Sie sind auf Seite 1von 9

Proceedings of the Sixth International Conference on Biaxial/Multiaxial Fatigue &

Fracture, 25-28 June 2001 –Lisboa, Portugal, Volume 1, pps. 363-370.

Code Based Fatigue Life Prediction of a Pressure


Vessel and its Validation

De Jesus, A.M.P.1, Ribeiro, A. S. 1, Fernandes, A. A.2


1
Departamento de Engenharias, Universidade de Trás-os-Montes e Alto
Douro, Quinta de Prados, 5000 Vila Real, Portugal
2
Instituto de Engenharia Mecânica – IDMEC, Faculdade de Engenharia da
Universidade do Porto, Rua Dr. Roberto Frias, 4200-465 Porto

ABSTRACT: In this work the fatigue assessment of a real size pressure vessel subject to
internal pressure fluctuations will be carried out. A comparison of the fatigue life of the
pressure vessel obtained using well established design by analysis techniques proposed in
ASME VIII – Division II, BS 5550 and AD-Merkblätter, and the new European CEN
procedures was undertaken. The stress analysis proposed in the design codes will be
performed using FEM applying submodeling techniques for calculating structural and peak
stresses in hot spot locations of the pressure vessel. The stresses obtained using the finite
element modelling were validated through a real size experimental test.

INTRODUCTION

Fatigue are not the most common cause of failures of pressure vessels, but
designers should always take it into account due to the possible serious
consequences of the failures of this type of structures and due to the fact
that all pressure vessels experience some kind of fatigue loading. Usually
pressure vessels are designed against fatigue, using well-established
methodologies included in design codes such as ASME VIII-Division 2 [1],
BS 5500 [2] and AD-Merkblätter [3]. New methodologies for fatigue
assessment of pressure vessels are now available in the CEN pre-standard
prEN1345-3 [4]. In this work the fatigue assessment of a real size pressure
vessel is carried out using previously referred methodologies. All these
methodologies require the evaluation of the stress distributions in the
critical details of the pressure vessel, normally welded joints. This task was
performed using linear elastic finite element calculations applying
submodeling techniques. This technique allows the structural details
modelling using solid elements, applying boundary conditions imported
from a global model based on shell elements. The numerical stresses were
compared with those obtained from strain measurements. Postprocessing of
stress results was required in order to determine fatigue actions as defined in
the various design procedures. The pressure vessel was subjected to a
fatigue test, applying successive start-up/shut-down cycles. The results of
this test were compared with code based predictions.

PRESSURE VESSEL DESCRIPTION

The pressure vessel was constructed using P420M 5 mm steel plate,


accordingly EN 10028. This steel presents a yield tensile stress of 420 MPa
and an ultimate tensile stress varying between 500 and 660 MPa. The
Young modulus and the Poisson coefficient were assumed to be equal to
210 GPa and 0.3 respectively. Figure 1 represents the pressure vessel, which
has two ellipsoidal heads and is mounted on two symmetrically placed
wood saddle supports. The pressure vessel has two types of seam welded
joints: two circunferencial joggle joints (C1 and C2) and one longitudinal
butt welded joint made from both sides (L1). Socket welded and screwed
nozzle connections are used. Two simple bracket attachments welded on the
top of the pressure vessel are used for handling the pressure vessel. The
previous referred details are marked on Figure 1, by means of
circumferences. These details are potential failure locations, so their stress
analyses were carried out.
= 1200 =
1000

±19
5m
ini.
Ø1250 ext.

75
25

326
5

326 20 1500 346


346 2000
2692

Figure 1: Pressure vessel dimensions.


NUMERICAL AND EXPERIMENTAL STRESS ANALYSIS OF
CRITICAL DETAILS

The numerical stress analysis of the critical details identified in previous


section was carried out using linear elastic finite element method. This
stress analysis was conducted using a shell-to-solid submodeling technique,
based on St. Venant’s principle. Submodeling is also known as the cut-
-boundary displacement method. The cut-boundary is the boundary of the
actual (solid) model which represents a cut through the global (shell) model.
Displacements calculated on the cut boundary of the global model are
specified as boundary conditions for the actual model [5]. The stress
distributions near to the weld toes were determined and compared with
experimental results obtained from strain gages measurements (SG). While
seam welds (C1 and L1) were instrumented using strip gages including 10
grids, the fillet welds were instrumented using single grid strain gages.

Stress distribution on the circunferencial welded joggle joint (C1)


Figure 2 illustrates the geometry of the joint as well as the mesh used in the
finite element model.

a) b)

Figure 2: Circunferencial welded joggle joint: a) dimensions; b) FE mesh.

Figure 3 represents the stress distributions on the outer surface of the


pressure vessel as a function of the distance from to the centre of the weld,
as defined in Figure 2a, for an internal pressure of 20 bar. The stresses
represented are hoop and meridional stresses. We observe a good agreement
between the numerical and experimental results in head side. The numerical
stresses observed in the cylinder side, are about 25 % greater than
experimental results, although the same evolution is observed.
Stress [MPa]

350 σhoop(SG)
Scirc Sσlong/merid
merid(SG)

300
σhoop(MEF)
Scirc(MEF) σmerid(MEF
Slong/Meri(MEF)
250

200

150

100 Weld profile


50

0
-35 -25 -15 -5 5 15 25 35
Distance to the weld centre [mm]
Figure 3: Stress distribution for the circunferencial welded joggle joint.

Stress distribution on longitudinal butt welded joint (L1)


Figures 4a and 4b represent respectively the geometry of the longitudinal
butt welded joint and the finite element mesh, used to compute the stress
distributions.

a) b)

Figure 4: Longitudinal butt welded joint: a) dimensions; b) FE mesh.

The first numerical models used for this type of welded joint didn’t include
any shape deviations. The obtained results, in the outer surface of the shell,
near the weld toe, are much lower than numerical ones. This is explained
from the secondary bending stresses resulting from an angular
misalignment. As a result of this angular misalignment, the stresses in the
inner surface are higher than expected. The geometry represented in Figure
4a already includes an angular misalignment. Figure 5 plots the numerical
and experimental stresses against the distance to the centre of the weld, in
the outer surface of the shell, for an internal pressure of 20 bar.
σlong(SG)
Slong -exp
Stress [MPa]
400

350 σhoop(SG)
Scirc_exp
300
σhoop(MEF)
Scir(MEF)
250
σlong(MEF)
Slong(MEF)
200

150 σhoopsup
Scir (Inner)
int (MEF)

100 σlong(Inner)
Slong sup. Int. MEF
50
-35 -25 -15 -5 5 15 25 35

Distance to the weld centre [mm]

Figure 5: Stress distribution near the weld toe for the butt welded joint.

The numerical and experimental results are in good agreement, which


means that the angular misalignment included in the model are very close to
the real value. The numerical stresses obtained in the inner surface are also
plotted because higher values are observed.

Stress distribution in the bracket attachment for pressure vessel handling


and socket welded and screwed nozzle connections
The stress distribution was also investigated in two other details namely the
bracket attachment used for handling the pressure vessel and the socket
welded and screwed attachment for nozzle connections. These details are
welded by means of a continuous all around edge fillet. The hoop and
longitudinal stresses were investigated in two main sections, namely a
circunferencial (CS) and a longitudinal section (LS). In Figure 6 the stress
evolution for these sections are plotted against the distance to the weld toe,
for the bracket attachment. In Figure 7 similar results are shown for the
socket welded and screwed attachment. For each one of these details only 2
single strain gages were used to validate the numerical results, placed at a
distance of 3 mm from the weld toe.
Stress [MPa]
400 σScir
hoop(L1)
(CS-MEF) σlong
Slong(L1)
(CS-MEF)
σScir
hoop(L2)
(LS-MEF) σlong (LS-MEF)
S long(L2)
350
σ
Scir (CS-SG)
hoop(EXP-L1) σlong (LS-SG)
S long (EXP-L2)
300

250

200

150

100

50
0 5 10 15 20 25 30
Distance to the weld toe [mm]

Figure 6: Stress distribution for the bracket attachment.

600 σScir
hoop(CS-MEF)
(L1) σlong (CS-MEF)
Slong(L1)
Stress[MPa]

500 σScir
hoop(LS-MEF)
(L2) σlong (LS-MEF)
S long(L2)

400
σScir
hoop(CS-SG)
(EXP-L1) σlong (LS-SG)
S long (EXP-L2)
300

200

100

0
0 5 10 15 20
Distance to the weld toe [mm]

Figure 7: Stress distribution for the socket welded and screwed nozzle.

The stress distributions obtained for the various details show that the
circunferencial welded joggle joint is the less critical. The stress values for
this detail did not exceed the nominal values. For this reason only the last
three details will be considered hereafter. The biggest stress observed in the
last three details occur at the inner surface, for the longitudinal butt welded
joint, and in the outer surface for the two last details, on the weld toe. In any
case the observed maximum values correspond to hoop stresses.
FATIGUE TEST DESCRIPTION

The pressure vessel was subjected to an internal pressure fluctuation


between 2 and 35 bar using water as hydraulic fluid. Each cycle has time
duration of about 3.75 minutes. The pressure test was continuously
registered as well as some strain values. It was observed that the first cycles
produce generalised plastic deformation of the vessel, but after few initial
cycles the elastic shakedown was achieved.
The pressure vessel fails after 5670 cycles, due to a crack initiated in the
inner surface of the pressure vessel, at the weld toe of the longitudinal butt
welded joint, that propagates with an elliptical front shape towards the outer
surface of the shell.

LIFE PREDICTIONS BASED ON DESIGN CODES

In this point the results of fatigue life predictions based on pressure vessels
design codes will be presented. The codes used in this task were the ASME
VIII–Division 2 (alternative rules), the BS 5500, the AD-Merkblätter, and
the new CEN procedures, included in the pre-standard prEN 13445-3. In
this predictions only the detailed assessment procedures were used, and the
details considered refers only to weld details. While the ASME code uses
the total notch stresses in fatigue assessment, the others use the structural
stresses, excluding the effect of the weld profile itself. For multiaxial
proportional stress fields, which is this case, the codes suggest the use of an
equivalent stress based on the Tresca criterion. Since the critical points are
located at the surface the equivalent stress will be equal to the maximum
principal stress. Structural stresses are determined using an extrapolation
technique based on two stress locations (0.4 t e t) [6]. Table 1 shows the
stress values obtained for life predictions and Figure 8 presents the fatigue
lives accordingly to various codes.

CONCLUSIONS

We can conclude that the most conservative code was the AD-Merkblätter,
followed by the new CEN procedures. All codes predict a global life for the
vessel bellow the observed value, although considering the butt welded joint
the component with higher life, which wasn’t verified in the fatigue test.
Design codes are not well suited for predicting fatigue lives involving
multiaxial stress fields, since S-N curves are based on uniaxial tensile
fatigue tests.

TABLE 1: STRESS IN CRITICAL LOCATIONS OF THE PRESSURE VESSEL

Design codes Stress types Stress values [MPa]

Longitudinal butt welded joint: 271


ASME VIII- Division 2- Alternative Rules Alternate notch stress Bracket attachment: 293
Socket welded and screwed: 424
BS 5500 Longitudinal butt welded joint: 482
AD-Merkblätter Structural stress range Bracket attachment: 599
prEN13445-3 (New CEN procedures) Socket welded and screwed: 790
Nº de Cycles

15000 13574 Longitudinal Butt welded joint

12500
Bracket attachment for handling the vessel
10307
Socket welded and screwed nozzle conections
10000 8182
7500
4465 4465
5000 3331
2945
2200
2500 1283 1163 1014
507
0
ASME VIII -D. 2 BS 5500 AD-Merkblätter New CEN procedures
Design Codes

Figure 8: Fatigue life predictions accordingly various codes.

REFERENCES

1. ASME Boiler and Pressure Vessel Code (1988), Section VIII - Rules for
Construction of Pressure Vessels, Division 2 – Alternative Rules.
2. BS 5500 (1997), Specifications for Unfired fusion welded pressure
vessels.
3. AD-Merkblätter, Technical Rules for Pressure Vessels, 1995.
4. prEN 1345-3 (1998), Unfired Pressure Vessels, European Committee for
Standardisation –CEN.
5. ANSYS, Inc. (1999), ANSYS Advanced Analysis Techniques Guide,
Fourth Edition.
6. Niemi, E. (1995), Stress determination for fatigue analysis of welded
components, International Institute of Welding, Abington Publishing.

Das könnte Ihnen auch gefallen