Professional Documents
Culture Documents
Abstract
A procedure for design against brittle failure based on a fracture mechanics model is available in the recent European pressure vessel
standard EN 13 445. In the present paper, this model is analysed in the light of more recent findings in fracture mechanics. In particular,
results from the SINTAP project and the Eurocode 3 procedure for selection of steel to avoid brittle fracture are considered.
Alternative models for the scatter in strength, fracture toughness—impact toughness correlations as well as their temperature dependence,
thickness dependence and strain rate dependence are compared. Different approaches for the handling of residual stresses are discussed.
When the effect of elasto-plastic corrections is taken into account both for primary and secondary stresses, the net effect is small. In relation to
more recent findings, the models used as a basis for EN 13 445 seem somewhat conservative, in particular for thinner gauges.
q 2005 Elsevier Ltd. All rights reserved.
strength steels will be proposed on the basis of their true non-redundant components, like the main girders of bridges,
properties and on the required safety level for pressure were selected for the fracture mechanics analysis. A semi-
vessels. elliptical failure geometry ahead of a significant construc-
A number of important new fracture mechanics tools has tional part like a stiffener and perpendicular to the maximum
been developed towards practical application since the load with an a/c ratio of 0.4 was assumed to be the most
model [2] was presented, such as the Master Curve concept critical one. Such an initial defect with a thickness related
[7] and the Failure Assessment Diagram (FAD) [8]. initial depth of a0Z0.5 ln(t/t0) (t0Z1 mm) with a0 in mm
Important results have also been generated in the European was allowed to grow during 500,000 load cycles before it
project SINTAP [9]. The purpose of the present paper is to becomes critical. This is a quarter of the assumed lifetime.
analyse how some of these results can be integrated into the After this time bridges are regularly inspected and a crack
model and how much the resulting minimum design would be found (depending on the part the critical crack
temperatures are affected. Comparisons will also be made depth is about aZ0.25–0.35t).
to the corresponding code for steel structures ‘Fracture In a separate paper the selection of the initial crack size is
avoidance concept for steel structures’—in prEN 1993-2 critically assessed [15].
(Eurocode 3) [10,11], which was developed in parallel to the
pressure vessel code and is also based on fracture 2.2. Loading assumptions
mechanics.
2.2.1. Primary stresses
The calculation of KI to be used in material selection for the
avoidance of brittle fracture requires knowledge of the
2. Basic assumptions and reltations to alternative
maximum principal stress perpendicular to the assumed
approaches
crack. In the design recommendations it was necessary to
assume a stress level, which covers all loading situations that
The model in [2] was proposed as a part of the basis for
might occur during the lifetime. In Eurocode 3, based on
EN 13 445 concerning design against brittle failure. The
typical loading for steel structures, three levels 0.25fy, 0.5fy
assumptions in this model will be analysed in relation to
and 0.75fy (fyZcharacteristic value of the nominal yield
more recent findings.
strength) were selected, representing different risk classes.
In practice only the nominal minimum yield strength ReL,
2.1. The fracture mechanics model
nom provided through the technical delivery condition is
known. This represents a lower bound for steel products.
Fracture mechanics based calculations generally use an Using this value for the load assumption would produce low
initial crack geometry of a certain geometrical size. This stress intensity factors and hence lead to unsafe predictions.
represents an assumed crack size, which might not be seen In [2], the nominal stress is therefore multiplied by a factor
during non-destructive testing and is large enough to cover fscatt of 1.4 to cover the typical range of yield strength values
possible ‘worst case’ scenarios that may occur during the in nominal production, which according to Sandström and
lifetime of a steel construction. Ivarsson [16] lies between 1.2 and 1.6 for low strength
The stress intensity factor of a component of isotropic steels. An alternative is to have a shift sshift. The value of
material is calculated for linear elastic component behaviour sshift is often taken as 100 MPa. This value was used in the
(Linear Elastic Fracture Mechanics, LEFW) case of the Eurocode-approach and covers both scatter and
pffiffiffiffiffiffi remote stresses [11]. For high strength steels this value is
KI Z sappl paYða; 2c; tÞ (1)
supported from studies of data from individual steel
where KI is the stress intensity factor, sappl the component producers. A common analysis of data from several
stress perpendicular to the crack, a crack depth of the design producers seems however to be lacking at the moment.
crack, 2c crack length of the design crack, t the component
width and Y a geometrical correction factor, which is taken 2.2.2. Secondary stresses, CEN-model
from handbooks. In EN 13 445 local residual stresses are fully taken into
For EN 13 445, Eq. (1) is selected using a/tZ0.25, where a account, whereas in Eurocode 3 only remote stresses are
is the crack depth and t is the gauge thickness. For the chosen considered as a constant of plus 100 MPa. The maximum
geometry of the assumed elliptical surface crack a/cZ0.4, residual stresses are regularly—as a simple rule of
Newman–Raju gives a value of YZ1.00092, which can be practice—assumed to have the same magnitude as the
taken as 1 [12]. Newman–Raju’s solution is reproduced in yield strength across the full thickness.2 Consequently, for
[13]. In comparison, the American ASME code [14] assumes
an a/c ratio of 0.33 which gives a Y-factor of 1.05. 2
In reality, the secondary stresses reach the yield strength only at certain
Eurocode 3 contains fatigue design rules, which relate points. The distribution of stresses across the thickness or length is in
typical constructional parts with a certain allowable stress equilibrium, which means that also compressive stresses can be found.
class. The most severe fatigue class, which occurs in Therefore, the proposed assumption is conservative.
874 R. Sandsträm et al. / International Journal of Pressure Vessels and Piping 82 (2005) 872–881
the as-welded (AW) condition, the maximum stress sappl,AW of the role of residual stresses. Based on the actual
will be the real (typical) yield strength of the material knowledge of residual stresses, several alternative
approaches are given. In the general design situation that
sappl;AW Z fscatt ReL;nom C sshift (2)
is considered here, the residual stresses are typically not
where either fscatt should be different from unity or sshift known. In that case if the expression in [19] is transferred to
different from zero. This holds also for the following the same form as Eq. (4) it can be written as
Eqs. (3)–(6).
ReL;nom
In the case of post-weld heat treatment (PWHT), the sres;AW Z fscatt K ReL;nom C sshift (5)
residual stresses are relieved to a certain extent. After sf fs
complete PWHT the residual stresses are typically of the where sf is the flow stress, which is often taken as the
order of a quarter of the maximum, namely the typical yield- average of the yield and tensile strengths. Eq. (5) gives a
strength [2]. Therefore, if the primary and secondary slightly more conservative value than Eq. (4).
stresses are added, the maximum stress sappl,PWHT on the BS 7910:1999 [20] recommends that in structures that
crack is obtained have been fully post-weld heat treated, the axial residual
sappl;PWHT Z ReL;nom =fs C ðfscatt ReL;nom C sshift Þ=4 (3) stress component is a factor 4Z30% of the yield strength of
the material in which the defect is located for stresses
where fs is a safety factor, which is chosen as 1.5 for parallel to the weld and 4Z20% of the lesser yield strength
pressure vessels. The first term on the right hand side of of the weld and parent material for stresses transverse to the
Eq. (3) is the maximum allowed design stress and the weld
second term an estimate of the maximum residual stress. For
non-welded components the last term in Eq. (3) should be sres;PWHT Z 4ðfscatt ReL;nom C sshift Þ (6)
left out. The primary stress is the same in the as-welded case
and consequently the secondary stress is then As can be seen, these values are quite close to those in
Eq. (3) where 4Z25%. We can conclude that these British
sres;AW Z ðfscatt K1=fs Þ=ReL;nom C sshift (4) procedures are almost identical to those in the CEN-model.
A linear superposition of primary and secondary stresses
is assumed. Fig. 1 shows a comparison of the stress 2.2.4. Secondary stresses, SINTAP
assumption for different strength pressure vessel steels. It In the Brite-Euram project SINTAP, an extensive study
becomes clear that the fscatt term punishes high strength steels on residual stresses and their significance in the design of
more than the sshift term in the AW-case. This effect is pressure vessels was carried out [21]. Existing experimental
becoming detrimental for yield strengths of 460 MPa and and numerically predicted residual stress profiles were
higher. The fscatt of 1.4 is not justified from the steel collected, and some new profiles were generated. Attempts
manufactures’ data and hence sshift is proposed alternatively. were also made to standardise residual stress profiles.
In the SINTAP assessment of residual stress profiles, it
2.2.3. Secondary stresses, R6 and BS 7910 was found that in many of the investigated weldments, the
In R6 [18] (and in BS PD6493:1991 [19] as well as in BS stresses reached the yield point at least at some position
7910:1999 [20]) procedures are given for the quantification although the variation between joints was quite large [21].
This can be considered as support for the assumptions
about the chosen stress levels in Eqs. (2)–(6). Those stress
levels will also be used in Eq. (7) when a direct
comparison is made between the CEN and the SINTAP
procedures.
In analysing the influence of the residual stresses, the
total stress intensity factor KI is given as a sum of a
contribution from the primary stresses, KIp , and a
contribution from the secondary stresses, KIs , in the
following way
KI
KR Z (14)
KIc
The same changes are obtained for the minimum design 4. Discussion
temperature in the AW and PWHT conditions, namely a
change of the minimum design temperature by 5, 0, and The total effect of the proposed modifications 3.3–3.5 is
K13 8C for a thickness of 60, 30, and 10 mm respectively, illustrated as ‘summary’ in Fig. 11. In the as-welded (AW)
see Fig. 10. These values can also be found from Fig. 5. The condition the change, is between K9 and K29 8C, and in
values are independent of the yield strength. the PWHT condition between 5 and K15 8C. The main part
of this effect is due to the modified thickness correction.
3.5. Scatter factor for the yield strength At other yield strengths these values are shifted but not more
than 5 8C for strengths up 640 MPa. In conclusion, the
The consequence of changing the scatter factor from 1.4 summary model is less conservative than the CEN-model in
to a scatter term of 100 MPa in Eqs. (2)–(6) is shown in the AW conditions and gives about the same values in the
Fig. 10. This change reduces the minimum design PWHT condition.
temperature by 11–16 8C in the as-welded condition and If the same assumptions are made as in Eurocode, slightly
by 4–7 8C in the post-weld heat treated condition. The higher minimum design temperatures are obtained. In the AW
changes are slightly larger for thinner than for thicker condition the values are higher than the summary values by
gauges. This effect increases slightly with increasing yield 15–25 8C and in the PWHT condition by 8–20 8C. The reason
strength. is that the elasto-plastic effect on the residual stresses is not
taken into account. This should not be considered as a direct
comparison with Eurocode, since the role of the residual
stresses is only included to a limited extent there.
5. Conclusions
The model given in [2], which was the basis for design
against brittle failure in EN 13 445 has been analysed. In
particular, the impact of newer findings in fracture
mechanics and applications in other areas of welded steel
construction such as Eurocode 3 have been studied:
† A modified and interpolated Sanz’ model for the [6] EN 13445-2. Unfired pressure vessels—part 2—materials. Brussels:
temperature dependence of the fracture toughness was CEN; 2002.
[7] Wallin K. Master curve method: a new concept for brittle fracture. Int
used for the CEN model. The values are almost J Mater Prod Technol (Switzerland) 1999;14:342–54.
identical to those of Wallin’s Master Curve approach, [8] Milne I, Ainsworth RA, Dowling AR, Stewart AT. Background to and
which is applied in Eurocode 3 and the SINTAP validation of CEGB report R/H/R6—revision 3. Int J Pres Ves Pip
approach. 1988;32:105–96.
† For the thickness dependence of fracture toughness, a [9] Structural Integrity Assessment Procedure (SINTAP). Final Revision,
EU-project BE 95-1462. Brite-Euram programme; 1999.
modified and interpolated model of Sanz was used for [10] Sedlacek G, Dahl W, Stötzel G, Liessem A. Improvement of the
the CEN work. Nowadays Wallin’s Master Curve is method given in Eurocode 3, Annex C for the choice of material to
often used. This gives a 5 8C higher minimum design avoid brittle fracture, International Institute of Welding document
temperatures at large gauges (60 mm) and 13 8C lower IIW-DOC-X-1274-93; 1993.
values at thinner gauges (10 mm). [11] EUROCODE 3, Design of steel structures (part 1) pr ENV 1993-1,
1998 and Design of steel bridges pr ENV 1993-2; 1999.
† Elasto-plastic corrections to the fracture toughness are [12] Raju IS, Newman JCJ. Stress-intensity factors for internal and
applied in Eurocode 3, the SINTAP approach and R6 external surface cracks in cylindrical vessels. J Pres Ves Technol
but not in the CEN model. If elasto-plastic corrections (Trans ASME) 1982;104:293–8.
also take account of the residual stresses, the change in [13] Metals handbook. 10th ed., vol. 19. Metals Park: ASM; 1996. p. 989.
the minimum design temperatures is reduced and [14] ASME Boiler and Vessel Code. Section XI, rules for in-service
inspection of nuclear power plant components. American Society for
increased of the order 5 8C in the AW and PWHT Mechanical Engineering, Articles A-3000, method of KI-determi-
conditions, respectively. nation; 1998.
† In the SINTAP approach the elasto-plastic corrections for [15] Sandström R, Langenberg P, Sieurin H. New brittle fracture model for
the residual stresses are taken into account by introducing the European pressure vessel standard. Int J Pres Ves Pip 2004;81:
separate stress intensity factors for the primary and the 837–45.
[16] Sandström R, Ivarsson B. Influence of scatter in yield stresses on
secondary stresses. In its original published form this design stresses at elevated temperatures. Mater Des 1986;7:95–100.
exaggerates the influence of the secondary stresses. [17] Barsom JM, Rolfe ST. Correlations between KIc and Charpy V-notch
Starting from other data also taken from the SINTAP test results in the transition-temperature range American Society for
project, a modified procedure was proposed, which gives Testing and Materials STP 466 1970 p. 281–302.
an improved description of the influence of plasticity on [18] Assessment of the integrity of structures containing defects. Barnwood,
Gloucester: British Energy Generation Ltd; 2001 R6-Revision 4.
the secondary stresses. The net effect on the minimum [19] Guidance on methods for assessing the acceptability of flaws in fusion
design temperatures in this case in relation to that in the welded structures. London: British Standards Institution, BS PD6493;
CEN model is however quite limited. 1991.
† The total effect of these modifications is K9 to K29 8C in [20] Guide on methods for assessing the acceptability of flaws in fusion
the AW condition and 5 to K15 8C in the PWHT welded structures. London: British Standards Institution, BS 7910;
1999.
condition. The largest effects appear for thinner gauges. [21] Stacey A, Barthelemy JY, Leggatt RH, Ainsworth RA. Incorporation
of residual stresses into the SINTAP defect assessment procedure.
Eng Fract Mech 2000;67:573–611.
[22] Smith SD. Comparison of the PD6493:1991 rho (p) factor with FEA
Acknowledgements results. TWI report SINTAP/TWI/1-2; 1997.
[23] Wallin K. Methodology for selecting Charpy toughness criteria for
thin high strength steels Jernkontorets forskning, report from working
Financial support from the EU Growth project ECO- group 4013/89 1990.
PRESS is gratefully acknowledged. [24] Liessem A. Bruchmechanische Sicherheitsanalysen von Stahlbauten
aus hoch-festen niedriglegierten Stählen.: Shaker Verlag; 1996 [Band
3/96].
[25] Sedlacek G, Stranghöner N, Stötzel G, Dahl W, Langenberg P,
References Liessem PA. Die Tragsicherkeit, die Ermüdungssicherheit und das
Sprödbruchproblem. Stahlbau 1996;65:407–14.
[1] Sanz G. Attempts to introduce a quantitative method of choosing steel [26] Stranghoner N, Sedlacek G, Stotzel G, Dahl W, Langenberg P. The
quality with reference to the risk of brittle fracture. Rev Metall CIT choice of steel material for steel bridges to avoid brittle fracture.
1980;77:621–42. J Constr Steel Res (UK) 1998;46:56.
[2] Sandström R. Minimum usage temperature for ferritic steels. Scand [27] Langenberg P, Sedlacek G, Dahl W, Feldmann M. Einsatz
J Metall 1987;16:242–52. bruchmechanischer Sicherheitskonzepte für Sonderfälle des Stahlbaus
[3] CODAP: French Code for the construction of unfired pressure vessels. im Bereich geschweißter Konstruktionen, AK Bruch 2002: For-
Part M. Appendix MAZ; 1984. tschritte in der Bruch- und Schädigungsmechanik. DVM Berichtsband
[4] Swedish pressure vessel code. Stockholm: Swedish pressure vessel 234. S. 223–34.
commission; 1987. [28] Falk J. Untersuchungen zum Einfluss der Belastungsgeschwindigkeit
[5] Wiesner CS, Sandström R, Garwood SJ, Street DM, Coulson KJ. auf das Verformungs- und Bruchverhalten an Stählen unterschiedli-
Background to requirements for the prevention of brittle fracture in cher Festigkeit. PhD Thesis, Institut für Eisenhüttenkunde, RWTH
the European standards for pressure vessels (prEN 13445) and Aachen; 1993.
metallic industrial piping (prEN 13480)). Int J Pres Ves Pip 2001;78: [29] Wallin K. The size effect in KIc results. Eng Fract Mech 1985;22:
391–9. 149–63.
Update
International Journal of Pressure Vessels and Piping
Volume 82, Issue 12, December 2005, Page 941
DOI: https://doi.org/10.1016/j.ijpvp.2005.09.001
International Journal of Pressure Vessels and Piping 82 (2005) 941
www.elsevier.com/locate/ijpvp
Erratum
Erratum to “Analysis of the brittle fracture avoidance model
for pressure vessels in European standard” [Int J Pressure Vessels
and Piping 82 (2005) 872–881]*
Rolf Sandström a,*, Peter Langenberg b, Henrik Sieurin a
a
Department of Materials Science and Engineering, Royal Institute of Technology (KTH), Brinellvagen 23, S-100 44 Stockholm, Sweden
b
IWT, Malmedyer Str. 30, D-52066 Aachen, Germany
The publisher regrets that in the above article the first author’s surname was printed incorrectly. It is now reproduced correctly,
above.
*
DOI of original article: 10.1016/j.ijpvp.2005.06.004
* Corresponding author. Tel.:C46 8 7908321; fax: C46 8 203107.
E-mail address: rsand@kth.se (R. Sandström).
0308-0161/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijpvp.2005.09.001