent upon the type of branch junction. The limit For computations that are performed using finite element
pressure can be established using a modified form of a standard techniques, it is difficult (if not impossible) to find a unique
solution limit load solution for a cylinder containing an axial maximum load. The way in which the limit load or pressure is
through-wall crack. This approximate method is shown to be defined in such circumstances is important. Definitions that
valid for a wide range of junction geometries. use a pressure-vs-deformation plot and establish the
intersection of a line whose gradient is a multiple of the elastic
INTRODUCTION slope with the pressure-vs-deformation curve itself have
become common in recent years. Clearly, the choice of the
Various investigations of the limit pressure and plastic deformation parameter is important and Gerdeen has
pressure of piping branch junctions have taken place at the recommended that for pressure loading, this should be the
University of Liverpool during the recent past. These volume change. However, this parameter is seldom available
investigations have been concerned primarily with the effect of and it has become common practice to use a local dimension
the presence of crack-like defects. Most work has been finite change or a strain in its place. The problem with using local
element (FE) based although there has been some experimental deformation changes is that they do not always represent the
work in support of these FE computations. The work reported global behaviour of the component as a whole. Also, the elastic
here is the consequence of a speculation that it may be possible slope based methods have been shown not to give a unique
to provide a simple way of determining the limit pressure of a representation of the limit or plastic loads [4]. This implies the

1 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


need for developing a limit load technique that is based on a PL PL
global, rather than a local deformation parameter. uncr
An excellent background to this area of study has been given by PL ( cyl)
Muscat et al [5] and in that paper the authors suggested an R, Rm Cylinder or run pipe mean radius
alternative approach based upon a plot of a load parameter and
the the total plastic work. Those ideas were subsequently Ri and Ro Inner and outer radii of cylinder or run pipe
extended by Li and Mackenzie [6] to characterise plastic T Cylinder or run pipe thickness
collapse (gross plastic deformation) using a plot of the
curvature of the load-plastic work curve. For several types of Y Yield strength
structure and material properties, the authors showed that Total plastic work at the limit pressure for
distinct points in these plots correspond to distinct stages in the cracked cylinders and branch junctions
deformation process. Lee et al [7] and Lee [8] proposed a Poissons ratio
plastic work criterion based upon a measure of the normalised
plastic work. The technique proposed in [7] has been employed crack length parameter defined as c/(RT)0.5
here to establish the limit pressures of some uncracked piping
branch junctions.
Subscripts and Superscripts
The results of a finite element parametric study of the limit cyl of cylinder
pressures of a range of piping branch junctions have been given
previously by Lynch [9]. The results of the FE computations L Limit
were compared with those from existing methods and with Mises associated with von-Mises yield criterion
some additional published data. In particular, the
recommendation of Rodabaugh [10] that was based upon the uncrCYL of uncracked cylinder
analysis of Cloud and Rodabaugh [11] and the Inverse Code
Method (creep rupture pressure) [12] were employed together FINITE ELEMENT COMPUTATIONS
with additional comparisons with data available from Robinson
[13] and Zahoor [14]. Some of these data will be used here for The finite element study reported here was part of a
comparison. In particular, the data of Lynch [9], Rodabaugh study which included both cracked and uncracked piping
[10] and the Inverse Code Method [12] will be used. branch junctions. Details of the mesh and convergence studies
have been given elsewhere (see e.g. Lee [8]). In brief, as the
NOMENCLATURE branch junction illustrated by Figure 1, is doubly symmetric, a
quarter model was considered sufficient. PATRAN [15] was
Principal Symbols used as the pre-processor for creating the mesh of all the FE
ALLPD total plastic work dissipated in a structure models. Brick elements, type C3D20R, were employed in
ABAQUS [16] and a typical mesh is illustrated in Figure 2.
c Half crack length Boundary conditions of zero displacement in the direction
d Branch pipe mean diameter perpendicular to the planes of symmetry were applied to the
symmetry faces. One node on the run pipe was constrained in
di and do Branch pipe inner and outer diameters three directions to prevent rigid body motion. Axial tensile
r Branch pipe mean radius stresses were applied to the ends of the run and branch pipes to
simulate the effect of end closures. The material was idealised
ri and ro Branch pipe inner and outer radii to be elastic/perfectly-plastic with E = 208 GPa, Y = 308 MPa
t Branch pipe thickness and = 0.28. Small displacement analysis was assumed and
the RIKS [16] algorithm was used to automatically determine
x Half crack length for plain cylinder the load increment.
D Cylinder or run pipe mean diameter Mesh sensitivity studies were undertaken to establish the effect
on the limit pressure of changing the number of elements used
Di and Do Inner and outer diameters of cylinder or run through the thickness of the model, using a coarser mesh in
pipe areas away from the branch junction, changing the length of
E Youngs Modulus both run pipe and branch pipe limbs and the effect of changing
L Half cylinder or run pipe length the element aspect ratio. Following these studies, the FE model
used contained three elements through the thickness of both run
P Applied pressure and branch pipes in the region of the junction for models
having D/T = 20 and 40 and one element through the thickness
for models having D/T = 60, 80 and 120. A typical mesh
arrangement contained 19 elements around the circumference

2 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


of the run pipe and 16 around the circumference of the branch length 2x should be taken as 2 x = d i where di is the inside
pipe in the region of the junction. The length of each run pipe
diameter of the branch pipe.
limb was at least 3D and the branch limb was at least 2D.
Typically, this gave models of the order of 3000 elements.
In order to demonstrate the credibility of the approximation for
junctions having t/T = 1, the approximation will be compared
A previous study [7] has shown that for such branch junctions,
with the FE solutions obtained using the techniques outlined
a reliable technique for assessing the limit pressure is to
above. For the approximation, a modified form of the solution
monitor the (global) plastic work ALLPD which is available as
given in R6 [1] as outlined in Annex A. Figure 9 shows the
computation precedes from the FE code ABAQUS [16]. In
limit pressure obtained using this approximation compared with
order to directly compare the plastic work for different branch
junctions, it was suggested ALLPD be normalised by a the FE solutions. Only data in the range 20 D / T 120
characteristic parameter for the branch junction. The parameter and 0.1 d / D 0.5 has been compared. Outside this
chosen was the plastic work, at the point of plastic collapse, of range, particularly for junctions having d / D 0.5 it has been
an uncracked cylinder (calculated using [17], the Mises yield found that the approximation gives only poor agreement.
criterion and L = D) and multiplied by (ri / T ) . Thus, this
parameter, termed , was defined as: For equal pressure strength junctions, Figure 10 provides a
comparison of the FE solutions obtained using the techniques
outlined above with the proposed approximation. The same
Mises
= ALLPD uncrCYL ri
T (1) range of geometry parameters has been used. The tables
included with Figures 9 and 10 show the errors incurred
calculated on the basis of equation (3) and rounded to the
And the criterion for the limit pressure of any branch junction nearest whole number. Hence, a positive difference implies
taken as that the proposed approximation underestimates the result
obtained using the FE solutions outlined above.
ALLPD = (2)
approximation
% difference = 1 (3)
computation
x 100
The results of these FE computations are compared with data FE
from other sources for equal thickness junctions (i.e. branch
thickness = run pipe thickness). Firstly, Figure 3 is a plot of For Figure 10, it will be noted that no computations have been
PL (the normalised limit pressure) plotted against d/D for a undertaken for those geometries in the lower triangle of the
range of D/T values. Here, the solid lines represent the result of error matrix. This is because these geometries have a very thin
using the recommendation of Rodabaugh [10] and the symbols branch and are considered to be unrealistic from the point-of-
are the result of the FE solutions outlined above. view of welding.

Similarly, Figures 4 and 5 show comparisons with the so-called Figures 11 and 12 show a comparison of the proposed
Inverse Code Method [12] and Lynch [9] respectively. Figures approximation with the FE solutions obtained by Lynch [9] for
6, 7 and 8 provide the same comparisons for junctions having branch junctions having a thicker pipe wall and larger branches.
equal pressure strength (t/d = T/D). This shows that the approximation proposed here generates a
similar error trend. This possibly extends the capability of the
THE PROPOSED APPROXIMATE METHOD AND ITS newly proposed method to cover the estimation of the limit
VERIFICATION pressures of thicker-walled junctions.

The proposal presented here is that for the purpose of CONCLUSIONS


determining the limit pressure, the piping branch junction may
be treated as a plain cylinder that contains a fully penetrating The Rodabaugh [10] recommendation and the Lynch
axial defect of length 2x. The limit pressure of this cylinder [9] data have been found to predict equal thickness branch
may be calculated using solutions given in the R6 defect junction limit pressures which are higher than those found
assessment procedure [1] or Miller [2]. However, it is clear that using the FE computations, reported here. The Inverse Code
the length of the crack to be used in this assessment is critical. [12] procedure has been found to give limit pressures lower
Experimentation with this idea has shown that for equal than those obtained by the FE studies reported here.
thickness branch junctions, the cylinder crack length 2x, should
An approximate method proposed here has been shown to
be taken as 2 x = ri , the inside radius of the branch junction.
allow estimation of the limit pressure to better than 10% for
For equal pressure strength branch junctions, the cylinder crack equal thickness branch junctions for 5 D / T 120 and
0.1 d / D 0.5 . For equal pressure strength junctions, the

3 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


approximation has been found to give acceptable results for [11] Cloud R L and Rodabaugh E C, Approximate
5 D / T 120 and 0.1 d / D 0.3 although for Analysis of the Plastic Limit Pressure of Nozzles
d / D 0.3 the error becomes significant. in Cylindrical Shells, Trans. ASME, J. Eng.
Power, vol 90, p 171 176, 1968.
In general, it has been found that the approximation proposed [12] Assessment Procedure R5, Calculation of
here has a smaller error than the Inverse Code method [12] for Reference Stresses and Stress Intensity Factors,
the range of branch junctions investigated. vol 7, Issue 2, Appendix A3 (draft), 2000.
[13] Robinson M, Lower Bound Limit Pressures for
the Cylinder-Cylinder Intersection: A Parametric
ACKNOWLEDGMENTS
Survey, Trans. ASME J. Pressure Vessel
Technology, vol 100, p 65 73, 1978.
The work reported here was supported by British Energy
[14] Zahoor A, Ductile Fracture Handbook, Report
Ltd. on behalf of the Industry Management Committee
EPRI NP 6301 D/ N14, 1 3, Electric Power
(HSE(NII), British Energy and BNFL). The authors
Institute, Palo Alto, CA, 1991.
acknowledge the support and advice of John Bouchard at
[15] PDA Engineering, PATRAN, version 7, Santa
British Energy throughout the project.
Ana, California.
[16] Hibbitt, Karlsson and Sorenson Inc.,
REFERENCES
Abaqus/Standard Users Manual, Version 5.8,
1998, Vol. 1-3.
[1] British Energy Generation Ltd, Assessment of
[17] Lee K S, Moreton D N and Moffat D G, The
the Integrity of Structures Containing Defects,
Plastic Work Required to Induce the Limit
R6, Revision 4, Gloucester, UK, 2006.
Pressure of a Plain Cylinder, Int J Pressure
[2] Miller A G, Review of Limit Loads of Structures
Vessels and Piping, vol 82, 2005, p 115 - 121.
Containing Defects, Int. J. Pressure Vessels and
Piping, 1988, Vol. 32, p 197 327.
ANNEX A
[3] Gerdeen J C, A Critical Evaluation of Plastic
Behaviour data and a United Definition of Plastic
The limit load cited in the R6 procedure [1] for a cracked plain
Loads for Pressure Components, Welding
cylinder is defined by:
Research Council Bulletin 254, 1979, p 1 63.
[4] Moffat D G, Hsieh M F and Lynch M, An
Assessment of ASME III and CEN TC54 Methods PL R 1
of Determining Plastic and Limit Loads for =
Pressure System Components, J of Strain YT M ()
Analysis vol 36, No 3, IMehcE, 2001, p 301 312.
[5] Muscat M, Mackenzie D and Hamilton R, A where M () = 1 + 1.05 2 (A1)
Work Criterion for Plastic Collapse, Int. J. of x
Pressure Vessel and Piping, 80, 2003, p 49-58. and =
[6] Li H and Mackenzie D, "Characterising gross RT
plastic deformation in design by analysis", Int J
Pressure Vessel and Piping 82 (2005) 777-786. This equation has been evaluated against the results of an
[7] Lee K S, Moreton D N and Moffat D G, A Limit extensive FE study and various modifications have been
Pressure Criterion for Cracked Cylinders and suggested to improve the results produced by this equation.
Piping Branch Junctions", ASME Pressure Vessel These have been employed in the current work. They are:
and Piping Conference, PVP-Vol. 481, RPV
Integrity and Fracture Mechanics, ed. Moinereau,
D., San Diego, 2004, 93-99. a. Rm has been used as the radius term R (rather than
[8] Lee KS, The Influence of Attachment Size on the Ri).
Limit Pressures of Defective Piping Components, b. The von Mises yield criterion has been employed
PhD Thesis, University of Liverpool, 2003. (rather than Tresca).
[9] Lynch M A, Limit Loads of Piping Branch c. The coefficient 1.05 has been changed to 1.4.
Junctions with Cracks, Thesis submitted for PhD,
University of Liverpool, June 2001. This results in the following:
[10] Rodabaugh E C, A Review of Area Replacement
Rules for Pipe Connections in Pressure Vessels
and Piping, WRC Bulletin 335, 1998.

4 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


PL R m 2 1 1.1
=
YT 3 M ( )
1
where M( ) = 1 + 1.4 2 (A2)
0.9
x
and =
R mT 0.8

P'L
do 0.7
d
di
0.6

T 0.5
t
D/T = 20
0.4 D/T = 40
D/T
0 = 60 0.1 0.2 0.3 0.4 0.5 0.6
Do D Di
2c D/T = 80
D/T = 120 d/D
Rodabaugh
2L

Fig. 1 Cross section of a branch junction model Fig. 3 Comparison of the limit pressures obtained through FE
computation with Rodabaugh [10] for equal thickness branch
junctions

1.1

0.9

0.8
P'L

0.7

0.6

0.5

D/T = 20
0.4 D/T = 40
D/T
0 = 60 0.1 0.2 0.3 0.4 0.5 0.6
D/T = 80
Fig. 2 Typical FE mesh for a branch junction D/T = 120 d/D
Inverse Code

Fig. 4 Comparison of the limit pressure obtained through FE


computation with the Inverse Code recommendation [12] for
equal thickness branch junctions

5 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1
1.1
0.9

1 0.8

0.7
0.9
0.6
0.8
0.5

P'L
P'L

0.7 0.4

0.3
0.6
0.2
0.5
0.1
D/T = 20
0.4
D/T = 20 0 D/T = 40
0.2= 40 0.3 0.4 0.5 0.6 D/T
0 = 60 0.1 0.2 0.3 0.4 0.5 0.6
D/T D/T = 80
D/T = 60 d/D D/T = 120 d/D
Lynch correlation Inverse Code

Fig. 5 Comparison of the limit pressure obtained through FE Fig. 7 Comparison of the limit pressure obtained through FE
computation with Lynch [9] for equal thickness branch computation with the Inverse Code recommendation [12] for
junctions equal pressure strength branch junctions

1
0.8
0.9
0.7
0.8
0.6
0.7
0.5
0.6
0.4
P'L

0.5
P'L

0.3
0.4
0.2
0.3

0.2
0.1

0D/T = 20
0.1
D/T = 20 0.2= 40
D/T 0.3 0.4 0.5 0.6
0 D/T = 40 D/T = 60
D/T
d/D
0 = 60 0.1 0.2 0.3 0.4 0.5 0.6 Lynch correlation
D/T = 80
D/T = 120 d/D
Rodabaugh

Fig. 6 Comparison of the limit pressures obtained through FE Fig. 8 Comparison of the limit pressure obtained through FE
computation with Rodabaugh [10] for equal pressure strength computation with Lynch [9] for equal pressure strength branch
branch junctions junctions

6 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1.1 1

0.9
1
0.8
0.9
0.7
0.8 0.6
P'L

0.5

P'L
0.7
0.4
0.6
0.3
0.5 0.2
D/T = 20
0.4D/T = 40 0.1 D/T = 20
D/T = 60
0 0.1 0.2 0.3 0.4 0.5 0.6 D/T = 40
D/T = 80 0
D/T = 60
D/T = 120 d/D 0 0.1 0.2 0.3 0.4 0.5 0.6
D/T = 80
new approx
D/T = 120 d/D
new approx

d/D 0.1 0.2 0.3 0.4 0.5 d/D 0.1 0.2 0.3 0.4 0.5
D/T
D/T
20 -3 -1 -2 -3 -3
20 -5 -6 -3 4 12

40 -1 -2 -3 -1 1
40 -6 1 10 20

60 -2 -3 -2 1 5
60 3 14 23

80 -2 -3 -1 3 7 80 15 25

120 -3 -3 1 6 8 120 27

Fig. 9 Comparison of the limit pressure obtained through FE Fig. 10 Comparison of the limit pressure obtained through FE
computation with that obtained using the proposed computation with that obtained using the proposed
approximation for equal thickness branch junctions. The table approximation for equal pressure strength junctions. The table
shows the difference as defined by equation (3). shows the difference as defined by equation (3).

7 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use


1.1 1

1 0.9

0.9 0.8

0.7
0.8
0.6
0.7
P'L

0.5

P'L
0.6
0.4
0.5
0.3
0.4
0.2
0.3 D/T = 5 D/T = 5
0.1 D/T = 10
D/T = 10 D/T = 20
0.2 D/T = 20 D/T = 30
0 D/T = 50
D/T
0 = 30 0.2 0.4 0.6 0.8 1 1.2
D/T = 50 0.2
D/T = 700.3 0.4 0.5 0.6 0.7 0.8
D/T = 70 d/D new approx
d/D
new approx

d/D 0.25 0.5 0.75 1 d/D 0.25 0.5 1


D/T D/T
5 3 1 -2 -3 5 -3 3 15

10 1 0 2 7 10 -4 9 29

20 1 2 8 17 20 -5 14 37

30 1 4 14 25 30 -5 17 39

50 0 5 17 31 50 -5 20 42

70 0 5 15 34 70 -6 20 45

Fig. 11 Comparison of the limit pressure obtained by Lynch [9] Fig. 12 Comparison of the limit pressure obtained by Lynch [9]
with that obtained using the proposed approximation for equal with that obtained using the proposed approximation for equal
thickness junctions. The table shows the difference as defined pressure strength junctions. The table shows the difference as
by equation (3). defined by equation (3).

8 Copyright 2007 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/29/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

You might also like