Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Hattori
Takasago R&D Center, Mitsubishi Heavy Industries, Ltd., 2-1-1, Shinhama, 2-Chome, Arai-cho, Takasago, Hyogo Pref., 676 Japan
K. lida
Department of Mechanical Engineering, Sibaura Institute of Technology, Shibaura, Minato-ku, Tokyo, Japan
T. Nomoto
Department of Naval Architecture and Ocean Engineering, University of Tokyo, Hongo, Bunkyo-ku, Tokyo, Japan
M. Sato
Welding Engineering Department, Japan Power Engineering and Inspection Corporation, Akasaka, Minato-ku, Tokyo, Japan
Introduction A power plant consists of many welded components; therefore, it is essential in establishing the reliability of the power plant to maintain the reliability of all welded components. The fatigue failures caused by mechanical vibrations of small-diameter welded joints, which are represented by socket welded joints, are one of the major causes of trouble for the welded parts of the power plant. In order to investigate the fatigue strength and to establish the standard for welding procedures, WSR (Welded Structure Reliability Verification Test) project (Higuchi et al 1995) has been promoted by the Japan Power Engineering and Inspection Corporation (JAPEIC) with financial support from the Ministry of International Trade and Industry (MITI). The four-point bending fatigue tests and other tests were conducted on socket welded joints, varying parameters such as material, pipe diameter, fillet weld shape, etc. It was necessary to investigate the bending stress and residual stress distributions at the toe and root quantitatively and evaluate the effect of those parameters on the fatigue strength systematically, since cracks may initiate at both locations. In this study, a finite element analysis was performed on socket welded joints to obtain the local stress and residual stress distributions. In addition, the residual stresses were calculated from the results of the sectioning method and X-ray diffraction method. The fatigue strength was evaluated by using these analytical results, taking the effects of local stress and residual stress distributions in socket welded joints into consideration.
als, pipe diameters, and fillet weld shapes to obtain the fatigue strength of socket welded joint. The stress analysis for local stress and residual stress distribution and the fatigue test data corresponding to these analyses are described hereinafter. 1 Materials and Specimen. For materials, the carbon steel JIS G3456 STPT410 and the stainless steel JIS G3459 SUS316LTP, which are generally used for pipe materials, were adopted. The chemical compositions and mechanical properties of these materials are shown in Table 1. Figure 1 shows the geometry and dimensions of the fourpoint bending fatigue test specimen, and the analytical model for the socket welded joint. Three dimensions of 20A, 25A, and 50A were adopted for pipe diameters and Sch80 and Schl60 were applied for pipe schedules (thickness of pipe wall). The fillet weld shapes of the socket welded joint are shown in Fig. 2. The convex and concave shape were applied for the fillet weld shape, and the throat depth was 0.85f (t: thickness of pipe wall). To make the throat depth 0.85f, the fillet weld was formed by machining from the socket side edge to the middle of the bead width after building up the weld to about 1.0?. The weld bead sequence is described in the following paragraphs. 2 Matrix for Fatigue Tests and Stress Analysis. The matrix for the four-point bending fatigue tests and stress analysis of socket welded joints are shown in Table 2. For the carbon steel STPT410, six cases where the pipe diameter is 20A, 25A and 50A and the fillet weld shapes are convex and concave were studied. For the stainless steel SUS316LTP, one case where the pipe diameter is 25A and the fillet weld shape is convex was studied. In those cases, the effects of material, pipe diameter, and fillet weld shape on the fatigue strength were evaluated. Local Stress Analysis Elastic FEM analysis was performed to investigate the stress distribution generated at the toe and root of the weld bead using the MARC program. Transactions of the ASME
the PVP Division, September 17, 1996; revised manuscript received June 6, 1997. Associate Technical Editor: Y. W. Kwon.
Downloaded 08 Sep 2008 to 128.206.119.65. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Machined off
SI
Mn
Ni
Or
Mo
T RT
EL
(%)
49 51 38 71
20A
STPT410 25A 0.18 0.25 0.67 0.007 0.001
RT RT
50A
SUS316L
RT
1 Cases and Model for Analysis. Table 2 shows the cases to be analyzed in the local stress analysis. The local stress analysis was carried out for the carbon steel STPT410. It was applied to four cases, which have different pipe diameters (20A, 25A, or 50A) and different fillet weld shapes (convex or concave). Figure 1 shows the dimensions for the numerical model. Fillet weld shape of the numerical model was determined by macroscopic observation of section at the welded part of the fatigue specimen. It is shown in Fig. 2. Figure 3 shows an example of a finite element mesh used for the standard type model (25A/Sch80, convex shape). Half of the specimen was modeled because of symmetry. Both the toe and the root of the welded parts were divided into small elements because of local stress concentrations. The effect of contact between pipe and socket was not considered in the boundary condition. 2 Analytical Procedure. In the elastic stress analysis, a pure bending moment is applied to the numerical model to correspond to the four-point bending fatigue test. The boundary
Machined off
1.48*t"
Clamped
condition of displacement and the loading condition are shown in Fig. 3. One end of the numerical model was completely fixed in the axial direction. On the other hand, the bending moment (M = 5.46 X 10" N-mm) was applied on the other end to generate the axial bending stress of about 20 MPa on the pipe surface. In addition, the symmetrical condition was adopted to the plane of 0-180 deg in the circumferential direction, and the shape of the cross section was held on the end surface of the loaded plane. The material properties of STPT410 and JIS G4051 S25C were used for the pipe and socket, respectively. For the weld metal, properties of the pipe material were used. In the local stress analysis, material properties at room temperature were used. 3 Analytical Results. Figure 4 shows the results of local stress analysis for the standard model (STPT410, 25A/Sch80, convex shape). The axial stress distribution is shown on the 180 deg plane of the welded part in Fig. 4. The highest axial stress appeared at the toe and root of welded part by application of bending moment. The maximum value of axial stress is 56.6 MPa at the toe on the 180-deg plane. It is 2.72 times the nominal axial stress <T on the outer surface of pipe. On the other hand, the maximum axial stress at the root is 1.91 times the nominal axial stress a.
20A
Sch80
25A
Sch80 Sch160
50A
Sch80
300
380
500
n
12 13
<tD1 4D2
40
36.5 31.0 27.2 30.0 39.0
50 50 40
34.0 40.0 47.0 59.0
80 70 60
60.5 70.0 77.0
Table 2
Material
0D3
t1 t2 t3
3.9
5.65
4.5
6.25
6.4
12.25 18.9 18.0
5.5 7.8
12.0 21.8
Fillet weld shape Convex F,S,R F,S,R Nominal diameter
20A
SchSO
25A
SchSO Sen160
50A
SchSO
25A
Sch80
8.8
13.5
9.9
17.0
Schedule
C G
R F
F,S,R
F,R
Concave
S,R
Fig. 1 Geometry and dimensions for fatigue specimen and analytical model of socket welded joint
Downloaded 08 Sep 2008 to 128.206.119.65. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
Completely fixed
I Pipe
^<<j Socket
Root
I.
Nominal Stress On /
20A
25A
SOA
Nominal diameter Fig. 5 Analytical results of local stress for socket welded joints (STPT410, SchBO)
calculated from the results of the sectioning method and X-ray diffraction method to compare with the analytical results.
(b) Enlarged view Fig. 3 Finite element mesh
The values of the stress concentration factor ala at the toe and. root are shown in Fig. 5 for all analyzed cases. The tendency of stress distribution and the location of high level stress in other cases were similar to those for the standard case (STPT410, 25A/Sch80, convex shape). Figure 5 also shows the dependence of local stress on the pipe diameter or fillet weld shape. The local stresses at the toe and root tend to increase with increase in pipe diameter. Furthermore, the local stresses for the concave-shaped model show a tendency to be lower than those for the convex-shaped one. Especially, decrease in the local stress at the toe is remarkable. In addition, for all analyzed cases, the local stress at the toe is higher than that at the root. Residual Stress Analysis Thermo-elasto-plastic analysis was performed on the same shaped model as the fatigue specimen using the MARC program. Residual stress distributions of the welded part after welding were obtained by the analysis. Measurements of temperature and radial deflection were carried out on the standard model to compare with the analysis. In addition, residual stresses were
1 Cases and Models for Analysis. Figure 1 shows the dimensions for the models used in the residual stress analysis. The models for six cases are analyzed to investigate the effects of pipe diameter, schedule, fillet weld shape, and the difference of materials, as shown in Table 2. Fillet weld shapes for the numerical model were shown in Fig. 2. For the convex shape model, the throat depth was formed to be 1.0; by building up and then finished to be 0.85r by machining it off. For the concave type of 25A/Sch80, the throat depth was finished to be 0.85? after welding the third layer at the toe. They have the same shapes as the fatigue specimens or the local stress analysis models. Table 3 shows the welding conditions for each FEM model. These welding conditions are the data obtained when the fatigue specimens were fabricated. 2 Analytical Procedure. Strictly speaking, three-dimensional analysis should be applied for the welding in the circumferential direction of pipe. However, the analysis can be carried out as an axisymmetric welding structure because pipe diameter is small and the welding speed for each layer is almost constant and can be simulated by an axisymmetric model. Actually, the analysis for this kind of weld joint has been carried out by many researchers under axisymmetric condition (Brusl et al, 1981; Josefson et al., 1989; Rybicki et al., 1979, 1982). In this analysis, the nonsteady temperature analysis was performed on the assumption that the weld metal is built up in the
Stress level (MPa) 1 43.9 2 31.3 3 18.6 4 5.9 S - 6.7 6 -19.4 Max. 56.6 Mln. -32.1
Fig. 4 Axial stress distribution under bending moment (STPT410, 25A/Soh80, 0.85t, convex shape)
Downloaded 08 Sep 2008 to 128.206.119.65. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
PaB8 No.
1
20A
SchSO convex
84
120
1.6 2
non
< 300C
2 3
16
53 43
4 A A A A
A..
- 440
481
25A
SohSO
convex
1~4
110
14
45
4,2
non
340
400
STPT410/
1
25A
SchOO concave
14
13.5
71 47 90 80
4,2 non non
< 300C
2 3
1~4
4,2
non
400
250
/B26C
15 14
25A
schieo
convex
180
300
5~8
61 72 46 50 48 67
4,2
4,1.6
SOA
SchSO
convex
1 2 3 4 1
120
16
360
400
SUS316L,
110
14 12
non
<117C
25A
Sch80
convex
- 320
319
90
34
circumferential direction at once. Then, thermo-elasto-plastic analysis was carried out using these results under axisymmetric conditions. The weld bead sequence for each model is shown in Table 3. The buildup depth and the heat input extent were modeled by macroscopic observation of the section at the welded part of fatigue specimen. The analysis was continuously carried out by giving the heat input to the built-up elements of the numerical model, from the first layer to the last. Therefore, the built-up elements of each layer were separated from the base metal in the initial stage. Then, it is connected to the base metal with every buildup and is activated. The heat input obtained from welding condition was given to the welded part instantaneously as an internal heat generation by the body flux. After welding, the surface is cooled by air. The following heat input was given to the built-up part taking the interpass temperature into consideration. The same model utilized in the temperature analysis was used for the residual stress analysis. The thermo-elasto-plastic analysis was carried out by using the results of the temperature analysis. The yield condition of Von Mises and the isotropic hardening rule were applied for the analysis. The analysis for the machined-off model was carried out by separating the nodes, which exist on the boundary between the machined-off part and the welded part, from the weld metal after analyzing the builtup model. The material properties, which depend on temperature, are used in the temperature analysis and the residual stress analysis. 3 Comparison of Temperature Obtained by Analysis With Measured. The temperature and radial deflection were measured for the standard model (STPT410, 25A/Sch80, convex shape) to confirm the suitability of the analytical results. They were compared with the analytical results using temperature analysis. Figure 6 shows the comparison of the measured temperature value with the analyzed one. These are the data for the building up of the first layer. The measured values of the peak temperature and the transient temperature in the process of cooling down corresponded to the analytical results. Figure 7 shows the comparison of the measured radial deflection with the analytical results. The variation of measured radial deflection shows agreement with the analytical results.
4 Analytical Results. An example of analytical results for the standard model (STPT410, 25A/Sch80, convex shape) is described. The bore of pipe tends to contract at the welded part and expand at both sides of the welded part after welding. The distribution of radial deflection obtained by analysis corre-
200
400
600
800
1000
1200
Time t (sec)
Fig. 6 Temperature history (1st pass)
40
60
80
100
120
140
160
180
200
Axial coordinate
Z (mm)
Downloaded 08 Sep 2008 to 128.206.119.65. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
M a t e r i a l Schedule
STPT410 Sch80
Toe
Root
Toe
Root
A
A
20A
25A
50A
To 0- 400
TiUelwek shape
Root
\
V
sponded to that obtained by measurement shown in Fig. 7. The diametral gap between pipe and socket showed a trend to get narrow after welding. Figure 8 shows the distribution of axial residual stress after machining off the weld metal following building up. The distributions of axial residual stress on the outer surface of pipe, the inner surface of the socket, and the cross section of the welded part after machining off the weld metal are shown in Fig. 9. A high level tensile stress was generated at the root. On the contrary, the axial residual stress at the toe was compressive. The location and magnitude of the maximum axial residual stress after welding were almost equivalent to those after machining off the weld metal, except that the latter was slightly higher than the former. The maximum axial residual stress was generated on the inner surface of socket near the root for the standard model (STPT410, 25A/Sch80, convex shape). Additionally, the calculation of residual stress by using the results of the sectioning method and the X-ray diffraction
200
"
\\
Toe V ,
a
-600
i i i
7.8
10
12.5
15
17.5
20
Coordinate
Z (mm)
method were carried out on the fatigue specimens. The results of these measurements are compared with the FEM results shown in Fig. 9. The measured values scattered a little, but corresponded well with the analytical results. As already mentioned, it was found that quite high axial stresses existed at the toe and root according to the residual stress analysis. All the analytical results of residual stress are shown in Table 3. It is observed that the compressive stress tends to be generated at the toe and the tensile stress tends to be generated at the root in many cases. However, the compressive residual stress is generated at the root in some cases. This is related to the location of the final bead and the number of layers. Figure 10 shows the results of residual stress analysis for all cases. The dependence of residual stress on pipe diameter and material is shown in Fig. 10(a). The high level tensile stress is found at the root for the convex type, and it tends to slightly decrease with an increase of pipe diameter. On the contrary, the high level compressive stress is found at the toe and it tends to slightly increase with an increase of pipe diameter. There is little difference in residual stresses between the carbon steel and the stainless steel. The residual stress at the root for the stainless steel is slightly lower than that for the carbon steel. Figure 10(>) shows the dependence of residual stress on fillet weld shape and schedule. For the effect of fillet weld shape, the residual stress for the concave type is compressive at both toe and root. It is greatly changed at the root by the final building up on the toe. For the effect of pipe schedule, the residual stresses at the toe and root are reversed with an increase of pipe schedule. The differences in weld bead dimension and number of layers are estimated to cause the dependence of residual stress on pipe schedule.
..
16
18
20
Coordinate f
Root
Fatigue Tests
The fully reversed four-point bending fatigue tests were carried out in air at room temperature using servo-hydraulic fatigue test equipment. The cycling frequencies were set up in the range Transactions of the ASME
Fig. 9 Residual stresses after machined off and welding ( S T P T 4 1 0 , 2 5 A / Sch80, 0.85f, Convex shape)
Downloaded 08 Sep 2008 to 128.206.119.65. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
I
S
103
O STPT410, 20A/SotlB0, Convex STPT410,25A/SCD80, Convex - - - STPT410,50Scri60, Convex V STPT410, 25A/ScM0, Concave QSUS316L, 2SA/Sch80, Convex I : Toe failure
!3r
HI
J.
Toe Root
Stress distribution Peak stress Secondary stress
A A
Nominal stress On
ul_
108
i.
108
as-"0 1 2
Fatigue life
Fig. 11
Nf
(Cycles)
of 3 to 20 Hz. A stress on the outer surface of the pipe was used as the nominal bending stress. It was calculated by dividing the bending moment M by the modulus of section Z (cross section of the pipe shown in Fig. 1). The fatigue life Nf was defined as the number of cycles when the pressure of nitrogen gas sealed in the specimen decreased from approximately 0.2 to 0.1 MPa by crack propagation through the thickness. Figure 11 shows the S-N curves for five cases of fatigue tests. From these results, the following summary related to the effects of each parameter was obtained: 1 The effect of fillet weld shape on fatigue strength was clearly observed. The fatigue strength for the specimen STPT410, 25A/Schl60 with concave fillet shape (V in Fig. 11) was higher than that for the specimen STPT410, 25A/ Sch80 with convex fillet shape (). Fatigue cracks initiated at the root for most of the convex types, but at the toe for the concave types. 2 The remarkable effect of pipe diameter on fatigue strength was also observed. The fatigue strength for the specimen 20A (O) was the highest of all. The fatigue strength for the specimen 25A () was higher'than that for the specimen 50A ( ) . For almost all specimens of 25A and 50A, failure occurred at the root, but failure occurred at the toe for specimen 20A. 3 The effect of material on fatigue strength was hardly observed. There was little difference of the fatigue strength between STPT410 () and SUS316LTP (ffl). From the foregoing summary, differences in fatigue strength using nominal stress amplitude aa i and crack initiation site were observed to depend on fillet weld shape or pipe diameter. The bending stress distribution and residual stress at the toe and root are estimated to make such differences. Discussion The effects of fillet weld shape and pipe diameter on fourpoint bending fatigue strength were discussed for socket welded joints. For the stress concentration, stress distribution of a welded part was considered. On the other hand, the effect of mean stress was considered by using the results of residual stress analysis. In this case, axial stresses in both analytical results were only used for evaluation of fatigue strength. The stress concentration factors at the toe and root are defined in the form of ratio to nominal stress, as mentioned previously. These values can be affected by the shape and FEM mesh size, so local stresses at the toe and root were calculated by multiplying the secondary stresses by the stress index (MITI, 1994) in the study.
Journal of Pressure Vessel Technology
Figure 12 shows an example of the axial stress distribution at the toe and root of socket welded joint. The secondary stress <7o2 at the toe and root was obtained from the analytical stress distribution. Then, stress amplitude was modified by using the factor, which is calculated by multiplying the ratio of secondary stress er2 to nominal stress amplitude aaX by stress index K. The stress index of 2, which is the value for fillet welded joint shown in Notification No. 501 of MITI was adopted for the convex type. For the concave type, the stress index of 1.2 was used by taking the difference of stress concentration factor and fatigue notch sensitivity factor into consideration because the radius of curvature p at the toe is about 5 mm for the concave type, and it is larger than that for the convex type (p = 0.5 ~ 1.0 mm). Figure 13 shows the S-N curves modified by the foregoing method. The differences between S-N curves are reduced a little compared with that in Fig. 11. However, S-N curves separated into the toe failure group and the root failure one. That is, it is impossible to account for the dependence of fatigue strength on fillet weld shape or pipe diameter by regarding the effect of stress concentration only. The fatigue data were modified by taking the effects of stress concentration and residual stress into account. It is supposed that residual stresses at the toe and root act on fatigue strength as mean stresses. The stress amplitude aa3 corresponding to the mean stress a, = 0 was estimated by using the Goodman diagram. That is, the stress amplitude at the cross point of vertical axis (a, = 0) with the line which contains
2
103
'
""I
11 n j
. .
11 M i |
1111
...0...
1 i
102 104
~i
1 : Toe failure
V.
ffl
ffl-
105
106
107
108
108
Fatigue life
Nf
(Cycles)
Fig. 13 Modified S-N curves for socket welded joints (effect of stress concentration was eliminated)
Downloaded 08 Sep 2008 to 128.206.119.65. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
' ;
'
ll|
1 1 1 1 M|
1 1 1
1 1
0 V EB
STPT410,20A/SchS0,Convtx 8TPT410,2SA)S<;heO, Convex 8TPT410,SOAISchBO, Convax STPI410,aA/Sch160, Concave SUS31L,28AIS(>h80,CMlwx 1: Toe failure
^ f^ e\ - % rL ^^ ^ T
a.= K
carried out, and the differences in fatigue strength due to the joint size, shape of welded part, etc., has been studied. In addition, the effect of residual stress on the fatigue strength has been investigated by comparing with the results of bending fatigue tests. The conclusion is as follows: 1 The local stress at toe and root of the socket welded joint increase with increase of pipe diameter. The local stresses decrease not only at the toe, but also at the root by building up the final bead at the toe. The local stress at the toe is larger than that at the root according to the adopted model. 2 For the residual stress in the welded part of the socket welded joint, large compressive stresses and tensile stresses were found at the toe and at the root, respectively. However, the residual stress at the root is changed to a compressive stress when the final bead is built up on the toe. In addition, in the case of 25A, residual stress became tensile at the toe and compressive at the root with increase of pipe schedule. Those changes are greatly dependent upon the location of final bead layer and the number of layers. 3 The stress level of tensile residual stress at the root decreases with increase of pipe diameter. Besides at the toe, the compressive stress has a tendency to change to a tensile stress. The tensile residual stress at the root of the stainless steel joint is lower than that of the carbon steel due to the difference in yield stress. 4 For four-point bending tests of socket welded joint, the large difference in fatigue strengths was observed by using the nominal stress amplitude of pipe, depending on the fillet weld shape and pipe diameter. However, they can be considerably reduced by taking the effects of stress concentration and mean stress due to residual stress into consideration. From the foregoing, it is estimated that the stress concentration and the residual stress affect the fatigue strength considerably.
"()<*
'
: Stress Index Oai Nominal stress amplitude a 0 j Secondary stress Oa3 Stress amplitude corresponding to om=0
'
108 100
104
106
10
107
Fatigue life
Nf (Cycles)
Fig. 14 Modified S-N curves for socket welded joints (effects of stress concentration and residual stress were eliminated)
UTS (ultimate tensile strength of pipe material) with no mean stress and nominal stress amplitude aa t with mean stress equivalent to the residual stress crR is adopted. The mean stress am is calculated according to the concept of ASME Section III (ASME, 1992) as follows: When the maximum stress crmax (=<7 + aai) is not more than YS (yield stress of pipe material), the residual stress aR acts as the mean stress. If the maximum stress amm satisfies YS umax and aai YS, residual stress aR is taken down to the yield stress. From the foregoing, the stress amplitude aa modified by taking the effects of stress concentration and residual stress into account can be expressed by the following equation:
Oa = K
Acknowledgment L^lJ
<Ta3
Figure 14 shows the fatigue data rearranged by using the modified stress amplitude aa. These data lay within the narrow scatter band. From the foregoing, it is estimated that both stress concentration and residual stress cause the fatigue strength to depend on fillet weld shape and pipe diameter. The difference in fatigue strength between the carbon steel STPT410 and the stainless steel SUS316LTP for the standard type (25A/Sch80, convex shape) is estimated to be caused by the difference in the residual stress at the root, fatigue notch sensitivity, and so on. Conclusions It has been pointed out that the fatigue strength of a pipe joint with small diameter, represented as socket welded joint which is used for the piping of power plant, is remarkably affected by the residual stress. In this study, the analysis for local stress and residual stress at the welded part has been
The present research was carried out in the joint study of WSR Committee by JAPEIC. The authors would like to express their great thanks to Professor Dr. Fukuhisa Matsuda (Osaka University), Prof. Dr. Shozaburo Ohta (Musashi Institute of Technology), and the other members of WSR Committee for their helpful comments. References
ASME, 1992, Boiler and Pressure Vessel Code, Section III. Brusl, F. W., and Stonesifer, R. B 1981, "Effect of Welding Parameters in Residual Stresses in BWR Piping Systems Report," ERPI NP-1743. Higuchi, T., Hayashi, M., Yamauchi, T lida, K., and Sato, M., 1995, "Fatigue Strength of Socket Welded Pipe Joint," ASME PVP-Vol. 313-1. Josefson, B.L., and Karlson, C. T., 1989, "FE-Calculated Stresses in MultiPass Butt-Welded Pipe-A Simplified Approach," International Journal of Pressure Vessels & Piping, Vol. 38. MITI, 1994, Technical Standard for Nuclear Power Plant, (in Japanese). Rybicki, E. F., Stonesifer, R. B., 1979, "Coputations of Residual Stresses due to Multi-Pass Welds in Piping Systems," ASME JOURNAL OF PRESSURE VESSEL
TECHNOLOGY, Vol. 101.
Rybicki, E. F., McGuire, P. A., Merrick, B., and Wert, J., 1982, "The Effect of Pipe Thickness on Residual Stresses due to Girth Welds," ASME JOURNAL
OF PRESSURE VESSEL TECHNOLOGY, Vol. 104.
Downloaded 08 Sep 2008 to 128.206.119.65. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm