Ductile-fatigue transition fracture mode of welded T-joints under quasi-static cyclic large plastic strain loading

Accepted Manuscript Ductile-fatigue transition fracture mode of welded T-joints under quasi-static cyclic large plastic strain loading Yan Liu, Liang-Jiu Jia, Hanbin Ge, Tomoya Kato, Toyoki Ikai PII: DOI: Reference:

S0013-7944(16)30733-0 http://dx.doi.org/10.1016/j.engfracmech.2017.02.018 EFM 5414

To appear in:

Engineering Fracture Mechanics

Received Date: Revised Date: Accepted Date:

9 December 2016 21 February 2017 21 February 2017

Please cite this article as: Liu, Y., Jia, L-J., Ge, H., Kato, T., Ikai, T., Ductile-fatigue transition fracture mode of welded T-joints under quasi-static cyclic large plastic strain loading, Engineering Fracture Mechanics (2017), doi: http://dx.doi.org/10.1016/j.engfracmech.2017.02.018

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Ductile-fatigue transition fracture mode of welded T-joints under quasi-static cyclic large plastic strain loading Yan Liu2, Liang-Jiu Jia1,*, Hanbin Ge2, Tomoya Kato2, Toyoki Ikai2

Abstract:

Welded structures experiencing a small number of large plastic strain

reversals during a strong earthquake, can lead to ductile or fatigue fracture. Experimental study on a series of welded T-joints under cyclic large plastic strain loading is carried out to clarify their failure mechanisms in the ultra low cycle fatigue range commonly with a crack initiation life of several dozens of cycles. Effects of post-weld treatment, notch size, notch location and loading history on the failure process of the joints are studied. The experimental results indicate that the fracture is a transition mode of ductile and fatigue fracture. It is also found that the post-weld treatment can greatly improve the crack initiation lives of the joints, and the Charpy impact energy has a great effect on the cracking propagation rate. Formulae to evaluate the crack initiation lives of the tested specimens respectively in terms of ductility ratio and equivalent plastic strain are also proposed, and the effect of small strain amplitudes on the damage accumulation is explained based on the test results under various loading protocols.

Keywords:

ductile fracture; ultra low cycle fatigue; transition; cyclic loading;

structural steel

1. Introduction Fracture of structural steel can be commonly classified into several categories, i.e., ductile fracture, brittle fracture and fatigue fracture[1]. Ductile fracture is commonly induced by void nucleation, void growth and void coalescence under large plastic 1

* Research Institute of Structural Engineering and Disaster Reduction, Tongji Univ., Shanghai, 200092, China. Email: [email protected] (corresponding author) 2 Dept. of Civil Eng., Meijo Univ., 1-501 Shiogamaguchi, Tempaku-ku, Nagoya, 468-8502, Japan.. 1

strain loading as illustrated in Fig. 1[2], which is commonly affected by the stress triaxiality and plastic strain. The fracture surface is the dimple pattern under scanning electron microscope observation. Brittle fracture can occur by cleavage for crystalline materials as a result of tensile stress acting normal to crystallographic planes, which is commonly controlled by stress. Fatigue fracture can be further divided into two categories, i.e., high cycle fatigue and low cycle fatigue, where the fatigue fracture surfaces are characterized by striations[3]. The one below 10000 cycles is termed low cycle fatigue, and is commonly associated with localized plastic behavior in metals. Typical fracture surfaces of the aforementioned three fracture mechanisms are illustrated in Fig. 2[4]. During a strong earthquake, metal structures are apt to fail in ductile fracture as observed during the 1994 Northridge earthquake[5] and the 1995 Kobe earthquake[6] due to the large plastic strain amplitudes. However, the fracture mode can transit from ductile fracture to fatigue fracture as the strain amplitude decreases[7]. The fracture mode is also termed as ultra low cycle fatigue or extremely low cycle fatigue in the literature[8-16]. Mechanisms of ductile fracture, e.g., [17-31], high- and low-cycle fatigue fracture have been intensively investigated, while corresponding research on the transition fracture mode between the two fracture modes is still limited[32-36]. To date, the corresponding fracture mechanism has not been comprehensively understood through experimental studies, especially for welded metal structures[37-41]. This paper aims to study the fracture mechanism of welded steel T-joints under cyclic large plastic strain loading, where the crack initiation lives are within 100 cycles. The effects of post-weld treatment, degree of local strain concentration, location of strain concentration and loading protocol on the fracture behavior were studied through both experimental and numerical studies. A ductile-fatigue transition fracture mode was identified based on observation of the specimens. Meanwhile, the post-weld treatment was found to be able to greatly improve the crack initiation lives of the joints. Formulae to evaluate the crack initiation lives of the welded T-joints

2

were also proposed, which can well predict the crack initiation life of the transition fracture mode.

2. Experimental Program 2.1 Specimens Experiments on 11 steel welded T-joints as shown in Fig. 3 were conducted under cyclic large plastic strain loading. All the specimens were manufactured from the same 16 mm thick steel plate along the rolling direction. The steel is made of SM490YA, which is a widely used steel type in civil engineering in Japan. The mechanical properties and chemical composition of the steel are given in Table 1. Three types of specimens were designed, i.e., as-welded, post-weld treated by filleting the top weld beads and post-weld treated notched specimens as shown in the Fig. 3. The as-welded specimens are shown in Fig. 3(b), where the T-joints were manufactured through welding of 3 plates with single bevel groove welds, and the bevel angles were 35 degrees. The configuration of the joint is to simulate a beam-flange-to-column-flange joint in beam-to-column connections in welded steel moment resisting frames. All the specimens were cut from the same welded assembly, and a 3-pass-3-layer welding was employed using the gas shielded arc welding method by an automatic welding process, and the inter-pass temperature was controlled to be below 200 °C and a heat input of about 40 kJ/cm was adopted according to engineering experience, which is below the upper bound, 100 kJ/cm, of the Japanese specification for steel bridges[42]. The material of the welding electrodes is SM-1S, and the wire diameter is 1.2 mm. The width of each T-joint is 100 mm to ensure the stress state at the joint region close to the mid-width be in a plane strain state. Post-weld treated specimens as illustrated in Fig. 3(b) were designed to simulate the effect of the post-weld treatment on crack initiation and crack propagation properties of the specimen, where only the top weld beads were removed through a mechanical process. Typical heights of the top and bottom weld beads are respective 3

about 4 mm and 1 mm. Three notch sizes illustrated in Fig. 3(c) were designed to simulate the effect of strain concentration on cracking behaviors of the specimens, where the notch depths were designed the same as 2.0 mm, and the radii at the notch roots are respective 0.25 mm, 1.0 mm and 4.0 mm. The notch roots are all located on the fusion line as shown in the figure. The specimens with different notch locations are shown in Fig. 3(d), where the notch root is respectively located within the weld deposit, fusion line, heat-affected zone (HAZ) and base metal, and their radii at the notch roots were all designed as 1.0 mm to investigate the effect of notch location on cracking behaviors of the specimens. It should be noted that the small bottom weld beads of the T-joints were not removed as illustrated in Fig. 4.

2.2 Manufacturing process of specimens The configurations of the fusion lines are not straight lines, and it took great effort to make the notches at the expected locations. Macroscopic examination tests on the two lateral surfaces of the welded T-joints were carried out to visualize boundaries of different subzones of the welds. The surfaces were ground using sandpapers with various grit sizes, from coarse to fine ones. The surfaces were then mirror finished using buffing cream. Finally, etching of the surfaces was carried out using Nital-4 (4% nitric acid in ethanol), and macrostructures as illustrated in Fig. 4 can be obtained. From the figure, one can observe the different subzones of the welds, and the sub-HAZ of a single weld pass can also be seen. To manufacture the notches at their expected positions, all the specimens were first cut in a factory, and macroscopic examination was conducted to visualize locations of the fusion lines in the laboratory of the university. Finally, the exact locations of the notches were verified and marked, and the notches were finally manufactured at the marked locations of the specimens in the factory. Through this process, all the notches can be located at their expected subzones. Meanwhile, macroscopic examination can also facilitate observation of crack initiation and crack propagation behaviors during the testing.

4

2.3 Test setup The cyclic tests were conducted using an MTS testing system as illustrated in Fig. 5, where the strength and displacement capacities are respective 500 kN and ±75mm. Each end of the specimen was clamped within two rolling round bars to realize a pin connection. It has to be noted that the bolts to clamp the rolling round bars should be tight slightly to avoid large frictional force between the rollers and the specimen. The top end of the joint was clamped by the loading head of the MTS, and all the tests were controlled using the net vertical displacement at the top end of the specimen. The specimens were supported by two supporting plates anchored at the machine bed of the MTS system through high strength bolts. Several LVDTs were employed to obtain the net vertical displacement of the specimen, and also to monitor the deformation of the supporting devices. The boundary conditions and the measurement scheme are illustrated in Fig. 6.

2.4 Loading protocol Four loading protocols shown in Fig. 7 were employed in the tests, where the former 3 ones are similar to each other, with an incremental loading followed by different constant-amplitude loadings. For the incremental loading stage, the adjacent increment is equal to the yield displacement, δy0. Since the displacement amplitude commonly has a great effect on the cumulative dissipated energy and cracking initiation and crack propagation properties for steel structures, which is thus varied in the 3 loading protocols. The values of the constant-amplitudes are respectively equal to 8δy0, 10δy0 and 12δy0. The last one is a random loading protocol, which is to simulate the displacement history during a realistic strong earthquake. The measured dimensions of the 11 specimens were listed in Table 2, where the numbering of the specimen “ROO-NWB(WB)-Ci-FL(BM, WD, HAZ)” denotes: ROO = curvature radius at the notch root, NWB = specimen without top weld beads (post-weld treated), WB = as-welded specimens with both top and bottom weld beads,

5

Ci = numbering of the loading protocol, FL = notch root location at the fusion line, BM = base metal, WD = weld deposit and HAZ = heat-affected zone.

3. Experimental results 3.1 Failure modes The main failure modes and cracking processes of the specimens are illustrated in Figs. 8 to 17. To determine the instant of crack initiation, a minimum size of a crack has to be defined. In this study, a crack of 1 mm can be observed by human eyes as well as a digital camera, which was employed in this study[15]. 3.1.1 Effect of post-weld treatment For the as-welded Specimen Rinf-WB-C1, crack initiated at the top weld toe as shown in Fig. 8. The crack initially propagated along the fusion line, and then propagated into the HAZ and base metal along the minimum cross-section of the specimen after the crack depth along the thickness direction reaches about 1.5 mm. The final rupture route is also generally along the minimum cross-section of the specimen. For the post-weld treatment, the crack initiation location changed from the fusion line to the weld deposit on the top surface of the weld as shown in Fig. 9 for Specimen Rinf-NWB-C1, where the red lines indicate the fusion lines. A crack also occurred at the weld toe of the bottom weld bead. The bottom crack initially propagated along the fusion line, and rupture of the specimen eventually occurred when the crack depth along the plate thickness direction increases to a certain size. The final rupture route is generally along the minimum cross-section of the specimen. The numbers of loading half cycles at critical instants, i.e., peak load, crack initiation and decrease of 50% peak load are also listed in Table 3. The number of loading half cycles till crack initiation also increases from 25 to 40 for the application of the post-weld treatment, indicating that the post-weld treatment can greatly delay the crack initiation. Numerical simulation result shown in Fig. 18(a) gives the comparison of the local equivalent plastic strain between the post-weld treated and as-welded specimens. It

6

can be found that filleting at the top weld bead shown in Fig. 3(b) can greatly reduce strain concentration at the top surface of the joints, which leads to high crack initiation life of the post-weld treated specimen. Before discussion on the crack propagation rate, an index termed as cumulative ductility ratio, μc, is defined to describe the cumulative ductility of the specimens under cyclic loading. It is defined as the cumulative plastic displacement non-dimensionalized by the initial yield displacement as follows: c 

  pi i

 y0

(1)

where δpi = plastic displacement of the i-th loading half cycle, δy0 = initial yield displacement of the first loading half cycle. The method to obtain the above variables is also illustrated in Fig. 19. During the experiments, the testing was stopped at the peak displacement of each half cycle for observation and measurement of cracks. The maximum crack depth along the plate thickness direction at the lateral surface was measured. The crack depth versus μc curves are given in Fig. 20, where the crack depth is non-dimensionalized by the thickness of the minimum cross section, which are respectively equal to 16 mm and 14 mm for the notchless and notched specimens. The crack propagation rate can be described by the difference between the number of half cycle till the instant of 50% peak load reduction and that till crack initiation. Large difference corresponds to a low crack propagation rate. Fig. 20(a) shows that the difference of the two instants is 88 for the as-welded specimen, while 53 for the post-weld treated one. This implies that the post-weld treatment has accelerate the crack propagation rate. It has been known that the fatigue crack propagation rate increases as the fracture toughness such as the Charpy impact energy decreases[43]. Charpy impact tests at temperatures ranging from -40 to 40 °C were conducted for the materials at each subzone. Macroscopic examination tests using the etching method were conducted before manufacturing of the notches for all the Charpy test coupons at the welds to ensure the notch roots exactly match their expected locations. The impact test results shown in Fig. 21 indicate that the Charpy impact energy at the weld 7

deposit is the lowest, where the Charpy impact energy at the weld deposit is almost half of the other regions. The Charpy impact energy at the HAZ is lower than that of the Base metal for low temperatures. In fact, original welding material commonly has high ductility and fracture resistance compared with base metal. After the welding process, there are sub-HAZs within two weld passes as illustrated in Fig. 4, and fracture toughness close to the sub-HAZs is commonly poor. The other regions at the weld deposit away from the sub-HAZs are commonly with good fracture toughness. After post-weld treatment, crack initiated and propagated at the weld deposit of the top surface instead of the fusion line. The low Charpy impact energy is the critical factor leading to the accelerated crack propagation rate for the post-weld treated specimen. 3.1.2 Effect of notch size For specimens with different notch sizes close to the fusion lines, i.e., R0.25-NWB-C1-FL, R1.0-NWB-C1-FL and R4.0-NWB-C1-FL, the crack initiation locations are different. From Figs. 10, 13 and 15, it can be found that cracks initiated from the

notch root

on the

fusion

line

for the

V-notched

specimen

R0.25-NWB-C1-FL, while initiated from both the weld deposit and the HAZ for the 2 other U-notched specimens with larger notch radii. The numbers of loading half cycles till critical instants, i.e., peak load, crack initiation and reduction of 50% peak load, are respectively close for the specimens with different notch sizes. The numbers of loading half cycles till crack initiation are the same as 15, which is much smaller than those of the notchless Specimen Rinf-NWB-C1. This indicates that the notches can lead to an early crack initiation of a specimen. However, the notch sizes investigated in this study are found to have unnoticeable effect on the crack initiation lives, though numerical results shown in Fig. 18(b) indicates that notch root strain increases significantly as the notch radius decreases. In this study, the notch roots of the 3 specimens with different notch sizes are all located on the fusion lines. As is known, the fusion line is very narrow, and crack was found to initiate close to the fusion line shown in Fig. 10(a) for the specimen with notch radii of 0.25 mm.

8

Numerical simulation results shown in Fig. 22 indicate that strain concentrates at a small region for the V-notched specimen, and at relative large areas for the 2 U-notched specimens. This makes cracks possible to initiate at a location (material) with the lowest ductility. The nonhomogeneous material close to the fusion line leads to the minor effect of the notch size. After crack initiation at the fusion line, crack propagated into the HAZ and base metal subsequently for Specimen R0.25-NWB-C1-FL as shown in Fig. 10. For the two other Specimens R1.0-NWB-C1-FL and R4-NWB-C1-FL, cracks initiated from the weld deposit subzone and propagated into the fusion line, HAZ and base metal subzones subsequently. Fig. 20(b) shows that the crack propagation rates are close to each

other,

where

the

ones

of

the

Specimens

R1.0-NWB-C1-FL

and

R4.0-NWB-C1-FL are a bit higher. The reason is also for the different Charpy impact energy at different subzones. Crack propagation at the weld deposit is the fastest, which explains the high crack propagation rates for Specimens R1.0-NWB-C1-FL and R4.0-NWB-C1-FL. 3.1.3 Effect of notch location For the U-notched specimens with different notch locations, the crack initiation and propagation paths are also different from each other. When the notch root is located at the base metal, crack initiated at the base metal close to the notch root, and crack propagated along the minimum cross section, which can be found from the cracking mode of Specimen R1-NWB-C1-BM shown in Fig. 11. For the one with the notch root at the weld deposit, crack initiated at the weld deposit close to the notch root and diagonally ran through the fusion line, HAZ and into the base metal as shown in Fig. 12. Subsequently, the crack propagated along the thickness direction in the base metal till failure. For the one with notch root at the HAZ, crack initiated at the HAZ and propagated into the base metal as shown in Fig. 14. The numbers of loading half cycles till crack initiation for the specimens with notches at weld deposit, fusion line, HAZ and base metal, are respective 15, 15, 15 and 17, and numbers of loading half cycles till reduction of 50% peak load are

9

respective 22, 24, 24 and 27. As illustrated in Fig. 3(d), the specimens with moment forces at the cross-sections through the notch roots according to an increasing sequence are: the one with the notch root at weld deposit, fusion line, HAZ and base metal. The maximum difference of moments at the notch root cross sections between the specimens with the notch root respectively at the weld metal and the base metal is about 13%. Numerical analysis results also indicate that the maximum difference of the notch root equivalent plastic strain shown in Fig. 18(c) between the specimens with notches located at the base metal and weld deposit is about 25%. The specimen with the notch root at weld deposit has the largest moment and local strain, while sustained the same loading half cycles till crack initiation as the other 2 specimens with notch roots at the fusion line and HAZ, indicating that the material at the weld deposit has larger resistance to crack initiation compared with those at the fusion line and HAZ. Crack initiation life is mainly dependent on ductility such as elongation of a material. It has been found that the original welding material commonly has excellent ductility over the base metal. After welding, the material at the weld deposit still can have good ductility except for the material close to the sub-HAZs as illustrated in Fig. 4 if the welding heat input is low. This explains why the material at the weld deposit has larger resistance to crack initiation than those at the fusion line and HAZ. This conclusion may not hold for the case of uniaxial tension, where a crack is apt to initiate at the sub-HAZ of the deposit due to the relative uniform stress and strain distribution. Different trends are found for the crack propagation rate. The number of loading half cycles till reduction of 50% peak load for Specimen R1.0-NWB-C1-WD is smaller than the other 2, indicating that the capacity to resist crack propagation of the material at the weld deposit is the poorest. This is also due to the low Charpy impact energy of the weld deposit shown in Fig. 21. 3.1.4 Effect of loading protocol For all the as-welded specimens under different loading protocols, cracks initiated at the top weld toes as shown in Figs. 9, 16 and 17. Cracks initially propagated along the

10

fusion lines, and finally propagated into the HAZ and base metal along the minimum cross-section of each specimen. Loading protocol was found to have minor effect on the failure mode. The number of loading half cycles till crack initiation decreases from 92 to 23 as the displacement amplitude increases, indicating that the crack initiation life is closely correlated with the displacement amplitude (strain amplitude). The same tendency can be found for the crack propagation rate. Crack depth versus cumulative ductility ratio curves shown in Fig. 20(d) indicates that cumulative ductility decreases significantly as the displacement amplitude increases. It can be found from Fig. 18 that the local equivalent plastic strain increases approximately linearly as the displacement increases except for the V-notched specimen. The specimen under random loading has a much larger cumulative ductility owing to its relative small displacement amplitudes. These can be both explained by the Mason-Coffin rule, where damage accumulates nonlinearly as the strain amplitude increases.

3.2 Fractographic observatoin Macro fracture surfaces of the specimens are given in Fig. 23, and all the surfaces can be divided into 2 types of typical subzones, i.e., top, bottom fatigue striations and central ductile fracture surfaces. Fractographic examination using a scanning electron microscope (SEM) was carried out for the fracture surfaces of the specimens. The observation results are illustrated in Figs. 24-26. The SEM observation results indicate that the fatigue striations are different from conventional ones in a high-cycle fatigue case. Dimples can be observed on the striations, which cannot be found in a high-cycle fatigue fracture surface as illustrated in Fig. 2(a). This is mainly due to the extremely large plastic strain amplitudes involved in this study, where the strain amplitude can reach as much as 150% illustrated in Fig. 18(b). The striations mixed with dimples indicates a ductile-fatigue transition fracture mode. Based on these observations, it can be implied that the fracture mode of structural steel can transfer from fatigue fracture to ductile fracture as the strain amplitude increases, and there is

11

also a ductile-fatigue transition fracture mode if the displacement amplitude is medium compared with those of fatigue and ductile fracture. Comparison among Figs. 23(a), (i) and (j) also shows that the portion of ductile fracture surface increases as the displacement amplitude increases. Figs. 24-26 also indicate that sizes of dimples on the striations are larger than those between striations. It can be deduced that the striations are generated at instants when peak displacement of each loading cycle is achieved. Comparison of Figs. 24 and 25 also indicates different micro-structure at the base metal and that at the weld. There are impurities or second-order particles at the weld, leading to poor ductility and fracture resistance of the material there.

3.3 Hysteretic curves The typical vertical load-vertical displacement curves of the specimens are illustrated in Fig. 27, and the absorbed plastic energies till different instants are given in Table 4. The typical mechanical properties of the specimens are given in Table 5. The tangent stiffness of the curve decreases drastically after yielding at small deformation stage, and increases remarkably as the deformation increases. The re-development of the stiffness and strength is mainly owing to the constraint effect of the rollers. During the experiments, the two ends of the specimens were clamped using two rollers and two bolts as illustrated in Figs. 5 and 6. The frictional and contact forces become remarkable as the deformation of the specimen increases. The effects of post-weld treatment, notch size, notch location and loading protocol on the vertical load-vertical displacement skeleton curves of the specimens are respectively shown in Figs. 28 to 31. In the skeleton curves, the instants of peak loads, crack initiation and decrease of 50% peak load are also marked. Since these instants can be at tension or compression half cycles, both the skeleton curves for the tension and compression half cycles were plotted for some of the specimens. 3.3.1 Effect of post-weld treatment Fig. 28 indicates that crack initiates before the peak load for the post-weld treated specimen, while contrary for the as-welded one. The peak load of the post-weld 12

treated one is 15% lower than that of the as-welded one. This is mainly due to the reduction of the cross-sectional area at the fusion regions for the post-weld treated specimens, where the top weld beads are removed as illustrated in Fig. 3(b). Strength deterioration of the post-weld treated one is much slower than the as-welded one. The post-weld treated one can sustain more loading cycles till the peak load and crack initiation than the one without the post-weld treatment. The absorbed energy till crack initiation reaches twice as much as that of the one without the post-weld treatment. These are all owing to the reduced strain concentration effect after the post-weld treatment. 3.3.2 Effect of notch size The load-displacement skeleton curves of the specimens with various notch sizes are shown in Fig. 29, where the notch roots were all designed to lie on the fusion lines. The overall configurations of the curves are similar and the absorbed energies till crack initiation are close to each other. Strength deterioration is closely related with the crack propagation since the load-carrying capacity is mainly determined by the residual area of the cracked cross section. Strength deterioration of the notched specimens develops rapidly compared with the notchless one as shown in Figs. 28 and 29 due to the high crack propagation rates at the notched specimen, and rupture of the whole cross sections at the notch roots occurs soon after the peak load is reached. 3.3.3 Effect of notch location The load-displacement skeleton curves of the specimens with various notch locations are shown in Fig. 30, where the curvature radii at the notch roots are all equal to 1.0 mm. Strength deterioration develops abruptly for the one with notch root at the weld deposit, while relatively slowly for the one with notch root at the base metal. This is mainly due to the low Charpy impact energy, leading to high crack propagation rate at the weld deposit. 3.3.4 Effect of loading protocol The load-displacement skeleton curves of the as-welded specimens under different loading protocols are shown in Fig. 31, where post-weld treatment was not conducted 13

for the specimens, and they are all notchless. Figs. 31(a) to (c) indicate that cracks all initiate at instants when the load decreases to 90% of the peak load. Strength deterioration becomes more rapidly as the displacement amplitude increases. The specimen under small displacement amplitudes can sustain much more loading half cycles than the one under large displacement amplitudes.

4. Ultra-low cycle fatigue life The Mason-Coffin rule is widely employed to evaluate low-cycle fatigue problems, where plastic straining occurs during the fatigue failure process. From the view point of structural engineers, the ductility ratio, μp, is more convenient for seismic design in practice, and global variables such as story drift angles have also been employed by other researchers[10, 44]. For cyclic loading protocols with various displacement amplitudes, the ductility ratio in each loading half cycle can be defined as: i   pi 

Pi Ke

(2)

 y0

According to the Mason-Coffin rule, the crack initiation life (loading half cycles Nfi) can be expressed in terms of the ductility ratio as follows: kd N fi  Cd （   pi）

(3)

where Cd and kd are material constants. However, Eq. (3) is applicable to cases under constant loading amplitudes. To apply to the ones under random loading amplitudes, a damage index D can be defined according to the linear damage accumulation rule, i.e., the Miner’s rule: Di 

1 N fi

(4)

14

where ΔDi is the incremental damage during the i-th loading half cycle. Failure is postulated to occur when D reaches 1. Eqs. (2) to (4) were employed to evaluate the crack initiation life of the as-welded specimens without notches. Based on fitting of the experimental results on as-welded specimens under different loading protocols, the material constants Cd and kd are respectively determined as 23600 and -2.6. The predicted crack initiation life to the actual one ratios are presented in Fig. 32, where the one under loading protocol B is underestimated by about 30%, and the other 3 can be evaluated with good accuracy within 5%. The main reason for this result is for the fact that cracks initiated at the left side of Specimen Rinf-WB-C2, but at the right side for the other 3 specimens. Generally, the Mason-Coffin rule can be employed to evaluate crack initiation life of the ductile-fatigue transition fracture mode observed in this study. Likewise, the crack initiation life can also be expressed in terms of plastic strain amplitude according to the Mason-Coffin rule as follows: p ks N fi  Cs  ( eq ,i )

where Cs and ks are material constants.

 eqp , i

(5)

is the local equivalent plastic strain

amplitude of the i-th cycle at the weld toe of the as-welded specimen, which can be obtained from numerical simulation. Likewise, the material constants Cs and ks are respectively determined as 2.46 and -0.6. The predicted crack initiation life to the actual one ratios for the 4 specimens are presented in Fig. 32, indicating that the strain based equation gives relative poor accuracy compared with the ductility ratio based one.

5. Conclusions

15

A series of experimental studies on steel welded T-joints with under various loading protocols were studied. Effects of post-weld treatment, notch size, notch location and loading protocol on crack initiation lives and crack propagation rates of the specimens were investigated through both experimental and numerical studies. Macroscopic examination tests were conducted for all the specimens to ensure the exact locations of the notch roots during the manufacturing process, and to facilitate confirming the cracking paths during the testing. Based on the above studies, the following conclusions can be drawn: (1) A ductile-fatigue transition fracture mode was confirmed for all the specimens investigated in this study, where both ductile dimples and fatigue striations can be observed at the top and bottom of the cracked surfaces. The central parts of the cracked surfaces were all ductile fracture surfaces. The portion of ductile fracture surface area increased as the displacement amplitude increases. (2) The crack initiation life was found to be dominated by ductility of a material, and crack propagation rate by the Charpy impact energy. This leads to opposite effects of the post-weld treatment on the crack initiation life of the welded T-joint and the crack propagation rate in this study. (3) Cracks didn’t initiate at the notch root for the U-notched specimens, but at the neighborhood of the notch root. However, cracks initiated at the notch root for the V-notched specimen. The notch size was found to have minor effect on the crack initiation lives of the specimens with notch roots at the fusion lines in this study. This is mainly due to the fact that the notch location, i.e., different material properties of each subzone, has more remarkable effects on the crack initiation and crack propagation of the welded specimens. (4) Loading protocol was found to have a significant effect on seismic performance of the T-joints, where the cracking initiation life, crack propagation rate, cumulative ductility ratio and energy dissipation capacity all decreased rapidly as the displacement amplitude increased. (5) Crack initiation life evaluation formulae respectively based on the ductility ratio and equivalent plastic strain were proposed for the as-welded T-joints fractured in 16

the ductile-fatigue transition fracture mode. The ductility ratio based formula gives better evaluation results than the equivalent plastic strain based one. The other effects such as weld heat input, structural steel type and welding material on seismic performance should be further studied. Numerical simulations using micro-mechanics based ductile fracture models[14, 16] are also in progress to examine the local stress and strain history of the specimens, aiming to evaluating effect of material properties such as the Charpy impact energy on seismic behaviors of welded structures under various loading protocols till rupture.

Acknowledgements The study is supported in part by grants from the Advanced Research Center for Natural Disaster Risk Reduction, Meijo University, which supported by Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan. This study is also partially supported by National Nature Science Foundation of China (51508401).

References [1] Anderson TL. Fracture mechanics : fundamentals and applications: Taylor & Francis; 2005. [2] Jia L-J, Kuwamura H. Ductile fracture simulation of structural steels under monotonic tension. J Struct Eng (ASCE). 2014;140:04013115. [3] Qian X, Nguyen CT, Petchdemaneengam Y, Ou Z, Swaddiwudhipong S, Marshall P. Fatigue performance of tubular X-joints with PJP+ welds: II — Numerical investigation. J Constr Steel Res. 2013;89:252-61. [4] Kobe Material Testing Laboratory Group. Available at: http://www.kmtl.co.jp/en/research. Accessed October 08, 2016. [5] Mahin SA. Lessons from damage to steel buildings during the Northridge earthquake. Eng Struct. 1998;20:261-70. [6] Kuwamura H. Fracture of steel during an earthquake—state-of-the-art in Japan. Eng Struct. 1998;20:310-22. [6] Kuwamura H. Transition between fatigue and ductile fracture in steel. J Struct Eng (ASCE). 1997;123:864-70. 17

[8] Zhang Q, Zhang J, Zhao P, Huang Y, Yu Z, Fang X. Low-cycle fatigue behaviors of a new type of 10% Cr martensitic steel and welded joint with Ni-based weld metal. Int J Fatigue. 2016;88:78-87. [9] Kamaya M. Fatigue properties of 316 stainless steel and its failure due to internal cracks in low-cycle and extremely low-cycle fatigue regimes. Int J Fatigue. 2010;32:1081-9. [10] Zhou H, Wang Y, Shi Y, Xiong J, Yang L. Extremely low cycle fatigue prediction of steel beam-to-column connection by using a micro-mechanics based fracture model. Int J Fatigue. 2013;48:90-100. [11] Nip KH, Gardner L, Davies CM, Elghazouli AY. Extremely low cycle fatigue tests on structural carbon steel and stainless steel. J Constr Steel Res. 2010;66:96-110. [12] Pereira JCR, de Jesus AMP, Xavier J, Fernandes AA. Ultra low-cycle fatigue behaviour of a structural steel. Eng Struct. 2014;60:214-22. [13] Kiran R, Khandelwal K. A micromechanical cyclic void growth model for ultra-low cycle fatigue. Int J Fatigue. 2015;70:24-37. [14] Jia L-J, Ikai T, Shinohara K, Ge H. Ductile crack initiation and propagation of structural steels under cyclic combined shear and normal stress loading. Constr Build Mater. 2016;112:69-83. [15] Jia L-J, Koyama T, Kuwamura H. Experimental and numerical study of postbuckling ductile fracture of heat-treated SHS stub columns. J Struct Eng (ASCE). 2014;140:04014044. [16] Jia L-J, Kuwamura H. Ductile fracture model for structural steel under cyclic large strain loading. J Constr Steel Res. 2015;106:110-21. [17] Hancock JW, Mackenzie AC. On the mechanisms of ductile failure in high-strength steels subjected to multi-axial stress-states. J Mech Phys Solids. 1976;24:147-60. [18] McClintock FA. A criterion for ductile fracture by the growth of holes. J Appl Mech. 1968;35:363-71. [19] Rousselier G. Ductile fracture models and their potential in local approach of fracture. Nucl Eng Des. 1987;105:97-111. [20] Jia L-J, Ge H, Shinohara K, Kato H. Experimental and numerical study on ductile fracture of structural steels under combined shear and tension. J Bridge Eng (ASCE). 2016:04016008. [21] Bao Y, Wierzbicki T. On fracture locus in the equivalent strain and stress triaxiality space. Int J Mech Sci. 2004;46:81-98. [22] Bao Y, Wierzbicki T. On the cut-off value of negative triaxiality for fracture. Eng Fract Mech. 2005;72:1049-69. [23] Xue L. Damage accumulation and fracture initiation in uncracked ductile solids subject to triaxial loading. Int J Solids Struct. 2007;44:5163-81. [24] Bao Y, Wierzbicki T. A comparative study on various ductile crack formation criteria. J Eng Mater Technol. 2004;126:314-24. [25] Xue L, Wierzbicki T. Ductile fracture initiation and propagation modeling using damage plasticity theory. Eng Fract Mech. 2008;75:3276-93. [26] Xue L, Wierzbicki T. Numerical simulation of fracture mode transition in ductile 18

plates. Int J Solids Struct. 2009;46:1423-35. [27] Khandelwal K, El-Tawil S. A finite strain continuum damage model for simulating ductile fracture in steels. Eng Fract Mech. 2014;116:172-89. [28] Kiran R, Khandelwal K. A micromechanical model for ductile fracture prediction in ASTM A992 steels. Eng Fract Mech. 2013;102:101-17. [29] Kiran R, Khandelwal K. A triaxiality and Lode parameter dependent ductile fracture criterion. Eng Fract Mech. 2014;128:121-38. [30] Jia L-J, Ge HB, Maruyama R, Shinohara K. Development of a novel high-performance all-steel fish-bone shaped buckling-restrained brace. Eng Struct. 2017;138:105-19. [31] Chen Y, Pan L, Jia L-J. Post-buckling ductile fracture analysis of panel zones in welded steel beam-to-column connections. J Constr Steel Res. 2017;132:117-29. [32] Pereira JCR, de Jesus AMP, Fernandes AA. A new ultra-low cycle fatigue model applied to the X60 piping steel. Int J Fatigue. 2016;93, Part 1:201-13. [33] Tateishi K, Hanji T, Minami K. A prediction model for extremely low cycle fatigue strength of structural steel. Int J Fatigue. 2007;29:887-96. [34] Xue L. A unified expression for low cycle fatigue and extremely low cycle fatigue and its implication for monotonic loading. Int J Fatigue. 2008;30:1691-8. [35] Algarni M, Choi Y, Bai Y. A unified material model for multiaxial ductile fracture and extremely low cycle fatigue of Inconel 718. Int J Fatigue. 2017;96:162-77. [36] Martinez X, Oller S, Barbu LG, Barbat AH, de Jesus AMP. Analysis of Ultra Low Cycle Fatigue problems with the Barcelona plastic damage model and a new isotropic hardening law. Int J Fatigue. 2015;73:132-42. [37] Kuroda M. Extremely low cycle fatigue life prediction based on a new cumulative fatigue damage model. Int J Fatigue. 2002;24:699-703. [38] Kermajani M, Ghaini FM, Miresmaeili R, Aghakouchak AA, Shadmand M. Effect of weld metal toughness on fracture behavior under ultra-low cycle fatigue loading (earthquake). Materials Science and Engineering: A. 2016;668:30-7. [39] Jia L-J, Ge HB, Suzuki T, Luo XQ. Experimental study on cracking of thick-walled welded beam-column connections with incomplete penetration in steel bridge piers. Journal of Bridge Engineering (ASCE). 2014;20:04014072. [40] Jia L-J, Ge HB, Suzuki T. Effect of post weld treatment on cracking behaviors of beam-column connections in steel bridge piers. Steel & Composite Structures. 2014;17:685-702. [41] Zhou H, Wang Y, Shi Y, Xiong J, Yang L. Extremely low cycle fatigue prediction of steel beam-to-column connection by using a micro-mechanics based fracture model. Int J Fatigue. 2013;48:90-100. [42] JRA. Specification for highway bridges. Tokyo: Japan Road Association; 2012. [43] Vergani L, Guagliano M, Souki I, Delagnes D, Lours P. 11th International Conference on the Mechanical Behavior of Materials (ICM11)Influence of heat treatment on the fracture toughness and crack propagation in 5% Cr martensitic steel. Procedia Engineering. 2011;10:631-7. [44] Xiang P, Nishitani A, Marutani S, Kodera K, Hatada T, Katamura R, et al. Identification of yield drift deformations and evaluation of the degree of damage 19

through the direct sensing of drift displacements. Earthquake Engineering & Structural Dynamics. 2016;45:2085-102.

20

Fig. 1 Illustration of ductile fracture process under monotonic tension[2]

Fig. 2 Different fracture modes of metal structures[4]

21

Fig. 3 Configurations of specimens

22

Fig. 4 Macrostructure of welded T-joints (as-welded specimen)

Fig. 5 Test setup

23

Fig. 6 Illustration for measurements and loading conditions

Fig. 7 Loading protocols

24

Fig. 8 Cracking process of Rinf-WB-C1

Fig. 9 Cracking process of Rinf-NWB-C1

Fig. 10 Cracking process of R0.25-NWB-C1-FL

Fig. 11 Cracking process of R1.0-NWB-C1-BM

25

Fig. 12 Cracking process of R1.0-NWB-C1-WD

Fig. 13 Cracking process of R1.0-NWB-C1-FL

Fig. 14 Cracking process of R1.0-NWB-C1-HAZ

26

Fig. 15 Cracking process of R4.0-NWB-C1-FL

Fig. 16 Cracking process of Rinf-WB-C2

Fig. 17 Cracking process of Rinf-WB-R

27

Fig. 18 Local strain at notch roots versus displacement curves

28

Fig. 19 Illustration of method to obtain cumulative ductility

Fig. 20 Crack depth versus cumulative ductility ratio curves

29

(a) Base metal

(c) HAZ

(b) Fusion line

(d) Weld deposit

Fig. 21 Charpy impact test results of different subzones

30

Fig. 22 Effect of notch size on strain concentration

31

Fig. 23 Macro fracture surfaces of specimens

32

Fig. 24 Fractographic observation of Specimen R1.0-NWB-C1-BM

Fig. 25 Fractographic observation of Specimen R1.0-NWB-C1-FL

Fig. 26 Fractographic observation of Specimen Rinf-WB-C2

33

Fig. 27 Typical load-displacement skeleton curves

Fig. 28 Load-displacement skeleton curves of specimens with different post-weld treatments

34

Fig. 29 Load-displacement skeleton curves of specimens with different notch sizes

35

Fig. 30 Load-displacement skeleton curves of specimens with different notch locations

36

Fig. 31 Load-displacement skeleton curves for tensile half cycles of specimens with different loading protocols

37

Predicted/Actual fatigue life

2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 0

Ductility ratio based formula Equivalent plastic strain based formula 12δy

10δy

8δy

Random

Fig. 32 Comparison of predicted and actual fatigue life of specimens with weld beads

38

Table 1. Mechanical properties and chemical composition of steel Mechanical Properties Steel

E

Poisson’s ratio

GPa SM490YA

201

0.28

Yield stress

Chemical Composition (weight%)

Tensile strength

Elongation

MPa

MPa

%

368

504

26.4

C

Si

Mn

P

S

0.12

0.24

1.29

0.018

0.005

Table 2. Measured dimensions of specimens (Unit: mm) Dimensions Specimens L

L1

L2

L3

h

t

b1

b2

f1

f2

f3

f4

w

d

Rinf-NWB-C1

650.3

317.2

318.0

15.1

200.3

16.0

98.9

98.6

24.8

24.0

13.0

13.0

/

/

Rinf-WB-C1

649.8

316.7

317.1

16.0

200.2

16.0

99.2

99.4

22.7

21.8

12.9

11.7

/

/

R0.25-NWB-C1-FL

650.2

317.4

317.0

15.8

200.1

15.9

99.2

100.0

21.3

21.9

14.5

12.0

2.5

2.0

R1.0-NWB-C1-BM

651.3

318.1

317.7

15.5

200.3

15.8

101.9

100.9

22.6

23.1

13.1

12.1

2.6

2.6

R1.0-NWB-C1-WD

650.5

317.5

317.8

15.2

201.0

15.8

99.5

100.5

22.1

23.0

12.8

12.8

2.4

2.1

R1.0-NWB-C1-FL

650.8

317.5

317.6

15.7

200.1

15.7

99.4

100.5

22.1

22.2

13.4

11.9

2.6

2.0

R1.0-NWB-C1-HAZ

650.6

317.6

317.5

15.5

200.0

15.8

99.6

100.3

21.5

21.4

13.8

12.2

2.5

2.3

R4.0-NWB-C1-FL

651.0

317.8

318.0

15.2

200.1

15.9

99.6

99.0

20.6

20.0

11.5

13.4

5.8

2.0

Rinf-WB-C2

650.8

317.8

317.3

15.7

199.9

15.9

99.7

99.6

23.0

23.2

13.4

12.0

/

/

Rinf-WB-C3

651.0

317.8

318.0

15.2

200.2

15.9

101.6

101.0

24.0

23.0

12.0

13.0

/

/

Rinf-WB-R

649.9

317.0

317.3

15.6

200.2

16.0

98.0

98.0

22.9

22.1

13.8

11.1

Notes: for the numbering of the specimens, R = notch size (“inf” denotes the ones without notches), WB and NWB respectively denotes specimens with and without weld bead.

39

Table 3. Number of half cycles at various instants No.

Specimens

Peak load

Crack initiation

Reduction of 50% peak load

1 2 3 4 5 6 7 8 9 10 11

Rinf-NWB-C1 Rinf-WB-C1 R0.25-NWB-C1-FL R1.0-NWB-C1-BM R1.0-NWB-C1-WD R1.0-NWB-C1-FL R1.0-NWB-C1-HAZ R4.0-NWB-C1-FL Rinf-WB-C2 Rinf-WB-C3 Rinf-WB-R

21

40

47

23

25

35

19

15

25

23

17

27

21

15

22

23

15

24

23

15

24

21

15

24

17

23

39

15

35

51

2

92

182

Table 4. Absorbed energy at various instants (Unit: kJ) Absorbed energy No.

Specimens

Peak load

Crack initiation

Reduction of 50% peak load

1 2 3 4 5 6 7 8 9 10 11

Rinf-NWB-C1 Rinf-WB-C1 R0.25-NWB-C1-FL R1.0-NWB-C1-BM R1.0-NWB-C1-WD R1.0-NWB-C1-FL R1.0-NWB-C1-HAZ R4.0-NWB-C1-FL Rinf-WB-C2 Rinf-WB-C3 Rinf-WB-R

19.3 25.8 11.8 21.8 16.3 20.5 20.5 19.2 9.0 12.2 4.7

72.6 31.6 8.7 10.6 8.1 8.4 8.3 8.6 24.9 37.9 55.9

83.3 53.7 24.3 26.9 17.7 21.5 21.5 24.8 57.9 57.0 94.8

Loading protocol

C1 C1 C1 C1 C1 C1 C1 C1 C2 C3 Random

40

Table 5. Mechanical properties of specimens No.

Specimen

Yield

Yield

strength

displacement

(kN)

(mm)

Peak tensile load

Displacement at peak

(kN)

tensile load (mm)

1

Rinf-NWB-C1

13.4

5.39

59.0

58.3

2

Rinf-WB-C1

15.1

5.38

69.3

64.6

3

R0.25-NWB-C1-FL

15.2

5.39

45.2

53.7

4

R1.0-NWB-C1-BM

12.6

5.38

49.6

64.5

5

R1.0-NWB-C1-WD

13.3

5.38

44.0

59.6

6

R1.0-NWB-C1-FL

14.2

5.39

52.4

64.6

7

R1.0-NWB-C1-HAZ

13.5

5.38

42.6

64.5

8

R4.0-NWB-C1-FL

14.4

5.41

51.3

59.2

9

Rinf-WB-C2

15.2

5.39

57.5

48.3

10

Rinf-WB-C3

16.1

5.39

50.3

43.1

11

Rinf-WB-R

15.9

5.38

76.6

80.7

41

Highlights: 1. Few studies on fracture of welded steel joints under cyclic large plastic strain loading conducted in the literature. 2. A ductile-fatigue transition fracture mode verified. 3. Effects of post-weld treatment, notch size, notch location and loading protocols on the fracture process investigated. 4. Macroscopic examination tests conducted for all the specimens to ensure accuracy of the tests. 5. Fracture life evaluation methods respectively based on ductility ratio and equivalent plastic strain proposed for the transition fracture mode.

42