Anda di halaman 1dari 8

Engineering Structures 24 (2002) 11331140 www.elsevier.

com/locate/engstruct

A simplied crack model for weld fracture in steel moment connections


T.D. Righiniotis a,, E. Omer b, A.Y. Elghazouli b
b a School of Engineering, University of Surrey, Guildford, Surrey GU2 7XH, UK Department of Civil and Environmental Engineering, Imperial College, London SW7 2BU, UK

Received 6 December 2001; received in revised form 11 March 2002; accepted 12 March 2002

Abstract This paper presents a simplied two-dimensional crack model for assessing the fracture of bottom ange welds in steel beamto-column connections. The formulation of the model includes the determination of approximate expressions for stress intensity factors related to the cracked geometry, accounting for typical stress conditions and taking due consideration of the presence of the backing bar. Idealized residual stress distributions are also incorporated in the model to examine their inuence on the behaviour. Particular attention is given to typical connection congurations, which have suffered considerable damage in the Northridge earthquake. Comparisons are made between the results obtained from the proposed model and those available from a number of experimental investigations as well as two-dimensional nite element analyses. Within the range of results examined in this study, the proposed model is shown to provide good, and generally conservative, predictions in terms of both the fracture moments and reduction in stiffness. 2002 Elsevier Science Ltd. All rights reserved.
Keywords: Steel connections; Northridge earthquake; Fracture mechanics; Weld failure

1. Introduction The 1994 Northridge earthquake resulted in widespread damage observed in moment resisting steel connections. The design of a typical connection conguration had been standardized through a series of experiments during the 70s [1,2] and 80s [3,4] and consisted of a shear plate, which was bolted on the beam and welded on the column ange while the beam anges were welded on the column ange as shown in Fig. 1. Typically, cracks were discovered in the column ange adjacent to the beam bottom ange and in some cases through the column web [5,6]. The alarming nature of the cracked steel moment resisting frames following the earthquake has attracted over the years considerable research interest. One of these research areas has concentrated on the study of the stress conditions typically encountered in the region of interest (see inset of Fig. 1).

Corresponding author. Tel.: +44-1483-879543; fax: +44-1483450984. E-mail address: t.righiniotis@surrey.ac.uk (T.D. Righiniotis).

These efforts have addressed questions regarding, among others, the weld metal fracture toughness [5], the effect of the backing bar [5,7], poor weld preparation [7,8], constraint conditions [8] and the effect of residual stresses [9]. Inevitably, given the existence of crack-like defects in the weld root, different fracture mechanics techniques were employed to model these cracks and hence explain the brittle fractures of Northridge. These efforts have employed, in some cases, solutions to simple crack models [5,7,8], while in other cases, two [10] and three-dimensional nite element crack models [8,9,11]. More recently, researchers have modelled the weld residual stress pattern in the beam ange [12] and used the local approach for cleavage fracture [13]. In this paper, a simplied two-dimensional crack model is developed, which is more robust than previously used simplied models because it addresses some of the main features of fracture at the bottom weld region. Based on the assumption of linear elastic material response, stress intensity factors are initially derived for the cracked geometry including the backing bar, and comparisons are made with published nite element results [10]. An assumed residual stress distribution through the beams

0141-0296/02/$ - see front matter 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 1 4 1 - 0 2 9 6 ( 0 2 ) 0 0 0 4 3 - 3

1134

T.D. Righiniotis et al. / Engineering Structures 24 (2002) 11331140

Fig. 1.

Typical pre-Northridge moment connection.

bottom ange is incorporated in the model formulation and is found to have a signicant effect on the failure stresses in the beam ange. Finally, comparisons with test results [14] are made in terms of both the fracture moment and the increased beam exibility caused by the presence of the crack. These comparisons demonstrate the ability of the proposed crack model to capture important features of the Northridge connections.

2. Crack model In order to investigate the behaviour of the typical moment resisting connection, several researchers employed fracture mechanics principles and assumed that the unfused material between the backing bar and the column ange as well as the aws existing at the weld root could be modelled as a single crack [5,7,8]. In doing so, the cracked geometry in the vicinity of the weld root was modelled as an edge crack in a semiinnite strip. In this study, the same region is modelled in two dimensions as a strip with a thickness tbf containing a crack of length a and connected to a semi-innite space while being subjected to tension p, bending m and a residual stress eld as indicated in Fig. 2. Additionally, a non-load carrying bar with thickness tbb is attached to

the strip (see Fig. 2). The material is assumed to be linear elastic, thus making the principles of Linear Elastic Fracture Mechanics (LEFM) pertinent. The geometry of Fig. 2 is decomposed in three separate geometries, which are shown in Fig. 3(a)(c). The geometry of Fig. 3(a) captures the condition of the crack being located at the base of the beam ange, which is here represented by the strip. The geometries of Fig. 3(b) and (c) are used here to evaluate the effect of the backing bar. Both p and m loading conditions are used for the model of Fig. 3(a) in order to describe the assumed elastic stress distribution through the beam ange accurately. However, given the stress distribution through the beam, the effect of m is generally small. 2.1. General formulation For Mode I, which corresponds to in-plane tensile crack opening, the stress intensity factors for arbitrary loading may be expressed in the general form K spaY(a / tbf) (1)

where s is the characteristic remotely applied stress and Y(a/tbf) is the Mode I stress magnication factor, which is dependent on the type of loading. The loading and geometry of the structure shown in Fig. 3(a) also gives rise to Mode II stress intensity factors, which have the same general form as Eq. (1) [15,16]. However, here, their contributions were found to be rather small [17], and for this reason, they are omitted from subsequent calculations. In the sequel, reference to any stress intensity factors is taken to imply Mode I. Stress magnication factors related to specic a/tbf values for the geometry of Fig. 3(a) under tension (Kt) and bending (Kb) may be found in tabular and graphical form in Reference [15]. Reference [16] provides in addition to this graphical information formulae for Y(a/tbf). Here, curve-tted expressions in polynomial form for the tension (Yt) and bending (Yb) numerical values of Reference [15] are given by [17] Yt(a / tbf) 2741

Fig. 2. Two-dimensional idealization of the beam to column connection under tension p, bending m and an arbitrary residual stress distribution sres.

a 6 a 6583.3 tbf tbf

T.D. Righiniotis et al. / Engineering Structures 24 (2002) 11331140

1135

27.098

a 2.1681 tbf

The corresponding characteristic stresses appearing in Eq. (1) are, for Kt calculation, s st p / (tbfbbf) and, for Kb, s sb 6m / (t2 bfbbf). The stress concentration effect associated with the backing bar may be obtained by considering the inuence the attachment of thickness tbb in Fig. 3(b) has on the semi-innite space of Fig. 3(c). Accordingly, the stress concentration due to the attachment may be approximated via the ratio Y1/Y2 where Y1 is the stress magnication factor associated with the geometry of Fig. 3(b) and Y2 is the stress magnication factor associated with the geometry of Fig. 3(c). A curve tted expression for the values of Y1, provided in References [15,16], is given by [17]

Y1 1.0807


a tbf tbb a tbf tbf

0.3299

(4)

while Y2 is given as [15,16] Y2 1.12 (5)

Fig. 3. (a) Cracked two-dimensional strip connected to a semi-innite space under tension p, bending m and an assumed uniform residual stress distribution sres. (b) Cracked semi-innite space with a nite thickness attachment under remote tension. (c) Cracked half space under remote tension.

Fig. 4 depicts the variation of the different Y factors used for the proposed model with a/tbf. The difference between the Yt and Yb curves illustrates the effect a gradient eld may have on the Y factors. On the other hand, Fig. 4 shows that the thicker the attachment (higher tbb/tbf ratio), the larger the Y1/Y2 factor. In addition to the external loading, a residual stress eld is present in the vicinity of the crack (see Fig. 2), which gives rise to a stress intensity factor Kres. The stress intensity factor Kres is of the general form of Eq. (1). The precise residual stress eld is not known but it is here conservatively assumed that the residual stresses in the beam ange are uniform throughout its thickness, as indicated in Fig. 3(a), and therefore Yres Yt. For a semi-elliptical weld root defect and under tension or bending, Eq. (1) can be extended to Ks

a 4 a 6140.1 2771.1 tbf tbf a 3.7757 64.493 tbf and

a 627.18 tbf

(2)

Q YY

pa Y1
2

(6)

where Y is Yt or Yb given by Eq. (2) or (3) and Q is the elliptical shape correction factor given by [18] Q(a / c) 1 1.464

Yb(a / tbf) 944.85 2168.2

a 6 a 2290.5 tbf tbf

a 4 a 1003.2 tbf tbf

a c

1.65

(7)

237.75

a tbf

(3)

and, as before, for the purposes of Kt calculation, s st p / (tbfbbf) and, for Kb, s sb 6m / (t2 bfbbf). In Eq. (7) , c is the half the width of the crack and is here taken equal to half the width of the beam web. This is a reasonable assumption, given that the crack depth,

1136

T.D. Righiniotis et al. / Engineering Structures 24 (2002) 11331140

Fig. 4.

Variation of Y factors with normalized crack depth a/tbf.

due to restricted welding access, is more likely to have a greater depth at the position of the cope hole. Similarly, for the stress intensity factor Kres related to the welding residual stresses Kres sres

1 a Y1 2 fbb 2 Yb da tbf Q Y2
0

QY
pa

(8)

fbt

1 a Y1 YtYb da 2 tbf Q Y2
0

(13)

where sres is the characteristic stress of the (uniform) residual stress eld and is assumed to be some fraction of the yield stress. 2.2. Crack stiffness The presence of a crack increases the exibility of the section and thus the former may be modelled as a spring. The additional rotation q, which is introduced along the line of the crack discontinuity, is given by [19] q Cbtp Cbbm (9)

By combining Eqs. (913) and using simple bending theory, the beams bending moment M is found to be, in terms of the strips rotation q, M EI [(Dtbf)fbt tbffbb]1 q 6p(1n2) (14)

Scrack q where I is the moment of inertia of the whole section, D is the depth of the entire section and Scrack is the spring stiffness of the crack. Using the above equations, the critical point at which unstable crack growth occurs can be found. This may be expressed in terms of the critical beam moment i.e. the beam moment Mf that causes fracture. Applying Irwins fracture criterion, the beams bending moment at failure is given by Mf 2Q(KICKres)I

where p and m are, respectively, the axial force and bending moment acting on the strip and Cij (i,j=t,b) are the exibility coefcients. Furthermore, p and m are given by p stbbftbf m sbbbft2 bf 6 (10) (11)

pa

[(Dtbf)Yt tbfYb]1

Y2 Y1

(15)

where bbf is the width of the beam ange and st and sb are the tensile and exural component, respectively, of the elastic stress distribution in the beam ange. The exibility coefcients Cij may be deduced as [19] Cbb 72 (1n2)p fbb bbft2 E 12 (1n2)p fbt bbft E (12)

where KIC is the plane strain fracture toughness of the material.

3. Results and discussion Fig. 5 depicts the variation of Kt (Eq. (6) for pure tension) with the depth of the weld root defect for a W36&times150 beam. The results of Chi et al. [10] based on two-dimensional nite element (FE) analyses

Cbt Ctb

where n is Poissons ratio and, here

T.D. Righiniotis et al. / Engineering Structures 24 (2002) 11331140

1137

Fig. 5. Variation of Kt with crack depth (W36150 beam, tbb=9.5 mm).

of the geometry of Fig. 2 are also presented. Since the FE study by Chi et al. did not incorporate any threedimensional effects, and in order to make a more direct comparison between the proposed model and their results, the factor Q appearing in Eq. (6) has been omitted. The proposed model overestimates Kt factors for this geometry up to crack depths of 1.2 mm and beyond 3.7 mm. Between 1.2 and 3.7 mm the proposed model slightly underestimates Kt. The interesting feature of the present results, which are shown in Fig. 5, is that Kt does not necessarily increase with crack depth. This is because proximity of the crack tip to the backing bar increases the stress concentration effect while reducing the effect of the upper ange boundary. The former is represented by the factor Y1/Y2, while the latter is represented by the factor Yt (see Fig. 4). Beyond approximately a 2 mm crack depth, the effect of the Y1/Y2 factor diminishes and the Kt solution is dominated by the factor Yt. Fig. 6 depicts the variation of the ange stress at failure versus crack depth for a W36150 beam with a 9.5 mm backing bar. Results are plotted for both cases when residual stresses are not considered and when a uniform residual stress pattern of half the yield magnitude (sres 0.5sy, sy 300 MPa) exists throughout the ange thickness. The predictions of the FE model of Chi

et al. [10] are plotted together with the experimental results reported in Reference [14]. The fracture toughness KIC is taken equal to 71.4 MPa m1/2 (65 Ksi in1/2) [10]. Here, the factor Q has been used with a xed c equal to half the beam web width (tbw=15.9mm). The results are here affected not only by the factors inuencing the results of Fig. 5, but also the ratio a/c (see Eqs. (6), (7) and (8)) since for a constant c, increase in a results in a higher Q factor (see Eq. (7)). Fig. 6 shows the profound effect residual stresses have on the ange stress at failure. Numerical analyses carried out by Matos and Dodds [12] suggest that, for a typical ange thickness, the residual stresses start at a very high tensile value at the lower ange surface and become very rapidly compressive (0.1 tbf). However, large variations in the residual stress pattern as well as the peak (characteristic) stress found on the ange surface can explain the differences between the results presented here and the experimental behaviour reported in Reference [14]. The maximum difference between the experimental results [14] and the present model with residual stresses is 36% and 27% when residual stresses are not included in the model. It has to be emphasized that the nite element results of Chi et al. depicted in Fig. 6 did not take into account any residual stresses and therefore a more direct comparison may be made with the present model when residual stresses are not considered. In this case, when compared with the results of Chi et al., the present model predicts a lower failure ange stress up to a 1.3 mm crack depth and a higher ange stress beyond this value. The effect of the backing bar thickness is illustrated in Fig. 7 where results are reported in terms of the normalized moment Mf/Mp, where Mf is given by Eq. (15) and Mp is the beams plastic moment, for a W36150 beam with an assumed fracture toughness of KIC=60 MPa m1/2 and two different depths of weld root defect (a0=0.5 mm and a0=2 mm). It is assumed here that sres 0.5sy (sy 300 MPa). As can be seen in Fig. 7, increase in the backing bar thickness reduces the nor-

Fig. 6. Variation of the failure beam ange stress with crack depth (W36150 beam, KIc=71.4 MPa, tbb= 9.5 mm, sres 0.5sy, sy 300 MPa).

Fig. 7. Effect of backing bar thickness on the normalized fracture moment (W36150 beam, KIC=60 MPa m1/2, sres 0.5sy, sy 300 MPa).

1138

T.D. Righiniotis et al. / Engineering Structures 24 (2002) 11331140

malized fracture moment but this reduction, which is caused by the dependence of Y1/Y2 on tbb/tbf (see Fig. 4), is rather small over the wide range of backing bar thickness shown in Fig. 7 (17% for a0=2 mm and 19.4% for a0=0.5 mm). This nding is in accord with the results of Chi et al. [10] in which it is reported that K values were insensitive to the thickness of the backing bar. Table 1 provides a comparison between the experimentally obtained [14] fracture moments Mexp and the theoretically determined fracture moments Mf (see Eq. (15)). Given the lack of information on crack depths for specimens PN2 and PN3 and the large variance observed for two of the UTA specimens (UTA1 and UTA2), a crack depth of 1 mm is assumed for both of these PN specimens. Crack depths are assumed also for specimens UCSD1 and UTA3. For the calculation of Mf, a uniform residual stress distribution is taken with sres 0.5sy with sy as determined in Reference [14] and these results are shown in the third column of Table 1. Under the assumption of the same fracture toughness for all specimens, very good agreement (16% difference) may be observed for three specimens (UCSD1, UTA1 and UTA3). For specimen UTA2, a difference of approximately 40% is observed, while for the three remaining specimens, the model grossly underestimates the fracture moment (slightly less than 50% difference). Column 4 of Table 1 presents the results for Mf/Mexp assuming zero residual stresses. Again, the fracture toughness is assumed to be constant and equal to 70 MPa m1/2. Under this assumption, excellent agreement (8%) may be observed for three of the specimens (UCSD1, UTA2 and UTA3), good agreement (approximately 20%) for one specimen (PN2), poor agreement (50%) for one specimen (UTA1) and fairly good agreement (approximately 30%) for two specimens (PN1 and PN3). Table 1, clearly demonstrates the signicant effect residual stresses can have on the fracture moment. Furthermore, given the
Table 1 Comparison between experimental and theoretical beam normalized fracture moments (KIC=70 MPa m1/2, tbb=9.5 mm, Mexp=experimental fracture moment, Mf =theoretical fracture moment (Eq. (15)) (1) Specimena (2) a (mm) (3) Mf/Mexp (sres 0.5sy)c 0.84 1.07 0.62 0.84 0.56 0.54 0.54 (4) Mf/Mexp sres 0 1.06 1.50 1.06 1.08 0.68 0.81 0.69

uncertainty associated with crack depth determination, fracture toughness and residual stress pattern, the model yields reasonable results. It should also be noted that in the experimental results reported in Table 1, fracture typically occurs before the occurrence of signicant yielding in the beam. In most cases the experimental fracture moment is lower than the yield moment and well below the plastic moment of the beam cross section. 3.1. Beam stiffness As mentioned in Section 2, the presence of the crack results in an increase of the exibility of the connection. Here comparisons are made between the model predictions based on Eq. (14) and the experimental results obtained from cantilevered beam specimens [14]. Assuming that the beam and crack behave as two springs in series, the total initial stiffness of the cantilevered beam will be given by
el Smod beam

1 Sfull beam

Scrack

Scrack

L 3EI

(16)

where Sfull beam 3EI L (17)

is the rotational beam stiffness, L is the length of the cantilever and Scrack is as given by Eq. (14). Furthermore, assuming that the beam anges carry the entire moment with no participation from the beam web, the partial cantilever beam stiffness without the crack will be given by Spartial beam 3EIf L (18)

where If is the moment of inertia of the beam anges. Fig. 8 depicts the variation of Sn, which is the crack

UCSD 1 UTA 1 UTA 2 UTA 3 UCB PN1 UCB PN2 UCB PN3
a b c

1b 3.7 9.7 1b 0.1 1b 1b

Specimen designation according to Reference [14]. Assumed crack depths. Yield stress as determined in Reference [14].

Fig. 8.

Variation of normalized stiffness Sn

(BeamW36150, tbb=9.5 mm, L=2 m).

Scrack with crack depth Sfull beam

T.D. Righiniotis et al. / Engineering Structures 24 (2002) 11331140

1139

stiffness Scrack normalised by the beam full stiffness Sfull beamfor a W36150 cantilevered beam (L=2 m). For small crack sizes (1 mm), the stiffness associated with the crack is large and as suggested by Eq. (16), the beam stiffness will tend to its full stiffness value. Beyond a crack depth of 1 mm, the decrease in Scrack is more gradual and for very large crack sizes (7 mm and 10 mm) becomes negligible. For cases where large crack sizes may be present in the weld root, as in the case of Northridge [5], the second term in Eq. (16) will dominate, thus resulting in signicant increases in exibility. The results obtained from Eqs. (1618) are shown in Table 2. The beam lengths L used for this study are the ones reported in Reference [14]. All stiffnesses predicted by Eqs. (1618) are normalized by the experimentally obtained stiffness Sexp Column 5 of Table 2 demonstrates that the assumption of full stiffness development is not accurate. Use of Eq. (17) overestimates the stiffness for all specimens and this overestimate varies between 23% and 67%. On the other hand, use of Eq. (18) results in Column 4, which is, with the exception of one specimen (EERC PN1, difference 32%), in very good (EERC PN2, difference 15%) to excellent agreement with the experimental results. However, Eq. (18) ignores the contribution by the beam web to the beams bending stiffness. By contrast, the model developed here implies that an additional exibility is introduced by the crack. These results are given in Column 3 of Table 2, for the crack depths shown in Column 2. The agreement is generally very good to excellent (maximum difference of 24% for specimen UCB PN1) but it has to be emphasized that the model predictions are highly dependent on the crack depth, which is here assumed in all but one case (UCB PN1). For this specimen, the presence of a small cracklike defect (depth 0.1 mm) has a very small inuence on the beam stiffness (compare Columns 3 and 5). However, given the large defect sizes observed in both laboratory specimens (see specimens UTA 1 and UTA 2 in Table 1) and in typical Northridge connections [5], even larger reductions in stiffness than those shown in Table

2 may be encountered. However, in view of Fig. 7, these reductions, beyond a certain crack size and within the crack size variability observed in typical Northridge connections, will be small.

4. Conclusions This paper has described a simplied two-dimensional crack model that captures some of the features of the Northridge moment resisting connections. The formulation was presented in terms of the bending moment at failure as well as the reduction in beam stiffness that the presence of a weld root defect induces. Stress intensity factor results were compared with the nite element results of Chi et al. [10] and good agreement was observed. When residual stresses were taken into account these were found to have a signicant effect on the failure beam ange stress. For most test results, the experimentally obtained [14] failure ange stresses where found to fall within a bracket, whose lower bound was dened by the use of welding residual stresses ( sres 0.5sy) and upper bound was dened by their omission. In the former case, the maximum difference was 36% and in the latter 27%. In general, incorporating residual stresses, results in conservative estimates of the failure moment, which can be as low as half of their experimental counterparts. However, these estimates were obtained in most cases by assuming the crack dimensions, the fracture toughness and the residual stresses on the beam ange surface. Finally, comparisons were made between the experimentally observed beam stiffness of six test specimens and the stiffness obtained using the proposed crack model. Very good agreement was observed, which may explain the increased exibility of the beam specimens. In general, the model accounts for the main underlying phenomena inuencing bottom ange weld fracture in moment resisting connections. It also provides a simplied alternative to complex nite element analysis.

Table 2 Comparison between normalized stiffnesses (tbb=9.5 mm, length of cantilever as given in Reference [14]) (1) Specimena (2) a (mm) (3) model Sbeam Sexp 1.15 1 0.95 1.24 0.92 0.96 (4) Spartial beam Sexp 1.32 1.15 1.09 1.03 0.99 1.04 (5) Sfull beam Sexp 1.67 1.45 1.38 1.27 1.23 1.28

EERC PN1 EERC PN2 EERC PN3 UCB PN1 UCB PN2 UCB PN3
a b

1b 1b 1b 0.1 1b 1b

Specimen designation according to Reference [14]. Assumed crack depths.

1140

T.D. Righiniotis et al. / Engineering Structures 24 (2002) 11331140

Within the range examined in this study, the model was shown to produce results on the conservative side, which is generally desirable, given the uncertainty in determining the parameters involved.

References
[1] Popov EP, Bertero VV. Cyclic loading of steel beams and connections. ASCE Journal of the Structural Division 1973;99(ST6):1189204. [2] Popov EP, Stephen, RM Cyclic loading of full size steel connections. Tech. rept. UCB/EERC/70/3. University of California, Berkeley, 1970. [3] Krawinkler H, Popov EP. Seismic behaviour of moment connections and joints. ASCE Journal of the Structural Division 1982;101(ST11):231736. [4] Popov EP, Amin NR, Louie JJC, Stephen RM. Cyclic behaviour of large beam-column assemblies. Earthquake Spectra 1985;1(2):20338. [5] Kaufmann EJ, Fisher JW, Di Julio Jr. RM, Gross JL. Failure analysis of welded steel moment resisting frames damaged in the Northridge earthquake. Tech. rept. 5944. NIST, 1997. [6] Miller DK. Northridge: the role of welding claried. The welding innovation quarterly 1994; Volume XI, No.2. [7] Fisher JW, Dexter RJ, Kaufmann EJ 1995. Fracture mechanics of welded structural steel connections. Background reports: Metallurgy, fracture mechanics, welding, moment connections and frame systems behavior. SAC/95-09. FEMA/SAC joint venture, 1995.

[8] Yang TS, Popov EP. Behaviour of pre-Northridge moment resisting connections. Tech. rept. UCB/EERC/95/08. University of California, Berkeley, 1995. [9] Righiniotis TD, Hobbs RE. Fracture strength of a moment resisting welded connection under combined loading; Part II, Results. Journal of Constructional Steel Research 2000;56(1):3145. [10] Chi WM, Deierlein GG, Ingraffea A. Finite-element fracture analyses of welded beam-column connections. Fatigue and Fracture Mechanics:30th Volume 2000. ASTM STP 1360: 439-455. [11] Chi WM, Deierlein GG, Ingraffea A. Fracture toughness demands in welded beam-column moment connections. ASCE Journal of Structural Engineering 2000;126(1):8897. [12] Matos CG, Dodds Jr. RH. Modeling the effects of residual stresses on defects in welds of steel frame connections. Engineering Structures 2000;22(9):110320. [13] Matos CG, Dodds Jr. RH. Probabilistic modelling of weld fracture in steel frame connections part I: quasi-static loading. Engineering Structures 2001;23(8):101130. [14] SAC. Experimental investigations of beam-column subassemblages. SAC/ 96-01. SAC joint venture, 1996. [15] Murakami Y. Stress intensity factors handbook. Oxford: Pergamon Press, 1987. [16] Tada H, Paris PC, Irwin GR. The stress analysis of cracks handbook. New York: ASME, 2000. [17] Omer E. Fracture of welded steel moment resisting connections. MSc dissertation, Imperial College, London, 2001. [18] Merkle JG. A review of some of the existing stress intensity factor solutions for part-through surface cracks. ORNL-TM 3983. US. Atomic Energy Commission, 1973. [19] Rice JR, Levy N. The part-through surface crack in an elastic plate. ASME Journal of Applied Mechanics 1972;39(3):18594.

Anda mungkin juga menyukai